]>
Commit | Line | Data |
---|---|---|
4a5fcab7 AL |
1 | /* |
2 | * JSON Parser | |
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 <stdbool.h> | |
c96c84a9 | 15 | #include <stdarg.h> |
4a5fcab7 AL |
16 | |
17 | #include "qemu-common.h" | |
18 | #include "qstring.h" | |
19 | #include "qint.h" | |
20 | #include "qdict.h" | |
21 | #include "qlist.h" | |
22 | #include "qfloat.h" | |
23 | #include "qbool.h" | |
24 | #include "json-parser.h" | |
25 | #include "json-lexer.h" | |
26 | ||
27 | typedef struct JSONParserContext | |
28 | { | |
29 | } JSONParserContext; | |
30 | ||
31 | #define BUG_ON(cond) assert(!(cond)) | |
32 | ||
33 | /** | |
34 | * TODO | |
35 | * | |
36 | * 0) make errors meaningful again | |
37 | * 1) add geometry information to tokens | |
38 | * 3) should we return a parsed size? | |
39 | * 4) deal with premature EOI | |
40 | */ | |
41 | ||
42 | static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap); | |
43 | ||
44 | /** | |
45 | * Token manipulators | |
46 | * | |
47 | * tokens are dictionaries that contain a type, a string value, and geometry information | |
48 | * about a token identified by the lexer. These are routines that make working with | |
49 | * these objects a bit easier. | |
50 | */ | |
51 | static const char *token_get_value(QObject *obj) | |
52 | { | |
53 | return qdict_get_str(qobject_to_qdict(obj), "token"); | |
54 | } | |
55 | ||
56 | static JSONTokenType token_get_type(QObject *obj) | |
57 | { | |
58 | return qdict_get_int(qobject_to_qdict(obj), "type"); | |
59 | } | |
60 | ||
61 | static int token_is_operator(QObject *obj, char op) | |
62 | { | |
63 | const char *val; | |
64 | ||
65 | if (token_get_type(obj) != JSON_OPERATOR) { | |
66 | return 0; | |
67 | } | |
68 | ||
69 | val = token_get_value(obj); | |
70 | ||
71 | return (val[0] == op) && (val[1] == 0); | |
72 | } | |
73 | ||
74 | static int token_is_keyword(QObject *obj, const char *value) | |
75 | { | |
76 | if (token_get_type(obj) != JSON_KEYWORD) { | |
77 | return 0; | |
78 | } | |
79 | ||
80 | return strcmp(token_get_value(obj), value) == 0; | |
81 | } | |
82 | ||
83 | static int token_is_escape(QObject *obj, const char *value) | |
84 | { | |
85 | if (token_get_type(obj) != JSON_ESCAPE) { | |
86 | return 0; | |
87 | } | |
88 | ||
89 | return (strcmp(token_get_value(obj), value) == 0); | |
90 | } | |
91 | ||
92 | /** | |
93 | * Error handler | |
94 | */ | |
95 | static void parse_error(JSONParserContext *ctxt, QObject *token, const char *msg, ...) | |
96 | { | |
c96c84a9 AK |
97 | va_list ap; |
98 | va_start(ap, msg); | |
99 | fprintf(stderr, "parse error: "); | |
100 | vfprintf(stderr, msg, ap); | |
101 | fprintf(stderr, "\n"); | |
102 | va_end(ap); | |
4a5fcab7 AL |
103 | } |
104 | ||
105 | /** | |
106 | * String helpers | |
107 | * | |
108 | * These helpers are used to unescape strings. | |
109 | */ | |
110 | static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length) | |
111 | { | |
112 | if (wchar <= 0x007F) { | |
113 | BUG_ON(buffer_length < 2); | |
114 | ||
115 | buffer[0] = wchar & 0x7F; | |
116 | buffer[1] = 0; | |
117 | } else if (wchar <= 0x07FF) { | |
118 | BUG_ON(buffer_length < 3); | |
119 | ||
120 | buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F); | |
121 | buffer[1] = 0x80 | (wchar & 0x3F); | |
122 | buffer[2] = 0; | |
123 | } else { | |
124 | BUG_ON(buffer_length < 4); | |
125 | ||
126 | buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F); | |
127 | buffer[1] = 0x80 | ((wchar >> 6) & 0x3F); | |
128 | buffer[2] = 0x80 | (wchar & 0x3F); | |
129 | buffer[3] = 0; | |
130 | } | |
131 | } | |
132 | ||
133 | static int hex2decimal(char ch) | |
134 | { | |
135 | if (ch >= '0' && ch <= '9') { | |
136 | return (ch - '0'); | |
137 | } else if (ch >= 'a' && ch <= 'f') { | |
138 | return 10 + (ch - 'a'); | |
139 | } else if (ch >= 'A' && ch <= 'F') { | |
140 | return 10 + (ch - 'A'); | |
141 | } | |
142 | ||
143 | return -1; | |
144 | } | |
145 | ||
146 | /** | |
147 | * parse_string(): Parse a json string and return a QObject | |
148 | * | |
149 | * string | |
150 | * "" | |
151 | * " chars " | |
152 | * chars | |
153 | * char | |
154 | * char chars | |
155 | * char | |
156 | * any-Unicode-character- | |
157 | * except-"-or-\-or- | |
158 | * control-character | |
159 | * \" | |
160 | * \\ | |
161 | * \/ | |
162 | * \b | |
163 | * \f | |
164 | * \n | |
165 | * \r | |
166 | * \t | |
167 | * \u four-hex-digits | |
168 | */ | |
169 | static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token) | |
170 | { | |
171 | const char *ptr = token_get_value(token); | |
172 | QString *str; | |
173 | int double_quote = 1; | |
174 | ||
175 | if (*ptr == '"') { | |
176 | double_quote = 1; | |
177 | } else { | |
178 | double_quote = 0; | |
179 | } | |
180 | ptr++; | |
181 | ||
182 | str = qstring_new(); | |
183 | while (*ptr && | |
184 | ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) { | |
185 | if (*ptr == '\\') { | |
186 | ptr++; | |
187 | ||
188 | switch (*ptr) { | |
189 | case '"': | |
190 | qstring_append(str, "\""); | |
191 | ptr++; | |
192 | break; | |
193 | case '\'': | |
194 | qstring_append(str, "'"); | |
195 | ptr++; | |
196 | break; | |
197 | case '\\': | |
198 | qstring_append(str, "\\"); | |
199 | ptr++; | |
200 | break; | |
201 | case '/': | |
202 | qstring_append(str, "/"); | |
203 | ptr++; | |
204 | break; | |
205 | case 'b': | |
206 | qstring_append(str, "\b"); | |
207 | ptr++; | |
208 | break; | |
209 | case 'n': | |
210 | qstring_append(str, "\n"); | |
211 | ptr++; | |
212 | break; | |
213 | case 'r': | |
214 | qstring_append(str, "\r"); | |
215 | ptr++; | |
216 | break; | |
217 | case 't': | |
218 | qstring_append(str, "\t"); | |
219 | ptr++; | |
220 | break; | |
221 | case 'u': { | |
222 | uint16_t unicode_char = 0; | |
223 | char utf8_char[4]; | |
224 | int i = 0; | |
225 | ||
226 | ptr++; | |
227 | ||
228 | for (i = 0; i < 4; i++) { | |
229 | if (qemu_isxdigit(*ptr)) { | |
230 | unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4); | |
231 | } else { | |
232 | parse_error(ctxt, token, | |
233 | "invalid hex escape sequence in string"); | |
234 | goto out; | |
235 | } | |
236 | ptr++; | |
237 | } | |
238 | ||
239 | wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char)); | |
240 | qstring_append(str, utf8_char); | |
241 | } break; | |
242 | default: | |
243 | parse_error(ctxt, token, "invalid escape sequence in string"); | |
244 | goto out; | |
245 | } | |
246 | } else { | |
247 | char dummy[2]; | |
248 | ||
249 | dummy[0] = *ptr++; | |
250 | dummy[1] = 0; | |
251 | ||
252 | qstring_append(str, dummy); | |
253 | } | |
254 | } | |
255 | ||
4a5fcab7 AL |
256 | return str; |
257 | ||
258 | out: | |
259 | QDECREF(str); | |
260 | return NULL; | |
261 | } | |
262 | ||
263 | /** | |
264 | * Parsing rules | |
265 | */ | |
266 | static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap) | |
267 | { | |
268 | QObject *key, *token = NULL, *value, *peek; | |
269 | QList *working = qlist_copy(*tokens); | |
270 | ||
271 | peek = qlist_peek(working); | |
272 | key = parse_value(ctxt, &working, ap); | |
d758d90f | 273 | if (!key || qobject_type(key) != QTYPE_QSTRING) { |
4a5fcab7 AL |
274 | parse_error(ctxt, peek, "key is not a string in object"); |
275 | goto out; | |
276 | } | |
277 | ||
278 | token = qlist_pop(working); | |
279 | if (!token_is_operator(token, ':')) { | |
280 | parse_error(ctxt, token, "missing : in object pair"); | |
281 | goto out; | |
282 | } | |
283 | ||
284 | value = parse_value(ctxt, &working, ap); | |
285 | if (value == NULL) { | |
286 | parse_error(ctxt, token, "Missing value in dict"); | |
287 | goto out; | |
288 | } | |
289 | ||
290 | qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value); | |
291 | ||
292 | qobject_decref(token); | |
293 | qobject_decref(key); | |
294 | QDECREF(*tokens); | |
295 | *tokens = working; | |
296 | ||
297 | return 0; | |
298 | ||
299 | out: | |
300 | qobject_decref(token); | |
301 | qobject_decref(key); | |
302 | QDECREF(working); | |
303 | ||
304 | return -1; | |
305 | } | |
306 | ||
307 | static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap) | |
308 | { | |
309 | QDict *dict = NULL; | |
310 | QObject *token, *peek; | |
311 | QList *working = qlist_copy(*tokens); | |
312 | ||
313 | token = qlist_pop(working); | |
314 | if (!token_is_operator(token, '{')) { | |
315 | goto out; | |
316 | } | |
317 | qobject_decref(token); | |
318 | token = NULL; | |
319 | ||
320 | dict = qdict_new(); | |
321 | ||
322 | peek = qlist_peek(working); | |
323 | if (!token_is_operator(peek, '}')) { | |
324 | if (parse_pair(ctxt, dict, &working, ap) == -1) { | |
325 | goto out; | |
326 | } | |
327 | ||
328 | token = qlist_pop(working); | |
329 | while (!token_is_operator(token, '}')) { | |
330 | if (!token_is_operator(token, ',')) { | |
331 | parse_error(ctxt, token, "expected separator in dict"); | |
332 | goto out; | |
333 | } | |
334 | qobject_decref(token); | |
335 | token = NULL; | |
336 | ||
337 | if (parse_pair(ctxt, dict, &working, ap) == -1) { | |
338 | goto out; | |
339 | } | |
340 | ||
341 | token = qlist_pop(working); | |
342 | } | |
343 | qobject_decref(token); | |
344 | token = NULL; | |
345 | } else { | |
346 | token = qlist_pop(working); | |
347 | qobject_decref(token); | |
348 | token = NULL; | |
349 | } | |
350 | ||
351 | QDECREF(*tokens); | |
352 | *tokens = working; | |
353 | ||
354 | return QOBJECT(dict); | |
355 | ||
356 | out: | |
357 | qobject_decref(token); | |
358 | QDECREF(working); | |
359 | QDECREF(dict); | |
360 | return NULL; | |
361 | } | |
362 | ||
363 | static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap) | |
364 | { | |
365 | QList *list = NULL; | |
366 | QObject *token, *peek; | |
367 | QList *working = qlist_copy(*tokens); | |
368 | ||
369 | token = qlist_pop(working); | |
370 | if (!token_is_operator(token, '[')) { | |
371 | goto out; | |
372 | } | |
373 | qobject_decref(token); | |
374 | token = NULL; | |
375 | ||
376 | list = qlist_new(); | |
377 | ||
378 | peek = qlist_peek(working); | |
379 | if (!token_is_operator(peek, ']')) { | |
380 | QObject *obj; | |
381 | ||
382 | obj = parse_value(ctxt, &working, ap); | |
383 | if (obj == NULL) { | |
384 | parse_error(ctxt, token, "expecting value"); | |
385 | goto out; | |
386 | } | |
387 | ||
388 | qlist_append_obj(list, obj); | |
389 | ||
390 | token = qlist_pop(working); | |
391 | while (!token_is_operator(token, ']')) { | |
392 | if (!token_is_operator(token, ',')) { | |
393 | parse_error(ctxt, token, "expected separator in list"); | |
394 | goto out; | |
395 | } | |
396 | ||
397 | qobject_decref(token); | |
398 | token = NULL; | |
399 | ||
400 | obj = parse_value(ctxt, &working, ap); | |
401 | if (obj == NULL) { | |
402 | parse_error(ctxt, token, "expecting value"); | |
403 | goto out; | |
404 | } | |
405 | ||
406 | qlist_append_obj(list, obj); | |
407 | ||
408 | token = qlist_pop(working); | |
409 | } | |
410 | ||
411 | qobject_decref(token); | |
412 | token = NULL; | |
413 | } else { | |
414 | token = qlist_pop(working); | |
415 | qobject_decref(token); | |
416 | token = NULL; | |
417 | } | |
418 | ||
419 | QDECREF(*tokens); | |
420 | *tokens = working; | |
421 | ||
422 | return QOBJECT(list); | |
423 | ||
424 | out: | |
425 | qobject_decref(token); | |
426 | QDECREF(working); | |
427 | QDECREF(list); | |
428 | return NULL; | |
429 | } | |
430 | ||
431 | static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens) | |
432 | { | |
433 | QObject *token, *ret; | |
434 | QList *working = qlist_copy(*tokens); | |
435 | ||
436 | token = qlist_pop(working); | |
437 | ||
438 | if (token_get_type(token) != JSON_KEYWORD) { | |
439 | goto out; | |
440 | } | |
441 | ||
442 | if (token_is_keyword(token, "true")) { | |
443 | ret = QOBJECT(qbool_from_int(true)); | |
444 | } else if (token_is_keyword(token, "false")) { | |
445 | ret = QOBJECT(qbool_from_int(false)); | |
446 | } else { | |
447 | parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token)); | |
448 | goto out; | |
449 | } | |
450 | ||
451 | qobject_decref(token); | |
452 | QDECREF(*tokens); | |
453 | *tokens = working; | |
454 | ||
455 | return ret; | |
456 | ||
457 | out: | |
458 | qobject_decref(token); | |
459 | QDECREF(working); | |
460 | ||
461 | return NULL; | |
462 | } | |
463 | ||
464 | static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap) | |
465 | { | |
466 | QObject *token = NULL, *obj; | |
467 | QList *working = qlist_copy(*tokens); | |
468 | ||
469 | if (ap == NULL) { | |
470 | goto out; | |
471 | } | |
472 | ||
473 | token = qlist_pop(working); | |
474 | ||
475 | if (token_is_escape(token, "%p")) { | |
476 | obj = va_arg(*ap, QObject *); | |
477 | } else if (token_is_escape(token, "%i")) { | |
478 | obj = QOBJECT(qbool_from_int(va_arg(*ap, int))); | |
479 | } else if (token_is_escape(token, "%d")) { | |
480 | obj = QOBJECT(qint_from_int(va_arg(*ap, int))); | |
481 | } else if (token_is_escape(token, "%ld")) { | |
482 | obj = QOBJECT(qint_from_int(va_arg(*ap, long))); | |
2c0d4b36 RT |
483 | } else if (token_is_escape(token, "%lld") || |
484 | token_is_escape(token, "%I64d")) { | |
4a5fcab7 AL |
485 | obj = QOBJECT(qint_from_int(va_arg(*ap, long long))); |
486 | } else if (token_is_escape(token, "%s")) { | |
487 | obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *))); | |
488 | } else if (token_is_escape(token, "%f")) { | |
489 | obj = QOBJECT(qfloat_from_double(va_arg(*ap, double))); | |
490 | } else { | |
491 | goto out; | |
492 | } | |
493 | ||
494 | qobject_decref(token); | |
495 | QDECREF(*tokens); | |
496 | *tokens = working; | |
497 | ||
498 | return obj; | |
499 | ||
500 | out: | |
501 | qobject_decref(token); | |
502 | QDECREF(working); | |
503 | ||
504 | return NULL; | |
505 | } | |
506 | ||
507 | static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens) | |
508 | { | |
509 | QObject *token, *obj; | |
510 | QList *working = qlist_copy(*tokens); | |
511 | ||
512 | token = qlist_pop(working); | |
513 | switch (token_get_type(token)) { | |
514 | case JSON_STRING: | |
515 | obj = QOBJECT(qstring_from_escaped_str(ctxt, token)); | |
516 | break; | |
517 | case JSON_INTEGER: | |
518 | obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10))); | |
519 | break; | |
520 | case JSON_FLOAT: | |
521 | /* FIXME dependent on locale */ | |
522 | obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL))); | |
523 | break; | |
524 | default: | |
525 | goto out; | |
526 | } | |
527 | ||
528 | qobject_decref(token); | |
529 | QDECREF(*tokens); | |
530 | *tokens = working; | |
531 | ||
532 | return obj; | |
533 | ||
534 | out: | |
535 | qobject_decref(token); | |
536 | QDECREF(working); | |
537 | ||
538 | return NULL; | |
539 | } | |
540 | ||
541 | static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap) | |
542 | { | |
543 | QObject *obj; | |
544 | ||
545 | obj = parse_object(ctxt, tokens, ap); | |
546 | if (obj == NULL) { | |
547 | obj = parse_array(ctxt, tokens, ap); | |
548 | } | |
549 | if (obj == NULL) { | |
550 | obj = parse_escape(ctxt, tokens, ap); | |
551 | } | |
552 | if (obj == NULL) { | |
553 | obj = parse_keyword(ctxt, tokens); | |
554 | } | |
555 | if (obj == NULL) { | |
556 | obj = parse_literal(ctxt, tokens); | |
557 | } | |
558 | ||
559 | return obj; | |
560 | } | |
561 | ||
562 | QObject *json_parser_parse(QList *tokens, va_list *ap) | |
563 | { | |
564 | JSONParserContext ctxt = {}; | |
565 | QList *working = qlist_copy(tokens); | |
566 | QObject *result; | |
567 | ||
568 | result = parse_value(&ctxt, &working, ap); | |
569 | ||
570 | QDECREF(working); | |
571 | ||
572 | return result; | |
573 | } |