Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 1 | /* |
| 2 | * JSON lexer |
| 3 | * |
| 4 | * Copyright IBM, Corp. 2009 |
| 5 | * |
| 6 | * Authors: |
| 7 | * Anthony Liguori <aliguori@us.ibm.com> |
| 8 | * |
| 9 | * This work is licensed under the terms of the GNU LGPL, version 2.1 or later. |
| 10 | * See the COPYING.LIB file in the top-level directory. |
| 11 | * |
| 12 | */ |
| 13 | |
| 14 | #include "qstring.h" |
| 15 | #include "qlist.h" |
| 16 | #include "qdict.h" |
| 17 | #include "qint.h" |
| 18 | #include "qemu-common.h" |
| 19 | #include "json-lexer.h" |
| 20 | |
| 21 | /* |
| 22 | * \"([^\\\"]|(\\\"\\'\\\\\\/\\b\\f\\n\\r\\t\\u[0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F]))*\" |
| 23 | * '([^\\']|(\\\"\\'\\\\\\/\\b\\f\\n\\r\\t\\u[0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F]))*' |
| 24 | * 0|([1-9][0-9]*(.[0-9]+)?([eE]([-+])?[0-9]+)) |
| 25 | * [{}\[\],:] |
| 26 | * [a-z]+ |
| 27 | * |
| 28 | */ |
| 29 | |
| 30 | enum json_lexer_state { |
| 31 | ERROR = 0, |
| 32 | IN_DONE_STRING, |
| 33 | IN_DQ_UCODE3, |
| 34 | IN_DQ_UCODE2, |
| 35 | IN_DQ_UCODE1, |
| 36 | IN_DQ_UCODE0, |
| 37 | IN_DQ_STRING_ESCAPE, |
| 38 | IN_DQ_STRING, |
| 39 | IN_SQ_UCODE3, |
| 40 | IN_SQ_UCODE2, |
| 41 | IN_SQ_UCODE1, |
| 42 | IN_SQ_UCODE0, |
| 43 | IN_SQ_STRING_ESCAPE, |
| 44 | IN_SQ_STRING, |
| 45 | IN_ZERO, |
| 46 | IN_DIGITS, |
| 47 | IN_DIGIT, |
| 48 | IN_EXP_E, |
| 49 | IN_MANTISSA, |
| 50 | IN_MANTISSA_DIGITS, |
| 51 | IN_NONZERO_NUMBER, |
| 52 | IN_NEG_NONZERO_NUMBER, |
| 53 | IN_KEYWORD, |
| 54 | IN_ESCAPE, |
| 55 | IN_ESCAPE_L, |
| 56 | IN_ESCAPE_LL, |
Roy Tam | 2c0d4b3 | 2010-02-04 10:30:30 +0800 | [diff] [blame] | 57 | IN_ESCAPE_I, |
| 58 | IN_ESCAPE_I6, |
| 59 | IN_ESCAPE_I64, |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 60 | IN_ESCAPE_DONE, |
| 61 | IN_WHITESPACE, |
| 62 | IN_OPERATOR_DONE, |
| 63 | IN_START, |
| 64 | }; |
| 65 | |
| 66 | #define TERMINAL(state) [0 ... 0x7F] = (state) |
| 67 | |
Paolo Bonzini | f7c0527 | 2010-05-24 09:39:52 +0200 | [diff] [blame^] | 68 | /* Return whether TERMINAL is a terminal state and the transition to it |
| 69 | from OLD_STATE required lookahead. This happens whenever the table |
| 70 | below uses the TERMINAL macro. */ |
| 71 | #define TERMINAL_NEEDED_LOOKAHEAD(old_state, terminal) \ |
| 72 | (json_lexer[(old_state)][0] == (terminal)) |
| 73 | |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 74 | static const uint8_t json_lexer[][256] = { |
| 75 | [IN_DONE_STRING] = { |
| 76 | TERMINAL(JSON_STRING), |
| 77 | }, |
| 78 | |
| 79 | /* double quote string */ |
| 80 | [IN_DQ_UCODE3] = { |
| 81 | ['0' ... '9'] = IN_DQ_STRING, |
| 82 | ['a' ... 'f'] = IN_DQ_STRING, |
| 83 | ['A' ... 'F'] = IN_DQ_STRING, |
| 84 | }, |
| 85 | [IN_DQ_UCODE2] = { |
| 86 | ['0' ... '9'] = IN_DQ_UCODE3, |
| 87 | ['a' ... 'f'] = IN_DQ_UCODE3, |
| 88 | ['A' ... 'F'] = IN_DQ_UCODE3, |
| 89 | }, |
| 90 | [IN_DQ_UCODE1] = { |
| 91 | ['0' ... '9'] = IN_DQ_UCODE2, |
| 92 | ['a' ... 'f'] = IN_DQ_UCODE2, |
| 93 | ['A' ... 'F'] = IN_DQ_UCODE2, |
| 94 | }, |
| 95 | [IN_DQ_UCODE0] = { |
| 96 | ['0' ... '9'] = IN_DQ_UCODE1, |
| 97 | ['a' ... 'f'] = IN_DQ_UCODE1, |
| 98 | ['A' ... 'F'] = IN_DQ_UCODE1, |
| 99 | }, |
| 100 | [IN_DQ_STRING_ESCAPE] = { |
| 101 | ['b'] = IN_DQ_STRING, |
| 102 | ['f'] = IN_DQ_STRING, |
| 103 | ['n'] = IN_DQ_STRING, |
| 104 | ['r'] = IN_DQ_STRING, |
| 105 | ['t'] = IN_DQ_STRING, |
Luiz Capitulino | 1041ba7 | 2010-05-19 16:57:28 -0300 | [diff] [blame] | 106 | ['/'] = IN_DQ_STRING, |
| 107 | ['\\'] = IN_DQ_STRING, |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 108 | ['\''] = IN_DQ_STRING, |
| 109 | ['\"'] = IN_DQ_STRING, |
| 110 | ['u'] = IN_DQ_UCODE0, |
| 111 | }, |
| 112 | [IN_DQ_STRING] = { |
| 113 | [1 ... 0xFF] = IN_DQ_STRING, |
| 114 | ['\\'] = IN_DQ_STRING_ESCAPE, |
| 115 | ['"'] = IN_DONE_STRING, |
| 116 | }, |
| 117 | |
| 118 | /* single quote string */ |
| 119 | [IN_SQ_UCODE3] = { |
| 120 | ['0' ... '9'] = IN_SQ_STRING, |
| 121 | ['a' ... 'f'] = IN_SQ_STRING, |
| 122 | ['A' ... 'F'] = IN_SQ_STRING, |
| 123 | }, |
| 124 | [IN_SQ_UCODE2] = { |
| 125 | ['0' ... '9'] = IN_SQ_UCODE3, |
| 126 | ['a' ... 'f'] = IN_SQ_UCODE3, |
| 127 | ['A' ... 'F'] = IN_SQ_UCODE3, |
| 128 | }, |
| 129 | [IN_SQ_UCODE1] = { |
| 130 | ['0' ... '9'] = IN_SQ_UCODE2, |
| 131 | ['a' ... 'f'] = IN_SQ_UCODE2, |
| 132 | ['A' ... 'F'] = IN_SQ_UCODE2, |
| 133 | }, |
| 134 | [IN_SQ_UCODE0] = { |
| 135 | ['0' ... '9'] = IN_SQ_UCODE1, |
| 136 | ['a' ... 'f'] = IN_SQ_UCODE1, |
| 137 | ['A' ... 'F'] = IN_SQ_UCODE1, |
| 138 | }, |
| 139 | [IN_SQ_STRING_ESCAPE] = { |
| 140 | ['b'] = IN_SQ_STRING, |
| 141 | ['f'] = IN_SQ_STRING, |
| 142 | ['n'] = IN_SQ_STRING, |
| 143 | ['r'] = IN_SQ_STRING, |
| 144 | ['t'] = IN_SQ_STRING, |
Luiz Capitulino | 1041ba7 | 2010-05-19 16:57:28 -0300 | [diff] [blame] | 145 | ['/'] = IN_DQ_STRING, |
| 146 | ['\\'] = IN_DQ_STRING, |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 147 | ['\''] = IN_SQ_STRING, |
| 148 | ['\"'] = IN_SQ_STRING, |
| 149 | ['u'] = IN_SQ_UCODE0, |
| 150 | }, |
| 151 | [IN_SQ_STRING] = { |
| 152 | [1 ... 0xFF] = IN_SQ_STRING, |
| 153 | ['\\'] = IN_SQ_STRING_ESCAPE, |
| 154 | ['\''] = IN_DONE_STRING, |
| 155 | }, |
| 156 | |
| 157 | /* Zero */ |
| 158 | [IN_ZERO] = { |
| 159 | TERMINAL(JSON_INTEGER), |
| 160 | ['0' ... '9'] = ERROR, |
| 161 | ['.'] = IN_MANTISSA, |
| 162 | }, |
| 163 | |
| 164 | /* Float */ |
| 165 | [IN_DIGITS] = { |
| 166 | TERMINAL(JSON_FLOAT), |
| 167 | ['0' ... '9'] = IN_DIGITS, |
| 168 | }, |
| 169 | |
| 170 | [IN_DIGIT] = { |
| 171 | ['0' ... '9'] = IN_DIGITS, |
| 172 | }, |
| 173 | |
| 174 | [IN_EXP_E] = { |
| 175 | ['-'] = IN_DIGIT, |
| 176 | ['+'] = IN_DIGIT, |
| 177 | ['0' ... '9'] = IN_DIGITS, |
| 178 | }, |
| 179 | |
| 180 | [IN_MANTISSA_DIGITS] = { |
| 181 | TERMINAL(JSON_FLOAT), |
| 182 | ['0' ... '9'] = IN_MANTISSA_DIGITS, |
| 183 | ['e'] = IN_EXP_E, |
| 184 | ['E'] = IN_EXP_E, |
| 185 | }, |
| 186 | |
| 187 | [IN_MANTISSA] = { |
| 188 | ['0' ... '9'] = IN_MANTISSA_DIGITS, |
| 189 | }, |
| 190 | |
| 191 | /* Number */ |
| 192 | [IN_NONZERO_NUMBER] = { |
| 193 | TERMINAL(JSON_INTEGER), |
| 194 | ['0' ... '9'] = IN_NONZERO_NUMBER, |
| 195 | ['e'] = IN_EXP_E, |
| 196 | ['E'] = IN_EXP_E, |
| 197 | ['.'] = IN_MANTISSA, |
| 198 | }, |
| 199 | |
| 200 | [IN_NEG_NONZERO_NUMBER] = { |
| 201 | ['0'] = IN_ZERO, |
| 202 | ['1' ... '9'] = IN_NONZERO_NUMBER, |
| 203 | }, |
| 204 | |
| 205 | /* keywords */ |
| 206 | [IN_KEYWORD] = { |
| 207 | TERMINAL(JSON_KEYWORD), |
| 208 | ['a' ... 'z'] = IN_KEYWORD, |
| 209 | }, |
| 210 | |
| 211 | /* whitespace */ |
| 212 | [IN_WHITESPACE] = { |
| 213 | TERMINAL(JSON_SKIP), |
| 214 | [' '] = IN_WHITESPACE, |
| 215 | ['\t'] = IN_WHITESPACE, |
| 216 | ['\r'] = IN_WHITESPACE, |
| 217 | ['\n'] = IN_WHITESPACE, |
| 218 | }, |
| 219 | |
| 220 | /* operator */ |
| 221 | [IN_OPERATOR_DONE] = { |
| 222 | TERMINAL(JSON_OPERATOR), |
| 223 | }, |
| 224 | |
| 225 | /* escape */ |
| 226 | [IN_ESCAPE_DONE] = { |
| 227 | TERMINAL(JSON_ESCAPE), |
| 228 | }, |
| 229 | |
| 230 | [IN_ESCAPE_LL] = { |
| 231 | ['d'] = IN_ESCAPE_DONE, |
| 232 | }, |
| 233 | |
| 234 | [IN_ESCAPE_L] = { |
| 235 | ['d'] = IN_ESCAPE_DONE, |
| 236 | ['l'] = IN_ESCAPE_LL, |
| 237 | }, |
| 238 | |
Roy Tam | 2c0d4b3 | 2010-02-04 10:30:30 +0800 | [diff] [blame] | 239 | [IN_ESCAPE_I64] = { |
| 240 | ['d'] = IN_ESCAPE_DONE, |
| 241 | }, |
| 242 | |
| 243 | [IN_ESCAPE_I6] = { |
| 244 | ['4'] = IN_ESCAPE_I64, |
| 245 | }, |
| 246 | |
| 247 | [IN_ESCAPE_I] = { |
| 248 | ['6'] = IN_ESCAPE_I6, |
| 249 | }, |
| 250 | |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 251 | [IN_ESCAPE] = { |
| 252 | ['d'] = IN_ESCAPE_DONE, |
| 253 | ['i'] = IN_ESCAPE_DONE, |
| 254 | ['p'] = IN_ESCAPE_DONE, |
| 255 | ['s'] = IN_ESCAPE_DONE, |
| 256 | ['f'] = IN_ESCAPE_DONE, |
| 257 | ['l'] = IN_ESCAPE_L, |
Roy Tam | 2c0d4b3 | 2010-02-04 10:30:30 +0800 | [diff] [blame] | 258 | ['I'] = IN_ESCAPE_I, |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 259 | }, |
| 260 | |
| 261 | /* top level rule */ |
| 262 | [IN_START] = { |
| 263 | ['"'] = IN_DQ_STRING, |
| 264 | ['\''] = IN_SQ_STRING, |
| 265 | ['0'] = IN_ZERO, |
| 266 | ['1' ... '9'] = IN_NONZERO_NUMBER, |
| 267 | ['-'] = IN_NEG_NONZERO_NUMBER, |
| 268 | ['{'] = IN_OPERATOR_DONE, |
| 269 | ['}'] = IN_OPERATOR_DONE, |
| 270 | ['['] = IN_OPERATOR_DONE, |
| 271 | [']'] = IN_OPERATOR_DONE, |
| 272 | [','] = IN_OPERATOR_DONE, |
| 273 | [':'] = IN_OPERATOR_DONE, |
| 274 | ['a' ... 'z'] = IN_KEYWORD, |
| 275 | ['%'] = IN_ESCAPE, |
| 276 | [' '] = IN_WHITESPACE, |
| 277 | ['\t'] = IN_WHITESPACE, |
| 278 | ['\r'] = IN_WHITESPACE, |
| 279 | ['\n'] = IN_WHITESPACE, |
| 280 | }, |
| 281 | }; |
| 282 | |
| 283 | void json_lexer_init(JSONLexer *lexer, JSONLexerEmitter func) |
| 284 | { |
| 285 | lexer->emit = func; |
| 286 | lexer->state = IN_START; |
| 287 | lexer->token = qstring_new(); |
Luiz Capitulino | 03308f6 | 2010-05-17 17:50:01 -0300 | [diff] [blame] | 288 | lexer->x = lexer->y = 0; |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 289 | } |
| 290 | |
| 291 | static int json_lexer_feed_char(JSONLexer *lexer, char ch) |
| 292 | { |
Paolo Bonzini | f7c0527 | 2010-05-24 09:39:52 +0200 | [diff] [blame^] | 293 | int char_consumed, new_state; |
| 294 | |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 295 | lexer->x++; |
| 296 | if (ch == '\n') { |
| 297 | lexer->x = 0; |
| 298 | lexer->y++; |
| 299 | } |
| 300 | |
Paolo Bonzini | f7c0527 | 2010-05-24 09:39:52 +0200 | [diff] [blame^] | 301 | do { |
| 302 | new_state = json_lexer[lexer->state][(uint8_t)ch]; |
| 303 | char_consumed = !TERMINAL_NEEDED_LOOKAHEAD(lexer->state, new_state); |
| 304 | if (char_consumed) { |
| 305 | qstring_append_chr(lexer->token, ch); |
| 306 | } |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 307 | |
Paolo Bonzini | f7c0527 | 2010-05-24 09:39:52 +0200 | [diff] [blame^] | 308 | switch (new_state) { |
| 309 | case JSON_OPERATOR: |
| 310 | case JSON_ESCAPE: |
| 311 | case JSON_INTEGER: |
| 312 | case JSON_FLOAT: |
| 313 | case JSON_KEYWORD: |
| 314 | case JSON_STRING: |
| 315 | lexer->emit(lexer, lexer->token, new_state, lexer->x, lexer->y); |
| 316 | case JSON_SKIP: |
| 317 | QDECREF(lexer->token); |
| 318 | lexer->token = qstring_new(); |
| 319 | new_state = IN_START; |
| 320 | break; |
| 321 | case ERROR: |
| 322 | return -EINVAL; |
| 323 | default: |
| 324 | break; |
| 325 | } |
| 326 | lexer->state = new_state; |
| 327 | } while (!char_consumed); |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 328 | return 0; |
| 329 | } |
| 330 | |
| 331 | int json_lexer_feed(JSONLexer *lexer, const char *buffer, size_t size) |
| 332 | { |
| 333 | size_t i; |
| 334 | |
| 335 | for (i = 0; i < size; i++) { |
| 336 | int err; |
| 337 | |
| 338 | err = json_lexer_feed_char(lexer, buffer[i]); |
| 339 | if (err < 0) { |
| 340 | return err; |
| 341 | } |
| 342 | } |
| 343 | |
| 344 | return 0; |
| 345 | } |
| 346 | |
| 347 | int json_lexer_flush(JSONLexer *lexer) |
| 348 | { |
Paolo Bonzini | f7c0527 | 2010-05-24 09:39:52 +0200 | [diff] [blame^] | 349 | return lexer->state == IN_START ? 0 : json_lexer_feed_char(lexer, 0); |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 350 | } |
| 351 | |
| 352 | void json_lexer_destroy(JSONLexer *lexer) |
| 353 | { |
| 354 | QDECREF(lexer->token); |
| 355 | } |