Integrate Poco::Web::JSON parser

JSON/include/Poco/JSON/Parser.h

@@ -53,19 +53,211 @@
 namespace Poco {
 namespace JSON {
 
+class JSON_API CharacterFeeder
+{
+public:
+	virtual ~CharacterFeeder() {}
+	virtual bool nextChar(int& c) = 0;
+};
+
+
+class JSON_API StreamCharacterFeeder : public CharacterFeeder
+{
+public:
+	StreamCharacterFeeder(std::istream& in) : _in(in) {}
+	virtual ~StreamCharacterFeeder() {}
+
+	bool nextChar(int& c);
+
+private:
+
+	std::istream& _in;
+};
+
+
+inline bool StreamCharacterFeeder::nextChar(int& c)
+{
+	if ( _in.good() )
+	{
+		c = _in.get();
+		return _in.good();
+	}
+	return false;
+}
+
+class JSON_API StringCharacterFeeder : public CharacterFeeder
+{
+public:
+	StringCharacterFeeder(std::string::const_iterator begin, std::string::const_iterator end) : _it(begin), _end(end)
+	{
+	}
+	virtual ~StringCharacterFeeder() {}
+
+	bool nextChar(int& c);
+
+private:
+
+	std::string::const_iterator _it;
+	std::string::const_iterator _end;
+
+};
+
+
+inline bool StringCharacterFeeder::nextChar(int& c)
+{
+	if ( _it == _end )
+		return false;
+
+	c = *_it++;
+	return true;
+}
 
 class JSON_API Parser
 	/// A class for passing JSON strings or streams.
 {
 public:
 
+	enum Classes
+	{
+		C_SPACE,  /* space */
+		C_WHITE,  /* other whitespace */
+		C_LCURB,  /* {  */
+		C_RCURB,  /* } */
+		C_LSQRB,  /* [ */
+		C_RSQRB,  /* ] */
+		C_COLON,  /* : */
+		C_COMMA,  /* , */
+		C_QUOTE,  /* " */
+		C_BACKS,  /* \ */
+		C_SLASH,  /* / */
+		C_PLUS,   /* + */
+		C_MINUS,  /* - */
+		C_POINT,  /* . */
+		C_ZERO ,  /* 0 */
+		C_DIGIT,  /* 123456789 */
+		C_LOW_A,  /* a */
+		C_LOW_B,  /* b */
+		C_LOW_C,  /* c */
+		C_LOW_D,  /* d */
+		C_LOW_E,  /* e */
+		C_LOW_F,  /* f */
+		C_LOW_L,  /* l */
+		C_LOW_N,  /* n */
+		C_LOW_R,  /* r */
+		C_LOW_S,  /* s */
+		C_LOW_T,  /* t */
+		C_LOW_U,  /* u */
+		C_ABCDF,  /* ABCDF */
+		C_E,      /* E */
+		C_ETC,    /* everything else */
+		C_STAR,   /* * */   
+		NR_CLASSES
+	};
+	
+	enum States
+		/// State codes
+	{
+		GO,  /* start    */
+		OK,  /* ok       */
+		OB,  /* object   */
+		KE,  /* key      */
+		CO,  /* colon    */
+		VA,  /* value    */
+		AR,  /* array    */
+		ST,  /* string   */
+		ES,  /* escape   */
+		U1,  /* u1       */
+		U2,  /* u2       */
+		U3,  /* u3       */
+		U4,  /* u4       */
+		MI,  /* minus    */
+		ZE,  /* zero     */
+		IT,  /* integer  */
+		FR,  /* fraction */
+		E1,  /* e        */
+		E2,  /* ex       */
+		E3,  /* exp      */
+		T1,  /* tr       */
+		T2,  /* tru      */
+		T3,  /* true     */
+		F1,  /* fa       */
+		F2,  /* fal      */
+		F3,  /* fals     */
+		F4,  /* false    */
+		N1,  /* nu       */
+		N2,  /* nul      */
+		N3,  /* null     */
+		C1,  /* /        */
+		C2,  /* / *     */
+		C3,  /* *        */
+		FX,  /* *.* *eE* */
+		D1,  /* second UTF-16 character decoding started by \ */
+		D2,  /* second UTF-16 character proceeded by u */
+		NR_STATES
+	};
+
+	enum Modes
+		/// Modes that can be pushed on the _pStack.
+	{
+		MODE_ARRAY = 1,
+		MODE_DONE = 2,
+		MODE_KEY = 3,
+		MODE_OBJECT = 4
+	};
+
+	enum Actions
+	{
+		CB = -10, /* _comment begin */
+		CE = -11, /* _comment end */
+		FA = -12, /* 0 */
+		TR = -13, /* 0 */
+		NU = -14, /* null */
+		DE = -15, /* double detected by exponent e E */
+		DF = -16, /* double detected by fraction . */
+		SB = -17, /* string begin */
+		MX = -18, /* integer detected by minus */
+		ZX = -19, /* integer detected by zero */
+		IX = -20, /* integer detected by 1-9 */
+		EX = -21, /* next char is _escaped */
+		UC = -22  /* Unicode character read */
+	};
+	
+	enum JSONType
+	{
+		JSON_T_NONE = 0,
+		JSON_T_ARRAY_BEGIN,
+		JSON_T_ARRAY_END,
+		JSON_T_OBJECT_BEGIN,
+		JSON_T_OBJECT_END,
+		JSON_T_INTEGER,
+		JSON_T_FLOAT,
+		JSON_T_NULL,
+		JSON_T_TRUE,
+		JSON_T_FALSE,
+		JSON_T_STRING,
+		JSON_T_KEY,
+		JSON_T_VALUE_SEPARATOR,
+		JSON_T_MAX
+	};
+
+
+	static const int PARSE_BUFFER_SIZE = 3500;
+	static const int PARSER_STACK_SIZE = 128;
+
 	Parser(const Handler::Ptr& pHandler = new ParseHandler);
 		/// Constructor.
 
 	virtual ~Parser();
 		/// Destructor.
 
-	Dynamic::Var parse(const std::string& source);
+
+	void allowComments(bool sw);
+		/// Allow comments. By default this is false.
+		
+	void allowNullByte(bool sw);
+		/// Allow a null byte in strings. By default this is true.
+
+	Dynamic::Var parse(const std::string& json);
 		/// Parses a string.
 
 	Dynamic::Var parse(std::istream& in);
@@ -81,33 +273,95 @@ public:
 		/// Returns the result of parsing;
 
 private:
-	const Token* nextToken();
-		/// Returns the next token.
-
-	void readObject();
-		/// Starts reading an object.
-
-	void readArray();
-		/// Starts reading an array.
-
-	bool readRow(bool firstCall = false);
-		/// Reads a property value pair. Returns true when a next row is expected.
-
-	void readValue(const Token* token);
-		/// Read a value from the token.
-
-	bool readElements(bool firstCall = false);
-		/// Read all elements of an array.
-
-	StreamTokenizer _tokenizer;
 	Handler::Ptr    _pHandler;
+
+
+	typedef Poco::Buffer<char> BufType;
+
+
+	bool push(int mode);
+		/// Push a mode onto the _pStack. Return false if there is overflow.
+
+	bool pop(int mode);
+		/// Pops the stack, assuring that the current mode matches the expectation.
+		/// Returns false if there is underflow or if the modes mismatch.
+
+	void growBuffer();
+
+
+	void clearBuffer();
+
+
+	void parseBufferPushBackChar(char c);
+
+
+	void parseBufferPopBackChar();
+
+
+	void addCharToParseBuffer(int nextChar, int nextClass);
+
+
+	void addEscapedCharToParseBuffer(int nextChar);
+
+
+	int decodeUnicodeChar();
+
+
+	void assertNotStringNullBool();
+
+
+	void assertNonContainer();
+
+
+	void parseBuffer();
+
+
+	bool parseChar(int nextChar);
+		/// Called for each character (or partial character) in JSON string.
+		/// It accepts UTF-8, UTF-16, or UTF-32. If it the character is accpeted,
+		/// it returns true, otherwise false.
+
+	bool done();
+
+
+	static int utf8_check_first(char byte);
+
+
+	static const int _asciiClass[128];
+		/// This array maps the 128 ASCII characters into character classes.
+		/// The remaining Unicode characters should be mapped to C_ETC.
+		/// Non-whitespace control characters are errors.
+	static const int _stateTransitionTable[NR_STATES][NR_CLASSES];
+	static const int xx = -1;
+
+	signed char _state;
+	signed char _beforeCommentState;
+	JSONType _type;
+	signed char _escaped;
+	signed char _comment;
+	unsigned short _utf16HighSurrogate;
+	long _depth;
+	long _top;
+	BufType _stack;
+	BufType _parseBuffer;
+	size_t _parseBufferCount;
+	size_t _commentBeginOffset;
+	char _decimalPoint;
+	bool _allowNullByte;
+	bool _allowComments;
+	SharedPtr<CharacterFeeder> feeder;
 };
 
 
-inline Dynamic::Var Parser::parse(const std::string& source)
+inline void Parser::allowComments(bool sw)
 {
-	std::istringstream is(source);
+	_allowComments = sw;
-	return parse(is);
+}
+
+
+inline void Parser::allowNullByte(bool sw)
+{
+	_allowNullByte = sw;
 }
 
 
@@ -129,6 +383,37 @@ inline Dynamic::Var Parser::result() const
 }
 
 
+inline bool Parser::done()
+{
+    return _state == OK && pop(MODE_DONE);
+}
+
+
+inline void Parser::assertNotStringNullBool()
+{
+	poco_assert(_type != JSON_T_FALSE &&
+		_type != JSON_T_TRUE &&
+		_type != JSON_T_NULL &&
+		_type != JSON_T_STRING);
+}
+
+
+inline void Parser::assertNonContainer()
+{
+	poco_assert(_type == JSON_T_NULL ||
+		_type == JSON_T_FALSE ||
+		_type == JSON_T_TRUE ||
+		_type == JSON_T_FLOAT ||
+		_type == JSON_T_INTEGER ||
+		_type == JSON_T_STRING);
+}
+
+
+inline void Parser::growBuffer()
+{
+	_parseBuffer.resize(_parseBuffer.size() * 2, true);
+}
+
 }} // namespace Poco::JSON
 
 

JSON/testsuite/src/JSONTest.cpp

@@ -1193,6 +1193,7 @@ void JSONTest::testInvalidJanssonFiles()
 				std::cout << filePath.toString() << std::endl;
 
 				Parser parser;
+				parser.allowNullByte(false);
 				Var result;
 
 				try
@@ -1235,6 +1236,7 @@ void JSONTest::testInvalidUnicodeJanssonFiles()
 				std::cout << filePath.toString() << std::endl;
 
 				Parser parser;
+				parser.allowNullByte(false);
 				Var result;
 
 				try