]> git.proxmox.com Git - mirror_qemu.git/blob - qobject/json-parser.c
qdict qlist: Make most helper macros functions
[mirror_qemu.git] / qobject / json-parser.c
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 "qemu/osdep.h"
15 #include "qemu/cutils.h"
16 #include "qapi/error.h"
17 #include "qemu-common.h"
18 #include "qapi/qmp/qbool.h"
19 #include "qapi/qmp/qnull.h"
20 #include "qapi/qmp/qnum.h"
21 #include "qapi/qmp/qstring.h"
22 #include "qapi/qmp/json-parser.h"
23 #include "qapi/qmp/json-lexer.h"
24 #include "qapi/qmp/json-streamer.h"
25
26 typedef struct JSONParserContext
27 {
28 Error *err;
29 JSONToken *current;
30 GQueue *buf;
31 } JSONParserContext;
32
33 #define BUG_ON(cond) assert(!(cond))
34
35 /**
36 * TODO
37 *
38 * 0) make errors meaningful again
39 * 1) add geometry information to tokens
40 * 3) should we return a parsed size?
41 * 4) deal with premature EOI
42 */
43
44 static QObject *parse_value(JSONParserContext *ctxt, va_list *ap);
45
46 /**
47 * Error handler
48 */
49 static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt,
50 JSONToken *token, const char *msg, ...)
51 {
52 va_list ap;
53 char message[1024];
54 va_start(ap, msg);
55 vsnprintf(message, sizeof(message), msg, ap);
56 va_end(ap);
57 if (ctxt->err) {
58 error_free(ctxt->err);
59 ctxt->err = NULL;
60 }
61 error_setg(&ctxt->err, "JSON parse error, %s", message);
62 }
63
64 /**
65 * String helpers
66 *
67 * These helpers are used to unescape strings.
68 */
69 static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length)
70 {
71 if (wchar <= 0x007F) {
72 BUG_ON(buffer_length < 2);
73
74 buffer[0] = wchar & 0x7F;
75 buffer[1] = 0;
76 } else if (wchar <= 0x07FF) {
77 BUG_ON(buffer_length < 3);
78
79 buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F);
80 buffer[1] = 0x80 | (wchar & 0x3F);
81 buffer[2] = 0;
82 } else {
83 BUG_ON(buffer_length < 4);
84
85 buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F);
86 buffer[1] = 0x80 | ((wchar >> 6) & 0x3F);
87 buffer[2] = 0x80 | (wchar & 0x3F);
88 buffer[3] = 0;
89 }
90 }
91
92 static int hex2decimal(char ch)
93 {
94 if (ch >= '0' && ch <= '9') {
95 return (ch - '0');
96 } else if (ch >= 'a' && ch <= 'f') {
97 return 10 + (ch - 'a');
98 } else if (ch >= 'A' && ch <= 'F') {
99 return 10 + (ch - 'A');
100 }
101
102 return -1;
103 }
104
105 /**
106 * parse_string(): Parse a json string and return a QObject
107 *
108 * string
109 * ""
110 * " chars "
111 * chars
112 * char
113 * char chars
114 * char
115 * any-Unicode-character-
116 * except-"-or-\-or-
117 * control-character
118 * \"
119 * \\
120 * \/
121 * \b
122 * \f
123 * \n
124 * \r
125 * \t
126 * \u four-hex-digits
127 */
128 static QString *qstring_from_escaped_str(JSONParserContext *ctxt,
129 JSONToken *token)
130 {
131 const char *ptr = token->str;
132 QString *str;
133 int double_quote = 1;
134
135 if (*ptr == '"') {
136 double_quote = 1;
137 } else {
138 double_quote = 0;
139 }
140 ptr++;
141
142 str = qstring_new();
143 while (*ptr &&
144 ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) {
145 if (*ptr == '\\') {
146 ptr++;
147
148 switch (*ptr) {
149 case '"':
150 qstring_append(str, "\"");
151 ptr++;
152 break;
153 case '\'':
154 qstring_append(str, "'");
155 ptr++;
156 break;
157 case '\\':
158 qstring_append(str, "\\");
159 ptr++;
160 break;
161 case '/':
162 qstring_append(str, "/");
163 ptr++;
164 break;
165 case 'b':
166 qstring_append(str, "\b");
167 ptr++;
168 break;
169 case 'f':
170 qstring_append(str, "\f");
171 ptr++;
172 break;
173 case 'n':
174 qstring_append(str, "\n");
175 ptr++;
176 break;
177 case 'r':
178 qstring_append(str, "\r");
179 ptr++;
180 break;
181 case 't':
182 qstring_append(str, "\t");
183 ptr++;
184 break;
185 case 'u': {
186 uint16_t unicode_char = 0;
187 char utf8_char[4];
188 int i = 0;
189
190 ptr++;
191
192 for (i = 0; i < 4; i++) {
193 if (qemu_isxdigit(*ptr)) {
194 unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4);
195 } else {
196 parse_error(ctxt, token,
197 "invalid hex escape sequence in string");
198 goto out;
199 }
200 ptr++;
201 }
202
203 wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
204 qstring_append(str, utf8_char);
205 } break;
206 default:
207 parse_error(ctxt, token, "invalid escape sequence in string");
208 goto out;
209 }
210 } else {
211 char dummy[2];
212
213 dummy[0] = *ptr++;
214 dummy[1] = 0;
215
216 qstring_append(str, dummy);
217 }
218 }
219
220 return str;
221
222 out:
223 QDECREF(str);
224 return NULL;
225 }
226
227 /* Note: the token object returned by parser_context_peek_token or
228 * parser_context_pop_token is deleted as soon as parser_context_pop_token
229 * is called again.
230 */
231 static JSONToken *parser_context_pop_token(JSONParserContext *ctxt)
232 {
233 g_free(ctxt->current);
234 assert(!g_queue_is_empty(ctxt->buf));
235 ctxt->current = g_queue_pop_head(ctxt->buf);
236 return ctxt->current;
237 }
238
239 static JSONToken *parser_context_peek_token(JSONParserContext *ctxt)
240 {
241 assert(!g_queue_is_empty(ctxt->buf));
242 return g_queue_peek_head(ctxt->buf);
243 }
244
245 static JSONParserContext *parser_context_new(GQueue *tokens)
246 {
247 JSONParserContext *ctxt;
248
249 if (!tokens) {
250 return NULL;
251 }
252
253 ctxt = g_malloc0(sizeof(JSONParserContext));
254 ctxt->buf = tokens;
255
256 return ctxt;
257 }
258
259 /* to support error propagation, ctxt->err must be freed separately */
260 static void parser_context_free(JSONParserContext *ctxt)
261 {
262 if (ctxt) {
263 while (!g_queue_is_empty(ctxt->buf)) {
264 parser_context_pop_token(ctxt);
265 }
266 g_free(ctxt->current);
267 g_queue_free(ctxt->buf);
268 g_free(ctxt);
269 }
270 }
271
272 /**
273 * Parsing rules
274 */
275 static int parse_pair(JSONParserContext *ctxt, QDict *dict, va_list *ap)
276 {
277 QObject *key = NULL, *value;
278 JSONToken *peek, *token;
279
280 peek = parser_context_peek_token(ctxt);
281 if (peek == NULL) {
282 parse_error(ctxt, NULL, "premature EOI");
283 goto out;
284 }
285
286 key = parse_value(ctxt, ap);
287 if (!key || qobject_type(key) != QTYPE_QSTRING) {
288 parse_error(ctxt, peek, "key is not a string in object");
289 goto out;
290 }
291
292 token = parser_context_pop_token(ctxt);
293 if (token == NULL) {
294 parse_error(ctxt, NULL, "premature EOI");
295 goto out;
296 }
297
298 if (token->type != JSON_COLON) {
299 parse_error(ctxt, token, "missing : in object pair");
300 goto out;
301 }
302
303 value = parse_value(ctxt, ap);
304 if (value == NULL) {
305 parse_error(ctxt, token, "Missing value in dict");
306 goto out;
307 }
308
309 qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
310
311 qobject_decref(key);
312
313 return 0;
314
315 out:
316 qobject_decref(key);
317
318 return -1;
319 }
320
321 static QObject *parse_object(JSONParserContext *ctxt, va_list *ap)
322 {
323 QDict *dict = NULL;
324 JSONToken *token, *peek;
325
326 token = parser_context_pop_token(ctxt);
327 assert(token && token->type == JSON_LCURLY);
328
329 dict = qdict_new();
330
331 peek = parser_context_peek_token(ctxt);
332 if (peek == NULL) {
333 parse_error(ctxt, NULL, "premature EOI");
334 goto out;
335 }
336
337 if (peek->type != JSON_RCURLY) {
338 if (parse_pair(ctxt, dict, ap) == -1) {
339 goto out;
340 }
341
342 token = parser_context_pop_token(ctxt);
343 if (token == NULL) {
344 parse_error(ctxt, NULL, "premature EOI");
345 goto out;
346 }
347
348 while (token->type != JSON_RCURLY) {
349 if (token->type != JSON_COMMA) {
350 parse_error(ctxt, token, "expected separator in dict");
351 goto out;
352 }
353
354 if (parse_pair(ctxt, dict, ap) == -1) {
355 goto out;
356 }
357
358 token = parser_context_pop_token(ctxt);
359 if (token == NULL) {
360 parse_error(ctxt, NULL, "premature EOI");
361 goto out;
362 }
363 }
364 } else {
365 (void)parser_context_pop_token(ctxt);
366 }
367
368 return QOBJECT(dict);
369
370 out:
371 QDECREF(dict);
372 return NULL;
373 }
374
375 static QObject *parse_array(JSONParserContext *ctxt, va_list *ap)
376 {
377 QList *list = NULL;
378 JSONToken *token, *peek;
379
380 token = parser_context_pop_token(ctxt);
381 assert(token && token->type == JSON_LSQUARE);
382
383 list = qlist_new();
384
385 peek = parser_context_peek_token(ctxt);
386 if (peek == NULL) {
387 parse_error(ctxt, NULL, "premature EOI");
388 goto out;
389 }
390
391 if (peek->type != JSON_RSQUARE) {
392 QObject *obj;
393
394 obj = parse_value(ctxt, ap);
395 if (obj == NULL) {
396 parse_error(ctxt, token, "expecting value");
397 goto out;
398 }
399
400 qlist_append_obj(list, obj);
401
402 token = parser_context_pop_token(ctxt);
403 if (token == NULL) {
404 parse_error(ctxt, NULL, "premature EOI");
405 goto out;
406 }
407
408 while (token->type != JSON_RSQUARE) {
409 if (token->type != JSON_COMMA) {
410 parse_error(ctxt, token, "expected separator in list");
411 goto out;
412 }
413
414 obj = parse_value(ctxt, ap);
415 if (obj == NULL) {
416 parse_error(ctxt, token, "expecting value");
417 goto out;
418 }
419
420 qlist_append_obj(list, obj);
421
422 token = parser_context_pop_token(ctxt);
423 if (token == NULL) {
424 parse_error(ctxt, NULL, "premature EOI");
425 goto out;
426 }
427 }
428 } else {
429 (void)parser_context_pop_token(ctxt);
430 }
431
432 return QOBJECT(list);
433
434 out:
435 QDECREF(list);
436 return NULL;
437 }
438
439 static QObject *parse_keyword(JSONParserContext *ctxt)
440 {
441 JSONToken *token;
442
443 token = parser_context_pop_token(ctxt);
444 assert(token && token->type == JSON_KEYWORD);
445
446 if (!strcmp(token->str, "true")) {
447 return QOBJECT(qbool_from_bool(true));
448 } else if (!strcmp(token->str, "false")) {
449 return QOBJECT(qbool_from_bool(false));
450 } else if (!strcmp(token->str, "null")) {
451 return QOBJECT(qnull());
452 }
453 parse_error(ctxt, token, "invalid keyword '%s'", token->str);
454 return NULL;
455 }
456
457 static QObject *parse_escape(JSONParserContext *ctxt, va_list *ap)
458 {
459 JSONToken *token;
460
461 if (ap == NULL) {
462 return NULL;
463 }
464
465 token = parser_context_pop_token(ctxt);
466 assert(token && token->type == JSON_ESCAPE);
467
468 if (!strcmp(token->str, "%p")) {
469 return va_arg(*ap, QObject *);
470 } else if (!strcmp(token->str, "%i")) {
471 return QOBJECT(qbool_from_bool(va_arg(*ap, int)));
472 } else if (!strcmp(token->str, "%d")) {
473 return QOBJECT(qnum_from_int(va_arg(*ap, int)));
474 } else if (!strcmp(token->str, "%ld")) {
475 return QOBJECT(qnum_from_int(va_arg(*ap, long)));
476 } else if (!strcmp(token->str, "%lld") ||
477 !strcmp(token->str, "%I64d")) {
478 return QOBJECT(qnum_from_int(va_arg(*ap, long long)));
479 } else if (!strcmp(token->str, "%u")) {
480 return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned int)));
481 } else if (!strcmp(token->str, "%lu")) {
482 return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long)));
483 } else if (!strcmp(token->str, "%llu") ||
484 !strcmp(token->str, "%I64u")) {
485 return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long long)));
486 } else if (!strcmp(token->str, "%s")) {
487 return QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
488 } else if (!strcmp(token->str, "%f")) {
489 return QOBJECT(qnum_from_double(va_arg(*ap, double)));
490 }
491 return NULL;
492 }
493
494 static QObject *parse_literal(JSONParserContext *ctxt)
495 {
496 JSONToken *token;
497
498 token = parser_context_pop_token(ctxt);
499 assert(token);
500
501 switch (token->type) {
502 case JSON_STRING:
503 return QOBJECT(qstring_from_escaped_str(ctxt, token));
504 case JSON_INTEGER: {
505 /*
506 * Represent JSON_INTEGER as QNUM_I64 if possible, else as
507 * QNUM_U64, else as QNUM_DOUBLE. Note that qemu_strtoi64()
508 * and qemu_strtou64() fail with ERANGE when it's not
509 * possible.
510 *
511 * qnum_get_int() will then work for any signed 64-bit
512 * JSON_INTEGER, qnum_get_uint() for any unsigned 64-bit
513 * integer, and qnum_get_double() both for any JSON_INTEGER
514 * and any JSON_FLOAT (with precision loss for integers beyond
515 * 53 bits)
516 */
517 int ret;
518 int64_t value;
519 uint64_t uvalue;
520
521 ret = qemu_strtoi64(token->str, NULL, 10, &value);
522 if (!ret) {
523 return QOBJECT(qnum_from_int(value));
524 }
525 assert(ret == -ERANGE);
526
527 if (token->str[0] != '-') {
528 ret = qemu_strtou64(token->str, NULL, 10, &uvalue);
529 if (!ret) {
530 return QOBJECT(qnum_from_uint(uvalue));
531 }
532 assert(ret == -ERANGE);
533 }
534 /* fall through to JSON_FLOAT */
535 }
536 case JSON_FLOAT:
537 /* FIXME dependent on locale; a pervasive issue in QEMU */
538 /* FIXME our lexer matches RFC 7159 in forbidding Inf or NaN,
539 * but those might be useful extensions beyond JSON */
540 return QOBJECT(qnum_from_double(strtod(token->str, NULL)));
541 default:
542 abort();
543 }
544 }
545
546 static QObject *parse_value(JSONParserContext *ctxt, va_list *ap)
547 {
548 JSONToken *token;
549
550 token = parser_context_peek_token(ctxt);
551 if (token == NULL) {
552 parse_error(ctxt, NULL, "premature EOI");
553 return NULL;
554 }
555
556 switch (token->type) {
557 case JSON_LCURLY:
558 return parse_object(ctxt, ap);
559 case JSON_LSQUARE:
560 return parse_array(ctxt, ap);
561 case JSON_ESCAPE:
562 return parse_escape(ctxt, ap);
563 case JSON_INTEGER:
564 case JSON_FLOAT:
565 case JSON_STRING:
566 return parse_literal(ctxt);
567 case JSON_KEYWORD:
568 return parse_keyword(ctxt);
569 default:
570 parse_error(ctxt, token, "expecting value");
571 return NULL;
572 }
573 }
574
575 QObject *json_parser_parse(GQueue *tokens, va_list *ap)
576 {
577 return json_parser_parse_err(tokens, ap, NULL);
578 }
579
580 QObject *json_parser_parse_err(GQueue *tokens, va_list *ap, Error **errp)
581 {
582 JSONParserContext *ctxt = parser_context_new(tokens);
583 QObject *result;
584
585 if (!ctxt) {
586 return NULL;
587 }
588
589 result = parse_value(ctxt, ap);
590
591 error_propagate(errp, ctxt->err);
592
593 parser_context_free(ctxt);
594
595 return result;
596 }