]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | * JSON Parser | |
3 | * | |
4 | * Copyright IBM, Corp. 2009 | |
5 | * | |
6 | * Authors: | |
7 | * Anthony Liguori <[email protected]> | |
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 "qemu/osdep.h" | |
15 | #include "qemu/cutils.h" | |
16 | #include "qemu/unicode.h" | |
17 | #include "qapi/error.h" | |
18 | #include "qemu-common.h" | |
19 | #include "qapi/qmp/qbool.h" | |
20 | #include "qapi/qmp/qdict.h" | |
21 | #include "qapi/qmp/qlist.h" | |
22 | #include "qapi/qmp/qnull.h" | |
23 | #include "qapi/qmp/qnum.h" | |
24 | #include "qapi/qmp/qstring.h" | |
25 | #include "qapi/qmp/json-parser.h" | |
26 | #include "qapi/qmp/json-lexer.h" | |
27 | #include "qapi/qmp/json-streamer.h" | |
28 | ||
29 | typedef struct JSONParserContext | |
30 | { | |
31 | Error *err; | |
32 | JSONToken *current; | |
33 | GQueue *buf; | |
34 | } JSONParserContext; | |
35 | ||
36 | #define BUG_ON(cond) assert(!(cond)) | |
37 | ||
38 | /** | |
39 | * TODO | |
40 | * | |
41 | * 0) make errors meaningful again | |
42 | * 1) add geometry information to tokens | |
43 | * 3) should we return a parsed size? | |
44 | * 4) deal with premature EOI | |
45 | */ | |
46 | ||
47 | static QObject *parse_value(JSONParserContext *ctxt, va_list *ap); | |
48 | ||
49 | /** | |
50 | * Error handler | |
51 | */ | |
52 | static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt, | |
53 | JSONToken *token, const char *msg, ...) | |
54 | { | |
55 | va_list ap; | |
56 | char message[1024]; | |
57 | ||
58 | if (ctxt->err) { | |
59 | return; | |
60 | } | |
61 | va_start(ap, msg); | |
62 | vsnprintf(message, sizeof(message), msg, ap); | |
63 | va_end(ap); | |
64 | error_setg(&ctxt->err, "JSON parse error, %s", message); | |
65 | } | |
66 | ||
67 | static int hex2decimal(char ch) | |
68 | { | |
69 | if (ch >= '0' && ch <= '9') { | |
70 | return (ch - '0'); | |
71 | } else if (ch >= 'a' && ch <= 'f') { | |
72 | return 10 + (ch - 'a'); | |
73 | } else if (ch >= 'A' && ch <= 'F') { | |
74 | return 10 + (ch - 'A'); | |
75 | } | |
76 | abort(); | |
77 | } | |
78 | ||
79 | /** | |
80 | * parse_string(): Parse a JSON string | |
81 | * | |
82 | * From RFC 8259 "The JavaScript Object Notation (JSON) Data | |
83 | * Interchange Format": | |
84 | * | |
85 | * char = unescaped / | |
86 | * escape ( | |
87 | * %x22 / ; " quotation mark U+0022 | |
88 | * %x5C / ; \ reverse solidus U+005C | |
89 | * %x2F / ; / solidus U+002F | |
90 | * %x62 / ; b backspace U+0008 | |
91 | * %x66 / ; f form feed U+000C | |
92 | * %x6E / ; n line feed U+000A | |
93 | * %x72 / ; r carriage return U+000D | |
94 | * %x74 / ; t tab U+0009 | |
95 | * %x75 4HEXDIG ) ; uXXXX U+XXXX | |
96 | * escape = %x5C ; \ | |
97 | * quotation-mark = %x22 ; " | |
98 | * unescaped = %x20-21 / %x23-5B / %x5D-10FFFF | |
99 | * | |
100 | * Extensions over RFC 8259: | |
101 | * - Extra escape sequence in strings: | |
102 | * 0x27 (apostrophe) is recognized after escape, too | |
103 | * - Single-quoted strings: | |
104 | * Like double-quoted strings, except they're delimited by %x27 | |
105 | * (apostrophe) instead of %x22 (quotation mark), and can't contain | |
106 | * unescaped apostrophe, but can contain unescaped quotation mark. | |
107 | * | |
108 | * Note: | |
109 | * - Encoding is modified UTF-8. | |
110 | * - Invalid Unicode characters are rejected. | |
111 | * - Control characters \x00..\x1F are rejected by the lexer. | |
112 | */ | |
113 | static QString *parse_string(JSONParserContext *ctxt, JSONToken *token) | |
114 | { | |
115 | const char *ptr = token->str; | |
116 | QString *str; | |
117 | char quote; | |
118 | int cp, i; | |
119 | char *end; | |
120 | ssize_t len; | |
121 | char utf8_buf[5]; | |
122 | ||
123 | assert(*ptr == '"' || *ptr == '\''); | |
124 | quote = *ptr++; | |
125 | str = qstring_new(); | |
126 | ||
127 | while (*ptr != quote) { | |
128 | assert(*ptr); | |
129 | if (*ptr == '\\') { | |
130 | ptr++; | |
131 | switch (*ptr++) { | |
132 | case '"': | |
133 | qstring_append_chr(str, '"'); | |
134 | break; | |
135 | case '\'': | |
136 | qstring_append_chr(str, '\''); | |
137 | break; | |
138 | case '\\': | |
139 | qstring_append_chr(str, '\\'); | |
140 | break; | |
141 | case '/': | |
142 | qstring_append_chr(str, '/'); | |
143 | break; | |
144 | case 'b': | |
145 | qstring_append_chr(str, '\b'); | |
146 | break; | |
147 | case 'f': | |
148 | qstring_append_chr(str, '\f'); | |
149 | break; | |
150 | case 'n': | |
151 | qstring_append_chr(str, '\n'); | |
152 | break; | |
153 | case 'r': | |
154 | qstring_append_chr(str, '\r'); | |
155 | break; | |
156 | case 't': | |
157 | qstring_append_chr(str, '\t'); | |
158 | break; | |
159 | case 'u': | |
160 | cp = 0; | |
161 | for (i = 0; i < 4; i++) { | |
162 | if (!qemu_isxdigit(*ptr)) { | |
163 | parse_error(ctxt, token, | |
164 | "invalid hex escape sequence in string"); | |
165 | goto out; | |
166 | } | |
167 | cp <<= 4; | |
168 | cp |= hex2decimal(*ptr); | |
169 | ptr++; | |
170 | } | |
171 | ||
172 | if (mod_utf8_encode(utf8_buf, sizeof(utf8_buf), cp) < 0) { | |
173 | parse_error(ctxt, token, | |
174 | "\\u%.4s is not a valid Unicode character", | |
175 | ptr - 3); | |
176 | goto out; | |
177 | } | |
178 | qstring_append(str, utf8_buf); | |
179 | break; | |
180 | default: | |
181 | parse_error(ctxt, token, "invalid escape sequence in string"); | |
182 | goto out; | |
183 | } | |
184 | } else { | |
185 | cp = mod_utf8_codepoint(ptr, 6, &end); | |
186 | if (cp < 0) { | |
187 | parse_error(ctxt, token, "invalid UTF-8 sequence in string"); | |
188 | goto out; | |
189 | } | |
190 | ptr = end; | |
191 | len = mod_utf8_encode(utf8_buf, sizeof(utf8_buf), cp); | |
192 | assert(len >= 0); | |
193 | qstring_append(str, utf8_buf); | |
194 | } | |
195 | } | |
196 | ||
197 | return str; | |
198 | ||
199 | out: | |
200 | qobject_unref(str); | |
201 | return NULL; | |
202 | } | |
203 | ||
204 | /* Note: the token object returned by parser_context_peek_token or | |
205 | * parser_context_pop_token is deleted as soon as parser_context_pop_token | |
206 | * is called again. | |
207 | */ | |
208 | static JSONToken *parser_context_pop_token(JSONParserContext *ctxt) | |
209 | { | |
210 | g_free(ctxt->current); | |
211 | assert(!g_queue_is_empty(ctxt->buf)); | |
212 | ctxt->current = g_queue_pop_head(ctxt->buf); | |
213 | return ctxt->current; | |
214 | } | |
215 | ||
216 | static JSONToken *parser_context_peek_token(JSONParserContext *ctxt) | |
217 | { | |
218 | assert(!g_queue_is_empty(ctxt->buf)); | |
219 | return g_queue_peek_head(ctxt->buf); | |
220 | } | |
221 | ||
222 | static JSONParserContext *parser_context_new(GQueue *tokens) | |
223 | { | |
224 | JSONParserContext *ctxt; | |
225 | ||
226 | if (!tokens) { | |
227 | return NULL; | |
228 | } | |
229 | ||
230 | ctxt = g_malloc0(sizeof(JSONParserContext)); | |
231 | ctxt->buf = tokens; | |
232 | ||
233 | return ctxt; | |
234 | } | |
235 | ||
236 | /* to support error propagation, ctxt->err must be freed separately */ | |
237 | static void parser_context_free(JSONParserContext *ctxt) | |
238 | { | |
239 | if (ctxt) { | |
240 | while (!g_queue_is_empty(ctxt->buf)) { | |
241 | parser_context_pop_token(ctxt); | |
242 | } | |
243 | g_free(ctxt->current); | |
244 | g_queue_free(ctxt->buf); | |
245 | g_free(ctxt); | |
246 | } | |
247 | } | |
248 | ||
249 | /** | |
250 | * Parsing rules | |
251 | */ | |
252 | static int parse_pair(JSONParserContext *ctxt, QDict *dict, va_list *ap) | |
253 | { | |
254 | QObject *value; | |
255 | QString *key = NULL; | |
256 | JSONToken *peek, *token; | |
257 | ||
258 | peek = parser_context_peek_token(ctxt); | |
259 | if (peek == NULL) { | |
260 | parse_error(ctxt, NULL, "premature EOI"); | |
261 | goto out; | |
262 | } | |
263 | ||
264 | key = qobject_to(QString, parse_value(ctxt, ap)); | |
265 | if (!key) { | |
266 | parse_error(ctxt, peek, "key is not a string in object"); | |
267 | goto out; | |
268 | } | |
269 | ||
270 | token = parser_context_pop_token(ctxt); | |
271 | if (token == NULL) { | |
272 | parse_error(ctxt, NULL, "premature EOI"); | |
273 | goto out; | |
274 | } | |
275 | ||
276 | if (token->type != JSON_COLON) { | |
277 | parse_error(ctxt, token, "missing : in object pair"); | |
278 | goto out; | |
279 | } | |
280 | ||
281 | value = parse_value(ctxt, ap); | |
282 | if (value == NULL) { | |
283 | parse_error(ctxt, token, "Missing value in dict"); | |
284 | goto out; | |
285 | } | |
286 | ||
287 | qdict_put_obj(dict, qstring_get_str(key), value); | |
288 | ||
289 | qobject_unref(key); | |
290 | ||
291 | return 0; | |
292 | ||
293 | out: | |
294 | qobject_unref(key); | |
295 | ||
296 | return -1; | |
297 | } | |
298 | ||
299 | static QObject *parse_object(JSONParserContext *ctxt, va_list *ap) | |
300 | { | |
301 | QDict *dict = NULL; | |
302 | JSONToken *token, *peek; | |
303 | ||
304 | token = parser_context_pop_token(ctxt); | |
305 | assert(token && token->type == JSON_LCURLY); | |
306 | ||
307 | dict = qdict_new(); | |
308 | ||
309 | peek = parser_context_peek_token(ctxt); | |
310 | if (peek == NULL) { | |
311 | parse_error(ctxt, NULL, "premature EOI"); | |
312 | goto out; | |
313 | } | |
314 | ||
315 | if (peek->type != JSON_RCURLY) { | |
316 | if (parse_pair(ctxt, dict, ap) == -1) { | |
317 | goto out; | |
318 | } | |
319 | ||
320 | token = parser_context_pop_token(ctxt); | |
321 | if (token == NULL) { | |
322 | parse_error(ctxt, NULL, "premature EOI"); | |
323 | goto out; | |
324 | } | |
325 | ||
326 | while (token->type != JSON_RCURLY) { | |
327 | if (token->type != JSON_COMMA) { | |
328 | parse_error(ctxt, token, "expected separator in dict"); | |
329 | goto out; | |
330 | } | |
331 | ||
332 | if (parse_pair(ctxt, dict, ap) == -1) { | |
333 | goto out; | |
334 | } | |
335 | ||
336 | token = parser_context_pop_token(ctxt); | |
337 | if (token == NULL) { | |
338 | parse_error(ctxt, NULL, "premature EOI"); | |
339 | goto out; | |
340 | } | |
341 | } | |
342 | } else { | |
343 | (void)parser_context_pop_token(ctxt); | |
344 | } | |
345 | ||
346 | return QOBJECT(dict); | |
347 | ||
348 | out: | |
349 | qobject_unref(dict); | |
350 | return NULL; | |
351 | } | |
352 | ||
353 | static QObject *parse_array(JSONParserContext *ctxt, va_list *ap) | |
354 | { | |
355 | QList *list = NULL; | |
356 | JSONToken *token, *peek; | |
357 | ||
358 | token = parser_context_pop_token(ctxt); | |
359 | assert(token && token->type == JSON_LSQUARE); | |
360 | ||
361 | list = qlist_new(); | |
362 | ||
363 | peek = parser_context_peek_token(ctxt); | |
364 | if (peek == NULL) { | |
365 | parse_error(ctxt, NULL, "premature EOI"); | |
366 | goto out; | |
367 | } | |
368 | ||
369 | if (peek->type != JSON_RSQUARE) { | |
370 | QObject *obj; | |
371 | ||
372 | obj = parse_value(ctxt, ap); | |
373 | if (obj == NULL) { | |
374 | parse_error(ctxt, token, "expecting value"); | |
375 | goto out; | |
376 | } | |
377 | ||
378 | qlist_append_obj(list, obj); | |
379 | ||
380 | token = parser_context_pop_token(ctxt); | |
381 | if (token == NULL) { | |
382 | parse_error(ctxt, NULL, "premature EOI"); | |
383 | goto out; | |
384 | } | |
385 | ||
386 | while (token->type != JSON_RSQUARE) { | |
387 | if (token->type != JSON_COMMA) { | |
388 | parse_error(ctxt, token, "expected separator in list"); | |
389 | goto out; | |
390 | } | |
391 | ||
392 | obj = parse_value(ctxt, ap); | |
393 | if (obj == NULL) { | |
394 | parse_error(ctxt, token, "expecting value"); | |
395 | goto out; | |
396 | } | |
397 | ||
398 | qlist_append_obj(list, obj); | |
399 | ||
400 | token = parser_context_pop_token(ctxt); | |
401 | if (token == NULL) { | |
402 | parse_error(ctxt, NULL, "premature EOI"); | |
403 | goto out; | |
404 | } | |
405 | } | |
406 | } else { | |
407 | (void)parser_context_pop_token(ctxt); | |
408 | } | |
409 | ||
410 | return QOBJECT(list); | |
411 | ||
412 | out: | |
413 | qobject_unref(list); | |
414 | return NULL; | |
415 | } | |
416 | ||
417 | static QObject *parse_keyword(JSONParserContext *ctxt) | |
418 | { | |
419 | JSONToken *token; | |
420 | ||
421 | token = parser_context_pop_token(ctxt); | |
422 | assert(token && token->type == JSON_KEYWORD); | |
423 | ||
424 | if (!strcmp(token->str, "true")) { | |
425 | return QOBJECT(qbool_from_bool(true)); | |
426 | } else if (!strcmp(token->str, "false")) { | |
427 | return QOBJECT(qbool_from_bool(false)); | |
428 | } else if (!strcmp(token->str, "null")) { | |
429 | return QOBJECT(qnull()); | |
430 | } | |
431 | parse_error(ctxt, token, "invalid keyword '%s'", token->str); | |
432 | return NULL; | |
433 | } | |
434 | ||
435 | static QObject *parse_escape(JSONParserContext *ctxt, va_list *ap) | |
436 | { | |
437 | JSONToken *token; | |
438 | ||
439 | if (ap == NULL) { | |
440 | return NULL; | |
441 | } | |
442 | ||
443 | token = parser_context_pop_token(ctxt); | |
444 | assert(token && token->type == JSON_ESCAPE); | |
445 | ||
446 | if (!strcmp(token->str, "%p")) { | |
447 | return va_arg(*ap, QObject *); | |
448 | } else if (!strcmp(token->str, "%i")) { | |
449 | return QOBJECT(qbool_from_bool(va_arg(*ap, int))); | |
450 | } else if (!strcmp(token->str, "%d")) { | |
451 | return QOBJECT(qnum_from_int(va_arg(*ap, int))); | |
452 | } else if (!strcmp(token->str, "%ld")) { | |
453 | return QOBJECT(qnum_from_int(va_arg(*ap, long))); | |
454 | } else if (!strcmp(token->str, "%lld") || | |
455 | !strcmp(token->str, "%I64d")) { | |
456 | return QOBJECT(qnum_from_int(va_arg(*ap, long long))); | |
457 | } else if (!strcmp(token->str, "%u")) { | |
458 | return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned int))); | |
459 | } else if (!strcmp(token->str, "%lu")) { | |
460 | return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long))); | |
461 | } else if (!strcmp(token->str, "%llu") || | |
462 | !strcmp(token->str, "%I64u")) { | |
463 | return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long long))); | |
464 | } else if (!strcmp(token->str, "%s")) { | |
465 | return QOBJECT(qstring_from_str(va_arg(*ap, const char *))); | |
466 | } else if (!strcmp(token->str, "%f")) { | |
467 | return QOBJECT(qnum_from_double(va_arg(*ap, double))); | |
468 | } | |
469 | return NULL; | |
470 | } | |
471 | ||
472 | static QObject *parse_literal(JSONParserContext *ctxt) | |
473 | { | |
474 | JSONToken *token; | |
475 | ||
476 | token = parser_context_pop_token(ctxt); | |
477 | assert(token); | |
478 | ||
479 | switch (token->type) { | |
480 | case JSON_STRING: | |
481 | return QOBJECT(parse_string(ctxt, token)); | |
482 | case JSON_INTEGER: { | |
483 | /* | |
484 | * Represent JSON_INTEGER as QNUM_I64 if possible, else as | |
485 | * QNUM_U64, else as QNUM_DOUBLE. Note that qemu_strtoi64() | |
486 | * and qemu_strtou64() fail with ERANGE when it's not | |
487 | * possible. | |
488 | * | |
489 | * qnum_get_int() will then work for any signed 64-bit | |
490 | * JSON_INTEGER, qnum_get_uint() for any unsigned 64-bit | |
491 | * integer, and qnum_get_double() both for any JSON_INTEGER | |
492 | * and any JSON_FLOAT (with precision loss for integers beyond | |
493 | * 53 bits) | |
494 | */ | |
495 | int ret; | |
496 | int64_t value; | |
497 | uint64_t uvalue; | |
498 | ||
499 | ret = qemu_strtoi64(token->str, NULL, 10, &value); | |
500 | if (!ret) { | |
501 | return QOBJECT(qnum_from_int(value)); | |
502 | } | |
503 | assert(ret == -ERANGE); | |
504 | ||
505 | if (token->str[0] != '-') { | |
506 | ret = qemu_strtou64(token->str, NULL, 10, &uvalue); | |
507 | if (!ret) { | |
508 | return QOBJECT(qnum_from_uint(uvalue)); | |
509 | } | |
510 | assert(ret == -ERANGE); | |
511 | } | |
512 | /* fall through to JSON_FLOAT */ | |
513 | } | |
514 | case JSON_FLOAT: | |
515 | /* FIXME dependent on locale; a pervasive issue in QEMU */ | |
516 | /* FIXME our lexer matches RFC 7159 in forbidding Inf or NaN, | |
517 | * but those might be useful extensions beyond JSON */ | |
518 | return QOBJECT(qnum_from_double(strtod(token->str, NULL))); | |
519 | default: | |
520 | abort(); | |
521 | } | |
522 | } | |
523 | ||
524 | static QObject *parse_value(JSONParserContext *ctxt, va_list *ap) | |
525 | { | |
526 | JSONToken *token; | |
527 | ||
528 | token = parser_context_peek_token(ctxt); | |
529 | if (token == NULL) { | |
530 | parse_error(ctxt, NULL, "premature EOI"); | |
531 | return NULL; | |
532 | } | |
533 | ||
534 | switch (token->type) { | |
535 | case JSON_LCURLY: | |
536 | return parse_object(ctxt, ap); | |
537 | case JSON_LSQUARE: | |
538 | return parse_array(ctxt, ap); | |
539 | case JSON_ESCAPE: | |
540 | return parse_escape(ctxt, ap); | |
541 | case JSON_INTEGER: | |
542 | case JSON_FLOAT: | |
543 | case JSON_STRING: | |
544 | return parse_literal(ctxt); | |
545 | case JSON_KEYWORD: | |
546 | return parse_keyword(ctxt); | |
547 | default: | |
548 | parse_error(ctxt, token, "expecting value"); | |
549 | return NULL; | |
550 | } | |
551 | } | |
552 | ||
553 | QObject *json_parser_parse(GQueue *tokens, va_list *ap) | |
554 | { | |
555 | return json_parser_parse_err(tokens, ap, NULL); | |
556 | } | |
557 | ||
558 | QObject *json_parser_parse_err(GQueue *tokens, va_list *ap, Error **errp) | |
559 | { | |
560 | JSONParserContext *ctxt = parser_context_new(tokens); | |
561 | QObject *result; | |
562 | ||
563 | if (!ctxt) { | |
564 | return NULL; | |
565 | } | |
566 | ||
567 | result = parse_value(ctxt, ap); | |
568 | ||
569 | error_propagate(errp, ctxt->err); | |
570 | ||
571 | parser_context_free(ctxt); | |
572 | ||
573 | return result; | |
574 | } |