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 | |
| 68 | static const uint8_t json_lexer[][256] = { |
| 69 | [IN_DONE_STRING] = { |
| 70 | TERMINAL(JSON_STRING), |
| 71 | }, |
| 72 | |
| 73 | /* double quote string */ |
| 74 | [IN_DQ_UCODE3] = { |
| 75 | ['0' ... '9'] = IN_DQ_STRING, |
| 76 | ['a' ... 'f'] = IN_DQ_STRING, |
| 77 | ['A' ... 'F'] = IN_DQ_STRING, |
| 78 | }, |
| 79 | [IN_DQ_UCODE2] = { |
| 80 | ['0' ... '9'] = IN_DQ_UCODE3, |
| 81 | ['a' ... 'f'] = IN_DQ_UCODE3, |
| 82 | ['A' ... 'F'] = IN_DQ_UCODE3, |
| 83 | }, |
| 84 | [IN_DQ_UCODE1] = { |
| 85 | ['0' ... '9'] = IN_DQ_UCODE2, |
| 86 | ['a' ... 'f'] = IN_DQ_UCODE2, |
| 87 | ['A' ... 'F'] = IN_DQ_UCODE2, |
| 88 | }, |
| 89 | [IN_DQ_UCODE0] = { |
| 90 | ['0' ... '9'] = IN_DQ_UCODE1, |
| 91 | ['a' ... 'f'] = IN_DQ_UCODE1, |
| 92 | ['A' ... 'F'] = IN_DQ_UCODE1, |
| 93 | }, |
| 94 | [IN_DQ_STRING_ESCAPE] = { |
| 95 | ['b'] = IN_DQ_STRING, |
| 96 | ['f'] = IN_DQ_STRING, |
| 97 | ['n'] = IN_DQ_STRING, |
| 98 | ['r'] = IN_DQ_STRING, |
| 99 | ['t'] = IN_DQ_STRING, |
| 100 | ['\''] = IN_DQ_STRING, |
| 101 | ['\"'] = IN_DQ_STRING, |
| 102 | ['u'] = IN_DQ_UCODE0, |
| 103 | }, |
| 104 | [IN_DQ_STRING] = { |
| 105 | [1 ... 0xFF] = IN_DQ_STRING, |
| 106 | ['\\'] = IN_DQ_STRING_ESCAPE, |
| 107 | ['"'] = IN_DONE_STRING, |
| 108 | }, |
| 109 | |
| 110 | /* single quote string */ |
| 111 | [IN_SQ_UCODE3] = { |
| 112 | ['0' ... '9'] = IN_SQ_STRING, |
| 113 | ['a' ... 'f'] = IN_SQ_STRING, |
| 114 | ['A' ... 'F'] = IN_SQ_STRING, |
| 115 | }, |
| 116 | [IN_SQ_UCODE2] = { |
| 117 | ['0' ... '9'] = IN_SQ_UCODE3, |
| 118 | ['a' ... 'f'] = IN_SQ_UCODE3, |
| 119 | ['A' ... 'F'] = IN_SQ_UCODE3, |
| 120 | }, |
| 121 | [IN_SQ_UCODE1] = { |
| 122 | ['0' ... '9'] = IN_SQ_UCODE2, |
| 123 | ['a' ... 'f'] = IN_SQ_UCODE2, |
| 124 | ['A' ... 'F'] = IN_SQ_UCODE2, |
| 125 | }, |
| 126 | [IN_SQ_UCODE0] = { |
| 127 | ['0' ... '9'] = IN_SQ_UCODE1, |
| 128 | ['a' ... 'f'] = IN_SQ_UCODE1, |
| 129 | ['A' ... 'F'] = IN_SQ_UCODE1, |
| 130 | }, |
| 131 | [IN_SQ_STRING_ESCAPE] = { |
| 132 | ['b'] = IN_SQ_STRING, |
| 133 | ['f'] = IN_SQ_STRING, |
| 134 | ['n'] = IN_SQ_STRING, |
| 135 | ['r'] = IN_SQ_STRING, |
| 136 | ['t'] = IN_SQ_STRING, |
| 137 | ['\''] = IN_SQ_STRING, |
| 138 | ['\"'] = IN_SQ_STRING, |
| 139 | ['u'] = IN_SQ_UCODE0, |
| 140 | }, |
| 141 | [IN_SQ_STRING] = { |
| 142 | [1 ... 0xFF] = IN_SQ_STRING, |
| 143 | ['\\'] = IN_SQ_STRING_ESCAPE, |
| 144 | ['\''] = IN_DONE_STRING, |
| 145 | }, |
| 146 | |
| 147 | /* Zero */ |
| 148 | [IN_ZERO] = { |
| 149 | TERMINAL(JSON_INTEGER), |
| 150 | ['0' ... '9'] = ERROR, |
| 151 | ['.'] = IN_MANTISSA, |
| 152 | }, |
| 153 | |
| 154 | /* Float */ |
| 155 | [IN_DIGITS] = { |
| 156 | TERMINAL(JSON_FLOAT), |
| 157 | ['0' ... '9'] = IN_DIGITS, |
| 158 | }, |
| 159 | |
| 160 | [IN_DIGIT] = { |
| 161 | ['0' ... '9'] = IN_DIGITS, |
| 162 | }, |
| 163 | |
| 164 | [IN_EXP_E] = { |
| 165 | ['-'] = IN_DIGIT, |
| 166 | ['+'] = IN_DIGIT, |
| 167 | ['0' ... '9'] = IN_DIGITS, |
| 168 | }, |
| 169 | |
| 170 | [IN_MANTISSA_DIGITS] = { |
| 171 | TERMINAL(JSON_FLOAT), |
| 172 | ['0' ... '9'] = IN_MANTISSA_DIGITS, |
| 173 | ['e'] = IN_EXP_E, |
| 174 | ['E'] = IN_EXP_E, |
| 175 | }, |
| 176 | |
| 177 | [IN_MANTISSA] = { |
| 178 | ['0' ... '9'] = IN_MANTISSA_DIGITS, |
| 179 | }, |
| 180 | |
| 181 | /* Number */ |
| 182 | [IN_NONZERO_NUMBER] = { |
| 183 | TERMINAL(JSON_INTEGER), |
| 184 | ['0' ... '9'] = IN_NONZERO_NUMBER, |
| 185 | ['e'] = IN_EXP_E, |
| 186 | ['E'] = IN_EXP_E, |
| 187 | ['.'] = IN_MANTISSA, |
| 188 | }, |
| 189 | |
| 190 | [IN_NEG_NONZERO_NUMBER] = { |
| 191 | ['0'] = IN_ZERO, |
| 192 | ['1' ... '9'] = IN_NONZERO_NUMBER, |
| 193 | }, |
| 194 | |
| 195 | /* keywords */ |
| 196 | [IN_KEYWORD] = { |
| 197 | TERMINAL(JSON_KEYWORD), |
| 198 | ['a' ... 'z'] = IN_KEYWORD, |
| 199 | }, |
| 200 | |
| 201 | /* whitespace */ |
| 202 | [IN_WHITESPACE] = { |
| 203 | TERMINAL(JSON_SKIP), |
| 204 | [' '] = IN_WHITESPACE, |
| 205 | ['\t'] = IN_WHITESPACE, |
| 206 | ['\r'] = IN_WHITESPACE, |
| 207 | ['\n'] = IN_WHITESPACE, |
| 208 | }, |
| 209 | |
| 210 | /* operator */ |
| 211 | [IN_OPERATOR_DONE] = { |
| 212 | TERMINAL(JSON_OPERATOR), |
| 213 | }, |
| 214 | |
| 215 | /* escape */ |
| 216 | [IN_ESCAPE_DONE] = { |
| 217 | TERMINAL(JSON_ESCAPE), |
| 218 | }, |
| 219 | |
| 220 | [IN_ESCAPE_LL] = { |
| 221 | ['d'] = IN_ESCAPE_DONE, |
| 222 | }, |
| 223 | |
| 224 | [IN_ESCAPE_L] = { |
| 225 | ['d'] = IN_ESCAPE_DONE, |
| 226 | ['l'] = IN_ESCAPE_LL, |
| 227 | }, |
| 228 | |
Roy Tam | 2c0d4b3 | 2010-02-04 10:30:30 +0800 | [diff] [blame] | 229 | [IN_ESCAPE_I64] = { |
| 230 | ['d'] = IN_ESCAPE_DONE, |
| 231 | }, |
| 232 | |
| 233 | [IN_ESCAPE_I6] = { |
| 234 | ['4'] = IN_ESCAPE_I64, |
| 235 | }, |
| 236 | |
| 237 | [IN_ESCAPE_I] = { |
| 238 | ['6'] = IN_ESCAPE_I6, |
| 239 | }, |
| 240 | |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 241 | [IN_ESCAPE] = { |
| 242 | ['d'] = IN_ESCAPE_DONE, |
| 243 | ['i'] = IN_ESCAPE_DONE, |
| 244 | ['p'] = IN_ESCAPE_DONE, |
| 245 | ['s'] = IN_ESCAPE_DONE, |
| 246 | ['f'] = IN_ESCAPE_DONE, |
| 247 | ['l'] = IN_ESCAPE_L, |
Roy Tam | 2c0d4b3 | 2010-02-04 10:30:30 +0800 | [diff] [blame] | 248 | ['I'] = IN_ESCAPE_I, |
Anthony Liguori | 5ab8558 | 2009-11-11 10:39:14 -0600 | [diff] [blame] | 249 | }, |
| 250 | |
| 251 | /* top level rule */ |
| 252 | [IN_START] = { |
| 253 | ['"'] = IN_DQ_STRING, |
| 254 | ['\''] = IN_SQ_STRING, |
| 255 | ['0'] = IN_ZERO, |
| 256 | ['1' ... '9'] = IN_NONZERO_NUMBER, |
| 257 | ['-'] = IN_NEG_NONZERO_NUMBER, |
| 258 | ['{'] = IN_OPERATOR_DONE, |
| 259 | ['}'] = IN_OPERATOR_DONE, |
| 260 | ['['] = IN_OPERATOR_DONE, |
| 261 | [']'] = IN_OPERATOR_DONE, |
| 262 | [','] = IN_OPERATOR_DONE, |
| 263 | [':'] = IN_OPERATOR_DONE, |
| 264 | ['a' ... 'z'] = IN_KEYWORD, |
| 265 | ['%'] = IN_ESCAPE, |
| 266 | [' '] = IN_WHITESPACE, |
| 267 | ['\t'] = IN_WHITESPACE, |
| 268 | ['\r'] = IN_WHITESPACE, |
| 269 | ['\n'] = IN_WHITESPACE, |
| 270 | }, |
| 271 | }; |
| 272 | |
| 273 | void json_lexer_init(JSONLexer *lexer, JSONLexerEmitter func) |
| 274 | { |
| 275 | lexer->emit = func; |
| 276 | lexer->state = IN_START; |
| 277 | lexer->token = qstring_new(); |
| 278 | } |
| 279 | |
| 280 | static int json_lexer_feed_char(JSONLexer *lexer, char ch) |
| 281 | { |
| 282 | char buf[2]; |
| 283 | |
| 284 | lexer->x++; |
| 285 | if (ch == '\n') { |
| 286 | lexer->x = 0; |
| 287 | lexer->y++; |
| 288 | } |
| 289 | |
| 290 | lexer->state = json_lexer[lexer->state][(uint8_t)ch]; |
| 291 | |
| 292 | switch (lexer->state) { |
| 293 | case JSON_OPERATOR: |
| 294 | case JSON_ESCAPE: |
| 295 | case JSON_INTEGER: |
| 296 | case JSON_FLOAT: |
| 297 | case JSON_KEYWORD: |
| 298 | case JSON_STRING: |
| 299 | lexer->emit(lexer, lexer->token, lexer->state, lexer->x, lexer->y); |
| 300 | case JSON_SKIP: |
| 301 | lexer->state = json_lexer[IN_START][(uint8_t)ch]; |
| 302 | QDECREF(lexer->token); |
| 303 | lexer->token = qstring_new(); |
| 304 | break; |
| 305 | case ERROR: |
| 306 | return -EINVAL; |
| 307 | default: |
| 308 | break; |
| 309 | } |
| 310 | |
| 311 | buf[0] = ch; |
| 312 | buf[1] = 0; |
| 313 | |
| 314 | qstring_append(lexer->token, buf); |
| 315 | |
| 316 | return 0; |
| 317 | } |
| 318 | |
| 319 | int json_lexer_feed(JSONLexer *lexer, const char *buffer, size_t size) |
| 320 | { |
| 321 | size_t i; |
| 322 | |
| 323 | for (i = 0; i < size; i++) { |
| 324 | int err; |
| 325 | |
| 326 | err = json_lexer_feed_char(lexer, buffer[i]); |
| 327 | if (err < 0) { |
| 328 | return err; |
| 329 | } |
| 330 | } |
| 331 | |
| 332 | return 0; |
| 333 | } |
| 334 | |
| 335 | int json_lexer_flush(JSONLexer *lexer) |
| 336 | { |
| 337 | return json_lexer_feed_char(lexer, 0); |
| 338 | } |
| 339 | |
| 340 | void json_lexer_destroy(JSONLexer *lexer) |
| 341 | { |
| 342 | QDECREF(lexer->token); |
| 343 | } |