Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 1 | // Copyright 2007-2011 Baptiste Lepilleur and The JsonCpp Authors |
| 2 | // Copyright (C) 2016 InfoTeCS JSC. All rights reserved. |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 3 | // Distributed under MIT license, or public domain if desired and |
| 4 | // recognized in your jurisdiction. |
| 5 | // See file LICENSE for detail or copy at http://jsoncpp.sourceforge.net/LICENSE |
| 6 | |
| 7 | #if !defined(JSON_IS_AMALGAMATION) |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 8 | #include "json_tool.h" |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 9 | #include <json/assertions.h> |
| 10 | #include <json/reader.h> |
| 11 | #include <json/value.h> |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 12 | #endif // if !defined(JSON_IS_AMALGAMATION) |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 13 | #include <algorithm> |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 14 | #include <cassert> |
| 15 | #include <cstring> |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 16 | #include <iostream> |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 17 | #include <istream> |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 18 | #include <limits> |
| 19 | #include <memory> |
| 20 | #include <set> |
| 21 | #include <sstream> |
| 22 | #include <utility> |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 23 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 24 | #include <cstdio> |
| 25 | #if __cplusplus >= 201103L |
| 26 | |
| 27 | #if !defined(sscanf) |
| 28 | #define sscanf std::sscanf |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 29 | #endif |
| 30 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 31 | #endif //__cplusplus |
| 32 | |
| 33 | #if defined(_MSC_VER) |
| 34 | #if !defined(_CRT_SECURE_CPP_OVERLOAD_STANDARD_NAMES) |
| 35 | #define _CRT_SECURE_CPP_OVERLOAD_STANDARD_NAMES 1 |
| 36 | #endif //_CRT_SECURE_CPP_OVERLOAD_STANDARD_NAMES |
| 37 | #endif //_MSC_VER |
| 38 | |
| 39 | #if defined(_MSC_VER) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 40 | // Disable warning about strdup being deprecated. |
| 41 | #pragma warning(disable : 4996) |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 42 | #endif |
| 43 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 44 | // Define JSONCPP_DEPRECATED_STACK_LIMIT as an appropriate integer at compile |
| 45 | // time to change the stack limit |
| 46 | #if !defined(JSONCPP_DEPRECATED_STACK_LIMIT) |
| 47 | #define JSONCPP_DEPRECATED_STACK_LIMIT 1000 |
| 48 | #endif |
| 49 | |
| 50 | static size_t const stackLimit_g = |
| 51 | JSONCPP_DEPRECATED_STACK_LIMIT; // see readValue() |
| 52 | |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 53 | namespace Json { |
| 54 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 55 | #if __cplusplus >= 201103L || (defined(_CPPLIB_VER) && _CPPLIB_VER >= 520) |
| 56 | using CharReaderPtr = std::unique_ptr<CharReader>; |
| 57 | #else |
| 58 | using CharReaderPtr = std::auto_ptr<CharReader>; |
| 59 | #endif |
| 60 | |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 61 | // Implementation of class Features |
| 62 | // //////////////////////////////// |
| 63 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 64 | Features::Features() = default; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 65 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 66 | Features Features::all() { return {}; } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 67 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 68 | Features Features::strictMode() { |
| 69 | Features features; |
| 70 | features.allowComments_ = false; |
| 71 | features.strictRoot_ = true; |
| 72 | features.allowDroppedNullPlaceholders_ = false; |
| 73 | features.allowNumericKeys_ = false; |
| 74 | return features; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 75 | } |
| 76 | |
| 77 | // Implementation of class Reader |
| 78 | // //////////////////////////////// |
| 79 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 80 | bool Reader::containsNewLine(Reader::Location begin, Reader::Location end) { |
| 81 | return std::any_of(begin, end, [](char b) { return b == '\n' || b == '\r'; }); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 82 | } |
| 83 | |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 84 | // Class Reader |
| 85 | // ////////////////////////////////////////////////////////////////// |
| 86 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 87 | Reader::Reader() : features_(Features::all()) {} |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 88 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 89 | Reader::Reader(const Features& features) : features_(features) {} |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 90 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 91 | bool Reader::parse(const std::string& document, Value& root, |
| 92 | bool collectComments) { |
| 93 | document_.assign(document.begin(), document.end()); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 94 | const char* begin = document_.c_str(); |
| 95 | const char* end = begin + document_.length(); |
| 96 | return parse(begin, end, root, collectComments); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 97 | } |
| 98 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 99 | bool Reader::parse(std::istream& is, Value& root, bool collectComments) { |
| 100 | // std::istream_iterator<char> begin(is); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 101 | // std::istream_iterator<char> end; |
| 102 | // Those would allow streamed input from a file, if parse() were a |
| 103 | // template function. |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 104 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 105 | // Since String is reference-counted, this at least does not |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 106 | // create an extra copy. |
Elliott Hughes | 1601ea0 | 2021-12-07 09:43:38 -0800 | [diff] [blame] | 107 | String doc(std::istreambuf_iterator<char>(is), {}); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 108 | return parse(doc.data(), doc.data() + doc.size(), root, collectComments); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 109 | } |
| 110 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 111 | bool Reader::parse(const char* beginDoc, const char* endDoc, Value& root, |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 112 | bool collectComments) { |
| 113 | if (!features_.allowComments_) { |
| 114 | collectComments = false; |
| 115 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 116 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 117 | begin_ = beginDoc; |
| 118 | end_ = endDoc; |
| 119 | collectComments_ = collectComments; |
| 120 | current_ = begin_; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 121 | lastValueEnd_ = nullptr; |
| 122 | lastValue_ = nullptr; |
| 123 | commentsBefore_.clear(); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 124 | errors_.clear(); |
| 125 | while (!nodes_.empty()) |
| 126 | nodes_.pop(); |
| 127 | nodes_.push(&root); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 128 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 129 | bool successful = readValue(); |
| 130 | Token token; |
| 131 | skipCommentTokens(token); |
| 132 | if (collectComments_ && !commentsBefore_.empty()) |
| 133 | root.setComment(commentsBefore_, commentAfter); |
| 134 | if (features_.strictRoot_) { |
| 135 | if (!root.isArray() && !root.isObject()) { |
| 136 | // Set error location to start of doc, ideally should be first token found |
| 137 | // in doc |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 138 | token.type_ = tokenError; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 139 | token.start_ = beginDoc; |
| 140 | token.end_ = endDoc; |
| 141 | addError( |
| 142 | "A valid JSON document must be either an array or an object value.", |
| 143 | token); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 144 | return false; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 145 | } |
| 146 | } |
| 147 | return successful; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 148 | } |
| 149 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 150 | bool Reader::readValue() { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 151 | // readValue() may call itself only if it calls readObject() or ReadArray(). |
| 152 | // These methods execute nodes_.push() just before and nodes_.pop)() just |
| 153 | // after calling readValue(). parse() executes one nodes_.push(), so > instead |
| 154 | // of >=. |
| 155 | if (nodes_.size() > stackLimit_g) |
| 156 | throwRuntimeError("Exceeded stackLimit in readValue()."); |
| 157 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 158 | Token token; |
| 159 | skipCommentTokens(token); |
| 160 | bool successful = true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 161 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 162 | if (collectComments_ && !commentsBefore_.empty()) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 163 | currentValue().setComment(commentsBefore_, commentBefore); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 164 | commentsBefore_.clear(); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 165 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 166 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 167 | switch (token.type_) { |
| 168 | case tokenObjectBegin: |
| 169 | successful = readObject(token); |
| 170 | currentValue().setOffsetLimit(current_ - begin_); |
| 171 | break; |
| 172 | case tokenArrayBegin: |
| 173 | successful = readArray(token); |
| 174 | currentValue().setOffsetLimit(current_ - begin_); |
| 175 | break; |
| 176 | case tokenNumber: |
| 177 | successful = decodeNumber(token); |
| 178 | break; |
| 179 | case tokenString: |
| 180 | successful = decodeString(token); |
| 181 | break; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 182 | case tokenTrue: { |
| 183 | Value v(true); |
| 184 | currentValue().swapPayload(v); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 185 | currentValue().setOffsetStart(token.start_ - begin_); |
| 186 | currentValue().setOffsetLimit(token.end_ - begin_); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 187 | } break; |
| 188 | case tokenFalse: { |
| 189 | Value v(false); |
| 190 | currentValue().swapPayload(v); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 191 | currentValue().setOffsetStart(token.start_ - begin_); |
| 192 | currentValue().setOffsetLimit(token.end_ - begin_); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 193 | } break; |
| 194 | case tokenNull: { |
| 195 | Value v; |
| 196 | currentValue().swapPayload(v); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 197 | currentValue().setOffsetStart(token.start_ - begin_); |
| 198 | currentValue().setOffsetLimit(token.end_ - begin_); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 199 | } break; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 200 | case tokenArraySeparator: |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 201 | case tokenObjectEnd: |
| 202 | case tokenArrayEnd: |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 203 | if (features_.allowDroppedNullPlaceholders_) { |
| 204 | // "Un-read" the current token and mark the current value as a null |
| 205 | // token. |
| 206 | current_--; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 207 | Value v; |
| 208 | currentValue().swapPayload(v); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 209 | currentValue().setOffsetStart(current_ - begin_ - 1); |
| 210 | currentValue().setOffsetLimit(current_ - begin_); |
| 211 | break; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 212 | } // Else, fall through... |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 213 | default: |
| 214 | currentValue().setOffsetStart(token.start_ - begin_); |
| 215 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 216 | return addError("Syntax error: value, object or array expected.", token); |
| 217 | } |
| 218 | |
| 219 | if (collectComments_) { |
| 220 | lastValueEnd_ = current_; |
| 221 | lastValue_ = ¤tValue(); |
| 222 | } |
| 223 | |
| 224 | return successful; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 225 | } |
| 226 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 227 | void Reader::skipCommentTokens(Token& token) { |
| 228 | if (features_.allowComments_) { |
| 229 | do { |
| 230 | readToken(token); |
| 231 | } while (token.type_ == tokenComment); |
| 232 | } else { |
| 233 | readToken(token); |
| 234 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 235 | } |
| 236 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 237 | bool Reader::readToken(Token& token) { |
| 238 | skipSpaces(); |
| 239 | token.start_ = current_; |
| 240 | Char c = getNextChar(); |
| 241 | bool ok = true; |
| 242 | switch (c) { |
| 243 | case '{': |
| 244 | token.type_ = tokenObjectBegin; |
| 245 | break; |
| 246 | case '}': |
| 247 | token.type_ = tokenObjectEnd; |
| 248 | break; |
| 249 | case '[': |
| 250 | token.type_ = tokenArrayBegin; |
| 251 | break; |
| 252 | case ']': |
| 253 | token.type_ = tokenArrayEnd; |
| 254 | break; |
| 255 | case '"': |
| 256 | token.type_ = tokenString; |
| 257 | ok = readString(); |
| 258 | break; |
| 259 | case '/': |
| 260 | token.type_ = tokenComment; |
| 261 | ok = readComment(); |
| 262 | break; |
| 263 | case '0': |
| 264 | case '1': |
| 265 | case '2': |
| 266 | case '3': |
| 267 | case '4': |
| 268 | case '5': |
| 269 | case '6': |
| 270 | case '7': |
| 271 | case '8': |
| 272 | case '9': |
| 273 | case '-': |
| 274 | token.type_ = tokenNumber; |
| 275 | readNumber(); |
| 276 | break; |
| 277 | case 't': |
| 278 | token.type_ = tokenTrue; |
| 279 | ok = match("rue", 3); |
| 280 | break; |
| 281 | case 'f': |
| 282 | token.type_ = tokenFalse; |
| 283 | ok = match("alse", 4); |
| 284 | break; |
| 285 | case 'n': |
| 286 | token.type_ = tokenNull; |
| 287 | ok = match("ull", 3); |
| 288 | break; |
| 289 | case ',': |
| 290 | token.type_ = tokenArraySeparator; |
| 291 | break; |
| 292 | case ':': |
| 293 | token.type_ = tokenMemberSeparator; |
| 294 | break; |
| 295 | case 0: |
| 296 | token.type_ = tokenEndOfStream; |
| 297 | break; |
| 298 | default: |
| 299 | ok = false; |
| 300 | break; |
| 301 | } |
| 302 | if (!ok) |
| 303 | token.type_ = tokenError; |
| 304 | token.end_ = current_; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 305 | return ok; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 306 | } |
| 307 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 308 | void Reader::skipSpaces() { |
| 309 | while (current_ != end_) { |
| 310 | Char c = *current_; |
| 311 | if (c == ' ' || c == '\t' || c == '\r' || c == '\n') |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 312 | ++current_; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 313 | else |
| 314 | break; |
| 315 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 316 | } |
| 317 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 318 | bool Reader::match(const Char* pattern, int patternLength) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 319 | if (end_ - current_ < patternLength) |
| 320 | return false; |
| 321 | int index = patternLength; |
| 322 | while (index--) |
| 323 | if (current_[index] != pattern[index]) |
| 324 | return false; |
| 325 | current_ += patternLength; |
| 326 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 327 | } |
| 328 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 329 | bool Reader::readComment() { |
| 330 | Location commentBegin = current_ - 1; |
| 331 | Char c = getNextChar(); |
| 332 | bool successful = false; |
| 333 | if (c == '*') |
| 334 | successful = readCStyleComment(); |
| 335 | else if (c == '/') |
| 336 | successful = readCppStyleComment(); |
| 337 | if (!successful) |
| 338 | return false; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 339 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 340 | if (collectComments_) { |
| 341 | CommentPlacement placement = commentBefore; |
| 342 | if (lastValueEnd_ && !containsNewLine(lastValueEnd_, commentBegin)) { |
| 343 | if (c != '*' || !containsNewLine(commentBegin, current_)) |
| 344 | placement = commentAfterOnSameLine; |
| 345 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 346 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 347 | addComment(commentBegin, current_, placement); |
| 348 | } |
| 349 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 350 | } |
| 351 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 352 | String Reader::normalizeEOL(Reader::Location begin, Reader::Location end) { |
| 353 | String normalized; |
| 354 | normalized.reserve(static_cast<size_t>(end - begin)); |
| 355 | Reader::Location current = begin; |
| 356 | while (current != end) { |
| 357 | char c = *current++; |
| 358 | if (c == '\r') { |
| 359 | if (current != end && *current == '\n') |
| 360 | // convert dos EOL |
| 361 | ++current; |
| 362 | // convert Mac EOL |
| 363 | normalized += '\n'; |
| 364 | } else { |
| 365 | normalized += c; |
| 366 | } |
| 367 | } |
| 368 | return normalized; |
| 369 | } |
| 370 | |
| 371 | void Reader::addComment(Location begin, Location end, |
| 372 | CommentPlacement placement) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 373 | assert(collectComments_); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 374 | const String& normalized = normalizeEOL(begin, end); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 375 | if (placement == commentAfterOnSameLine) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 376 | assert(lastValue_ != nullptr); |
| 377 | lastValue_->setComment(normalized, placement); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 378 | } else { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 379 | commentsBefore_ += normalized; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 380 | } |
| 381 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 382 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 383 | bool Reader::readCStyleComment() { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 384 | while ((current_ + 1) < end_) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 385 | Char c = getNextChar(); |
| 386 | if (c == '*' && *current_ == '/') |
| 387 | break; |
| 388 | } |
| 389 | return getNextChar() == '/'; |
| 390 | } |
| 391 | |
| 392 | bool Reader::readCppStyleComment() { |
| 393 | while (current_ != end_) { |
| 394 | Char c = getNextChar(); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 395 | if (c == '\n') |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 396 | break; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 397 | if (c == '\r') { |
| 398 | // Consume DOS EOL. It will be normalized in addComment. |
| 399 | if (current_ != end_ && *current_ == '\n') |
| 400 | getNextChar(); |
| 401 | // Break on Moc OS 9 EOL. |
| 402 | break; |
| 403 | } |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 404 | } |
| 405 | return true; |
| 406 | } |
| 407 | |
| 408 | void Reader::readNumber() { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 409 | Location p = current_; |
| 410 | char c = '0'; // stopgap for already consumed character |
| 411 | // integral part |
| 412 | while (c >= '0' && c <= '9') |
| 413 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 414 | // fractional part |
| 415 | if (c == '.') { |
| 416 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 417 | while (c >= '0' && c <= '9') |
| 418 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 419 | } |
| 420 | // exponential part |
| 421 | if (c == 'e' || c == 'E') { |
| 422 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 423 | if (c == '+' || c == '-') |
| 424 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 425 | while (c >= '0' && c <= '9') |
| 426 | c = (current_ = p) < end_ ? *p++ : '\0'; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 427 | } |
| 428 | } |
| 429 | |
| 430 | bool Reader::readString() { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 431 | Char c = '\0'; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 432 | while (current_ != end_) { |
| 433 | c = getNextChar(); |
| 434 | if (c == '\\') |
| 435 | getNextChar(); |
| 436 | else if (c == '"') |
| 437 | break; |
| 438 | } |
| 439 | return c == '"'; |
| 440 | } |
| 441 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 442 | bool Reader::readObject(Token& token) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 443 | Token tokenName; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 444 | String name; |
| 445 | Value init(objectValue); |
| 446 | currentValue().swapPayload(init); |
| 447 | currentValue().setOffsetStart(token.start_ - begin_); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 448 | while (readToken(tokenName)) { |
| 449 | bool initialTokenOk = true; |
| 450 | while (tokenName.type_ == tokenComment && initialTokenOk) |
| 451 | initialTokenOk = readToken(tokenName); |
| 452 | if (!initialTokenOk) |
| 453 | break; |
| 454 | if (tokenName.type_ == tokenObjectEnd && name.empty()) // empty object |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 455 | return true; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 456 | name.clear(); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 457 | if (tokenName.type_ == tokenString) { |
| 458 | if (!decodeString(tokenName, name)) |
| 459 | return recoverFromError(tokenObjectEnd); |
| 460 | } else if (tokenName.type_ == tokenNumber && features_.allowNumericKeys_) { |
| 461 | Value numberName; |
| 462 | if (!decodeNumber(tokenName, numberName)) |
| 463 | return recoverFromError(tokenObjectEnd); |
| 464 | name = numberName.asString(); |
| 465 | } else { |
| 466 | break; |
| 467 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 468 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 469 | Token colon; |
| 470 | if (!readToken(colon) || colon.type_ != tokenMemberSeparator) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 471 | return addErrorAndRecover("Missing ':' after object member name", colon, |
| 472 | tokenObjectEnd); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 473 | } |
| 474 | Value& value = currentValue()[name]; |
| 475 | nodes_.push(&value); |
| 476 | bool ok = readValue(); |
| 477 | nodes_.pop(); |
| 478 | if (!ok) // error already set |
| 479 | return recoverFromError(tokenObjectEnd); |
| 480 | |
| 481 | Token comma; |
| 482 | if (!readToken(comma) || |
| 483 | (comma.type_ != tokenObjectEnd && comma.type_ != tokenArraySeparator && |
| 484 | comma.type_ != tokenComment)) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 485 | return addErrorAndRecover("Missing ',' or '}' in object declaration", |
| 486 | comma, tokenObjectEnd); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 487 | } |
| 488 | bool finalizeTokenOk = true; |
| 489 | while (comma.type_ == tokenComment && finalizeTokenOk) |
| 490 | finalizeTokenOk = readToken(comma); |
| 491 | if (comma.type_ == tokenObjectEnd) |
| 492 | return true; |
| 493 | } |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 494 | return addErrorAndRecover("Missing '}' or object member name", tokenName, |
| 495 | tokenObjectEnd); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 496 | } |
| 497 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 498 | bool Reader::readArray(Token& token) { |
| 499 | Value init(arrayValue); |
| 500 | currentValue().swapPayload(init); |
| 501 | currentValue().setOffsetStart(token.start_ - begin_); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 502 | skipSpaces(); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 503 | if (current_ != end_ && *current_ == ']') // empty array |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 504 | { |
| 505 | Token endArray; |
| 506 | readToken(endArray); |
| 507 | return true; |
| 508 | } |
| 509 | int index = 0; |
| 510 | for (;;) { |
| 511 | Value& value = currentValue()[index++]; |
| 512 | nodes_.push(&value); |
| 513 | bool ok = readValue(); |
| 514 | nodes_.pop(); |
| 515 | if (!ok) // error already set |
| 516 | return recoverFromError(tokenArrayEnd); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 517 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 518 | Token currentToken; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 519 | // Accept Comment after last item in the array. |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 520 | ok = readToken(currentToken); |
| 521 | while (currentToken.type_ == tokenComment && ok) { |
| 522 | ok = readToken(currentToken); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 523 | } |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 524 | bool badTokenType = (currentToken.type_ != tokenArraySeparator && |
| 525 | currentToken.type_ != tokenArrayEnd); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 526 | if (!ok || badTokenType) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 527 | return addErrorAndRecover("Missing ',' or ']' in array declaration", |
| 528 | currentToken, tokenArrayEnd); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 529 | } |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 530 | if (currentToken.type_ == tokenArrayEnd) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 531 | break; |
| 532 | } |
| 533 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 534 | } |
| 535 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 536 | bool Reader::decodeNumber(Token& token) { |
| 537 | Value decoded; |
| 538 | if (!decodeNumber(token, decoded)) |
| 539 | return false; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 540 | currentValue().swapPayload(decoded); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 541 | currentValue().setOffsetStart(token.start_ - begin_); |
| 542 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 543 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 544 | } |
| 545 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 546 | bool Reader::decodeNumber(Token& token, Value& decoded) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 547 | // Attempts to parse the number as an integer. If the number is |
| 548 | // larger than the maximum supported value of an integer then |
| 549 | // we decode the number as a double. |
| 550 | Location current = token.start_; |
| 551 | bool isNegative = *current == '-'; |
| 552 | if (isNegative) |
| 553 | ++current; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 554 | // TODO: Help the compiler do the div and mod at compile time or get rid of |
| 555 | // them. |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 556 | Value::LargestUInt maxIntegerValue = |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 557 | isNegative ? Value::LargestUInt(Value::maxLargestInt) + 1 |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 558 | : Value::maxLargestUInt; |
| 559 | Value::LargestUInt threshold = maxIntegerValue / 10; |
| 560 | Value::LargestUInt value = 0; |
| 561 | while (current < token.end_) { |
| 562 | Char c = *current++; |
| 563 | if (c < '0' || c > '9') |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 564 | return decodeDouble(token, decoded); |
| 565 | auto digit(static_cast<Value::UInt>(c - '0')); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 566 | if (value >= threshold) { |
| 567 | // We've hit or exceeded the max value divided by 10 (rounded down). If |
| 568 | // a) we've only just touched the limit, b) this is the last digit, and |
| 569 | // c) it's small enough to fit in that rounding delta, we're okay. |
| 570 | // Otherwise treat this number as a double to avoid overflow. |
| 571 | if (value > threshold || current != token.end_ || |
| 572 | digit > maxIntegerValue % 10) { |
| 573 | return decodeDouble(token, decoded); |
| 574 | } |
| 575 | } |
| 576 | value = value * 10 + digit; |
| 577 | } |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 578 | if (isNegative && value == maxIntegerValue) |
| 579 | decoded = Value::minLargestInt; |
| 580 | else if (isNegative) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 581 | decoded = -Value::LargestInt(value); |
| 582 | else if (value <= Value::LargestUInt(Value::maxInt)) |
| 583 | decoded = Value::LargestInt(value); |
| 584 | else |
| 585 | decoded = value; |
| 586 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 587 | } |
| 588 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 589 | bool Reader::decodeDouble(Token& token) { |
| 590 | Value decoded; |
| 591 | if (!decodeDouble(token, decoded)) |
| 592 | return false; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 593 | currentValue().swapPayload(decoded); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 594 | currentValue().setOffsetStart(token.start_ - begin_); |
| 595 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 596 | return true; |
| 597 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 598 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 599 | bool Reader::decodeDouble(Token& token, Value& decoded) { |
| 600 | double value = 0; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 601 | String buffer(token.start_, token.end_); |
| 602 | IStringStream is(buffer); |
| 603 | if (!(is >> value)) |
| 604 | return addError( |
| 605 | "'" + String(token.start_, token.end_) + "' is not a number.", token); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 606 | decoded = value; |
| 607 | return true; |
| 608 | } |
| 609 | |
| 610 | bool Reader::decodeString(Token& token) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 611 | String decoded_string; |
| 612 | if (!decodeString(token, decoded_string)) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 613 | return false; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 614 | Value decoded(decoded_string); |
| 615 | currentValue().swapPayload(decoded); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 616 | currentValue().setOffsetStart(token.start_ - begin_); |
| 617 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 618 | return true; |
| 619 | } |
| 620 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 621 | bool Reader::decodeString(Token& token, String& decoded) { |
| 622 | decoded.reserve(static_cast<size_t>(token.end_ - token.start_ - 2)); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 623 | Location current = token.start_ + 1; // skip '"' |
| 624 | Location end = token.end_ - 1; // do not include '"' |
| 625 | while (current != end) { |
| 626 | Char c = *current++; |
| 627 | if (c == '"') |
| 628 | break; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 629 | if (c == '\\') { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 630 | if (current == end) |
| 631 | return addError("Empty escape sequence in string", token, current); |
| 632 | Char escape = *current++; |
| 633 | switch (escape) { |
| 634 | case '"': |
| 635 | decoded += '"'; |
| 636 | break; |
| 637 | case '/': |
| 638 | decoded += '/'; |
| 639 | break; |
| 640 | case '\\': |
| 641 | decoded += '\\'; |
| 642 | break; |
| 643 | case 'b': |
| 644 | decoded += '\b'; |
| 645 | break; |
| 646 | case 'f': |
| 647 | decoded += '\f'; |
| 648 | break; |
| 649 | case 'n': |
| 650 | decoded += '\n'; |
| 651 | break; |
| 652 | case 'r': |
| 653 | decoded += '\r'; |
| 654 | break; |
| 655 | case 't': |
| 656 | decoded += '\t'; |
| 657 | break; |
| 658 | case 'u': { |
| 659 | unsigned int unicode; |
| 660 | if (!decodeUnicodeCodePoint(token, current, end, unicode)) |
| 661 | return false; |
| 662 | decoded += codePointToUTF8(unicode); |
| 663 | } break; |
| 664 | default: |
| 665 | return addError("Bad escape sequence in string", token, current); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 666 | } |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 667 | } else { |
| 668 | decoded += c; |
| 669 | } |
| 670 | } |
| 671 | return true; |
| 672 | } |
| 673 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 674 | bool Reader::decodeUnicodeCodePoint(Token& token, Location& current, |
| 675 | Location end, unsigned int& unicode) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 676 | |
| 677 | if (!decodeUnicodeEscapeSequence(token, current, end, unicode)) |
| 678 | return false; |
| 679 | if (unicode >= 0xD800 && unicode <= 0xDBFF) { |
| 680 | // surrogate pairs |
| 681 | if (end - current < 6) |
| 682 | return addError( |
| 683 | "additional six characters expected to parse unicode surrogate pair.", |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 684 | token, current); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 685 | if (*(current++) == '\\' && *(current++) == 'u') { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 686 | unsigned int surrogatePair; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 687 | if (decodeUnicodeEscapeSequence(token, current, end, surrogatePair)) { |
| 688 | unicode = 0x10000 + ((unicode & 0x3FF) << 10) + (surrogatePair & 0x3FF); |
| 689 | } else |
| 690 | return false; |
| 691 | } else |
| 692 | return addError("expecting another \\u token to begin the second half of " |
| 693 | "a unicode surrogate pair", |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 694 | token, current); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 695 | } |
| 696 | return true; |
| 697 | } |
| 698 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 699 | bool Reader::decodeUnicodeEscapeSequence(Token& token, Location& current, |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 700 | Location end, |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 701 | unsigned int& ret_unicode) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 702 | if (end - current < 4) |
| 703 | return addError( |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 704 | "Bad unicode escape sequence in string: four digits expected.", token, |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 705 | current); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 706 | int unicode = 0; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 707 | for (int index = 0; index < 4; ++index) { |
| 708 | Char c = *current++; |
| 709 | unicode *= 16; |
| 710 | if (c >= '0' && c <= '9') |
| 711 | unicode += c - '0'; |
| 712 | else if (c >= 'a' && c <= 'f') |
| 713 | unicode += c - 'a' + 10; |
| 714 | else if (c >= 'A' && c <= 'F') |
| 715 | unicode += c - 'A' + 10; |
| 716 | else |
| 717 | return addError( |
| 718 | "Bad unicode escape sequence in string: hexadecimal digit expected.", |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 719 | token, current); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 720 | } |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 721 | ret_unicode = static_cast<unsigned int>(unicode); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 722 | return true; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 723 | } |
| 724 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 725 | bool Reader::addError(const String& message, Token& token, Location extra) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 726 | ErrorInfo info; |
| 727 | info.token_ = token; |
| 728 | info.message_ = message; |
| 729 | info.extra_ = extra; |
| 730 | errors_.push_back(info); |
| 731 | return false; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 732 | } |
| 733 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 734 | bool Reader::recoverFromError(TokenType skipUntilToken) { |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 735 | size_t const errorCount = errors_.size(); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 736 | Token skip; |
| 737 | for (;;) { |
| 738 | if (!readToken(skip)) |
| 739 | errors_.resize(errorCount); // discard errors caused by recovery |
| 740 | if (skip.type_ == skipUntilToken || skip.type_ == tokenEndOfStream) |
| 741 | break; |
| 742 | } |
| 743 | errors_.resize(errorCount); |
| 744 | return false; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 745 | } |
| 746 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 747 | bool Reader::addErrorAndRecover(const String& message, Token& token, |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 748 | TokenType skipUntilToken) { |
| 749 | addError(message, token); |
| 750 | return recoverFromError(skipUntilToken); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 751 | } |
| 752 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 753 | Value& Reader::currentValue() { return *(nodes_.top()); } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 754 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 755 | Reader::Char Reader::getNextChar() { |
| 756 | if (current_ == end_) |
| 757 | return 0; |
| 758 | return *current_++; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 759 | } |
| 760 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 761 | void Reader::getLocationLineAndColumn(Location location, int& line, |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 762 | int& column) const { |
| 763 | Location current = begin_; |
| 764 | Location lastLineStart = current; |
| 765 | line = 0; |
| 766 | while (current < location && current != end_) { |
| 767 | Char c = *current++; |
| 768 | if (c == '\r') { |
| 769 | if (*current == '\n') |
| 770 | ++current; |
| 771 | lastLineStart = current; |
| 772 | ++line; |
| 773 | } else if (c == '\n') { |
| 774 | lastLineStart = current; |
| 775 | ++line; |
| 776 | } |
| 777 | } |
| 778 | // column & line start at 1 |
| 779 | column = int(location - lastLineStart) + 1; |
| 780 | ++line; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 781 | } |
| 782 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 783 | String Reader::getLocationLineAndColumn(Location location) const { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 784 | int line, column; |
| 785 | getLocationLineAndColumn(location, line, column); |
| 786 | char buffer[18 + 16 + 16 + 1]; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 787 | jsoncpp_snprintf(buffer, sizeof(buffer), "Line %d, Column %d", line, column); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 788 | return buffer; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 789 | } |
| 790 | |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 791 | // Deprecated. Preserved for backward compatibility |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 792 | String Reader::getFormatedErrorMessages() const { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 793 | return getFormattedErrorMessages(); |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 794 | } |
| 795 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 796 | String Reader::getFormattedErrorMessages() const { |
| 797 | String formattedMessage; |
| 798 | for (const auto& error : errors_) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 799 | formattedMessage += |
| 800 | "* " + getLocationLineAndColumn(error.token_.start_) + "\n"; |
| 801 | formattedMessage += " " + error.message_ + "\n"; |
| 802 | if (error.extra_) |
| 803 | formattedMessage += |
| 804 | "See " + getLocationLineAndColumn(error.extra_) + " for detail.\n"; |
| 805 | } |
| 806 | return formattedMessage; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 807 | } |
| 808 | |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 809 | std::vector<Reader::StructuredError> Reader::getStructuredErrors() const { |
| 810 | std::vector<Reader::StructuredError> allErrors; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 811 | for (const auto& error : errors_) { |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 812 | Reader::StructuredError structured; |
| 813 | structured.offset_start = error.token_.start_ - begin_; |
| 814 | structured.offset_limit = error.token_.end_ - begin_; |
| 815 | structured.message = error.message_; |
| 816 | allErrors.push_back(structured); |
| 817 | } |
| 818 | return allErrors; |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 819 | } |
| 820 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 821 | bool Reader::pushError(const Value& value, const String& message) { |
| 822 | ptrdiff_t const length = end_ - begin_; |
| 823 | if (value.getOffsetStart() > length || value.getOffsetLimit() > length) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 824 | return false; |
| 825 | Token token; |
| 826 | token.type_ = tokenError; |
| 827 | token.start_ = begin_ + value.getOffsetStart(); |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 828 | token.end_ = begin_ + value.getOffsetLimit(); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 829 | ErrorInfo info; |
| 830 | info.token_ = token; |
| 831 | info.message_ = message; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 832 | info.extra_ = nullptr; |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 833 | errors_.push_back(info); |
| 834 | return true; |
| 835 | } |
| 836 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 837 | bool Reader::pushError(const Value& value, const String& message, |
| 838 | const Value& extra) { |
| 839 | ptrdiff_t const length = end_ - begin_; |
| 840 | if (value.getOffsetStart() > length || value.getOffsetLimit() > length || |
| 841 | extra.getOffsetLimit() > length) |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 842 | return false; |
| 843 | Token token; |
| 844 | token.type_ = tokenError; |
| 845 | token.start_ = begin_ + value.getOffsetStart(); |
| 846 | token.end_ = begin_ + value.getOffsetLimit(); |
| 847 | ErrorInfo info; |
| 848 | info.token_ = token; |
| 849 | info.message_ = message; |
| 850 | info.extra_ = begin_ + extra.getOffsetStart(); |
| 851 | errors_.push_back(info); |
| 852 | return true; |
| 853 | } |
| 854 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 855 | bool Reader::good() const { return errors_.empty(); } |
| 856 | |
| 857 | // Originally copied from the Features class (now deprecated), used internally |
| 858 | // for features implementation. |
| 859 | class OurFeatures { |
| 860 | public: |
| 861 | static OurFeatures all(); |
| 862 | bool allowComments_; |
| 863 | bool allowTrailingCommas_; |
| 864 | bool strictRoot_; |
| 865 | bool allowDroppedNullPlaceholders_; |
| 866 | bool allowNumericKeys_; |
| 867 | bool allowSingleQuotes_; |
| 868 | bool failIfExtra_; |
| 869 | bool rejectDupKeys_; |
| 870 | bool allowSpecialFloats_; |
| 871 | bool skipBom_; |
| 872 | size_t stackLimit_; |
| 873 | }; // OurFeatures |
| 874 | |
| 875 | OurFeatures OurFeatures::all() { return {}; } |
| 876 | |
| 877 | // Implementation of class Reader |
| 878 | // //////////////////////////////// |
| 879 | |
| 880 | // Originally copied from the Reader class (now deprecated), used internally |
| 881 | // for implementing JSON reading. |
| 882 | class OurReader { |
| 883 | public: |
| 884 | using Char = char; |
| 885 | using Location = const Char*; |
| 886 | struct StructuredError { |
| 887 | ptrdiff_t offset_start; |
| 888 | ptrdiff_t offset_limit; |
| 889 | String message; |
| 890 | }; |
| 891 | |
| 892 | explicit OurReader(OurFeatures const& features); |
| 893 | bool parse(const char* beginDoc, const char* endDoc, Value& root, |
| 894 | bool collectComments = true); |
| 895 | String getFormattedErrorMessages() const; |
| 896 | std::vector<StructuredError> getStructuredErrors() const; |
| 897 | |
| 898 | private: |
| 899 | OurReader(OurReader const&); // no impl |
| 900 | void operator=(OurReader const&); // no impl |
| 901 | |
| 902 | enum TokenType { |
| 903 | tokenEndOfStream = 0, |
| 904 | tokenObjectBegin, |
| 905 | tokenObjectEnd, |
| 906 | tokenArrayBegin, |
| 907 | tokenArrayEnd, |
| 908 | tokenString, |
| 909 | tokenNumber, |
| 910 | tokenTrue, |
| 911 | tokenFalse, |
| 912 | tokenNull, |
| 913 | tokenNaN, |
| 914 | tokenPosInf, |
| 915 | tokenNegInf, |
| 916 | tokenArraySeparator, |
| 917 | tokenMemberSeparator, |
| 918 | tokenComment, |
| 919 | tokenError |
| 920 | }; |
| 921 | |
| 922 | class Token { |
| 923 | public: |
| 924 | TokenType type_; |
| 925 | Location start_; |
| 926 | Location end_; |
| 927 | }; |
| 928 | |
| 929 | class ErrorInfo { |
| 930 | public: |
| 931 | Token token_; |
| 932 | String message_; |
| 933 | Location extra_; |
| 934 | }; |
| 935 | |
| 936 | using Errors = std::deque<ErrorInfo>; |
| 937 | |
| 938 | bool readToken(Token& token); |
| 939 | void skipSpaces(); |
| 940 | void skipBom(bool skipBom); |
| 941 | bool match(const Char* pattern, int patternLength); |
| 942 | bool readComment(); |
| 943 | bool readCStyleComment(bool* containsNewLineResult); |
| 944 | bool readCppStyleComment(); |
| 945 | bool readString(); |
| 946 | bool readStringSingleQuote(); |
| 947 | bool readNumber(bool checkInf); |
| 948 | bool readValue(); |
| 949 | bool readObject(Token& token); |
| 950 | bool readArray(Token& token); |
| 951 | bool decodeNumber(Token& token); |
| 952 | bool decodeNumber(Token& token, Value& decoded); |
| 953 | bool decodeString(Token& token); |
| 954 | bool decodeString(Token& token, String& decoded); |
| 955 | bool decodeDouble(Token& token); |
| 956 | bool decodeDouble(Token& token, Value& decoded); |
| 957 | bool decodeUnicodeCodePoint(Token& token, Location& current, Location end, |
| 958 | unsigned int& unicode); |
| 959 | bool decodeUnicodeEscapeSequence(Token& token, Location& current, |
| 960 | Location end, unsigned int& unicode); |
| 961 | bool addError(const String& message, Token& token, Location extra = nullptr); |
| 962 | bool recoverFromError(TokenType skipUntilToken); |
| 963 | bool addErrorAndRecover(const String& message, Token& token, |
| 964 | TokenType skipUntilToken); |
| 965 | void skipUntilSpace(); |
| 966 | Value& currentValue(); |
| 967 | Char getNextChar(); |
| 968 | void getLocationLineAndColumn(Location location, int& line, |
| 969 | int& column) const; |
| 970 | String getLocationLineAndColumn(Location location) const; |
| 971 | void addComment(Location begin, Location end, CommentPlacement placement); |
| 972 | void skipCommentTokens(Token& token); |
| 973 | |
| 974 | static String normalizeEOL(Location begin, Location end); |
| 975 | static bool containsNewLine(Location begin, Location end); |
| 976 | |
| 977 | using Nodes = std::stack<Value*>; |
| 978 | |
| 979 | Nodes nodes_{}; |
| 980 | Errors errors_{}; |
| 981 | String document_{}; |
| 982 | Location begin_ = nullptr; |
| 983 | Location end_ = nullptr; |
| 984 | Location current_ = nullptr; |
| 985 | Location lastValueEnd_ = nullptr; |
| 986 | Value* lastValue_ = nullptr; |
| 987 | bool lastValueHasAComment_ = false; |
| 988 | String commentsBefore_{}; |
| 989 | |
| 990 | OurFeatures const features_; |
| 991 | bool collectComments_ = false; |
| 992 | }; // OurReader |
| 993 | |
| 994 | // complete copy of Read impl, for OurReader |
| 995 | |
| 996 | bool OurReader::containsNewLine(OurReader::Location begin, |
| 997 | OurReader::Location end) { |
| 998 | return std::any_of(begin, end, [](char b) { return b == '\n' || b == '\r'; }); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 999 | } |
| 1000 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 1001 | OurReader::OurReader(OurFeatures const& features) : features_(features) {} |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 1002 | |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 1003 | bool OurReader::parse(const char* beginDoc, const char* endDoc, Value& root, |
| 1004 | bool collectComments) { |
| 1005 | if (!features_.allowComments_) { |
| 1006 | collectComments = false; |
| 1007 | } |
| 1008 | |
| 1009 | begin_ = beginDoc; |
| 1010 | end_ = endDoc; |
| 1011 | collectComments_ = collectComments; |
| 1012 | current_ = begin_; |
| 1013 | lastValueEnd_ = nullptr; |
| 1014 | lastValue_ = nullptr; |
| 1015 | commentsBefore_.clear(); |
| 1016 | errors_.clear(); |
| 1017 | while (!nodes_.empty()) |
| 1018 | nodes_.pop(); |
| 1019 | nodes_.push(&root); |
| 1020 | |
| 1021 | // skip byte order mark if it exists at the beginning of the UTF-8 text. |
| 1022 | skipBom(features_.skipBom_); |
| 1023 | bool successful = readValue(); |
| 1024 | nodes_.pop(); |
| 1025 | Token token; |
| 1026 | skipCommentTokens(token); |
| 1027 | if (features_.failIfExtra_ && (token.type_ != tokenEndOfStream)) { |
| 1028 | addError("Extra non-whitespace after JSON value.", token); |
| 1029 | return false; |
| 1030 | } |
| 1031 | if (collectComments_ && !commentsBefore_.empty()) |
| 1032 | root.setComment(commentsBefore_, commentAfter); |
| 1033 | if (features_.strictRoot_) { |
| 1034 | if (!root.isArray() && !root.isObject()) { |
| 1035 | // Set error location to start of doc, ideally should be first token found |
| 1036 | // in doc |
| 1037 | token.type_ = tokenError; |
| 1038 | token.start_ = beginDoc; |
| 1039 | token.end_ = endDoc; |
| 1040 | addError( |
| 1041 | "A valid JSON document must be either an array or an object value.", |
| 1042 | token); |
| 1043 | return false; |
| 1044 | } |
| 1045 | } |
| 1046 | return successful; |
| 1047 | } |
| 1048 | |
| 1049 | bool OurReader::readValue() { |
| 1050 | // To preserve the old behaviour we cast size_t to int. |
| 1051 | if (nodes_.size() > features_.stackLimit_) |
| 1052 | throwRuntimeError("Exceeded stackLimit in readValue()."); |
| 1053 | Token token; |
| 1054 | skipCommentTokens(token); |
| 1055 | bool successful = true; |
| 1056 | |
| 1057 | if (collectComments_ && !commentsBefore_.empty()) { |
| 1058 | currentValue().setComment(commentsBefore_, commentBefore); |
| 1059 | commentsBefore_.clear(); |
| 1060 | } |
| 1061 | |
| 1062 | switch (token.type_) { |
| 1063 | case tokenObjectBegin: |
| 1064 | successful = readObject(token); |
| 1065 | currentValue().setOffsetLimit(current_ - begin_); |
| 1066 | break; |
| 1067 | case tokenArrayBegin: |
| 1068 | successful = readArray(token); |
| 1069 | currentValue().setOffsetLimit(current_ - begin_); |
| 1070 | break; |
| 1071 | case tokenNumber: |
| 1072 | successful = decodeNumber(token); |
| 1073 | break; |
| 1074 | case tokenString: |
| 1075 | successful = decodeString(token); |
| 1076 | break; |
| 1077 | case tokenTrue: { |
| 1078 | Value v(true); |
| 1079 | currentValue().swapPayload(v); |
| 1080 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1081 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1082 | } break; |
| 1083 | case tokenFalse: { |
| 1084 | Value v(false); |
| 1085 | currentValue().swapPayload(v); |
| 1086 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1087 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1088 | } break; |
| 1089 | case tokenNull: { |
| 1090 | Value v; |
| 1091 | currentValue().swapPayload(v); |
| 1092 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1093 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1094 | } break; |
| 1095 | case tokenNaN: { |
| 1096 | Value v(std::numeric_limits<double>::quiet_NaN()); |
| 1097 | currentValue().swapPayload(v); |
| 1098 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1099 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1100 | } break; |
| 1101 | case tokenPosInf: { |
| 1102 | Value v(std::numeric_limits<double>::infinity()); |
| 1103 | currentValue().swapPayload(v); |
| 1104 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1105 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1106 | } break; |
| 1107 | case tokenNegInf: { |
| 1108 | Value v(-std::numeric_limits<double>::infinity()); |
| 1109 | currentValue().swapPayload(v); |
| 1110 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1111 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1112 | } break; |
| 1113 | case tokenArraySeparator: |
| 1114 | case tokenObjectEnd: |
| 1115 | case tokenArrayEnd: |
| 1116 | if (features_.allowDroppedNullPlaceholders_) { |
| 1117 | // "Un-read" the current token and mark the current value as a null |
| 1118 | // token. |
| 1119 | current_--; |
| 1120 | Value v; |
| 1121 | currentValue().swapPayload(v); |
| 1122 | currentValue().setOffsetStart(current_ - begin_ - 1); |
| 1123 | currentValue().setOffsetLimit(current_ - begin_); |
| 1124 | break; |
| 1125 | } // else, fall through ... |
| 1126 | default: |
| 1127 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1128 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1129 | return addError("Syntax error: value, object or array expected.", token); |
| 1130 | } |
| 1131 | |
| 1132 | if (collectComments_) { |
| 1133 | lastValueEnd_ = current_; |
| 1134 | lastValueHasAComment_ = false; |
| 1135 | lastValue_ = ¤tValue(); |
| 1136 | } |
| 1137 | |
| 1138 | return successful; |
| 1139 | } |
| 1140 | |
| 1141 | void OurReader::skipCommentTokens(Token& token) { |
| 1142 | if (features_.allowComments_) { |
| 1143 | do { |
| 1144 | readToken(token); |
| 1145 | } while (token.type_ == tokenComment); |
| 1146 | } else { |
| 1147 | readToken(token); |
| 1148 | } |
| 1149 | } |
| 1150 | |
| 1151 | bool OurReader::readToken(Token& token) { |
| 1152 | skipSpaces(); |
| 1153 | token.start_ = current_; |
| 1154 | Char c = getNextChar(); |
| 1155 | bool ok = true; |
| 1156 | switch (c) { |
| 1157 | case '{': |
| 1158 | token.type_ = tokenObjectBegin; |
| 1159 | break; |
| 1160 | case '}': |
| 1161 | token.type_ = tokenObjectEnd; |
| 1162 | break; |
| 1163 | case '[': |
| 1164 | token.type_ = tokenArrayBegin; |
| 1165 | break; |
| 1166 | case ']': |
| 1167 | token.type_ = tokenArrayEnd; |
| 1168 | break; |
| 1169 | case '"': |
| 1170 | token.type_ = tokenString; |
| 1171 | ok = readString(); |
| 1172 | break; |
| 1173 | case '\'': |
| 1174 | if (features_.allowSingleQuotes_) { |
| 1175 | token.type_ = tokenString; |
| 1176 | ok = readStringSingleQuote(); |
| 1177 | } else { |
| 1178 | // If we don't allow single quotes, this is a failure case. |
| 1179 | ok = false; |
| 1180 | } |
| 1181 | break; |
| 1182 | case '/': |
| 1183 | token.type_ = tokenComment; |
| 1184 | ok = readComment(); |
| 1185 | break; |
| 1186 | case '0': |
| 1187 | case '1': |
| 1188 | case '2': |
| 1189 | case '3': |
| 1190 | case '4': |
| 1191 | case '5': |
| 1192 | case '6': |
| 1193 | case '7': |
| 1194 | case '8': |
| 1195 | case '9': |
| 1196 | token.type_ = tokenNumber; |
| 1197 | readNumber(false); |
| 1198 | break; |
| 1199 | case '-': |
| 1200 | if (readNumber(true)) { |
| 1201 | token.type_ = tokenNumber; |
| 1202 | } else { |
| 1203 | token.type_ = tokenNegInf; |
| 1204 | ok = features_.allowSpecialFloats_ && match("nfinity", 7); |
| 1205 | } |
| 1206 | break; |
| 1207 | case '+': |
| 1208 | if (readNumber(true)) { |
| 1209 | token.type_ = tokenNumber; |
| 1210 | } else { |
| 1211 | token.type_ = tokenPosInf; |
| 1212 | ok = features_.allowSpecialFloats_ && match("nfinity", 7); |
| 1213 | } |
| 1214 | break; |
| 1215 | case 't': |
| 1216 | token.type_ = tokenTrue; |
| 1217 | ok = match("rue", 3); |
| 1218 | break; |
| 1219 | case 'f': |
| 1220 | token.type_ = tokenFalse; |
| 1221 | ok = match("alse", 4); |
| 1222 | break; |
| 1223 | case 'n': |
| 1224 | token.type_ = tokenNull; |
| 1225 | ok = match("ull", 3); |
| 1226 | break; |
| 1227 | case 'N': |
| 1228 | if (features_.allowSpecialFloats_) { |
| 1229 | token.type_ = tokenNaN; |
| 1230 | ok = match("aN", 2); |
| 1231 | } else { |
| 1232 | ok = false; |
| 1233 | } |
| 1234 | break; |
| 1235 | case 'I': |
| 1236 | if (features_.allowSpecialFloats_) { |
| 1237 | token.type_ = tokenPosInf; |
| 1238 | ok = match("nfinity", 7); |
| 1239 | } else { |
| 1240 | ok = false; |
| 1241 | } |
| 1242 | break; |
| 1243 | case ',': |
| 1244 | token.type_ = tokenArraySeparator; |
| 1245 | break; |
| 1246 | case ':': |
| 1247 | token.type_ = tokenMemberSeparator; |
| 1248 | break; |
| 1249 | case 0: |
| 1250 | token.type_ = tokenEndOfStream; |
| 1251 | break; |
| 1252 | default: |
| 1253 | ok = false; |
| 1254 | break; |
| 1255 | } |
| 1256 | if (!ok) |
| 1257 | token.type_ = tokenError; |
| 1258 | token.end_ = current_; |
| 1259 | return ok; |
| 1260 | } |
| 1261 | |
| 1262 | void OurReader::skipSpaces() { |
| 1263 | while (current_ != end_) { |
| 1264 | Char c = *current_; |
| 1265 | if (c == ' ' || c == '\t' || c == '\r' || c == '\n') |
| 1266 | ++current_; |
| 1267 | else |
| 1268 | break; |
| 1269 | } |
| 1270 | } |
| 1271 | |
| 1272 | void OurReader::skipBom(bool skipBom) { |
| 1273 | // The default behavior is to skip BOM. |
| 1274 | if (skipBom) { |
| 1275 | if ((end_ - begin_) >= 3 && strncmp(begin_, "\xEF\xBB\xBF", 3) == 0) { |
| 1276 | begin_ += 3; |
| 1277 | current_ = begin_; |
| 1278 | } |
| 1279 | } |
| 1280 | } |
| 1281 | |
| 1282 | bool OurReader::match(const Char* pattern, int patternLength) { |
| 1283 | if (end_ - current_ < patternLength) |
| 1284 | return false; |
| 1285 | int index = patternLength; |
| 1286 | while (index--) |
| 1287 | if (current_[index] != pattern[index]) |
| 1288 | return false; |
| 1289 | current_ += patternLength; |
| 1290 | return true; |
| 1291 | } |
| 1292 | |
| 1293 | bool OurReader::readComment() { |
| 1294 | const Location commentBegin = current_ - 1; |
| 1295 | const Char c = getNextChar(); |
| 1296 | bool successful = false; |
| 1297 | bool cStyleWithEmbeddedNewline = false; |
| 1298 | |
| 1299 | const bool isCStyleComment = (c == '*'); |
| 1300 | const bool isCppStyleComment = (c == '/'); |
| 1301 | if (isCStyleComment) { |
| 1302 | successful = readCStyleComment(&cStyleWithEmbeddedNewline); |
| 1303 | } else if (isCppStyleComment) { |
| 1304 | successful = readCppStyleComment(); |
| 1305 | } |
| 1306 | |
| 1307 | if (!successful) |
| 1308 | return false; |
| 1309 | |
| 1310 | if (collectComments_) { |
| 1311 | CommentPlacement placement = commentBefore; |
| 1312 | |
| 1313 | if (!lastValueHasAComment_) { |
| 1314 | if (lastValueEnd_ && !containsNewLine(lastValueEnd_, commentBegin)) { |
| 1315 | if (isCppStyleComment || !cStyleWithEmbeddedNewline) { |
| 1316 | placement = commentAfterOnSameLine; |
| 1317 | lastValueHasAComment_ = true; |
| 1318 | } |
| 1319 | } |
| 1320 | } |
| 1321 | |
| 1322 | addComment(commentBegin, current_, placement); |
| 1323 | } |
| 1324 | return true; |
| 1325 | } |
| 1326 | |
| 1327 | String OurReader::normalizeEOL(OurReader::Location begin, |
| 1328 | OurReader::Location end) { |
| 1329 | String normalized; |
| 1330 | normalized.reserve(static_cast<size_t>(end - begin)); |
| 1331 | OurReader::Location current = begin; |
| 1332 | while (current != end) { |
| 1333 | char c = *current++; |
| 1334 | if (c == '\r') { |
| 1335 | if (current != end && *current == '\n') |
| 1336 | // convert dos EOL |
| 1337 | ++current; |
| 1338 | // convert Mac EOL |
| 1339 | normalized += '\n'; |
| 1340 | } else { |
| 1341 | normalized += c; |
| 1342 | } |
| 1343 | } |
| 1344 | return normalized; |
| 1345 | } |
| 1346 | |
| 1347 | void OurReader::addComment(Location begin, Location end, |
| 1348 | CommentPlacement placement) { |
| 1349 | assert(collectComments_); |
| 1350 | const String& normalized = normalizeEOL(begin, end); |
| 1351 | if (placement == commentAfterOnSameLine) { |
| 1352 | assert(lastValue_ != nullptr); |
| 1353 | lastValue_->setComment(normalized, placement); |
| 1354 | } else { |
| 1355 | commentsBefore_ += normalized; |
| 1356 | } |
| 1357 | } |
| 1358 | |
| 1359 | bool OurReader::readCStyleComment(bool* containsNewLineResult) { |
| 1360 | *containsNewLineResult = false; |
| 1361 | |
| 1362 | while ((current_ + 1) < end_) { |
| 1363 | Char c = getNextChar(); |
| 1364 | if (c == '*' && *current_ == '/') |
| 1365 | break; |
| 1366 | if (c == '\n') |
| 1367 | *containsNewLineResult = true; |
| 1368 | } |
| 1369 | |
| 1370 | return getNextChar() == '/'; |
| 1371 | } |
| 1372 | |
| 1373 | bool OurReader::readCppStyleComment() { |
| 1374 | while (current_ != end_) { |
| 1375 | Char c = getNextChar(); |
| 1376 | if (c == '\n') |
| 1377 | break; |
| 1378 | if (c == '\r') { |
| 1379 | // Consume DOS EOL. It will be normalized in addComment. |
| 1380 | if (current_ != end_ && *current_ == '\n') |
| 1381 | getNextChar(); |
| 1382 | // Break on Moc OS 9 EOL. |
| 1383 | break; |
| 1384 | } |
| 1385 | } |
| 1386 | return true; |
| 1387 | } |
| 1388 | |
| 1389 | bool OurReader::readNumber(bool checkInf) { |
| 1390 | Location p = current_; |
| 1391 | if (checkInf && p != end_ && *p == 'I') { |
| 1392 | current_ = ++p; |
| 1393 | return false; |
| 1394 | } |
| 1395 | char c = '0'; // stopgap for already consumed character |
| 1396 | // integral part |
| 1397 | while (c >= '0' && c <= '9') |
| 1398 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1399 | // fractional part |
| 1400 | if (c == '.') { |
| 1401 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1402 | while (c >= '0' && c <= '9') |
| 1403 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1404 | } |
| 1405 | // exponential part |
| 1406 | if (c == 'e' || c == 'E') { |
| 1407 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1408 | if (c == '+' || c == '-') |
| 1409 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1410 | while (c >= '0' && c <= '9') |
| 1411 | c = (current_ = p) < end_ ? *p++ : '\0'; |
| 1412 | } |
| 1413 | return true; |
| 1414 | } |
| 1415 | bool OurReader::readString() { |
| 1416 | Char c = 0; |
| 1417 | while (current_ != end_) { |
| 1418 | c = getNextChar(); |
| 1419 | if (c == '\\') |
| 1420 | getNextChar(); |
| 1421 | else if (c == '"') |
| 1422 | break; |
| 1423 | } |
| 1424 | return c == '"'; |
| 1425 | } |
| 1426 | |
| 1427 | bool OurReader::readStringSingleQuote() { |
| 1428 | Char c = 0; |
| 1429 | while (current_ != end_) { |
| 1430 | c = getNextChar(); |
| 1431 | if (c == '\\') |
| 1432 | getNextChar(); |
| 1433 | else if (c == '\'') |
| 1434 | break; |
| 1435 | } |
| 1436 | return c == '\''; |
| 1437 | } |
| 1438 | |
| 1439 | bool OurReader::readObject(Token& token) { |
| 1440 | Token tokenName; |
| 1441 | String name; |
| 1442 | Value init(objectValue); |
| 1443 | currentValue().swapPayload(init); |
| 1444 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1445 | while (readToken(tokenName)) { |
| 1446 | bool initialTokenOk = true; |
| 1447 | while (tokenName.type_ == tokenComment && initialTokenOk) |
| 1448 | initialTokenOk = readToken(tokenName); |
| 1449 | if (!initialTokenOk) |
| 1450 | break; |
| 1451 | if (tokenName.type_ == tokenObjectEnd && |
| 1452 | (name.empty() || |
| 1453 | features_.allowTrailingCommas_)) // empty object or trailing comma |
| 1454 | return true; |
| 1455 | name.clear(); |
| 1456 | if (tokenName.type_ == tokenString) { |
| 1457 | if (!decodeString(tokenName, name)) |
| 1458 | return recoverFromError(tokenObjectEnd); |
| 1459 | } else if (tokenName.type_ == tokenNumber && features_.allowNumericKeys_) { |
| 1460 | Value numberName; |
| 1461 | if (!decodeNumber(tokenName, numberName)) |
| 1462 | return recoverFromError(tokenObjectEnd); |
| 1463 | name = numberName.asString(); |
| 1464 | } else { |
| 1465 | break; |
| 1466 | } |
| 1467 | if (name.length() >= (1U << 30)) |
| 1468 | throwRuntimeError("keylength >= 2^30"); |
| 1469 | if (features_.rejectDupKeys_ && currentValue().isMember(name)) { |
| 1470 | String msg = "Duplicate key: '" + name + "'"; |
| 1471 | return addErrorAndRecover(msg, tokenName, tokenObjectEnd); |
| 1472 | } |
| 1473 | |
| 1474 | Token colon; |
| 1475 | if (!readToken(colon) || colon.type_ != tokenMemberSeparator) { |
| 1476 | return addErrorAndRecover("Missing ':' after object member name", colon, |
| 1477 | tokenObjectEnd); |
| 1478 | } |
| 1479 | Value& value = currentValue()[name]; |
| 1480 | nodes_.push(&value); |
| 1481 | bool ok = readValue(); |
| 1482 | nodes_.pop(); |
| 1483 | if (!ok) // error already set |
| 1484 | return recoverFromError(tokenObjectEnd); |
| 1485 | |
| 1486 | Token comma; |
| 1487 | if (!readToken(comma) || |
| 1488 | (comma.type_ != tokenObjectEnd && comma.type_ != tokenArraySeparator && |
| 1489 | comma.type_ != tokenComment)) { |
| 1490 | return addErrorAndRecover("Missing ',' or '}' in object declaration", |
| 1491 | comma, tokenObjectEnd); |
| 1492 | } |
| 1493 | bool finalizeTokenOk = true; |
| 1494 | while (comma.type_ == tokenComment && finalizeTokenOk) |
| 1495 | finalizeTokenOk = readToken(comma); |
| 1496 | if (comma.type_ == tokenObjectEnd) |
| 1497 | return true; |
| 1498 | } |
| 1499 | return addErrorAndRecover("Missing '}' or object member name", tokenName, |
| 1500 | tokenObjectEnd); |
| 1501 | } |
| 1502 | |
| 1503 | bool OurReader::readArray(Token& token) { |
| 1504 | Value init(arrayValue); |
| 1505 | currentValue().swapPayload(init); |
| 1506 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1507 | int index = 0; |
| 1508 | for (;;) { |
| 1509 | skipSpaces(); |
| 1510 | if (current_ != end_ && *current_ == ']' && |
| 1511 | (index == 0 || |
| 1512 | (features_.allowTrailingCommas_ && |
| 1513 | !features_.allowDroppedNullPlaceholders_))) // empty array or trailing |
| 1514 | // comma |
| 1515 | { |
| 1516 | Token endArray; |
| 1517 | readToken(endArray); |
| 1518 | return true; |
| 1519 | } |
| 1520 | Value& value = currentValue()[index++]; |
| 1521 | nodes_.push(&value); |
| 1522 | bool ok = readValue(); |
| 1523 | nodes_.pop(); |
| 1524 | if (!ok) // error already set |
| 1525 | return recoverFromError(tokenArrayEnd); |
| 1526 | |
| 1527 | Token currentToken; |
| 1528 | // Accept Comment after last item in the array. |
| 1529 | ok = readToken(currentToken); |
| 1530 | while (currentToken.type_ == tokenComment && ok) { |
| 1531 | ok = readToken(currentToken); |
| 1532 | } |
| 1533 | bool badTokenType = (currentToken.type_ != tokenArraySeparator && |
| 1534 | currentToken.type_ != tokenArrayEnd); |
| 1535 | if (!ok || badTokenType) { |
| 1536 | return addErrorAndRecover("Missing ',' or ']' in array declaration", |
| 1537 | currentToken, tokenArrayEnd); |
| 1538 | } |
| 1539 | if (currentToken.type_ == tokenArrayEnd) |
| 1540 | break; |
| 1541 | } |
| 1542 | return true; |
| 1543 | } |
| 1544 | |
| 1545 | bool OurReader::decodeNumber(Token& token) { |
| 1546 | Value decoded; |
| 1547 | if (!decodeNumber(token, decoded)) |
| 1548 | return false; |
| 1549 | currentValue().swapPayload(decoded); |
| 1550 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1551 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1552 | return true; |
| 1553 | } |
| 1554 | |
| 1555 | bool OurReader::decodeNumber(Token& token, Value& decoded) { |
| 1556 | // Attempts to parse the number as an integer. If the number is |
| 1557 | // larger than the maximum supported value of an integer then |
| 1558 | // we decode the number as a double. |
| 1559 | Location current = token.start_; |
| 1560 | const bool isNegative = *current == '-'; |
| 1561 | if (isNegative) { |
| 1562 | ++current; |
| 1563 | } |
| 1564 | |
| 1565 | // We assume we can represent the largest and smallest integer types as |
| 1566 | // unsigned integers with separate sign. This is only true if they can fit |
| 1567 | // into an unsigned integer. |
| 1568 | static_assert(Value::maxLargestInt <= Value::maxLargestUInt, |
| 1569 | "Int must be smaller than UInt"); |
| 1570 | |
| 1571 | // We need to convert minLargestInt into a positive number. The easiest way |
| 1572 | // to do this conversion is to assume our "threshold" value of minLargestInt |
| 1573 | // divided by 10 can fit in maxLargestInt when absolute valued. This should |
| 1574 | // be a safe assumption. |
| 1575 | static_assert(Value::minLargestInt <= -Value::maxLargestInt, |
| 1576 | "The absolute value of minLargestInt must be greater than or " |
| 1577 | "equal to maxLargestInt"); |
| 1578 | static_assert(Value::minLargestInt / 10 >= -Value::maxLargestInt, |
| 1579 | "The absolute value of minLargestInt must be only 1 magnitude " |
| 1580 | "larger than maxLargest Int"); |
| 1581 | |
| 1582 | static constexpr Value::LargestUInt positive_threshold = |
| 1583 | Value::maxLargestUInt / 10; |
| 1584 | static constexpr Value::UInt positive_last_digit = Value::maxLargestUInt % 10; |
| 1585 | |
| 1586 | // For the negative values, we have to be more careful. Since typically |
| 1587 | // -Value::minLargestInt will cause an overflow, we first divide by 10 and |
| 1588 | // then take the inverse. This assumes that minLargestInt is only a single |
| 1589 | // power of 10 different in magnitude, which we check above. For the last |
| 1590 | // digit, we take the modulus before negating for the same reason. |
| 1591 | static constexpr auto negative_threshold = |
| 1592 | Value::LargestUInt(-(Value::minLargestInt / 10)); |
| 1593 | static constexpr auto negative_last_digit = |
| 1594 | Value::UInt(-(Value::minLargestInt % 10)); |
| 1595 | |
| 1596 | const Value::LargestUInt threshold = |
| 1597 | isNegative ? negative_threshold : positive_threshold; |
| 1598 | const Value::UInt max_last_digit = |
| 1599 | isNegative ? negative_last_digit : positive_last_digit; |
| 1600 | |
| 1601 | Value::LargestUInt value = 0; |
| 1602 | while (current < token.end_) { |
| 1603 | Char c = *current++; |
| 1604 | if (c < '0' || c > '9') |
| 1605 | return decodeDouble(token, decoded); |
| 1606 | |
| 1607 | const auto digit(static_cast<Value::UInt>(c - '0')); |
| 1608 | if (value >= threshold) { |
| 1609 | // We've hit or exceeded the max value divided by 10 (rounded down). If |
| 1610 | // a) we've only just touched the limit, meaing value == threshold, |
| 1611 | // b) this is the last digit, or |
| 1612 | // c) it's small enough to fit in that rounding delta, we're okay. |
| 1613 | // Otherwise treat this number as a double to avoid overflow. |
| 1614 | if (value > threshold || current != token.end_ || |
| 1615 | digit > max_last_digit) { |
| 1616 | return decodeDouble(token, decoded); |
| 1617 | } |
| 1618 | } |
| 1619 | value = value * 10 + digit; |
| 1620 | } |
| 1621 | |
| 1622 | if (isNegative) { |
| 1623 | // We use the same magnitude assumption here, just in case. |
| 1624 | const auto last_digit = static_cast<Value::UInt>(value % 10); |
| 1625 | decoded = -Value::LargestInt(value / 10) * 10 - last_digit; |
| 1626 | } else if (value <= Value::LargestUInt(Value::maxLargestInt)) { |
| 1627 | decoded = Value::LargestInt(value); |
| 1628 | } else { |
| 1629 | decoded = value; |
| 1630 | } |
| 1631 | |
| 1632 | return true; |
| 1633 | } |
| 1634 | |
| 1635 | bool OurReader::decodeDouble(Token& token) { |
| 1636 | Value decoded; |
| 1637 | if (!decodeDouble(token, decoded)) |
| 1638 | return false; |
| 1639 | currentValue().swapPayload(decoded); |
| 1640 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1641 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1642 | return true; |
| 1643 | } |
| 1644 | |
| 1645 | bool OurReader::decodeDouble(Token& token, Value& decoded) { |
| 1646 | double value = 0; |
| 1647 | const String buffer(token.start_, token.end_); |
| 1648 | IStringStream is(buffer); |
| 1649 | if (!(is >> value)) { |
| 1650 | return addError( |
| 1651 | "'" + String(token.start_, token.end_) + "' is not a number.", token); |
| 1652 | } |
| 1653 | decoded = value; |
| 1654 | return true; |
| 1655 | } |
| 1656 | |
| 1657 | bool OurReader::decodeString(Token& token) { |
| 1658 | String decoded_string; |
| 1659 | if (!decodeString(token, decoded_string)) |
| 1660 | return false; |
| 1661 | Value decoded(decoded_string); |
| 1662 | currentValue().swapPayload(decoded); |
| 1663 | currentValue().setOffsetStart(token.start_ - begin_); |
| 1664 | currentValue().setOffsetLimit(token.end_ - begin_); |
| 1665 | return true; |
| 1666 | } |
| 1667 | |
| 1668 | bool OurReader::decodeString(Token& token, String& decoded) { |
| 1669 | decoded.reserve(static_cast<size_t>(token.end_ - token.start_ - 2)); |
| 1670 | Location current = token.start_ + 1; // skip '"' |
| 1671 | Location end = token.end_ - 1; // do not include '"' |
| 1672 | while (current != end) { |
| 1673 | Char c = *current++; |
| 1674 | if (c == '"') |
| 1675 | break; |
| 1676 | if (c == '\\') { |
| 1677 | if (current == end) |
| 1678 | return addError("Empty escape sequence in string", token, current); |
| 1679 | Char escape = *current++; |
| 1680 | switch (escape) { |
| 1681 | case '"': |
| 1682 | decoded += '"'; |
| 1683 | break; |
| 1684 | case '/': |
| 1685 | decoded += '/'; |
| 1686 | break; |
| 1687 | case '\\': |
| 1688 | decoded += '\\'; |
| 1689 | break; |
| 1690 | case 'b': |
| 1691 | decoded += '\b'; |
| 1692 | break; |
| 1693 | case 'f': |
| 1694 | decoded += '\f'; |
| 1695 | break; |
| 1696 | case 'n': |
| 1697 | decoded += '\n'; |
| 1698 | break; |
| 1699 | case 'r': |
| 1700 | decoded += '\r'; |
| 1701 | break; |
| 1702 | case 't': |
| 1703 | decoded += '\t'; |
| 1704 | break; |
| 1705 | case 'u': { |
| 1706 | unsigned int unicode; |
| 1707 | if (!decodeUnicodeCodePoint(token, current, end, unicode)) |
| 1708 | return false; |
| 1709 | decoded += codePointToUTF8(unicode); |
| 1710 | } break; |
| 1711 | default: |
| 1712 | return addError("Bad escape sequence in string", token, current); |
| 1713 | } |
| 1714 | } else { |
| 1715 | decoded += c; |
| 1716 | } |
| 1717 | } |
| 1718 | return true; |
| 1719 | } |
| 1720 | |
| 1721 | bool OurReader::decodeUnicodeCodePoint(Token& token, Location& current, |
| 1722 | Location end, unsigned int& unicode) { |
| 1723 | |
| 1724 | if (!decodeUnicodeEscapeSequence(token, current, end, unicode)) |
| 1725 | return false; |
| 1726 | if (unicode >= 0xD800 && unicode <= 0xDBFF) { |
| 1727 | // surrogate pairs |
| 1728 | if (end - current < 6) |
| 1729 | return addError( |
| 1730 | "additional six characters expected to parse unicode surrogate pair.", |
| 1731 | token, current); |
| 1732 | if (*(current++) == '\\' && *(current++) == 'u') { |
| 1733 | unsigned int surrogatePair; |
| 1734 | if (decodeUnicodeEscapeSequence(token, current, end, surrogatePair)) { |
| 1735 | unicode = 0x10000 + ((unicode & 0x3FF) << 10) + (surrogatePair & 0x3FF); |
| 1736 | } else |
| 1737 | return false; |
| 1738 | } else |
| 1739 | return addError("expecting another \\u token to begin the second half of " |
| 1740 | "a unicode surrogate pair", |
| 1741 | token, current); |
| 1742 | } |
| 1743 | return true; |
| 1744 | } |
| 1745 | |
| 1746 | bool OurReader::decodeUnicodeEscapeSequence(Token& token, Location& current, |
| 1747 | Location end, |
| 1748 | unsigned int& ret_unicode) { |
| 1749 | if (end - current < 4) |
| 1750 | return addError( |
| 1751 | "Bad unicode escape sequence in string: four digits expected.", token, |
| 1752 | current); |
| 1753 | int unicode = 0; |
| 1754 | for (int index = 0; index < 4; ++index) { |
| 1755 | Char c = *current++; |
| 1756 | unicode *= 16; |
| 1757 | if (c >= '0' && c <= '9') |
| 1758 | unicode += c - '0'; |
| 1759 | else if (c >= 'a' && c <= 'f') |
| 1760 | unicode += c - 'a' + 10; |
| 1761 | else if (c >= 'A' && c <= 'F') |
| 1762 | unicode += c - 'A' + 10; |
| 1763 | else |
| 1764 | return addError( |
| 1765 | "Bad unicode escape sequence in string: hexadecimal digit expected.", |
| 1766 | token, current); |
| 1767 | } |
| 1768 | ret_unicode = static_cast<unsigned int>(unicode); |
| 1769 | return true; |
| 1770 | } |
| 1771 | |
| 1772 | bool OurReader::addError(const String& message, Token& token, Location extra) { |
| 1773 | ErrorInfo info; |
| 1774 | info.token_ = token; |
| 1775 | info.message_ = message; |
| 1776 | info.extra_ = extra; |
| 1777 | errors_.push_back(info); |
| 1778 | return false; |
| 1779 | } |
| 1780 | |
| 1781 | bool OurReader::recoverFromError(TokenType skipUntilToken) { |
| 1782 | size_t errorCount = errors_.size(); |
| 1783 | Token skip; |
| 1784 | for (;;) { |
| 1785 | if (!readToken(skip)) |
| 1786 | errors_.resize(errorCount); // discard errors caused by recovery |
| 1787 | if (skip.type_ == skipUntilToken || skip.type_ == tokenEndOfStream) |
| 1788 | break; |
| 1789 | } |
| 1790 | errors_.resize(errorCount); |
| 1791 | return false; |
| 1792 | } |
| 1793 | |
| 1794 | bool OurReader::addErrorAndRecover(const String& message, Token& token, |
| 1795 | TokenType skipUntilToken) { |
| 1796 | addError(message, token); |
| 1797 | return recoverFromError(skipUntilToken); |
| 1798 | } |
| 1799 | |
| 1800 | Value& OurReader::currentValue() { return *(nodes_.top()); } |
| 1801 | |
| 1802 | OurReader::Char OurReader::getNextChar() { |
| 1803 | if (current_ == end_) |
| 1804 | return 0; |
| 1805 | return *current_++; |
| 1806 | } |
| 1807 | |
| 1808 | void OurReader::getLocationLineAndColumn(Location location, int& line, |
| 1809 | int& column) const { |
| 1810 | Location current = begin_; |
| 1811 | Location lastLineStart = current; |
| 1812 | line = 0; |
| 1813 | while (current < location && current != end_) { |
| 1814 | Char c = *current++; |
| 1815 | if (c == '\r') { |
| 1816 | if (*current == '\n') |
| 1817 | ++current; |
| 1818 | lastLineStart = current; |
| 1819 | ++line; |
| 1820 | } else if (c == '\n') { |
| 1821 | lastLineStart = current; |
| 1822 | ++line; |
| 1823 | } |
| 1824 | } |
| 1825 | // column & line start at 1 |
| 1826 | column = int(location - lastLineStart) + 1; |
| 1827 | ++line; |
| 1828 | } |
| 1829 | |
| 1830 | String OurReader::getLocationLineAndColumn(Location location) const { |
| 1831 | int line, column; |
| 1832 | getLocationLineAndColumn(location, line, column); |
| 1833 | char buffer[18 + 16 + 16 + 1]; |
| 1834 | jsoncpp_snprintf(buffer, sizeof(buffer), "Line %d, Column %d", line, column); |
| 1835 | return buffer; |
| 1836 | } |
| 1837 | |
| 1838 | String OurReader::getFormattedErrorMessages() const { |
| 1839 | String formattedMessage; |
| 1840 | for (const auto& error : errors_) { |
| 1841 | formattedMessage += |
| 1842 | "* " + getLocationLineAndColumn(error.token_.start_) + "\n"; |
| 1843 | formattedMessage += " " + error.message_ + "\n"; |
| 1844 | if (error.extra_) |
| 1845 | formattedMessage += |
| 1846 | "See " + getLocationLineAndColumn(error.extra_) + " for detail.\n"; |
| 1847 | } |
| 1848 | return formattedMessage; |
| 1849 | } |
| 1850 | |
| 1851 | std::vector<OurReader::StructuredError> OurReader::getStructuredErrors() const { |
| 1852 | std::vector<OurReader::StructuredError> allErrors; |
| 1853 | for (const auto& error : errors_) { |
| 1854 | OurReader::StructuredError structured; |
| 1855 | structured.offset_start = error.token_.start_ - begin_; |
| 1856 | structured.offset_limit = error.token_.end_ - begin_; |
| 1857 | structured.message = error.message_; |
| 1858 | allErrors.push_back(structured); |
| 1859 | } |
| 1860 | return allErrors; |
| 1861 | } |
| 1862 | |
| 1863 | class OurCharReader : public CharReader { |
| 1864 | bool const collectComments_; |
| 1865 | OurReader reader_; |
| 1866 | |
| 1867 | public: |
| 1868 | OurCharReader(bool collectComments, OurFeatures const& features) |
| 1869 | : collectComments_(collectComments), reader_(features) {} |
| 1870 | bool parse(char const* beginDoc, char const* endDoc, Value* root, |
| 1871 | String* errs) override { |
| 1872 | bool ok = reader_.parse(beginDoc, endDoc, *root, collectComments_); |
| 1873 | if (errs) { |
| 1874 | *errs = reader_.getFormattedErrorMessages(); |
| 1875 | } |
| 1876 | return ok; |
| 1877 | } |
| 1878 | }; |
| 1879 | |
| 1880 | CharReaderBuilder::CharReaderBuilder() { setDefaults(&settings_); } |
| 1881 | CharReaderBuilder::~CharReaderBuilder() = default; |
| 1882 | CharReader* CharReaderBuilder::newCharReader() const { |
| 1883 | bool collectComments = settings_["collectComments"].asBool(); |
| 1884 | OurFeatures features = OurFeatures::all(); |
| 1885 | features.allowComments_ = settings_["allowComments"].asBool(); |
| 1886 | features.allowTrailingCommas_ = settings_["allowTrailingCommas"].asBool(); |
| 1887 | features.strictRoot_ = settings_["strictRoot"].asBool(); |
| 1888 | features.allowDroppedNullPlaceholders_ = |
| 1889 | settings_["allowDroppedNullPlaceholders"].asBool(); |
| 1890 | features.allowNumericKeys_ = settings_["allowNumericKeys"].asBool(); |
| 1891 | features.allowSingleQuotes_ = settings_["allowSingleQuotes"].asBool(); |
| 1892 | |
| 1893 | // Stack limit is always a size_t, so we get this as an unsigned int |
| 1894 | // regardless of it we have 64-bit integer support enabled. |
| 1895 | features.stackLimit_ = static_cast<size_t>(settings_["stackLimit"].asUInt()); |
| 1896 | features.failIfExtra_ = settings_["failIfExtra"].asBool(); |
| 1897 | features.rejectDupKeys_ = settings_["rejectDupKeys"].asBool(); |
| 1898 | features.allowSpecialFloats_ = settings_["allowSpecialFloats"].asBool(); |
| 1899 | features.skipBom_ = settings_["skipBom"].asBool(); |
| 1900 | return new OurCharReader(collectComments, features); |
| 1901 | } |
| 1902 | |
| 1903 | bool CharReaderBuilder::validate(Json::Value* invalid) const { |
| 1904 | static const auto& valid_keys = *new std::set<String>{ |
| 1905 | "collectComments", |
| 1906 | "allowComments", |
| 1907 | "allowTrailingCommas", |
| 1908 | "strictRoot", |
| 1909 | "allowDroppedNullPlaceholders", |
| 1910 | "allowNumericKeys", |
| 1911 | "allowSingleQuotes", |
| 1912 | "stackLimit", |
| 1913 | "failIfExtra", |
| 1914 | "rejectDupKeys", |
| 1915 | "allowSpecialFloats", |
| 1916 | "skipBom", |
| 1917 | }; |
| 1918 | for (auto si = settings_.begin(); si != settings_.end(); ++si) { |
| 1919 | auto key = si.name(); |
| 1920 | if (valid_keys.count(key)) |
| 1921 | continue; |
| 1922 | if (invalid) |
Elliott Hughes | 1601ea0 | 2021-12-07 09:43:38 -0800 | [diff] [blame] | 1923 | (*invalid)[key] = *si; |
Haibo Huang | b0bee82 | 2021-02-24 15:40:15 -0800 | [diff] [blame] | 1924 | else |
| 1925 | return false; |
| 1926 | } |
| 1927 | return invalid ? invalid->empty() : true; |
| 1928 | } |
| 1929 | |
| 1930 | Value& CharReaderBuilder::operator[](const String& key) { |
| 1931 | return settings_[key]; |
| 1932 | } |
| 1933 | // static |
| 1934 | void CharReaderBuilder::strictMode(Json::Value* settings) { |
| 1935 | //! [CharReaderBuilderStrictMode] |
| 1936 | (*settings)["allowComments"] = false; |
| 1937 | (*settings)["allowTrailingCommas"] = false; |
| 1938 | (*settings)["strictRoot"] = true; |
| 1939 | (*settings)["allowDroppedNullPlaceholders"] = false; |
| 1940 | (*settings)["allowNumericKeys"] = false; |
| 1941 | (*settings)["allowSingleQuotes"] = false; |
| 1942 | (*settings)["stackLimit"] = 1000; |
| 1943 | (*settings)["failIfExtra"] = true; |
| 1944 | (*settings)["rejectDupKeys"] = true; |
| 1945 | (*settings)["allowSpecialFloats"] = false; |
| 1946 | (*settings)["skipBom"] = true; |
| 1947 | //! [CharReaderBuilderStrictMode] |
| 1948 | } |
| 1949 | // static |
| 1950 | void CharReaderBuilder::setDefaults(Json::Value* settings) { |
| 1951 | //! [CharReaderBuilderDefaults] |
| 1952 | (*settings)["collectComments"] = true; |
| 1953 | (*settings)["allowComments"] = true; |
| 1954 | (*settings)["allowTrailingCommas"] = true; |
| 1955 | (*settings)["strictRoot"] = false; |
| 1956 | (*settings)["allowDroppedNullPlaceholders"] = false; |
| 1957 | (*settings)["allowNumericKeys"] = false; |
| 1958 | (*settings)["allowSingleQuotes"] = false; |
| 1959 | (*settings)["stackLimit"] = 1000; |
| 1960 | (*settings)["failIfExtra"] = false; |
| 1961 | (*settings)["rejectDupKeys"] = false; |
| 1962 | (*settings)["allowSpecialFloats"] = false; |
| 1963 | (*settings)["skipBom"] = true; |
| 1964 | //! [CharReaderBuilderDefaults] |
| 1965 | } |
| 1966 | |
| 1967 | ////////////////////////////////// |
| 1968 | // global functions |
| 1969 | |
| 1970 | bool parseFromStream(CharReader::Factory const& fact, IStream& sin, Value* root, |
| 1971 | String* errs) { |
| 1972 | OStringStream ssin; |
| 1973 | ssin << sin.rdbuf(); |
| 1974 | String doc = ssin.str(); |
| 1975 | char const* begin = doc.data(); |
| 1976 | char const* end = begin + doc.size(); |
| 1977 | // Note that we do not actually need a null-terminator. |
| 1978 | CharReaderPtr const reader(fact.newCharReader()); |
| 1979 | return reader->parse(begin, end, root, errs); |
| 1980 | } |
| 1981 | |
| 1982 | IStream& operator>>(IStream& sin, Value& root) { |
| 1983 | CharReaderBuilder b; |
| 1984 | String errs; |
| 1985 | bool ok = parseFromStream(b, sin, &root, &errs); |
| 1986 | if (!ok) { |
| 1987 | throwRuntimeError(errs); |
Derek Sollenberger | 2eb3b4d | 2016-01-11 14:41:40 -0500 | [diff] [blame] | 1988 | } |
| 1989 | return sin; |
| 1990 | } |
Leon Scroggins III | f59fb0e | 2014-05-28 15:19:42 -0400 | [diff] [blame] | 1991 | |
| 1992 | } // namespace Json |