Merge pull request #81 from miloyip/IterativeParseOptimization

Iterative parse optimization

include/rapidjson/document.h

@@ -1221,13 +1221,12 @@ public:
 		\tparam SourceEncoding Encoding of input stream
 		\tparam InputStream Type of input stream, implementing Stream concept
 		\param is Input stream to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
 	template <unsigned parseFlags, typename SourceEncoding, typename InputStream>
-	GenericDocument& ParseStream(InputStream& is, size_t limit = 0) {
+	GenericDocument& ParseStream(InputStream& is) {
 		ValueType::SetNull(); // Remove existing root if exist
-		GenericReader<SourceEncoding, Encoding, Allocator> reader(limit, &GetAllocator());
+		GenericReader<SourceEncoding, Encoding, Allocator> reader(&GetAllocator());
 		ClearStackOnExit scope(*this);
 		parseResult_ = reader.template Parse<parseFlags>(is, *this);
 		if (parseResult_) {
@@ -1241,23 +1240,21 @@ public:
 	/*! \tparam parseFlags Combination of \ref ParseFlag.
 		\tparam InputStream Type of input stream, implementing Stream concept
 		\param is Input stream to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
 	template <unsigned parseFlags, typename InputStream>
-	GenericDocument& ParseStream(InputStream& is, size_t limit = 0) {
-		return ParseStream<parseFlags,Encoding,InputStream>(is, limit);
+	GenericDocument& ParseStream(InputStream& is) {
+		return ParseStream<parseFlags,Encoding,InputStream>(is);
 	}
 
 	//! Parse JSON text from an input stream (with \ref kParseDefaultFlags)
 	/*! \tparam InputStream Type of input stream, implementing Stream concept
 		\param is Input stream to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
 	template <typename InputStream>
-	GenericDocument& ParseStream(InputStream& is, size_t limit = 0) {
-		return ParseStream<kParseDefaultFlags, Encoding, InputStream>(is, limit);
+	GenericDocument& ParseStream(InputStream& is) {
+		return ParseStream<kParseDefaultFlags, Encoding, InputStream>(is);
 	}
 	//!@}
 
@@ -1268,33 +1265,30 @@ public:
 	/*! \tparam parseFlags Combination of \ref ParseFlag.
 		\tparam SourceEncoding Transcoding from input Encoding
 		\param str Mutable zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
 	template <unsigned parseFlags, typename SourceEncoding>
-	GenericDocument& ParseInsitu(Ch* str, size_t limit = 0) {
+	GenericDocument& ParseInsitu(Ch* str) {
 		GenericInsituStringStream<Encoding> s(str);
-		return ParseStream<parseFlags | kParseInsituFlag, SourceEncoding>(s, limit);
+		return ParseStream<parseFlags | kParseInsituFlag, SourceEncoding>(s);
 	}
 
 	//! Parse JSON text from a mutable string
 	/*! \tparam parseFlags Combination of \ref ParseFlag.
 		\param str Mutable zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
 	template <unsigned parseFlags>
-	GenericDocument& ParseInsitu(Ch* str, size_t limit = 0) {
-		return ParseInsitu<parseFlags, Encoding>(str, limit);
+	GenericDocument& ParseInsitu(Ch* str) {
+		return ParseInsitu<parseFlags, Encoding>(str);
 	}
 
 	//! Parse JSON text from a mutable string (with \ref kParseDefaultFlags)
 	/*! \param str Mutable zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 		\return The document itself for fluent API.
 	*/
-	GenericDocument& ParseInsitu(Ch* str, size_t limit = 0) {
-		return ParseInsitu<kParseDefaultFlags, Encoding>(str, limit);
+	GenericDocument& ParseInsitu(Ch* str) {
+		return ParseInsitu<kParseDefaultFlags, Encoding>(str);
 	}
 	//!@}
 
@@ -1305,31 +1299,28 @@ public:
 	/*! \tparam parseFlags Combination of \ref ParseFlag (must not contain \ref kParseInsituFlag).
 		\tparam SourceEncoding Transcoding from input Encoding
 		\param str Read-only zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 	*/
 	template <unsigned parseFlags, typename SourceEncoding>
-	GenericDocument& Parse(const Ch* str, size_t limit = 0) {
+	GenericDocument& Parse(const Ch* str) {
 		RAPIDJSON_ASSERT(!(parseFlags & kParseInsituFlag));
 		GenericStringStream<SourceEncoding> s(str);
-		return ParseStream<parseFlags, SourceEncoding>(s, limit);
+		return ParseStream<parseFlags, SourceEncoding>(s);
 	}
 
 	//! Parse JSON text from a read-only string
 	/*! \tparam parseFlags Combination of \ref ParseFlag (must not contain \ref kParseInsituFlag).
 		\param str Read-only zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 	*/
 	template <unsigned parseFlags>
-	GenericDocument& Parse(const Ch* str, size_t limit = 0) {
-		return Parse<parseFlags, Encoding>(str, limit);
+	GenericDocument& Parse(const Ch* str) {
+		return Parse<parseFlags, Encoding>(str);
 	}
 
 	//! Parse JSON text from a read-only string (with \ref kParseDefaultFlags)
 	/*! \param str Read-only zero-terminated string to be parsed.
-		\param limit Parsing stack size limit(in bytes). Pass 0 means no limit.
 	*/
-	GenericDocument& Parse(const Ch* str, size_t limit = 0) {
-		return Parse<kParseDefaultFlags>(str, limit);
+	GenericDocument& Parse(const Ch* str) {
+		return Parse<kParseDefaultFlags>(str);
 	}
 	//!@}
 

include/rapidjson/error/en.h

@@ -40,7 +40,6 @@ inline const RAPIDJSON_ERROR_CHARTYPE* GetParseError_En(ParseErrorCode parseErro
 
 		case kParseErrorTermination:					return RAPIDJSON_ERROR_STRING("Terminate parsing due to Handler error.");
 		case kParseErrorUnspecificSyntaxError:			return RAPIDJSON_ERROR_STRING("Unspecific syntax error.");
-		case kParseErrorStackSizeLimitExceeded:			return RAPIDJSON_ERROR_STRING("Parsing stack size limit is exceeded.");
 
 		default:
 			return RAPIDJSON_ERROR_STRING("Unknown error.");

include/rapidjson/error/error.h

@@ -60,7 +60,6 @@ enum ParseErrorCode {
 
 	kParseErrorTermination,						//!< Parsing was terminated.
 	kParseErrorUnspecificSyntaxError,			//!< Unspecific syntax error.
-	kParseErrorStackSizeLimitExceeded			//!< Parsing stack size limit is exceeded.
 };
 
 //! Result of parsing (wraps ParseErrorCode)

include/rapidjson/internal/stack.h

@@ -28,20 +28,14 @@ public:
 
 	void Clear() { /*stack_top_ = 0;*/ stack_top_ = stack_; }
 
+	// Optimization note: try to minimize the size of this function for force inline.
+	// Expansion is run very infrequently, so it is moved to another (probably non-inline) function.
 	template<typename T>
-	T* Push(size_t count = 1) {
+	RAPIDJSON_FORCEINLINE T* Push(size_t count = 1) {
 		 // Expand the stack if needed
-		if (stack_top_ + sizeof(T) * count >= stack_end_) {
-			size_t new_capacity = stack_capacity_ * 2;
-			size_t size = GetSize();
-			size_t new_size = GetSize() + sizeof(T) * count;
-			if (new_capacity < new_size)
-				new_capacity = new_size;
-			stack_ = (char*)allocator_->Realloc(stack_, stack_capacity_, new_capacity);
-			stack_capacity_ = new_capacity;
-			stack_top_ = stack_ + size;
-			stack_end_ = stack_ + stack_capacity_;
-		}
+		if (stack_top_ + sizeof(T) * count >= stack_end_)
+			Expand<T>(count);
+
 		T* ret = reinterpret_cast<T*>(stack_top_);
 		stack_top_ += sizeof(T) * count;
 		return ret;
@@ -69,6 +63,19 @@ public:
 	size_t GetCapacity() const { return stack_capacity_; }
 
 private:
+	template<typename T>
+	void Expand(size_t count) {
+		size_t new_capacity = stack_capacity_ * 2;
+		size_t size = GetSize();
+		size_t new_size = GetSize() + sizeof(T) * count;
+		if (new_capacity < new_size)
+			new_capacity = new_size;
+		stack_ = (char*)allocator_->Realloc(stack_, stack_capacity_, new_capacity);
+		stack_capacity_ = new_capacity;
+		stack_top_ = stack_ + size;
+		stack_end_ = stack_ + stack_capacity_;
+	}
+
 	// Prohibit copy constructor & assignment operator.
 	Stack(const Stack&);
 	Stack& operator=(const Stack&);

include/rapidjson/rapidjson.h

@@ -41,6 +41,8 @@
 #ifndef RAPIDJSON_FORCEINLINE
 #ifdef _MSC_VER
 #define RAPIDJSON_FORCEINLINE __forceinline
+#elif defined(__GNUC__) && __GNUC__ >= 4
+#define RAPIDJSON_FORCEINLINE __attribute__((always_inline))
 #else
 #define RAPIDJSON_FORCEINLINE
 #endif

test/perftest/rapidjsontest.cpp

@@ -104,48 +104,37 @@ TEST_F(RapidJson, SIMD_SUFFIX(ReaderParse_DummyHandler_ValidateEncoding)) {
 }
 
 TEST_F(RapidJson, SIMD_SUFFIX(DocumentParseInsitu_MemoryPoolAllocator)) {
-	//const size_t userBufferSize = 128 * 1024;
-	//char* userBuffer = (char*)malloc(userBufferSize);
-
 	for (size_t i = 0; i < kTrialCount; i++) {
 		memcpy(temp_, json_, length_ + 1);
-		//MemoryPoolAllocator<> allocator(userBuffer, userBufferSize);
-		//Document doc(&allocator);
 		Document doc;
 		doc.ParseInsitu(temp_);
 		ASSERT_TRUE(doc.IsObject());
-		//if (i == 0) {
-		//	size_t size = doc.GetAllocator().Size();
-		//	size_t capacity = doc.GetAllocator().Capacity();
-		//	size_t stack_capacity = doc.GetStackCapacity();
-		//	size_t actual = size - stack_capacity;
-		//	std::cout << "Size:" << size << " Capacity:" << capacity  << " Stack:" << stack_capacity << " Actual:" << actual << std::endl;
-		//}
 	}
+}
 
-	//free(userBuffer);
+TEST_F(RapidJson, SIMD_SUFFIX(DocumentParseIterativeInsitu_MemoryPoolAllocator)) {
+	for (size_t i = 0; i < kTrialCount; i++) {
+		memcpy(temp_, json_, length_ + 1);
+		Document doc;
+		doc.ParseInsitu<kParseIterativeFlag>(temp_);
+		ASSERT_TRUE(doc.IsObject());
+	}
 }
 
 TEST_F(RapidJson, SIMD_SUFFIX(DocumentParse_MemoryPoolAllocator)) {
-	//const size_t userBufferSize = 128 * 1024;
-	//char* userBuffer = (char*)malloc(userBufferSize);
-
 	for (size_t i = 0; i < kTrialCount; i++) {
-		//MemoryPoolAllocator<> allocator(userBuffer, userBufferSize);
-		//Document doc(&allocator);
 		Document doc;
 		doc.Parse(json_);
 		ASSERT_TRUE(doc.IsObject());
-		//if (i == 0) {
-		//	size_t size = doc.GetAllocator().Size();
-		//	size_t capacity = doc.GetAllocator().Capacity();
-		//	size_t stack_capacity = doc.GetStackCapacity();
-		//	size_t actual = size - stack_capacity;
-		//	std::cout << "Size:" << size << " Capacity:" << capacity  << " Stack:" << stack_capacity << " Actual:" << actual << std::endl;
-		//}
 	}
+}
 
-	//free(userBuffer);
+TEST_F(RapidJson, SIMD_SUFFIX(DocumentParseIterative_MemoryPoolAllocator)) {
+	for (size_t i = 0; i < kTrialCount; i++) {
+		Document doc;
+		doc.Parse<kParseIterativeFlag>(json_);
+		ASSERT_TRUE(doc.IsObject());
+	}
 }
 
 TEST_F(RapidJson, SIMD_SUFFIX(DocumentParse_CrtAllocator)) {

test/unittest/readertest.cpp

@@ -932,18 +932,6 @@ TEST(Reader, IterativeParsing_ShortCircuit) {
 	}
 }
 
-TEST(Reader, IterativeParsing_LimitStackSize) {
-	BaseReaderHandler<> handler;
-	Reader reader(20);
-	StringStream is("[[[]]]");
-
-	ParseResult r = reader.Parse<kParseIterativeFlag>(is, handler);
-
-	EXPECT_TRUE(reader.HasParseError());
-	EXPECT_EQ(kParseErrorStackSizeLimitExceeded, r.Code());
-	EXPECT_EQ(2u, r.Offset());
-}
-
 #ifdef __GNUC__
 RAPIDJSON_DIAG_POP
 #endif