From 821d5b0a2859097628c84b7ff5ff41614133b936 Mon Sep 17 00:00:00 2001 From: Bassem Girgis Date: Wed, 22 Jan 2020 15:55:26 -0600 Subject: [PATCH] upgrade cereal --- noarch/include/cereal/access.hpp | 107 +-- noarch/include/cereal/archives/binary.hpp | 12 +- noarch/include/cereal/archives/json.hpp | 42 +- .../cereal/archives/portable_binary.hpp | 26 +- noarch/include/cereal/archives/xml.hpp | 91 +- noarch/include/cereal/cereal.hpp | 120 ++- noarch/include/cereal/details/helpers.hpp | 43 +- .../cereal/details/polymorphic_impl.hpp | 90 +- .../include/cereal/details/static_object.hpp | 8 +- noarch/include/cereal/details/traits.hpp | 40 +- noarch/include/cereal/external/base64.hpp | 25 +- .../cereal/external/rapidjson/allocators.h | 17 +- .../external/rapidjson/cursorstreamwrapper.h | 78 ++ .../cereal/external/rapidjson/document.h | 347 +++++--- .../cereal/external/rapidjson/encodedstream.h | 2 +- .../cereal/external/rapidjson/encodings.h | 86 +- .../cereal/external/rapidjson/error/error.h | 10 +- .../external/rapidjson/filereadstream.h | 4 +- .../external/rapidjson/filewritestream.h | 4 +- .../external/rapidjson/internal/biginteger.h | 4 +- .../external/rapidjson/internal/diyfp.h | 37 +- .../cereal/external/rapidjson/internal/dtoa.h | 8 +- .../external/rapidjson/internal/ieee754.h | 4 +- .../cereal/external/rapidjson/internal/itoa.h | 82 +- .../cereal/external/rapidjson/internal/meta.h | 9 +- .../external/rapidjson/internal/regex.h | 319 ++++--- .../external/rapidjson/internal/stack.h | 10 +- .../external/rapidjson/internal/strfunc.h | 14 + .../external/rapidjson/internal/strtod.h | 111 ++- .../external/rapidjson/istreamwrapper.h | 98 ++- .../cereal/external/rapidjson/memorystream.h | 15 +- .../cereal/external/rapidjson/pointer.h | 92 +- .../cereal/external/rapidjson/prettywriter.h | 60 +- .../cereal/external/rapidjson/rapidjson.h | 113 ++- .../cereal/external/rapidjson/reader.h | 599 ++++++++++--- .../cereal/external/rapidjson/schema.h | 717 +++++++++++++--- .../cereal/external/rapidjson/stream.h | 52 +- .../cereal/external/rapidjson/stringbuffer.h | 4 + .../cereal/external/rapidjson/writer.h | 161 +++- .../cereal/external/rapidxml/license.txt | 104 +-- .../cereal/external/rapidxml/manual.html | 810 +++++++++--------- .../cereal/external/rapidxml/rapidxml.hpp | 88 +- .../external/rapidxml/rapidxml_print.hpp | 2 + noarch/include/cereal/macros.hpp | 19 + noarch/include/cereal/specialize.hpp | 139 +++ noarch/include/cereal/types/atomic.hpp | 55 ++ noarch/include/cereal/types/base_class.hpp | 6 +- noarch/include/cereal/types/bitset.hpp | 4 +- noarch/include/cereal/types/boost_variant.hpp | 128 ++- noarch/include/cereal/types/memory.hpp | 54 +- noarch/include/cereal/types/optional.hpp | 66 ++ noarch/include/cereal/types/polymorphic.hpp | 24 +- noarch/include/cereal/types/tuple.hpp | 8 +- noarch/include/cereal/types/variant.hpp | 109 +++ noarch/include/cereal/types/vector.hpp | 12 +- noarch/include/cereal/version.hpp | 52 ++ 56 files changed, 3717 insertions(+), 1624 deletions(-) create mode 100644 noarch/include/cereal/external/rapidjson/cursorstreamwrapper.h create mode 100644 noarch/include/cereal/specialize.hpp create mode 100644 noarch/include/cereal/types/atomic.hpp create mode 100644 noarch/include/cereal/types/optional.hpp create mode 100644 noarch/include/cereal/types/variant.hpp create mode 100644 noarch/include/cereal/version.hpp diff --git a/noarch/include/cereal/access.hpp b/noarch/include/cereal/access.hpp index 63c1f74a..d7917bdf 100644 --- a/noarch/include/cereal/access.hpp +++ b/noarch/include/cereal/access.hpp @@ -1,5 +1,5 @@ /*! \file access.hpp - \brief Access control, default construction, and serialization disambiguation */ + \brief Access control and default construction */ /* Copyright (c) 2014, Randolph Voorhies, Shane Grant All rights reserved. @@ -35,6 +35,7 @@ #include #include "cereal/macros.hpp" +#include "cereal/specialize.hpp" #include "cereal/details/helpers.hpp" namespace cereal @@ -110,6 +111,7 @@ namespace cereal // forward decl for construct //! @cond PRIVATE_NEVERDEFINED namespace memory_detail{ template struct LoadAndConstructLoadWrapper; } + namespace boost_variant_detail{ template struct LoadAndConstructLoadWrapper; } //! @endcond //! Used to construct types with no default constructor @@ -202,7 +204,8 @@ namespace cereal } private: - template friend struct ::cereal::memory_detail::LoadAndConstructLoadWrapper; + template friend struct ::cereal::memory_detail::LoadAndConstructLoadWrapper; + template friend struct ::cereal::boost_variant_detail::LoadAndConstructLoadWrapper; construct( T * p ) : itsPtr( p ), itsEnableSharedRestoreFunction( [](){} ), itsValid( false ) {} construct( T * p, std::function enableSharedFunc ) : // g++4.7 ice with default lambda to std func @@ -331,106 +334,6 @@ namespace cereal } }; // end class access - // ###################################################################### - //! A specifier used in conjunction with cereal::specialize to disambiguate - //! serialization in special cases - /*! @relates specialize - @ingroup Access */ - enum class specialization - { - member_serialize, //!< Force the use of a member serialize function - member_load_save, //!< Force the use of a member load/save pair - member_load_save_minimal, //!< Force the use of a member minimal load/save pair - non_member_serialize, //!< Force the use of a non-member serialize function - non_member_load_save, //!< Force the use of a non-member load/save pair - non_member_load_save_minimal //!< Force the use of a non-member minimal load/save pair - }; - - //! A class used to disambiguate cases where cereal cannot detect a unique way of serializing a class - /*! cereal attempts to figure out which method of serialization (member vs. non-member serialize - or load/save pair) at compile time. If for some reason cereal cannot find a non-ambiguous way - of serializing a type, it will produce a static assertion complaining about this. - - This can happen because you have both a serialize and load/save pair, or even because a base - class has a serialize (public or private with friend access) and a derived class does not - overwrite this due to choosing some other serialization type. - - Specializing this class will tell cereal to explicitly use the serialization type you specify - and it will not complain about ambiguity in its compile time selection. However, if cereal detects - an ambiguity in specializations, it will continue to issue a static assertion. - - @code{.cpp} - class MyParent - { - friend class cereal::access; - template - void serialize( Archive & ar ) {} - }; - - // Although serialize is private in MyParent, to cereal::access it will look public, - // even through MyDerived - class MyDerived : public MyParent - { - public: - template - void load( Archive & ar ) {} - - template - void save( Archive & ar ) {} - }; - - // The load/save pair in MyDerived is ambiguous because serialize in MyParent can - // be accessed from cereal::access. This looks the same as making serialize public - // in MyParent, making it seem as though MyDerived has both a serialize and a load/save pair. - // cereal will complain about this at compile time unless we disambiguate: - - namespace cereal - { - // This struct specialization will tell cereal which is the right way to serialize the ambiguity - template struct specialize {}; - - // If we only had a disambiguation for a specific archive type, it would look something like this - template <> struct specialize {}; - } - @endcode - - You can also choose to use the macros CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES or - CEREAL_SPECIALIZE_FOR_ARCHIVE if you want to type a little bit less. - - @tparam T The type to specialize the serialization for - @tparam S The specialization type to use for T - @ingroup Access */ - template - struct specialize : public std::false_type {}; - - //! Convenient macro for performing specialization for all archive types - /*! This performs specialization for the specific type for all types of archives. - This macro should be placed at the global namespace. - - @code{cpp} - struct MyType {}; - CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES( MyType, cereal::specialization::member_load_save ); - @endcode - - @relates specialize - @ingroup Access */ - #define CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES( Type, Specialization ) \ - namespace cereal { template struct specialize {}; } - - //! Convenient macro for performing specialization for a single archive type - /*! This performs specialization for the specific type for a single type of archive. - This macro should be placed at the global namespace. - - @code{cpp} - struct MyType {}; - CEREAL_SPECIALIZE_FOR_ARCHIVE( cereal::XMLInputArchive, MyType, cereal::specialization::member_load_save ); - @endcode - - @relates specialize - @ingroup Access */ - #define CEREAL_SPECIALIZE_FOR_ARCHIVE( Archive, Type, Specialization ) \ - namespace cereal { template <> struct specialize {}; } - // ###################################################################### // Deferred Implementation, see construct for more information template template inline diff --git a/noarch/include/cereal/archives/binary.hpp b/noarch/include/cereal/archives/binary.hpp index 7e003226..d8c53183 100644 --- a/noarch/include/cereal/archives/binary.hpp +++ b/noarch/include/cereal/archives/binary.hpp @@ -62,9 +62,9 @@ namespace cereal ~BinaryOutputArchive() CEREAL_NOEXCEPT = default; //! Writes size bytes of data to the output stream - void saveBinary( const void * data, std::size_t size ) + void saveBinary( const void * data, std::streamsize size ) { - auto const writtenSize = static_cast( itsStream.rdbuf()->sputn( reinterpret_cast( data ), size ) ); + auto const writtenSize = itsStream.rdbuf()->sputn( reinterpret_cast( data ), size ); if(writtenSize != size) throw Exception("Failed to write " + std::to_string(size) + " bytes to output stream! Wrote " + std::to_string(writtenSize)); @@ -97,9 +97,9 @@ namespace cereal ~BinaryInputArchive() CEREAL_NOEXCEPT = default; //! Reads size bytes of data from the input stream - void loadBinary( void * const data, std::size_t size ) + void loadBinary( void * const data, std::streamsize size ) { - auto const readSize = static_cast( itsStream.rdbuf()->sgetn( reinterpret_cast( data ), size ) ); + auto const readSize = itsStream.rdbuf()->sgetn( reinterpret_cast( data ), size ); if(readSize != size) throw Exception("Failed to read " + std::to_string(size) + " bytes from input stream! Read " + std::to_string(readSize)); @@ -148,14 +148,14 @@ namespace cereal template inline void CEREAL_SAVE_FUNCTION_NAME(BinaryOutputArchive & ar, BinaryData const & bd) { - ar.saveBinary( bd.data, static_cast( bd.size ) ); + ar.saveBinary( bd.data, static_cast( bd.size ) ); } //! Loading binary data template inline void CEREAL_LOAD_FUNCTION_NAME(BinaryInputArchive & ar, BinaryData & bd) { - ar.loadBinary(bd.data, static_cast(bd.size)); + ar.loadBinary(bd.data, static_cast( bd.size ) ); } } // namespace cereal diff --git a/noarch/include/cereal/archives/json.hpp b/noarch/include/cereal/archives/json.hpp index 9d57434a..82312e32 100644 --- a/noarch/include/cereal/archives/json.hpp +++ b/noarch/include/cereal/archives/json.hpp @@ -40,6 +40,11 @@ namespace cereal { RapidJSONException( const char * what_ ) : Exception( what_ ) {} }; } +// Inform rapidjson that assert will throw +#ifndef CEREAL_RAPIDJSON_ASSERT_THROWS +#define CEREAL_RAPIDJSON_ASSERT_THROWS +#endif // CEREAL_RAPIDJSON_ASSERT_THROWS + // Override rapidjson assertions to throw exceptions by default #ifndef CEREAL_RAPIDJSON_ASSERT #define CEREAL_RAPIDJSON_ASSERT(x) if(!(x)){ \ @@ -47,8 +52,14 @@ namespace cereal #endif // RAPIDJSON_ASSERT // Enable support for parsing of nan, inf, -inf +#ifndef CEREAL_RAPIDJSON_WRITE_DEFAULT_FLAGS #define CEREAL_RAPIDJSON_WRITE_DEFAULT_FLAGS kWriteNanAndInfFlag +#endif + +// Enable support for parsing of nan, inf, -inf +#ifndef CEREAL_RAPIDJSON_PARSE_DEFAULT_FLAGS #define CEREAL_RAPIDJSON_PARSE_DEFAULT_FLAGS kParseFullPrecisionFlag | kParseNanAndInfFlag +#endif #include "cereal/external/rapidjson/prettywriter.h" #include "cereal/external/rapidjson/ostreamwrapper.h" @@ -209,11 +220,13 @@ namespace cereal { case NodeType::StartArray: itsWriter.StartArray(); + // fall through case NodeType::InArray: itsWriter.EndArray(); break; case NodeType::StartObject: itsWriter.StartObject(); + // fall through case NodeType::InObject: itsWriter.EndObject(); break; @@ -477,7 +490,7 @@ namespace cereal } Iterator(ValueIterator begin, ValueIterator end) : - itsValueItBegin(begin), itsValueItEnd(end), itsIndex(0), itsType(Value) + itsValueItBegin(begin), itsIndex(0), itsType(Value) { if( std::distance( begin, end ) == 0 ) itsType = Null_; @@ -532,7 +545,7 @@ namespace cereal private: MemberIterator itsMemberItBegin, itsMemberItEnd; //!< The member iterator (object) - ValueIterator itsValueItBegin, itsValueItEnd; //!< The value iterator (array) + ValueIterator itsValueItBegin; //!< The value iterator (array) size_t itsIndex; //!< The current index of this iterator enum Type {Value, Member, Null_} itsType; //!< Whether this holds values (array) or members (objects) or nothing }; @@ -751,6 +764,31 @@ namespace cereal void epilogue( JSONInputArchive &, NameValuePair const & ) { } + // ###################################################################### + //! Prologue for deferred data for JSON archives + /*! Do nothing for the defer wrapper */ + template inline + void prologue( JSONOutputArchive &, DeferredData const & ) + { } + + //! Prologue for deferred data for JSON archives + template inline + void prologue( JSONInputArchive &, DeferredData const & ) + { } + + // ###################################################################### + //! Epilogue for deferred for JSON archives + /*! NVPs do not start or finish nodes - they just set up the names */ + template inline + void epilogue( JSONOutputArchive &, DeferredData const & ) + { } + + //! Epilogue for deferred for JSON archives + /*! Do nothing for the defer wrapper */ + template inline + void epilogue( JSONInputArchive &, DeferredData const & ) + { } + // ###################################################################### //! Prologue for SizeTags for JSON archives /*! SizeTags are strictly ignored for JSON, they just indicate diff --git a/noarch/include/cereal/archives/portable_binary.hpp b/noarch/include/cereal/archives/portable_binary.hpp index 76e634ba..59b72a6c 100644 --- a/noarch/include/cereal/archives/portable_binary.hpp +++ b/noarch/include/cereal/archives/portable_binary.hpp @@ -128,19 +128,19 @@ namespace cereal ~PortableBinaryOutputArchive() CEREAL_NOEXCEPT = default; //! Writes size bytes of data to the output stream - template inline - void saveBinary( const void * data, std::size_t size ) + template inline + void saveBinary( const void * data, std::streamsize size ) { - std::size_t writtenSize = 0; + std::streamsize writtenSize = 0; if( itsConvertEndianness ) { - for( std::size_t i = 0; i < size; i += DataSize ) - for( std::size_t j = 0; j < DataSize; ++j ) - writtenSize += static_cast( itsStream.rdbuf()->sputn( reinterpret_cast( data ) + DataSize - j - 1 + i, 1 ) ); + for( std::streamsize i = 0; i < size; i += DataSize ) + for( std::streamsize j = 0; j < DataSize; ++j ) + writtenSize += itsStream.rdbuf()->sputn( reinterpret_cast( data ) + DataSize - j - 1 + i, 1 ); } else - writtenSize = static_cast( itsStream.rdbuf()->sputn( reinterpret_cast( data ), size ) ); + writtenSize = itsStream.rdbuf()->sputn( reinterpret_cast( data ), size ); if(writtenSize != size) throw Exception("Failed to write " + std::to_string(size) + " bytes to output stream! Wrote " + std::to_string(writtenSize)); @@ -235,11 +235,11 @@ namespace cereal /*! @param data The data to save @param size The number of bytes in the data @tparam DataSize T The size of the actual type of the data elements being loaded */ - template inline - void loadBinary( void * const data, std::size_t size ) + template inline + void loadBinary( void * const data, std::streamsize size ) { // load data - auto const readSize = static_cast( itsStream.rdbuf()->sgetn( reinterpret_cast( data ), size ) ); + auto const readSize = itsStream.rdbuf()->sgetn( reinterpret_cast( data ), size ); if(readSize != size) throw Exception("Failed to read " + std::to_string(size) + " bytes from input stream! Read " + std::to_string(readSize)); @@ -248,7 +248,7 @@ namespace cereal if( itsConvertEndianness ) { std::uint8_t * ptr = reinterpret_cast( data ); - for( std::size_t i = 0; i < size; i += DataSize ) + for( std::streamsize i = 0; i < size; i += DataSize ) portable_binary_detail::swap_bytes( ptr + i ); } } @@ -308,7 +308,7 @@ namespace cereal (std::is_floating_point::value && std::numeric_limits::is_iec559), "Portable binary only supports IEEE 754 standardized floating point" ); - ar.template saveBinary( bd.data, static_cast( bd.size ) ); + ar.template saveBinary( bd.data, static_cast( bd.size ) ); } //! Loading binary data from portable binary @@ -320,7 +320,7 @@ namespace cereal (std::is_floating_point::value && std::numeric_limits::is_iec559), "Portable binary only supports IEEE 754 standardized floating point" ); - ar.template loadBinary( bd.data, static_cast( bd.size ) ); + ar.template loadBinary( bd.data, static_cast( bd.size ) ); } } // namespace cereal diff --git a/noarch/include/cereal/archives/xml.hpp b/noarch/include/cereal/archives/xml.hpp index c520f02e..23b41269 100644 --- a/noarch/include/cereal/archives/xml.hpp +++ b/noarch/include/cereal/archives/xml.hpp @@ -101,31 +101,58 @@ namespace cereal //! @{ //! A class containing various advanced options for the XML archive + /*! Options can either be directly passed to the constructor, or chained using the + modifier functions for an interface analogous to named parameters */ class Options { public: //! Default options static Options Default(){ return Options(); } - //! Default options with no indentation - static Options NoIndent(){ return Options( std::numeric_limits::max_digits10, false ); } - //! Specify specific options for the XMLOutputArchive - /*! @param precision The precision used for floating point numbers - @param indent Whether to indent each line of XML - @param outputType Whether to output the type of each serialized object as an attribute */ - explicit Options( int precision = std::numeric_limits::max_digits10, - bool indent = true, - bool outputType = false ) : - itsPrecision( precision ), - itsIndent( indent ), - itsOutputType( outputType ) { } + /*! @param precision_ The precision used for floating point numbers + @param indent_ Whether to indent each line of XML + @param outputType_ Whether to output the type of each serialized object as an attribute + @param sizeAttributes_ Whether dynamically sized containers output the size=dynamic attribute */ + explicit Options( int precision_ = std::numeric_limits::max_digits10, + bool indent_ = true, + bool outputType_ = false, + bool sizeAttributes_ = true ) : + itsPrecision( precision_ ), + itsIndent( indent_ ), + itsOutputType( outputType_ ), + itsSizeAttributes( sizeAttributes_ ) + { } + + /*! @name Option Modifiers + An interface for setting option settings analogous to named parameters. + + @code{cpp} + cereal::XMLOutputArchive ar( myStream, + cereal::XMLOutputArchive::Options() + .indent(true) + .sizeAttributes(false) ); + @endcode + */ + //! @{ + + //! Sets the precision used for floaing point numbers + Options & precision( int value ){ itsPrecision = value; return * this; } + //! Whether to indent each line of XML + Options & indent( bool enable ){ itsIndent = enable; return *this; } + //! Whether to output the type of each serialized object as an attribute + Options & outputType( bool enable ){ itsOutputType = enable; return *this; } + //! Whether dynamically sized containers (e.g. vector) output the size=dynamic attribute + Options & sizeAttributes( bool enable ){ itsSizeAttributes = enable; return *this; } + + //! @} private: friend class XMLOutputArchive; int itsPrecision; bool itsIndent; bool itsOutputType; + bool itsSizeAttributes; }; //! Construct, outputting to the provided stream upon destruction @@ -137,7 +164,8 @@ namespace cereal OutputArchive(this), itsStream(stream), itsOutputType( options.itsOutputType ), - itsIndent( options.itsIndent ) + itsIndent( options.itsIndent ), + itsSizeAttributes(options.itsSizeAttributes) { // rapidxml will delete all allocations when xml_document is cleared auto node = itsXML.allocate_node( rapidxml::node_declaration ); @@ -182,7 +210,7 @@ namespace cereal itsNodes.top().node->append_attribute( itsXML.allocate_attribute( "type", "cereal binary data" ) ); finishNode(); - }; + } //! @} /*! @name Internal Functionality @@ -289,6 +317,8 @@ namespace cereal itsNodes.top().node->append_attribute( itsXML.allocate_attribute( namePtr, valuePtr ) ); } + bool hasSizeAttributes() const { return itsSizeAttributes; } + protected: //! A struct that contains metadata about a node struct NodeInfo @@ -330,6 +360,7 @@ namespace cereal std::ostringstream itsOS; //!< Used to format strings internally bool itsOutputType; //!< Controls whether type information is printed bool itsIndent; //!< Controls whether indenting is used + bool itsSizeAttributes; //!< Controls whether lists have a size attribute }; // XMLOutputArchive // ###################################################################### @@ -436,7 +467,7 @@ namespace cereal std::memcpy( data, decoded.data(), decoded.size() ); finishNode(); - }; + } //! @} /*! @name Internal Functionality @@ -758,13 +789,41 @@ namespace cereal void epilogue( XMLInputArchive &, NameValuePair const & ) { } + // ###################################################################### + //! Prologue for deferred data for XML archives + /*! Do nothing for the defer wrapper */ + template inline + void prologue( XMLOutputArchive &, DeferredData const & ) + { } + + //! Prologue for deferred data for XML archives + template inline + void prologue( XMLInputArchive &, DeferredData const & ) + { } + + // ###################################################################### + //! Epilogue for deferred for XML archives + /*! NVPs do not start or finish nodes - they just set up the names */ + template inline + void epilogue( XMLOutputArchive &, DeferredData const & ) + { } + + //! Epilogue for deferred for XML archives + /*! Do nothing for the defer wrapper */ + template inline + void epilogue( XMLInputArchive &, DeferredData const & ) + { } + // ###################################################################### //! Prologue for SizeTags for XML output archives /*! SizeTags do not start or finish nodes */ template inline void prologue( XMLOutputArchive & ar, SizeTag const & ) { - ar.appendAttribute( "size", "dynamic" ); + if (ar.hasSizeAttributes()) + { + ar.appendAttribute("size", "dynamic"); + } } template inline diff --git a/noarch/include/cereal/cereal.hpp b/noarch/include/cereal/cereal.hpp index a7770103..50a76ee5 100644 --- a/noarch/include/cereal/cereal.hpp +++ b/noarch/include/cereal/cereal.hpp @@ -32,6 +32,7 @@ #include #include #include +#include #include #include #include @@ -97,6 +98,59 @@ namespace cereal return {std::forward(sz)}; } + // ###################################################################### + //! Marks data for deferred serialization + /*! cereal performs a recursive depth-first traversal of data it serializes. When + serializing smart pointers to large, nested, or cyclical data structures, it + is possible to encounter a stack overflow from excessive recursion when following + a chain of pointers. + + Deferment can help in these situations if the data can be serialized separately from + the pointers used to traverse the structure. For example, a graph structure can have its + nodes serialized before its edges: + + @code{.cpp} + struct MyEdge + { + std::shared_ptr connection; + int some_value; + + template + void serialize(Archive & archive) + { + // when we serialize an edge, we'll defer serializing the associated node + archive( cereal::defer( connection ), + some_value ); + } + }; + + struct MyGraphStructure + { + std::vector edges; + std::vector nodes; + + template + void serialize(Archive & archive) + { + // because of the deferment, we ensure all nodes are fully serialized + // before any connection pointers to those nodes are serialized + archive( edges, nodes ); + + // we have to explicitly inform the archive when it is safe to serialize + // the deferred data + archive.serializeDeferments(); + } + }; + @endcode + + @relates DeferredData + @ingroup Utility */ + template inline + DeferredData defer( T && value ) + { + return {std::forward(value)}; + } + // ###################################################################### //! Called before a type is serialized to set up any special archive state //! for processing some type @@ -145,6 +199,14 @@ namespace cereal instantiate_polymorphic_binding( T*, Archive*, BindingTag, adl_tag ); \ } } /* end namespaces */ + //! Helper macro to omit unused warning + #if defined(__GNUC__) + // GCC / clang don't want the function + #define CEREAL_UNUSED_FUNCTION + #else + #define CEREAL_UNUSED_FUNCTION static void unused() { (void)version; } + #endif + // ###################################################################### //! Defines a class version for some type /*! Versioning information is optional and adds some small amount of @@ -207,7 +269,7 @@ namespace cereal std::type_index(typeid(TYPE)).hash_code(), VERSION_NUMBER ); \ return VERSION_NUMBER; \ } \ - static void unused() { (void)version; } \ + CEREAL_UNUSED_FUNCTION \ }; /* end Version */ \ const std::uint32_t Version::version = \ Version::registerVersion(); \ @@ -250,6 +312,14 @@ namespace cereal return *self; } + //! Serializes any data marked for deferment using defer + /*! This will cause any data wrapped in DeferredData to be immediately serialized */ + void serializeDeferments() + { + for( auto & deferment : itsDeferments ) + deferment(); + } + /*! @name Boost Transition Layer Functionality that mirrors the syntax for Boost. This is useful if you are transitioning a large project from Boost to cereal. The preferred interface for cereal is using operator(). */ @@ -258,14 +328,14 @@ namespace cereal //! Indicates this archive is not intended for loading /*! This ensures compatibility with boost archive types. If you are transitioning from boost, you can check this value within a member or external serialize function - (i.e., Archive::is_loading::value) to disable behavior specific to loading, until + (i.e., Archive::is_loading::value) to disable behavior specific to loading, until you can transition to split save/load or save_minimal/load_minimal functions */ using is_loading = std::false_type; //! Indicates this archive is intended for saving /*! This ensures compatibility with boost archive types. If you are transitioning from boost, you can check this value within a member or external serialize function - (i.e., Archive::is_saving::value) to enable behavior specific to loading, until + (i.e., Archive::is_saving::value) to enable behavior specific to loading, until you can transition to split save/load or save_minimal/load_minimal functions */ using is_saving = std::true_type; @@ -379,6 +449,17 @@ namespace cereal return *self; } + std::vector> itsDeferments; + + template inline + ArchiveType & processImpl(DeferredData const & d) + { + std::function deferment( [=](){ self->process( d.value ); } ); + itsDeferments.emplace_back( std::move(deferment) ); + + return *self; + } + //! Helper macro that expands the requirements for activating an overload /*! Requirements: Has the requested serialization function @@ -483,8 +564,7 @@ namespace cereal /*! If this is the first time this class has been serialized, we will record its version number and serialize that. - @tparam T The type of the class being serialized - @param version The version number associated with it */ + @tparam T The type of the class being serialized */ template inline std::uint32_t registerClassVersion() { @@ -620,6 +700,14 @@ namespace cereal return *self; } + //! Serializes any data marked for deferment using defer + /*! This will cause any data wrapped in DeferredData to be immediately serialized */ + void serializeDeferments() + { + for( auto & deferment : itsDeferments ) + deferment(); + } + /*! @name Boost Transition Layer Functionality that mirrors the syntax for Boost. This is useful if you are transitioning a large project from Boost to cereal. The preferred interface for cereal is using operator(). */ @@ -628,14 +716,14 @@ namespace cereal //! Indicates this archive is intended for loading /*! This ensures compatibility with boost archive types. If you are transitioning from boost, you can check this value within a member or external serialize function - (i.e., Archive::is_loading::value) to enable behavior specific to loading, until + (i.e., Archive::is_loading::value) to enable behavior specific to loading, until you can transition to split save/load or save_minimal/load_minimal functions */ using is_loading = std::true_type; //! Indicates this archive is not intended for saving /*! This ensures compatibility with boost archive types. If you are transitioning from boost, you can check this value within a member or external serialize function - (i.e., Archive::is_saving::value) to disable behavior specific to loading, until + (i.e., Archive::is_saving::value) to disable behavior specific to loading, until you can transition to split save/load or save_minimal/load_minimal functions */ using is_saving = std::false_type; @@ -667,6 +755,7 @@ namespace cereal /*! This is used to retrieve a previously registered shared_ptr which has already been loaded. + @internal @param id The unique id that was serialized for the pointer @return A shared pointer to the data @throw Exception if the id does not exist */ @@ -685,6 +774,7 @@ namespace cereal /*! After a shared pointer has been allocated for the first time, it should be registered with its loaded id for future references to it. + @internal @param id The unique identifier for the shared pointer @param ptr The actual shared pointer */ inline void registerSharedPointer(std::uint32_t const id, std::shared_ptr ptr) @@ -697,6 +787,7 @@ namespace cereal /*! This is used to retrieve a string previously registered during a polymorphic load. + @internal @param id The unique id that was serialized for the polymorphic type @return The string identifier for the tyep */ inline std::string getPolymorphicName(std::uint32_t const id) @@ -713,6 +804,7 @@ namespace cereal /*! After a polymorphic type has been loaded for the first time, it should be registered with its loaded id for future references to it. + @internal @param id The unique identifier for the polymorphic type @param name The name associated with the tyep */ inline void registerPolymorphicName(std::uint32_t const id, std::string const & name) @@ -762,6 +854,17 @@ namespace cereal return *self; } + std::vector> itsDeferments; + + template inline + ArchiveType & processImpl(DeferredData const & d) + { + std::function deferment( [=](){ self->process( d.value ); } ); + itsDeferments.emplace_back( std::move(deferment) ); + + return *self; + } + //! Helper macro that expands the requirements for activating an overload /*! Requirements: Has the requested serialization function @@ -875,8 +978,7 @@ namespace cereal /*! If this is the first time this class has been serialized, we will record its version number and serialize that. - @tparam T The type of the class being serialized - @param version The version number associated with it */ + @tparam T The type of the class being serialized */ template inline std::uint32_t loadClassVersion() { diff --git a/noarch/include/cereal/details/helpers.hpp b/noarch/include/cereal/details/helpers.hpp index 90926880..7b4c4ad3 100644 --- a/noarch/include/cereal/details/helpers.hpp +++ b/noarch/include/cereal/details/helpers.hpp @@ -68,8 +68,10 @@ namespace cereal namespace detail { struct NameValuePairCore {}; //!< Traits struct for NVPs + struct DeferredDataCore {}; //!< Traits struct for DeferredData } + // ###################################################################### //! For holding name value pairs /*! This pairs a name (some string) with some value such that an archive can potentially take advantage of the pairing. @@ -210,7 +212,7 @@ namespace cereal { //! Internally store the pointer as a void *, keeping const if created with //! a const pointer - using PT = typename std::conditional::type>::value, + using PT = typename std::conditional::type>::type>::value, const void *, void *>::type; @@ -220,6 +222,43 @@ namespace cereal uint64_t size; //!< size in bytes }; + // ###################################################################### + //! A wrapper around data that should be serialized after all non-deferred data + /*! This class is used to demarcate data that can only be safely serialized after + any data not wrapped in this class. + + @internal */ + template + class DeferredData : detail::DeferredDataCore + { + private: + // If we get passed an array, keep the type as is, otherwise store + // a reference if we were passed an l value reference, else copy the value + using Type = typename std::conditional::type>::value, + typename std::remove_cv::type, + typename std::conditional::value, + T, + typename std::decay::type>::type>::type; + + // prevent nested nvps + static_assert( !std::is_base_of::value, + "Cannot defer DeferredData" ); + + DeferredData & operator=( DeferredData const & ) = delete; + + public: + //! Constructs a new NameValuePair + /*! @param v The value to defer. Ideally this should be an l-value reference so that + the value can be both loaded and saved to. If you pass an r-value reference, + the DeferredData will store a copy of it instead of a reference. Thus you should + only pass r-values in cases where this makes sense, such as the result of some + size() call. + @internal */ + DeferredData( T && v ) : value(std::forward(v)) {} + + Type value; + }; + // ###################################################################### namespace detail { @@ -256,7 +295,7 @@ namespace cereal struct adl_tag; // used during saving pointers - static const int32_t msb_32bit = 0x80000000; + static const uint32_t msb_32bit = 0x80000000; static const int32_t msb2_32bit = 0x40000000; } diff --git a/noarch/include/cereal/details/polymorphic_impl.hpp b/noarch/include/cereal/details/polymorphic_impl.hpp index 81128c8b..bee13ab4 100644 --- a/noarch/include/cereal/details/polymorphic_impl.hpp +++ b/noarch/include/cereal/details/polymorphic_impl.hpp @@ -56,6 +56,14 @@ #include #include +//! Helper macro to omit unused warning +#if defined(__GNUC__) + // GCC / clang don't want the function + #define CEREAL_BIND_TO_ARCHIVES_UNUSED_FUNCTION +#else + #define CEREAL_BIND_TO_ARCHIVES_UNUSED_FUNCTION static void unused() { (void)b; } +#endif + //! Binds a polymorhic type to all registered archives /*! This binds a polymorphic type to all compatible registered archives that have been registered with CEREAL_REGISTER_ARCHIVE. This must be called @@ -67,7 +75,7 @@ template<> \ struct init_binding<__VA_ARGS__> { \ static bind_to_archives<__VA_ARGS__> const & b; \ - static void unused() { (void)b; } \ + CEREAL_BIND_TO_ARCHIVES_UNUSED_FUNCTION \ }; \ bind_to_archives<__VA_ARGS__> const & init_binding<__VA_ARGS__>::b = \ ::cereal::detail::StaticObject< \ @@ -115,8 +123,10 @@ namespace cereal all registered mappings between base and derived types. */ struct PolymorphicCasters { + //! Maps from a derived type index to a set of chainable casters + using DerivedCasterMap = std::unordered_map>; //! Maps from base type index to a map from derived type index to caster - std::map>> map; + std::unordered_map map; std::multimap reverseMap; @@ -127,24 +137,26 @@ namespace cereal "Make sure you either serialize the base class at some point via cereal::base_class or cereal::virtual_base_class.\n" \ "Alternatively, manually register the association with CEREAL_REGISTER_POLYMORPHIC_RELATION."); - //! Checks if the mapping object that can perform the upcast or downcast + //! Checks if the mapping object that can perform the upcast or downcast exists, and returns it if so /*! Uses the type index from the base and derived class to find the matching - registered caster. If no matching caster exists, returns false. */ - static bool exists( std::type_index const & baseIndex, std::type_index const & derivedIndex ) + registered caster. If no matching caster exists, the bool in the pair will be false and the vector + reference should not be used. */ + static std::pair const &> + lookup_if_exists( std::type_index const & baseIndex, std::type_index const & derivedIndex ) { // First phase of lookup - match base type index auto const & baseMap = StaticObject::getInstance().map; auto baseIter = baseMap.find( baseIndex ); if (baseIter == baseMap.end()) - return false; + return {false, {}}; // Second phase - find a match from base to derived - auto & derivedMap = baseIter->second; + auto const & derivedMap = baseIter->second; auto derivedIter = derivedMap.find( derivedIndex ); if (derivedIter == derivedMap.end()) - return false; + return {false, {}}; - return true; + return {true, derivedIter->second}; } //! Gets the mapping object that can perform the upcast or downcast @@ -162,7 +174,7 @@ namespace cereal exceptionFunc(); // Second phase - find a match from base to derived - auto & derivedMap = baseIter->second; + auto const & derivedMap = baseIter->second; auto derivedIter = derivedMap.find( derivedIndex ); if( derivedIter == derivedMap.end() ) exceptionFunc(); @@ -176,8 +188,8 @@ namespace cereal { auto const & mapping = lookup( baseInfo, typeid(Derived), [&](){ UNREGISTERED_POLYMORPHIC_CAST_EXCEPTION(save) } ); - for( auto const * map : mapping ) - dptr = map->downcast( dptr ); + for( auto const * dmap : mapping ) + dptr = dmap->downcast( dptr ); return static_cast( dptr ); } @@ -213,6 +225,14 @@ namespace cereal #undef UNREGISTERED_POLYMORPHIC_CAST_EXCEPTION }; + #ifdef CEREAL_OLDER_GCC + #define CEREAL_EMPLACE_MAP(map, key, value) \ + map.insert( std::make_pair(std::move(key), std::move(value)) ); + #else // NOT CEREAL_OLDER_GCC + #define CEREAL_EMPLACE_MAP(map, key, value) \ + map.emplace( key, value ); + #endif // NOT_CEREAL_OLDER_GCC + //! Strongly typed derivation of PolymorphicCaster template struct PolymorphicVirtualCaster : PolymorphicCaster @@ -229,18 +249,16 @@ namespace cereal // First insert the relation Base->Derived const auto lock = StaticObject::lock(); auto & baseMap = StaticObject::getInstance().map; - auto lb = baseMap.lower_bound(baseKey); { - auto & derivedMap = baseMap.insert( lb, {baseKey, {}} )->second; - auto lbd = derivedMap.lower_bound(derivedKey); - auto & derivedVec = derivedMap.insert( lbd, { std::move(derivedKey), {}} )->second; + auto & derivedMap = baseMap.insert( {baseKey, PolymorphicCasters::DerivedCasterMap{}} ).first->second; + auto & derivedVec = derivedMap.insert( {derivedKey, {}} ).first->second; derivedVec.push_back( this ); } // Insert reverse relation Derived->Base auto & reverseMap = StaticObject::getInstance().reverseMap; - reverseMap.insert( {derivedKey, baseKey} ); + CEREAL_EMPLACE_MAP(reverseMap, derivedKey, baseKey); // Find all chainable unregistered relations /* The strategy here is to process only the nodes in the class hierarchy graph that have been @@ -254,28 +272,40 @@ namespace cereal // Checks whether there is a path from parent->child and returns a pair // dist is set to MAX if the path does not exist auto checkRelation = [](std::type_index const & parentInfo, std::type_index const & childInfo) -> - std::pair> + std::pair const &> { - if( PolymorphicCasters::exists( parentInfo, childInfo ) ) + auto result = PolymorphicCasters::lookup_if_exists( parentInfo, childInfo ); + if( result.first ) { - auto const & path = PolymorphicCasters::lookup( parentInfo, childInfo, [](){} ); + auto const & path = result.second; return {path.size(), path}; } else - return {std::numeric_limits::max(), {}}; + return {(std::numeric_limits::max)(), {}}; }; - std::stack parentStack; // Holds the parent nodes to be processed - std::set dirtySet; // Marks child nodes that have been changed - std::set processedParents; // Marks parent nodes that have been processed + std::stack parentStack; // Holds the parent nodes to be processed + std::vector dirtySet; // Marks child nodes that have been changed + std::unordered_set processedParents; // Marks parent nodes that have been processed + + // Checks if a child has been marked dirty + auto isDirty = [&](std::type_index const & c) + { + auto const dirtySetSize = dirtySet.size(); + for( size_t i = 0; i < dirtySetSize; ++i ) + if( dirtySet[i] == c ) + return true; + + return false; + }; // Begin processing the base key and mark derived as dirty parentStack.push( baseKey ); - dirtySet.insert( derivedKey ); + dirtySet.emplace_back( derivedKey ); while( !parentStack.empty() ) { - using Relations = std::multimap>>; + using Relations = std::unordered_multimap>>; Relations unregisteredRelations; // Defer insertions until after main loop to prevent iterator invalidation const auto parent = parentStack.top(); @@ -285,7 +315,7 @@ namespace cereal for( auto const & childPair : baseMap[parent] ) { const auto child = childPair.first; - if( dirtySet.count( child ) && baseMap.count( child ) ) + if( isDirty( child ) && baseMap.count( child ) ) { auto parentChildPath = checkRelation( parent, child ); @@ -340,11 +370,11 @@ namespace cereal { auto & derivedMap = baseMap.find( it.first )->second; derivedMap[it.second.first] = it.second.second; - reverseMap.insert( {it.second.first, it.first} ); + CEREAL_EMPLACE_MAP(reverseMap, it.second.first, it.first ); } // Mark current parent as modified - dirtySet.insert( parent ); + dirtySet.emplace_back( parent ); // Insert all parents of the current parent node that haven't yet been processed auto parentRange = reverseMap.equal_range( parent ); @@ -361,6 +391,8 @@ namespace cereal } // end chainable relations } // end PolymorphicVirtualCaster() + #undef CEREAL_EMPLACE_MAP + //! Performs the proper downcast with the templated types void const * downcast( void const * const ptr ) const override { diff --git a/noarch/include/cereal/details/static_object.hpp b/noarch/include/cereal/details/static_object.hpp index c9c888d7..cd30348c 100644 --- a/noarch/include/cereal/details/static_object.hpp +++ b/noarch/include/cereal/details/static_object.hpp @@ -48,7 +48,7 @@ # define CEREAL_DLL_EXPORT __declspec(dllexport) # define CEREAL_USED #else // clang or gcc -# define CEREAL_DLL_EXPORT +# define CEREAL_DLL_EXPORT __attribute__ ((visibility("default"))) # define CEREAL_USED __attribute__ ((__used__)) #endif @@ -67,13 +67,12 @@ namespace cereal class CEREAL_DLL_EXPORT StaticObject { private: - //! Forces instantiation at pre-execution time - static void instantiate( T const & ) {} static T & create() { static T t; - instantiate(instance); + //! Forces instantiation at pre-execution time + (void)instance; return t; } @@ -95,6 +94,7 @@ namespace cereal std::unique_lock lock; #else public: + LockGuard(LockGuard const &) = default; // prevents implicit copy ctor warning ~LockGuard() CEREAL_NOEXCEPT {} // prevents variable not used #endif }; diff --git a/noarch/include/cereal/details/traits.hpp b/noarch/include/cereal/details/traits.hpp index 3390bbd3..06fe424c 100644 --- a/noarch/include/cereal/details/traits.hpp +++ b/noarch/include/cereal/details/traits.hpp @@ -800,6 +800,18 @@ namespace cereal See notes from member load_minimal implementation. + Note that there should be an additional const check on load_minimal after the valid check, + but this currently interferes with many valid uses of minimal serialization. It has been + removed (see #565 on github) and previously was: + + @code + static_assert( check::const_valid || !check::exists, + "cereal detected an invalid serialization type parameter in non-member " #test_name ". " + #test_name " non-member functions must accept their serialization type by non-const reference" ); + @endcode + + See #132, #436, #263, and #565 on https://github.com/USCiLab/cereal for more details. + @param test_name The name to give the test (e.g. load_minimal or versioned_load_minimal) @param save_name The corresponding name the save test would have (e.g. save_minimal or versioned_save_minimal) @param versioned Either blank or the macro CEREAL_MAKE_VERSIONED_TEST */ @@ -847,9 +859,6 @@ namespace cereal static_assert( check::valid || !check::exists, "cereal detected different types in corresponding non-member " \ #test_name " and " #save_name " functions. \n " \ "the paramater to " #test_name " must be a constant reference to the type that " #save_name " returns." ); \ - static_assert( check::const_valid || !check::exists, \ - "cereal detected an invalid serialization type parameter in non-member " #test_name ". " \ - #test_name " non-member functions must accept their serialization type by non-const reference" ); \ }; \ } /* namespace detail */ \ \ @@ -869,17 +878,29 @@ namespace cereal #undef CEREAL_MAKE_HAS_NON_MEMBER_LOAD_MINIMAL_TEST // ###################################################################### + namespace detail + { + // const stripped away before reaching here, prevents errors on conversion from + // construct to construct + template + struct has_member_load_and_construct_impl : std::integral_constant( std::declval(), std::declval< ::cereal::construct&>() ) ), void>::value> + { }; + + template + struct has_member_versioned_load_and_construct_impl : std::integral_constant( std::declval(), std::declval< ::cereal::construct&>(), 0 ) ), void>::value> + { }; + } // namespace detail + //! Member load and construct check template - struct has_member_load_and_construct : std::integral_constant( std::declval(), std::declval< ::cereal::construct&>() ) ), void>::value> + struct has_member_load_and_construct : detail::has_member_load_and_construct_impl::type, A> { }; - // ###################################################################### //! Member load and construct check (versioned) template - struct has_member_versioned_load_and_construct : std::integral_constant( std::declval(), std::declval< ::cereal::construct&>(), 0 ) ), void>::value> + struct has_member_versioned_load_and_construct : detail::has_member_versioned_load_and_construct_impl::type, A> { }; // ###################################################################### @@ -901,7 +922,8 @@ namespace cereal }; \ } /* end namespace detail */ \ template \ - struct has_non_member_##test_name : std::integral_constant::value> {}; + struct has_non_member_##test_name : \ + std::integral_constant::type, A>::value> {}; // ###################################################################### //! Non member load and construct check diff --git a/noarch/include/cereal/external/base64.hpp b/noarch/include/cereal/external/base64.hpp index 0d63991e..ce32324f 100644 --- a/noarch/include/cereal/external/base64.hpp +++ b/noarch/include/cereal/external/base64.hpp @@ -25,6 +25,11 @@ #ifndef CEREAL_EXTERNAL_BASE64_HPP_ #define CEREAL_EXTERNAL_BASE64_HPP_ +#ifdef __GNUC__ +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wconversion" +#endif + #include namespace cereal @@ -67,8 +72,8 @@ namespace cereal char_array_3[j] = '\0'; char_array_4[0] = (char_array_3[0] & 0xfc) >> 2; - char_array_4[1] = static_cast(((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4)); - char_array_4[2] = static_cast(((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6)); + char_array_4[1] = ((char_array_3[0] & 0x03) << 4) + ((char_array_3[1] & 0xf0) >> 4); + char_array_4[2] = ((char_array_3[1] & 0x0f) << 2) + ((char_array_3[2] & 0xc0) >> 6); char_array_4[3] = char_array_3[2] & 0x3f; for (j = 0; (j < i + 1); j++) @@ -95,9 +100,9 @@ namespace cereal for (i = 0; i <4; i++) char_array_4[i] = static_cast(chars.find( char_array_4[i] )); - char_array_3[0] = static_cast((char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4)); - char_array_3[1] = static_cast(((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2)); - char_array_3[2] = static_cast(((char_array_4[2] & 0x3) << 6) + char_array_4[3]); + char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4); + char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2); + char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3]; for (i = 0; (i < 3); i++) ret += char_array_3[i]; @@ -112,9 +117,9 @@ namespace cereal for (j = 0; j <4; j++) char_array_4[j] = static_cast(chars.find( char_array_4[j] )); - char_array_3[0] = static_cast((char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4)); - char_array_3[1] = static_cast(((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2)); - char_array_3[2] = static_cast(((char_array_4[2] & 0x3) << 6) + char_array_4[3]); + char_array_3[0] = (char_array_4[0] << 2) + ((char_array_4[1] & 0x30) >> 4); + char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2); + char_array_3[2] = ((char_array_4[2] & 0x3) << 6) + char_array_4[3]; for (j = 0; (j < i - 1); j++) ret += char_array_3[j]; } @@ -123,5 +128,7 @@ namespace cereal } } // namespace base64 } // namespace cereal - +#ifdef __GNUC__ +#pragma GCC diagnostic pop +#endif #endif // CEREAL_EXTERNAL_BASE64_HPP_ diff --git a/noarch/include/cereal/external/rapidjson/allocators.h b/noarch/include/cereal/external/rapidjson/allocators.h index 554dc4a2..d375e28a 100644 --- a/noarch/include/cereal/external/rapidjson/allocators.h +++ b/noarch/include/cereal/external/rapidjson/allocators.h @@ -52,6 +52,19 @@ concept Allocator { \endcode */ + +/*! \def CEREAL_RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY + \ingroup CEREAL_RAPIDJSON_CONFIG + \brief User-defined kDefaultChunkCapacity definition. + + User can define this as any \c size that is a power of 2. +*/ + +#ifndef CEREAL_RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY +#define CEREAL_RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY (64 * 1024) +#endif + + /////////////////////////////////////////////////////////////////////////////// // CrtAllocator @@ -236,7 +249,7 @@ private: */ bool AddChunk(size_t capacity) { if (!baseAllocator_) - ownBaseAllocator_ = baseAllocator_ = CEREAL_RAPIDJSON_NEW(BaseAllocator()); + ownBaseAllocator_ = baseAllocator_ = CEREAL_RAPIDJSON_NEW(BaseAllocator)(); if (ChunkHeader* chunk = reinterpret_cast(baseAllocator_->Malloc(CEREAL_RAPIDJSON_ALIGN(sizeof(ChunkHeader)) + capacity))) { chunk->capacity = capacity; chunk->size = 0; @@ -248,7 +261,7 @@ private: return false; } - static const int kDefaultChunkCapacity = 64 * 1024; //!< Default chunk capacity. + static const int kDefaultChunkCapacity = CEREAL_RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY; //!< Default chunk capacity. //! Chunk header for perpending to each chunk. /*! Chunks are stored as a singly linked list. diff --git a/noarch/include/cereal/external/rapidjson/cursorstreamwrapper.h b/noarch/include/cereal/external/rapidjson/cursorstreamwrapper.h new file mode 100644 index 00000000..f3d20f7f --- /dev/null +++ b/noarch/include/cereal/external/rapidjson/cursorstreamwrapper.h @@ -0,0 +1,78 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef CEREAL_RAPIDJSON_CURSORSTREAMWRAPPER_H_ +#define CEREAL_RAPIDJSON_CURSORSTREAMWRAPPER_H_ + +#include "stream.h" + +#if defined(__GNUC__) +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(effc++) +#endif + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(4702) // unreachable code +CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +CEREAL_RAPIDJSON_NAMESPACE_BEGIN + + +//! Cursor stream wrapper for counting line and column number if error exists. +/*! + \tparam InputStream Any stream that implements Stream Concept +*/ +template > +class CursorStreamWrapper : public GenericStreamWrapper { +public: + typedef typename Encoding::Ch Ch; + + CursorStreamWrapper(InputStream& is): + GenericStreamWrapper(is), line_(1), col_(0) {} + + // counting line and column number + Ch Take() { + Ch ch = this->is_.Take(); + if(ch == '\n') { + line_ ++; + col_ = 0; + } else { + col_ ++; + } + return ch; + } + + //! Get the error line number, if error exists. + size_t GetLine() const { return line_; } + //! Get the error column number, if error exists. + size_t GetColumn() const { return col_; } + +private: + size_t line_; //!< Current Line + size_t col_; //!< Current Column +}; + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +CEREAL_RAPIDJSON_DIAG_POP +#endif + +#if defined(__GNUC__) +CEREAL_RAPIDJSON_DIAG_POP +#endif + +CEREAL_RAPIDJSON_NAMESPACE_END + +#endif // CEREAL_RAPIDJSON_CURSORSTREAMWRAPPER_H_ diff --git a/noarch/include/cereal/external/rapidjson/document.h b/noarch/include/cereal/external/rapidjson/document.h index 46faeb13..ca6da63d 100644 --- a/noarch/include/cereal/external/rapidjson/document.h +++ b/noarch/include/cereal/external/rapidjson/document.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef CEREAL_RAPIDJSON_DOCUMENT_H_ @@ -26,26 +26,21 @@ #include CEREAL_RAPIDJSON_DIAG_PUSH -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant -CEREAL_RAPIDJSON_DIAG_OFF(4244) // conversion from kXxxFlags to 'uint16_t', possible loss of data -#endif - #ifdef __clang__ CEREAL_RAPIDJSON_DIAG_OFF(padded) CEREAL_RAPIDJSON_DIAG_OFF(switch-enum) CEREAL_RAPIDJSON_DIAG_OFF(c++98-compat) +#elif defined(_MSC_VER) +CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +CEREAL_RAPIDJSON_DIAG_OFF(4244) // conversion from kXxxFlags to 'uint16_t', possible loss of data #endif #ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_OFF(effc++) -#if __GNUC__ >= 6 -CEREAL_RAPIDJSON_DIAG_OFF(terminate) // ignore throwing CEREAL_RAPIDJSON_ASSERT in CEREAL_RAPIDJSON_NOEXCEPT functions -#endif #endif // __GNUC__ #ifndef CEREAL_RAPIDJSON_NOMEMBERITERATORCLASS -#include // std::iterator, std::random_access_iterator_tag +#include // std::random_access_iterator_tag #endif #if CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS @@ -67,10 +62,16 @@ class GenericDocument; But a compiler (IBM XL C/C++ for AIX) have reported to have problem with that so it moved as a namespace scope struct. https://code.google.com/p/rapidjson/issues/detail?id=64 */ -template -struct GenericMember { +template +struct GenericMember { GenericValue name; //!< name of member (must be a string) GenericValue value; //!< value of member. + + // swap() for std::sort() and other potential use in STL. + friend inline void swap(GenericMember& a, GenericMember& b) CEREAL_RAPIDJSON_NOEXCEPT { + a.name.Swap(b.name); + a.value.Swap(b.value); + } }; /////////////////////////////////////////////////////////////////////////////// @@ -98,16 +99,13 @@ struct GenericMember { \see GenericMember, GenericValue::MemberIterator, GenericValue::ConstMemberIterator */ template -class GenericMemberIterator - : public std::iterator >::Type> { +class GenericMemberIterator { friend class GenericValue; template friend class GenericMemberIterator; typedef GenericMember PlainType; typedef typename internal::MaybeAddConst::Type ValueType; - typedef std::iterator BaseType; public: //! Iterator type itself @@ -117,12 +115,21 @@ public: //! Non-constant iterator type typedef GenericMemberIterator NonConstIterator; + /** \name std::iterator_traits support */ + //@{ + typedef ValueType value_type; + typedef ValueType * pointer; + typedef ValueType & reference; + typedef std::ptrdiff_t difference_type; + typedef std::random_access_iterator_tag iterator_category; + //@} + //! Pointer to (const) GenericMember - typedef typename BaseType::pointer Pointer; + typedef pointer Pointer; //! Reference to (const) GenericMember - typedef typename BaseType::reference Reference; + typedef reference Reference; //! Signed integer type (e.g. \c ptrdiff_t) - typedef typename BaseType::difference_type DifferenceType; + typedef difference_type DifferenceType; //! Default constructor (singular value) /*! Creates an iterator pointing to no element. @@ -198,17 +205,17 @@ private: // class-based member iterator implementation disabled, use plain pointers template -struct GenericMemberIterator; +class GenericMemberIterator; //! non-const GenericMemberIterator template -struct GenericMemberIterator { +class GenericMemberIterator { //! use plain pointer as iterator type typedef GenericMember* Iterator; }; //! const GenericMemberIterator template -struct GenericMemberIterator { +class GenericMemberIterator { //! use plain const pointer as iterator type typedef const GenericMember* Iterator; }; @@ -300,7 +307,7 @@ struct GenericStringRef { */ #endif explicit GenericStringRef(const CharType* str) - : s(str), length(internal::StrLen(str)){ CEREAL_RAPIDJSON_ASSERT(s != 0); } + : s(str), length(NotNullStrLen(str)) {} //! Create constant string reference from pointer and length #ifndef __clang__ // -Wdocumentation @@ -312,12 +319,10 @@ struct GenericStringRef { */ #endif GenericStringRef(const CharType* str, SizeType len) - : s(str), length(len) { CEREAL_RAPIDJSON_ASSERT(s != 0); } + : s(CEREAL_RAPIDJSON_LIKELY(str) ? str : emptyString), length(len) { CEREAL_RAPIDJSON_ASSERT(str != 0 || len == 0u); } GenericStringRef(const GenericStringRef& rhs) : s(rhs.s), length(rhs.length) {} - GenericStringRef& operator=(const GenericStringRef& rhs) { s = rhs.s; length = rhs.length; } - //! implicit conversion to plain CharType pointer operator const Ch *() const { return s; } @@ -325,11 +330,24 @@ struct GenericStringRef { const SizeType length; //!< length of the string (excluding the trailing NULL terminator) private: + SizeType NotNullStrLen(const CharType* str) { + CEREAL_RAPIDJSON_ASSERT(str != 0); + return internal::StrLen(str); + } + + /// Empty string - used when passing in a NULL pointer + static const Ch emptyString[]; + //! Disallow construction from non-const array template GenericStringRef(CharType (&str)[N]) /* = delete */; + //! Copy assignment operator not permitted - immutable type + GenericStringRef& operator=(const GenericStringRef& rhs) /* = delete */; }; +template +const CharType GenericStringRef::emptyString[] = { CharType() }; + //! Mark a character pointer as constant string /*! Mark a plain character pointer as a "string literal". This function can be used to avoid copying a character string to be referenced as a @@ -344,7 +362,7 @@ private: */ template inline GenericStringRef StringRef(const CharType* str) { - return GenericStringRef(str, internal::StrLen(str)); + return GenericStringRef(str); } //! Mark a character pointer as constant string @@ -410,7 +428,7 @@ namespace internal { template struct TypeHelper {}; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsBool(); } static bool Get(const ValueType& v) { return v.GetBool(); } @@ -418,7 +436,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, bool data, typename ValueType::AllocatorType&) { return v.SetBool(data); } }; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsInt(); } static int Get(const ValueType& v) { return v.GetInt(); } @@ -426,7 +444,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, int data, typename ValueType::AllocatorType&) { return v.SetInt(data); } }; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsUint(); } static unsigned Get(const ValueType& v) { return v.GetUint(); } @@ -434,7 +452,27 @@ struct TypeHelper { static ValueType& Set(ValueType& v, unsigned data, typename ValueType::AllocatorType&) { return v.SetUint(data); } }; -template +#ifdef _MSC_VER +CEREAL_RAPIDJSON_STATIC_ASSERT(sizeof(long) == sizeof(int)); +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsInt(); } + static long Get(const ValueType& v) { return v.GetInt(); } + static ValueType& Set(ValueType& v, long data) { return v.SetInt(data); } + static ValueType& Set(ValueType& v, long data, typename ValueType::AllocatorType&) { return v.SetInt(data); } +}; + +CEREAL_RAPIDJSON_STATIC_ASSERT(sizeof(unsigned long) == sizeof(unsigned)); +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsUint(); } + static unsigned long Get(const ValueType& v) { return v.GetUint(); } + static ValueType& Set(ValueType& v, unsigned long data) { return v.SetUint(data); } + static ValueType& Set(ValueType& v, unsigned long data, typename ValueType::AllocatorType&) { return v.SetUint(data); } +}; +#endif + +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsInt64(); } static int64_t Get(const ValueType& v) { return v.GetInt64(); } @@ -442,7 +480,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, int64_t data, typename ValueType::AllocatorType&) { return v.SetInt64(data); } }; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsUint64(); } static uint64_t Get(const ValueType& v) { return v.GetUint64(); } @@ -450,7 +488,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, uint64_t data, typename ValueType::AllocatorType&) { return v.SetUint64(data); } }; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsDouble(); } static double Get(const ValueType& v) { return v.GetDouble(); } @@ -458,7 +496,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, double data, typename ValueType::AllocatorType&) { return v.SetDouble(data); } }; -template +template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsFloat(); } static float Get(const ValueType& v) { return v.GetFloat(); } @@ -466,7 +504,7 @@ struct TypeHelper { static ValueType& Set(ValueType& v, float data, typename ValueType::AllocatorType&) { return v.SetFloat(data); } }; -template +template struct TypeHelper { typedef const typename ValueType::Ch* StringType; static bool Is(const ValueType& v) { return v.IsString(); } @@ -476,7 +514,7 @@ struct TypeHelper { }; #if CEREAL_RAPIDJSON_HAS_STDSTRING -template +template struct TypeHelper > { typedef std::basic_string StringType; static bool Is(const ValueType& v) { return v.IsString(); } @@ -485,7 +523,7 @@ struct TypeHelper > { }; #endif -template +template struct TypeHelper { typedef typename ValueType::Array ArrayType; static bool Is(const ValueType& v) { return v.IsArray(); } @@ -494,23 +532,23 @@ struct TypeHelper { static ValueType& Set(ValueType& v, ArrayType data, typename ValueType::AllocatorType&) { return v = data; } }; -template +template struct TypeHelper { typedef typename ValueType::ConstArray ArrayType; static bool Is(const ValueType& v) { return v.IsArray(); } static ArrayType Get(const ValueType& v) { return v.GetArray(); } }; -template +template struct TypeHelper { typedef typename ValueType::Object ObjectType; static bool Is(const ValueType& v) { return v.IsObject(); } static ObjectType Get(ValueType& v) { return v.GetObject(); } static ValueType& Set(ValueType& v, ObjectType data) { return v = data; } - static ValueType& Set(ValueType& v, ObjectType data, typename ValueType::AllocatorType&) { v = data; } + static ValueType& Set(ValueType& v, ObjectType data, typename ValueType::AllocatorType&) { return v = data; } }; -template +template struct TypeHelper { typedef typename ValueType::ConstObject ObjectType; static bool Is(const ValueType& v) { return v.IsObject(); } @@ -536,7 +574,7 @@ template class GenericObject; \tparam Encoding Encoding of the value. (Even non-string values need to have the same encoding in a document) \tparam Allocator Allocator type for allocating memory of object, array and string. */ -template > +template > class GenericValue { public: //! Name-value pair in an object. @@ -590,11 +628,11 @@ public: \note Default content for number is zero. */ explicit GenericValue(Type type) CEREAL_RAPIDJSON_NOEXCEPT : data_() { - static const uint16_t defaultFlags[7] = { + static const uint16_t defaultFlags[] = { kNullFlag, kFalseFlag, kTrueFlag, kObjectFlag, kArrayFlag, kShortStringFlag, kNumberAnyFlag }; - CEREAL_RAPIDJSON_ASSERT(type <= kNumberType); + CEREAL_RAPIDJSON_NOEXCEPT_ASSERT(type >= kNullType && type <= kNumberType); data_.f.flags = defaultFlags[type]; // Use ShortString to store empty string. @@ -607,10 +645,50 @@ public: \tparam SourceAllocator allocator of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator for allocating copied elements and buffers. Commonly use GenericDocument::GetAllocator(). + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) \see CopyFrom() */ - template< typename SourceAllocator > - GenericValue(const GenericValue& rhs, Allocator & allocator); + template + GenericValue(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { + switch (rhs.GetType()) { + case kObjectType: { + SizeType count = rhs.data_.o.size; + Member* lm = reinterpret_cast(allocator.Malloc(count * sizeof(Member))); + const typename GenericValue::Member* rm = rhs.GetMembersPointer(); + for (SizeType i = 0; i < count; i++) { + new (&lm[i].name) GenericValue(rm[i].name, allocator, copyConstStrings); + new (&lm[i].value) GenericValue(rm[i].value, allocator, copyConstStrings); + } + data_.f.flags = kObjectFlag; + data_.o.size = data_.o.capacity = count; + SetMembersPointer(lm); + } + break; + case kArrayType: { + SizeType count = rhs.data_.a.size; + GenericValue* le = reinterpret_cast(allocator.Malloc(count * sizeof(GenericValue))); + const GenericValue* re = rhs.GetElementsPointer(); + for (SizeType i = 0; i < count; i++) + new (&le[i]) GenericValue(re[i], allocator, copyConstStrings); + data_.f.flags = kArrayFlag; + data_.a.size = data_.a.capacity = count; + SetElementsPointer(le); + } + break; + case kStringType: + if (rhs.data_.f.flags == kConstStringFlag && !copyConstStrings) { + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + } + else + SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); + break; + default: + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + break; + } + } //! Constructor for boolean value. /*! \param b Boolean value @@ -638,7 +716,7 @@ public: //! Constructor for unsigned value. explicit GenericValue(unsigned u) CEREAL_RAPIDJSON_NOEXCEPT : data_() { - data_.n.u64 = u; + data_.n.u64 = u; data_.f.flags = (u & 0x80000000) ? kNumberUintFlag : (kNumberUintFlag | kIntFlag | kInt64Flag); } @@ -672,6 +750,9 @@ public: //! Constructor for double value. explicit GenericValue(double d) CEREAL_RAPIDJSON_NOEXCEPT : data_() { data_.n.d = d; data_.f.flags = kNumberDoubleFlag; } + //! Constructor for float value. + explicit GenericValue(float f) CEREAL_RAPIDJSON_NOEXCEPT : data_() { data_.n.d = static_cast(f); data_.f.flags = kNumberDoubleFlag; } + //! Constructor for constant string (i.e. do not make a copy of string) GenericValue(const Ch* s, SizeType length) CEREAL_RAPIDJSON_NOEXCEPT : data_() { SetStringRaw(StringRef(s, length)); } @@ -753,9 +834,10 @@ public: /*! \param rhs Source of the assignment. It will become a null value after assignment. */ GenericValue& operator=(GenericValue& rhs) CEREAL_RAPIDJSON_NOEXCEPT { - CEREAL_RAPIDJSON_ASSERT(this != &rhs); - this->~GenericValue(); - RawAssign(rhs); + if (CEREAL_RAPIDJSON_LIKELY(this != &rhs)) { + this->~GenericValue(); + RawAssign(rhs); + } return *this; } @@ -800,12 +882,13 @@ public: \tparam SourceAllocator Allocator type of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator to use for copying + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) */ template - GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator) { + GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { CEREAL_RAPIDJSON_ASSERT(static_cast(this) != static_cast(&rhs)); this->~GenericValue(); - new (this) GenericValue(rhs, allocator); + new (this) GenericValue(rhs, allocator, copyConstStrings); return *this; } @@ -846,7 +929,7 @@ public: //! Equal-to operator /*! \note If an object contains duplicated named member, comparing equality with any object is always \c false. - \note Linear time complexity (number of all values in the subtree and total lengths of all strings). + \note Complexity is quadratic in Object's member number and linear for the rest (number of all values in the subtree and total lengths of all strings). */ template bool operator==(const GenericValue& rhs) const { @@ -857,14 +940,14 @@ public: switch (GetType()) { case kObjectType: // Warning: O(n^2) inner-loop if (data_.o.size != rhs.data_.o.size) - return false; + return false; for (ConstMemberIterator lhsMemberItr = MemberBegin(); lhsMemberItr != MemberEnd(); ++lhsMemberItr) { typename RhsType::ConstMemberIterator rhsMemberItr = rhs.FindMember(lhsMemberItr->name); if (rhsMemberItr == rhs.MemberEnd() || lhsMemberItr->value != rhsMemberItr->value) return false; } return true; - + case kArrayType: if (data_.a.size != rhs.data_.a.size) return false; @@ -955,14 +1038,14 @@ public: uint64_t u = GetUint64(); volatile double d = static_cast(u); return (d >= 0.0) - && (d < static_cast(std::numeric_limits::max())) + && (d < static_cast((std::numeric_limits::max)())) && (u == static_cast(d)); } if (IsInt64()) { int64_t i = GetInt64(); volatile double d = static_cast(i); - return (d >= static_cast(std::numeric_limits::min())) - && (d < static_cast(std::numeric_limits::max())) + return (d >= static_cast((std::numeric_limits::min)())) + && (d < static_cast((std::numeric_limits::max)())) && (i == static_cast(d)); } return true; // double, int, uint are always lossless @@ -979,8 +1062,8 @@ public: bool IsLosslessFloat() const { if (!IsNumber()) return false; double a = GetDouble(); - if (a < static_cast(-std::numeric_limits::max()) - || a > static_cast(std::numeric_limits::max())) + if (a < static_cast(-(std::numeric_limits::max)()) + || a > static_cast((std::numeric_limits::max)())) return false; double b = static_cast(static_cast(a)); return a >= b && a <= b; // Prevent -Wfloat-equal @@ -1015,6 +1098,9 @@ public: //! Get the number of members in the object. SizeType MemberCount() const { CEREAL_RAPIDJSON_ASSERT(IsObject()); return data_.o.size; } + //! Get the capacity of object. + SizeType MemberCapacity() const { CEREAL_RAPIDJSON_ASSERT(IsObject()); return data_.o.capacity; } + //! Check whether the object is empty. bool ObjectEmpty() const { CEREAL_RAPIDJSON_ASSERT(IsObject()); return data_.o.size == 0; } @@ -1083,6 +1169,21 @@ public: /*! \pre IsObject() == true */ MemberIterator MemberEnd() { CEREAL_RAPIDJSON_ASSERT(IsObject()); return MemberIterator(GetMembersPointer() + data_.o.size); } + //! Request the object to have enough capacity to store members. + /*! \param newCapacity The capacity that the object at least need to have. + \param allocator Allocator for reallocating memory. It must be the same one as used before. Commonly use GenericDocument::GetAllocator(). + \return The value itself for fluent API. + \note Linear time complexity. + */ + GenericValue& MemberReserve(SizeType newCapacity, Allocator &allocator) { + CEREAL_RAPIDJSON_ASSERT(IsObject()); + if (newCapacity > data_.o.capacity) { + SetMembersPointer(reinterpret_cast(allocator.Realloc(GetMembersPointer(), data_.o.capacity * sizeof(Member), newCapacity * sizeof(Member)))); + data_.o.capacity = newCapacity; + } + return *this; + } + //! Check whether a member exists in the object. /*! \param name Member name to be searched. @@ -1188,17 +1289,8 @@ public: CEREAL_RAPIDJSON_ASSERT(name.IsString()); ObjectData& o = data_.o; - if (o.size >= o.capacity) { - if (o.capacity == 0) { - o.capacity = kDefaultObjectCapacity; - SetMembersPointer(reinterpret_cast(allocator.Malloc(o.capacity * sizeof(Member)))); - } - else { - SizeType oldCapacity = o.capacity; - o.capacity += (oldCapacity + 1) / 2; // grow by factor 1.5 - SetMembersPointer(reinterpret_cast(allocator.Realloc(GetMembersPointer(), oldCapacity * sizeof(Member), o.capacity * sizeof(Member)))); - } - } + if (o.size >= o.capacity) + MemberReserve(o.capacity == 0 ? kDefaultObjectCapacity : (o.capacity + (o.capacity + 1) / 2), allocator); Member* members = GetMembersPointer(); members[o.size].name.RawAssign(name); members[o.size].value.RawAssign(value); @@ -1335,7 +1427,7 @@ public: \note Linear time complexity. */ void RemoveAllMembers() { - CEREAL_RAPIDJSON_ASSERT(IsObject()); + CEREAL_RAPIDJSON_ASSERT(IsObject()); for (MemberIterator m = MemberBegin(); m != MemberEnd(); ++m) m->~Member(); data_.o.size = 0; @@ -1425,7 +1517,7 @@ public: MemberIterator pos = MemberBegin() + (first - MemberBegin()); for (MemberIterator itr = pos; itr != last; ++itr) itr->~Member(); - std::memmove(&*pos, &*last, static_cast(MemberEnd() - last) * sizeof(Member)); + std::memmove(static_cast(&*pos), &*last, static_cast(MemberEnd() - last) * sizeof(Member)); data_.o.size -= static_cast(last - first); return pos; } @@ -1481,7 +1573,7 @@ public: \note Linear time complexity. */ void Clear() { - CEREAL_RAPIDJSON_ASSERT(IsArray()); + CEREAL_RAPIDJSON_ASSERT(IsArray()); GenericValue* e = GetElementsPointer(); for (GenericValue* v = e; v != e + data_.a.size; ++v) v->~GenericValue(); @@ -1628,8 +1720,8 @@ public: CEREAL_RAPIDJSON_ASSERT(last <= End()); ValueIterator pos = Begin() + (first - Begin()); for (ValueIterator itr = pos; itr != last; ++itr) - itr->~GenericValue(); - std::memmove(pos, last, static_cast(End() - last) * sizeof(GenericValue)); + itr->~GenericValue(); + std::memmove(static_cast(pos), last, static_cast(End() - last) * sizeof(GenericValue)); data_.a.size -= static_cast(last - first); return pos; } @@ -1671,7 +1763,7 @@ public: GenericValue& SetInt64(int64_t i64) { this->~GenericValue(); new (this) GenericValue(i64); return *this; } GenericValue& SetUint64(uint64_t u64) { this->~GenericValue(); new (this) GenericValue(u64); return *this; } GenericValue& SetDouble(double d) { this->~GenericValue(); new (this) GenericValue(d); return *this; } - GenericValue& SetFloat(float f) { this->~GenericValue(); new (this) GenericValue(f); return *this; } + GenericValue& SetFloat(float f) { this->~GenericValue(); new (this) GenericValue(static_cast(f)); return *this; } //@} @@ -1687,7 +1779,7 @@ public: //! Set this value as a string without copying source string. /*! This version has better performance with supplied length, and also support string containing null character. - \param s source string pointer. + \param s source string pointer. \param length The length of source string, excluding the trailing null terminator. \return The value itself for fluent API. \post IsString() == true && GetString() == s && GetStringLength() == length @@ -1704,21 +1796,29 @@ public: //! Set this value as a string by copying from source string. /*! This version has better performance with supplied length, and also support string containing null character. - \param s source string. + \param s source string. \param length The length of source string, excluding the trailing null terminator. \param allocator Allocator for allocating copied buffer. Commonly use GenericDocument::GetAllocator(). \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { this->~GenericValue(); SetStringRaw(StringRef(s, length), allocator); return *this; } + GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { return SetString(StringRef(s, length), allocator); } //! Set this value as a string by copying from source string. - /*! \param s source string. + /*! \param s source string. \param allocator Allocator for allocating copied buffer. Commonly use GenericDocument::GetAllocator(). \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(s, internal::StrLen(s), allocator); } + GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(StringRef(s), allocator); } + + //! Set this value as a string by copying from source string. + /*! \param s source string reference + \param allocator Allocator for allocating copied buffer. Commonly use GenericDocument::GetAllocator(). + \return The value itself for fluent API. + \post IsString() == true && GetString() != s.s && strcmp(GetString(),s) == 0 && GetStringLength() == length + */ + GenericValue& SetString(StringRefType s, Allocator& allocator) { this->~GenericValue(); SetStringRaw(s, allocator); return *this; } #if CEREAL_RAPIDJSON_HAS_STDSTRING //! Set this value as a string by copying from source string. @@ -1728,7 +1828,7 @@ public: \post IsString() == true && GetString() != s.data() && strcmp(GetString(),s.data() == 0 && GetStringLength() == s.size() \note Requires the definition of the preprocessor symbol \ref CEREAL_RAPIDJSON_HAS_STDSTRING. */ - GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(s.data(), SizeType(s.size()), allocator); } + GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(StringRef(s), allocator); } #endif //@} @@ -1790,10 +1890,10 @@ public: if (CEREAL_RAPIDJSON_UNLIKELY(!v->Accept(handler))) return false; return handler.EndArray(data_.a.size); - + case kStringType: return handler.String(GetString(), GetStringLength(), (data_.f.flags & kCopyFlag) != 0); - + default: CEREAL_RAPIDJSON_ASSERT(GetType() == kNumberType); if (IsDouble()) return handler.Double(data_.n.d); @@ -1936,7 +2036,7 @@ private: if (count) { GenericValue* e = static_cast(allocator.Malloc(count * sizeof(GenericValue))); SetElementsPointer(e); - std::memcpy(e, values, count * sizeof(GenericValue)); + std::memcpy(static_cast(e), values, count * sizeof(GenericValue)); } else SetElementsPointer(0); @@ -1949,7 +2049,7 @@ private: if (count) { Member* m = static_cast(allocator.Malloc(count * sizeof(Member))); SetMembersPointer(m); - std::memcpy(m, members, count * sizeof(Member)); + std::memcpy(static_cast(m), members, count * sizeof(Member)); } else SetMembersPointer(0); @@ -2010,7 +2110,7 @@ private: typedef GenericValue > Value; /////////////////////////////////////////////////////////////////////////////// -// GenericDocument +// GenericDocument //! A document for parsing JSON text as DOM. /*! @@ -2038,20 +2138,20 @@ public: GenericValue(type), allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() { if (!allocator_) - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); } //! Constructor - /*! Creates an empty document which type is Null. + /*! Creates an empty document which type is Null. \param allocator Optional allocator for allocating memory. \param stackCapacity Optional initial capacity of stack in bytes. \param stackAllocator Optional allocator for allocating memory for stack. */ - GenericDocument(Allocator* allocator = 0, size_t stackCapacity = kDefaultStackCapacity, StackAllocator* stackAllocator = 0) : + GenericDocument(Allocator* allocator = 0, size_t stackCapacity = kDefaultStackCapacity, StackAllocator* stackAllocator = 0) : allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() { if (!allocator_) - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); } #if CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS @@ -2112,6 +2212,10 @@ public: return *this; } + // Allow Swap with ValueType. + // Refer to Effective C++ 3rd Edition/Item 33: Avoid hiding inherited names. + using ValueType::Swap; + //! free-standing swap function helper /*! Helper function to enable support for common swap implementation pattern based on \c std::swap: @@ -2243,7 +2347,7 @@ public: template GenericDocument& Parse(const typename SourceEncoding::Ch* str, size_t length) { CEREAL_RAPIDJSON_ASSERT(!(parseFlags & kParseInsituFlag)); - MemoryStream ms(static_cast(str), length * sizeof(typename SourceEncoding::Ch)); + MemoryStream ms(reinterpret_cast(str), length * sizeof(typename SourceEncoding::Ch)); EncodedInputStream is(ms); ParseStream(is); return *this; @@ -2253,7 +2357,7 @@ public: GenericDocument& Parse(const Ch* str, size_t length) { return Parse(str, length); } - + GenericDocument& Parse(const Ch* str, size_t length) { return Parse(str, length); } @@ -2273,14 +2377,14 @@ public: GenericDocument& Parse(const std::basic_string& str) { return Parse(str); } -#endif // CEREAL_RAPIDJSON_HAS_STDSTRING +#endif // CEREAL_RAPIDJSON_HAS_STDSTRING //!@} //!@name Handling parse errors //!@{ - //! Whether a parse error has occured in the last parsing. + //! Whether a parse error has occurred in the last parsing. bool HasParseError() const { return parseResult_.IsError(); } //! Get the \ref ParseErrorCode of last parsing. @@ -2338,16 +2442,16 @@ public: bool Uint64(uint64_t i) { new (stack_.template Push()) ValueType(i); return true; } bool Double(double d) { new (stack_.template Push()) ValueType(d); return true; } - bool RawNumber(const Ch* str, SizeType length, bool copy) { - if (copy) + bool RawNumber(const Ch* str, SizeType length, bool copy) { + if (copy) new (stack_.template Push()) ValueType(str, length, GetAllocator()); else new (stack_.template Push()) ValueType(str, length); return true; } - bool String(const Ch* str, SizeType length, bool copy) { - if (copy) + bool String(const Ch* str, SizeType length, bool copy) { + if (copy) new (stack_.template Push()) ValueType(str, length, GetAllocator()); else new (stack_.template Push()) ValueType(str, length); @@ -2355,7 +2459,7 @@ public: } bool StartObject() { new (stack_.template Push()) ValueType(kObjectType); return true; } - + bool Key(const Ch* str, SizeType length, bool copy) { return String(str, length, copy); } bool EndObject(SizeType memberCount) { @@ -2365,7 +2469,7 @@ public: } bool StartArray() { new (stack_.template Push()) ValueType(kArrayType); return true; } - + bool EndArray(SizeType elementCount) { ValueType* elements = stack_.template Pop(elementCount); stack_.template Top()->SetArrayRaw(elements, elementCount, GetAllocator()); @@ -2401,35 +2505,6 @@ private: //! GenericDocument with UTF8 encoding typedef GenericDocument > Document; -// defined here due to the dependency on GenericDocument -template -template -inline -GenericValue::GenericValue(const GenericValue& rhs, Allocator& allocator) -{ - switch (rhs.GetType()) { - case kObjectType: - case kArrayType: { // perform deep copy via SAX Handler - GenericDocument d(&allocator); - rhs.Accept(d); - RawAssign(*d.stack_.template Pop(1)); - } - break; - case kStringType: - if (rhs.data_.f.flags == kConstStringFlag) { - data_.f.flags = rhs.data_.f.flags; - data_ = *reinterpret_cast(&rhs.data_); - } else { - SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); - } - break; - default: - data_.f.flags = rhs.data_.f.flags; - data_ = *reinterpret_cast(&rhs.data_); - break; - } -} - //! Helper class for accessing Value of array type. /*! Instance of this helper class is obtained by \c GenericValue::GetArray(). @@ -2510,6 +2585,7 @@ public: ~GenericObject() {} SizeType MemberCount() const { return value_.MemberCount(); } + SizeType MemberCapacity() const { return value_.MemberCapacity(); } bool ObjectEmpty() const { return value_.ObjectEmpty(); } template ValueType& operator[](T* name) const { return value_[name]; } template ValueType& operator[](const GenericValue& name) const { return value_[name]; } @@ -2518,6 +2594,7 @@ public: #endif MemberIterator MemberBegin() const { return value_.MemberBegin(); } MemberIterator MemberEnd() const { return value_.MemberEnd(); } + GenericObject MemberReserve(SizeType newCapacity, AllocatorType &allocator) const { value_.MemberReserve(newCapacity, allocator); return *this; } bool HasMember(const Ch* name) const { return value_.HasMember(name); } #if CEREAL_RAPIDJSON_HAS_STDSTRING bool HasMember(const std::basic_string& name) const { return value_.HasMember(name); } @@ -2543,7 +2620,7 @@ public: GenericObject AddMember(StringRefType name, ValueType& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } GenericObject AddMember(StringRefType name, StringRefType value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } template CEREAL_RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (GenericObject)) AddMember(StringRefType name, T value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } - void RemoveAllMembers() { return value_.RemoveAllMembers(); } + void RemoveAllMembers() { value_.RemoveAllMembers(); } bool RemoveMember(const Ch* name) const { return value_.RemoveMember(name); } #if CEREAL_RAPIDJSON_HAS_STDSTRING bool RemoveMember(const std::basic_string& name) const { return value_.RemoveMember(name); } diff --git a/noarch/include/cereal/external/rapidjson/encodedstream.h b/noarch/include/cereal/external/rapidjson/encodedstream.h index 96713948..b04774be 100644 --- a/noarch/include/cereal/external/rapidjson/encodedstream.h +++ b/noarch/include/cereal/external/rapidjson/encodedstream.h @@ -200,7 +200,7 @@ private: // xx xx xx xx UTF-8 if (!hasBOM_) { - unsigned pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); + int pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); switch (pattern) { case 0x08: type_ = kUTF32BE; break; case 0x0A: type_ = kUTF16BE; break; diff --git a/noarch/include/cereal/external/rapidjson/encodings.h b/noarch/include/cereal/external/rapidjson/encodings.h index ed00b976..a1c27ecf 100644 --- a/noarch/include/cereal/external/rapidjson/encodings.h +++ b/noarch/include/cereal/external/rapidjson/encodings.h @@ -17,7 +17,7 @@ #include "rapidjson.h" -#ifdef _MSC_VER +#if defined(_MSC_VER) && !defined(__clang__) CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(4244) // conversion from 'type1' to 'type2', possible loss of data CEREAL_RAPIDJSON_DIAG_OFF(4702) // unreachable code @@ -144,9 +144,9 @@ struct UTF8 { template static bool Decode(InputStream& is, unsigned* codepoint) { -#define COPY() c = is.Take(); *codepoint = (*codepoint << 6) | (static_cast(c) & 0x3Fu) -#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) -#define TAIL() COPY(); TRANS(0x70) +#define CEREAL_RAPIDJSON_COPY() c = is.Take(); *codepoint = (*codepoint << 6) | (static_cast(c) & 0x3Fu) +#define CEREAL_RAPIDJSON_TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) +#define CEREAL_RAPIDJSON_TAIL() CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x70) typename InputStream::Ch c = is.Take(); if (!(c & 0x80)) { *codepoint = static_cast(c); @@ -157,48 +157,48 @@ struct UTF8 { if (type >= 32) { *codepoint = 0; } else { - *codepoint = (0xFF >> type) & static_cast(c); + *codepoint = (0xFFu >> type) & static_cast(c); } bool result = true; switch (type) { - case 2: TAIL(); return result; - case 3: TAIL(); TAIL(); return result; - case 4: COPY(); TRANS(0x50); TAIL(); return result; - case 5: COPY(); TRANS(0x10); TAIL(); TAIL(); return result; - case 6: TAIL(); TAIL(); TAIL(); return result; - case 10: COPY(); TRANS(0x20); TAIL(); return result; - case 11: COPY(); TRANS(0x60); TAIL(); TAIL(); return result; + case 2: CEREAL_RAPIDJSON_TAIL(); return result; + case 3: CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 4: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x50); CEREAL_RAPIDJSON_TAIL(); return result; + case 5: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x10); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 6: CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 10: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x20); CEREAL_RAPIDJSON_TAIL(); return result; + case 11: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x60); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; default: return false; } -#undef COPY -#undef TRANS -#undef TAIL +#undef CEREAL_RAPIDJSON_COPY +#undef CEREAL_RAPIDJSON_TRANS +#undef CEREAL_RAPIDJSON_TAIL } template static bool Validate(InputStream& is, OutputStream& os) { -#define COPY() os.Put(c = is.Take()) -#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) -#define TAIL() COPY(); TRANS(0x70) +#define CEREAL_RAPIDJSON_COPY() os.Put(c = is.Take()) +#define CEREAL_RAPIDJSON_TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) +#define CEREAL_RAPIDJSON_TAIL() CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x70) Ch c; - COPY(); + CEREAL_RAPIDJSON_COPY(); if (!(c & 0x80)) return true; bool result = true; switch (GetRange(static_cast(c))) { - case 2: TAIL(); return result; - case 3: TAIL(); TAIL(); return result; - case 4: COPY(); TRANS(0x50); TAIL(); return result; - case 5: COPY(); TRANS(0x10); TAIL(); TAIL(); return result; - case 6: TAIL(); TAIL(); TAIL(); return result; - case 10: COPY(); TRANS(0x20); TAIL(); return result; - case 11: COPY(); TRANS(0x60); TAIL(); TAIL(); return result; + case 2: CEREAL_RAPIDJSON_TAIL(); return result; + case 3: CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 4: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x50); CEREAL_RAPIDJSON_TAIL(); return result; + case 5: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x10); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 6: CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; + case 10: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x20); CEREAL_RAPIDJSON_TAIL(); return result; + case 11: CEREAL_RAPIDJSON_COPY(); CEREAL_RAPIDJSON_TRANS(0x60); CEREAL_RAPIDJSON_TAIL(); CEREAL_RAPIDJSON_TAIL(); return result; default: return false; } -#undef COPY -#undef TRANS -#undef TAIL +#undef CEREAL_RAPIDJSON_COPY +#undef CEREAL_RAPIDJSON_TRANS +#undef CEREAL_RAPIDJSON_TAIL } static unsigned char GetRange(unsigned char c) { @@ -283,7 +283,7 @@ struct UTF16 { CEREAL_RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); unsigned v = codepoint - 0x10000; os.Put(static_cast((v >> 10) | 0xD800)); - os.Put((v & 0x3FF) | 0xDC00); + os.Put(static_cast((v & 0x3FF) | 0xDC00)); } } @@ -299,7 +299,7 @@ struct UTF16 { CEREAL_RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); unsigned v = codepoint - 0x10000; PutUnsafe(os, static_cast((v >> 10) | 0xD800)); - PutUnsafe(os, (v & 0x3FF) | 0xDC00); + PutUnsafe(os, static_cast((v & 0x3FF) | 0xDC00)); } } @@ -384,7 +384,7 @@ struct UTF16BE : UTF16 { static CharType Take(InputByteStream& is) { CEREAL_RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); unsigned c = static_cast(static_cast(is.Take())) << 8; - c |= static_cast(is.Take()); + c |= static_cast(static_cast(is.Take())); return static_cast(c); } @@ -620,28 +620,28 @@ struct AutoUTF { #define CEREAL_RAPIDJSON_ENCODINGS_FUNC(x) UTF8::x, UTF16LE::x, UTF16BE::x, UTF32LE::x, UTF32BE::x template - CEREAL_RAPIDJSON_FORCEINLINE static void Encode(OutputStream& os, unsigned codepoint) { + static CEREAL_RAPIDJSON_FORCEINLINE void Encode(OutputStream& os, unsigned codepoint) { typedef void (*EncodeFunc)(OutputStream&, unsigned); static const EncodeFunc f[] = { CEREAL_RAPIDJSON_ENCODINGS_FUNC(Encode) }; (*f[os.GetType()])(os, codepoint); } template - CEREAL_RAPIDJSON_FORCEINLINE static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + static CEREAL_RAPIDJSON_FORCEINLINE void EncodeUnsafe(OutputStream& os, unsigned codepoint) { typedef void (*EncodeFunc)(OutputStream&, unsigned); static const EncodeFunc f[] = { CEREAL_RAPIDJSON_ENCODINGS_FUNC(EncodeUnsafe) }; (*f[os.GetType()])(os, codepoint); } template - CEREAL_RAPIDJSON_FORCEINLINE static bool Decode(InputStream& is, unsigned* codepoint) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Decode(InputStream& is, unsigned* codepoint) { typedef bool (*DecodeFunc)(InputStream&, unsigned*); static const DecodeFunc f[] = { CEREAL_RAPIDJSON_ENCODINGS_FUNC(Decode) }; return (*f[is.GetType()])(is, codepoint); } template - CEREAL_RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { typedef bool (*ValidateFunc)(InputStream&, OutputStream&); static const ValidateFunc f[] = { CEREAL_RAPIDJSON_ENCODINGS_FUNC(Validate) }; return (*f[is.GetType()])(is, os); @@ -658,7 +658,7 @@ template struct Transcoder { //! Take one Unicode codepoint from source encoding, convert it to target encoding and put it to the output stream. template - CEREAL_RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { unsigned codepoint; if (!SourceEncoding::Decode(is, &codepoint)) return false; @@ -667,7 +667,7 @@ struct Transcoder { } template - CEREAL_RAPIDJSON_FORCEINLINE static bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { unsigned codepoint; if (!SourceEncoding::Decode(is, &codepoint)) return false; @@ -677,7 +677,7 @@ struct Transcoder { //! Validate one Unicode codepoint from an encoded stream. template - CEREAL_RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Transcode(is, os); // Since source/target encoding is different, must transcode. } }; @@ -690,26 +690,26 @@ inline void PutUnsafe(Stream& stream, typename Stream::Ch c); template struct Transcoder { template - CEREAL_RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { os.Put(is.Take()); // Just copy one code unit. This semantic is different from primary template class. return true; } template - CEREAL_RAPIDJSON_FORCEINLINE static bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { PutUnsafe(os, is.Take()); // Just copy one code unit. This semantic is different from primary template class. return true; } template - CEREAL_RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static CEREAL_RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Encoding::Validate(is, os); // source/target encoding are the same } }; CEREAL_RAPIDJSON_NAMESPACE_END -#if defined(__GNUC__) || defined(_MSC_VER) +#if defined(__GNUC__) || (defined(_MSC_VER) && !defined(__clang__)) CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/error/error.h b/noarch/include/cereal/external/rapidjson/error/error.h index c2638f14..7e7cc243 100644 --- a/noarch/include/cereal/external/rapidjson/error/error.h +++ b/noarch/include/cereal/external/rapidjson/error/error.h @@ -104,6 +104,8 @@ enum ParseErrorCode { \see GenericReader::Parse, GenericDocument::Parse */ struct ParseResult { + //!! Unspecified boolean type + typedef bool (ParseResult::*BooleanType)() const; public: //! Default constructor, no error. ParseResult() : code_(kParseErrorNone), offset_(0) {} @@ -115,8 +117,8 @@ public: //! Get the error offset, if \ref IsError(), 0 otherwise. size_t Offset() const { return offset_; } - //! Conversion to \c bool, returns \c true, iff !\ref IsError(). - operator bool() const { return !IsError(); } + //! Explicit conversion to \c bool, returns \c true, iff !\ref IsError(). + operator BooleanType() const { return !IsError() ? &ParseResult::IsError : NULL; } //! Whether the result is an error. bool IsError() const { return code_ != kParseErrorNone; } @@ -124,6 +126,10 @@ public: bool operator==(ParseErrorCode code) const { return code_ == code; } friend bool operator==(ParseErrorCode code, const ParseResult & err) { return code == err.code_; } + bool operator!=(const ParseResult& that) const { return !(*this == that); } + bool operator!=(ParseErrorCode code) const { return !(*this == code); } + friend bool operator!=(ParseErrorCode code, const ParseResult & err) { return err != code; } + //! Reset error code. void Clear() { Set(kParseErrorNone); } //! Update error code and offset. diff --git a/noarch/include/cereal/external/rapidjson/filereadstream.h b/noarch/include/cereal/external/rapidjson/filereadstream.h index abafb549..253a7f9a 100644 --- a/noarch/include/cereal/external/rapidjson/filereadstream.h +++ b/noarch/include/cereal/external/rapidjson/filereadstream.h @@ -59,7 +59,7 @@ public: // For encoding detection only. const Ch* Peek4() const { - return (current_ + 4 <= bufferLast_) ? current_ : 0; + return (current_ + 4 - !eof_ <= bufferLast_) ? current_ : 0; } private: @@ -68,7 +68,7 @@ private: ++current_; else if (!eof_) { count_ += readCount_; - readCount_ = fread(buffer_, 1, bufferSize_, fp_); + readCount_ = std::fread(buffer_, 1, bufferSize_, fp_); bufferLast_ = buffer_ + readCount_ - 1; current_ = buffer_; diff --git a/noarch/include/cereal/external/rapidjson/filewritestream.h b/noarch/include/cereal/external/rapidjson/filewritestream.h index 9a600994..44ae1279 100644 --- a/noarch/include/cereal/external/rapidjson/filewritestream.h +++ b/noarch/include/cereal/external/rapidjson/filewritestream.h @@ -25,7 +25,7 @@ CEREAL_RAPIDJSON_DIAG_OFF(unreachable-code) CEREAL_RAPIDJSON_NAMESPACE_BEGIN -//! Wrapper of C file stream for input using fread(). +//! Wrapper of C file stream for output using fwrite(). /*! \note implements Stream concept */ @@ -62,7 +62,7 @@ public: void Flush() { if (current_ != buffer_) { - size_t result = fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); + size_t result = std::fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); if (result < static_cast(current_ - buffer_)) { // failure deliberately ignored at this time // added to avoid warn_unused_result build errors diff --git a/noarch/include/cereal/external/rapidjson/internal/biginteger.h b/noarch/include/cereal/external/rapidjson/internal/biginteger.h index f8a5a50f..77a6519d 100644 --- a/noarch/include/cereal/external/rapidjson/internal/biginteger.h +++ b/noarch/include/cereal/external/rapidjson/internal/biginteger.h @@ -17,7 +17,7 @@ #include "../rapidjson.h" -#if defined(_MSC_VER) && defined(_M_AMD64) +#if defined(_MSC_VER) && !__INTEL_COMPILER && defined(_M_AMD64) #include // for _umul128 #pragma intrinsic(_umul128) #endif @@ -133,7 +133,7 @@ public: CEREAL_RAPIDJSON_ASSERT(count_ + offset <= kCapacity); if (interShift == 0) { - std::memmove(&digits_[count_ - 1 + offset], &digits_[count_ - 1], count_ * sizeof(Type)); + std::memmove(digits_ + offset, digits_, count_ * sizeof(Type)); count_ += offset; } else { diff --git a/noarch/include/cereal/external/rapidjson/internal/diyfp.h b/noarch/include/cereal/external/rapidjson/internal/diyfp.h index ab7d66bc..3f92a078 100644 --- a/noarch/include/cereal/external/rapidjson/internal/diyfp.h +++ b/noarch/include/cereal/external/rapidjson/internal/diyfp.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. // This is a C++ header-only implementation of Grisu2 algorithm from the publication: @@ -20,8 +20,9 @@ #define CEREAL_RAPIDJSON_DIYFP_H_ #include "../rapidjson.h" +#include -#if defined(_MSC_VER) && defined(_M_AMD64) +#if defined(_MSC_VER) && defined(_M_AMD64) && !defined(__INTEL_COMPILER) #include #pragma intrinsic(_BitScanReverse64) #pragma intrinsic(_umul128) @@ -56,7 +57,7 @@ struct DiyFp { if (biased_e != 0) { f = significand + kDpHiddenBit; e = biased_e - kDpExponentBias; - } + } else { f = significand; e = kDpMinExponent + 1; @@ -99,6 +100,7 @@ struct DiyFp { } DiyFp Normalize() const { + CEREAL_RAPIDJSON_ASSERT(f != 0); // https://stackoverflow.com/a/26809183/291737 #if defined(_MSC_VER) && defined(_M_AMD64) unsigned long index; _BitScanReverse64(&index, f); @@ -141,7 +143,16 @@ struct DiyFp { double d; uint64_t u64; }u; - const uint64_t be = (e == kDpDenormalExponent && (f & kDpHiddenBit) == 0) ? 0 : + CEREAL_RAPIDJSON_ASSERT(f <= kDpHiddenBit + kDpSignificandMask); + if (e < kDpDenormalExponent) { + // Underflow. + return 0.0; + } + if (e >= kDpMaxExponent) { + // Overflow. + return std::numeric_limits::infinity(); + } + const uint64_t be = (e == kDpDenormalExponent && (f & kDpHiddenBit) == 0) ? 0 : static_cast(e + kDpExponentBias); u.u64 = (f & kDpSignificandMask) | (be << kDpSignificandSize); return u.d; @@ -220,9 +231,10 @@ inline DiyFp GetCachedPowerByIndex(size_t index) { 641, 667, 694, 720, 747, 774, 800, 827, 853, 880, 907, 933, 960, 986, 1013, 1039, 1066 }; + CEREAL_RAPIDJSON_ASSERT(index < 87); return DiyFp(kCachedPowers_F[index], kCachedPowers_E[index]); } - + inline DiyFp GetCachedPower(int e, int* K) { //int k = static_cast(ceil((-61 - e) * 0.30102999566398114)) + 374; @@ -238,10 +250,11 @@ inline DiyFp GetCachedPower(int e, int* K) { } inline DiyFp GetCachedPower10(int exp, int *outExp) { - unsigned index = (static_cast(exp) + 348u) / 8u; - *outExp = -348 + static_cast(index) * 8; - return GetCachedPowerByIndex(index); - } + CEREAL_RAPIDJSON_ASSERT(exp >= -348); + unsigned index = static_cast(exp + 348) / 8u; + *outExp = -348 + static_cast(index) * 8; + return GetCachedPowerByIndex(index); +} #ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_POP diff --git a/noarch/include/cereal/external/rapidjson/internal/dtoa.h b/noarch/include/cereal/external/rapidjson/internal/dtoa.h index c59c92e2..ea62b349 100644 --- a/noarch/include/cereal/external/rapidjson/internal/dtoa.h +++ b/noarch/include/cereal/external/rapidjson/internal/dtoa.h @@ -41,7 +41,7 @@ inline void GrisuRound(char* buffer, int len, uint64_t delta, uint64_t rest, uin } } -inline unsigned CountDecimalDigit32(uint32_t n) { +inline int CountDecimalDigit32(uint32_t n) { // Simple pure C++ implementation was faster than __builtin_clz version in this situation. if (n < 10) return 1; if (n < 100) return 2; @@ -63,7 +63,7 @@ inline void DigitGen(const DiyFp& W, const DiyFp& Mp, uint64_t delta, char* buff const DiyFp wp_w = Mp - W; uint32_t p1 = static_cast(Mp.f >> -one.e); uint64_t p2 = Mp.f & (one.f - 1); - unsigned kappa = CountDecimalDigit32(p1); // kappa in [0, 9] + int kappa = CountDecimalDigit32(p1); // kappa in [0, 9] *len = 0; while (kappa > 0) { @@ -102,8 +102,8 @@ inline void DigitGen(const DiyFp& W, const DiyFp& Mp, uint64_t delta, char* buff kappa--; if (p2 < delta) { *K += kappa; - int index = -static_cast(kappa); - GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * (index < 9 ? kPow10[-static_cast(kappa)] : 0)); + int index = -kappa; + GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * (index < 9 ? kPow10[index] : 0)); return; } } diff --git a/noarch/include/cereal/external/rapidjson/internal/ieee754.h b/noarch/include/cereal/external/rapidjson/internal/ieee754.h index 35845a15..6730dfdc 100644 --- a/noarch/include/cereal/external/rapidjson/internal/ieee754.h +++ b/noarch/include/cereal/external/rapidjson/internal/ieee754.h @@ -48,13 +48,13 @@ public: int IntegerExponent() const { return (IsNormal() ? Exponent() : kDenormalExponent) - kSignificandSize; } uint64_t ToBias() const { return (u_ & kSignMask) ? ~u_ + 1 : u_ | kSignMask; } - static unsigned EffectiveSignificandSize(int order) { + static int EffectiveSignificandSize(int order) { if (order >= -1021) return 53; else if (order <= -1074) return 0; else - return static_cast(order) + 1074; + return order + 1074; } private: diff --git a/noarch/include/cereal/external/rapidjson/internal/itoa.h b/noarch/include/cereal/external/rapidjson/internal/itoa.h index 3e83658f..f04e3fa8 100644 --- a/noarch/include/cereal/external/rapidjson/internal/itoa.h +++ b/noarch/include/cereal/external/rapidjson/internal/itoa.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef CEREAL_RAPIDJSON_ITOA_ @@ -37,12 +37,14 @@ inline const char* GetDigitsLut() { } inline char* u32toa(uint32_t value, char* buffer) { + CEREAL_RAPIDJSON_ASSERT(buffer != 0); + const char* cDigitsLut = GetDigitsLut(); if (value < 10000) { const uint32_t d1 = (value / 100) << 1; const uint32_t d2 = (value % 100) << 1; - + if (value >= 1000) *buffer++ = cDigitsLut[d1]; if (value >= 100) @@ -55,13 +57,13 @@ inline char* u32toa(uint32_t value, char* buffer) { // value = bbbbcccc const uint32_t b = value / 10000; const uint32_t c = value % 10000; - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + if (value >= 10000000) *buffer++ = cDigitsLut[d1]; if (value >= 1000000) @@ -69,7 +71,7 @@ inline char* u32toa(uint32_t value, char* buffer) { if (value >= 100000) *buffer++ = cDigitsLut[d2]; *buffer++ = cDigitsLut[d2 + 1]; - + *buffer++ = cDigitsLut[d3]; *buffer++ = cDigitsLut[d3 + 1]; *buffer++ = cDigitsLut[d4]; @@ -77,10 +79,10 @@ inline char* u32toa(uint32_t value, char* buffer) { } else { // value = aabbbbcccc in decimal - + const uint32_t a = value / 100000000; // 1 to 42 value %= 100000000; - + if (a >= 10) { const unsigned i = a << 1; *buffer++ = cDigitsLut[i]; @@ -91,13 +93,13 @@ inline char* u32toa(uint32_t value, char* buffer) { const uint32_t b = value / 10000; // 0 to 9999 const uint32_t c = value % 10000; // 0 to 9999 - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + *buffer++ = cDigitsLut[d1]; *buffer++ = cDigitsLut[d1 + 1]; *buffer++ = cDigitsLut[d2]; @@ -111,6 +113,7 @@ inline char* u32toa(uint32_t value, char* buffer) { } inline char* i32toa(int32_t value, char* buffer) { + CEREAL_RAPIDJSON_ASSERT(buffer != 0); uint32_t u = static_cast(value); if (value < 0) { *buffer++ = '-'; @@ -121,6 +124,7 @@ inline char* i32toa(int32_t value, char* buffer) { } inline char* u64toa(uint64_t value, char* buffer) { + CEREAL_RAPIDJSON_ASSERT(buffer != 0); const char* cDigitsLut = GetDigitsLut(); const uint64_t kTen8 = 100000000; const uint64_t kTen9 = kTen8 * 10; @@ -131,13 +135,13 @@ inline char* u64toa(uint64_t value, char* buffer) { const uint64_t kTen14 = kTen8 * 1000000; const uint64_t kTen15 = kTen8 * 10000000; const uint64_t kTen16 = kTen8 * kTen8; - + if (value < kTen8) { uint32_t v = static_cast(value); if (v < 10000) { const uint32_t d1 = (v / 100) << 1; const uint32_t d2 = (v % 100) << 1; - + if (v >= 1000) *buffer++ = cDigitsLut[d1]; if (v >= 100) @@ -150,13 +154,13 @@ inline char* u64toa(uint64_t value, char* buffer) { // value = bbbbcccc const uint32_t b = v / 10000; const uint32_t c = v % 10000; - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + if (value >= 10000000) *buffer++ = cDigitsLut[d1]; if (value >= 1000000) @@ -164,7 +168,7 @@ inline char* u64toa(uint64_t value, char* buffer) { if (value >= 100000) *buffer++ = cDigitsLut[d2]; *buffer++ = cDigitsLut[d2 + 1]; - + *buffer++ = cDigitsLut[d3]; *buffer++ = cDigitsLut[d3 + 1]; *buffer++ = cDigitsLut[d4]; @@ -174,22 +178,22 @@ inline char* u64toa(uint64_t value, char* buffer) { else if (value < kTen16) { const uint32_t v0 = static_cast(value / kTen8); const uint32_t v1 = static_cast(value % kTen8); - + const uint32_t b0 = v0 / 10000; const uint32_t c0 = v0 % 10000; - + const uint32_t d1 = (b0 / 100) << 1; const uint32_t d2 = (b0 % 100) << 1; - + const uint32_t d3 = (c0 / 100) << 1; const uint32_t d4 = (c0 % 100) << 1; const uint32_t b1 = v1 / 10000; const uint32_t c1 = v1 % 10000; - + const uint32_t d5 = (b1 / 100) << 1; const uint32_t d6 = (b1 % 100) << 1; - + const uint32_t d7 = (c1 / 100) << 1; const uint32_t d8 = (c1 % 100) << 1; @@ -207,9 +211,8 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[d3 + 1]; if (value >= kTen9) *buffer++ = cDigitsLut[d4]; - if (value >= kTen8) - *buffer++ = cDigitsLut[d4 + 1]; - + + *buffer++ = cDigitsLut[d4 + 1]; *buffer++ = cDigitsLut[d5]; *buffer++ = cDigitsLut[d5 + 1]; *buffer++ = cDigitsLut[d6]; @@ -222,7 +225,7 @@ inline char* u64toa(uint64_t value, char* buffer) { else { const uint32_t a = static_cast(value / kTen16); // 1 to 1844 value %= kTen16; - + if (a < 10) *buffer++ = static_cast('0' + static_cast(a)); else if (a < 100) { @@ -232,7 +235,7 @@ inline char* u64toa(uint64_t value, char* buffer) { } else if (a < 1000) { *buffer++ = static_cast('0' + static_cast(a / 100)); - + const uint32_t i = (a % 100) << 1; *buffer++ = cDigitsLut[i]; *buffer++ = cDigitsLut[i + 1]; @@ -245,28 +248,28 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[j]; *buffer++ = cDigitsLut[j + 1]; } - + const uint32_t v0 = static_cast(value / kTen8); const uint32_t v1 = static_cast(value % kTen8); - + const uint32_t b0 = v0 / 10000; const uint32_t c0 = v0 % 10000; - + const uint32_t d1 = (b0 / 100) << 1; const uint32_t d2 = (b0 % 100) << 1; - + const uint32_t d3 = (c0 / 100) << 1; const uint32_t d4 = (c0 % 100) << 1; - + const uint32_t b1 = v1 / 10000; const uint32_t c1 = v1 % 10000; - + const uint32_t d5 = (b1 / 100) << 1; const uint32_t d6 = (b1 % 100) << 1; - + const uint32_t d7 = (c1 / 100) << 1; const uint32_t d8 = (c1 % 100) << 1; - + *buffer++ = cDigitsLut[d1]; *buffer++ = cDigitsLut[d1 + 1]; *buffer++ = cDigitsLut[d2]; @@ -284,11 +287,12 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[d8]; *buffer++ = cDigitsLut[d8 + 1]; } - + return buffer; } inline char* i64toa(int64_t value, char* buffer) { + CEREAL_RAPIDJSON_ASSERT(buffer != 0); uint64_t u = static_cast(value); if (value < 0) { *buffer++ = '-'; diff --git a/noarch/include/cereal/external/rapidjson/internal/meta.h b/noarch/include/cereal/external/rapidjson/internal/meta.h index c98db906..f511ff8f 100644 --- a/noarch/include/cereal/external/rapidjson/internal/meta.h +++ b/noarch/include/cereal/external/rapidjson/internal/meta.h @@ -21,7 +21,8 @@ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(effc++) #endif -#if defined(_MSC_VER) + +#if defined(_MSC_VER) && !defined(__clang__) CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(6334) #endif @@ -174,7 +175,11 @@ template struct RemoveSfinaeTag { typedef T Type; CEREAL_RAPIDJSON_NAMESPACE_END //@endcond -#if defined(__GNUC__) || defined(_MSC_VER) +#if defined(_MSC_VER) && !defined(__clang__) +CEREAL_RAPIDJSON_DIAG_POP +#endif + +#ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/internal/regex.h b/noarch/include/cereal/external/rapidjson/internal/regex.h index 3bfa3ec3..6c2ed6ef 100644 --- a/noarch/include/cereal/external/rapidjson/internal/regex.h +++ b/noarch/include/cereal/external/rapidjson/internal/regex.h @@ -24,16 +24,17 @@ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(padded) CEREAL_RAPIDJSON_DIAG_OFF(switch-enum) CEREAL_RAPIDJSON_DIAG_OFF(implicit-fallthrough) +#elif defined(_MSC_VER) +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif #ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(effc++) +#if __GNUC__ >= 7 +CEREAL_RAPIDJSON_DIAG_OFF(implicit-fallthrough) #endif - -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_PUSH -CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif #ifndef CEREAL_RAPIDJSON_REGEX_VERBOSE @@ -43,12 +44,40 @@ CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated CEREAL_RAPIDJSON_NAMESPACE_BEGIN namespace internal { +/////////////////////////////////////////////////////////////////////////////// +// DecodedStream + +template +class DecodedStream { +public: + DecodedStream(SourceStream& ss) : ss_(ss), codepoint_() { Decode(); } + unsigned Peek() { return codepoint_; } + unsigned Take() { + unsigned c = codepoint_; + if (c) // No further decoding when '\0' + Decode(); + return c; + } + +private: + void Decode() { + if (!Encoding::Decode(ss_, &codepoint_)) + codepoint_ = 0; + } + + SourceStream& ss_; + unsigned codepoint_; +}; + /////////////////////////////////////////////////////////////////////////////// // GenericRegex static const SizeType kRegexInvalidState = ~SizeType(0); //!< Represents an invalid index in GenericRegex::State::out, out1 static const SizeType kRegexInvalidRange = ~SizeType(0); +template +class GenericRegexSearch; + //! Regular expression engine with subset of ECMAscript grammar. /*! Supported regular expression syntax: @@ -84,45 +113,29 @@ static const SizeType kRegexInvalidRange = ~SizeType(0); template class GenericRegex { public: + typedef Encoding EncodingType; typedef typename Encoding::Ch Ch; + template friend class GenericRegexSearch; GenericRegex(const Ch* source, Allocator* allocator = 0) : - states_(allocator, 256), ranges_(allocator, 256), root_(kRegexInvalidState), stateCount_(), rangeCount_(), - stateSet_(), state0_(allocator, 0), state1_(allocator, 0), anchorBegin_(), anchorEnd_() + ownAllocator_(allocator ? 0 : CEREAL_RAPIDJSON_NEW(Allocator)()), allocator_(allocator ? allocator : ownAllocator_), + states_(allocator_, 256), ranges_(allocator_, 256), root_(kRegexInvalidState), stateCount_(), rangeCount_(), + anchorBegin_(), anchorEnd_() { GenericStringStream ss(source); - DecodedStream > ds(ss); + DecodedStream, Encoding> ds(ss); Parse(ds); } - ~GenericRegex() { - Allocator::Free(stateSet_); + ~GenericRegex() + { + CEREAL_RAPIDJSON_DELETE(ownAllocator_); } bool IsValid() const { return root_ != kRegexInvalidState; } - template - bool Match(InputStream& is) const { - return SearchWithAnchoring(is, true, true); - } - - bool Match(const Ch* s) const { - GenericStringStream is(s); - return Match(is); - } - - template - bool Search(InputStream& is) const { - return SearchWithAnchoring(is, anchorBegin_, anchorEnd_); - } - - bool Search(const Ch* s) const { - GenericStringStream is(s); - return Search(is); - } - private: enum Operator { kZeroOrOne, @@ -157,28 +170,6 @@ private: SizeType minIndex; }; - template - class DecodedStream { - public: - DecodedStream(SourceStream& ss) : ss_(ss), codepoint_() { Decode(); } - unsigned Peek() { return codepoint_; } - unsigned Take() { - unsigned c = codepoint_; - if (c) // No further decoding when '\0' - Decode(); - return c; - } - - private: - void Decode() { - if (!Encoding::Decode(ss_, &codepoint_)) - codepoint_ = 0; - } - - SourceStream& ss_; - unsigned codepoint_; - }; - State& GetState(SizeType index) { CEREAL_RAPIDJSON_ASSERT(index < stateCount_); return states_.template Bottom()[index]; @@ -200,11 +191,10 @@ private: } template - void Parse(DecodedStream& ds) { - Allocator allocator; - Stack operandStack(&allocator, 256); // Frag - Stack operatorStack(&allocator, 256); // Operator - Stack atomCountStack(&allocator, 256); // unsigned (Atom per parenthesis) + void Parse(DecodedStream& ds) { + Stack operandStack(allocator_, 256); // Frag + Stack operatorStack(allocator_, 256); // Operator + Stack atomCountStack(allocator_, 256); // unsigned (Atom per parenthesis) *atomCountStack.template Push() = 0; @@ -327,14 +317,6 @@ private: printf("\n"); #endif } - - // Preallocate buffer for SearchWithAnchoring() - CEREAL_RAPIDJSON_ASSERT(stateSet_ == 0); - if (stateCount_ > 0) { - stateSet_ = static_cast(states_.GetAllocator().Malloc(GetStateSetSize())); - state0_.template Reserve(stateCount_); - state1_.template Reserve(stateCount_); - } } SizeType NewState(SizeType out, SizeType out1, unsigned codepoint) { @@ -413,8 +395,7 @@ private: } return false; - default: - CEREAL_RAPIDJSON_ASSERT(op == kOneOrMore); + case kOneOrMore: if (operandStack.GetSize() >= sizeof(Frag)) { Frag e = *operandStack.template Pop(1); SizeType s = NewState(kRegexInvalidState, e.start, 0); @@ -423,6 +404,10 @@ private: return true; } return false; + + default: + // syntax error (e.g. unclosed kLeftParenthesis) + return false; } } @@ -483,7 +468,7 @@ private: } template - bool ParseUnsigned(DecodedStream& ds, unsigned* u) { + bool ParseUnsigned(DecodedStream& ds, unsigned* u) { unsigned r = 0; if (ds.Peek() < '0' || ds.Peek() > '9') return false; @@ -497,7 +482,7 @@ private: } template - bool ParseRange(DecodedStream& ds, SizeType* range) { + bool ParseRange(DecodedStream& ds, SizeType* range) { bool isBegin = true; bool negate = false; int step = 0; @@ -575,7 +560,7 @@ private: } template - bool CharacterEscape(DecodedStream& ds, unsigned* escapedCodepoint) { + bool CharacterEscape(DecodedStream& ds, unsigned* escapedCodepoint) { unsigned codepoint; switch (codepoint = ds.Take()) { case '^': @@ -603,72 +588,8 @@ private: } } - template - bool SearchWithAnchoring(InputStream& is, bool anchorBegin, bool anchorEnd) const { - CEREAL_RAPIDJSON_ASSERT(IsValid()); - DecodedStream ds(is); - - state0_.Clear(); - Stack *current = &state0_, *next = &state1_; - const size_t stateSetSize = GetStateSetSize(); - std::memset(stateSet_, 0, stateSetSize); - - bool matched = AddState(*current, root_); - unsigned codepoint; - while (!current->Empty() && (codepoint = ds.Take()) != 0) { - std::memset(stateSet_, 0, stateSetSize); - next->Clear(); - matched = false; - for (const SizeType* s = current->template Bottom(); s != current->template End(); ++s) { - const State& sr = GetState(*s); - if (sr.codepoint == codepoint || - sr.codepoint == kAnyCharacterClass || - (sr.codepoint == kRangeCharacterClass && MatchRange(sr.rangeStart, codepoint))) - { - matched = AddState(*next, sr.out) || matched; - if (!anchorEnd && matched) - return true; - } - if (!anchorBegin) - AddState(*next, root_); - } - internal::Swap(current, next); - } - - return matched; - } - - size_t GetStateSetSize() const { - return (stateCount_ + 31) / 32 * 4; - } - - // Return whether the added states is a match state - bool AddState(Stack& l, SizeType index) const { - CEREAL_RAPIDJSON_ASSERT(index != kRegexInvalidState); - - const State& s = GetState(index); - if (s.out1 != kRegexInvalidState) { // Split - bool matched = AddState(l, s.out); - return AddState(l, s.out1) || matched; - } - else if (!(stateSet_[index >> 5] & (1 << (index & 31)))) { - stateSet_[index >> 5] |= (1 << (index & 31)); - *l.template PushUnsafe() = index; - } - return s.out == kRegexInvalidState; // by using PushUnsafe() above, we can ensure s is not validated due to reallocation. - } - - bool MatchRange(SizeType rangeIndex, unsigned codepoint) const { - bool yes = (GetRange(rangeIndex).start & kRangeNegationFlag) == 0; - while (rangeIndex != kRegexInvalidRange) { - const Range& r = GetRange(rangeIndex); - if (codepoint >= (r.start & ~kRangeNegationFlag) && codepoint <= r.end) - return yes; - rangeIndex = r.next; - } - return !yes; - } - + Allocator* ownAllocator_; + Allocator* allocator_; Stack states_; Stack ranges_; SizeType root_; @@ -678,23 +599,141 @@ private: static const unsigned kInfinityQuantifier = ~0u; // For SearchWithAnchoring() - uint32_t* stateSet_; // allocated by states_.GetAllocator() - mutable Stack state0_; - mutable Stack state1_; bool anchorBegin_; bool anchorEnd_; }; +template +class GenericRegexSearch { +public: + typedef typename RegexType::EncodingType Encoding; + typedef typename Encoding::Ch Ch; + + GenericRegexSearch(const RegexType& regex, Allocator* allocator = 0) : + regex_(regex), allocator_(allocator), ownAllocator_(0), + state0_(allocator, 0), state1_(allocator, 0), stateSet_() + { + CEREAL_RAPIDJSON_ASSERT(regex_.IsValid()); + if (!allocator_) + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); + stateSet_ = static_cast(allocator_->Malloc(GetStateSetSize())); + state0_.template Reserve(regex_.stateCount_); + state1_.template Reserve(regex_.stateCount_); + } + + ~GenericRegexSearch() { + Allocator::Free(stateSet_); + CEREAL_RAPIDJSON_DELETE(ownAllocator_); + } + + template + bool Match(InputStream& is) { + return SearchWithAnchoring(is, true, true); + } + + bool Match(const Ch* s) { + GenericStringStream is(s); + return Match(is); + } + + template + bool Search(InputStream& is) { + return SearchWithAnchoring(is, regex_.anchorBegin_, regex_.anchorEnd_); + } + + bool Search(const Ch* s) { + GenericStringStream is(s); + return Search(is); + } + +private: + typedef typename RegexType::State State; + typedef typename RegexType::Range Range; + + template + bool SearchWithAnchoring(InputStream& is, bool anchorBegin, bool anchorEnd) { + DecodedStream ds(is); + + state0_.Clear(); + Stack *current = &state0_, *next = &state1_; + const size_t stateSetSize = GetStateSetSize(); + std::memset(stateSet_, 0, stateSetSize); + + bool matched = AddState(*current, regex_.root_); + unsigned codepoint; + while (!current->Empty() && (codepoint = ds.Take()) != 0) { + std::memset(stateSet_, 0, stateSetSize); + next->Clear(); + matched = false; + for (const SizeType* s = current->template Bottom(); s != current->template End(); ++s) { + const State& sr = regex_.GetState(*s); + if (sr.codepoint == codepoint || + sr.codepoint == RegexType::kAnyCharacterClass || + (sr.codepoint == RegexType::kRangeCharacterClass && MatchRange(sr.rangeStart, codepoint))) + { + matched = AddState(*next, sr.out) || matched; + if (!anchorEnd && matched) + return true; + } + if (!anchorBegin) + AddState(*next, regex_.root_); + } + internal::Swap(current, next); + } + + return matched; + } + + size_t GetStateSetSize() const { + return (regex_.stateCount_ + 31) / 32 * 4; + } + + // Return whether the added states is a match state + bool AddState(Stack& l, SizeType index) { + CEREAL_RAPIDJSON_ASSERT(index != kRegexInvalidState); + + const State& s = regex_.GetState(index); + if (s.out1 != kRegexInvalidState) { // Split + bool matched = AddState(l, s.out); + return AddState(l, s.out1) || matched; + } + else if (!(stateSet_[index >> 5] & (1u << (index & 31)))) { + stateSet_[index >> 5] |= (1u << (index & 31)); + *l.template PushUnsafe() = index; + } + return s.out == kRegexInvalidState; // by using PushUnsafe() above, we can ensure s is not validated due to reallocation. + } + + bool MatchRange(SizeType rangeIndex, unsigned codepoint) const { + bool yes = (regex_.GetRange(rangeIndex).start & RegexType::kRangeNegationFlag) == 0; + while (rangeIndex != kRegexInvalidRange) { + const Range& r = regex_.GetRange(rangeIndex); + if (codepoint >= (r.start & ~RegexType::kRangeNegationFlag) && codepoint <= r.end) + return yes; + rangeIndex = r.next; + } + return !yes; + } + + const RegexType& regex_; + Allocator* allocator_; + Allocator* ownAllocator_; + Stack state0_; + Stack state1_; + uint32_t* stateSet_; +}; + typedef GenericRegex > Regex; +typedef GenericRegexSearch RegexSearch; } // namespace internal CEREAL_RAPIDJSON_NAMESPACE_END -#ifdef __clang__ +#ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_POP #endif -#ifdef _MSC_VER +#if defined(__clang__) || defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/internal/stack.h b/noarch/include/cereal/external/rapidjson/internal/stack.h index 768551e9..be77088f 100644 --- a/noarch/include/cereal/external/rapidjson/internal/stack.h +++ b/noarch/include/cereal/external/rapidjson/internal/stack.h @@ -17,6 +17,7 @@ #include "../allocators.h" #include "swap.h" +#include #if defined(__clang__) CEREAL_RAPIDJSON_DIAG_PUSH @@ -100,7 +101,7 @@ public: void ShrinkToFit() { if (Empty()) { // If the stack is empty, completely deallocate the memory. - Allocator::Free(stack_); + Allocator::Free(stack_); // NOLINT (+clang-analyzer-unix.Malloc) stack_ = 0; stackTop_ = 0; stackEnd_ = 0; @@ -114,7 +115,7 @@ public: template CEREAL_RAPIDJSON_FORCEINLINE void Reserve(size_t count = 1) { // Expand the stack if needed - if (CEREAL_RAPIDJSON_UNLIKELY(stackTop_ + sizeof(T) * count > stackEnd_)) + if (CEREAL_RAPIDJSON_UNLIKELY(static_cast(sizeof(T) * count) > (stackEnd_ - stackTop_))) Expand(count); } @@ -126,7 +127,8 @@ public: template CEREAL_RAPIDJSON_FORCEINLINE T* PushUnsafe(size_t count = 1) { - CEREAL_RAPIDJSON_ASSERT(stackTop_ + sizeof(T) * count <= stackEnd_); + CEREAL_RAPIDJSON_ASSERT(stackTop_); + CEREAL_RAPIDJSON_ASSERT(static_cast(sizeof(T) * count) <= (stackEnd_ - stackTop_)); T* ret = reinterpret_cast(stackTop_); stackTop_ += sizeof(T) * count; return ret; @@ -183,7 +185,7 @@ private: size_t newCapacity; if (stack_ == 0) { if (!allocator_) - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); newCapacity = initialCapacity_; } else { newCapacity = GetCapacity(); diff --git a/noarch/include/cereal/external/rapidjson/internal/strfunc.h b/noarch/include/cereal/external/rapidjson/internal/strfunc.h index 0c6973c6..44af229d 100644 --- a/noarch/include/cereal/external/rapidjson/internal/strfunc.h +++ b/noarch/include/cereal/external/rapidjson/internal/strfunc.h @@ -16,6 +16,7 @@ #define CEREAL_RAPIDJSON_INTERNAL_STRFUNC_H_ #include "../stream.h" +#include CEREAL_RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -28,14 +29,27 @@ namespace internal { */ template inline SizeType StrLen(const Ch* s) { + CEREAL_RAPIDJSON_ASSERT(s != 0); const Ch* p = s; while (*p) ++p; return SizeType(p - s); } +template <> +inline SizeType StrLen(const char* s) { + return SizeType(std::strlen(s)); +} + +template <> +inline SizeType StrLen(const wchar_t* s) { + return SizeType(std::wcslen(s)); +} + //! Returns number of code points in a encoded string. template bool CountStringCodePoint(const typename Encoding::Ch* s, SizeType length, SizeType* outCount) { + CEREAL_RAPIDJSON_ASSERT(s != 0); + CEREAL_RAPIDJSON_ASSERT(outCount != 0); GenericStringStream is(s); const typename Encoding::Ch* end = s + length; SizeType count = 0; diff --git a/noarch/include/cereal/external/rapidjson/internal/strtod.h b/noarch/include/cereal/external/rapidjson/internal/strtod.h index 076434c3..d60c740b 100644 --- a/noarch/include/cereal/external/rapidjson/internal/strtod.h +++ b/noarch/include/cereal/external/rapidjson/internal/strtod.h @@ -19,6 +19,8 @@ #include "biginteger.h" #include "diyfp.h" #include "pow10.h" +#include +#include CEREAL_RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -126,46 +128,46 @@ inline bool StrtodFast(double d, int p, double* result) { } // Compute an approximation and see if it is within 1/2 ULP -inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosition, int exp, double* result) { +inline bool StrtodDiyFp(const char* decimals, int dLen, int dExp, double* result) { uint64_t significand = 0; - size_t i = 0; // 2^64 - 1 = 18446744073709551615, 1844674407370955161 = 0x1999999999999999 - for (; i < length; i++) { + int i = 0; // 2^64 - 1 = 18446744073709551615, 1844674407370955161 = 0x1999999999999999 + for (; i < dLen; i++) { if (significand > CEREAL_RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) || (significand == CEREAL_RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) && decimals[i] > '5')) break; significand = significand * 10u + static_cast(decimals[i] - '0'); } - if (i < length && decimals[i] >= '5') // Rounding + if (i < dLen && decimals[i] >= '5') // Rounding significand++; - size_t remaining = length - i; - const unsigned kUlpShift = 3; - const unsigned kUlp = 1 << kUlpShift; + int remaining = dLen - i; + const int kUlpShift = 3; + const int kUlp = 1 << kUlpShift; int64_t error = (remaining == 0) ? 0 : kUlp / 2; DiyFp v(significand, 0); v = v.Normalize(); error <<= -v.e; - const int dExp = static_cast(decimalPosition) - static_cast(i) + exp; + dExp += remaining; int actualExp; DiyFp cachedPower = GetCachedPower10(dExp, &actualExp); if (actualExp != dExp) { static const DiyFp kPow10[] = { - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xa0000000, 00000000), -60), // 10^1 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xc8000000, 00000000), -57), // 10^2 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xfa000000, 00000000), -54), // 10^3 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0x9c400000, 00000000), -50), // 10^4 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xc3500000, 00000000), -47), // 10^5 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xf4240000, 00000000), -44), // 10^6 - DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0x98968000, 00000000), -40) // 10^7 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xa0000000, 0x00000000), -60), // 10^1 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xc8000000, 0x00000000), -57), // 10^2 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xfa000000, 0x00000000), -54), // 10^3 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0x9c400000, 0x00000000), -50), // 10^4 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xc3500000, 0x00000000), -47), // 10^5 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0xf4240000, 0x00000000), -44), // 10^6 + DiyFp(CEREAL_RAPIDJSON_UINT64_C2(0x98968000, 0x00000000), -40) // 10^7 }; - int adjustment = dExp - actualExp - 1; - CEREAL_RAPIDJSON_ASSERT(adjustment >= 0 && adjustment < 7); - v = v * kPow10[adjustment]; - if (length + static_cast(adjustment)> 19u) // has more digits than decimal digits in 64-bit + int adjustment = dExp - actualExp; + CEREAL_RAPIDJSON_ASSERT(adjustment >= 1 && adjustment < 8); + v = v * kPow10[adjustment - 1]; + if (dLen + adjustment > 19) // has more digits than decimal digits in 64-bit error += kUlp / 2; } @@ -177,17 +179,17 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit v = v.Normalize(); error <<= oldExp - v.e; - const unsigned effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); - unsigned precisionSize = 64 - effectiveSignificandSize; + const int effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); + int precisionSize = 64 - effectiveSignificandSize; if (precisionSize + kUlpShift >= 64) { - unsigned scaleExp = (precisionSize + kUlpShift) - 63; + int scaleExp = (precisionSize + kUlpShift) - 63; v.f >>= scaleExp; v.e += scaleExp; - error = (error >> scaleExp) + 1 + static_cast(kUlp); + error = (error >> scaleExp) + 1 + kUlp; precisionSize -= scaleExp; } - DiyFp rounded(v.f >> precisionSize, v.e + static_cast(precisionSize)); + DiyFp rounded(v.f >> precisionSize, v.e + precisionSize); const uint64_t precisionBits = (v.f & ((uint64_t(1) << precisionSize) - 1)) * kUlp; const uint64_t halfWay = (uint64_t(1) << (precisionSize - 1)) * kUlp; if (precisionBits >= halfWay + static_cast(error)) { @@ -203,9 +205,9 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit return halfWay - static_cast(error) >= precisionBits || precisionBits >= halfWay + static_cast(error); } -inline double StrtodBigInteger(double approx, const char* decimals, size_t length, size_t decimalPosition, int exp) { - const BigInteger dInt(decimals, length); - const int dExp = static_cast(decimalPosition) - static_cast(length) + exp; +inline double StrtodBigInteger(double approx, const char* decimals, int dLen, int dExp) { + CEREAL_RAPIDJSON_ASSERT(dLen >= 0); + const BigInteger dInt(decimals, static_cast(dLen)); Double a(approx); int cmp = CheckWithinHalfULP(a.Value(), dInt, dExp); if (cmp < 0) @@ -225,42 +227,61 @@ inline double StrtodFullPrecision(double d, int p, const char* decimals, size_t CEREAL_RAPIDJSON_ASSERT(d >= 0.0); CEREAL_RAPIDJSON_ASSERT(length >= 1); - double result; + double result = 0.0; if (StrtodFast(d, p, &result)) return result; + CEREAL_RAPIDJSON_ASSERT(length <= INT_MAX); + int dLen = static_cast(length); + + CEREAL_RAPIDJSON_ASSERT(length >= decimalPosition); + CEREAL_RAPIDJSON_ASSERT(length - decimalPosition <= INT_MAX); + int dExpAdjust = static_cast(length - decimalPosition); + + CEREAL_RAPIDJSON_ASSERT(exp >= INT_MIN + dExpAdjust); + int dExp = exp - dExpAdjust; + + // Make sure length+dExp does not overflow + CEREAL_RAPIDJSON_ASSERT(dExp <= INT_MAX - dLen); + // Trim leading zeros - while (*decimals == '0' && length > 1) { - length--; + while (dLen > 0 && *decimals == '0') { + dLen--; decimals++; - decimalPosition--; } // Trim trailing zeros - while (decimals[length - 1] == '0' && length > 1) { - length--; - decimalPosition--; - exp++; + while (dLen > 0 && decimals[dLen - 1] == '0') { + dLen--; + dExp++; + } + + if (dLen == 0) { // Buffer only contains zeros. + return 0.0; } // Trim right-most digits - const int kMaxDecimalDigit = 780; - if (static_cast(length) > kMaxDecimalDigit) { - int delta = (static_cast(length) - kMaxDecimalDigit); - exp += delta; - decimalPosition -= static_cast(delta); - length = kMaxDecimalDigit; + const int kMaxDecimalDigit = 767 + 1; + if (dLen > kMaxDecimalDigit) { + dExp += dLen - kMaxDecimalDigit; + dLen = kMaxDecimalDigit; } - // If too small, underflow to zero - if (int(length) + exp < -324) + // If too small, underflow to zero. + // Any x <= 10^-324 is interpreted as zero. + if (dLen + dExp <= -324) return 0.0; - if (StrtodDiyFp(decimals, length, decimalPosition, exp, &result)) + // If too large, overflow to infinity. + // Any x >= 10^309 is interpreted as +infinity. + if (dLen + dExp > 309) + return std::numeric_limits::infinity(); + + if (StrtodDiyFp(decimals, dLen, dExp, &result)) return result; // Use approximation from StrtodDiyFp and make adjustment with BigInteger comparison - return StrtodBigInteger(result, decimals, length, decimalPosition, exp); + return StrtodBigInteger(result, decimals, dLen, dExp); } } // namespace internal diff --git a/noarch/include/cereal/external/rapidjson/istreamwrapper.h b/noarch/include/cereal/external/rapidjson/istreamwrapper.h index e1c7ea8d..4df7ce3e 100644 --- a/noarch/include/cereal/external/rapidjson/istreamwrapper.h +++ b/noarch/include/cereal/external/rapidjson/istreamwrapper.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef CEREAL_RAPIDJSON_ISTREAMWRAPPER_H_ @@ -17,16 +17,14 @@ #include "stream.h" #include +#include #ifdef __clang__ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(padded) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(4351) // new behavior: elements of array 'array' will be default initialized -CEREAL_RAPIDJSON_DIAG_OFF(4127) // ignore assert(false) for triggering exception #endif CEREAL_RAPIDJSON_NAMESPACE_BEGIN @@ -46,62 +44,76 @@ CEREAL_RAPIDJSON_NAMESPACE_BEGIN \tparam StreamType Class derived from \c std::basic_istream. */ - + template class BasicIStreamWrapper { public: typedef typename StreamType::char_type Ch; - BasicIStreamWrapper(StreamType& stream) : stream_(stream), count_(), peekBuffer_() {} - Ch Peek() const { - typename StreamType::int_type c = stream_.peek(); - return CEREAL_RAPIDJSON_LIKELY(c != StreamType::traits_type::eof()) ? static_cast(c) : '\0'; + //! Constructor. + /*! + \param stream stream opened for read. + */ + BasicIStreamWrapper(StreamType &stream) : stream_(stream), buffer_(peekBuffer_), bufferSize_(4), bufferLast_(0), current_(buffer_), readCount_(0), count_(0), eof_(false) { + Read(); } - Ch Take() { - typename StreamType::int_type c = stream_.get(); - if (CEREAL_RAPIDJSON_LIKELY(c != StreamType::traits_type::eof())) { - count_++; - return static_cast(c); - } - else - return '\0'; + //! Constructor. + /*! + \param stream stream opened for read. + \param buffer user-supplied buffer. + \param bufferSize size of buffer in bytes. Must >=4 bytes. + */ + BasicIStreamWrapper(StreamType &stream, char* buffer, size_t bufferSize) : stream_(stream), buffer_(buffer), bufferSize_(bufferSize), bufferLast_(0), current_(buffer_), readCount_(0), count_(0), eof_(false) { + CEREAL_RAPIDJSON_ASSERT(bufferSize >= 4); + Read(); } - // tellg() may return -1 when failed. So we count by ourself. - size_t Tell() const { return count_; } + Ch Peek() const { return *current_; } + Ch Take() { Ch c = *current_; Read(); return c; } + size_t Tell() const { return count_ + static_cast(current_ - buffer_); } - Ch* PutBegin() { CEREAL_RAPIDJSON_ASSERT(false); return 0; } + // Not implemented void Put(Ch) { CEREAL_RAPIDJSON_ASSERT(false); } - void Flush() { CEREAL_RAPIDJSON_ASSERT(false); } + void Flush() { CEREAL_RAPIDJSON_ASSERT(false); } + Ch* PutBegin() { CEREAL_RAPIDJSON_ASSERT(false); return 0; } size_t PutEnd(Ch*) { CEREAL_RAPIDJSON_ASSERT(false); return 0; } // For encoding detection only. const Ch* Peek4() const { - CEREAL_RAPIDJSON_ASSERT(sizeof(Ch) == 1); // Only usable for byte stream. - int i; - bool hasError = false; - for (i = 0; i < 4; ++i) { - typename StreamType::int_type c = stream_.get(); - if (c == StreamType::traits_type::eof()) { - hasError = true; - stream_.clear(); - break; - } - peekBuffer_[i] = static_cast(c); - } - for (--i; i >= 0; --i) - stream_.putback(peekBuffer_[i]); - return !hasError ? peekBuffer_ : 0; + return (current_ + 4 - !eof_ <= bufferLast_) ? current_ : 0; } private: + BasicIStreamWrapper(); BasicIStreamWrapper(const BasicIStreamWrapper&); BasicIStreamWrapper& operator=(const BasicIStreamWrapper&); - StreamType& stream_; - size_t count_; //!< Number of characters read. Note: - mutable Ch peekBuffer_[4]; + void Read() { + if (current_ < bufferLast_) + ++current_; + else if (!eof_) { + count_ += readCount_; + readCount_ = bufferSize_; + bufferLast_ = buffer_ + readCount_ - 1; + current_ = buffer_; + + if (!stream_.read(buffer_, static_cast(bufferSize_))) { + readCount_ = static_cast(stream_.gcount()); + *(bufferLast_ = buffer_ + readCount_) = '\0'; + eof_ = true; + } + } + } + + StreamType &stream_; + Ch peekBuffer_[4], *buffer_; + size_t bufferSize_; + Ch *bufferLast_; + Ch *current_; + size_t readCount_; + size_t count_; //!< Number of characters read + bool eof_; }; typedef BasicIStreamWrapper IStreamWrapper; diff --git a/noarch/include/cereal/external/rapidjson/memorystream.h b/noarch/include/cereal/external/rapidjson/memorystream.h index e48988cd..326bda54 100644 --- a/noarch/include/cereal/external/rapidjson/memorystream.h +++ b/noarch/include/cereal/external/rapidjson/memorystream.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef CEREAL_RAPIDJSON_MEMORYSTREAM_H_ @@ -23,11 +23,6 @@ CEREAL_RAPIDJSON_DIAG_OFF(unreachable-code) CEREAL_RAPIDJSON_DIAG_OFF(missing-noreturn) #endif -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_PUSH -CEREAL_RAPIDJSON_DIAG_OFF( 4127 ) // ignore assert(false) for triggering exception -#endif - CEREAL_RAPIDJSON_NAMESPACE_BEGIN //! Represents an in-memory input byte stream. @@ -69,7 +64,7 @@ struct MemoryStream { CEREAL_RAPIDJSON_NAMESPACE_END -#if defined(__clang__) || defined(_MSC_VER) +#ifdef __clang__ CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/pointer.h b/noarch/include/cereal/external/rapidjson/pointer.h index 1e4836cd..930970f9 100644 --- a/noarch/include/cereal/external/rapidjson/pointer.h +++ b/noarch/include/cereal/external/rapidjson/pointer.h @@ -21,9 +21,7 @@ #ifdef __clang__ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(switch-enum) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif @@ -165,7 +163,12 @@ public: GenericPointer(const Token* tokens, size_t tokenCount) : allocator_(), ownAllocator_(), nameBuffer_(), tokens_(const_cast(tokens)), tokenCount_(tokenCount), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) {} //! Copy constructor. - GenericPointer(const GenericPointer& rhs, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + GenericPointer(const GenericPointer& rhs) : allocator_(rhs.allocator_), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + *this = rhs; + } + + //! Copy constructor. + GenericPointer(const GenericPointer& rhs, Allocator* allocator) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { *this = rhs; } @@ -197,6 +200,36 @@ public: return *this; } + //! Swap the content of this pointer with an other. + /*! + \param other The pointer to swap with. + \note Constant complexity. + */ + GenericPointer& Swap(GenericPointer& other) CEREAL_RAPIDJSON_NOEXCEPT { + internal::Swap(allocator_, other.allocator_); + internal::Swap(ownAllocator_, other.ownAllocator_); + internal::Swap(nameBuffer_, other.nameBuffer_); + internal::Swap(tokens_, other.tokens_); + internal::Swap(tokenCount_, other.tokenCount_); + internal::Swap(parseErrorOffset_, other.parseErrorOffset_); + internal::Swap(parseErrorCode_, other.parseErrorCode_); + return *this; + } + + //! free-standing swap function helper + /*! + Helper function to enable support for common swap implementation pattern based on \c std::swap: + \code + void swap(MyClass& a, MyClass& b) { + using std::swap; + swap(a.pointer, b.pointer); + // ... + } + \endcode + \see Swap() + */ + friend inline void swap(GenericPointer& a, GenericPointer& b) CEREAL_RAPIDJSON_NOEXCEPT { a.Swap(b); } + //@} //!@name Append token @@ -240,7 +273,7 @@ public: template CEREAL_RAPIDJSON_DISABLEIF_RETURN((internal::NotExpr::Type, Ch> >), (GenericPointer)) Append(T* name, Allocator* allocator = 0) const { - return Append(name, StrLen(name), allocator); + return Append(name, internal::StrLen(name), allocator); } #if CEREAL_RAPIDJSON_HAS_STDSTRING @@ -274,7 +307,7 @@ public: else { Ch name[21]; for (size_t i = 0; i <= length; i++) - name[i] = buffer[i]; + name[i] = static_cast(buffer[i]); Token token = { name, length, index }; return Append(token, allocator); } @@ -353,6 +386,33 @@ public: */ bool operator!=(const GenericPointer& rhs) const { return !(*this == rhs); } + //! Less than operator. + /*! + \note Invalid pointers are always greater than valid ones. + */ + bool operator<(const GenericPointer& rhs) const { + if (!IsValid()) + return false; + if (!rhs.IsValid()) + return true; + + if (tokenCount_ != rhs.tokenCount_) + return tokenCount_ < rhs.tokenCount_; + + for (size_t i = 0; i < tokenCount_; i++) { + if (tokens_[i].index != rhs.tokens_[i].index) + return tokens_[i].index < rhs.tokens_[i].index; + + if (tokens_[i].length != rhs.tokens_[i].length) + return tokens_[i].length < rhs.tokens_[i].length; + + if (int cmp = std::memcmp(tokens_[i].name, rhs.tokens_[i].name, sizeof(Ch) * tokens_[i].length)) + return cmp < 0; + } + + return false; + } + //@} //!@name Stringify @@ -532,14 +592,14 @@ public: */ ValueType& GetWithDefault(ValueType& root, const ValueType& defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.CopyFrom(defaultValue, allocator); } //! Query a value in a subtree with default null-terminated string. ValueType& GetWithDefault(ValueType& root, const Ch* defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.SetString(defaultValue, allocator); } @@ -547,7 +607,7 @@ public: //! Query a value in a subtree with default std::basic_string. ValueType& GetWithDefault(ValueType& root, const std::basic_string& defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.SetString(defaultValue, allocator); } #endif @@ -758,7 +818,7 @@ private: */ Ch* CopyFromRaw(const GenericPointer& rhs, size_t extraToken = 0, size_t extraNameBufferSize = 0) { if (!allocator_) // allocator is independently owned. - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); size_t nameBufferSize = rhs.tokenCount_; // null terminators for tokens for (Token *t = rhs.tokens_; t != rhs.tokens_ + rhs.tokenCount_; ++t) @@ -806,7 +866,7 @@ private: // Create own allocator if user did not supply. if (!allocator_) - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); // Count number of '/' as tokenCount tokenCount_ = 0; @@ -1029,8 +1089,8 @@ private: unsigned char u = static_cast(c); static const char hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; os_.Put('%'); - os_.Put(hexDigits[u >> 4]); - os_.Put(hexDigits[u & 15]); + os_.Put(static_cast(hexDigits[u >> 4])); + os_.Put(static_cast(hexDigits[u & 15])); } private: OutputStream& os_; @@ -1347,11 +1407,7 @@ bool EraseValueByPointer(T& root, const CharType(&source)[N]) { CEREAL_RAPIDJSON_NAMESPACE_END -#ifdef __clang__ -CEREAL_RAPIDJSON_DIAG_POP -#endif - -#ifdef _MSC_VER +#if defined(__clang__) || defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/prettywriter.h b/noarch/include/cereal/external/rapidjson/prettywriter.h index 50a24fc6..0c2beb49 100644 --- a/noarch/include/cereal/external/rapidjson/prettywriter.h +++ b/noarch/include/cereal/external/rapidjson/prettywriter.h @@ -22,6 +22,11 @@ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(effc++) #endif +#if defined(__clang__) +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + CEREAL_RAPIDJSON_NAMESPACE_BEGIN //! Combination of PrettyWriter format flags. @@ -34,7 +39,7 @@ enum PrettyFormatOptions { //! Writer with indentation and spacing. /*! - \tparam OutputStream Type of ouptut os. + \tparam OutputStream Type of output os. \tparam SourceEncoding Encoding of source string. \tparam TargetEncoding Encoding of output stream. \tparam StackAllocator Type of allocator for allocating memory of stack. @@ -42,7 +47,7 @@ enum PrettyFormatOptions { template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags> class PrettyWriter : public Writer { public: - typedef Writer Base; + typedef Writer Base; typedef typename Base::Ch Ch; //! Constructor @@ -57,6 +62,11 @@ public: explicit PrettyWriter(StackAllocator* allocator = 0, size_t levelDepth = Base::kDefaultLevelDepth) : Base(allocator, levelDepth), indentChar_(' '), indentCharCount_(4) {} +#if CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS + PrettyWriter(PrettyWriter&& rhs) : + Base(std::forward(rhs)), indentChar_(rhs.indentChar_), indentCharCount_(rhs.indentCharCount_), formatOptions_(rhs.formatOptions_) {} +#endif + //! Set custom indentation. /*! \param indentChar Character for indentation. Must be whitespace character (' ', '\\t', '\\n', '\\r'). \param indentCharCount Number of indent characters for each indentation level. @@ -82,24 +92,26 @@ public: */ //@{ - bool Null() { PrettyPrefix(kNullType); return Base::WriteNull(); } - bool Bool(bool b) { PrettyPrefix(b ? kTrueType : kFalseType); return Base::WriteBool(b); } - bool Int(int i) { PrettyPrefix(kNumberType); return Base::WriteInt(i); } - bool Uint(unsigned u) { PrettyPrefix(kNumberType); return Base::WriteUint(u); } - bool Int64(int64_t i64) { PrettyPrefix(kNumberType); return Base::WriteInt64(i64); } - bool Uint64(uint64_t u64) { PrettyPrefix(kNumberType); return Base::WriteUint64(u64); } - bool Double(double d) { PrettyPrefix(kNumberType); return Base::WriteDouble(d); } + bool Null() { PrettyPrefix(kNullType); return Base::EndValue(Base::WriteNull()); } + bool Bool(bool b) { PrettyPrefix(b ? kTrueType : kFalseType); return Base::EndValue(Base::WriteBool(b)); } + bool Int(int i) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteInt(i)); } + bool Uint(unsigned u) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteUint(u)); } + bool Int64(int64_t i64) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteInt64(i64)); } + bool Uint64(uint64_t u64) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteUint64(u64)); } + bool Double(double d) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteDouble(d)); } bool RawNumber(const Ch* str, SizeType length, bool copy = false) { + CEREAL_RAPIDJSON_ASSERT(str != 0); (void)copy; PrettyPrefix(kNumberType); - return Base::WriteString(str, length); + return Base::EndValue(Base::WriteString(str, length)); } bool String(const Ch* str, SizeType length, bool copy = false) { + CEREAL_RAPIDJSON_ASSERT(str != 0); (void)copy; PrettyPrefix(kStringType); - return Base::WriteString(str, length); + return Base::EndValue(Base::WriteString(str, length)); } #if CEREAL_RAPIDJSON_HAS_STDSTRING @@ -124,19 +136,21 @@ public: bool EndObject(SizeType memberCount = 0) { (void)memberCount; - CEREAL_RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); - CEREAL_RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); + CEREAL_RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); // not inside an Object + CEREAL_RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); // currently inside an Array, not Object + CEREAL_RAPIDJSON_ASSERT(0 == Base::level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value + bool empty = Base::level_stack_.template Pop(1)->valueCount == 0; if (!empty) { Base::os_->Put('\n'); WriteIndent(); } - bool ret = Base::WriteEndObject(); + bool ret = Base::EndValue(Base::WriteEndObject()); (void)ret; CEREAL_RAPIDJSON_ASSERT(ret == true); if (Base::level_stack_.Empty()) // end of json text - Base::os_->Flush(); + Base::Flush(); return true; } @@ -156,11 +170,11 @@ public: Base::os_->Put('\n'); WriteIndent(); } - bool ret = Base::WriteEndArray(); + bool ret = Base::EndValue(Base::WriteEndArray()); (void)ret; CEREAL_RAPIDJSON_ASSERT(ret == true); if (Base::level_stack_.Empty()) // end of json text - Base::os_->Flush(); + Base::Flush(); return true; } @@ -184,7 +198,11 @@ public: \param type Type of the root of json. \note When using PrettyWriter::RawValue(), the result json may not be indented correctly. */ - bool RawValue(const Ch* json, size_t length, Type type) { PrettyPrefix(type); return Base::WriteRawValue(json, length); } + bool RawValue(const Ch* json, size_t length, Type type) { + CEREAL_RAPIDJSON_ASSERT(json != 0); + PrettyPrefix(type); + return Base::EndValue(Base::WriteRawValue(json, length)); + } protected: void PrettyPrefix(Type type) { @@ -233,7 +251,7 @@ protected: void WriteIndent() { size_t count = (Base::level_stack_.GetSize() / sizeof(typename Base::Level)) * indentCharCount_; - PutN(*Base::os_, static_cast(indentChar_), count); + PutN(*Base::os_, static_cast(indentChar_), count); } Ch indentChar_; @@ -248,6 +266,10 @@ private: CEREAL_RAPIDJSON_NAMESPACE_END +#if defined(__clang__) +CEREAL_RAPIDJSON_DIAG_POP +#endif + #ifdef __GNUC__ CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidjson/rapidjson.h b/noarch/include/cereal/external/rapidjson/rapidjson.h index b7286fe2..3eefe60a 100644 --- a/noarch/include/cereal/external/rapidjson/rapidjson.h +++ b/noarch/include/cereal/external/rapidjson/rapidjson.h @@ -26,7 +26,7 @@ Some RapidJSON features are configurable to adapt the library to a wide variety of platforms, environments and usage scenarios. Most of the - features can be configured in terms of overriden or predefined + features can be configured in terms of overridden or predefined preprocessor macros at compile-time. Some additional customization is available in the \ref CEREAL_RAPIDJSON_ERRORS APIs. @@ -49,6 +49,11 @@ // token stringification #define CEREAL_RAPIDJSON_STRINGIFY(x) CEREAL_RAPIDJSON_DO_STRINGIFY(x) #define CEREAL_RAPIDJSON_DO_STRINGIFY(x) #x + +// token concatenation +#define CEREAL_RAPIDJSON_JOIN(X, Y) CEREAL_RAPIDJSON_DO_JOIN(X, Y) +#define CEREAL_RAPIDJSON_DO_JOIN(X, Y) CEREAL_RAPIDJSON_DO_JOIN2(X, Y) +#define CEREAL_RAPIDJSON_DO_JOIN2(X, Y) X##Y //!@endcond /*! \def CEREAL_RAPIDJSON_MAJOR_VERSION @@ -68,8 +73,8 @@ \brief Version of RapidJSON in ".." string format. */ #define CEREAL_RAPIDJSON_MAJOR_VERSION 1 -#define CEREAL_RAPIDJSON_MINOR_VERSION 0 -#define CEREAL_RAPIDJSON_PATCH_VERSION 2 +#define CEREAL_RAPIDJSON_MINOR_VERSION 1 +#define CEREAL_RAPIDJSON_PATCH_VERSION 0 #define CEREAL_RAPIDJSON_VERSION_STRING \ CEREAL_RAPIDJSON_STRINGIFY(CEREAL_RAPIDJSON_MAJOR_VERSION.CEREAL_RAPIDJSON_MINOR_VERSION.CEREAL_RAPIDJSON_PATCH_VERSION) @@ -214,7 +219,7 @@ # elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ # define CEREAL_RAPIDJSON_ENDIAN CEREAL_RAPIDJSON_BIGENDIAN # else -# error Unknown machine endianess detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. # endif // __BYTE_ORDER__ // Detect with GLIBC's endian.h # elif defined(__GLIBC__) @@ -224,7 +229,7 @@ # elif (__BYTE_ORDER == __BIG_ENDIAN) # define CEREAL_RAPIDJSON_ENDIAN CEREAL_RAPIDJSON_BIGENDIAN # else -# error Unknown machine endianess detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. # endif // __GLIBC__ // Detect with _LITTLE_ENDIAN and _BIG_ENDIAN macro # elif defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN) @@ -236,12 +241,12 @@ # define CEREAL_RAPIDJSON_ENDIAN CEREAL_RAPIDJSON_BIGENDIAN # elif defined(__i386__) || defined(__alpha__) || defined(__ia64) || defined(__ia64__) || defined(_M_IX86) || defined(_M_IA64) || defined(_M_ALPHA) || defined(__amd64) || defined(__amd64__) || defined(_M_AMD64) || defined(__x86_64) || defined(__x86_64__) || defined(_M_X64) || defined(__bfin__) # define CEREAL_RAPIDJSON_ENDIAN CEREAL_RAPIDJSON_LITTLEENDIAN -# elif defined(_MSC_VER) && defined(_M_ARM) +# elif defined(_MSC_VER) && (defined(_M_ARM) || defined(_M_ARM64)) # define CEREAL_RAPIDJSON_ENDIAN CEREAL_RAPIDJSON_LITTLEENDIAN # elif defined(CEREAL_RAPIDJSON_DOXYGEN_RUNNING) # define CEREAL_RAPIDJSON_ENDIAN # else -# error Unknown machine endianess detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define CEREAL_RAPIDJSON_ENDIAN. # endif #endif // CEREAL_RAPIDJSON_ENDIAN @@ -264,16 +269,11 @@ /*! \ingroup CEREAL_RAPIDJSON_CONFIG \param x pointer to align - Some machines require strict data alignment. Currently the default uses 4 bytes - alignment on 32-bit platforms and 8 bytes alignment for 64-bit platforms. + Some machines require strict data alignment. The default is 8 bytes. User can customize by defining the CEREAL_RAPIDJSON_ALIGN function macro. */ #ifndef CEREAL_RAPIDJSON_ALIGN -#if CEREAL_RAPIDJSON_64BIT == 1 -#define CEREAL_RAPIDJSON_ALIGN(x) (((x) + static_cast(7u)) & ~static_cast(7u)) -#else -#define CEREAL_RAPIDJSON_ALIGN(x) (((x) + 3u) & ~3u) -#endif +#define CEREAL_RAPIDJSON_ALIGN(x) (((x) + static_cast(7u)) & ~static_cast(7u)) #endif /////////////////////////////////////////////////////////////////////////////// @@ -320,17 +320,17 @@ #endif /////////////////////////////////////////////////////////////////////////////// -// CEREAL_RAPIDJSON_SSE2/CEREAL_RAPIDJSON_SSE42/CEREAL_RAPIDJSON_SIMD +// CEREAL_RAPIDJSON_SSE2/CEREAL_RAPIDJSON_SSE42/CEREAL_RAPIDJSON_NEON/CEREAL_RAPIDJSON_SIMD /*! \def CEREAL_RAPIDJSON_SIMD \ingroup CEREAL_RAPIDJSON_CONFIG - \brief Enable SSE2/SSE4.2 optimization. + \brief Enable SSE2/SSE4.2/Neon optimization. RapidJSON supports optimized implementations for some parsing operations - based on the SSE2 or SSE4.2 SIMD extensions on modern Intel-compatible - processors. + based on the SSE2, SSE4.2 or NEon SIMD extensions on modern Intel + or ARM compatible processors. - To enable these optimizations, two different symbols can be defined; + To enable these optimizations, three different symbols can be defined; \code // Enable SSE2 optimization. #define CEREAL_RAPIDJSON_SSE2 @@ -339,13 +339,17 @@ #define CEREAL_RAPIDJSON_SSE42 \endcode - \c CEREAL_RAPIDJSON_SSE42 takes precedence, if both are defined. + // Enable ARM Neon optimization. + #define CEREAL_RAPIDJSON_NEON + \endcode + + \c CEREAL_RAPIDJSON_SSE42 takes precedence over SSE2, if both are defined. If any of these symbols is defined, RapidJSON defines the macro \c CEREAL_RAPIDJSON_SIMD to indicate the availability of the optimized code. */ #if defined(CEREAL_RAPIDJSON_SSE2) || defined(CEREAL_RAPIDJSON_SSE42) \ - || defined(CEREAL_RAPIDJSON_DOXYGEN_RUNNING) + || defined(CEREAL_RAPIDJSON_NEON) || defined(CEREAL_RAPIDJSON_DOXYGEN_RUNNING) #define CEREAL_RAPIDJSON_SIMD #endif @@ -405,7 +409,15 @@ CEREAL_RAPIDJSON_NAMESPACE_END /////////////////////////////////////////////////////////////////////////////// // CEREAL_RAPIDJSON_STATIC_ASSERT -// Adopt from boost +// Prefer C++11 static_assert, if available +#ifndef CEREAL_RAPIDJSON_STATIC_ASSERT +#if __cplusplus >= 201103L || ( defined(_MSC_VER) && _MSC_VER >= 1800 ) +#define CEREAL_RAPIDJSON_STATIC_ASSERT(x) \ + static_assert(x, CEREAL_RAPIDJSON_STRINGIFY(x)) +#endif // C++11 +#endif // CEREAL_RAPIDJSON_STATIC_ASSERT + +// Adopt C++03 implementation from boost #ifndef CEREAL_RAPIDJSON_STATIC_ASSERT #ifndef __clang__ //!@cond CEREAL_RAPIDJSON_HIDDEN_FROM_DOXYGEN @@ -413,14 +425,10 @@ CEREAL_RAPIDJSON_NAMESPACE_END CEREAL_RAPIDJSON_NAMESPACE_BEGIN template struct STATIC_ASSERTION_FAILURE; template <> struct STATIC_ASSERTION_FAILURE { enum { value = 1 }; }; -template struct StaticAssertTest {}; +template struct StaticAssertTest {}; CEREAL_RAPIDJSON_NAMESPACE_END -#define CEREAL_RAPIDJSON_JOIN(X, Y) CEREAL_RAPIDJSON_DO_JOIN(X, Y) -#define CEREAL_RAPIDJSON_DO_JOIN(X, Y) CEREAL_RAPIDJSON_DO_JOIN2(X, Y) -#define CEREAL_RAPIDJSON_DO_JOIN2(X, Y) X##Y - -#if defined(__GNUC__) +#if defined(__GNUC__) || defined(__clang__) #define CEREAL_RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE __attribute__((unused)) #else #define CEREAL_RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE @@ -438,7 +446,7 @@ CEREAL_RAPIDJSON_NAMESPACE_END typedef ::CEREAL_RAPIDJSON_NAMESPACE::StaticAssertTest< \ sizeof(::CEREAL_RAPIDJSON_NAMESPACE::STATIC_ASSERTION_FAILURE)> \ CEREAL_RAPIDJSON_JOIN(StaticAssertTypedef, __LINE__) CEREAL_RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE -#endif +#endif // CEREAL_RAPIDJSON_STATIC_ASSERT /////////////////////////////////////////////////////////////////////////////// // CEREAL_RAPIDJSON_LIKELY, CEREAL_RAPIDJSON_UNLIKELY @@ -530,13 +538,14 @@ CEREAL_RAPIDJSON_NAMESPACE_END #ifndef CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS #if defined(__clang__) #if __has_feature(cxx_rvalue_references) && \ - (defined(_LIBCPP_VERSION) || defined(__GLIBCXX__) && __GLIBCXX__ >= 20080306) + (defined(_MSC_VER) || defined(_LIBCPP_VERSION) || defined(__GLIBCXX__) && __GLIBCXX__ >= 20080306) #define CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS 1 #else #define CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS 0 #endif #elif (defined(CEREAL_RAPIDJSON_GNUC) && (CEREAL_RAPIDJSON_GNUC >= CEREAL_RAPIDJSON_VERSION_CODE(4,3,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ - (defined(_MSC_VER) && _MSC_VER >= 1600) + (defined(_MSC_VER) && _MSC_VER >= 1600) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS 1 #else @@ -547,8 +556,9 @@ CEREAL_RAPIDJSON_NAMESPACE_END #ifndef CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT #if defined(__clang__) #define CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT __has_feature(cxx_noexcept) -#elif (defined(CEREAL_RAPIDJSON_GNUC) && (CEREAL_RAPIDJSON_GNUC >= CEREAL_RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) -// (defined(_MSC_VER) && _MSC_VER >= ????) // not yet supported +#elif (defined(CEREAL_RAPIDJSON_GNUC) && (CEREAL_RAPIDJSON_GNUC >= CEREAL_RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ + (defined(_MSC_VER) && _MSC_VER >= 1900) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT 1 #else #define CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT 0 @@ -562,14 +572,19 @@ CEREAL_RAPIDJSON_NAMESPACE_END // no automatic detection, yet #ifndef CEREAL_RAPIDJSON_HAS_CXX11_TYPETRAITS +#if (defined(_MSC_VER) && _MSC_VER >= 1700) +#define CEREAL_RAPIDJSON_HAS_CXX11_TYPETRAITS 1 +#else #define CEREAL_RAPIDJSON_HAS_CXX11_TYPETRAITS 0 #endif +#endif #ifndef CEREAL_RAPIDJSON_HAS_CXX11_RANGE_FOR #if defined(__clang__) #define CEREAL_RAPIDJSON_HAS_CXX11_RANGE_FOR __has_feature(cxx_range_for) -#elif (defined(CEREAL_RAPIDJSON_GNUC) && (CEREAL_RAPIDJSON_GNUC >= CEREAL_RAPIDJSON_VERSION_CODE(4,3,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ - (defined(_MSC_VER) && _MSC_VER >= 1700) +#elif (defined(CEREAL_RAPIDJSON_GNUC) && (CEREAL_RAPIDJSON_GNUC >= CEREAL_RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ + (defined(_MSC_VER) && _MSC_VER >= 1700) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define CEREAL_RAPIDJSON_HAS_CXX11_RANGE_FOR 1 #else #define CEREAL_RAPIDJSON_HAS_CXX11_RANGE_FOR 0 @@ -578,12 +593,38 @@ CEREAL_RAPIDJSON_NAMESPACE_END //!@endcond +//! Assertion (in non-throwing contexts). + /*! \ingroup CEREAL_RAPIDJSON_CONFIG + Some functions provide a \c noexcept guarantee, if the compiler supports it. + In these cases, the \ref CEREAL_RAPIDJSON_ASSERT macro cannot be overridden to + throw an exception. This macro adds a separate customization point for + such cases. + + Defaults to C \c assert() (as \ref CEREAL_RAPIDJSON_ASSERT), if \c noexcept is + supported, and to \ref CEREAL_RAPIDJSON_ASSERT otherwise. + */ + +/////////////////////////////////////////////////////////////////////////////// +// CEREAL_RAPIDJSON_NOEXCEPT_ASSERT + +#ifndef CEREAL_RAPIDJSON_NOEXCEPT_ASSERT +#ifdef CEREAL_RAPIDJSON_ASSERT_THROWS +#if CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT +#define CEREAL_RAPIDJSON_NOEXCEPT_ASSERT(x) +#else +#define CEREAL_RAPIDJSON_NOEXCEPT_ASSERT(x) CEREAL_RAPIDJSON_ASSERT(x) +#endif // CEREAL_RAPIDJSON_HAS_CXX11_NOEXCEPT +#else +#define CEREAL_RAPIDJSON_NOEXCEPT_ASSERT(x) CEREAL_RAPIDJSON_ASSERT(x) +#endif // CEREAL_RAPIDJSON_ASSERT_THROWS +#endif // CEREAL_RAPIDJSON_NOEXCEPT_ASSERT + /////////////////////////////////////////////////////////////////////////////// // new/delete #ifndef CEREAL_RAPIDJSON_NEW ///! customization point for global \c new -#define CEREAL_RAPIDJSON_NEW(x) new x +#define CEREAL_RAPIDJSON_NEW(TypeName) new TypeName #endif #ifndef CEREAL_RAPIDJSON_DELETE ///! customization point for global \c delete diff --git a/noarch/include/cereal/external/rapidjson/reader.h b/noarch/include/cereal/external/rapidjson/reader.h index ad8ff332..43eb5259 100644 --- a/noarch/include/cereal/external/rapidjson/reader.h +++ b/noarch/include/cereal/external/rapidjson/reader.h @@ -33,12 +33,8 @@ #include #elif defined(CEREAL_RAPIDJSON_SSE2) #include -#endif - -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_PUSH -CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant -CEREAL_RAPIDJSON_DIAG_OFF(4702) // unreachable code +#elif defined(CEREAL_RAPIDJSON_NEON) +#include #endif #ifdef __clang__ @@ -46,6 +42,10 @@ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(old-style-cast) CEREAL_RAPIDJSON_DIAG_OFF(padded) CEREAL_RAPIDJSON_DIAG_OFF(switch-enum) +#elif defined(_MSC_VER) +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +CEREAL_RAPIDJSON_DIAG_OFF(4702) // unreachable code #endif #ifdef __GNUC__ @@ -299,16 +299,9 @@ inline const char *SkipWhitespace_SIMD(const char* p) { for (;; p += 16) { const __m128i s = _mm_load_si128(reinterpret_cast(p)); - const int r = _mm_cvtsi128_si32(_mm_cmpistrm(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK | _SIDD_NEGATIVE_POLARITY)); - if (r != 0) { // some of characters is non-whitespace -#ifdef _MSC_VER // Find the index of first non-whitespace - unsigned long offset; - _BitScanForward(&offset, r); - return p + offset; -#else - return p + __builtin_ffs(r) - 1; -#endif - } + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; } } @@ -325,16 +318,9 @@ inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { for (; p <= end - 16; p += 16) { const __m128i s = _mm_loadu_si128(reinterpret_cast(p)); - const int r = _mm_cvtsi128_si32(_mm_cmpistrm(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK | _SIDD_NEGATIVE_POLARITY)); - if (r != 0) { // some of characters is non-whitespace -#ifdef _MSC_VER // Find the index of first non-whitespace - unsigned long offset; - _BitScanForward(&offset, r); - return p + offset; -#else - return p + __builtin_ffs(r) - 1; -#endif - } + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; } return SkipWhitespace(p, end); @@ -425,7 +411,92 @@ inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { return SkipWhitespace(p, end); } -#endif // CEREAL_RAPIDJSON_SSE2 +#elif defined(CEREAL_RAPIDJSON_NEON) + +//! Skip whitespace with ARM Neon instructions, testing 16 8-byte characters at once. +inline const char *SkipWhitespace_SIMD(const char* p) { + // Fast return for single non-whitespace + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + // 16-byte align to the next boundary + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz =__builtin_clzll(high);; + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low);; + return p + (lz >> 3); + } + } +} + +inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { + // Fast return for single non-whitespace + if (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (; p <= end - 16; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low); + return p + (lz >> 3); + } + } + + return SkipWhitespace(p, end); +} + +#endif // CEREAL_RAPIDJSON_NEON #ifdef CEREAL_RAPIDJSON_SIMD //! Template function specialization for InsituStringStream @@ -471,7 +542,8 @@ public: /*! \param stackAllocator Optional allocator for allocating stack memory. (Only use for non-destructive parsing) \param stackCapacity stack capacity in bytes for storing a single decoded string. (Only use for non-destructive parsing) */ - GenericReader(StackAllocator* stackAllocator = 0, size_t stackCapacity = kDefaultStackCapacity) : stack_(stackAllocator, stackCapacity), parseResult_() {} + GenericReader(StackAllocator* stackAllocator = 0, size_t stackCapacity = kDefaultStackCapacity) : + stack_(stackAllocator, stackCapacity), parseResult_(), state_(IterativeParsingStartState) {} //! Parse JSON text. /*! \tparam parseFlags Combination of \ref ParseFlag. @@ -527,7 +599,84 @@ public: return Parse(is, handler); } - //! Whether a parse error has occured in the last parsing. + //! Initialize JSON text token-by-token parsing + /*! + */ + void IterativeParseInit() { + parseResult_.Clear(); + state_ = IterativeParsingStartState; + } + + //! Parse one token from JSON text + /*! \tparam InputStream Type of input stream, implementing Stream concept + \tparam Handler Type of handler, implementing Handler concept. + \param is Input stream to be parsed. + \param handler The handler to receive events. + \return Whether the parsing is successful. + */ + template + bool IterativeParseNext(InputStream& is, Handler& handler) { + while (CEREAL_RAPIDJSON_LIKELY(is.Peek() != '\0')) { + SkipWhitespaceAndComments(is); + + Token t = Tokenize(is.Peek()); + IterativeParsingState n = Predict(state_, t); + IterativeParsingState d = Transit(state_, t, n, is, handler); + + // If we've finished or hit an error... + if (CEREAL_RAPIDJSON_UNLIKELY(IsIterativeParsingCompleteState(d))) { + // Report errors. + if (d == IterativeParsingErrorState) { + HandleError(state_, is); + return false; + } + + // Transition to the finish state. + CEREAL_RAPIDJSON_ASSERT(d == IterativeParsingFinishState); + state_ = d; + + // If StopWhenDone is not set... + if (!(parseFlags & kParseStopWhenDoneFlag)) { + // ... and extra non-whitespace data is found... + SkipWhitespaceAndComments(is); + if (is.Peek() != '\0') { + // ... this is considered an error. + HandleError(state_, is); + return false; + } + } + + // Success! We are done! + return true; + } + + // Transition to the new state. + state_ = d; + + // If we parsed anything other than a delimiter, we invoked the handler, so we can return true now. + if (!IsIterativeParsingDelimiterState(n)) + return true; + } + + // We reached the end of file. + stack_.Clear(); + + if (state_ != IterativeParsingFinishState) { + HandleError(state_, is); + return false; + } + + return true; + } + + //! Check if token-by-token parsing JSON text is complete + /*! \return Whether the JSON has been fully decoded. + */ + CEREAL_RAPIDJSON_FORCEINLINE bool IterativeParseComplete() const { + return IsIterativeParsingCompleteState(state_); + } + + //! Whether a parse error has occurred in the last parsing. bool HasParseError() const { return parseResult_.IsError(); } //! Get the \ref ParseErrorCode of last parsing. @@ -575,7 +724,7 @@ private: } } else if (CEREAL_RAPIDJSON_LIKELY(Consume(is, '/'))) - while (is.Peek() != '\0' && is.Take() != '\n'); + while (is.Peek() != '\0' && is.Take() != '\n') {} else CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorUnspecificSyntaxError, is.Tell()); @@ -750,7 +899,7 @@ private: return false; } - // Helper function to parse four hexidecimal digits in \uXXXX in ParseString(). + // Helper function to parse four hexadecimal digits in \uXXXX in ParseString(). template unsigned ParseHex4(InputStream& is, size_t escapeOffset) { unsigned codepoint = 0; @@ -857,7 +1006,7 @@ private: Ch c = is.Peek(); if (CEREAL_RAPIDJSON_UNLIKELY(c == '\\')) { // Escape - size_t escapeOffset = is.Tell(); // For invalid escaping, report the inital '\\' as error offset + size_t escapeOffset = is.Tell(); // For invalid escaping, report the initial '\\' as error offset is.Take(); Ch e = is.Peek(); if ((sizeof(Ch) == 1 || unsigned(e) < 256) && CEREAL_RAPIDJSON_LIKELY(escape[static_cast(e)])) { @@ -892,7 +1041,7 @@ private: if (c == '\0') CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell()); else - CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, is.Tell()); + CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorStringInvalidEncoding, is.Tell()); } else { size_t offset = is.Tell(); @@ -927,7 +1076,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -936,7 +1085,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (CEREAL_RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -948,11 +1097,13 @@ private: #else length = static_cast(__builtin_ffs(r) - 1); #endif - char* q = reinterpret_cast(os.Push(length)); - for (size_t i = 0; i < length; i++) - q[i] = p[i]; + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; - p += length; + p += length; + } break; } _mm_storeu_si128(reinterpret_cast<__m128i *>(os.Push(16)), s); @@ -988,7 +1139,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -997,7 +1148,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (CEREAL_RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -1036,7 +1187,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -1045,7 +1196,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (CEREAL_RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -1064,7 +1215,180 @@ private: is.src_ = is.dst_ = p; } -#endif +#elif defined(CEREAL_RAPIDJSON_NEON) + // StringStream -> StackStream + static CEREAL_RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(StringStream& is, StackStream& os) { + const char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (CEREAL_RAPIDJSON_UNLIKELY(*p == '\"') || CEREAL_RAPIDJSON_UNLIKELY(*p == '\\') || CEREAL_RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + return; + } + else + os.Put(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high);; + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low);; + length = lz >> 3; + escaped = true; + } + if (CEREAL_RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; + + p += length; + } + break; + } + vst1q_u8(reinterpret_cast(os.Push(16)), s); + } + + is.src_ = p; + } + + // InsituStringStream -> InsituStringStream + static CEREAL_RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(InsituStringStream& is, InsituStringStream& os) { + CEREAL_RAPIDJSON_ASSERT(&is == &os); + (void)os; + + if (is.src_ == is.dst_) { + SkipUnescapedString(is); + return; + } + + char* p = is.src_; + char *q = is.dst_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (CEREAL_RAPIDJSON_UNLIKELY(*p == '\"') || CEREAL_RAPIDJSON_UNLIKELY(*p == '\\') || CEREAL_RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + is.dst_ = q; + return; + } + else + *q++ = *p++; + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16, q += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + length = lz >> 3; + escaped = true; + } + if (CEREAL_RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + for (const char* pend = p + length; p != pend; ) { + *q++ = *p++; + } + break; + } + vst1q_u8(reinterpret_cast(q), s); + } + + is.src_ = p; + is.dst_ = q; + } + + // When read/write pointers are the same for insitu stream, just skip unescaped characters + static CEREAL_RAPIDJSON_FORCEINLINE void SkipUnescapedString(InsituStringStream& is) { + CEREAL_RAPIDJSON_ASSERT(is.src_ == is.dst_); + char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + for (; p != nextAligned; p++) + if (CEREAL_RAPIDJSON_UNLIKELY(*p == '\"') || CEREAL_RAPIDJSON_UNLIKELY(*p == '\\') || CEREAL_RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = is.dst_ = p; + return; + } + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + p += 8 + (lz >> 3); + break; + } + } else { + int lz = __builtin_clzll(low); + p += lz >> 3; + break; + } + } + + is.src_ = is.dst_ = p; + } +#endif // CEREAL_RAPIDJSON_NEON template class NumberStream; @@ -1075,7 +1399,6 @@ private: typedef typename InputStream::Ch Ch; NumberStream(GenericReader& reader, InputStream& s) : is(s) { (void)reader; } - ~NumberStream() {} CEREAL_RAPIDJSON_FORCEINLINE Ch Peek() const { return is.Peek(); } CEREAL_RAPIDJSON_FORCEINLINE Ch TakePush() { return is.Take(); } @@ -1097,7 +1420,6 @@ private: typedef NumberStream Base; public: NumberStream(GenericReader& reader, InputStream& s) : Base(reader, s), stackStream(reader.stack_) {} - ~NumberStream() {} CEREAL_RAPIDJSON_FORCEINLINE Ch TakePush() { stackStream.Put(static_cast(Base::is.Peek())); @@ -1124,7 +1446,6 @@ private: typedef NumberStream Base; public: NumberStream(GenericReader& reader, InputStream& is) : Base(reader, is) {} - ~NumberStream() {} CEREAL_RAPIDJSON_FORCEINLINE Ch Take() { return Base::TakePush(); } }; @@ -1185,18 +1506,27 @@ private: } // Parse NaN or Infinity here else if ((parseFlags & kParseNanAndInfFlag) && CEREAL_RAPIDJSON_LIKELY((s.Peek() == 'I' || s.Peek() == 'N'))) { - useNanOrInf = true; - if (CEREAL_RAPIDJSON_LIKELY(Consume(s, 'N') && Consume(s, 'a') && Consume(s, 'N'))) { - d = std::numeric_limits::quiet_NaN(); + if (Consume(s, 'N')) { + if (Consume(s, 'a') && Consume(s, 'N')) { + d = std::numeric_limits::quiet_NaN(); + useNanOrInf = true; + } } - else if (CEREAL_RAPIDJSON_LIKELY(Consume(s, 'I') && Consume(s, 'n') && Consume(s, 'f'))) { - d = (minus ? -std::numeric_limits::infinity() : std::numeric_limits::infinity()); - if (CEREAL_RAPIDJSON_UNLIKELY(s.Peek() == 'i' && !(Consume(s, 'i') && Consume(s, 'n') - && Consume(s, 'i') && Consume(s, 't') && Consume(s, 'y')))) - CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + else if (CEREAL_RAPIDJSON_LIKELY(Consume(s, 'I'))) { + if (Consume(s, 'n') && Consume(s, 'f')) { + d = (minus ? -std::numeric_limits::infinity() : std::numeric_limits::infinity()); + useNanOrInf = true; + + if (CEREAL_RAPIDJSON_UNLIKELY(s.Peek() == 'i' && !(Consume(s, 'i') && Consume(s, 'n') + && Consume(s, 'i') && Consume(s, 't') && Consume(s, 'y')))) { + CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } + } } - else + + if (CEREAL_RAPIDJSON_UNLIKELY(!useNanOrInf)) { CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } } else CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); @@ -1231,8 +1561,6 @@ private: // Force double for big integer if (useDouble) { while (CEREAL_RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { - if (CEREAL_RAPIDJSON_UNLIKELY(d >= 1.7976931348623157e307)) // DBL_MAX / 10.0 - CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); d = d * 10 + (s.TakePush() - '0'); } } @@ -1302,9 +1630,18 @@ private: if (CEREAL_RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { exp = static_cast(s.Take() - '0'); if (expMinus) { + // (exp + expFrac) must not underflow int => we're detecting when -exp gets + // dangerously close to INT_MIN (a pessimistic next digit 9 would push it into + // underflow territory): + // + // -(exp * 10 + 9) + expFrac >= INT_MIN + // <=> exp <= (expFrac - INT_MIN - 9) / 10 + CEREAL_RAPIDJSON_ASSERT(expFrac <= 0); + int maxExp = (expFrac + 2147483639) / 10; + while (CEREAL_RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { exp = exp * 10 + static_cast(s.Take() - '0'); - if (exp >= 214748364) { // Issue #313: prevent overflow exponent + if (CEREAL_RAPIDJSON_UNLIKELY(exp > maxExp)) { while (CEREAL_RAPIDJSON_UNLIKELY(s.Peek() >= '0' && s.Peek() <= '9')) // Consume the rest of exponent s.Take(); } @@ -1363,6 +1700,13 @@ private: else d = internal::StrtodNormalPrecision(d, p); + // Use > max, instead of == inf, to fix bogus warning -Wfloat-equal + if (d > (std::numeric_limits::max)()) { + // Overflow + // TODO: internal::StrtodX should report overflow (or underflow) + CEREAL_RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); + } + cont = handler.Double(minus ? -d : d); } else if (useNanOrInf) { @@ -1408,29 +1752,31 @@ private: // States enum IterativeParsingState { - IterativeParsingStartState = 0, - IterativeParsingFinishState, - IterativeParsingErrorState, + IterativeParsingFinishState = 0, // sink states at top + IterativeParsingErrorState, // sink states at top + IterativeParsingStartState, // Object states IterativeParsingObjectInitialState, IterativeParsingMemberKeyState, - IterativeParsingKeyValueDelimiterState, IterativeParsingMemberValueState, - IterativeParsingMemberDelimiterState, IterativeParsingObjectFinishState, // Array states IterativeParsingArrayInitialState, IterativeParsingElementState, - IterativeParsingElementDelimiterState, IterativeParsingArrayFinishState, // Single value state - IterativeParsingValueState - }; + IterativeParsingValueState, - enum { cIterativeParsingStateCount = IterativeParsingValueState + 1 }; + // Delimiter states (at bottom) + IterativeParsingElementDelimiterState, + IterativeParsingMemberDelimiterState, + IterativeParsingKeyValueDelimiterState, + + cIterativeParsingStateCount + }; // Tokens enum Token { @@ -1452,7 +1798,7 @@ private: kTokenCount }; - CEREAL_RAPIDJSON_FORCEINLINE Token Tokenize(Ch c) { + CEREAL_RAPIDJSON_FORCEINLINE Token Tokenize(Ch c) const { //!@cond CEREAL_RAPIDJSON_HIDDEN_FROM_DOXYGEN #define N NumberToken @@ -1479,9 +1825,21 @@ private: return NumberToken; } - CEREAL_RAPIDJSON_FORCEINLINE IterativeParsingState Predict(IterativeParsingState state, Token token) { + CEREAL_RAPIDJSON_FORCEINLINE IterativeParsingState Predict(IterativeParsingState state, Token token) const { // current state x one lookahead token -> new state static const char G[cIterativeParsingStateCount][kTokenCount] = { + // Finish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Error(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // Start { IterativeParsingArrayInitialState, // Left bracket @@ -1496,18 +1854,6 @@ private: IterativeParsingValueState, // Null IterativeParsingValueState // Number }, - // Finish(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, - // Error(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, // ObjectInitial { IterativeParsingErrorState, // Left bracket @@ -1536,20 +1882,6 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // KeyValueDelimiter - { - IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) - IterativeParsingErrorState, // Right bracket - IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) - IterativeParsingErrorState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberValueState, // String - IterativeParsingMemberValueState, // False - IterativeParsingMemberValueState, // True - IterativeParsingMemberValueState, // Null - IterativeParsingMemberValueState // Number - }, // MemberValue { IterativeParsingErrorState, // Left bracket @@ -1564,20 +1896,6 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // MemberDelimiter - { - IterativeParsingErrorState, // Left bracket - IterativeParsingErrorState, // Right bracket - IterativeParsingErrorState, // Left curly bracket - IterativeParsingObjectFinishState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberKeyState, // String - IterativeParsingErrorState, // False - IterativeParsingErrorState, // True - IterativeParsingErrorState, // Null - IterativeParsingErrorState // Number - }, // ObjectFinish(sink state) { IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, @@ -1612,6 +1930,18 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, + // ArrayFinish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Single Value (sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // ElementDelimiter { IterativeParsingArrayInitialState, // Left bracket(push Element state) @@ -1626,18 +1956,34 @@ private: IterativeParsingElementState, // Null IterativeParsingElementState // Number }, - // ArrayFinish(sink state) + // MemberDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState + IterativeParsingErrorState, // Left bracket + IterativeParsingErrorState, // Right bracket + IterativeParsingErrorState, // Left curly bracket + IterativeParsingObjectFinishState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberKeyState, // String + IterativeParsingErrorState, // False + IterativeParsingErrorState, // True + IterativeParsingErrorState, // Null + IterativeParsingErrorState // Number }, - // Single Value (sink state) + // KeyValueDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - } + IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) + IterativeParsingErrorState, // Right bracket + IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) + IterativeParsingErrorState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberValueState, // String + IterativeParsingMemberValueState, // False + IterativeParsingMemberValueState, // True + IterativeParsingMemberValueState, // Null + IterativeParsingMemberValueState // Number + }, }; // End of G return static_cast(G[state][token]); @@ -1818,6 +2164,14 @@ private: } } + CEREAL_RAPIDJSON_FORCEINLINE bool IsIterativeParsingDelimiterState(IterativeParsingState s) const { + return s >= IterativeParsingElementDelimiterState; + } + + CEREAL_RAPIDJSON_FORCEINLINE bool IsIterativeParsingCompleteState(IterativeParsingState s) const { + return s <= IterativeParsingErrorState; + } + template ParseResult IterativeParse(InputStream& is, Handler& handler) { parseResult_.Clear(); @@ -1856,6 +2210,7 @@ private: static const size_t kDefaultStackCapacity = 256; //!< Default stack capacity in bytes for storing a single decoded string. internal::Stack stack_; //!< A stack for storing decoded string temporarily during non-destructive parsing. ParseResult parseResult_; + IterativeParsingState state_; }; // class GenericReader //! Reader with UTF8 encoding and default allocator. @@ -1863,7 +2218,7 @@ typedef GenericReader, UTF8<> > Reader; CEREAL_RAPIDJSON_NAMESPACE_END -#ifdef __clang__ +#if defined(__clang__) || defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_POP #endif @@ -1872,8 +2227,4 @@ CEREAL_RAPIDJSON_DIAG_POP CEREAL_RAPIDJSON_DIAG_POP #endif -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_POP -#endif - #endif // CEREAL_RAPIDJSON_READER_H_ diff --git a/noarch/include/cereal/external/rapidjson/schema.h b/noarch/include/cereal/external/rapidjson/schema.h index 12987550..17366c67 100644 --- a/noarch/include/cereal/external/rapidjson/schema.h +++ b/noarch/include/cereal/external/rapidjson/schema.h @@ -17,6 +17,7 @@ #include "document.h" #include "pointer.h" +#include "stringbuffer.h" #include // abs, floor #if !defined(CEREAL_RAPIDJSON_SCHEMA_USE_INTERNALREGEX) @@ -25,7 +26,7 @@ #define CEREAL_RAPIDJSON_SCHEMA_USE_INTERNALREGEX 0 #endif -#if !CEREAL_RAPIDJSON_SCHEMA_USE_INTERNALREGEX && !defined(CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX) && (__cplusplus >=201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)) +#if !CEREAL_RAPIDJSON_SCHEMA_USE_INTERNALREGEX && defined(CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX) && (__cplusplus >=201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)) #define CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX 1 #else #define CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX 0 @@ -62,9 +63,7 @@ CEREAL_RAPIDJSON_DIAG_OFF(weak-vtables) CEREAL_RAPIDJSON_DIAG_OFF(exit-time-destructors) CEREAL_RAPIDJSON_DIAG_OFF(c++98-compat-pedantic) CEREAL_RAPIDJSON_DIAG_OFF(variadic-macros) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif @@ -157,6 +156,62 @@ public: virtual void FreeState(void* p) = 0; }; +/////////////////////////////////////////////////////////////////////////////// +// IValidationErrorHandler + +template +class IValidationErrorHandler { +public: + typedef typename SchemaType::Ch Ch; + typedef typename SchemaType::SValue SValue; + + virtual ~IValidationErrorHandler() {} + + virtual void NotMultipleOf(int64_t actual, const SValue& expected) = 0; + virtual void NotMultipleOf(uint64_t actual, const SValue& expected) = 0; + virtual void NotMultipleOf(double actual, const SValue& expected) = 0; + virtual void AboveMaximum(int64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void AboveMaximum(uint64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void AboveMaximum(double actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(int64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(uint64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(double actual, const SValue& expected, bool exclusive) = 0; + + virtual void TooLong(const Ch* str, SizeType length, SizeType expected) = 0; + virtual void TooShort(const Ch* str, SizeType length, SizeType expected) = 0; + virtual void DoesNotMatch(const Ch* str, SizeType length) = 0; + + virtual void DisallowedItem(SizeType index) = 0; + virtual void TooFewItems(SizeType actualCount, SizeType expectedCount) = 0; + virtual void TooManyItems(SizeType actualCount, SizeType expectedCount) = 0; + virtual void DuplicateItems(SizeType index1, SizeType index2) = 0; + + virtual void TooManyProperties(SizeType actualCount, SizeType expectedCount) = 0; + virtual void TooFewProperties(SizeType actualCount, SizeType expectedCount) = 0; + virtual void StartMissingProperties() = 0; + virtual void AddMissingProperty(const SValue& name) = 0; + virtual bool EndMissingProperties() = 0; + virtual void PropertyViolations(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void DisallowedProperty(const Ch* name, SizeType length) = 0; + + virtual void StartDependencyErrors() = 0; + virtual void StartMissingDependentProperties() = 0; + virtual void AddMissingDependentProperty(const SValue& targetName) = 0; + virtual void EndMissingDependentProperties(const SValue& sourceName) = 0; + virtual void AddDependencySchemaError(const SValue& souceName, ISchemaValidator* subvalidator) = 0; + virtual bool EndDependencyErrors() = 0; + + virtual void DisallowedValue() = 0; + virtual void StartDisallowedType() = 0; + virtual void AddExpectedType(const typename SchemaType::ValueType& expectedType) = 0; + virtual void EndDisallowedType(const typename SchemaType::ValueType& actualType) = 0; + virtual void NotAllOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void NoneOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void NotOneOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void Disallowed() = 0; +}; + + /////////////////////////////////////////////////////////////////////////////// // Hasher @@ -261,6 +316,7 @@ template struct SchemaValidationContext { typedef Schema SchemaType; typedef ISchemaStateFactory SchemaValidatorFactoryType; + typedef IValidationErrorHandler ErrorHandlerType; typedef typename SchemaType::ValueType ValueType; typedef typename ValueType::Ch Ch; @@ -270,8 +326,9 @@ struct SchemaValidationContext { kPatternValidatorWithAdditionalProperty }; - SchemaValidationContext(SchemaValidatorFactoryType& f, const SchemaType* s) : + SchemaValidationContext(SchemaValidatorFactoryType& f, ErrorHandlerType& eh, const SchemaType* s) : factory(f), + error_handler(eh), schema(s), valueSchema(), invalidKeyword(), @@ -311,6 +368,7 @@ struct SchemaValidationContext { } SchemaValidatorFactoryType& factory; + ErrorHandlerType& error_handler; const SchemaType* schema; const SchemaType* valueSchema; const Ch* invalidKeyword; @@ -345,15 +403,20 @@ public: typedef SchemaValidationContext Context; typedef Schema SchemaType; typedef GenericValue SValue; + typedef IValidationErrorHandler ErrorHandler; friend class GenericSchemaDocument; Schema(SchemaDocumentType* schemaDocument, const PointerType& p, const ValueType& value, const ValueType& document, AllocatorType* allocator) : allocator_(allocator), + uri_(schemaDocument->GetURI(), *allocator), + pointer_(p, allocator), + typeless_(schemaDocument->GetTypeless()), enum_(), enumCount_(), not_(), type_((1 << kTotalSchemaType) - 1), // typeless validatorCount_(), + notValidatorIndex_(), properties_(), additionalPropertiesSchema_(), patternProperties_(), @@ -377,7 +440,8 @@ public: minLength_(0), maxLength_(~SizeType(0)), exclusiveMinimum_(false), - exclusiveMaximum_(false) + exclusiveMaximum_(false), + defaultValueLength_(0) { typedef typename SchemaDocumentType::ValueType ValueType; typedef typename ValueType::ConstValueIterator ConstValueIterator; @@ -400,7 +464,7 @@ public: enum_ = static_cast(allocator_->Malloc(sizeof(uint64_t) * v->Size())); for (ConstValueIterator itr = v->Begin(); itr != v->End(); ++itr) { typedef Hasher > EnumHasherType; - char buffer[256 + 24]; + char buffer[256u + 24]; MemoryPoolAllocator<> hasherAllocator(buffer, sizeof(buffer)); EnumHasherType h(&hasherAllocator, 256); itr->Accept(h); @@ -453,7 +517,7 @@ public: for (SizeType i = 0; i < propertyCount_; i++) { new (&properties_[i]) Property(); properties_[i].name = allProperties[i]; - properties_[i].schema = GetTypeless(); + properties_[i].schema = typeless_; } } } @@ -572,12 +636,16 @@ public: if (const ValueType* v = GetMember(value, GetMultipleOfString())) if (v->IsNumber() && v->GetDouble() > 0.0) multipleOf_.CopyFrom(*v, *allocator_); + + // Default + if (const ValueType* v = GetMember(value, GetDefaultValueString())) + if (v->IsString()) + defaultValueLength_ = v->GetStringLength(); + } ~Schema() { - if (allocator_) { - allocator_->Free(enum_); - } + AllocatorType::Free(enum_); if (properties_) { for (SizeType i = 0; i < propertyCount_; i++) properties_[i].~Property(); @@ -592,11 +660,19 @@ public: #if CEREAL_RAPIDJSON_SCHEMA_HAS_REGEX if (pattern_) { pattern_->~RegexType(); - allocator_->Free(pattern_); + AllocatorType::Free(pattern_); } #endif } + const SValue& GetURI() const { + return uri_; + } + + const PointerType& GetPointer() const { + return pointer_; + } + bool BeginValue(Context& context) const { if (context.inArray) { if (uniqueItems_) @@ -610,12 +686,14 @@ public: else if (additionalItemsSchema_) context.valueSchema = additionalItemsSchema_; else if (additionalItems_) - context.valueSchema = GetTypeless(); - else + context.valueSchema = typeless_; + else { + context.error_handler.DisallowedItem(context.arrayElementIndex); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetItemsString()); + } } else - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.arrayElementIndex++; } @@ -637,15 +715,21 @@ public: } if (context.objectPatternValidatorType == Context::kPatternValidatorOnly) { - if (!patternValid) + if (!patternValid) { + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } else if (context.objectPatternValidatorType == Context::kPatternValidatorWithProperty) { - if (!patternValid || !otherValid) + if (!patternValid || !otherValid) { + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count + 1); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } - else if (!patternValid && !otherValid) // kPatternValidatorWithAdditionalProperty) + else if (!patternValid && !otherValid) { // kPatternValidatorWithAdditionalProperty) + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count + 1); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } if (enum_) { @@ -653,19 +737,23 @@ public: for (SizeType i = 0; i < enumCount_; i++) if (enum_[i] == h) goto foundEnum; + context.error_handler.DisallowedValue(); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetEnumString()); foundEnum:; } if (allOf_.schemas) for (SizeType i = allOf_.begin; i < allOf_.begin + allOf_.count; i++) - if (!context.validators[i]->IsValid()) + if (!context.validators[i]->IsValid()) { + context.error_handler.NotAllOf(&context.validators[allOf_.begin], allOf_.count); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetAllOfString()); + } if (anyOf_.schemas) { for (SizeType i = anyOf_.begin; i < anyOf_.begin + anyOf_.count; i++) if (context.validators[i]->IsValid()) goto foundAny; + context.error_handler.NoneOf(&context.validators[anyOf_.begin], anyOf_.count); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetAnyOfString()); foundAny:; } @@ -674,30 +762,39 @@ public: bool oneValid = false; for (SizeType i = oneOf_.begin; i < oneOf_.begin + oneOf_.count; i++) if (context.validators[i]->IsValid()) { - if (oneValid) + if (oneValid) { + context.error_handler.NotOneOf(&context.validators[oneOf_.begin], oneOf_.count); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); - else + } else oneValid = true; } - if (!oneValid) + if (!oneValid) { + context.error_handler.NotOneOf(&context.validators[oneOf_.begin], oneOf_.count); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); + } } - if (not_ && context.validators[notValidatorIndex_]->IsValid()) + if (not_ && context.validators[notValidatorIndex_]->IsValid()) { + context.error_handler.Disallowed(); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetNotString()); + } return true; } - bool Null(Context& context) const { - if (!(type_ & (1 << kNullSchemaType))) + bool Null(Context& context) const { + if (!(type_ & (1 << kNullSchemaType))) { + DisallowedType(context, GetNullString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } return CreateParallelValidator(context); } - bool Bool(Context& context, bool) const { - if (!(type_ & (1 << kBooleanSchemaType))) + bool Bool(Context& context, bool) const { + if (!(type_ & (1 << kBooleanSchemaType))) { + DisallowedType(context, GetBooleanString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } return CreateParallelValidator(context); } @@ -726,8 +823,10 @@ public: } bool Double(Context& context, double d) const { - if (!(type_ & (1 << kNumberSchemaType))) + if (!(type_ & (1 << kNumberSchemaType))) { + DisallowedType(context, GetNumberString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull() && !CheckDoubleMinimum(context, d)) return false; @@ -742,28 +841,38 @@ public: } bool String(Context& context, const Ch* str, SizeType length, bool) const { - if (!(type_ & (1 << kStringSchemaType))) + if (!(type_ & (1 << kStringSchemaType))) { + DisallowedType(context, GetStringString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (minLength_ != 0 || maxLength_ != SizeType(~0)) { SizeType count; if (internal::CountStringCodePoint(str, length, &count)) { - if (count < minLength_) + if (count < minLength_) { + context.error_handler.TooShort(str, length, minLength_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinLengthString()); - if (count > maxLength_) + } + if (count > maxLength_) { + context.error_handler.TooLong(str, length, maxLength_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxLengthString()); + } } } - if (pattern_ && !IsPatternMatch(pattern_, str, length)) + if (pattern_ && !IsPatternMatch(pattern_, str, length)) { + context.error_handler.DoesNotMatch(str, length); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternString()); + } return CreateParallelValidator(context); } - bool StartObject(Context& context) const { - if (!(type_ & (1 << kObjectSchemaType))) + bool StartObject(Context& context) const { + if (!(type_ & (1 << kObjectSchemaType))) { + DisallowedType(context, GetObjectString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (hasDependencies_ || hasRequired_) { context.propertyExist = static_cast(context.factory.MallocState(sizeof(bool) * propertyCount_)); @@ -784,15 +893,17 @@ public: if (patternProperties_) { context.patternPropertiesSchemaCount = 0; for (SizeType i = 0; i < patternPropertyCount_; i++) - if (patternProperties_[i].pattern && IsPatternMatch(patternProperties_[i].pattern, str, len)) + if (patternProperties_[i].pattern && IsPatternMatch(patternProperties_[i].pattern, str, len)) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = patternProperties_[i].schema; + context.valueSchema = typeless_; + } } SizeType index; if (FindPropertyIndex(ValueType(str, len).Move(), &index)) { if (context.patternPropertiesSchemaCount > 0) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = properties_[index].schema; - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.valuePatternValidatorType = Context::kPatternValidatorWithProperty; } else @@ -807,7 +918,7 @@ public: if (additionalPropertiesSchema_) { if (additionalPropertiesSchema_ && context.patternPropertiesSchemaCount > 0) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = additionalPropertiesSchema_; - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.valuePatternValidatorType = Context::kPatternValidatorWithAdditionalProperty; } else @@ -815,49 +926,70 @@ public: return true; } else if (additionalProperties_) { - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; return true; } - if (context.patternPropertiesSchemaCount == 0) // patternProperties are not additional properties + if (context.patternPropertiesSchemaCount == 0) { // patternProperties are not additional properties + context.error_handler.DisallowedProperty(str, len); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetAdditionalPropertiesString()); + } return true; } bool EndObject(Context& context, SizeType memberCount) const { - if (hasRequired_) + if (hasRequired_) { + context.error_handler.StartMissingProperties(); for (SizeType index = 0; index < propertyCount_; index++) - if (properties_[index].required) - if (!context.propertyExist[index]) - CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetRequiredString()); + if (properties_[index].required && !context.propertyExist[index]) + if (properties_[index].schema->defaultValueLength_ == 0 ) + context.error_handler.AddMissingProperty(properties_[index].name); + if (context.error_handler.EndMissingProperties()) + CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetRequiredString()); + } - if (memberCount < minProperties_) + if (memberCount < minProperties_) { + context.error_handler.TooFewProperties(memberCount, minProperties_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinPropertiesString()); + } - if (memberCount > maxProperties_) + if (memberCount > maxProperties_) { + context.error_handler.TooManyProperties(memberCount, maxProperties_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxPropertiesString()); + } if (hasDependencies_) { - for (SizeType sourceIndex = 0; sourceIndex < propertyCount_; sourceIndex++) + context.error_handler.StartDependencyErrors(); + for (SizeType sourceIndex = 0; sourceIndex < propertyCount_; sourceIndex++) { + const Property& source = properties_[sourceIndex]; if (context.propertyExist[sourceIndex]) { - if (properties_[sourceIndex].dependencies) { + if (source.dependencies) { + context.error_handler.StartMissingDependentProperties(); for (SizeType targetIndex = 0; targetIndex < propertyCount_; targetIndex++) - if (properties_[sourceIndex].dependencies[targetIndex] && !context.propertyExist[targetIndex]) - CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); + if (source.dependencies[targetIndex] && !context.propertyExist[targetIndex]) + context.error_handler.AddMissingDependentProperty(properties_[targetIndex].name); + context.error_handler.EndMissingDependentProperties(source.name); + } + else if (source.dependenciesSchema) { + ISchemaValidator* dependenciesValidator = context.validators[source.dependenciesValidatorIndex]; + if (!dependenciesValidator->IsValid()) + context.error_handler.AddDependencySchemaError(source.name, dependenciesValidator); } - else if (properties_[sourceIndex].dependenciesSchema) - if (!context.validators[properties_[sourceIndex].dependenciesValidatorIndex]->IsValid()) - CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); } + } + if (context.error_handler.EndDependencyErrors()) + CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); } return true; } - bool StartArray(Context& context) const { - if (!(type_ & (1 << kArraySchemaType))) + bool StartArray(Context& context) const { + if (!(type_ & (1 << kArraySchemaType))) { + DisallowedType(context, GetArrayString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } context.arrayElementIndex = 0; context.inArray = true; @@ -865,14 +997,18 @@ public: return CreateParallelValidator(context); } - bool EndArray(Context& context, SizeType elementCount) const { + bool EndArray(Context& context, SizeType elementCount) const { context.inArray = false; - if (elementCount < minItems_) + if (elementCount < minItems_) { + context.error_handler.TooFewItems(elementCount, minItems_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinItemsString()); + } - if (elementCount > maxItems_) + if (elementCount > maxItems_) { + context.error_handler.TooManyItems(elementCount, maxItems_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxItemsString()); + } return true; } @@ -881,7 +1017,7 @@ public: #define CEREAL_RAPIDJSON_STRING_(name, ...) \ static const ValueType& Get##name##String() {\ static const Ch s[] = { __VA_ARGS__, '\0' };\ - static const ValueType v(s, sizeof(s) / sizeof(Ch) - 1);\ + static const ValueType v(s, static_cast(sizeof(s) / sizeof(Ch) - 1));\ return v;\ } @@ -918,6 +1054,7 @@ public: CEREAL_RAPIDJSON_STRING_(ExclusiveMinimum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'i', 'n', 'i', 'm', 'u', 'm') CEREAL_RAPIDJSON_STRING_(ExclusiveMaximum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'a', 'x', 'i', 'm', 'u', 'm') CEREAL_RAPIDJSON_STRING_(MultipleOf, 'm', 'u', 'l', 't', 'i', 'p', 'l', 'e', 'O', 'f') + CEREAL_RAPIDJSON_STRING_(DefaultValue, 'd', 'e', 'f', 'a', 'u', 'l', 't') #undef CEREAL_RAPIDJSON_STRING_ @@ -934,7 +1071,7 @@ private: }; #if CEREAL_RAPIDJSON_SCHEMA_USE_INTERNALREGEX - typedef internal::GenericRegex RegexType; + typedef internal::GenericRegex RegexType; #elif CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX typedef std::basic_regex RegexType; #else @@ -949,11 +1086,6 @@ private: SizeType count; }; - static const SchemaType* GetTypeless() { - static SchemaType typeless(0, PointerType(), ValueType(kObjectType).Move(), ValueType(kObjectType).Move(), 0); - return &typeless; - } - template void AddUniqueElement(V1& a, const V2& v) { for (typename V1::ConstValueIterator itr = a.Begin(); itr != a.End(); ++itr) @@ -999,7 +1131,7 @@ private: template RegexType* CreatePattern(const ValueType& value) { if (value.IsString()) { - RegexType* r = new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString()); + RegexType* r = new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString(), allocator_); if (!r->IsValid()) { r->~RegexType(); AllocatorType::Free(r); @@ -1011,17 +1143,21 @@ private: } static bool IsPatternMatch(const RegexType* pattern, const Ch *str, SizeType) { - return pattern->Search(str); + GenericRegexSearch rs(*pattern); + return rs.Search(str); } #elif CEREAL_RAPIDJSON_SCHEMA_USE_STDREGEX template RegexType* CreatePattern(const ValueType& value) { - if (value.IsString()) + if (value.IsString()) { + RegexType *r = static_cast(allocator_->Malloc(sizeof(RegexType))); try { - return new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString(), std::size_t(value.GetStringLength()), std::regex_constants::ECMAScript); + return new (r) RegexType(value.GetString(), std::size_t(value.GetStringLength()), std::regex_constants::ECMAScript); } catch (const std::regex_error&) { + AllocatorType::Free(r); } + } return 0; } @@ -1097,15 +1233,20 @@ private: } bool CheckInt(Context& context, int64_t i) const { - if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) { + DisallowedType(context, GetIntegerString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull()) { if (minimum_.IsInt64()) { - if (exclusiveMinimum_ ? i <= minimum_.GetInt64() : i < minimum_.GetInt64()) + if (exclusiveMinimum_ ? i <= minimum_.GetInt64() : i < minimum_.GetInt64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } } else if (minimum_.IsUint64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); // i <= max(int64_t) < minimum.GetUint64() } else if (!CheckDoubleMinimum(context, static_cast(i))) @@ -1114,19 +1255,23 @@ private: if (!maximum_.IsNull()) { if (maximum_.IsInt64()) { - if (exclusiveMaximum_ ? i >= maximum_.GetInt64() : i > maximum_.GetInt64()) + if (exclusiveMaximum_ ? i >= maximum_.GetInt64() : i > maximum_.GetInt64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } } - else if (maximum_.IsUint64()) - /* do nothing */; // i <= max(int64_t) < maximum_.GetUint64() + else if (maximum_.IsUint64()) { } + /* do nothing */ // i <= max(int64_t) < maximum_.GetUint64() else if (!CheckDoubleMaximum(context, static_cast(i))) return false; } if (!multipleOf_.IsNull()) { if (multipleOf_.IsUint64()) { - if (static_cast(i >= 0 ? i : -i) % multipleOf_.GetUint64() != 0) + if (static_cast(i >= 0 ? i : -i) % multipleOf_.GetUint64() != 0) { + context.error_handler.NotMultipleOf(i, multipleOf_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } } else if (!CheckDoubleMultipleOf(context, static_cast(i))) return false; @@ -1136,13 +1281,17 @@ private: } bool CheckUint(Context& context, uint64_t i) const { - if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) { + DisallowedType(context, GetIntegerString()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull()) { if (minimum_.IsUint64()) { - if (exclusiveMinimum_ ? i <= minimum_.GetUint64() : i < minimum_.GetUint64()) + if (exclusiveMinimum_ ? i <= minimum_.GetUint64() : i < minimum_.GetUint64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } } else if (minimum_.IsInt64()) /* do nothing */; // i >= 0 > minimum.Getint64() @@ -1152,19 +1301,25 @@ private: if (!maximum_.IsNull()) { if (maximum_.IsUint64()) { - if (exclusiveMaximum_ ? i >= maximum_.GetUint64() : i > maximum_.GetUint64()) + if (exclusiveMaximum_ ? i >= maximum_.GetUint64() : i > maximum_.GetUint64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } } - else if (maximum_.IsInt64()) + else if (maximum_.IsInt64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); // i >= 0 > maximum_ + } else if (!CheckDoubleMaximum(context, static_cast(i))) return false; } if (!multipleOf_.IsNull()) { if (multipleOf_.IsUint64()) { - if (i % multipleOf_.GetUint64() != 0) + if (i % multipleOf_.GetUint64() != 0) { + context.error_handler.NotMultipleOf(i, multipleOf_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } } else if (!CheckDoubleMultipleOf(context, static_cast(i))) return false; @@ -1174,14 +1329,18 @@ private: } bool CheckDoubleMinimum(Context& context, double d) const { - if (exclusiveMinimum_ ? d <= minimum_.GetDouble() : d < minimum_.GetDouble()) + if (exclusiveMinimum_ ? d <= minimum_.GetDouble() : d < minimum_.GetDouble()) { + context.error_handler.BelowMinimum(d, minimum_, exclusiveMinimum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } return true; } bool CheckDoubleMaximum(Context& context, double d) const { - if (exclusiveMaximum_ ? d >= maximum_.GetDouble() : d > maximum_.GetDouble()) + if (exclusiveMaximum_ ? d >= maximum_.GetDouble() : d > maximum_.GetDouble()) { + context.error_handler.AboveMaximum(d, maximum_, exclusiveMaximum_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } return true; } @@ -1189,11 +1348,29 @@ private: double a = std::abs(d), b = std::abs(multipleOf_.GetDouble()); double q = std::floor(a / b); double r = a - q * b; - if (r > 0.0) + if (r > 0.0) { + context.error_handler.NotMultipleOf(d, multipleOf_); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } return true; } + void DisallowedType(Context& context, const ValueType& actualType) const { + ErrorHandler& eh = context.error_handler; + eh.StartDisallowedType(); + + if (type_ & (1 << kNullSchemaType)) eh.AddExpectedType(GetNullString()); + if (type_ & (1 << kBooleanSchemaType)) eh.AddExpectedType(GetBooleanString()); + if (type_ & (1 << kObjectSchemaType)) eh.AddExpectedType(GetObjectString()); + if (type_ & (1 << kArraySchemaType)) eh.AddExpectedType(GetArrayString()); + if (type_ & (1 << kStringSchemaType)) eh.AddExpectedType(GetStringString()); + + if (type_ & (1 << kNumberSchemaType)) eh.AddExpectedType(GetNumberString()); + else if (type_ & (1 << kIntegerSchemaType)) eh.AddExpectedType(GetIntegerString()); + + eh.EndDisallowedType(actualType); + } + struct Property { Property() : schema(), dependenciesSchema(), dependenciesValidatorIndex(), dependencies(), required(false) {} ~Property() { AllocatorType::Free(dependencies); } @@ -1218,6 +1395,9 @@ private: }; AllocatorType* allocator_; + SValue uri_; + PointerType pointer_; + const SchemaType* typeless_; uint64_t* enum_; SizeType enumCount_; SchemaArray allOf_; @@ -1258,6 +1438,8 @@ private: SValue multipleOf_; bool exclusiveMinimum_; bool exclusiveMaximum_; + + SizeType defaultValueLength_; }; template @@ -1267,7 +1449,7 @@ struct TokenHelper { char buffer[21]; size_t length = static_cast((sizeof(SizeType) == 4 ? u32toa(index, buffer) : u64toa(index, buffer)) - buffer); for (size_t i = 0; i < length; i++) - *documentStack.template Push() = buffer[i]; + *documentStack.template Push() = static_cast(buffer[i]); } }; @@ -1326,6 +1508,7 @@ public: typedef typename EncodingType::Ch Ch; typedef internal::Schema SchemaType; typedef GenericPointer PointerType; + typedef GenericValue URIType; friend class internal::Schema; template friend class GenericSchemaValidator; @@ -1335,19 +1518,29 @@ public: Compile a JSON document into schema document. \param document A JSON document as source. + \param uri The base URI of this schema document for purposes of violation reporting. + \param uriLength Length of \c name, in code points. \param remoteProvider An optional remote schema document provider for resolving remote reference. Can be null. \param allocator An optional allocator instance for allocating memory. Can be null. */ - explicit GenericSchemaDocument(const ValueType& document, IRemoteSchemaDocumentProviderType* remoteProvider = 0, Allocator* allocator = 0) : + explicit GenericSchemaDocument(const ValueType& document, const Ch* uri = 0, SizeType uriLength = 0, + IRemoteSchemaDocumentProviderType* remoteProvider = 0, Allocator* allocator = 0) : remoteProvider_(remoteProvider), allocator_(allocator), ownAllocator_(), root_(), + typeless_(), schemaMap_(allocator, kInitialSchemaMapSize), schemaRef_(allocator, kInitialSchemaRefSize) { if (!allocator_) - ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = CEREAL_RAPIDJSON_NEW(Allocator)(); + + Ch noUri[1] = {0}; + uri_.SetString(uri ? uri : noUri, uriLength, *allocator_); + + typeless_ = static_cast(allocator_->Malloc(sizeof(SchemaType))); + new (typeless_) SchemaType(this, PointerType(), ValueType(kObjectType).Move(), ValueType(kObjectType).Move(), allocator_); // Generate root schema, it will call CreateSchema() to create sub-schemas, // And call AddRefSchema() if there are $ref. @@ -1365,6 +1558,9 @@ public: new (schemaMap_.template Push()) SchemaEntry(refEntry->source, const_cast(s), false, allocator_); } } + else if (refEntry->schema) + *refEntry->schema = typeless_; + refEntry->~SchemaRefEntry(); } @@ -1380,12 +1576,15 @@ public: allocator_(rhs.allocator_), ownAllocator_(rhs.ownAllocator_), root_(rhs.root_), + typeless_(rhs.typeless_), schemaMap_(std::move(rhs.schemaMap_)), - schemaRef_(std::move(rhs.schemaRef_)) + schemaRef_(std::move(rhs.schemaRef_)), + uri_(std::move(rhs.uri_)) { rhs.remoteProvider_ = 0; rhs.allocator_ = 0; rhs.ownAllocator_ = 0; + rhs.typeless_ = 0; } #endif @@ -1394,9 +1593,16 @@ public: while (!schemaMap_.Empty()) schemaMap_.template Pop(1)->~SchemaEntry(); + if (typeless_) { + typeless_->~SchemaType(); + Allocator::Free(typeless_); + } + CEREAL_RAPIDJSON_DELETE(ownAllocator_); } + const URIType& GetURI() const { return uri_; } + //! Get the root schema. const SchemaType& GetRoot() const { return *root_; } @@ -1428,7 +1634,7 @@ private: void CreateSchemaRecursive(const SchemaType** schema, const PointerType& pointer, const ValueType& v, const ValueType& document) { if (schema) - *schema = SchemaType::GetTypeless(); + *schema = typeless_; if (v.GetType() == kObjectType) { const SchemaType* s = GetSchema(pointer); @@ -1473,12 +1679,13 @@ private: if (i > 0) { // Remote reference, resolve immediately if (remoteProvider_) { - if (const GenericSchemaDocument* remoteDocument = remoteProvider_->GetRemoteDocument(s, i - 1)) { + if (const GenericSchemaDocument* remoteDocument = remoteProvider_->GetRemoteDocument(s, i)) { PointerType pointer(&s[i], len - i, allocator_); if (pointer.IsValid()) { if (const SchemaType* sc = remoteDocument->GetSchema(pointer)) { if (schema) *schema = sc; + new (schemaMap_.template Push()) SchemaEntry(source, const_cast(sc), false, allocator_); return true; } } @@ -1515,6 +1722,8 @@ private: return PointerType(); } + const SchemaType* GetTypeless() const { return typeless_; } + static const size_t kInitialSchemaMapSize = 64; static const size_t kInitialSchemaRefSize = 64; @@ -1522,8 +1731,10 @@ private: Allocator *allocator_; Allocator *ownAllocator_; const SchemaType* root_; //!< Root schema. + SchemaType* typeless_; internal::Stack schemaMap_; // Stores created Pointer -> Schemas internal::Stack schemaRef_; // Stores Pointer from $ref and schema which holds the $ref + URIType uri_; }; //! GenericSchemaDocument using Value type. @@ -1552,13 +1763,17 @@ template < typename StateAllocator = CrtAllocator> class GenericSchemaValidator : public internal::ISchemaStateFactory, - public internal::ISchemaValidator + public internal::ISchemaValidator, + public internal::IValidationErrorHandler { public: typedef typename SchemaDocumentType::SchemaType SchemaType; typedef typename SchemaDocumentType::PointerType PointerType; typedef typename SchemaType::EncodingType EncodingType; + typedef typename SchemaType::SValue SValue; typedef typename EncodingType::Ch Ch; + typedef GenericStringRef StringRefType; + typedef GenericValue ValueType; //! Constructor without output handler. /*! @@ -1575,11 +1790,14 @@ public: : schemaDocument_(&schemaDocument), root_(schemaDocument.GetRoot()), - outputHandler_(GetNullHandler()), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE , depth_(0) @@ -1603,11 +1821,14 @@ public: : schemaDocument_(&schemaDocument), root_(schemaDocument.GetRoot()), - outputHandler_(outputHandler), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(&outputHandler), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE , depth_(0) @@ -1626,6 +1847,9 @@ public: while (!schemaStack_.Empty()) PopSchema(); documentStack_.Clear(); + error_.SetObject(); + currentError_.SetNull(); + missingDependents_.SetNull(); valid_ = true; } @@ -1633,9 +1857,13 @@ public: // Implementation of ISchemaValidator virtual bool IsValid() const { return valid_; } + //! Gets the error object. + ValueType& GetError() { return error_; } + const ValueType& GetError() const { return error_; } + //! Gets the JSON pointer pointed to the invalid schema. PointerType GetInvalidSchemaPointer() const { - return schemaStack_.Empty() ? PointerType() : schemaDocument_->GetPointer(&CurrentSchema()); + return schemaStack_.Empty() ? PointerType() : CurrentSchema().GetPointer(); } //! Gets the keyword of invalid schema. @@ -1645,9 +1873,196 @@ public: //! Gets the JSON pointer pointed to the invalid value. PointerType GetInvalidDocumentPointer() const { - return documentStack_.Empty() ? PointerType() : PointerType(documentStack_.template Bottom(), documentStack_.GetSize() / sizeof(Ch)); + if (documentStack_.Empty()) { + return PointerType(); + } + else { + return PointerType(documentStack_.template Bottom(), documentStack_.GetSize() / sizeof(Ch)); + } } + void NotMultipleOf(int64_t actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void NotMultipleOf(uint64_t actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void NotMultipleOf(double actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void AboveMaximum(int64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void AboveMaximum(uint64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void AboveMaximum(double actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void BelowMinimum(int64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + void BelowMinimum(uint64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + void BelowMinimum(double actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + + void TooLong(const Ch* str, SizeType length, SizeType expected) { + AddNumberError(SchemaType::GetMaxLengthString(), + ValueType(str, length, GetStateAllocator()).Move(), SValue(expected).Move()); + } + void TooShort(const Ch* str, SizeType length, SizeType expected) { + AddNumberError(SchemaType::GetMinLengthString(), + ValueType(str, length, GetStateAllocator()).Move(), SValue(expected).Move()); + } + void DoesNotMatch(const Ch* str, SizeType length) { + currentError_.SetObject(); + currentError_.AddMember(GetActualString(), ValueType(str, length, GetStateAllocator()).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetPatternString()); + } + + void DisallowedItem(SizeType index) { + currentError_.SetObject(); + currentError_.AddMember(GetDisallowedString(), ValueType(index).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetAdditionalItemsString(), true); + } + void TooFewItems(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMinItemsString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void TooManyItems(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMaxItemsString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void DuplicateItems(SizeType index1, SizeType index2) { + ValueType duplicates(kArrayType); + duplicates.PushBack(index1, GetStateAllocator()); + duplicates.PushBack(index2, GetStateAllocator()); + currentError_.SetObject(); + currentError_.AddMember(GetDuplicatesString(), duplicates, GetStateAllocator()); + AddCurrentError(SchemaType::GetUniqueItemsString(), true); + } + + void TooManyProperties(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMaxPropertiesString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void TooFewProperties(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMinPropertiesString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void StartMissingProperties() { + currentError_.SetArray(); + } + void AddMissingProperty(const SValue& name) { + currentError_.PushBack(ValueType(name, GetStateAllocator()).Move(), GetStateAllocator()); + } + bool EndMissingProperties() { + if (currentError_.Empty()) + return false; + ValueType error(kObjectType); + error.AddMember(GetMissingString(), currentError_, GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetRequiredString()); + return true; + } + void PropertyViolations(ISchemaValidator** subvalidators, SizeType count) { + for (SizeType i = 0; i < count; ++i) + MergeError(static_cast(subvalidators[i])->GetError()); + } + void DisallowedProperty(const Ch* name, SizeType length) { + currentError_.SetObject(); + currentError_.AddMember(GetDisallowedString(), ValueType(name, length, GetStateAllocator()).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetAdditionalPropertiesString(), true); + } + + void StartDependencyErrors() { + currentError_.SetObject(); + } + void StartMissingDependentProperties() { + missingDependents_.SetArray(); + } + void AddMissingDependentProperty(const SValue& targetName) { + missingDependents_.PushBack(ValueType(targetName, GetStateAllocator()).Move(), GetStateAllocator()); + } + void EndMissingDependentProperties(const SValue& sourceName) { + if (!missingDependents_.Empty()) + currentError_.AddMember(ValueType(sourceName, GetStateAllocator()).Move(), + missingDependents_, GetStateAllocator()); + } + void AddDependencySchemaError(const SValue& sourceName, ISchemaValidator* subvalidator) { + currentError_.AddMember(ValueType(sourceName, GetStateAllocator()).Move(), + static_cast(subvalidator)->GetError(), GetStateAllocator()); + } + bool EndDependencyErrors() { + if (currentError_.ObjectEmpty()) + return false; + ValueType error(kObjectType); + error.AddMember(GetErrorsString(), currentError_, GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetDependenciesString()); + return true; + } + + void DisallowedValue() { + currentError_.SetObject(); + AddCurrentError(SchemaType::GetEnumString()); + } + void StartDisallowedType() { + currentError_.SetArray(); + } + void AddExpectedType(const typename SchemaType::ValueType& expectedType) { + currentError_.PushBack(ValueType(expectedType, GetStateAllocator()).Move(), GetStateAllocator()); + } + void EndDisallowedType(const typename SchemaType::ValueType& actualType) { + ValueType error(kObjectType); + error.AddMember(GetExpectedString(), currentError_, GetStateAllocator()); + error.AddMember(GetActualString(), ValueType(actualType, GetStateAllocator()).Move(), GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetTypeString()); + } + void NotAllOf(ISchemaValidator** subvalidators, SizeType count) { + for (SizeType i = 0; i < count; ++i) { + MergeError(static_cast(subvalidators[i])->GetError()); + } + } + void NoneOf(ISchemaValidator** subvalidators, SizeType count) { + AddErrorArray(SchemaType::GetAnyOfString(), subvalidators, count); + } + void NotOneOf(ISchemaValidator** subvalidators, SizeType count) { + AddErrorArray(SchemaType::GetOneOfString(), subvalidators, count); + } + void Disallowed() { + currentError_.SetObject(); + AddCurrentError(SchemaType::GetNotString()); + } + +#define CEREAL_RAPIDJSON_STRING_(name, ...) \ + static const StringRefType& Get##name##String() {\ + static const Ch s[] = { __VA_ARGS__, '\0' };\ + static const StringRefType v(s, static_cast(sizeof(s) / sizeof(Ch) - 1)); \ + return v;\ + } + + CEREAL_RAPIDJSON_STRING_(InstanceRef, 'i', 'n', 's', 't', 'a', 'n', 'c', 'e', 'R', 'e', 'f') + CEREAL_RAPIDJSON_STRING_(SchemaRef, 's', 'c', 'h', 'e', 'm', 'a', 'R', 'e', 'f') + CEREAL_RAPIDJSON_STRING_(Expected, 'e', 'x', 'p', 'e', 'c', 't', 'e', 'd') + CEREAL_RAPIDJSON_STRING_(Actual, 'a', 'c', 't', 'u', 'a', 'l') + CEREAL_RAPIDJSON_STRING_(Disallowed, 'd', 'i', 's', 'a', 'l', 'l', 'o', 'w', 'e', 'd') + CEREAL_RAPIDJSON_STRING_(Missing, 'm', 'i', 's', 's', 'i', 'n', 'g') + CEREAL_RAPIDJSON_STRING_(Errors, 'e', 'r', 'r', 'o', 'r', 's') + CEREAL_RAPIDJSON_STRING_(Duplicates, 'd', 'u', 'p', 'l', 'i', 'c', 'a', 't', 'e', 's') + +#undef CEREAL_RAPIDJSON_STRING_ + #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE #define CEREAL_RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_() \ CEREAL_RAPIDJSON_MULTILINEMACRO_BEGIN\ @@ -1679,14 +2094,14 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END } #define CEREAL_RAPIDJSON_SCHEMA_HANDLE_END_(method, arg2)\ - return valid_ = EndValue() && outputHandler_.method arg2 + return valid_ = EndValue() && (!outputHandler_ || outputHandler_->method arg2) #define CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(method, arg1, arg2) \ CEREAL_RAPIDJSON_SCHEMA_HANDLE_BEGIN_ (method, arg1);\ CEREAL_RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(method, arg2);\ CEREAL_RAPIDJSON_SCHEMA_HANDLE_END_ (method, arg2) - bool Null() { CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(Null, (CurrentContext() ), ( )); } + bool Null() { CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(Null, (CurrentContext()), ( )); } bool Bool(bool b) { CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(Bool, (CurrentContext(), b), (b)); } bool Int(int i) { CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(Int, (CurrentContext(), i), (i)); } bool Uint(unsigned u) { CEREAL_RAPIDJSON_SCHEMA_HANDLE_VALUE_(Uint, (CurrentContext(), u), (u)); } @@ -1701,7 +2116,7 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END bool StartObject() { CEREAL_RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartObject, (CurrentContext())); CEREAL_RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartObject, ()); - return valid_ = outputHandler_.StartObject(); + return valid_ = !outputHandler_ || outputHandler_->StartObject(); } bool Key(const Ch* str, SizeType len, bool copy) { @@ -1709,7 +2124,7 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END AppendToken(str, len); if (!CurrentSchema().Key(CurrentContext(), str, len, copy)) return valid_ = false; CEREAL_RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(Key, (str, len, copy)); - return valid_ = outputHandler_.Key(str, len, copy); + return valid_ = !outputHandler_ || outputHandler_->Key(str, len, copy); } bool EndObject(SizeType memberCount) { @@ -1722,7 +2137,7 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END bool StartArray() { CEREAL_RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartArray, (CurrentContext())); CEREAL_RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartArray, ()); - return valid_ = outputHandler_.StartArray(); + return valid_ = !outputHandler_ || outputHandler_->StartArray(); } bool EndArray(SizeType elementCount) { @@ -1739,7 +2154,7 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END // Implementation of ISchemaStateFactory virtual ISchemaValidator* CreateSchemaValidator(const SchemaType& root) { - return new (GetStateAllocator().Malloc(sizeof(GenericSchemaValidator))) GenericSchemaValidator(*schemaDocument_, root, + return new (GetStateAllocator().Malloc(sizeof(GenericSchemaValidator))) GenericSchemaValidator(*schemaDocument_, root, documentStack_.template Bottom(), documentStack_.GetSize(), #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE depth_ + 1, #endif @@ -1771,7 +2186,7 @@ CEREAL_RAPIDJSON_MULTILINEMACRO_END } virtual void FreeState(void* p) { - return StateAllocator::Free(p); + StateAllocator::Free(p); } private: @@ -1782,6 +2197,7 @@ private: GenericSchemaValidator( const SchemaDocumentType& schemaDocument, const SchemaType& root, + const char* basePath, size_t basePathSize, #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE unsigned depth, #endif @@ -1791,21 +2207,26 @@ private: : schemaDocument_(&schemaDocument), root_(root), - outputHandler_(GetNullHandler()), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE , depth_(depth) #endif { + if (basePath && basePathSize) + memcpy(documentStack_.template Push(basePathSize), basePath, basePathSize); } StateAllocator& GetStateAllocator() { if (!stateAllocator_) - stateAllocator_ = ownStateAllocator_ = CEREAL_RAPIDJSON_NEW(StateAllocator()); + stateAllocator_ = ownStateAllocator_ = CEREAL_RAPIDJSON_NEW(StateAllocator)(); return *stateAllocator_; } @@ -1823,8 +2244,8 @@ private: const SchemaType** sa = CurrentContext().patternPropertiesSchemas; typename Context::PatternValidatorType patternValidatorType = CurrentContext().valuePatternValidatorType; bool valueUniqueness = CurrentContext().valueUniqueness; - if (CurrentContext().valueSchema) - PushSchema(*CurrentContext().valueSchema); + CEREAL_RAPIDJSON_ASSERT(CurrentContext().valueSchema); + PushSchema(*CurrentContext().valueSchema); if (count > 0) { CurrentContext().objectPatternValidatorType = patternValidatorType; @@ -1864,8 +2285,10 @@ private: if (!a) CurrentContext().arrayElementHashCodes = a = new (GetStateAllocator().Malloc(sizeof(HashCodeArray))) HashCodeArray(kArrayType); for (typename HashCodeArray::ConstValueIterator itr = a->Begin(); itr != a->End(); ++itr) - if (itr->GetUint64() == h) + if (itr->GetUint64() == h) { + DuplicateItems(static_cast(itr - a->Begin()), a->Size()); CEREAL_RAPIDJSON_INVALID_KEYWORD_RETURN(SchemaType::GetUniqueItemsString()); + } a->PushBack(h, GetStateAllocator()); } } @@ -1894,7 +2317,7 @@ private: } } - CEREAL_RAPIDJSON_FORCEINLINE void PushSchema(const SchemaType& schema) { new (schemaStack_.template Push()) Context(*this, &schema); } + CEREAL_RAPIDJSON_FORCEINLINE void PushSchema(const SchemaType& schema) { new (schemaStack_.template Push()) Context(*this, *this, &schema); } CEREAL_RAPIDJSON_FORCEINLINE void PopSchema() { Context* c = schemaStack_.template Pop(1); @@ -1905,24 +2328,86 @@ private: c->~Context(); } + void AddErrorLocation(ValueType& result, bool parent) { + GenericStringBuffer sb; + PointerType instancePointer = GetInvalidDocumentPointer(); + ((parent && instancePointer.GetTokenCount() > 0) + ? PointerType(instancePointer.GetTokens(), instancePointer.GetTokenCount() - 1) + : instancePointer).StringifyUriFragment(sb); + ValueType instanceRef(sb.GetString(), static_cast(sb.GetSize() / sizeof(Ch)), + GetStateAllocator()); + result.AddMember(GetInstanceRefString(), instanceRef, GetStateAllocator()); + sb.Clear(); + memcpy(sb.Push(CurrentSchema().GetURI().GetStringLength()), + CurrentSchema().GetURI().GetString(), + CurrentSchema().GetURI().GetStringLength() * sizeof(Ch)); + GetInvalidSchemaPointer().StringifyUriFragment(sb); + ValueType schemaRef(sb.GetString(), static_cast(sb.GetSize() / sizeof(Ch)), + GetStateAllocator()); + result.AddMember(GetSchemaRefString(), schemaRef, GetStateAllocator()); + } + + void AddError(ValueType& keyword, ValueType& error) { + typename ValueType::MemberIterator member = error_.FindMember(keyword); + if (member == error_.MemberEnd()) + error_.AddMember(keyword, error, GetStateAllocator()); + else { + if (member->value.IsObject()) { + ValueType errors(kArrayType); + errors.PushBack(member->value, GetStateAllocator()); + member->value = errors; + } + member->value.PushBack(error, GetStateAllocator()); + } + } + + void AddCurrentError(const typename SchemaType::ValueType& keyword, bool parent = false) { + AddErrorLocation(currentError_, parent); + AddError(ValueType(keyword, GetStateAllocator(), false).Move(), currentError_); + } + + void MergeError(ValueType& other) { + for (typename ValueType::MemberIterator it = other.MemberBegin(), end = other.MemberEnd(); it != end; ++it) { + AddError(it->name, it->value); + } + } + + void AddNumberError(const typename SchemaType::ValueType& keyword, ValueType& actual, const SValue& expected, + const typename SchemaType::ValueType& (*exclusive)() = 0) { + currentError_.SetObject(); + currentError_.AddMember(GetActualString(), actual, GetStateAllocator()); + currentError_.AddMember(GetExpectedString(), ValueType(expected, GetStateAllocator()).Move(), GetStateAllocator()); + if (exclusive) + currentError_.AddMember(ValueType(exclusive(), GetStateAllocator()).Move(), true, GetStateAllocator()); + AddCurrentError(keyword); + } + + void AddErrorArray(const typename SchemaType::ValueType& keyword, + ISchemaValidator** subvalidators, SizeType count) { + ValueType errors(kArrayType); + for (SizeType i = 0; i < count; ++i) + errors.PushBack(static_cast(subvalidators[i])->GetError(), GetStateAllocator()); + currentError_.SetObject(); + currentError_.AddMember(GetErrorsString(), errors, GetStateAllocator()); + AddCurrentError(keyword); + } + const SchemaType& CurrentSchema() const { return *schemaStack_.template Top()->schema; } Context& CurrentContext() { return *schemaStack_.template Top(); } const Context& CurrentContext() const { return *schemaStack_.template Top(); } - static OutputHandler& GetNullHandler() { - static OutputHandler nullHandler; - return nullHandler; - } - static const size_t kDefaultSchemaStackCapacity = 1024; static const size_t kDefaultDocumentStackCapacity = 256; const SchemaDocumentType* schemaDocument_; const SchemaType& root_; - OutputHandler& outputHandler_; StateAllocator* stateAllocator_; StateAllocator* ownStateAllocator_; internal::Stack schemaStack_; //!< stack to store the current path of schema (BaseSchemaType *) internal::Stack documentStack_; //!< stack to store the current path of validating document (Ch) + OutputHandler* outputHandler_; + ValueType error_; + ValueType currentError_; + ValueType missingDependents_; bool valid_; #if CEREAL_RAPIDJSON_SCHEMA_VERBOSE unsigned depth_; @@ -1954,13 +2439,14 @@ class SchemaValidatingReader { public: typedef typename SchemaDocumentType::PointerType PointerType; typedef typename InputStream::Ch Ch; + typedef GenericValue ValueType; //! Constructor /*! \param is Input stream. \param sd Schema document. */ - SchemaValidatingReader(InputStream& is, const SchemaDocumentType& sd) : is_(is), sd_(sd), invalidSchemaKeyword_(), isValid_(true) {} + SchemaValidatingReader(InputStream& is, const SchemaDocumentType& sd) : is_(is), sd_(sd), invalidSchemaKeyword_(), error_(kObjectType), isValid_(true) {} template bool operator()(Handler& handler) { @@ -1973,11 +2459,13 @@ public: invalidSchemaPointer_ = PointerType(); invalidSchemaKeyword_ = 0; invalidDocumentPointer_ = PointerType(); + error_.SetObject(); } else { invalidSchemaPointer_ = validator.GetInvalidSchemaPointer(); invalidSchemaKeyword_ = validator.GetInvalidSchemaKeyword(); invalidDocumentPointer_ = validator.GetInvalidDocumentPointer(); + error_.CopyFrom(validator.GetError(), allocator_); } return parseResult_; @@ -1988,6 +2476,7 @@ public: const PointerType& GetInvalidSchemaPointer() const { return invalidSchemaPointer_; } const Ch* GetInvalidSchemaKeyword() const { return invalidSchemaKeyword_; } const PointerType& GetInvalidDocumentPointer() const { return invalidDocumentPointer_; } + const ValueType& GetError() const { return error_; } private: InputStream& is_; @@ -1997,6 +2486,8 @@ private: PointerType invalidSchemaPointer_; const Ch* invalidSchemaKeyword_; PointerType invalidDocumentPointer_; + StackAllocator allocator_; + ValueType error_; bool isValid_; }; diff --git a/noarch/include/cereal/external/rapidjson/stream.h b/noarch/include/cereal/external/rapidjson/stream.h index c05c2d4e..abc0153b 100644 --- a/noarch/include/cereal/external/rapidjson/stream.h +++ b/noarch/include/cereal/external/rapidjson/stream.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #include "rapidjson.h" @@ -100,6 +100,50 @@ inline void PutN(Stream& stream, Ch c, size_t n) { PutUnsafe(stream, c); } +/////////////////////////////////////////////////////////////////////////////// +// GenericStreamWrapper + +//! A Stream Wrapper +/*! \tThis string stream is a wrapper for any stream by just forwarding any + \treceived message to the origin stream. + \note implements Stream concept +*/ + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(4702) // unreachable code +CEREAL_RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +template > +class GenericStreamWrapper { +public: + typedef typename Encoding::Ch Ch; + GenericStreamWrapper(InputStream& is): is_(is) {} + + Ch Peek() const { return is_.Peek(); } + Ch Take() { return is_.Take(); } + size_t Tell() { return is_.Tell(); } + Ch* PutBegin() { return is_.PutBegin(); } + void Put(Ch ch) { is_.Put(ch); } + void Flush() { is_.Flush(); } + size_t PutEnd(Ch* ch) { return is_.PutEnd(ch); } + + // wrapper for MemoryStream + const Ch* Peek4() const { return is_.Peek4(); } + + // wrapper for AutoUTFInputStream + UTFType GetType() const { return is_.GetType(); } + bool HasBOM() const { return is_.HasBOM(); } + +protected: + InputStream& is_; +}; + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +CEREAL_RAPIDJSON_DIAG_POP +#endif + /////////////////////////////////////////////////////////////////////////////// // StringStream diff --git a/noarch/include/cereal/external/rapidjson/stringbuffer.h b/noarch/include/cereal/external/rapidjson/stringbuffer.h index 76eb7309..d010477a 100644 --- a/noarch/include/cereal/external/rapidjson/stringbuffer.h +++ b/noarch/include/cereal/external/rapidjson/stringbuffer.h @@ -78,8 +78,12 @@ public: return stack_.template Bottom(); } + //! Get the size of string in bytes in the string buffer. size_t GetSize() const { return stack_.GetSize(); } + //! Get the length of string in Ch in the string buffer. + size_t GetLength() const { return stack_.GetSize() / sizeof(Ch); } + static const size_t kDefaultCapacity = 256; mutable internal::Stack stack_; diff --git a/noarch/include/cereal/external/rapidjson/writer.h b/noarch/include/cereal/external/rapidjson/writer.h index ae86cb57..b8da773e 100644 --- a/noarch/include/cereal/external/rapidjson/writer.h +++ b/noarch/include/cereal/external/rapidjson/writer.h @@ -16,6 +16,7 @@ #define CEREAL_RAPIDJSON_WRITER_H_ #include "stream.h" +#include "internal/meta.h" #include "internal/stack.h" #include "internal/strfunc.h" #include "internal/dtoa.h" @@ -31,17 +32,18 @@ #include #elif defined(CEREAL_RAPIDJSON_SSE2) #include -#endif - -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_PUSH -CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +#elif defined(CEREAL_RAPIDJSON_NEON) +#include #endif #ifdef __clang__ CEREAL_RAPIDJSON_DIAG_PUSH CEREAL_RAPIDJSON_DIAG_OFF(padded) CEREAL_RAPIDJSON_DIAG_OFF(unreachable-code) +CEREAL_RAPIDJSON_DIAG_OFF(c++98-compat) +#elif defined(_MSC_VER) +CEREAL_RAPIDJSON_DIAG_PUSH +CEREAL_RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant #endif CEREAL_RAPIDJSON_NAMESPACE_BEGIN @@ -63,7 +65,7 @@ CEREAL_RAPIDJSON_NAMESPACE_BEGIN enum WriteFlag { kWriteNoFlags = 0, //!< No flags are set. kWriteValidateEncodingFlag = 1, //!< Validate encoding of JSON strings. - kWriteNanAndInfFlag = 2, //!< Allow writing of Inf, -Inf and NaN. + kWriteNanAndInfFlag = 2, //!< Allow writing of Infinity, -Infinity and NaN. kWriteDefaultFlags = CEREAL_RAPIDJSON_WRITE_DEFAULT_FLAGS //!< Default write flags. Can be customized by defining CEREAL_RAPIDJSON_WRITE_DEFAULT_FLAGS }; @@ -103,6 +105,13 @@ public: Writer(StackAllocator* allocator = 0, size_t levelDepth = kDefaultLevelDepth) : os_(0), level_stack_(allocator, levelDepth * sizeof(Level)), maxDecimalPlaces_(kDefaultMaxDecimalPlaces), hasRoot_(false) {} +#if CEREAL_RAPIDJSON_HAS_CXX11_RVALUE_REFS + Writer(Writer&& rhs) : + os_(rhs.os_), level_stack_(std::move(rhs.level_stack_)), maxDecimalPlaces_(rhs.maxDecimalPlaces_), hasRoot_(rhs.hasRoot_) { + rhs.os_ = 0; + } +#endif + //! Reset the writer with a new stream. /*! This function reset the writer with a new stream and default settings, @@ -184,12 +193,14 @@ public: bool Double(double d) { Prefix(kNumberType); return EndValue(WriteDouble(d)); } bool RawNumber(const Ch* str, SizeType length, bool copy = false) { + CEREAL_RAPIDJSON_ASSERT(str != 0); (void)copy; Prefix(kNumberType); return EndValue(WriteString(str, length)); } bool String(const Ch* str, SizeType length, bool copy = false) { + CEREAL_RAPIDJSON_ASSERT(str != 0); (void)copy; Prefix(kStringType); return EndValue(WriteString(str, length)); @@ -209,10 +220,18 @@ public: bool Key(const Ch* str, SizeType length, bool copy = false) { return String(str, length, copy); } +#if CEREAL_RAPIDJSON_HAS_STDSTRING + bool Key(const std::basic_string& str) + { + return Key(str.data(), SizeType(str.size())); + } +#endif + bool EndObject(SizeType memberCount = 0) { (void)memberCount; - CEREAL_RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); - CEREAL_RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); + CEREAL_RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); // not inside an Object + CEREAL_RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); // currently inside an Array, not Object + CEREAL_RAPIDJSON_ASSERT(0 == level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value level_stack_.template Pop(1); return EndValue(WriteEndObject()); } @@ -236,9 +255,9 @@ public: //@{ //! Simpler but slower overload. - bool String(const Ch* str) { return String(str, internal::StrLen(str)); } - bool Key(const Ch* str) { return Key(str, internal::StrLen(str)); } - + bool String(const Ch* const& str) { return String(str, internal::StrLen(str)); } + bool Key(const Ch* const& str) { return Key(str, internal::StrLen(str)); } + //@} //! Write a raw JSON value. @@ -249,7 +268,19 @@ public: \param length Length of the json. \param type Type of the root of json. */ - bool RawValue(const Ch* json, size_t length, Type type) { Prefix(type); return EndValue(WriteRawValue(json, length)); } + bool RawValue(const Ch* json, size_t length, Type type) { + CEREAL_RAPIDJSON_ASSERT(json != 0); + Prefix(type); + return EndValue(WriteRawValue(json, length)); + } + + //! Flush the output stream. + /*! + Allows the user to flush the output stream immediately. + */ + void Flush() { + os_->Flush(); + } protected: //! Information for each nested level @@ -283,7 +314,7 @@ protected: const char* end = internal::i32toa(i, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -292,7 +323,7 @@ protected: const char* end = internal::u32toa(u, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -301,7 +332,7 @@ protected: const char* end = internal::i64toa(i64, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -310,7 +341,7 @@ protected: char* end = internal::u64toa(u64, buffer); PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -338,12 +369,12 @@ protected: char* end = internal::dtoa(d, buffer, maxDecimalPlaces_); PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteString(const Ch* str, SizeType length) { - static const typename TargetEncoding::Ch hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; + static const typename OutputStream::Ch hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; static const char escape[256] = { #define Z16 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 //0 1 2 3 4 5 6 7 8 9 A B C D E F @@ -399,7 +430,7 @@ protected: else if ((sizeof(Ch) == 1 || static_cast(c) < 256) && CEREAL_RAPIDJSON_UNLIKELY(escape[static_cast(c)])) { is.Take(); PutUnsafe(*os_, '\\'); - PutUnsafe(*os_, static_cast(escape[static_cast(c)])); + PutUnsafe(*os_, static_cast(escape[static_cast(c)])); if (escape[static_cast(c)] == 'u') { PutUnsafe(*os_, '0'); PutUnsafe(*os_, '0'); @@ -427,9 +458,13 @@ protected: bool WriteRawValue(const Ch* json, size_t length) { PutReserve(*os_, length); - for (size_t i = 0; i < length; i++) { - CEREAL_RAPIDJSON_ASSERT(json[i] != '\0'); - PutUnsafe(*os_, json[i]); + GenericStringStream is(json); + while (CEREAL_RAPIDJSON_LIKELY(is.Tell() < length)) { + CEREAL_RAPIDJSON_ASSERT(is.Peek() != '\0'); + if (CEREAL_RAPIDJSON_UNLIKELY(!(writeFlags & kWriteValidateEncodingFlag ? + Transcoder::Validate(is, *os_) : + Transcoder::TranscodeUnsafe(is, *os_)))) + return false; } return true; } @@ -457,7 +492,7 @@ protected: // Flush the value if it is the top level one. bool EndValue(bool ret) { if (CEREAL_RAPIDJSON_UNLIKELY(level_stack_.Empty())) // end of json text - os_->Flush(); + Flush(); return ret; } @@ -561,7 +596,7 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -570,7 +605,7 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (CEREAL_RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -595,15 +630,79 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz is.src_ = p; return CEREAL_RAPIDJSON_LIKELY(is.Tell() < length); } -#endif // defined(CEREAL_RAPIDJSON_SSE2) || defined(CEREAL_RAPIDJSON_SSE42) +#elif defined(CEREAL_RAPIDJSON_NEON) +template<> +inline bool Writer::ScanWriteUnescapedString(StringStream& is, size_t length) { + if (length < 16) + return CEREAL_RAPIDJSON_LIKELY(is.Tell() < length); + + if (!CEREAL_RAPIDJSON_LIKELY(is.Tell() < length)) + return false; + + const char* p = is.src_; + const char* end = is.head_ + length; + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + const char* endAligned = reinterpret_cast(reinterpret_cast(end) & static_cast(~15)); + if (nextAligned > end) + return true; + + while (p != nextAligned) + if (*p < 0x20 || *p == '\"' || *p == '\\') { + is.src_ = p; + return CEREAL_RAPIDJSON_LIKELY(is.Tell() < length); + } + else + os_->PutUnsafe(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (; p != endAligned; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType len = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + len = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + len = lz >> 3; + escaped = true; + } + if (CEREAL_RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + char* q = reinterpret_cast(os_->PushUnsafe(len)); + for (size_t i = 0; i < len; i++) + q[i] = p[i]; + + p += len; + break; + } + vst1q_u8(reinterpret_cast(os_->PushUnsafe(16)), s); + } + + is.src_ = p; + return CEREAL_RAPIDJSON_LIKELY(is.Tell() < length); +} +#endif // CEREAL_RAPIDJSON_NEON CEREAL_RAPIDJSON_NAMESPACE_END -#ifdef _MSC_VER -CEREAL_RAPIDJSON_DIAG_POP -#endif - -#ifdef __clang__ +#if defined(_MSC_VER) || defined(__clang__) CEREAL_RAPIDJSON_DIAG_POP #endif diff --git a/noarch/include/cereal/external/rapidxml/license.txt b/noarch/include/cereal/external/rapidxml/license.txt index 14098318..0095bc72 100644 --- a/noarch/include/cereal/external/rapidxml/license.txt +++ b/noarch/include/cereal/external/rapidxml/license.txt @@ -1,52 +1,52 @@ -Use of this software is granted under one of the following two licenses, -to be chosen freely by the user. - -1. Boost Software License - Version 1.0 - August 17th, 2003 -=============================================================================== - -Copyright (c) 2006, 2007 Marcin Kalicinski - -Permission is hereby granted, free of charge, to any person or organization -obtaining a copy of the software and accompanying documentation covered by -this license (the "Software") to use, reproduce, display, distribute, -execute, and transmit the Software, and to prepare derivative works of the -Software, and to permit third-parties to whom the Software is furnished to -do so, all subject to the following: - -The copyright notices in the Software and this entire statement, including -the above license grant, this restriction and the following disclaimer, -must be included in all copies of the Software, in whole or in part, and -all derivative works of the Software, unless such copies or derivative -works are solely in the form of machine-executable object code generated by -a source language processor. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT -SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE -FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, -ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER -DEALINGS IN THE SOFTWARE. - -2. The MIT License -=============================================================================== - -Copyright (c) 2006, 2007 Marcin Kalicinski - -Permission is hereby granted, free of charge, to any person obtaining a copy -of this software and associated documentation files (the "Software"), to deal -in the Software without restriction, including without limitation the rights -to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies -of the Software, and to permit persons to whom the Software is furnished to do so, -subject to the following conditions: - -The above copyright notice and this permission notice shall be included in all -copies or substantial portions of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL -THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS -IN THE SOFTWARE. +Use of this software is granted under one of the following two licenses, +to be chosen freely by the user. + +1. Boost Software License - Version 1.0 - August 17th, 2003 +=============================================================================== + +Copyright (c) 2006, 2007 Marcin Kalicinski + +Permission is hereby granted, free of charge, to any person or organization +obtaining a copy of the software and accompanying documentation covered by +this license (the "Software") to use, reproduce, display, distribute, +execute, and transmit the Software, and to prepare derivative works of the +Software, and to permit third-parties to whom the Software is furnished to +do so, all subject to the following: + +The copyright notices in the Software and this entire statement, including +the above license grant, this restriction and the following disclaimer, +must be included in all copies of the Software, in whole or in part, and +all derivative works of the Software, unless such copies or derivative +works are solely in the form of machine-executable object code generated by +a source language processor. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT +SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE +FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, +ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER +DEALINGS IN THE SOFTWARE. + +2. The MIT License +=============================================================================== + +Copyright (c) 2006, 2007 Marcin Kalicinski + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies +of the Software, and to permit persons to whom the Software is furnished to do so, +subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS +IN THE SOFTWARE. diff --git a/noarch/include/cereal/external/rapidxml/manual.html b/noarch/include/cereal/external/rapidxml/manual.html index 2c422703..4e6a5f1a 100644 --- a/noarch/include/cereal/external/rapidxml/manual.html +++ b/noarch/include/cereal/external/rapidxml/manual.html @@ -1,406 +1,406 @@ -

RAPIDXML Manual

Version 1.13

Copyright (C) 2006, 2009 Marcin Kalicinski
See accompanying file license.txt for license information.

Table of Contents

1. What is RapidXml?
1.1 Dependencies And Compatibility
1.2 Character Types And Encodings
1.3 Error Handling
1.4 Memory Allocation
1.5 W3C Compliance
1.6 API Design
1.7 Reliability
1.8 Acknowledgements
2. Two Minute Tutorial
2.1 Parsing
2.2 Accessing The DOM Tree
2.3 Modifying The DOM Tree
2.4 Printing XML
3. Differences From Regular XML Parsers
3.1 Lifetime Of Source Text
3.2 Ownership Of Strings
3.3 Destructive Vs Non-Destructive Mode
4. Performance
4.1 Comparison With Other Parsers
5. Reference

1. What is RapidXml?

RapidXml is an attempt to create the fastest XML DOM parser possible, while retaining useability, portability and reasonable W3C compatibility. It is an in-situ parser written in C++, with parsing speed approaching that of strlen() function executed on the same data.

- Entire parser is contained in a single header file, so no building or linking is neccesary. To use it you just need to copy rapidxml.hpp file to a convenient place (such as your project directory), and include it where needed. You may also want to use printing functions contained in header rapidxml_print.hpp.

1.1 Dependencies And Compatibility

RapidXml has no dependencies other than a very small subset of standard C++ library (<cassert>, <cstdlib>, <new> and <exception>, unless exceptions are disabled). It should compile on any reasonably conformant compiler, and was tested on Visual C++ 2003, Visual C++ 2005, Visual C++ 2008, gcc 3, gcc 4, and Comeau 4.3.3. Care was taken that no warnings are produced on these compilers, even with highest warning levels enabled.

1.2 Character Types And Encodings

RapidXml is character type agnostic, and can work both with narrow and wide characters. Current version does not fully support UTF-16 or UTF-32, so use of wide characters is somewhat incapacitated. However, it should succesfully parse wchar_t strings containing UTF-16 or UTF-32 if endianness of the data matches that of the machine. UTF-8 is fully supported, including all numeric character references, which are expanded into appropriate UTF-8 byte sequences (unless you enable parse_no_utf8 flag).

- Note that RapidXml performs no decoding - strings returned by name() and value() functions will contain text encoded using the same encoding as source file. Rapidxml understands and expands the following character references: &apos; &amp; &quot; &lt; &gt; &#...; Other character references are not expanded.

1.3 Error Handling

By default, RapidXml uses C++ exceptions to report errors. If this behaviour is undesirable, RAPIDXML_NO_EXCEPTIONS can be defined to suppress exception code. See parse_error class and parse_error_handler() function for more information.

1.4 Memory Allocation

RapidXml uses a special memory pool object to allocate nodes and attributes, because direct allocation using new operator would be far too slow. Underlying memory allocations performed by the pool can be customized by use of memory_pool::set_allocator() function. See class memory_pool for more information.

1.5 W3C Compliance

RapidXml is not a W3C compliant parser, primarily because it ignores DOCTYPE declarations. There is a number of other, minor incompatibilities as well. Still, it can successfully parse and produce complete trees of all valid XML files in W3C conformance suite (over 1000 files specially designed to find flaws in XML processors). In destructive mode it performs whitespace normalization and character entity substitution for a small set of built-in entities.

1.6 API Design

RapidXml API is minimalistic, to reduce code size as much as possible, and facilitate use in embedded environments. Additional convenience functions are provided in separate headers: rapidxml_utils.hpp and rapidxml_print.hpp. Contents of these headers is not an essential part of the library, and is currently not documented (otherwise than with comments in code).

1.7 Reliability

RapidXml is very robust and comes with a large harness of unit tests. Special care has been taken to ensure stability of the parser no matter what source text is thrown at it. One of the unit tests produces 100,000 randomly corrupted variants of XML document, which (when uncorrupted) contains all constructs recognized by RapidXml. RapidXml passes this test when it correctly recognizes that errors have been introduced, and does not crash or loop indefinitely.

- Another unit test puts RapidXml head-to-head with another, well estabilished XML parser, and verifies that their outputs match across a wide variety of small and large documents.

- Yet another test feeds RapidXml with over 1000 test files from W3C compliance suite, and verifies that correct results are obtained. There are also additional tests that verify each API function separately, and test that various parsing modes work as expected.

1.8 Acknowledgements

I would like to thank Arseny Kapoulkine for his work on pugixml, which was an inspiration for this project. Additional thanks go to Kristen Wegner for creating pugxml, from which pugixml was derived. Janusz Wohlfeil kindly ran RapidXml speed tests on hardware that I did not have access to, allowing me to expand performance comparison table.

2. Two Minute Tutorial

2.1 Parsing

The following code causes RapidXml to parse a zero-terminated string named text: 
using namespace rapidxml;
-xml_document<> doc;    // character type defaults to char
-doc.parse<0>(text);    // 0 means default parse flags
-
doc object is now a root of DOM tree containing representation of the parsed XML. Because all RapidXml interface is contained inside namespace rapidxml, users must either bring contents of this namespace into scope, or fully qualify all the names. Class xml_document represents a root of the DOM hierarchy. By means of public inheritance, it is also an xml_node and a memory_pool. Template parameter of xml_document::parse() function is used to specify parsing flags, with which you can fine-tune behaviour of the parser. Note that flags must be a compile-time constant.

2.2 Accessing The DOM Tree

To access the DOM tree, use methods of xml_node and xml_attribute classes: 
cout << "Name of my first node is: " << doc.first_node()->name() << "\n";
-xml_node<> *node = doc.first_node("foobar");
-cout << "Node foobar has value " << node->value() << "\n";
-for (xml_attribute<> *attr = node->first_attribute();
-     attr; attr = attr->next_attribute())
-{
-    cout << "Node foobar has attribute " << attr->name() << " ";
-    cout << "with value " << attr->value() << "\n";
-}
-

2.3 Modifying The DOM Tree

DOM tree produced by the parser is fully modifiable. Nodes and attributes can be added/removed, and their contents changed. The below example creates a HTML document, whose sole contents is a link to google.com website: 
xml_document<> doc;
-xml_node<> *node = doc.allocate_node(node_element, "a", "Google");
-doc.append_node(node);
-xml_attribute<> *attr = doc.allocate_attribute("href", "google.com");
-node->append_attribute(attr);
-
One quirk is that nodes and attributes do not own the text of their names and values. This is because normally they only store pointers to the source text. So, when assigning a new name or value to the node, care must be taken to ensure proper lifetime of the string. The easiest way to achieve it is to allocate the string from the xml_document memory pool. In the above example this is not necessary, because we are only assigning character constants. But the code below uses memory_pool::allocate_string() function to allocate node name (which will have the same lifetime as the document), and assigns it to a new node: 
xml_document<> doc;
-char *node_name = doc.allocate_string(name);        // Allocate string and copy name into it
-xml_node<> *node = doc.allocate_node(node_element, node_name);  // Set node name to node_name
-
Check Reference section for description of the entire interface.

2.4 Printing XML

You can print xml_document and xml_node objects into an XML string. Use print() function or operator <<, which are defined in rapidxml_print.hpp header. 
using namespace rapidxml;
-xml_document<> doc;    // character type defaults to char
-// ... some code to fill the document
-
-// Print to stream using operator <<
-std::cout << doc;   
-
-// Print to stream using print function, specifying printing flags
-print(std::cout, doc, 0);   // 0 means default printing flags
-
-// Print to string using output iterator
-std::string s;
-print(std::back_inserter(s), doc, 0);
-
-// Print to memory buffer using output iterator
-char buffer[4096];                      // You are responsible for making the buffer large enough!
-char *end = print(buffer, doc, 0);      // end contains pointer to character after last printed character
-*end = 0;                               // Add string terminator after XML
-

3. Differences From Regular XML Parsers

RapidXml is an in-situ parser, which allows it to achieve very high parsing speed. In-situ means that parser does not make copies of strings. Instead, it places pointers to the source text in the DOM hierarchy.

3.1 Lifetime Of Source Text

In-situ parsing requires that source text lives at least as long as the document object. If source text is destroyed, names and values of nodes in DOM tree will become destroyed as well. Additionally, whitespace processing, character entity translation, and zero-termination of strings require that source text be modified during parsing (but see non-destructive mode). This makes the text useless for further processing once it was parsed by RapidXml.

- In many cases however, these are not serious issues.

3.2 Ownership Of Strings

Nodes and attributes produced by RapidXml do not own their name and value strings. They merely hold the pointers to them. This means you have to be careful when setting these values manually, by using xml_base::name(const Ch *) or xml_base::value(const Ch *) functions. Care must be taken to ensure that lifetime of the string passed is at least as long as lifetime of the node/attribute. The easiest way to achieve it is to allocate the string from memory_pool owned by the document. Use memory_pool::allocate_string() function for this purpose.

3.3 Destructive Vs Non-Destructive Mode

By default, the parser modifies source text during the parsing process. This is required to achieve character entity translation, whitespace normalization, and zero-termination of strings.

- In some cases this behaviour may be undesirable, for example if source text resides in read only memory, or is mapped to memory directly from file. By using appropriate parser flags (parse_non_destructive), source text modifications can be disabled. However, because RapidXml does in-situ parsing, it obviously has the following side-effects:

4. Performance

RapidXml achieves its speed through use of several techniques:
  • In-situ parsing. When building DOM tree, RapidXml does not make copies of string data, such as node names and values. Instead, it stores pointers to interior of the source text.
  • Use of template metaprogramming techniques. This allows it to move much of the work to compile time. Through magic of the templates, C++ compiler generates a separate copy of parsing code for any combination of parser flags you use. In each copy, all possible decisions are made at compile time and all unused code is omitted.
  • Extensive use of lookup tables for parsing.
  • Hand-tuned C++ with profiling done on several most popular CPUs.
This results in a very small and fast code: a parser which is custom tailored to exact needs with each invocation.

4.1 Comparison With Other Parsers

The table below compares speed of RapidXml to some other parsers, and to strlen() function executed on the same data. On a modern CPU (as of 2007), you can expect parsing throughput to be close to 1 GB/s. As a rule of thumb, parsing speed is about 50-100x faster than Xerces DOM, 30-60x faster than TinyXml, 3-12x faster than pugxml, and about 5% - 30% faster than pugixml, the fastest XML parser I know of.
  • The test file is a real-world, 50kB large, moderately dense XML file.
  • All timing is done by using RDTSC instruction present in Pentium-compatible CPUs.
  • No profile-guided optimizations are used.
  • All parsers are running in their fastest modes.
  • The results are given in CPU cycles per character, so frequency of CPUs is irrelevant.
  • The results are minimum values from a large number of runs, to minimize effects of operating system activity, task switching, interrupt handling etc.
  • A single parse of the test file takes about 1/10th of a millisecond, so with large number of runs there is a good chance of hitting at least one no-interrupt streak, and obtaining undisturbed results.
Platform
Compiler
strlen() RapidXml pugixml 0.3 pugxml TinyXml
Pentium 4
MSVC 8.0
2.5
5.4
7.0
61.7
298.8
Pentium 4
gcc 4.1.1
0.8
6.1
9.5
67.0
413.2
Core 2
MSVC 8.0
1.0
4.5
5.0
24.6
154.8
Core 2
gcc 4.1.1
0.6
4.6
5.4
28.3
229.3
Athlon XP
MSVC 8.0
3.1
7.7
8.0
25.5
182.6
Athlon XP
gcc 4.1.1
0.9
8.2
9.2
33.7
265.2
Pentium 3
MSVC 8.0
2.0
6.3
7.0
30.9
211.9
Pentium 3
gcc 4.1.1
1.0
6.7
8.9
35.3
316.0
(*) All results are in CPU cycles per character of source text

5. Reference

This section lists all classes, functions, constants etc. and describes them in detail.
class - template - rapidxml::memory_pool
- constructor - memory_pool()
- destructor - ~memory_pool()
function allocate_node(node_type type, const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0)
function allocate_attribute(const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0)
function allocate_string(const Ch *source=0, std::size_t size=0)
function clone_node(const xml_node< Ch > *source, xml_node< Ch > *result=0)
function clear()
function set_allocator(alloc_func *af, free_func *ff)

class rapidxml::parse_error
- constructor - parse_error(const char *what, void *where)
function what() const
function where() const

class - template - rapidxml::xml_attribute
- constructor - xml_attribute()
function document() const
function previous_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function next_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const

class - template - rapidxml::xml_base
- constructor - xml_base()
function name() const
function name_size() const
function value() const
function value_size() const
function name(const Ch *name, std::size_t size)
function name(const Ch *name)
function value(const Ch *value, std::size_t size)
function value(const Ch *value)
function parent() const

class - template - rapidxml::xml_document
- constructor - xml_document()
function parse(Ch *text)
function clear()

class - template - rapidxml::xml_node
- constructor - xml_node(node_type type)
function type() const
function document() const
function first_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function last_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function previous_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function next_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function first_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function last_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function type(node_type type)
function prepend_node(xml_node< Ch > *child)
function append_node(xml_node< Ch > *child)
function insert_node(xml_node< Ch > *where, xml_node< Ch > *child)
function remove_first_node()
function remove_last_node()
function remove_node(xml_node< Ch > *where)
function remove_all_nodes()
function prepend_attribute(xml_attribute< Ch > *attribute)
function append_attribute(xml_attribute< Ch > *attribute)
function insert_attribute(xml_attribute< Ch > *where, xml_attribute< Ch > *attribute)
function remove_first_attribute()
function remove_last_attribute()
function remove_attribute(xml_attribute< Ch > *where)
function remove_all_attributes()

namespace rapidxml
enum node_type
function parse_error_handler(const char *what, void *where)
function print(OutIt out, const xml_node< Ch > &node, int flags=0)
function print(std::basic_ostream< Ch > &out, const xml_node< Ch > &node, int flags=0)
function operator<<(std::basic_ostream< Ch > &out, const xml_node< Ch > &node)
- constant - parse_no_data_nodes
- constant - parse_no_element_values
- constant - parse_no_string_terminators
- constant - parse_no_entity_translation
- constant - parse_no_utf8
- constant - parse_declaration_node
- constant - parse_comment_nodes
- constant - parse_doctype_node
- constant - parse_pi_nodes
- constant - parse_validate_closing_tags
- constant - parse_trim_whitespace
- constant - parse_normalize_whitespace
- constant - parse_default
- constant - parse_non_destructive
- constant - parse_fastest
- constant - parse_full
- constant - print_no_indenting

class - template - rapidxml::memory_pool

- - Defined in rapidxml.hpp
- Base class for - xml_document

Description

This class is used by the parser to create new nodes and attributes, without overheads of dynamic memory allocation. In most cases, you will not need to use this class directly. However, if you need to create nodes manually or modify names/values of nodes, you are encouraged to use memory_pool of relevant xml_document to allocate the memory. Not only is this faster than allocating them by using new operator, but also their lifetime will be tied to the lifetime of document, possibly simplyfing memory management.

- Call allocate_node() or allocate_attribute() functions to obtain new nodes or attributes from the pool. You can also call allocate_string() function to allocate strings. Such strings can then be used as names or values of nodes without worrying about their lifetime. Note that there is no free() function -- all allocations are freed at once when clear() function is called, or when the pool is destroyed.

- It is also possible to create a standalone memory_pool, and use it to allocate nodes, whose lifetime will not be tied to any document.

- Pool maintains RAPIDXML_STATIC_POOL_SIZE bytes of statically allocated memory. Until static memory is exhausted, no dynamic memory allocations are done. When static memory is exhausted, pool allocates additional blocks of memory of size RAPIDXML_DYNAMIC_POOL_SIZE each, by using global new[] and delete[] operators. This behaviour can be changed by setting custom allocation routines. Use set_allocator() function to set them.

- Allocations for nodes, attributes and strings are aligned at RAPIDXML_ALIGNMENT bytes. This value defaults to the size of pointer on target architecture.

- To obtain absolutely top performance from the parser, it is important that all nodes are allocated from a single, contiguous block of memory. Otherwise, cache misses when jumping between two (or more) disjoint blocks of memory can slow down parsing quite considerably. If required, you can tweak RAPIDXML_STATIC_POOL_SIZE, RAPIDXML_DYNAMIC_POOL_SIZE and RAPIDXML_ALIGNMENT to obtain best wasted memory to performance compromise. To do it, define their values before rapidxml.hpp file is included.

Parameters

Ch
Character type of created nodes.

- constructor - memory_pool::memory_pool

Synopsis

memory_pool(); -

Description

Constructs empty pool with default allocator functions.

- destructor - memory_pool::~memory_pool

Synopsis

~memory_pool(); -

Description

Destroys pool and frees all the memory. This causes memory occupied by nodes allocated by the pool to be freed. Nodes allocated from the pool are no longer valid.

function memory_pool::allocate_node

Synopsis

xml_node<Ch>* allocate_node(node_type type, const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0); -

Description

Allocates a new node from the pool, and optionally assigns name and value to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

type
Type of node to create.
name
Name to assign to the node, or 0 to assign no name.
value
Value to assign to the node, or 0 to assign no value.
name_size
Size of name to assign, or 0 to automatically calculate size from name string.
value_size
Size of value to assign, or 0 to automatically calculate size from value string.

Returns

Pointer to allocated node. This pointer will never be NULL.

function memory_pool::allocate_attribute

Synopsis

xml_attribute<Ch>* allocate_attribute(const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0); -

Description

Allocates a new attribute from the pool, and optionally assigns name and value to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

name
Name to assign to the attribute, or 0 to assign no name.
value
Value to assign to the attribute, or 0 to assign no value.
name_size
Size of name to assign, or 0 to automatically calculate size from name string.
value_size
Size of value to assign, or 0 to automatically calculate size from value string.

Returns

Pointer to allocated attribute. This pointer will never be NULL.

function memory_pool::allocate_string

Synopsis

Ch* allocate_string(const Ch *source=0, std::size_t size=0); -

Description

Allocates a char array of given size from the pool, and optionally copies a given string to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

source
String to initialize the allocated memory with, or 0 to not initialize it.
size
Number of characters to allocate, or zero to calculate it automatically from source string length; if size is 0, source string must be specified and null terminated.

Returns

Pointer to allocated char array. This pointer will never be NULL.

function memory_pool::clone_node

Synopsis

xml_node<Ch>* clone_node(const xml_node< Ch > *source, xml_node< Ch > *result=0); -

Description

Clones an xml_node and its hierarchy of child nodes and attributes. Nodes and attributes are allocated from this memory pool. Names and values are not cloned, they are shared between the clone and the source. Result node can be optionally specified as a second parameter, in which case its contents will be replaced with cloned source node. This is useful when you want to clone entire document.

Parameters

source
Node to clone.
result
Node to put results in, or 0 to automatically allocate result node

Returns

Pointer to cloned node. This pointer will never be NULL.

function memory_pool::clear

Synopsis

void clear(); -

Description

Clears the pool. This causes memory occupied by nodes allocated by the pool to be freed. Any nodes or strings allocated from the pool will no longer be valid.

function memory_pool::set_allocator

Synopsis

void set_allocator(alloc_func *af, free_func *ff); -

Description

Sets or resets the user-defined memory allocation functions for the pool. This can only be called when no memory is allocated from the pool yet, otherwise results are undefined. Allocation function must not return invalid pointer on failure. It should either throw, stop the program, or use longjmp() function to pass control to other place of program. If it returns invalid pointer, results are undefined.

- User defined allocation functions must have the following forms:

-void *allocate(std::size_t size);
-void free(void *pointer);

Parameters

af
Allocation function, or 0 to restore default function
ff
Free function, or 0 to restore default function

class rapidxml::parse_error

- - Defined in rapidxml.hpp

Description

Parse error exception. This exception is thrown by the parser when an error occurs. Use what() function to get human-readable error message. Use where() function to get a pointer to position within source text where error was detected.

- If throwing exceptions by the parser is undesirable, it can be disabled by defining RAPIDXML_NO_EXCEPTIONS macro before rapidxml.hpp is included. This will cause the parser to call rapidxml::parse_error_handler() function instead of throwing an exception. This function must be defined by the user.

- This class derives from std::exception class.

- constructor - parse_error::parse_error

Synopsis

parse_error(const char *what, void *where); -

Description

Constructs parse error.

function parse_error::what

Synopsis

virtual const char* what() const; -

Description

Gets human readable description of error.

Returns

Pointer to null terminated description of the error.

function parse_error::where

Synopsis

Ch* where() const; -

Description

Gets pointer to character data where error happened. Ch should be the same as char type of xml_document that produced the error.

Returns

Pointer to location within the parsed string where error occured.

class - template - rapidxml::xml_attribute

- - Defined in rapidxml.hpp
- Inherits from - xml_base

Description

Class representing attribute node of XML document. Each attribute has name and value strings, which are available through name() and value() functions (inherited from xml_base). Note that after parse, both name and value of attribute will point to interior of source text used for parsing. Thus, this text must persist in memory for the lifetime of attribute.

Parameters

Ch
Character type to use.

- constructor - xml_attribute::xml_attribute

Synopsis

xml_attribute(); -

Description

Constructs an empty attribute with the specified type. Consider using memory_pool of appropriate xml_document if allocating attributes manually.

function xml_attribute::document

Synopsis

xml_document<Ch>* document() const; -

Description

Gets document of which attribute is a child.

Returns

Pointer to document that contains this attribute, or 0 if there is no parent document.

function xml_attribute::previous_attribute

Synopsis

xml_attribute<Ch>* previous_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets previous attribute, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return previous attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_attribute::next_attribute

Synopsis

xml_attribute<Ch>* next_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets next attribute, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return next attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

class - template - rapidxml::xml_base

- - Defined in rapidxml.hpp
- Base class for - xml_attribute xml_node

Description

Base class for xml_node and xml_attribute implementing common functions: name(), name_size(), value(), value_size() and parent().

Parameters

Ch
Character type to use

- constructor - xml_base::xml_base

Synopsis

xml_base(); -

function xml_base::name

Synopsis

Ch* name() const; -

Description

Gets name of the node. Interpretation of name depends on type of node. Note that name will not be zero-terminated if rapidxml::parse_no_string_terminators option was selected during parse.

- Use name_size() function to determine length of the name.

Returns

Name of node, or empty string if node has no name.

function xml_base::name_size

Synopsis

std::size_t name_size() const; -

Description

Gets size of node name, not including terminator character. This function works correctly irrespective of whether name is or is not zero terminated.

Returns

Size of node name, in characters.

function xml_base::value

Synopsis

Ch* value() const; -

Description

Gets value of node. Interpretation of value depends on type of node. Note that value will not be zero-terminated if rapidxml::parse_no_string_terminators option was selected during parse.

- Use value_size() function to determine length of the value.

Returns

Value of node, or empty string if node has no value.

function xml_base::value_size

Synopsis

std::size_t value_size() const; -

Description

Gets size of node value, not including terminator character. This function works correctly irrespective of whether value is or is not zero terminated.

Returns

Size of node value, in characters.

function xml_base::name

Synopsis

void name(const Ch *name, std::size_t size); -

Description

Sets name of node to a non zero-terminated string. See Ownership Of Strings .

- Note that node does not own its name or value, it only stores a pointer to it. It will not delete or otherwise free the pointer on destruction. It is reponsibility of the user to properly manage lifetime of the string. The easiest way to achieve it is to use memory_pool of the document to allocate the string - on destruction of the document the string will be automatically freed.

- Size of name must be specified separately, because name does not have to be zero terminated. Use name(const Ch *) function to have the length automatically calculated (string must be zero terminated).

Parameters

name
Name of node to set. Does not have to be zero terminated.
size
Size of name, in characters. This does not include zero terminator, if one is present.

function xml_base::name

Synopsis

void name(const Ch *name); -

Description

Sets name of node to a zero-terminated string. See also Ownership Of Strings and xml_node::name(const Ch *, std::size_t).

Parameters

name
Name of node to set. Must be zero terminated.

function xml_base::value

Synopsis

void value(const Ch *value, std::size_t size); -

Description

Sets value of node to a non zero-terminated string. See Ownership Of Strings .

- Note that node does not own its name or value, it only stores a pointer to it. It will not delete or otherwise free the pointer on destruction. It is reponsibility of the user to properly manage lifetime of the string. The easiest way to achieve it is to use memory_pool of the document to allocate the string - on destruction of the document the string will be automatically freed.

- Size of value must be specified separately, because it does not have to be zero terminated. Use value(const Ch *) function to have the length automatically calculated (string must be zero terminated).

- If an element has a child node of type node_data, it will take precedence over element value when printing. If you want to manipulate data of elements using values, use parser flag rapidxml::parse_no_data_nodes to prevent creation of data nodes by the parser.

Parameters

value
value of node to set. Does not have to be zero terminated.
size
Size of value, in characters. This does not include zero terminator, if one is present.

function xml_base::value

Synopsis

void value(const Ch *value); -

Description

Sets value of node to a zero-terminated string. See also Ownership Of Strings and xml_node::value(const Ch *, std::size_t).

Parameters

value
Vame of node to set. Must be zero terminated.

function xml_base::parent

Synopsis

xml_node<Ch>* parent() const; -

Description

Gets node parent.

Returns

Pointer to parent node, or 0 if there is no parent.

class - template - rapidxml::xml_document

- - Defined in rapidxml.hpp
- Inherits from - xml_node memory_pool

Description

This class represents root of the DOM hierarchy. It is also an xml_node and a memory_pool through public inheritance. Use parse() function to build a DOM tree from a zero-terminated XML text string. parse() function allocates memory for nodes and attributes by using functions of xml_document, which are inherited from memory_pool. To access root node of the document, use the document itself, as if it was an xml_node.

Parameters

Ch
Character type to use.

- constructor - xml_document::xml_document

Synopsis

xml_document(); -

Description

Constructs empty XML document.

function xml_document::parse

Synopsis

void parse(Ch *text); -

Description

Parses zero-terminated XML string according to given flags. Passed string will be modified by the parser, unless rapidxml::parse_non_destructive flag is used. The string must persist for the lifetime of the document. In case of error, rapidxml::parse_error exception will be thrown.

- If you want to parse contents of a file, you must first load the file into the memory, and pass pointer to its beginning. Make sure that data is zero-terminated.

- Document can be parsed into multiple times. Each new call to parse removes previous nodes and attributes (if any), but does not clear memory pool.

Parameters

text
XML data to parse; pointer is non-const to denote fact that this data may be modified by the parser.

function xml_document::clear

Synopsis

void clear(); -

Description

Clears the document by deleting all nodes and clearing the memory pool. All nodes owned by document pool are destroyed.

class - template - rapidxml::xml_node

- - Defined in rapidxml.hpp
- Inherits from - xml_base
- Base class for - xml_document

Description

Class representing a node of XML document. Each node may have associated name and value strings, which are available through name() and value() functions. Interpretation of name and value depends on type of the node. Type of node can be determined by using type() function.

- Note that after parse, both name and value of node, if any, will point interior of source text used for parsing. Thus, this text must persist in the memory for the lifetime of node.

Parameters

Ch
Character type to use.

- constructor - xml_node::xml_node

Synopsis

xml_node(node_type type); -

Description

Constructs an empty node with the specified type. Consider using memory_pool of appropriate document to allocate nodes manually.

Parameters

type
Type of node to construct.

function xml_node::type

Synopsis

node_type type() const; -

Description

Gets type of node.

Returns

Type of node.

function xml_node::document

Synopsis

xml_document<Ch>* document() const; -

Description

Gets document of which node is a child.

Returns

Pointer to document that contains this node, or 0 if there is no parent document.

function xml_node::first_node

Synopsis

xml_node<Ch>* first_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets first child node, optionally matching node name.

Parameters

name
Name of child to find, or 0 to return first child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found child, or 0 if not found.

function xml_node::last_node

Synopsis

xml_node<Ch>* last_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets last child node, optionally matching node name. Behaviour is undefined if node has no children. Use first_node() to test if node has children.

Parameters

name
Name of child to find, or 0 to return last child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found child, or 0 if not found.

function xml_node::previous_sibling

Synopsis

xml_node<Ch>* previous_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets previous sibling node, optionally matching node name. Behaviour is undefined if node has no parent. Use parent() to test if node has a parent.

Parameters

name
Name of sibling to find, or 0 to return previous sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found sibling, or 0 if not found.

function xml_node::next_sibling

Synopsis

xml_node<Ch>* next_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets next sibling node, optionally matching node name. Behaviour is undefined if node has no parent. Use parent() to test if node has a parent.

Parameters

name
Name of sibling to find, or 0 to return next sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found sibling, or 0 if not found.

function xml_node::first_attribute

Synopsis

xml_attribute<Ch>* first_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets first attribute of node, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return first attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_node::last_attribute

Synopsis

xml_attribute<Ch>* last_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; -

Description

Gets last attribute of node, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return last attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_node::type

Synopsis

void type(node_type type); -

Description

Sets type of node.

Parameters

type
Type of node to set.

function xml_node::prepend_node

Synopsis

void prepend_node(xml_node< Ch > *child); -

Description

Prepends a new child node. The prepended child becomes the first child, and all existing children are moved one position back.

Parameters

child
Node to prepend.

function xml_node::append_node

Synopsis

void append_node(xml_node< Ch > *child); -

Description

Appends a new child node. The appended child becomes the last child.

Parameters

child
Node to append.

function xml_node::insert_node

Synopsis

void insert_node(xml_node< Ch > *where, xml_node< Ch > *child); -

Description

Inserts a new child node at specified place inside the node. All children after and including the specified node are moved one position back.

Parameters

where
Place where to insert the child, or 0 to insert at the back.
child
Node to insert.

function xml_node::remove_first_node

Synopsis

void remove_first_node(); -

Description

Removes first child node. If node has no children, behaviour is undefined. Use first_node() to test if node has children.

function xml_node::remove_last_node

Synopsis

void remove_last_node(); -

Description

Removes last child of the node. If node has no children, behaviour is undefined. Use first_node() to test if node has children.

function xml_node::remove_node

Synopsis

void remove_node(xml_node< Ch > *where); -

Description

Removes specified child from the node.

function xml_node::remove_all_nodes

Synopsis

void remove_all_nodes(); -

Description

Removes all child nodes (but not attributes).

function xml_node::prepend_attribute

Synopsis

void prepend_attribute(xml_attribute< Ch > *attribute); -

Description

Prepends a new attribute to the node.

Parameters

attribute
Attribute to prepend.

function xml_node::append_attribute

Synopsis

void append_attribute(xml_attribute< Ch > *attribute); -

Description

Appends a new attribute to the node.

Parameters

attribute
Attribute to append.

function xml_node::insert_attribute

Synopsis

void insert_attribute(xml_attribute< Ch > *where, xml_attribute< Ch > *attribute); -

Description

Inserts a new attribute at specified place inside the node. All attributes after and including the specified attribute are moved one position back.

Parameters

where
Place where to insert the attribute, or 0 to insert at the back.
attribute
Attribute to insert.

function xml_node::remove_first_attribute

Synopsis

void remove_first_attribute(); -

Description

Removes first attribute of the node. If node has no attributes, behaviour is undefined. Use first_attribute() to test if node has attributes.

function xml_node::remove_last_attribute

Synopsis

void remove_last_attribute(); -

Description

Removes last attribute of the node. If node has no attributes, behaviour is undefined. Use first_attribute() to test if node has attributes.

function xml_node::remove_attribute

Synopsis

void remove_attribute(xml_attribute< Ch > *where); -

Description

Removes specified attribute from node.

Parameters

where
Pointer to attribute to be removed.

function xml_node::remove_all_attributes

Synopsis

void remove_all_attributes(); -

Description

Removes all attributes of node.

enum node_type

Description

Enumeration listing all node types produced by the parser. Use xml_node::type() function to query node type.

Values

node_document
A document node. Name and value are empty.
node_element
An element node. Name contains element name. Value contains text of first data node.
node_data
A data node. Name is empty. Value contains data text.
node_cdata
A CDATA node. Name is empty. Value contains data text.
node_comment
A comment node. Name is empty. Value contains comment text.
node_declaration
A declaration node. Name and value are empty. Declaration parameters (version, encoding and standalone) are in node attributes.
node_doctype
A DOCTYPE node. Name is empty. Value contains DOCTYPE text.
node_pi
A PI node. Name contains target. Value contains instructions.

function parse_error_handler

Synopsis

void rapidxml::parse_error_handler(const char *what, void *where); -

Description

When exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function is called to notify user about the error. It must be defined by the user.

- This function cannot return. If it does, the results are undefined.

- A very simple definition might look like that: - void rapidxml::parse_error_handler(const char *what, void *where) - { - std::cout << "Parse error: " << what << "\n"; - std::abort(); - } -

Parameters

what
Human readable description of the error.
where
Pointer to character data where error was detected.

function print

Synopsis

OutIt rapidxml::print(OutIt out, const xml_node< Ch > &node, int flags=0); -

Description

Prints XML to given output iterator.

Parameters

out
Output iterator to print to.
node
Node to be printed. Pass xml_document to print entire document.
flags
Flags controlling how XML is printed.

Returns

Output iterator pointing to position immediately after last character of printed text.

function print

Synopsis

std::basic_ostream<Ch>& rapidxml::print(std::basic_ostream< Ch > &out, const xml_node< Ch > &node, int flags=0); -

Description

Prints XML to given output stream.

Parameters

out
Output stream to print to.
node
Node to be printed. Pass xml_document to print entire document.
flags
Flags controlling how XML is printed.

Returns

Output stream.

function operator<<

Synopsis

std::basic_ostream<Ch>& rapidxml::operator<<(std::basic_ostream< Ch > &out, const xml_node< Ch > &node); -

Description

Prints formatted XML to given output stream. Uses default printing flags. Use print() function to customize printing process.

Parameters

out
Output stream to print to.
node
Node to be printed.

Returns

Output stream.

- constant - parse_no_data_nodes

Synopsis

const int parse_no_data_nodes - = 0x1; -

Description

Parse flag instructing the parser to not create data nodes. Text of first data node will still be placed in value of parent element, unless rapidxml::parse_no_element_values flag is also specified. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_no_element_values

Synopsis

const int parse_no_element_values - = 0x2; -

Description

Parse flag instructing the parser to not use text of first data node as a value of parent element. Can be combined with other flags by use of | operator. Note that child data nodes of element node take precendence over its value when printing. That is, if element has one or more child data nodes and a value, the value will be ignored. Use rapidxml::parse_no_data_nodes flag to prevent creation of data nodes if you want to manipulate data using values of elements.

- See xml_document::parse() function.

- constant - parse_no_string_terminators

Synopsis

const int parse_no_string_terminators - = 0x4; -

Description

Parse flag instructing the parser to not place zero terminators after strings in the source text. By default zero terminators are placed, modifying source text. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_no_entity_translation

Synopsis

const int parse_no_entity_translation - = 0x8; -

Description

Parse flag instructing the parser to not translate entities in the source text. By default entities are translated, modifying source text. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_no_utf8

Synopsis

const int parse_no_utf8 - = 0x10; -

Description

Parse flag instructing the parser to disable UTF-8 handling and assume plain 8 bit characters. By default, UTF-8 handling is enabled. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_declaration_node

Synopsis

const int parse_declaration_node - = 0x20; -

Description

Parse flag instructing the parser to create XML declaration node. By default, declaration node is not created. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_comment_nodes

Synopsis

const int parse_comment_nodes - = 0x40; -

Description

Parse flag instructing the parser to create comments nodes. By default, comment nodes are not created. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_doctype_node

Synopsis

const int parse_doctype_node - = 0x80; -

Description

Parse flag instructing the parser to create DOCTYPE node. By default, doctype node is not created. Although W3C specification allows at most one DOCTYPE node, RapidXml will silently accept documents with more than one. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_pi_nodes

Synopsis

const int parse_pi_nodes - = 0x100; -

Description

Parse flag instructing the parser to create PI nodes. By default, PI nodes are not created. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_validate_closing_tags

Synopsis

const int parse_validate_closing_tags - = 0x200; -

Description

Parse flag instructing the parser to validate closing tag names. If not set, name inside closing tag is irrelevant to the parser. By default, closing tags are not validated. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_trim_whitespace

Synopsis

const int parse_trim_whitespace - = 0x400; -

Description

Parse flag instructing the parser to trim all leading and trailing whitespace of data nodes. By default, whitespace is not trimmed. This flag does not cause the parser to modify source text. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_normalize_whitespace

Synopsis

const int parse_normalize_whitespace - = 0x800; -

Description

Parse flag instructing the parser to condense all whitespace runs of data nodes to a single space character. Trimming of leading and trailing whitespace of data is controlled by rapidxml::parse_trim_whitespace flag. By default, whitespace is not normalized. If this flag is specified, source text will be modified. Can be combined with other flags by use of | operator.

- See xml_document::parse() function.

- constant - parse_default

Synopsis

const int parse_default - = 0; -

Description

Parse flags which represent default behaviour of the parser. This is always equal to 0, so that all other flags can be simply ored together. Normally there is no need to inconveniently disable flags by anding with their negated (~) values. This also means that meaning of each flag is a negation of the default setting. For example, if flag name is rapidxml::parse_no_utf8, it means that utf-8 is enabled by default, and using the flag will disable it.

- See xml_document::parse() function.

- constant - parse_non_destructive

Synopsis

const int parse_non_destructive - = parse_no_string_terminators | parse_no_entity_translation; -

Description

A combination of parse flags that forbids any modifications of the source text. This also results in faster parsing. However, note that the following will occur:
  • names and values of nodes will not be zero terminated, you have to use xml_base::name_size() and xml_base::value_size() functions to determine where name and value ends
  • entities will not be translated
  • whitespace will not be normalized
-See xml_document::parse() function.

- constant - parse_fastest

Synopsis

const int parse_fastest - = parse_non_destructive | parse_no_data_nodes; -

Description

A combination of parse flags resulting in fastest possible parsing, without sacrificing important data.

- See xml_document::parse() function.

- constant - parse_full

Synopsis

const int parse_full - = parse_declaration_node | parse_comment_nodes | parse_doctype_node | parse_pi_nodes | parse_validate_closing_tags; -

Description

A combination of parse flags resulting in largest amount of data being extracted. This usually results in slowest parsing.

- See xml_document::parse() function.

- constant - print_no_indenting

Synopsis

const int print_no_indenting - = 0x1; +

RAPIDXML Manual

Version 1.13

Copyright (C) 2006, 2009 Marcin Kalicinski
See accompanying file license.txt for license information.

Table of Contents

1. What is RapidXml?
1.1 Dependencies And Compatibility
1.2 Character Types And Encodings
1.3 Error Handling
1.4 Memory Allocation
1.5 W3C Compliance
1.6 API Design
1.7 Reliability
1.8 Acknowledgements
2. Two Minute Tutorial
2.1 Parsing
2.2 Accessing The DOM Tree
2.3 Modifying The DOM Tree
2.4 Printing XML
3. Differences From Regular XML Parsers
3.1 Lifetime Of Source Text
3.2 Ownership Of Strings
3.3 Destructive Vs Non-Destructive Mode
4. Performance
4.1 Comparison With Other Parsers
5. Reference

1. What is RapidXml?

RapidXml is an attempt to create the fastest XML DOM parser possible, while retaining useability, portability and reasonable W3C compatibility. It is an in-situ parser written in C++, with parsing speed approaching that of strlen() function executed on the same data.

+ Entire parser is contained in a single header file, so no building or linking is neccesary. To use it you just need to copy rapidxml.hpp file to a convenient place (such as your project directory), and include it where needed. You may also want to use printing functions contained in header rapidxml_print.hpp.

1.1 Dependencies And Compatibility

RapidXml has no dependencies other than a very small subset of standard C++ library (<cassert>, <cstdlib>, <new> and <exception>, unless exceptions are disabled). It should compile on any reasonably conformant compiler, and was tested on Visual C++ 2003, Visual C++ 2005, Visual C++ 2008, gcc 3, gcc 4, and Comeau 4.3.3. Care was taken that no warnings are produced on these compilers, even with highest warning levels enabled.

1.2 Character Types And Encodings

RapidXml is character type agnostic, and can work both with narrow and wide characters. Current version does not fully support UTF-16 or UTF-32, so use of wide characters is somewhat incapacitated. However, it should succesfully parse wchar_t strings containing UTF-16 or UTF-32 if endianness of the data matches that of the machine. UTF-8 is fully supported, including all numeric character references, which are expanded into appropriate UTF-8 byte sequences (unless you enable parse_no_utf8 flag).

+ Note that RapidXml performs no decoding - strings returned by name() and value() functions will contain text encoded using the same encoding as source file. Rapidxml understands and expands the following character references: &apos; &amp; &quot; &lt; &gt; &#...; Other character references are not expanded.

1.3 Error Handling

By default, RapidXml uses C++ exceptions to report errors. If this behaviour is undesirable, RAPIDXML_NO_EXCEPTIONS can be defined to suppress exception code. See parse_error class and parse_error_handler() function for more information.

1.4 Memory Allocation

RapidXml uses a special memory pool object to allocate nodes and attributes, because direct allocation using new operator would be far too slow. Underlying memory allocations performed by the pool can be customized by use of memory_pool::set_allocator() function. See class memory_pool for more information.

1.5 W3C Compliance

RapidXml is not a W3C compliant parser, primarily because it ignores DOCTYPE declarations. There is a number of other, minor incompatibilities as well. Still, it can successfully parse and produce complete trees of all valid XML files in W3C conformance suite (over 1000 files specially designed to find flaws in XML processors). In destructive mode it performs whitespace normalization and character entity substitution for a small set of built-in entities.

1.6 API Design

RapidXml API is minimalistic, to reduce code size as much as possible, and facilitate use in embedded environments. Additional convenience functions are provided in separate headers: rapidxml_utils.hpp and rapidxml_print.hpp. Contents of these headers is not an essential part of the library, and is currently not documented (otherwise than with comments in code).

1.7 Reliability

RapidXml is very robust and comes with a large harness of unit tests. Special care has been taken to ensure stability of the parser no matter what source text is thrown at it. One of the unit tests produces 100,000 randomly corrupted variants of XML document, which (when uncorrupted) contains all constructs recognized by RapidXml. RapidXml passes this test when it correctly recognizes that errors have been introduced, and does not crash or loop indefinitely.

+ Another unit test puts RapidXml head-to-head with another, well estabilished XML parser, and verifies that their outputs match across a wide variety of small and large documents.

+ Yet another test feeds RapidXml with over 1000 test files from W3C compliance suite, and verifies that correct results are obtained. There are also additional tests that verify each API function separately, and test that various parsing modes work as expected.

1.8 Acknowledgements

I would like to thank Arseny Kapoulkine for his work on pugixml, which was an inspiration for this project. Additional thanks go to Kristen Wegner for creating pugxml, from which pugixml was derived. Janusz Wohlfeil kindly ran RapidXml speed tests on hardware that I did not have access to, allowing me to expand performance comparison table.

2. Two Minute Tutorial

2.1 Parsing

The following code causes RapidXml to parse a zero-terminated string named text: 
using namespace rapidxml;
+xml_document<> doc;    // character type defaults to char
+doc.parse<0>(text);    // 0 means default parse flags
+
doc object is now a root of DOM tree containing representation of the parsed XML. Because all RapidXml interface is contained inside namespace rapidxml, users must either bring contents of this namespace into scope, or fully qualify all the names. Class xml_document represents a root of the DOM hierarchy. By means of public inheritance, it is also an xml_node and a memory_pool. Template parameter of xml_document::parse() function is used to specify parsing flags, with which you can fine-tune behaviour of the parser. Note that flags must be a compile-time constant.

2.2 Accessing The DOM Tree

To access the DOM tree, use methods of xml_node and xml_attribute classes: 
cout << "Name of my first node is: " << doc.first_node()->name() << "\n";
+xml_node<> *node = doc.first_node("foobar");
+cout << "Node foobar has value " << node->value() << "\n";
+for (xml_attribute<> *attr = node->first_attribute();
+     attr; attr = attr->next_attribute())
+{
+    cout << "Node foobar has attribute " << attr->name() << " ";
+    cout << "with value " << attr->value() << "\n";
+}
+

2.3 Modifying The DOM Tree

DOM tree produced by the parser is fully modifiable. Nodes and attributes can be added/removed, and their contents changed. The below example creates a HTML document, whose sole contents is a link to google.com website: 
xml_document<> doc;
+xml_node<> *node = doc.allocate_node(node_element, "a", "Google");
+doc.append_node(node);
+xml_attribute<> *attr = doc.allocate_attribute("href", "google.com");
+node->append_attribute(attr);
+
One quirk is that nodes and attributes do not own the text of their names and values. This is because normally they only store pointers to the source text. So, when assigning a new name or value to the node, care must be taken to ensure proper lifetime of the string. The easiest way to achieve it is to allocate the string from the xml_document memory pool. In the above example this is not necessary, because we are only assigning character constants. But the code below uses memory_pool::allocate_string() function to allocate node name (which will have the same lifetime as the document), and assigns it to a new node: 
xml_document<> doc;
+char *node_name = doc.allocate_string(name);        // Allocate string and copy name into it
+xml_node<> *node = doc.allocate_node(node_element, node_name);  // Set node name to node_name
+
Check Reference section for description of the entire interface.

2.4 Printing XML

You can print xml_document and xml_node objects into an XML string. Use print() function or operator <<, which are defined in rapidxml_print.hpp header. 
using namespace rapidxml;
+xml_document<> doc;    // character type defaults to char
+// ... some code to fill the document
+
+// Print to stream using operator <<
+std::cout << doc;   
+
+// Print to stream using print function, specifying printing flags
+print(std::cout, doc, 0);   // 0 means default printing flags
+
+// Print to string using output iterator
+std::string s;
+print(std::back_inserter(s), doc, 0);
+
+// Print to memory buffer using output iterator
+char buffer[4096];                      // You are responsible for making the buffer large enough!
+char *end = print(buffer, doc, 0);      // end contains pointer to character after last printed character
+*end = 0;                               // Add string terminator after XML
+

3. Differences From Regular XML Parsers

RapidXml is an in-situ parser, which allows it to achieve very high parsing speed. In-situ means that parser does not make copies of strings. Instead, it places pointers to the source text in the DOM hierarchy.

3.1 Lifetime Of Source Text

In-situ parsing requires that source text lives at least as long as the document object. If source text is destroyed, names and values of nodes in DOM tree will become destroyed as well. Additionally, whitespace processing, character entity translation, and zero-termination of strings require that source text be modified during parsing (but see non-destructive mode). This makes the text useless for further processing once it was parsed by RapidXml.

+ In many cases however, these are not serious issues.

3.2 Ownership Of Strings

Nodes and attributes produced by RapidXml do not own their name and value strings. They merely hold the pointers to them. This means you have to be careful when setting these values manually, by using xml_base::name(const Ch *) or xml_base::value(const Ch *) functions. Care must be taken to ensure that lifetime of the string passed is at least as long as lifetime of the node/attribute. The easiest way to achieve it is to allocate the string from memory_pool owned by the document. Use memory_pool::allocate_string() function for this purpose.

3.3 Destructive Vs Non-Destructive Mode

By default, the parser modifies source text during the parsing process. This is required to achieve character entity translation, whitespace normalization, and zero-termination of strings.

+ In some cases this behaviour may be undesirable, for example if source text resides in read only memory, or is mapped to memory directly from file. By using appropriate parser flags (parse_non_destructive), source text modifications can be disabled. However, because RapidXml does in-situ parsing, it obviously has the following side-effects:

4. Performance

RapidXml achieves its speed through use of several techniques:
  • In-situ parsing. When building DOM tree, RapidXml does not make copies of string data, such as node names and values. Instead, it stores pointers to interior of the source text.
  • Use of template metaprogramming techniques. This allows it to move much of the work to compile time. Through magic of the templates, C++ compiler generates a separate copy of parsing code for any combination of parser flags you use. In each copy, all possible decisions are made at compile time and all unused code is omitted.
  • Extensive use of lookup tables for parsing.
  • Hand-tuned C++ with profiling done on several most popular CPUs.
This results in a very small and fast code: a parser which is custom tailored to exact needs with each invocation.

4.1 Comparison With Other Parsers

The table below compares speed of RapidXml to some other parsers, and to strlen() function executed on the same data. On a modern CPU (as of 2007), you can expect parsing throughput to be close to 1 GB/s. As a rule of thumb, parsing speed is about 50-100x faster than Xerces DOM, 30-60x faster than TinyXml, 3-12x faster than pugxml, and about 5% - 30% faster than pugixml, the fastest XML parser I know of.
  • The test file is a real-world, 50kB large, moderately dense XML file.
  • All timing is done by using RDTSC instruction present in Pentium-compatible CPUs.
  • No profile-guided optimizations are used.
  • All parsers are running in their fastest modes.
  • The results are given in CPU cycles per character, so frequency of CPUs is irrelevant.
  • The results are minimum values from a large number of runs, to minimize effects of operating system activity, task switching, interrupt handling etc.
  • A single parse of the test file takes about 1/10th of a millisecond, so with large number of runs there is a good chance of hitting at least one no-interrupt streak, and obtaining undisturbed results.
Platform
Compiler
strlen() RapidXml pugixml 0.3 pugxml TinyXml
Pentium 4
MSVC 8.0
2.5
5.4
7.0
61.7
298.8
Pentium 4
gcc 4.1.1
0.8
6.1
9.5
67.0
413.2
Core 2
MSVC 8.0
1.0
4.5
5.0
24.6
154.8
Core 2
gcc 4.1.1
0.6
4.6
5.4
28.3
229.3
Athlon XP
MSVC 8.0
3.1
7.7
8.0
25.5
182.6
Athlon XP
gcc 4.1.1
0.9
8.2
9.2
33.7
265.2
Pentium 3
MSVC 8.0
2.0
6.3
7.0
30.9
211.9
Pentium 3
gcc 4.1.1
1.0
6.7
8.9
35.3
316.0
(*) All results are in CPU cycles per character of source text

5. Reference

This section lists all classes, functions, constants etc. and describes them in detail.
class + template + rapidxml::memory_pool
+ constructor + memory_pool()
+ destructor + ~memory_pool()
function allocate_node(node_type type, const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0)
function allocate_attribute(const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0)
function allocate_string(const Ch *source=0, std::size_t size=0)
function clone_node(const xml_node< Ch > *source, xml_node< Ch > *result=0)
function clear()
function set_allocator(alloc_func *af, free_func *ff)

class rapidxml::parse_error
+ constructor + parse_error(const char *what, void *where)
function what() const
function where() const

class + template + rapidxml::xml_attribute
+ constructor + xml_attribute()
function document() const
function previous_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function next_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const

class + template + rapidxml::xml_base
+ constructor + xml_base()
function name() const
function name_size() const
function value() const
function value_size() const
function name(const Ch *name, std::size_t size)
function name(const Ch *name)
function value(const Ch *value, std::size_t size)
function value(const Ch *value)
function parent() const

class + template + rapidxml::xml_document
+ constructor + xml_document()
function parse(Ch *text)
function clear()

class + template + rapidxml::xml_node
+ constructor + xml_node(node_type type)
function type() const
function document() const
function first_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function last_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function previous_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function next_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function first_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function last_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const
function type(node_type type)
function prepend_node(xml_node< Ch > *child)
function append_node(xml_node< Ch > *child)
function insert_node(xml_node< Ch > *where, xml_node< Ch > *child)
function remove_first_node()
function remove_last_node()
function remove_node(xml_node< Ch > *where)
function remove_all_nodes()
function prepend_attribute(xml_attribute< Ch > *attribute)
function append_attribute(xml_attribute< Ch > *attribute)
function insert_attribute(xml_attribute< Ch > *where, xml_attribute< Ch > *attribute)
function remove_first_attribute()
function remove_last_attribute()
function remove_attribute(xml_attribute< Ch > *where)
function remove_all_attributes()

namespace rapidxml
enum node_type
function parse_error_handler(const char *what, void *where)
function print(OutIt out, const xml_node< Ch > &node, int flags=0)
function print(std::basic_ostream< Ch > &out, const xml_node< Ch > &node, int flags=0)
function operator<<(std::basic_ostream< Ch > &out, const xml_node< Ch > &node)
+ constant + parse_no_data_nodes
+ constant + parse_no_element_values
+ constant + parse_no_string_terminators
+ constant + parse_no_entity_translation
+ constant + parse_no_utf8
+ constant + parse_declaration_node
+ constant + parse_comment_nodes
+ constant + parse_doctype_node
+ constant + parse_pi_nodes
+ constant + parse_validate_closing_tags
+ constant + parse_trim_whitespace
+ constant + parse_normalize_whitespace
+ constant + parse_default
+ constant + parse_non_destructive
+ constant + parse_fastest
+ constant + parse_full
+ constant + print_no_indenting

class + template + rapidxml::memory_pool

+ + Defined in rapidxml.hpp
+ Base class for + xml_document

Description

This class is used by the parser to create new nodes and attributes, without overheads of dynamic memory allocation. In most cases, you will not need to use this class directly. However, if you need to create nodes manually or modify names/values of nodes, you are encouraged to use memory_pool of relevant xml_document to allocate the memory. Not only is this faster than allocating them by using new operator, but also their lifetime will be tied to the lifetime of document, possibly simplyfing memory management.

+ Call allocate_node() or allocate_attribute() functions to obtain new nodes or attributes from the pool. You can also call allocate_string() function to allocate strings. Such strings can then be used as names or values of nodes without worrying about their lifetime. Note that there is no free() function -- all allocations are freed at once when clear() function is called, or when the pool is destroyed.

+ It is also possible to create a standalone memory_pool, and use it to allocate nodes, whose lifetime will not be tied to any document.

+ Pool maintains RAPIDXML_STATIC_POOL_SIZE bytes of statically allocated memory. Until static memory is exhausted, no dynamic memory allocations are done. When static memory is exhausted, pool allocates additional blocks of memory of size RAPIDXML_DYNAMIC_POOL_SIZE each, by using global new[] and delete[] operators. This behaviour can be changed by setting custom allocation routines. Use set_allocator() function to set them.

+ Allocations for nodes, attributes and strings are aligned at RAPIDXML_ALIGNMENT bytes. This value defaults to the size of pointer on target architecture.

+ To obtain absolutely top performance from the parser, it is important that all nodes are allocated from a single, contiguous block of memory. Otherwise, cache misses when jumping between two (or more) disjoint blocks of memory can slow down parsing quite considerably. If required, you can tweak RAPIDXML_STATIC_POOL_SIZE, RAPIDXML_DYNAMIC_POOL_SIZE and RAPIDXML_ALIGNMENT to obtain best wasted memory to performance compromise. To do it, define their values before rapidxml.hpp file is included.

Parameters

Ch
Character type of created nodes.

+ constructor + memory_pool::memory_pool

Synopsis

memory_pool(); +

Description

Constructs empty pool with default allocator functions.

+ destructor + memory_pool::~memory_pool

Synopsis

~memory_pool(); +

Description

Destroys pool and frees all the memory. This causes memory occupied by nodes allocated by the pool to be freed. Nodes allocated from the pool are no longer valid.

function memory_pool::allocate_node

Synopsis

xml_node<Ch>* allocate_node(node_type type, const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0); +

Description

Allocates a new node from the pool, and optionally assigns name and value to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

type
Type of node to create.
name
Name to assign to the node, or 0 to assign no name.
value
Value to assign to the node, or 0 to assign no value.
name_size
Size of name to assign, or 0 to automatically calculate size from name string.
value_size
Size of value to assign, or 0 to automatically calculate size from value string.

Returns

Pointer to allocated node. This pointer will never be NULL.

function memory_pool::allocate_attribute

Synopsis

xml_attribute<Ch>* allocate_attribute(const Ch *name=0, const Ch *value=0, std::size_t name_size=0, std::size_t value_size=0); +

Description

Allocates a new attribute from the pool, and optionally assigns name and value to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

name
Name to assign to the attribute, or 0 to assign no name.
value
Value to assign to the attribute, or 0 to assign no value.
name_size
Size of name to assign, or 0 to automatically calculate size from name string.
value_size
Size of value to assign, or 0 to automatically calculate size from value string.

Returns

Pointer to allocated attribute. This pointer will never be NULL.

function memory_pool::allocate_string

Synopsis

Ch* allocate_string(const Ch *source=0, std::size_t size=0); +

Description

Allocates a char array of given size from the pool, and optionally copies a given string to it. If the allocation request cannot be accomodated, this function will throw std::bad_alloc. If exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function will call rapidxml::parse_error_handler() function.

Parameters

source
String to initialize the allocated memory with, or 0 to not initialize it.
size
Number of characters to allocate, or zero to calculate it automatically from source string length; if size is 0, source string must be specified and null terminated.

Returns

Pointer to allocated char array. This pointer will never be NULL.

function memory_pool::clone_node

Synopsis

xml_node<Ch>* clone_node(const xml_node< Ch > *source, xml_node< Ch > *result=0); +

Description

Clones an xml_node and its hierarchy of child nodes and attributes. Nodes and attributes are allocated from this memory pool. Names and values are not cloned, they are shared between the clone and the source. Result node can be optionally specified as a second parameter, in which case its contents will be replaced with cloned source node. This is useful when you want to clone entire document.

Parameters

source
Node to clone.
result
Node to put results in, or 0 to automatically allocate result node

Returns

Pointer to cloned node. This pointer will never be NULL.

function memory_pool::clear

Synopsis

void clear(); +

Description

Clears the pool. This causes memory occupied by nodes allocated by the pool to be freed. Any nodes or strings allocated from the pool will no longer be valid.

function memory_pool::set_allocator

Synopsis

void set_allocator(alloc_func *af, free_func *ff); +

Description

Sets or resets the user-defined memory allocation functions for the pool. This can only be called when no memory is allocated from the pool yet, otherwise results are undefined. Allocation function must not return invalid pointer on failure. It should either throw, stop the program, or use longjmp() function to pass control to other place of program. If it returns invalid pointer, results are undefined.

+ User defined allocation functions must have the following forms:

+void *allocate(std::size_t size);
+void free(void *pointer);

Parameters

af
Allocation function, or 0 to restore default function
ff
Free function, or 0 to restore default function

class rapidxml::parse_error

+ + Defined in rapidxml.hpp

Description

Parse error exception. This exception is thrown by the parser when an error occurs. Use what() function to get human-readable error message. Use where() function to get a pointer to position within source text where error was detected.

+ If throwing exceptions by the parser is undesirable, it can be disabled by defining RAPIDXML_NO_EXCEPTIONS macro before rapidxml.hpp is included. This will cause the parser to call rapidxml::parse_error_handler() function instead of throwing an exception. This function must be defined by the user.

+ This class derives from std::exception class.

+ constructor + parse_error::parse_error

Synopsis

parse_error(const char *what, void *where); +

Description

Constructs parse error.

function parse_error::what

Synopsis

virtual const char* what() const; +

Description

Gets human readable description of error.

Returns

Pointer to null terminated description of the error.

function parse_error::where

Synopsis

Ch* where() const; +

Description

Gets pointer to character data where error happened. Ch should be the same as char type of xml_document that produced the error.

Returns

Pointer to location within the parsed string where error occured.

class + template + rapidxml::xml_attribute

+ + Defined in rapidxml.hpp
+ Inherits from + xml_base

Description

Class representing attribute node of XML document. Each attribute has name and value strings, which are available through name() and value() functions (inherited from xml_base). Note that after parse, both name and value of attribute will point to interior of source text used for parsing. Thus, this text must persist in memory for the lifetime of attribute.

Parameters

Ch
Character type to use.

+ constructor + xml_attribute::xml_attribute

Synopsis

xml_attribute(); +

Description

Constructs an empty attribute with the specified type. Consider using memory_pool of appropriate xml_document if allocating attributes manually.

function xml_attribute::document

Synopsis

xml_document<Ch>* document() const; +

Description

Gets document of which attribute is a child.

Returns

Pointer to document that contains this attribute, or 0 if there is no parent document.

function xml_attribute::previous_attribute

Synopsis

xml_attribute<Ch>* previous_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets previous attribute, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return previous attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_attribute::next_attribute

Synopsis

xml_attribute<Ch>* next_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets next attribute, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return next attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

class + template + rapidxml::xml_base

+ + Defined in rapidxml.hpp
+ Base class for + xml_attribute xml_node

Description

Base class for xml_node and xml_attribute implementing common functions: name(), name_size(), value(), value_size() and parent().

Parameters

Ch
Character type to use

+ constructor + xml_base::xml_base

Synopsis

xml_base(); +

function xml_base::name

Synopsis

Ch* name() const; +

Description

Gets name of the node. Interpretation of name depends on type of node. Note that name will not be zero-terminated if rapidxml::parse_no_string_terminators option was selected during parse.

+ Use name_size() function to determine length of the name.

Returns

Name of node, or empty string if node has no name.

function xml_base::name_size

Synopsis

std::size_t name_size() const; +

Description

Gets size of node name, not including terminator character. This function works correctly irrespective of whether name is or is not zero terminated.

Returns

Size of node name, in characters.

function xml_base::value

Synopsis

Ch* value() const; +

Description

Gets value of node. Interpretation of value depends on type of node. Note that value will not be zero-terminated if rapidxml::parse_no_string_terminators option was selected during parse.

+ Use value_size() function to determine length of the value.

Returns

Value of node, or empty string if node has no value.

function xml_base::value_size

Synopsis

std::size_t value_size() const; +

Description

Gets size of node value, not including terminator character. This function works correctly irrespective of whether value is or is not zero terminated.

Returns

Size of node value, in characters.

function xml_base::name

Synopsis

void name(const Ch *name, std::size_t size); +

Description

Sets name of node to a non zero-terminated string. See Ownership Of Strings .

+ Note that node does not own its name or value, it only stores a pointer to it. It will not delete or otherwise free the pointer on destruction. It is reponsibility of the user to properly manage lifetime of the string. The easiest way to achieve it is to use memory_pool of the document to allocate the string - on destruction of the document the string will be automatically freed.

+ Size of name must be specified separately, because name does not have to be zero terminated. Use name(const Ch *) function to have the length automatically calculated (string must be zero terminated).

Parameters

name
Name of node to set. Does not have to be zero terminated.
size
Size of name, in characters. This does not include zero terminator, if one is present.

function xml_base::name

Synopsis

void name(const Ch *name); +

Description

Sets name of node to a zero-terminated string. See also Ownership Of Strings and xml_node::name(const Ch *, std::size_t).

Parameters

name
Name of node to set. Must be zero terminated.

function xml_base::value

Synopsis

void value(const Ch *value, std::size_t size); +

Description

Sets value of node to a non zero-terminated string. See Ownership Of Strings .

+ Note that node does not own its name or value, it only stores a pointer to it. It will not delete or otherwise free the pointer on destruction. It is reponsibility of the user to properly manage lifetime of the string. The easiest way to achieve it is to use memory_pool of the document to allocate the string - on destruction of the document the string will be automatically freed.

+ Size of value must be specified separately, because it does not have to be zero terminated. Use value(const Ch *) function to have the length automatically calculated (string must be zero terminated).

+ If an element has a child node of type node_data, it will take precedence over element value when printing. If you want to manipulate data of elements using values, use parser flag rapidxml::parse_no_data_nodes to prevent creation of data nodes by the parser.

Parameters

value
value of node to set. Does not have to be zero terminated.
size
Size of value, in characters. This does not include zero terminator, if one is present.

function xml_base::value

Synopsis

void value(const Ch *value); +

Description

Sets value of node to a zero-terminated string. See also Ownership Of Strings and xml_node::value(const Ch *, std::size_t).

Parameters

value
Vame of node to set. Must be zero terminated.

function xml_base::parent

Synopsis

xml_node<Ch>* parent() const; +

Description

Gets node parent.

Returns

Pointer to parent node, or 0 if there is no parent.

class + template + rapidxml::xml_document

+ + Defined in rapidxml.hpp
+ Inherits from + xml_node memory_pool

Description

This class represents root of the DOM hierarchy. It is also an xml_node and a memory_pool through public inheritance. Use parse() function to build a DOM tree from a zero-terminated XML text string. parse() function allocates memory for nodes and attributes by using functions of xml_document, which are inherited from memory_pool. To access root node of the document, use the document itself, as if it was an xml_node.

Parameters

Ch
Character type to use.

+ constructor + xml_document::xml_document

Synopsis

xml_document(); +

Description

Constructs empty XML document.

function xml_document::parse

Synopsis

void parse(Ch *text); +

Description

Parses zero-terminated XML string according to given flags. Passed string will be modified by the parser, unless rapidxml::parse_non_destructive flag is used. The string must persist for the lifetime of the document. In case of error, rapidxml::parse_error exception will be thrown.

+ If you want to parse contents of a file, you must first load the file into the memory, and pass pointer to its beginning. Make sure that data is zero-terminated.

+ Document can be parsed into multiple times. Each new call to parse removes previous nodes and attributes (if any), but does not clear memory pool.

Parameters

text
XML data to parse; pointer is non-const to denote fact that this data may be modified by the parser.

function xml_document::clear

Synopsis

void clear(); +

Description

Clears the document by deleting all nodes and clearing the memory pool. All nodes owned by document pool are destroyed.

class + template + rapidxml::xml_node

+ + Defined in rapidxml.hpp
+ Inherits from + xml_base
+ Base class for + xml_document

Description

Class representing a node of XML document. Each node may have associated name and value strings, which are available through name() and value() functions. Interpretation of name and value depends on type of the node. Type of node can be determined by using type() function.

+ Note that after parse, both name and value of node, if any, will point interior of source text used for parsing. Thus, this text must persist in the memory for the lifetime of node.

Parameters

Ch
Character type to use.

+ constructor + xml_node::xml_node

Synopsis

xml_node(node_type type); +

Description

Constructs an empty node with the specified type. Consider using memory_pool of appropriate document to allocate nodes manually.

Parameters

type
Type of node to construct.

function xml_node::type

Synopsis

node_type type() const; +

Description

Gets type of node.

Returns

Type of node.

function xml_node::document

Synopsis

xml_document<Ch>* document() const; +

Description

Gets document of which node is a child.

Returns

Pointer to document that contains this node, or 0 if there is no parent document.

function xml_node::first_node

Synopsis

xml_node<Ch>* first_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets first child node, optionally matching node name.

Parameters

name
Name of child to find, or 0 to return first child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found child, or 0 if not found.

function xml_node::last_node

Synopsis

xml_node<Ch>* last_node(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets last child node, optionally matching node name. Behaviour is undefined if node has no children. Use first_node() to test if node has children.

Parameters

name
Name of child to find, or 0 to return last child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found child, or 0 if not found.

function xml_node::previous_sibling

Synopsis

xml_node<Ch>* previous_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets previous sibling node, optionally matching node name. Behaviour is undefined if node has no parent. Use parent() to test if node has a parent.

Parameters

name
Name of sibling to find, or 0 to return previous sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found sibling, or 0 if not found.

function xml_node::next_sibling

Synopsis

xml_node<Ch>* next_sibling(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets next sibling node, optionally matching node name. Behaviour is undefined if node has no parent. Use parent() to test if node has a parent.

Parameters

name
Name of sibling to find, or 0 to return next sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found sibling, or 0 if not found.

function xml_node::first_attribute

Synopsis

xml_attribute<Ch>* first_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets first attribute of node, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return first attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_node::last_attribute

Synopsis

xml_attribute<Ch>* last_attribute(const Ch *name=0, std::size_t name_size=0, bool case_sensitive=true) const; +

Description

Gets last attribute of node, optionally matching attribute name.

Parameters

name
Name of attribute to find, or 0 to return last attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero
name_size
Size of name, in characters, or 0 to have size calculated automatically from string
case_sensitive
Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters

Returns

Pointer to found attribute, or 0 if not found.

function xml_node::type

Synopsis

void type(node_type type); +

Description

Sets type of node.

Parameters

type
Type of node to set.

function xml_node::prepend_node

Synopsis

void prepend_node(xml_node< Ch > *child); +

Description

Prepends a new child node. The prepended child becomes the first child, and all existing children are moved one position back.

Parameters

child
Node to prepend.

function xml_node::append_node

Synopsis

void append_node(xml_node< Ch > *child); +

Description

Appends a new child node. The appended child becomes the last child.

Parameters

child
Node to append.

function xml_node::insert_node

Synopsis

void insert_node(xml_node< Ch > *where, xml_node< Ch > *child); +

Description

Inserts a new child node at specified place inside the node. All children after and including the specified node are moved one position back.

Parameters

where
Place where to insert the child, or 0 to insert at the back.
child
Node to insert.

function xml_node::remove_first_node

Synopsis

void remove_first_node(); +

Description

Removes first child node. If node has no children, behaviour is undefined. Use first_node() to test if node has children.

function xml_node::remove_last_node

Synopsis

void remove_last_node(); +

Description

Removes last child of the node. If node has no children, behaviour is undefined. Use first_node() to test if node has children.

function xml_node::remove_node

Synopsis

void remove_node(xml_node< Ch > *where); +

Description

Removes specified child from the node.

function xml_node::remove_all_nodes

Synopsis

void remove_all_nodes(); +

Description

Removes all child nodes (but not attributes).

function xml_node::prepend_attribute

Synopsis

void prepend_attribute(xml_attribute< Ch > *attribute); +

Description

Prepends a new attribute to the node.

Parameters

attribute
Attribute to prepend.

function xml_node::append_attribute

Synopsis

void append_attribute(xml_attribute< Ch > *attribute); +

Description

Appends a new attribute to the node.

Parameters

attribute
Attribute to append.

function xml_node::insert_attribute

Synopsis

void insert_attribute(xml_attribute< Ch > *where, xml_attribute< Ch > *attribute); +

Description

Inserts a new attribute at specified place inside the node. All attributes after and including the specified attribute are moved one position back.

Parameters

where
Place where to insert the attribute, or 0 to insert at the back.
attribute
Attribute to insert.

function xml_node::remove_first_attribute

Synopsis

void remove_first_attribute(); +

Description

Removes first attribute of the node. If node has no attributes, behaviour is undefined. Use first_attribute() to test if node has attributes.

function xml_node::remove_last_attribute

Synopsis

void remove_last_attribute(); +

Description

Removes last attribute of the node. If node has no attributes, behaviour is undefined. Use first_attribute() to test if node has attributes.

function xml_node::remove_attribute

Synopsis

void remove_attribute(xml_attribute< Ch > *where); +

Description

Removes specified attribute from node.

Parameters

where
Pointer to attribute to be removed.

function xml_node::remove_all_attributes

Synopsis

void remove_all_attributes(); +

Description

Removes all attributes of node.

enum node_type

Description

Enumeration listing all node types produced by the parser. Use xml_node::type() function to query node type.

Values

node_document
A document node. Name and value are empty.
node_element
An element node. Name contains element name. Value contains text of first data node.
node_data
A data node. Name is empty. Value contains data text.
node_cdata
A CDATA node. Name is empty. Value contains data text.
node_comment
A comment node. Name is empty. Value contains comment text.
node_declaration
A declaration node. Name and value are empty. Declaration parameters (version, encoding and standalone) are in node attributes.
node_doctype
A DOCTYPE node. Name is empty. Value contains DOCTYPE text.
node_pi
A PI node. Name contains target. Value contains instructions.

function parse_error_handler

Synopsis

void rapidxml::parse_error_handler(const char *what, void *where); +

Description

When exceptions are disabled by defining RAPIDXML_NO_EXCEPTIONS, this function is called to notify user about the error. It must be defined by the user.

+ This function cannot return. If it does, the results are undefined.

+ A very simple definition might look like that: + void rapidxml::parse_error_handler(const char *what, void *where) + { + std::cout << "Parse error: " << what << "\n"; + std::abort(); + } +

Parameters

what
Human readable description of the error.
where
Pointer to character data where error was detected.

function print

Synopsis

OutIt rapidxml::print(OutIt out, const xml_node< Ch > &node, int flags=0); +

Description

Prints XML to given output iterator.

Parameters

out
Output iterator to print to.
node
Node to be printed. Pass xml_document to print entire document.
flags
Flags controlling how XML is printed.

Returns

Output iterator pointing to position immediately after last character of printed text.

function print

Synopsis

std::basic_ostream<Ch>& rapidxml::print(std::basic_ostream< Ch > &out, const xml_node< Ch > &node, int flags=0); +

Description

Prints XML to given output stream.

Parameters

out
Output stream to print to.
node
Node to be printed. Pass xml_document to print entire document.
flags
Flags controlling how XML is printed.

Returns

Output stream.

function operator<<

Synopsis

std::basic_ostream<Ch>& rapidxml::operator<<(std::basic_ostream< Ch > &out, const xml_node< Ch > &node); +

Description

Prints formatted XML to given output stream. Uses default printing flags. Use print() function to customize printing process.

Parameters

out
Output stream to print to.
node
Node to be printed.

Returns

Output stream.

+ constant + parse_no_data_nodes

Synopsis

const int parse_no_data_nodes + = 0x1; +

Description

Parse flag instructing the parser to not create data nodes. Text of first data node will still be placed in value of parent element, unless rapidxml::parse_no_element_values flag is also specified. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_no_element_values

Synopsis

const int parse_no_element_values + = 0x2; +

Description

Parse flag instructing the parser to not use text of first data node as a value of parent element. Can be combined with other flags by use of | operator. Note that child data nodes of element node take precendence over its value when printing. That is, if element has one or more child data nodes and a value, the value will be ignored. Use rapidxml::parse_no_data_nodes flag to prevent creation of data nodes if you want to manipulate data using values of elements.

+ See xml_document::parse() function.

+ constant + parse_no_string_terminators

Synopsis

const int parse_no_string_terminators + = 0x4; +

Description

Parse flag instructing the parser to not place zero terminators after strings in the source text. By default zero terminators are placed, modifying source text. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_no_entity_translation

Synopsis

const int parse_no_entity_translation + = 0x8; +

Description

Parse flag instructing the parser to not translate entities in the source text. By default entities are translated, modifying source text. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_no_utf8

Synopsis

const int parse_no_utf8 + = 0x10; +

Description

Parse flag instructing the parser to disable UTF-8 handling and assume plain 8 bit characters. By default, UTF-8 handling is enabled. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_declaration_node

Synopsis

const int parse_declaration_node + = 0x20; +

Description

Parse flag instructing the parser to create XML declaration node. By default, declaration node is not created. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_comment_nodes

Synopsis

const int parse_comment_nodes + = 0x40; +

Description

Parse flag instructing the parser to create comments nodes. By default, comment nodes are not created. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_doctype_node

Synopsis

const int parse_doctype_node + = 0x80; +

Description

Parse flag instructing the parser to create DOCTYPE node. By default, doctype node is not created. Although W3C specification allows at most one DOCTYPE node, RapidXml will silently accept documents with more than one. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_pi_nodes

Synopsis

const int parse_pi_nodes + = 0x100; +

Description

Parse flag instructing the parser to create PI nodes. By default, PI nodes are not created. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_validate_closing_tags

Synopsis

const int parse_validate_closing_tags + = 0x200; +

Description

Parse flag instructing the parser to validate closing tag names. If not set, name inside closing tag is irrelevant to the parser. By default, closing tags are not validated. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_trim_whitespace

Synopsis

const int parse_trim_whitespace + = 0x400; +

Description

Parse flag instructing the parser to trim all leading and trailing whitespace of data nodes. By default, whitespace is not trimmed. This flag does not cause the parser to modify source text. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_normalize_whitespace

Synopsis

const int parse_normalize_whitespace + = 0x800; +

Description

Parse flag instructing the parser to condense all whitespace runs of data nodes to a single space character. Trimming of leading and trailing whitespace of data is controlled by rapidxml::parse_trim_whitespace flag. By default, whitespace is not normalized. If this flag is specified, source text will be modified. Can be combined with other flags by use of | operator.

+ See xml_document::parse() function.

+ constant + parse_default

Synopsis

const int parse_default + = 0; +

Description

Parse flags which represent default behaviour of the parser. This is always equal to 0, so that all other flags can be simply ored together. Normally there is no need to inconveniently disable flags by anding with their negated (~) values. This also means that meaning of each flag is a negation of the default setting. For example, if flag name is rapidxml::parse_no_utf8, it means that utf-8 is enabled by default, and using the flag will disable it.

+ See xml_document::parse() function.

+ constant + parse_non_destructive

Synopsis

const int parse_non_destructive + = parse_no_string_terminators | parse_no_entity_translation; +

Description

A combination of parse flags that forbids any modifications of the source text. This also results in faster parsing. However, note that the following will occur:
  • names and values of nodes will not be zero terminated, you have to use xml_base::name_size() and xml_base::value_size() functions to determine where name and value ends
  • entities will not be translated
  • whitespace will not be normalized
+See xml_document::parse() function.

+ constant + parse_fastest

Synopsis

const int parse_fastest + = parse_non_destructive | parse_no_data_nodes; +

Description

A combination of parse flags resulting in fastest possible parsing, without sacrificing important data.

+ See xml_document::parse() function.

+ constant + parse_full

Synopsis

const int parse_full + = parse_declaration_node | parse_comment_nodes | parse_doctype_node | parse_pi_nodes | parse_validate_closing_tags; +

Description

A combination of parse flags resulting in largest amount of data being extracted. This usually results in slowest parsing.

+ See xml_document::parse() function.

+ constant + print_no_indenting

Synopsis

const int print_no_indenting + = 0x1;

Description

Printer flag instructing the printer to suppress indenting of XML. See print() function.

\ No newline at end of file diff --git a/noarch/include/cereal/external/rapidxml/rapidxml.hpp b/noarch/include/cereal/external/rapidxml/rapidxml.hpp index 61013a94..d82893a6 100644 --- a/noarch/include/cereal/external/rapidxml/rapidxml.hpp +++ b/noarch/include/cereal/external/rapidxml/rapidxml.hpp @@ -85,7 +85,7 @@ namespace rapidxml //! Gets human readable description of error. //! \return Pointer to null terminated description of the error. - virtual const char *what() const throw() + virtual const char *what() const CEREAL_NOEXCEPT override { return m_what; } @@ -317,7 +317,7 @@ namespace rapidxml const Ch *tmp = p; while (*tmp) ++tmp; - return tmp - p; + return static_cast(tmp - p); } // Compare strings for equality @@ -387,7 +387,7 @@ namespace rapidxml //! If required, you can tweak CEREAL_RAPIDXML_STATIC_POOL_SIZE, CEREAL_RAPIDXML_DYNAMIC_POOL_SIZE and CEREAL_RAPIDXML_ALIGNMENT //! to obtain best wasted memory to performance compromise. //! To do it, define their values before rapidxml.hpp file is included. - //! \param Ch Character type of created nodes. + //! \tparam Ch Character type of created nodes. template class memory_pool { @@ -656,7 +656,7 @@ namespace rapidxml //! Base class for xml_node and xml_attribute implementing common functions: //! name(), name_size(), value(), value_size() and parent(). - //! \param Ch Character type to use + //! \tparam Ch Character type to use template class xml_base { @@ -729,7 +729,7 @@ namespace rapidxml //!

//! Size of name must be specified separately, because name does not have to be zero terminated. //! Use name(const Ch *) function to have the length automatically calculated (string must be zero terminated). - //! \param name Name of node to set. Does not have to be zero terminated. + //! \param name_ Name of node to set. Does not have to be zero terminated. //! \param size Size of name, in characters. This does not include zero terminator, if one is present. void name(const Ch *name_, std::size_t size) { @@ -739,7 +739,7 @@ namespace rapidxml //! Sets name of node to a zero-terminated string. //! See also \ref ownership_of_strings and xml_node::name(const Ch *, std::size_t). - //! \param name Name of node to set. Must be zero terminated. + //! \param name_ Name of node to set. Must be zero terminated. void name(const Ch *name_) { this->name(name_, internal::measure(name_)); @@ -759,7 +759,7 @@ namespace rapidxml //!

//! If an element has a child node of type node_data, it will take precedence over element value when printing. //! If you want to manipulate data of elements using values, use parser flag rapidxml::parse_no_data_nodes to prevent creation of data nodes by the parser. - //! \param value value of node to set. Does not have to be zero terminated. + //! \param value_ value of node to set. Does not have to be zero terminated. //! \param size Size of value, in characters. This does not include zero terminator, if one is present. void value(const Ch *value_, std::size_t size) { @@ -769,7 +769,7 @@ namespace rapidxml //! Sets value of node to a zero-terminated string. //! See also \ref ownership_of_strings and xml_node::value(const Ch *, std::size_t). - //! \param value Vame of node to set. Must be zero terminated. + //! \param value_ Vame of node to set. Must be zero terminated. void value(const Ch *value_) { this->value(value_, internal::measure(value_)); @@ -806,7 +806,7 @@ namespace rapidxml //! Each attribute has name and value strings, which are available through name() and value() functions (inherited from xml_base). //! Note that after parse, both name and value of attribute will point to interior of source text used for parsing. //! Thus, this text must persist in memory for the lifetime of attribute. - //! \param Ch Character type to use. + //! \tparam Ch Character type to use. template class xml_attribute: public xml_base { @@ -862,8 +862,8 @@ namespace rapidxml } //! Gets next attribute, optionally matching attribute name. - //! \param name Name of attribute to find, or 0 to return next attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero - //! \param name_size Size of name, in characters, or 0 to have size calculated automatically from string + //! \param name_ Name of attribute to find, or 0 to return next attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero + //! \param name_size_ Size of name, in characters, or 0 to have size calculated automatically from string //! \param case_sensitive Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters //! \return Pointer to found attribute, or 0 if not found. xml_attribute *next_attribute(const Ch *name_ = 0, std::size_t name_size_ = 0, bool case_sensitive = true) const @@ -898,7 +898,7 @@ namespace rapidxml //!

//! Note that after parse, both name and value of node, if any, will point interior of source text used for parsing. //! Thus, this text must persist in the memory for the lifetime of node. - //! \param Ch Character type to use. + //! \tparam Ch Character type to use. template class xml_node: public xml_base { @@ -910,7 +910,7 @@ namespace rapidxml //! Constructs an empty node with the specified type. //! Consider using memory_pool of appropriate document to allocate nodes manually. - //! \param type Type of node to construct. + //! \param type_ Type of node to construct. xml_node(node_type type_) : m_type(type_) , m_first_node(0) @@ -942,8 +942,8 @@ namespace rapidxml } //! Gets first child node, optionally matching node name. - //! \param name Name of child to find, or 0 to return first child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero - //! \param name_size Size of name, in characters, or 0 to have size calculated automatically from string + //! \param name_ Name of child to find, or 0 to return first child regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero + //! \param name_size_ Size of name, in characters, or 0 to have size calculated automatically from string //! \param case_sensitive Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters //! \return Pointer to found child, or 0 if not found. xml_node *first_node(const Ch *name_ = 0, std::size_t name_size_ = 0, bool case_sensitive = true) const @@ -1010,8 +1010,8 @@ namespace rapidxml //! Gets next sibling node, optionally matching node name. //! Behaviour is undefined if node has no parent. //! Use parent() to test if node has a parent. - //! \param name Name of sibling to find, or 0 to return next sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero - //! \param name_size Size of name, in characters, or 0 to have size calculated automatically from string + //! \param name_ Name of sibling to find, or 0 to return next sibling regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero + //! \param name_size_ Size of name, in characters, or 0 to have size calculated automatically from string //! \param case_sensitive Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters //! \return Pointer to found sibling, or 0 if not found. xml_node *next_sibling(const Ch *name_ = 0, std::size_t name_size_ = 0, bool case_sensitive = true) const @@ -1031,8 +1031,8 @@ namespace rapidxml } //! Gets first attribute of node, optionally matching attribute name. - //! \param name Name of attribute to find, or 0 to return first attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero - //! \param name_size Size of name, in characters, or 0 to have size calculated automatically from string + //! \param name_ Name of attribute to find, or 0 to return first attribute regardless of its name; this string doesn't have to be zero-terminated if name_size is non-zero + //! \param name_size_ Size of name, in characters, or 0 to have size calculated automatically from string //! \param case_sensitive Should name comparison be case-sensitive; non case-sensitive comparison works properly only for ASCII characters //! \return Pointer to found attribute, or 0 if not found. xml_attribute *first_attribute(const Ch *name_ = 0, std::size_t name_size_ = 0, bool case_sensitive = true) const @@ -1074,7 +1074,7 @@ namespace rapidxml // Node modification //! Sets type of node. - //! \param type Type of node to set. + //! \param type_ Type of node to set. void type(node_type type_) { m_type = type_; @@ -1366,7 +1366,7 @@ namespace rapidxml //! parse() function allocates memory for nodes and attributes by using functions of xml_document, //! which are inherited from memory_pool. //! To access root node of the document, use the document itself, as if it was an xml_node. - //! \param Ch Character type to use. + //! \tparam Ch Character type to use. template class xml_document: public xml_node, public memory_pool { @@ -1527,7 +1527,7 @@ namespace rapidxml { // Insert 8-bit ASCII character // Todo: possibly verify that code is less than 256 and use replacement char otherwise? - text[0] = static_cast(code); + text[0] = static_cast(code); text += 1; } else @@ -1535,28 +1535,28 @@ namespace rapidxml // Insert UTF8 sequence if (code < 0x80) // 1 byte sequence { - text[0] = static_cast(code); + text[0] = static_cast(code); text += 1; } else if (code < 0x800) // 2 byte sequence { - text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[0] = static_cast(code | 0xC0); + text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[0] = static_cast(code | 0xC0); text += 2; } else if (code < 0x10000) // 3 byte sequence { - text[2] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[0] = static_cast(code | 0xE0); + text[2] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[0] = static_cast(code | 0xE0); text += 3; } else if (code < 0x110000) // 4 byte sequence { - text[3] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[2] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; - text[0] = static_cast(code | 0xF0); + text[3] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[2] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[1] = static_cast((code | 0x80) & 0xBF); code >>= 6; + text[0] = static_cast(code | 0xF0); text += 4; } else // Invalid, only codes up to 0x10FFFF are allowed in Unicode @@ -1812,7 +1812,7 @@ namespace rapidxml // Create comment node xml_node *comment = this->allocate_node(node_comment); - comment->value(value_, text - value_); + comment->value(value_, static_cast(text - value_)); // Place zero terminator after comment value if (!(Flags & parse_no_string_terminators)) @@ -1871,7 +1871,7 @@ namespace rapidxml { // Create a new doctype node xml_node *doctype = this->allocate_node(node_doctype); - doctype->value(value_, text - value_); + doctype->value(value_, static_cast(text - value_)); // Place zero terminator after value if (!(Flags & parse_no_string_terminators)) @@ -1903,7 +1903,7 @@ namespace rapidxml skip(text); if (text == name_) CEREAL_RAPIDXML_PARSE_ERROR("expected PI target", text); - pi->name(name_, text - name_); + pi->name(name_, static_cast(text - name_)); // Skip whitespace between pi target and pi skip(text); @@ -1920,7 +1920,7 @@ namespace rapidxml } // Set pi value (verbatim, no entity expansion or whitespace normalization) - pi->value(value_, text - value_); + pi->value(value_, static_cast(text - value_)); // Place zero terminator after name and value if (!(Flags & parse_no_string_terminators)) @@ -1987,14 +1987,14 @@ namespace rapidxml if (!(Flags & parse_no_data_nodes)) { xml_node *data = this->allocate_node(node_data); - data->value(value_, end - value_); + data->value(value_, static_cast(end - value_)); node->append_node(data); } // Add data to parent node if no data exists yet if (!(Flags & parse_no_element_values)) if (*node->value() == Ch('\0')) - node->value(value_, end - value_); + node->value(value_, static_cast(end - value_)); // Place zero terminator after value if (!(Flags & parse_no_string_terminators)) @@ -2037,7 +2037,7 @@ namespace rapidxml // Create new cdata node xml_node *cdata = this->allocate_node(node_cdata); - cdata->value(value_, text - value_); + cdata->value(value_, static_cast(text - value_)); // Place zero terminator after value if (!(Flags & parse_no_string_terminators)) @@ -2059,7 +2059,7 @@ namespace rapidxml skip(text); if (text == name_) CEREAL_RAPIDXML_PARSE_ERROR("expected element name", text); - element->name(name_, text - name_); + element->name(name_, static_cast(text - name_)); // Skip whitespace between element name and attributes or > skip(text); @@ -2216,7 +2216,7 @@ namespace rapidxml // Skip and validate closing tag name Ch *closing_name = text; skip(text); - if (!internal::compare(node->name(), node->name_size(), closing_name, text - closing_name, true)) + if (!internal::compare(node->name(), node->name_size(), closing_name, static_cast(text - closing_name), true)) CEREAL_RAPIDXML_PARSE_ERROR("invalid closing tag name", text); } else @@ -2232,7 +2232,7 @@ namespace rapidxml if (contents_end && contents_end != contents_start) { - node->value(contents_start, contents_end - contents_start); + node->value(contents_start, static_cast(contents_end - contents_start)); node->value()[node->value_size()] = Ch('\0'); } return; // Node closed, finished parsing contents @@ -2275,7 +2275,7 @@ namespace rapidxml // Create new attribute xml_attribute *attribute = this->allocate_attribute(); - attribute->name(name_, text - name_); + attribute->name(name_, static_cast(text - name_)); node->append_attribute(attribute); // Skip whitespace after attribute name @@ -2308,7 +2308,7 @@ namespace rapidxml end = skip_and_expand_character_refs, attribute_value_pure_pred, AttFlags>(text, false); // Set attribute value - attribute->value(value_, end - value_); + attribute->value(value_, static_cast(end - value_)); // Make sure that end quote is present if (*text != quote) diff --git a/noarch/include/cereal/external/rapidxml/rapidxml_print.hpp b/noarch/include/cereal/external/rapidxml/rapidxml_print.hpp index ce10dbb0..7fbb962e 100644 --- a/noarch/include/cereal/external/rapidxml/rapidxml_print.hpp +++ b/noarch/include/cereal/external/rapidxml/rapidxml_print.hpp @@ -363,10 +363,12 @@ namespace rapidxml out = print_pi_node(out, node, flags, indent); break; +#ifndef __GNUC__ // Unknown default: assert(0); break; +#endif } // If indenting not disabled, add line break after node diff --git a/noarch/include/cereal/macros.hpp b/noarch/include/cereal/macros.hpp index cfdc14c5..96ca7516 100644 --- a/noarch/include/cereal/macros.hpp +++ b/noarch/include/cereal/macros.hpp @@ -132,4 +132,23 @@ #endif // end !defined(CEREAL_HAS_NOEXCEPT) #endif // ifndef CEREAL_NOEXCEPT +// ###################################################################### +//! Checks if C++17 is available +#if __cplusplus >= 201703L || (defined(_MSVC_LANG) && _MSVC_LANG >= 201703L) +#define CEREAL_HAS_CPP17 +#endif + +//! Checks if C++14 is available +#if __cplusplus >= 201402L +#define CEREAL_HAS_CPP14 +#endif + +// ###################################################################### +//! Defines the CEREAL_ALIGNOF macro to use instead of alignof +#if defined(_MSC_VER) && _MSC_VER < 1900 +#define CEREAL_ALIGNOF __alignof +#else // not MSVC 2013 or older +#define CEREAL_ALIGNOF alignof +#endif // end MSVC check + #endif // CEREAL_MACROS_HPP_ diff --git a/noarch/include/cereal/specialize.hpp b/noarch/include/cereal/specialize.hpp new file mode 100644 index 00000000..b9276db2 --- /dev/null +++ b/noarch/include/cereal/specialize.hpp @@ -0,0 +1,139 @@ +/*! \file specialize.hpp + \brief Serialization disambiguation */ +/* + Copyright (c) 2014, Randolph Voorhies, Shane Grant + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of cereal nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND + ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY + DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef CEREAL_SPECIALIZE_HPP_ +#define CEREAL_SPECIALIZE_HPP_ + +namespace cereal +{ + // Forward declaration of access class that users can become friends with + class access; + + // ###################################################################### + //! A specifier used in conjunction with cereal::specialize to disambiguate + //! serialization in special cases + /*! @relates specialize + @ingroup Access */ + enum class specialization + { + member_serialize, //!< Force the use of a member serialize function + member_load_save, //!< Force the use of a member load/save pair + member_load_save_minimal, //!< Force the use of a member minimal load/save pair + non_member_serialize, //!< Force the use of a non-member serialize function + non_member_load_save, //!< Force the use of a non-member load/save pair + non_member_load_save_minimal //!< Force the use of a non-member minimal load/save pair + }; + + //! A class used to disambiguate cases where cereal cannot detect a unique way of serializing a class + /*! cereal attempts to figure out which method of serialization (member vs. non-member serialize + or load/save pair) at compile time. If for some reason cereal cannot find a non-ambiguous way + of serializing a type, it will produce a static assertion complaining about this. + + This can happen because you have both a serialize and load/save pair, or even because a base + class has a serialize (public or private with friend access) and a derived class does not + overwrite this due to choosing some other serialization type. + + Specializing this class will tell cereal to explicitly use the serialization type you specify + and it will not complain about ambiguity in its compile time selection. However, if cereal detects + an ambiguity in specializations, it will continue to issue a static assertion. + + @code{.cpp} + class MyParent + { + friend class cereal::access; + template + void serialize( Archive & ar ) {} + }; + + // Although serialize is private in MyParent, to cereal::access it will look public, + // even through MyDerived + class MyDerived : public MyParent + { + public: + template + void load( Archive & ar ) {} + + template + void save( Archive & ar ) {} + }; + + // The load/save pair in MyDerived is ambiguous because serialize in MyParent can + // be accessed from cereal::access. This looks the same as making serialize public + // in MyParent, making it seem as though MyDerived has both a serialize and a load/save pair. + // cereal will complain about this at compile time unless we disambiguate: + + namespace cereal + { + // This struct specialization will tell cereal which is the right way to serialize the ambiguity + template struct specialize {}; + + // If we only had a disambiguation for a specific archive type, it would look something like this + template <> struct specialize {}; + } + @endcode + + You can also choose to use the macros CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES or + CEREAL_SPECIALIZE_FOR_ARCHIVE if you want to type a little bit less. + + @tparam T The type to specialize the serialization for + @tparam S The specialization type to use for T + @ingroup Access */ + template + struct specialize : public std::false_type {}; + + //! Convenient macro for performing specialization for all archive types + /*! This performs specialization for the specific type for all types of archives. + This macro should be placed at the global namespace. + + @code{cpp} + struct MyType {}; + CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES( MyType, cereal::specialization::member_load_save ); + @endcode + + @relates specialize + @ingroup Access */ + #define CEREAL_SPECIALIZE_FOR_ALL_ARCHIVES( Type, Specialization ) \ + namespace cereal { template struct specialize {}; } + + //! Convenient macro for performing specialization for a single archive type + /*! This performs specialization for the specific type for a single type of archive. + This macro should be placed at the global namespace. + + @code{cpp} + struct MyType {}; + CEREAL_SPECIALIZE_FOR_ARCHIVE( cereal::XMLInputArchive, MyType, cereal::specialization::member_load_save ); + @endcode + + @relates specialize + @ingroup Access */ + #define CEREAL_SPECIALIZE_FOR_ARCHIVE( Archive, Type, Specialization ) \ + namespace cereal { template <> struct specialize {}; } +} + +#endif diff --git a/noarch/include/cereal/types/atomic.hpp b/noarch/include/cereal/types/atomic.hpp new file mode 100644 index 00000000..03d8b872 --- /dev/null +++ b/noarch/include/cereal/types/atomic.hpp @@ -0,0 +1,55 @@ +/*! \file atomic.hpp + \brief Support for types found in \ + \ingroup STLSupport */ +/* + Copyright (c) 2014, Randolph Voorhies, Shane Grant + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of cereal nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND + ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY + DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ +#ifndef CEREAL_TYPES_ATOMIC_HPP_ +#define CEREAL_TYPES_ATOMIC_HPP_ + +#include +#include + +namespace cereal +{ + //! Serializing (save) for std::atomic + template inline + void CEREAL_SAVE_FUNCTION_NAME( Archive & ar, std::atomic const & a ) + { + ar( CEREAL_NVP_("atomic_data", a.load()) ); + } + + //! Serializing (load) for std::atomic + template inline + void CEREAL_LOAD_FUNCTION_NAME( Archive & ar, std::atomic & a ) + { + T tmp; + ar( CEREAL_NVP_("atomic_data", tmp) ); + a.store( tmp ); + } +} // namespace cereal + +#endif // CEREAL_TYPES_ATOMIC_HPP_ diff --git a/noarch/include/cereal/types/base_class.hpp b/noarch/include/cereal/types/base_class.hpp index 37b551ed..fd7423e6 100644 --- a/noarch/include/cereal/types/base_class.hpp +++ b/noarch/include/cereal/types/base_class.hpp @@ -63,7 +63,8 @@ namespace cereal This also automatically registers polymorphic relation between the base and derived class, assuming they are indeed polymorphic. Note this is not the same as polymorphic type registration. For more information - see the documentation on polymorphism. + see the documentation on polymorphism. If using a polymorphic class, be sure to include support for + polymorphism (cereal/types/polymorphic.hpp). \sa virtual_base_class @@ -120,7 +121,8 @@ namespace cereal This also automatically registers polymorphic relation between the base and derived class, assuming they are indeed polymorphic. Note this is not the same as polymorphic type registration. For more information - see the documentation on polymorphism. + see the documentation on polymorphism. If using a polymorphic class, be sure to include support for + polymorphism (cereal/types/polymorphic.hpp). \sa base_class diff --git a/noarch/include/cereal/types/bitset.hpp b/noarch/include/cereal/types/bitset.hpp index 12d3a827..22ff534f 100644 --- a/noarch/include/cereal/types/bitset.hpp +++ b/noarch/include/cereal/types/bitset.hpp @@ -67,7 +67,7 @@ namespace cereal if( bits[i] ) chunk |= mask; - mask >>= 1; + mask = static_cast(mask >> 1); // output current chunk when mask is empty (8 bits) if( mask == 0 ) @@ -163,7 +163,7 @@ namespace cereal if( chunk & mask ) bits[i] = 1; - mask >>= 1; + mask = static_cast(mask >> 1); } break; } diff --git a/noarch/include/cereal/types/boost_variant.hpp b/noarch/include/cereal/types/boost_variant.hpp index e1863c14..5f64de50 100644 --- a/noarch/include/cereal/types/boost_variant.hpp +++ b/noarch/include/cereal/types/boost_variant.hpp @@ -30,13 +30,20 @@ #ifndef CEREAL_TYPES_BOOST_VARIANT_HPP_ #define CEREAL_TYPES_BOOST_VARIANT_HPP_ +//! @internal +#if defined(_MSC_VER) && _MSC_VER < 1911 +#define CEREAL_CONSTEXPR_LAMBDA +#else // MSVC 2017 or newer, all other compilers +#define CEREAL_CONSTEXPR_LAMBDA constexpr +#endif + #include "cereal/cereal.hpp" -#include -#include +#include +#include namespace cereal { - namespace variant_detail + namespace boost_variant_detail { //! @internal template @@ -45,62 +52,113 @@ namespace cereal variant_save_visitor(Archive & ar_) : ar(ar_) {} template - void operator()(T const & value) const - { - ar( CEREAL_NVP_("data", value) ); - } + void operator()(T const & value) const + { + ar( CEREAL_NVP_("data", value) ); + } Archive & ar; }; //! @internal - template - typename std::enable_if::value, void>::type - load_variant(Archive & /*ar*/, int /*target*/, Variant & /*variant*/) + template + struct LoadAndConstructLoadWrapper { - throw ::cereal::Exception("Error traversing variant during load"); - } + using ST = typename std::aligned_storage::type; + + LoadAndConstructLoadWrapper() : + construct( reinterpret_cast( &st ) ) + { } + + ~LoadAndConstructLoadWrapper() + { + if (construct.itsValid) + { + construct->~T(); + } + } + + void CEREAL_SERIALIZE_FUNCTION_NAME( Archive & ar ) + { + ::cereal::detail::Construct::load_andor_construct( ar, construct ); + } + + ST st; + ::cereal::construct construct; + }; //! @internal - template - typename std::enable_if::value, void>::type - load_variant(Archive & ar, int target, Variant & variant) - { - if(N == target) - { - H value; - ar( CEREAL_NVP_("data", value) ); - variant = value; - } - else - load_variant(ar, target, variant); - } + template struct load_variant_wrapper; - } // namespace variant_detail + //! Avoid serializing variant void_ type + /*! @internal */ + template <> + struct load_variant_wrapper + { + template + static void load_variant( Archive &, Variant & ) + { } + }; + + //! @internal + template + struct load_variant_wrapper + { + // default constructible + template + static void load_variant_impl( Archive & ar, Variant & variant, std::true_type ) + { + T value; + ar( CEREAL_NVP_("data", value) ); + variant = std::move(value); + } + + // not default constructible + template + static void load_variant_impl(Archive & ar, Variant & variant, std::false_type ) + { + LoadAndConstructLoadWrapper loadWrapper; + + ar( CEREAL_NVP_("data", loadWrapper) ); + variant = std::move(*loadWrapper.construct.ptr()); + } + + //! @internal + template + static void load_variant(Archive & ar, Variant & variant) + { + load_variant_impl( ar, variant, typename std::is_default_constructible::type() ); + } + }; + } // namespace boost_variant_detail //! Saving for boost::variant - template inline - void CEREAL_SAVE_FUNCTION_NAME( Archive & ar, boost::variant const & variant ) + template inline + void CEREAL_SAVE_FUNCTION_NAME( Archive & ar, boost::variant const & variant ) { int32_t which = variant.which(); ar( CEREAL_NVP_("which", which) ); - variant_detail::variant_save_visitor visitor(ar); + boost_variant_detail::variant_save_visitor visitor(ar); variant.apply_visitor(visitor); } //! Loading for boost::variant - template inline - void CEREAL_LOAD_FUNCTION_NAME( Archive & ar, boost::variant & variant ) + template inline + void CEREAL_LOAD_FUNCTION_NAME( Archive & ar, boost::variant & variant ) { - typedef typename boost::variant::types types; - int32_t which; ar( CEREAL_NVP_("which", which) ); - if(which >= boost::mpl::size::value) + + using LoadFuncType = void(*)(Archive &, boost::variant &); + CEREAL_CONSTEXPR_LAMBDA LoadFuncType loadFuncArray[] = {&boost_variant_detail::load_variant_wrapper::load_variant...}; + + if(which >= int32_t(sizeof(loadFuncArray)/sizeof(loadFuncArray[0]))) throw Exception("Invalid 'which' selector when deserializing boost::variant"); - variant_detail::load_variant<0, boost::variant, VariantType1, VariantTypes...>(ar, which, variant); + loadFuncArray[which](ar, variant); } } // namespace cereal +#undef CEREAL_CONSTEXPR_LAMBDA + #endif // CEREAL_TYPES_BOOST_VARIANT_HPP_ diff --git a/noarch/include/cereal/types/memory.hpp b/noarch/include/cereal/types/memory.hpp index 40c965a6..59e9da9b 100644 --- a/noarch/include/cereal/types/memory.hpp +++ b/noarch/include/cereal/types/memory.hpp @@ -34,13 +34,6 @@ #include #include -// Work around MSVC not having alignof -#if defined(_MSC_VER) && _MSC_VER < 1900 -#define CEREAL_ALIGNOF __alignof -#else // not MSVC 2013 or older -#define CEREAL_ALIGNOF alignof -#endif // end MSVC check - namespace cereal { namespace memory_detail @@ -54,6 +47,7 @@ namespace cereal PtrWrapper(T && p) : ptr(std::forward(p)) {} T & ptr; + PtrWrapper( PtrWrapper const & ) = default; PtrWrapper & operator=( PtrWrapper const & ) = delete; }; @@ -110,6 +104,11 @@ namespace cereal portion of the class and replace it after whatever happens to modify it (e.g. the user performing construction or the wrapper shared_ptr in saving). + Note that this goes into undefined behavior territory, but as of the initial writing + of this, all standard library implementations of std::enable_shared_from_this are + compatible with this memory manipulation. It is entirely possible that this may someday + break or may not work with convoluted use cases. + Example usage: @code{.cpp} @@ -136,7 +135,7 @@ namespace cereal using BaseType = typename ::cereal::traits::get_shared_from_this_base::type; using ParentType = std::enable_shared_from_this; using StorageType = typename std::aligned_storage::type; - + public: //! Saves the state of some type inheriting from enable_shared_from_this /*! @param ptr The raw pointer held by the shared_ptr */ @@ -153,7 +152,8 @@ namespace cereal { if( !itsRestored ) { - std::memcpy( itsPtr, &itsState, sizeof(ParentType) ); + // void * cast needed when type has no trivial copy-assignment + std::memcpy( static_cast(itsPtr), &itsState, sizeof(ParentType) ); itsRestored = true; } } @@ -278,8 +278,6 @@ namespace cereal typename std::enable_if::value, void>::type CEREAL_LOAD_FUNCTION_NAME( Archive & ar, memory_detail::PtrWrapper &> & wrapper ) { - auto & ptr = wrapper.ptr; - uint32_t id; ar( CEREAL_NVP_("id", id) ); @@ -297,8 +295,9 @@ namespace cereal // Allocate our storage, which we will treat as // uninitialized until initialized with placement new - ptr.reset( reinterpret_cast( new ST() ), - [=]( T * t ) + using NonConstT = typename std::remove_const::type; + std::shared_ptr ptr(reinterpret_cast(new ST()), + [=]( NonConstT * t ) { if( *valid ) t->~T(); @@ -310,13 +309,14 @@ namespace cereal ar.registerSharedPointer( id, ptr ); // Perform the actual loading and allocation - memory_detail::loadAndConstructSharedPtr( ar, ptr.get(), typename ::cereal::traits::has_shared_from_this::type() ); + memory_detail::loadAndConstructSharedPtr( ar, ptr.get(), typename ::cereal::traits::has_shared_from_this::type() ); // Mark pointer as valid (initialized) *valid = true; + wrapper.ptr = std::move(ptr); } else - ptr = std::static_pointer_cast(ar.getSharedPointer(id)); + wrapper.ptr = std::static_pointer_cast(ar.getSharedPointer(id)); } //! Loading std::shared_ptr, case when no user load and construct (wrapper implementation) @@ -325,20 +325,20 @@ namespace cereal typename std::enable_if::value, void>::type CEREAL_LOAD_FUNCTION_NAME( Archive & ar, memory_detail::PtrWrapper &> & wrapper ) { - auto & ptr = wrapper.ptr; - uint32_t id; ar( CEREAL_NVP_("id", id) ); if( id & detail::msb_32bit ) { - ptr.reset( detail::Construct::load_andor_construct() ); + using NonConstT = typename std::remove_const::type; + std::shared_ptr ptr( detail::Construct::load_andor_construct() ); ar.registerSharedPointer( id, ptr ); ar( CEREAL_NVP_("data", *ptr) ); + wrapper.ptr = std::move(ptr); } else - ptr = std::static_pointer_cast(ar.getSharedPointer(id)); + wrapper.ptr = std::static_pointer_cast(ar.getSharedPointer(id)); } //! Saving std::unique_ptr (wrapper implementation) @@ -374,16 +374,17 @@ namespace cereal if( isValid ) { + using NonConstT = typename std::remove_const::type; // Storage type for the pointer - since we can't default construct this type, // we'll allocate it using std::aligned_storage - using ST = typename std::aligned_storage::type; + using ST = typename std::aligned_storage::type; // Allocate storage - note the ST type so that deleter is correct if // an exception is thrown before we are initialized std::unique_ptr stPtr( new ST() ); // Use wrapper to enter into "data" nvp of ptr_wrapper - memory_detail::LoadAndConstructLoadWrapper loadWrapper( reinterpret_cast( stPtr.get() ) ); + memory_detail::LoadAndConstructLoadWrapper loadWrapper( reinterpret_cast( stPtr.get() ) ); // Initialize storage ar( CEREAL_NVP_("data", loadWrapper) ); @@ -404,16 +405,16 @@ namespace cereal uint8_t isValid; ar( CEREAL_NVP_("valid", isValid) ); - auto & ptr = wrapper.ptr; - if( isValid ) { - ptr.reset( detail::Construct::load_andor_construct() ); + using NonConstT = typename std::remove_const::type; + std::unique_ptr ptr( detail::Construct::load_andor_construct() ); ar( CEREAL_NVP_( "data", *ptr ) ); + wrapper.ptr = std::move(ptr); } else { - ptr.reset( nullptr ); + wrapper.ptr.reset( nullptr ); } } } // namespace cereal @@ -421,5 +422,4 @@ namespace cereal // automatically include polymorphic support #include "cereal/types/polymorphic.hpp" -#undef CEREAL_ALIGNOF -#endif // CEREAL_TYPES_SHARED_PTR_HPP_ \ No newline at end of file +#endif // CEREAL_TYPES_SHARED_PTR_HPP_ diff --git a/noarch/include/cereal/types/optional.hpp b/noarch/include/cereal/types/optional.hpp new file mode 100644 index 00000000..3b244465 --- /dev/null +++ b/noarch/include/cereal/types/optional.hpp @@ -0,0 +1,66 @@ +/*! \file optional.hpp + \brief Support for std::optional + \ingroup STLSupport */ +/* + Copyright (c) 2017, Juan Pedro Bolivar Puente + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of cereal nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND + ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY + DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ +#ifndef CEREAL_TYPES_STD_OPTIONAL_ +#define CEREAL_TYPES_STD_OPTIONAL_ + +#include "cereal/cereal.hpp" +#include + +namespace cereal { + //! Saving for std::optional + template inline + void CEREAL_SAVE_FUNCTION_NAME(Archive& ar, const std::optional& optional) + { + if(!optional) { + ar(CEREAL_NVP_("nullopt", true)); + } else { + ar(CEREAL_NVP_("nullopt", false), + CEREAL_NVP_("data", *optional)); + } + } + + //! Loading for std::optional + template inline + void CEREAL_LOAD_FUNCTION_NAME(Archive& ar, std::optional& optional) + { + bool nullopt; + ar(CEREAL_NVP_("nullopt", nullopt)); + + if (nullopt) { + optional = std::nullopt; + } else { + T value; + ar(CEREAL_NVP_("data", value)); + optional = std::move(value); + } + } +} // namespace cereal + +#endif // CEREAL_TYPES_STD_OPTIONAL_ diff --git a/noarch/include/cereal/types/polymorphic.hpp b/noarch/include/cereal/types/polymorphic.hpp index 5acc11f8..f3db78f6 100644 --- a/noarch/include/cereal/types/polymorphic.hpp +++ b/noarch/include/cereal/types/polymorphic.hpp @@ -165,17 +165,19 @@ See CEREAL_REGISTER_DYNAMIC_INIT for detailed explanation of how this macro should be used. The name used should match that for CEREAL_REGISTER_DYNAMIC_INIT. */ -#define CEREAL_FORCE_DYNAMIC_INIT(LibName) \ - namespace cereal { \ - namespace detail { \ - void dynamic_init_dummy_##LibName(); \ - } /* end detail */ \ - namespace { \ - void dynamic_init_##LibName() \ - { \ - ::cereal::detail::dynamic_init_dummy_##LibName(); \ - } \ - } } /* end namespaces */ +#define CEREAL_FORCE_DYNAMIC_INIT(LibName) \ + namespace cereal { \ + namespace detail { \ + void CEREAL_DLL_EXPORT dynamic_init_dummy_##LibName(); \ + } /* end detail */ \ + } /* end cereal */ \ + namespace { \ + struct dynamic_init_##LibName { \ + dynamic_init_##LibName() { \ + ::cereal::detail::dynamic_init_dummy_##LibName(); \ + } \ + } dynamic_init_instance_##LibName; \ + } /* end anonymous namespace */ namespace cereal { diff --git a/noarch/include/cereal/types/tuple.hpp b/noarch/include/cereal/types/tuple.hpp index 7e56f0dc..23013cdb 100644 --- a/noarch/include/cereal/types/tuple.hpp +++ b/noarch/include/cereal/types/tuple.hpp @@ -38,7 +38,7 @@ namespace cereal namespace tuple_detail { //! Creates a c string from a sequence of characters - /*! The c string created will alwas be prefixed by "tuple_element" + /*! The c string created will always be prefixed by "tuple_element" Based on code from: http://stackoverflow/a/20973438/710791 @internal */ template @@ -63,7 +63,7 @@ namespace cereal template struct to_string_impl { - using type = typename to_string_impl::type; + using type = typename to_string_impl(R+std::size_t{'0'}), C...>::type; }; //! Base case with no quotient @@ -71,7 +71,7 @@ namespace cereal template struct to_string_impl<0, R, C...> { - using type = char_seq_to_c_str; + using type = char_seq_to_c_str(R+std::size_t{'0'}), C...>; }; //! Generates a c string for a given index of a tuple @@ -84,7 +84,7 @@ namespace cereal struct tuple_element_name { using type = typename to_string_impl::type; - static const typename type::arr_type c_str(){ return type::str; }; + static const typename type::arr_type c_str(){ return type::str; } }; // unwinds a tuple to save it diff --git a/noarch/include/cereal/types/variant.hpp b/noarch/include/cereal/types/variant.hpp new file mode 100644 index 00000000..8bea397b --- /dev/null +++ b/noarch/include/cereal/types/variant.hpp @@ -0,0 +1,109 @@ +/*! \file variant.hpp + \brief Support for std::variant + \ingroup STLSupport */ +/* + Copyright (c) 2014, 2017, Randolph Voorhies, Shane Grant, Juan Pedro + Bolivar Puente. All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of cereal nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND + ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY + DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ +#ifndef CEREAL_TYPES_STD_VARIANT_HPP_ +#define CEREAL_TYPES_STD_VARIANT_HPP_ + +#include "cereal/cereal.hpp" +#include +#include + +namespace cereal +{ + namespace variant_detail + { + //! @internal + template + struct variant_save_visitor + { + variant_save_visitor(Archive & ar_) : ar(ar_) {} + + template + void operator()(T const & value) const + { + ar( CEREAL_NVP_("data", value) ); + } + + Archive & ar; + }; + + //! @internal + template + typename std::enable_if, void>::type + load_variant(Archive & /*ar*/, int /*target*/, Variant & /*variant*/) + { + throw ::cereal::Exception("Error traversing variant during load"); + } + //! @internal + template + typename std::enable_if, void>::type + load_variant(Archive & ar, int target, Variant & variant) + { + if(N == target) + { + H value; + ar( CEREAL_NVP_("data", value) ); + variant = std::move(value); + } + else + load_variant(ar, target, variant); + } + + } // namespace variant_detail + + //! Saving for std::variant + template inline + void CEREAL_SAVE_FUNCTION_NAME( Archive & ar, std::variant const & variant ) + { + std::int32_t index = static_cast(variant.index()); + ar( CEREAL_NVP_("index", index) ); + variant_detail::variant_save_visitor visitor(ar); + std::visit(visitor, variant); + } + + //! Loading for std::variant + template inline + void CEREAL_LOAD_FUNCTION_NAME( Archive & ar, std::variant & variant ) + { + using variant_t = typename std::variant; + + std::int32_t index; + ar( CEREAL_NVP_("index", index) ); + if(index >= static_cast(std::variant_size_v)) + throw Exception("Invalid 'index' selector when deserializing std::variant"); + + variant_detail::load_variant<0, variant_t, VariantTypes...>(ar, index, variant); + } + + //! Serializing a std::monostate + template + void CEREAL_SERIALIZE_FUNCTION_NAME( Archive &, std::monostate const & ) {} +} // namespace cereal + +#endif // CEREAL_TYPES_STD_VARIANT_HPP_ diff --git a/noarch/include/cereal/types/vector.hpp b/noarch/include/cereal/types/vector.hpp index 8302cb90..4984bd3e 100644 --- a/noarch/include/cereal/types/vector.hpp +++ b/noarch/include/cereal/types/vector.hpp @@ -60,8 +60,8 @@ namespace cereal //! Serialization for non-arithmetic vector types template inline - typename std::enable_if, Archive>::value - || !std::is_arithmetic::value, void>::type + typename std::enable_if<(!traits::is_output_serializable, Archive>::value + || !std::is_arithmetic::value) && !std::is_same::value, void>::type CEREAL_SAVE_FUNCTION_NAME( Archive & ar, std::vector const & vector ) { ar( make_size_tag( static_cast(vector.size()) ) ); // number of elements @@ -71,8 +71,8 @@ namespace cereal //! Serialization for non-arithmetic vector types template inline - typename std::enable_if, Archive>::value - || !std::is_arithmetic::value, void>::type + typename std::enable_if<(!traits::is_input_serializable, Archive>::value + || !std::is_arithmetic::value) && !std::is_same::value, void>::type CEREAL_LOAD_FUNCTION_NAME( Archive & ar, std::vector & vector ) { size_type size; @@ -88,7 +88,7 @@ namespace cereal void CEREAL_SAVE_FUNCTION_NAME( Archive & ar, std::vector const & vector ) { ar( make_size_tag( static_cast(vector.size()) ) ); // number of elements - for(auto && v : vector) + for(const auto v : vector) ar( static_cast(v) ); } @@ -100,7 +100,7 @@ namespace cereal ar( make_size_tag( size ) ); vector.resize( static_cast( size ) ); - for(auto && v : vector) + for(auto v : vector) { bool b; ar( b ); diff --git a/noarch/include/cereal/version.hpp b/noarch/include/cereal/version.hpp new file mode 100644 index 00000000..2a9dada9 --- /dev/null +++ b/noarch/include/cereal/version.hpp @@ -0,0 +1,52 @@ +/*! \file version.hpp + \brief Macros to detect cereal version + + These macros can assist in determining the version of cereal. Be + warned that cereal is not guaranteed to be compatible across + different versions. For more information on releases of cereal, + see https://github.com/USCiLab/cereal/releases. + + \ingroup utility */ +/* + Copyright (c) 2018, Shane Grant + All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of cereal nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND + ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + DISCLAIMED. IN NO EVENT SHALL RANDOLPH VOORHIES OR SHANE GRANT BE LIABLE FOR ANY + DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef CEREAL_VERSION_HPP_ +#define CEREAL_VERSION_HPP_ + +//! The major version +#define CEREAL_VERSION_MAJOR 1 +//! The minor version +#define CEREAL_VERSION_MINOR 3 +//! The patch version +#define CEREAL_VERSION_PATCH 0 + +//! The full version as a single number +#define CEREAL_VERSION (CEREAL_VERSION_MAJOR * 10000 \ + + CEREAL_VERSION_MINOR * 100 \ + + CEREAL_VERSION_PATCH) + +#endif // CEREAL_VERSION_HPP_