diff options
author | 2015-07-20 08:11:41 +0200 | |
---|---|---|
committer | 2015-07-20 08:11:41 +0200 | |
commit | 4bcb0c13f50ddb1cd3fd0341b4bb31d9c3c7fc7a (patch) | |
tree | 55eccacfca0e69435d0d4e6615205310ca11f271 /3rdparty/jsoncpp/src/lib_json/json_reader.cpp | |
parent | d132946c6f45aa8d271c6180e86f72dd08243048 (diff) | |
parent | 229785f695b26c702c8490792611f59d0366a933 (diff) |
Merge pull request #5 from mamedev/master
Sync to base master
Diffstat (limited to '3rdparty/jsoncpp/src/lib_json/json_reader.cpp')
-rw-r--r-- | 3rdparty/jsoncpp/src/lib_json/json_reader.cpp | 1272 |
1 files changed, 1182 insertions, 90 deletions
diff --git a/3rdparty/jsoncpp/src/lib_json/json_reader.cpp b/3rdparty/jsoncpp/src/lib_json/json_reader.cpp index c5111f8d708..ceff66c3134 100644 --- a/3rdparty/jsoncpp/src/lib_json/json_reader.cpp +++ b/3rdparty/jsoncpp/src/lib_json/json_reader.cpp @@ -14,6 +14,9 @@ #include <cassert> #include <cstring> #include <istream> +#include <sstream> +#include <memory> +#include <set> #if defined(_MSC_VER) && _MSC_VER < 1500 // VC++ 8.0 and below #define snprintf _snprintf @@ -24,8 +27,17 @@ #pragma warning(disable : 4996) #endif +static int const stackLimit_g = 1000; +static int stackDepth_g = 0; // see readValue() + namespace Json { +#if __cplusplus >= 201103L +typedef std::unique_ptr<CharReader> CharReaderPtr; +#else +typedef std::auto_ptr<CharReader> CharReaderPtr; +#endif + // Implementation of class Features // //////////////////////////////// @@ -47,23 +59,6 @@ Features Features::strictMode() { // Implementation of class Reader // //////////////////////////////// -static inline bool in(Reader::Char c, - Reader::Char c1, - Reader::Char c2, - Reader::Char c3, - Reader::Char c4) { - return c == c1 || c == c2 || c == c3 || c == c4; -} - -static inline bool in(Reader::Char c, - Reader::Char c1, - Reader::Char c2, - Reader::Char c3, - Reader::Char c4, - Reader::Char c5) { - return c == c1 || c == c2 || c == c3 || c == c4 || c == c5; -} - static bool containsNewLine(Reader::Location begin, Reader::Location end) { for (; begin < end; ++begin) if (*begin == '\n' || *begin == '\r') @@ -125,6 +120,7 @@ bool Reader::parse(const char* beginDoc, nodes_.pop(); nodes_.push(&root); + stackDepth_g = 0; // Yes, this is bad coding, but options are limited. bool successful = readValue(); Token token; skipCommentTokens(token); @@ -147,19 +143,18 @@ bool Reader::parse(const char* beginDoc, } bool Reader::readValue() { + // This is a non-reentrant way to support a stackLimit. Terrible! + // But this deprecated class has a security problem: Bad input can + // cause a seg-fault. This seems like a fair, binary-compatible way + // to prevent the problem. + if (stackDepth_g >= stackLimit_g) throwRuntimeError("Exceeded stackLimit in readValue()."); + ++stackDepth_g; + Token token; skipCommentTokens(token); bool successful = true; if (collectComments_ && !commentsBefore_.empty()) { - // Remove newline characters at the end of the comments - size_t lastNonNewline = commentsBefore_.find_last_not_of("\r\n"); - if (lastNonNewline != std::string::npos) { - commentsBefore_.erase(lastNonNewline + 1); - } else { - commentsBefore_.clear(); - } - currentValue().setComment(commentsBefore_, commentBefore); commentsBefore_ = ""; } @@ -180,31 +175,42 @@ bool Reader::readValue() { successful = decodeString(token); break; case tokenTrue: - currentValue() = true; + { + Value v(true); + currentValue().swapPayload(v); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); + } break; case tokenFalse: - currentValue() = false; + { + Value v(false); + currentValue().swapPayload(v); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); + } break; case tokenNull: - currentValue() = Value(); + { + Value v; + currentValue().swapPayload(v); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); + } break; case tokenArraySeparator: + case tokenObjectEnd: + case tokenArrayEnd: if (features_.allowDroppedNullPlaceholders_) { // "Un-read" the current token and mark the current value as a null // token. current_--; - currentValue() = Value(); + Value v; + currentValue().swapPayload(v); currentValue().setOffsetStart(current_ - begin_ - 1); currentValue().setOffsetLimit(current_ - begin_); break; - } - // Else, fall through... + } // Else, fall through... default: currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); @@ -216,6 +222,7 @@ bool Reader::readValue() { lastValue_ = ¤tValue(); } + --stackDepth_g; return successful; } @@ -229,13 +236,6 @@ void Reader::skipCommentTokens(Token& token) { } } -bool Reader::expectToken(TokenType type, Token& token, const char* message) { - readToken(token); - if (token.type_ != type) - return addError(message, token); - return true; -} - bool Reader::readToken(Token& token) { skipSpaces(); token.start_ = current_; @@ -351,14 +351,34 @@ bool Reader::readComment() { return true; } +static std::string normalizeEOL(Reader::Location begin, Reader::Location end) { + std::string normalized; + normalized.reserve(end - begin); + Reader::Location current = begin; + while (current != end) { + char c = *current++; + if (c == '\r') { + if (current != end && *current == '\n') + // convert dos EOL + ++current; + // convert Mac EOL + normalized += '\n'; + } else { + normalized += c; + } + } + return normalized; +} + void Reader::addComment(Location begin, Location end, CommentPlacement placement) { assert(collectComments_); + const std::string& normalized = normalizeEOL(begin, end); if (placement == commentAfterOnSameLine) { assert(lastValue_ != 0); - lastValue_->setComment(std::string(begin, end), placement); + lastValue_->setComment(normalized, placement); } else { - commentsBefore_ += std::string(begin, end); + commentsBefore_ += normalized; } } @@ -374,18 +394,38 @@ bool Reader::readCStyleComment() { bool Reader::readCppStyleComment() { while (current_ != end_) { Char c = getNextChar(); - if (c == '\r' || c == '\n') + if (c == '\n') + break; + if (c == '\r') { + // Consume DOS EOL. It will be normalized in addComment. + if (current_ != end_ && *current_ == '\n') + getNextChar(); + // Break on Moc OS 9 EOL. break; + } } return true; } void Reader::readNumber() { - while (current_ != end_) { - if (!(*current_ >= '0' && *current_ <= '9') && - !in(*current_, '.', 'e', 'E', '+', '-')) - break; - ++current_; + const char *p = current_; + char c = '0'; // stopgap for already consumed character + // integral part + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; + // fractional part + if (c == '.') { + c = (current_ = p) < end_ ? *p++ : 0; + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; + } + // exponential part + if (c == 'e' || c == 'E') { + c = (current_ = p) < end_ ? *p++ : 0; + if (c == '+' || c == '-') + c = (current_ = p) < end_ ? *p++ : 0; + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; } } @@ -404,7 +444,8 @@ bool Reader::readString() { bool Reader::readObject(Token& tokenStart) { Token tokenName; std::string name; - currentValue() = Value(objectValue); + Value init(objectValue); + currentValue().swapPayload(init); currentValue().setOffsetStart(tokenStart.start_ - begin_); while (readToken(tokenName)) { bool initialTokenOk = true; @@ -457,7 +498,8 @@ bool Reader::readObject(Token& tokenStart) { } bool Reader::readArray(Token& tokenStart) { - currentValue() = Value(arrayValue); + Value init(arrayValue); + currentValue().swapPayload(init); currentValue().setOffsetStart(tokenStart.start_ - begin_); skipSpaces(); if (*current_ == ']') // empty array @@ -497,20 +539,13 @@ bool Reader::decodeNumber(Token& token) { Value decoded; if (!decodeNumber(token, decoded)) return false; - currentValue() = decoded; + currentValue().swapPayload(decoded); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); return true; } bool Reader::decodeNumber(Token& token, Value& decoded) { - bool isDouble = false; - for (Location inspect = token.start_; inspect != token.end_; ++inspect) { - isDouble = isDouble || in(*inspect, '.', 'e', 'E', '+') || - (*inspect == '-' && inspect != token.start_); - } - if (isDouble) - return decodeDouble(token, decoded); // Attempts to parse the number as an integer. If the number is // larger than the maximum supported value of an integer then // we decode the number as a double. @@ -518,6 +553,7 @@ bool Reader::decodeNumber(Token& token, Value& decoded) { bool isNegative = *current == '-'; if (isNegative) ++current; + // TODO: Help the compiler do the div and mod at compile time or get rid of them. Value::LargestUInt maxIntegerValue = isNegative ? Value::LargestUInt(-Value::minLargestInt) : Value::maxLargestUInt; @@ -526,9 +562,7 @@ bool Reader::decodeNumber(Token& token, Value& decoded) { while (current < token.end_) { Char c = *current++; if (c < '0' || c > '9') - return addError("'" + std::string(token.start_, token.end_) + - "' is not a number.", - token); + return decodeDouble(token, decoded); Value::UInt digit(c - '0'); if (value >= threshold) { // We've hit or exceeded the max value divided by 10 (rounded down). If @@ -555,7 +589,7 @@ bool Reader::decodeDouble(Token& token) { Value decoded; if (!decodeDouble(token, decoded)) return false; - currentValue() = decoded; + currentValue().swapPayload(decoded); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); return true; @@ -563,6 +597,945 @@ bool Reader::decodeDouble(Token& token) { bool Reader::decodeDouble(Token& token, Value& decoded) { double value = 0; + std::string buffer(token.start_, token.end_); + std::istringstream is(buffer); + if (!(is >> value)) + return addError("'" + std::string(token.start_, token.end_) + + "' is not a number.", + token); + decoded = value; + return true; +} + +bool Reader::decodeString(Token& token) { + std::string decoded_string; + if (!decodeString(token, decoded_string)) + return false; + Value decoded(decoded_string); + currentValue().swapPayload(decoded); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + return true; +} + +bool Reader::decodeString(Token& token, std::string& decoded) { + decoded.reserve(token.end_ - token.start_ - 2); + Location current = token.start_ + 1; // skip '"' + Location end = token.end_ - 1; // do not include '"' + while (current != end) { + Char c = *current++; + if (c == '"') + break; + else if (c == '\\') { + if (current == end) + return addError("Empty escape sequence in string", token, current); + Char escape = *current++; + switch (escape) { + case '"': + decoded += '"'; + break; + case '/': + decoded += '/'; + break; + case '\\': + decoded += '\\'; + break; + case 'b': + decoded += '\b'; + break; + case 'f': + decoded += '\f'; + break; + case 'n': + decoded += '\n'; + break; + case 'r': + decoded += '\r'; + break; + case 't': + decoded += '\t'; + break; + case 'u': { + unsigned int unicode; + if (!decodeUnicodeCodePoint(token, current, end, unicode)) + return false; + decoded += codePointToUTF8(unicode); + } break; + default: + return addError("Bad escape sequence in string", token, current); + } + } else { + decoded += c; + } + } + return true; +} + +bool Reader::decodeUnicodeCodePoint(Token& token, + Location& current, + Location end, + unsigned int& unicode) { + + if (!decodeUnicodeEscapeSequence(token, current, end, unicode)) + return false; + if (unicode >= 0xD800 && unicode <= 0xDBFF) { + // surrogate pairs + if (end - current < 6) + return addError( + "additional six characters expected to parse unicode surrogate pair.", + token, + current); + unsigned int surrogatePair; + if (*(current++) == '\\' && *(current++) == 'u') { + if (decodeUnicodeEscapeSequence(token, current, end, surrogatePair)) { + unicode = 0x10000 + ((unicode & 0x3FF) << 10) + (surrogatePair & 0x3FF); + } else + return false; + } else + return addError("expecting another \\u token to begin the second half of " + "a unicode surrogate pair", + token, + current); + } + return true; +} + +bool Reader::decodeUnicodeEscapeSequence(Token& token, + Location& current, + Location end, + unsigned int& unicode) { + if (end - current < 4) + return addError( + "Bad unicode escape sequence in string: four digits expected.", + token, + current); + unicode = 0; + for (int index = 0; index < 4; ++index) { + Char c = *current++; + unicode *= 16; + if (c >= '0' && c <= '9') + unicode += c - '0'; + else if (c >= 'a' && c <= 'f') + unicode += c - 'a' + 10; + else if (c >= 'A' && c <= 'F') + unicode += c - 'A' + 10; + else + return addError( + "Bad unicode escape sequence in string: hexadecimal digit expected.", + token, + current); + } + return true; +} + +bool +Reader::addError(const std::string& message, Token& token, Location extra) { + ErrorInfo info; + info.token_ = token; + info.message_ = message; + info.extra_ = extra; + errors_.push_back(info); + return false; +} + +bool Reader::recoverFromError(TokenType skipUntilToken) { + int errorCount = int(errors_.size()); + Token skip; + for (;;) { + if (!readToken(skip)) + errors_.resize(errorCount); // discard errors caused by recovery + if (skip.type_ == skipUntilToken || skip.type_ == tokenEndOfStream) + break; + } + errors_.resize(errorCount); + return false; +} + +bool Reader::addErrorAndRecover(const std::string& message, + Token& token, + TokenType skipUntilToken) { + addError(message, token); + return recoverFromError(skipUntilToken); +} + +Value& Reader::currentValue() { return *(nodes_.top()); } + +Reader::Char Reader::getNextChar() { + if (current_ == end_) + return 0; + return *current_++; +} + +void Reader::getLocationLineAndColumn(Location location, + int& line, + int& column) const { + Location current = begin_; + Location lastLineStart = current; + line = 0; + while (current < location && current != end_) { + Char c = *current++; + if (c == '\r') { + if (*current == '\n') + ++current; + lastLineStart = current; + ++line; + } else if (c == '\n') { + lastLineStart = current; + ++line; + } + } + // column & line start at 1 + column = int(location - lastLineStart) + 1; + ++line; +} + +std::string Reader::getLocationLineAndColumn(Location location) const { + int line, column; + getLocationLineAndColumn(location, line, column); + char buffer[18 + 16 + 16 + 1]; +#if defined(_MSC_VER) && defined(__STDC_SECURE_LIB__) +#if defined(WINCE) + _snprintf(buffer, sizeof(buffer), "Line %d, Column %d", line, column); +#else + sprintf_s(buffer, sizeof(buffer), "Line %d, Column %d", line, column); +#endif +#else + snprintf(buffer, sizeof(buffer), "Line %d, Column %d", line, column); +#endif + return buffer; +} + +// Deprecated. Preserved for backward compatibility +std::string Reader::getFormatedErrorMessages() const { + return getFormattedErrorMessages(); +} + +std::string Reader::getFormattedErrorMessages() const { + std::string formattedMessage; + for (Errors::const_iterator itError = errors_.begin(); + itError != errors_.end(); + ++itError) { + const ErrorInfo& error = *itError; + formattedMessage += + "* " + getLocationLineAndColumn(error.token_.start_) + "\n"; + formattedMessage += " " + error.message_ + "\n"; + if (error.extra_) + formattedMessage += + "See " + getLocationLineAndColumn(error.extra_) + " for detail.\n"; + } + return formattedMessage; +} + +std::vector<Reader::StructuredError> Reader::getStructuredErrors() const { + std::vector<Reader::StructuredError> allErrors; + for (Errors::const_iterator itError = errors_.begin(); + itError != errors_.end(); + ++itError) { + const ErrorInfo& error = *itError; + Reader::StructuredError structured; + structured.offset_start = error.token_.start_ - begin_; + structured.offset_limit = error.token_.end_ - begin_; + structured.message = error.message_; + allErrors.push_back(structured); + } + return allErrors; +} + +bool Reader::pushError(const Value& value, const std::string& message) { + size_t length = end_ - begin_; + if(value.getOffsetStart() > length + || value.getOffsetLimit() > length) + return false; + Token token; + token.type_ = tokenError; + token.start_ = begin_ + value.getOffsetStart(); + token.end_ = end_ + value.getOffsetLimit(); + ErrorInfo info; + info.token_ = token; + info.message_ = message; + info.extra_ = 0; + errors_.push_back(info); + return true; +} + +bool Reader::pushError(const Value& value, const std::string& message, const Value& extra) { + size_t length = end_ - begin_; + if(value.getOffsetStart() > length + || value.getOffsetLimit() > length + || extra.getOffsetLimit() > length) + return false; + Token token; + token.type_ = tokenError; + token.start_ = begin_ + value.getOffsetStart(); + token.end_ = begin_ + value.getOffsetLimit(); + ErrorInfo info; + info.token_ = token; + info.message_ = message; + info.extra_ = begin_ + extra.getOffsetStart(); + errors_.push_back(info); + return true; +} + +bool Reader::good() const { + return !errors_.size(); +} + +// exact copy of Features +class OurFeatures { +public: + static OurFeatures all(); + OurFeatures(); + bool allowComments_; + bool strictRoot_; + bool allowDroppedNullPlaceholders_; + bool allowNumericKeys_; + bool allowSingleQuotes_; + bool failIfExtra_; + bool rejectDupKeys_; + int stackLimit_; +}; // OurFeatures + +// exact copy of Implementation of class Features +// //////////////////////////////// + +OurFeatures::OurFeatures() + : allowComments_(true), strictRoot_(false) + , allowDroppedNullPlaceholders_(false), allowNumericKeys_(false) + , allowSingleQuotes_(false) + , failIfExtra_(false) +{ +} + +OurFeatures OurFeatures::all() { return OurFeatures(); } + +// Implementation of class Reader +// //////////////////////////////// + +// exact copy of Reader, renamed to OurReader +class OurReader { +public: + typedef char Char; + typedef const Char* Location; + struct StructuredError { + size_t offset_start; + size_t offset_limit; + std::string message; + }; + + OurReader(OurFeatures const& features); + bool parse(const char* beginDoc, + const char* endDoc, + Value& root, + bool collectComments = true); + std::string getFormattedErrorMessages() const; + std::vector<StructuredError> getStructuredErrors() const; + bool pushError(const Value& value, const std::string& message); + bool pushError(const Value& value, const std::string& message, const Value& extra); + bool good() const; + +private: + OurReader(OurReader const&); // no impl + void operator=(OurReader const&); // no impl + + enum TokenType { + tokenEndOfStream = 0, + tokenObjectBegin, + tokenObjectEnd, + tokenArrayBegin, + tokenArrayEnd, + tokenString, + tokenNumber, + tokenTrue, + tokenFalse, + tokenNull, + tokenArraySeparator, + tokenMemberSeparator, + tokenComment, + tokenError + }; + + class Token { + public: + TokenType type_; + Location start_; + Location end_; + }; + + class ErrorInfo { + public: + Token token_; + std::string message_; + Location extra_; + }; + + typedef std::deque<ErrorInfo> Errors; + + bool readToken(Token& token); + void skipSpaces(); + bool match(Location pattern, int patternLength); + bool readComment(); + bool readCStyleComment(); + bool readCppStyleComment(); + bool readString(); + bool readStringSingleQuote(); + void readNumber(); + bool readValue(); + bool readObject(Token& token); + bool readArray(Token& token); + bool decodeNumber(Token& token); + bool decodeNumber(Token& token, Value& decoded); + bool decodeString(Token& token); + bool decodeString(Token& token, std::string& decoded); + bool decodeDouble(Token& token); + bool decodeDouble(Token& token, Value& decoded); + bool decodeUnicodeCodePoint(Token& token, + Location& current, + Location end, + unsigned int& unicode); + bool decodeUnicodeEscapeSequence(Token& token, + Location& current, + Location end, + unsigned int& unicode); + bool addError(const std::string& message, Token& token, Location extra = 0); + bool recoverFromError(TokenType skipUntilToken); + bool addErrorAndRecover(const std::string& message, + Token& token, + TokenType skipUntilToken); + void skipUntilSpace(); + Value& currentValue(); + Char getNextChar(); + void + getLocationLineAndColumn(Location location, int& line, int& column) const; + std::string getLocationLineAndColumn(Location location) const; + void addComment(Location begin, Location end, CommentPlacement placement); + void skipCommentTokens(Token& token); + + typedef std::stack<Value*> Nodes; + Nodes nodes_; + Errors errors_; + std::string document_; + Location begin_; + Location end_; + Location current_; + Location lastValueEnd_; + Value* lastValue_; + std::string commentsBefore_; + int stackDepth_; + + OurFeatures const features_; + bool collectComments_; +}; // OurReader + +// complete copy of Read impl, for OurReader + +OurReader::OurReader(OurFeatures const& features) + : errors_(), document_(), begin_(), end_(), current_(), lastValueEnd_(), + lastValue_(), commentsBefore_(), features_(features), collectComments_() { +} + +bool OurReader::parse(const char* beginDoc, + const char* endDoc, + Value& root, + bool collectComments) { + if (!features_.allowComments_) { + collectComments = false; + } + + begin_ = beginDoc; + end_ = endDoc; + collectComments_ = collectComments; + current_ = begin_; + lastValueEnd_ = 0; + lastValue_ = 0; + commentsBefore_ = ""; + errors_.clear(); + while (!nodes_.empty()) + nodes_.pop(); + nodes_.push(&root); + + stackDepth_ = 0; + bool successful = readValue(); + Token token; + skipCommentTokens(token); + if (features_.failIfExtra_) { + if (token.type_ != tokenError && token.type_ != tokenEndOfStream) { + addError("Extra non-whitespace after JSON value.", token); + return false; + } + } + if (collectComments_ && !commentsBefore_.empty()) + root.setComment(commentsBefore_, commentAfter); + if (features_.strictRoot_) { + if (!root.isArray() && !root.isObject()) { + // Set error location to start of doc, ideally should be first token found + // in doc + token.type_ = tokenError; + token.start_ = beginDoc; + token.end_ = endDoc; + addError( + "A valid JSON document must be either an array or an object value.", + token); + return false; + } + } + return successful; +} + +bool OurReader::readValue() { + if (stackDepth_ >= features_.stackLimit_) throwRuntimeError("Exceeded stackLimit in readValue()."); + ++stackDepth_; + Token token; + skipCommentTokens(token); + bool successful = true; + + if (collectComments_ && !commentsBefore_.empty()) { + currentValue().setComment(commentsBefore_, commentBefore); + commentsBefore_ = ""; + } + + switch (token.type_) { + case tokenObjectBegin: + successful = readObject(token); + currentValue().setOffsetLimit(current_ - begin_); + break; + case tokenArrayBegin: + successful = readArray(token); + currentValue().setOffsetLimit(current_ - begin_); + break; + case tokenNumber: + successful = decodeNumber(token); + break; + case tokenString: + successful = decodeString(token); + break; + case tokenTrue: + { + Value v(true); + currentValue().swapPayload(v); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + } + break; + case tokenFalse: + { + Value v(false); + currentValue().swapPayload(v); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + } + break; + case tokenNull: + { + Value v; + currentValue().swapPayload(v); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + } + break; + case tokenArraySeparator: + case tokenObjectEnd: + case tokenArrayEnd: + if (features_.allowDroppedNullPlaceholders_) { + // "Un-read" the current token and mark the current value as a null + // token. + current_--; + Value v; + currentValue().swapPayload(v); + currentValue().setOffsetStart(current_ - begin_ - 1); + currentValue().setOffsetLimit(current_ - begin_); + break; + } // else, fall through ... + default: + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + return addError("Syntax error: value, object or array expected.", token); + } + + if (collectComments_) { + lastValueEnd_ = current_; + lastValue_ = ¤tValue(); + } + + --stackDepth_; + return successful; +} + +void OurReader::skipCommentTokens(Token& token) { + if (features_.allowComments_) { + do { + readToken(token); + } while (token.type_ == tokenComment); + } else { + readToken(token); + } +} + +bool OurReader::readToken(Token& token) { + skipSpaces(); + token.start_ = current_; + Char c = getNextChar(); + bool ok = true; + switch (c) { + case '{': + token.type_ = tokenObjectBegin; + break; + case '}': + token.type_ = tokenObjectEnd; + break; + case '[': + token.type_ = tokenArrayBegin; + break; + case ']': + token.type_ = tokenArrayEnd; + break; + case '"': + token.type_ = tokenString; + ok = readString(); + break; + case '\'': + if (features_.allowSingleQuotes_) { + token.type_ = tokenString; + ok = readStringSingleQuote(); + break; + } // else continue + case '/': + token.type_ = tokenComment; + ok = readComment(); + break; + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + case '8': + case '9': + case '-': + token.type_ = tokenNumber; + readNumber(); + break; + case 't': + token.type_ = tokenTrue; + ok = match("rue", 3); + break; + case 'f': + token.type_ = tokenFalse; + ok = match("alse", 4); + break; + case 'n': + token.type_ = tokenNull; + ok = match("ull", 3); + break; + case ',': + token.type_ = tokenArraySeparator; + break; + case ':': + token.type_ = tokenMemberSeparator; + break; + case 0: + token.type_ = tokenEndOfStream; + break; + default: + ok = false; + break; + } + if (!ok) + token.type_ = tokenError; + token.end_ = current_; + return true; +} + +void OurReader::skipSpaces() { + while (current_ != end_) { + Char c = *current_; + if (c == ' ' || c == '\t' || c == '\r' || c == '\n') + ++current_; + else + break; + } +} + +bool OurReader::match(Location pattern, int patternLength) { + if (end_ - current_ < patternLength) + return false; + int index = patternLength; + while (index--) + if (current_[index] != pattern[index]) + return false; + current_ += patternLength; + return true; +} + +bool OurReader::readComment() { + Location commentBegin = current_ - 1; + Char c = getNextChar(); + bool successful = false; + if (c == '*') + successful = readCStyleComment(); + else if (c == '/') + successful = readCppStyleComment(); + if (!successful) + return false; + + if (collectComments_) { + CommentPlacement placement = commentBefore; + if (lastValueEnd_ && !containsNewLine(lastValueEnd_, commentBegin)) { + if (c != '*' || !containsNewLine(commentBegin, current_)) + placement = commentAfterOnSameLine; + } + + addComment(commentBegin, current_, placement); + } + return true; +} + +void +OurReader::addComment(Location begin, Location end, CommentPlacement placement) { + assert(collectComments_); + const std::string& normalized = normalizeEOL(begin, end); + if (placement == commentAfterOnSameLine) { + assert(lastValue_ != 0); + lastValue_->setComment(normalized, placement); + } else { + commentsBefore_ += normalized; + } +} + +bool OurReader::readCStyleComment() { + while (current_ != end_) { + Char c = getNextChar(); + if (c == '*' && *current_ == '/') + break; + } + return getNextChar() == '/'; +} + +bool OurReader::readCppStyleComment() { + while (current_ != end_) { + Char c = getNextChar(); + if (c == '\n') + break; + if (c == '\r') { + // Consume DOS EOL. It will be normalized in addComment. + if (current_ != end_ && *current_ == '\n') + getNextChar(); + // Break on Moc OS 9 EOL. + break; + } + } + return true; +} + +void OurReader::readNumber() { + const char *p = current_; + char c = '0'; // stopgap for already consumed character + // integral part + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; + // fractional part + if (c == '.') { + c = (current_ = p) < end_ ? *p++ : 0; + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; + } + // exponential part + if (c == 'e' || c == 'E') { + c = (current_ = p) < end_ ? *p++ : 0; + if (c == '+' || c == '-') + c = (current_ = p) < end_ ? *p++ : 0; + while (c >= '0' && c <= '9') + c = (current_ = p) < end_ ? *p++ : 0; + } +} +bool OurReader::readString() { + Char c = 0; + while (current_ != end_) { + c = getNextChar(); + if (c == '\\') + getNextChar(); + else if (c == '"') + break; + } + return c == '"'; +} + + +bool OurReader::readStringSingleQuote() { + Char c = 0; + while (current_ != end_) { + c = getNextChar(); + if (c == '\\') + getNextChar(); + else if (c == '\'') + break; + } + return c == '\''; +} + +bool OurReader::readObject(Token& tokenStart) { + Token tokenName; + std::string name; + Value init(objectValue); + currentValue().swapPayload(init); + currentValue().setOffsetStart(tokenStart.start_ - begin_); + while (readToken(tokenName)) { + bool initialTokenOk = true; + while (tokenName.type_ == tokenComment && initialTokenOk) + initialTokenOk = readToken(tokenName); + if (!initialTokenOk) + break; + if (tokenName.type_ == tokenObjectEnd && name.empty()) // empty object + return true; + name = ""; + if (tokenName.type_ == tokenString) { + if (!decodeString(tokenName, name)) + return recoverFromError(tokenObjectEnd); + } else if (tokenName.type_ == tokenNumber && features_.allowNumericKeys_) { + Value numberName; + if (!decodeNumber(tokenName, numberName)) + return recoverFromError(tokenObjectEnd); + name = numberName.asString(); + } else { + break; + } + + Token colon; + if (!readToken(colon) || colon.type_ != tokenMemberSeparator) { + return addErrorAndRecover( + "Missing ':' after object member name", colon, tokenObjectEnd); + } + if (name.length() >= (1U<<30)) throwRuntimeError("keylength >= 2^30"); + if (features_.rejectDupKeys_ && currentValue().isMember(name)) { + std::string msg = "Duplicate key: '" + name + "'"; + return addErrorAndRecover( + msg, tokenName, tokenObjectEnd); + } + Value& value = currentValue()[name]; + nodes_.push(&value); + bool ok = readValue(); + nodes_.pop(); + if (!ok) // error already set + return recoverFromError(tokenObjectEnd); + + Token comma; + if (!readToken(comma) || + (comma.type_ != tokenObjectEnd && comma.type_ != tokenArraySeparator && + comma.type_ != tokenComment)) { + return addErrorAndRecover( + "Missing ',' or '}' in object declaration", comma, tokenObjectEnd); + } + bool finalizeTokenOk = true; + while (comma.type_ == tokenComment && finalizeTokenOk) + finalizeTokenOk = readToken(comma); + if (comma.type_ == tokenObjectEnd) + return true; + } + return addErrorAndRecover( + "Missing '}' or object member name", tokenName, tokenObjectEnd); +} + +bool OurReader::readArray(Token& tokenStart) { + Value init(arrayValue); + currentValue().swapPayload(init); + currentValue().setOffsetStart(tokenStart.start_ - begin_); + skipSpaces(); + if (*current_ == ']') // empty array + { + Token endArray; + readToken(endArray); + return true; + } + int index = 0; + for (;;) { + Value& value = currentValue()[index++]; + nodes_.push(&value); + bool ok = readValue(); + nodes_.pop(); + if (!ok) // error already set + return recoverFromError(tokenArrayEnd); + + Token token; + // Accept Comment after last item in the array. + ok = readToken(token); + while (token.type_ == tokenComment && ok) { + ok = readToken(token); + } + bool badTokenType = + (token.type_ != tokenArraySeparator && token.type_ != tokenArrayEnd); + if (!ok || badTokenType) { + return addErrorAndRecover( + "Missing ',' or ']' in array declaration", token, tokenArrayEnd); + } + if (token.type_ == tokenArrayEnd) + break; + } + return true; +} + +bool OurReader::decodeNumber(Token& token) { + Value decoded; + if (!decodeNumber(token, decoded)) + return false; + currentValue().swapPayload(decoded); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + return true; +} + +bool OurReader::decodeNumber(Token& token, Value& decoded) { + // Attempts to parse the number as an integer. If the number is + // larger than the maximum supported value of an integer then + // we decode the number as a double. + Location current = token.start_; + bool isNegative = *current == '-'; + if (isNegative) + ++current; + // TODO: Help the compiler do the div and mod at compile time or get rid of them. + Value::LargestUInt maxIntegerValue = + isNegative ? Value::LargestUInt(-Value::minLargestInt) + : Value::maxLargestUInt; + Value::LargestUInt threshold = maxIntegerValue / 10; + Value::LargestUInt value = 0; + while (current < token.end_) { + Char c = *current++; + if (c < '0' || c > '9') + return decodeDouble(token, decoded); + Value::UInt digit(c - '0'); + if (value >= threshold) { + // We've hit or exceeded the max value divided by 10 (rounded down). If + // a) we've only just touched the limit, b) this is the last digit, and + // c) it's small enough to fit in that rounding delta, we're okay. + // Otherwise treat this number as a double to avoid overflow. + if (value > threshold || current != token.end_ || + digit > maxIntegerValue % 10) { + return decodeDouble(token, decoded); + } + } + value = value * 10 + digit; + } + if (isNegative) + decoded = -Value::LargestInt(value); + else if (value <= Value::LargestUInt(Value::maxInt)) + decoded = Value::LargestInt(value); + else + decoded = value; + return true; +} + +bool OurReader::decodeDouble(Token& token) { + Value decoded; + if (!decodeDouble(token, decoded)) + return false; + currentValue().swapPayload(decoded); + currentValue().setOffsetStart(token.start_ - begin_); + currentValue().setOffsetLimit(token.end_ - begin_); + return true; +} + +bool OurReader::decodeDouble(Token& token, Value& decoded) { + double value = 0; const int bufferSize = 32; int count; int length = int(token.end_ - token.start_); @@ -597,17 +1570,18 @@ bool Reader::decodeDouble(Token& token, Value& decoded) { return true; } -bool Reader::decodeString(Token& token) { - std::string decoded; - if (!decodeString(token, decoded)) +bool OurReader::decodeString(Token& token) { + std::string decoded_string; + if (!decodeString(token, decoded_string)) return false; - currentValue() = decoded; + Value decoded(decoded_string); + currentValue().swapPayload(decoded); currentValue().setOffsetStart(token.start_ - begin_); currentValue().setOffsetLimit(token.end_ - begin_); return true; } -bool Reader::decodeString(Token& token, std::string& decoded) { +bool OurReader::decodeString(Token& token, std::string& decoded) { decoded.reserve(token.end_ - token.start_ - 2); Location current = token.start_ + 1; // skip '"' Location end = token.end_ - 1; // do not include '"' @@ -660,7 +1634,7 @@ bool Reader::decodeString(Token& token, std::string& decoded) { return true; } -bool Reader::decodeUnicodeCodePoint(Token& token, +bool OurReader::decodeUnicodeCodePoint(Token& token, Location& current, Location end, unsigned int& unicode) { @@ -689,7 +1663,7 @@ bool Reader::decodeUnicodeCodePoint(Token& token, return true; } -bool Reader::decodeUnicodeEscapeSequence(Token& token, +bool OurReader::decodeUnicodeEscapeSequence(Token& token, Location& current, Location end, unsigned int& unicode) { @@ -718,7 +1692,7 @@ bool Reader::decodeUnicodeEscapeSequence(Token& token, } bool -Reader::addError(const std::string& message, Token& token, Location extra) { +OurReader::addError(const std::string& message, Token& token, Location extra) { ErrorInfo info; info.token_ = token; info.message_ = message; @@ -727,7 +1701,7 @@ Reader::addError(const std::string& message, Token& token, Location extra) { return false; } -bool Reader::recoverFromError(TokenType skipUntilToken) { +bool OurReader::recoverFromError(TokenType skipUntilToken) { int errorCount = int(errors_.size()); Token skip; for (;;) { @@ -740,22 +1714,22 @@ bool Reader::recoverFromError(TokenType skipUntilToken) { return false; } -bool Reader::addErrorAndRecover(const std::string& message, +bool OurReader::addErrorAndRecover(const std::string& message, Token& token, TokenType skipUntilToken) { addError(message, token); return recoverFromError(skipUntilToken); } -Value& Reader::currentValue() { return *(nodes_.top()); } +Value& OurReader::currentValue() { return *(nodes_.top()); } -Reader::Char Reader::getNextChar() { +OurReader::Char OurReader::getNextChar() { if (current_ == end_) return 0; return *current_++; } -void Reader::getLocationLineAndColumn(Location location, +void OurReader::getLocationLineAndColumn(Location location, int& line, int& column) const { Location current = begin_; @@ -778,7 +1752,7 @@ void Reader::getLocationLineAndColumn(Location location, ++line; } -std::string Reader::getLocationLineAndColumn(Location location) const { +std::string OurReader::getLocationLineAndColumn(Location location) const { int line, column; getLocationLineAndColumn(location, line, column); char buffer[18 + 16 + 16 + 1]; @@ -794,12 +1768,7 @@ std::string Reader::getLocationLineAndColumn(Location location) const { return buffer; } -// Deprecated. Preserved for backward compatibility -std::string Reader::getFormatedErrorMessages() const { - return getFormattedErrorMessages(); -} - -std::string Reader::getFormattedErrorMessages() const { +std::string OurReader::getFormattedErrorMessages() const { std::string formattedMessage; for (Errors::const_iterator itError = errors_.begin(); itError != errors_.end(); @@ -815,13 +1784,13 @@ std::string Reader::getFormattedErrorMessages() const { return formattedMessage; } -std::vector<Reader::StructuredError> Reader::getStructuredErrors() const { - std::vector<Reader::StructuredError> allErrors; +std::vector<OurReader::StructuredError> OurReader::getStructuredErrors() const { + std::vector<OurReader::StructuredError> allErrors; for (Errors::const_iterator itError = errors_.begin(); itError != errors_.end(); ++itError) { const ErrorInfo& error = *itError; - Reader::StructuredError structured; + OurReader::StructuredError structured; structured.offset_start = error.token_.start_ - begin_; structured.offset_limit = error.token_.end_ - begin_; structured.message = error.message_; @@ -830,7 +1799,7 @@ std::vector<Reader::StructuredError> Reader::getStructuredErrors() const { return allErrors; } -bool Reader::pushError(const Value& value, const std::string& message) { +bool OurReader::pushError(const Value& value, const std::string& message) { size_t length = end_ - begin_; if(value.getOffsetStart() > length || value.getOffsetLimit() > length) @@ -847,7 +1816,7 @@ bool Reader::pushError(const Value& value, const std::string& message) { return true; } -bool Reader::pushError(const Value& value, const std::string& message, const Value& extra) { +bool OurReader::pushError(const Value& value, const std::string& message, const Value& extra) { size_t length = end_ - begin_; if(value.getOffsetStart() > length || value.getOffsetLimit() > length @@ -865,19 +1834,142 @@ bool Reader::pushError(const Value& value, const std::string& message, const Val return true; } -bool Reader::good() const { +bool OurReader::good() const { return !errors_.size(); } + +class OurCharReader : public CharReader { + bool const collectComments_; + OurReader reader_; +public: + OurCharReader( + bool collectComments, + OurFeatures const& features) + : collectComments_(collectComments) + , reader_(features) + {} + virtual bool parse( + char const* beginDoc, char const* endDoc, + Value* root, std::string* errs) { + bool ok = reader_.parse(beginDoc, endDoc, *root, collectComments_); + if (errs) { + *errs = reader_.getFormattedErrorMessages(); + } + return ok; + } +}; + +CharReaderBuilder::CharReaderBuilder() +{ + setDefaults(&settings_); +} +CharReaderBuilder::~CharReaderBuilder() +{} +CharReader* CharReaderBuilder::newCharReader() const +{ + bool collectComments = settings_["collectComments"].asBool(); + OurFeatures features = OurFeatures::all(); + features.allowComments_ = settings_["allowComments"].asBool(); + features.strictRoot_ = settings_["strictRoot"].asBool(); + features.allowDroppedNullPlaceholders_ = settings_["allowDroppedNullPlaceholders"].asBool(); + features.allowNumericKeys_ = settings_["allowNumericKeys"].asBool(); + features.allowSingleQuotes_ = settings_["allowSingleQuotes"].asBool(); + features.stackLimit_ = settings_["stackLimit"].asInt(); + features.failIfExtra_ = settings_["failIfExtra"].asBool(); + features.rejectDupKeys_ = settings_["rejectDupKeys"].asBool(); + return new OurCharReader(collectComments, features); +} +static void getValidReaderKeys(std::set<std::string>* valid_keys) +{ + valid_keys->clear(); + valid_keys->insert("collectComments"); + valid_keys->insert("allowComments"); + valid_keys->insert("strictRoot"); + valid_keys->insert("allowDroppedNullPlaceholders"); + valid_keys->insert("allowNumericKeys"); + valid_keys->insert("allowSingleQuotes"); + valid_keys->insert("stackLimit"); + valid_keys->insert("failIfExtra"); + valid_keys->insert("rejectDupKeys"); +} +bool CharReaderBuilder::validate(Json::Value* invalid) const +{ + Json::Value my_invalid; + if (!invalid) invalid = &my_invalid; // so we do not need to test for NULL + Json::Value& inv = *invalid; + std::set<std::string> valid_keys; + getValidReaderKeys(&valid_keys); + Value::Members keys = settings_.getMemberNames(); + size_t n = keys.size(); + for (size_t i = 0; i < n; ++i) { + std::string const& key = keys[i]; + if (valid_keys.find(key) == valid_keys.end()) { + inv[key] = settings_[key]; + } + } + return 0u == inv.size(); +} +Value& CharReaderBuilder::operator[](std::string key) +{ + return settings_[key]; +} +// static +void CharReaderBuilder::strictMode(Json::Value* settings) +{ +//! [CharReaderBuilderStrictMode] + (*settings)["allowComments"] = false; + (*settings)["strictRoot"] = true; + (*settings)["allowDroppedNullPlaceholders"] = false; + (*settings)["allowNumericKeys"] = false; + (*settings)["allowSingleQuotes"] = false; + (*settings)["failIfExtra"] = true; + (*settings)["rejectDupKeys"] = true; +//! [CharReaderBuilderStrictMode] +} +// static +void CharReaderBuilder::setDefaults(Json::Value* settings) +{ +//! [CharReaderBuilderDefaults] + (*settings)["collectComments"] = true; + (*settings)["allowComments"] = true; + (*settings)["strictRoot"] = false; + (*settings)["allowDroppedNullPlaceholders"] = false; + (*settings)["allowNumericKeys"] = false; + (*settings)["allowSingleQuotes"] = false; + (*settings)["stackLimit"] = 1000; + (*settings)["failIfExtra"] = false; + (*settings)["rejectDupKeys"] = false; +//! [CharReaderBuilderDefaults] +} + +////////////////////////////////// +// global functions + +bool parseFromStream( + CharReader::Factory const& fact, std::istream& sin, + Value* root, std::string* errs) +{ + std::ostringstream ssin; + ssin << sin.rdbuf(); + std::string doc = ssin.str(); + char const* begin = doc.data(); + char const* end = begin + doc.size(); + // Note that we do not actually need a null-terminator. + CharReaderPtr const reader(fact.newCharReader()); + return reader->parse(begin, end, root, errs); +} + std::istream& operator>>(std::istream& sin, Value& root) { - Json::Reader reader; - bool ok = reader.parse(sin, root, true); + CharReaderBuilder b; + std::string errs; + bool ok = parseFromStream(b, sin, &root, &errs); if (!ok) { fprintf(stderr, "Error from reader: %s", - reader.getFormattedErrorMessages().c_str()); + errs.c_str()); - JSON_FAIL_MESSAGE("reader error"); + throwRuntimeError("reader error"); } return sin; } |