]> git.proxmox.com Git - mirror_qemu.git/blame - json-parser.c
json-streamer: make sure to reset token_size after emitting a token list
[mirror_qemu.git] / json-parser.c
CommitLineData
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
c96c84a9 14#include <stdarg.h>
4a5fcab7
AL
15
16#include "qemu-common.h"
17#include "qstring.h"
18#include "qint.h"
19#include "qdict.h"
20#include "qlist.h"
21#include "qfloat.h"
22#include "qbool.h"
23#include "json-parser.h"
24#include "json-lexer.h"
ef749d07 25#include "qerror.h"
4a5fcab7
AL
26
27typedef struct JSONParserContext
28{
ef749d07 29 Error *err;
4a5fcab7
AL
30} JSONParserContext;
31
32#define BUG_ON(cond) assert(!(cond))
33
34/**
35 * TODO
36 *
37 * 0) make errors meaningful again
38 * 1) add geometry information to tokens
39 * 3) should we return a parsed size?
40 * 4) deal with premature EOI
41 */
42
43static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap);
44
45/**
46 * Token manipulators
47 *
48 * tokens are dictionaries that contain a type, a string value, and geometry information
49 * about a token identified by the lexer. These are routines that make working with
50 * these objects a bit easier.
51 */
52static const char *token_get_value(QObject *obj)
53{
54 return qdict_get_str(qobject_to_qdict(obj), "token");
55}
56
57static JSONTokenType token_get_type(QObject *obj)
58{
59 return qdict_get_int(qobject_to_qdict(obj), "type");
60}
61
62static int token_is_operator(QObject *obj, char op)
63{
64 const char *val;
65
66 if (token_get_type(obj) != JSON_OPERATOR) {
67 return 0;
68 }
69
70 val = token_get_value(obj);
71
72 return (val[0] == op) && (val[1] == 0);
73}
74
75static int token_is_keyword(QObject *obj, const char *value)
76{
77 if (token_get_type(obj) != JSON_KEYWORD) {
78 return 0;
79 }
80
81 return strcmp(token_get_value(obj), value) == 0;
82}
83
84static int token_is_escape(QObject *obj, const char *value)
85{
86 if (token_get_type(obj) != JSON_ESCAPE) {
87 return 0;
88 }
89
90 return (strcmp(token_get_value(obj), value) == 0);
91}
92
93/**
94 * Error handler
95 */
8b7968f7
SW
96static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt,
97 QObject *token, const char *msg, ...)
4a5fcab7 98{
c96c84a9 99 va_list ap;
ef749d07 100 char message[1024];
c96c84a9 101 va_start(ap, msg);
ef749d07 102 vsnprintf(message, sizeof(message), msg, ap);
c96c84a9 103 va_end(ap);
ef749d07
AL
104 if (ctxt->err) {
105 error_free(ctxt->err);
106 ctxt->err = NULL;
107 }
108 error_set(&ctxt->err, QERR_JSON_PARSE_ERROR, message);
4a5fcab7
AL
109}
110
111/**
112 * String helpers
113 *
114 * These helpers are used to unescape strings.
115 */
116static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length)
117{
118 if (wchar <= 0x007F) {
119 BUG_ON(buffer_length < 2);
120
121 buffer[0] = wchar & 0x7F;
122 buffer[1] = 0;
123 } else if (wchar <= 0x07FF) {
124 BUG_ON(buffer_length < 3);
125
126 buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F);
127 buffer[1] = 0x80 | (wchar & 0x3F);
128 buffer[2] = 0;
129 } else {
130 BUG_ON(buffer_length < 4);
131
132 buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F);
133 buffer[1] = 0x80 | ((wchar >> 6) & 0x3F);
134 buffer[2] = 0x80 | (wchar & 0x3F);
135 buffer[3] = 0;
136 }
137}
138
139static int hex2decimal(char ch)
140{
141 if (ch >= '0' && ch <= '9') {
142 return (ch - '0');
143 } else if (ch >= 'a' && ch <= 'f') {
144 return 10 + (ch - 'a');
145 } else if (ch >= 'A' && ch <= 'F') {
146 return 10 + (ch - 'A');
147 }
148
149 return -1;
150}
151
152/**
153 * parse_string(): Parse a json string and return a QObject
154 *
155 * string
156 * ""
157 * " chars "
158 * chars
159 * char
160 * char chars
161 * char
162 * any-Unicode-character-
163 * except-"-or-\-or-
164 * control-character
165 * \"
166 * \\
167 * \/
168 * \b
169 * \f
170 * \n
171 * \r
172 * \t
173 * \u four-hex-digits
174 */
175static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token)
176{
177 const char *ptr = token_get_value(token);
178 QString *str;
179 int double_quote = 1;
180
181 if (*ptr == '"') {
182 double_quote = 1;
183 } else {
184 double_quote = 0;
185 }
186 ptr++;
187
188 str = qstring_new();
189 while (*ptr &&
190 ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) {
191 if (*ptr == '\\') {
192 ptr++;
193
194 switch (*ptr) {
195 case '"':
196 qstring_append(str, "\"");
197 ptr++;
198 break;
199 case '\'':
200 qstring_append(str, "'");
201 ptr++;
202 break;
203 case '\\':
204 qstring_append(str, "\\");
205 ptr++;
206 break;
207 case '/':
208 qstring_append(str, "/");
209 ptr++;
210 break;
211 case 'b':
212 qstring_append(str, "\b");
213 ptr++;
214 break;
bd032695
LC
215 case 'f':
216 qstring_append(str, "\f");
217 ptr++;
218 break;
4a5fcab7
AL
219 case 'n':
220 qstring_append(str, "\n");
221 ptr++;
222 break;
223 case 'r':
224 qstring_append(str, "\r");
225 ptr++;
226 break;
227 case 't':
228 qstring_append(str, "\t");
229 ptr++;
230 break;
231 case 'u': {
232 uint16_t unicode_char = 0;
233 char utf8_char[4];
234 int i = 0;
235
236 ptr++;
237
238 for (i = 0; i < 4; i++) {
239 if (qemu_isxdigit(*ptr)) {
240 unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4);
241 } else {
242 parse_error(ctxt, token,
243 "invalid hex escape sequence in string");
244 goto out;
245 }
246 ptr++;
247 }
248
249 wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
250 qstring_append(str, utf8_char);
251 } break;
252 default:
253 parse_error(ctxt, token, "invalid escape sequence in string");
254 goto out;
255 }
256 } else {
257 char dummy[2];
258
259 dummy[0] = *ptr++;
260 dummy[1] = 0;
261
262 qstring_append(str, dummy);
263 }
264 }
265
4a5fcab7
AL
266 return str;
267
268out:
269 QDECREF(str);
270 return NULL;
271}
272
273/**
274 * Parsing rules
275 */
276static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap)
277{
278 QObject *key, *token = NULL, *value, *peek;
279 QList *working = qlist_copy(*tokens);
280
281 peek = qlist_peek(working);
282 key = parse_value(ctxt, &working, ap);
d758d90f 283 if (!key || qobject_type(key) != QTYPE_QSTRING) {
4a5fcab7
AL
284 parse_error(ctxt, peek, "key is not a string in object");
285 goto out;
286 }
287
288 token = qlist_pop(working);
289 if (!token_is_operator(token, ':')) {
290 parse_error(ctxt, token, "missing : in object pair");
291 goto out;
292 }
293
294 value = parse_value(ctxt, &working, ap);
295 if (value == NULL) {
296 parse_error(ctxt, token, "Missing value in dict");
297 goto out;
298 }
299
300 qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
301
302 qobject_decref(token);
303 qobject_decref(key);
304 QDECREF(*tokens);
305 *tokens = working;
306
307 return 0;
308
309out:
310 qobject_decref(token);
311 qobject_decref(key);
312 QDECREF(working);
313
314 return -1;
315}
316
317static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
318{
319 QDict *dict = NULL;
320 QObject *token, *peek;
321 QList *working = qlist_copy(*tokens);
322
323 token = qlist_pop(working);
324 if (!token_is_operator(token, '{')) {
325 goto out;
326 }
327 qobject_decref(token);
328 token = NULL;
329
330 dict = qdict_new();
331
332 peek = qlist_peek(working);
333 if (!token_is_operator(peek, '}')) {
334 if (parse_pair(ctxt, dict, &working, ap) == -1) {
335 goto out;
336 }
337
338 token = qlist_pop(working);
339 while (!token_is_operator(token, '}')) {
340 if (!token_is_operator(token, ',')) {
341 parse_error(ctxt, token, "expected separator in dict");
342 goto out;
343 }
344 qobject_decref(token);
345 token = NULL;
346
347 if (parse_pair(ctxt, dict, &working, ap) == -1) {
348 goto out;
349 }
350
351 token = qlist_pop(working);
352 }
353 qobject_decref(token);
354 token = NULL;
355 } else {
356 token = qlist_pop(working);
357 qobject_decref(token);
358 token = NULL;
359 }
360
361 QDECREF(*tokens);
362 *tokens = working;
363
364 return QOBJECT(dict);
365
366out:
367 qobject_decref(token);
368 QDECREF(working);
369 QDECREF(dict);
370 return NULL;
371}
372
373static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
374{
375 QList *list = NULL;
376 QObject *token, *peek;
377 QList *working = qlist_copy(*tokens);
378
379 token = qlist_pop(working);
380 if (!token_is_operator(token, '[')) {
381 goto out;
382 }
383 qobject_decref(token);
384 token = NULL;
385
386 list = qlist_new();
387
388 peek = qlist_peek(working);
389 if (!token_is_operator(peek, ']')) {
390 QObject *obj;
391
392 obj = parse_value(ctxt, &working, 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 = qlist_pop(working);
401 while (!token_is_operator(token, ']')) {
402 if (!token_is_operator(token, ',')) {
403 parse_error(ctxt, token, "expected separator in list");
404 goto out;
405 }
406
407 qobject_decref(token);
408 token = NULL;
409
410 obj = parse_value(ctxt, &working, ap);
411 if (obj == NULL) {
412 parse_error(ctxt, token, "expecting value");
413 goto out;
414 }
415
416 qlist_append_obj(list, obj);
417
418 token = qlist_pop(working);
419 }
420
421 qobject_decref(token);
422 token = NULL;
423 } else {
424 token = qlist_pop(working);
425 qobject_decref(token);
426 token = NULL;
427 }
428
429 QDECREF(*tokens);
430 *tokens = working;
431
432 return QOBJECT(list);
433
434out:
435 qobject_decref(token);
436 QDECREF(working);
437 QDECREF(list);
438 return NULL;
439}
440
441static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
442{
443 QObject *token, *ret;
444 QList *working = qlist_copy(*tokens);
445
446 token = qlist_pop(working);
447
448 if (token_get_type(token) != JSON_KEYWORD) {
449 goto out;
450 }
451
452 if (token_is_keyword(token, "true")) {
453 ret = QOBJECT(qbool_from_int(true));
454 } else if (token_is_keyword(token, "false")) {
455 ret = QOBJECT(qbool_from_int(false));
456 } else {
457 parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
458 goto out;
459 }
460
461 qobject_decref(token);
462 QDECREF(*tokens);
463 *tokens = working;
464
465 return ret;
466
467out:
468 qobject_decref(token);
469 QDECREF(working);
470
471 return NULL;
472}
473
474static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
475{
476 QObject *token = NULL, *obj;
477 QList *working = qlist_copy(*tokens);
478
479 if (ap == NULL) {
480 goto out;
481 }
482
483 token = qlist_pop(working);
484
485 if (token_is_escape(token, "%p")) {
486 obj = va_arg(*ap, QObject *);
487 } else if (token_is_escape(token, "%i")) {
488 obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
489 } else if (token_is_escape(token, "%d")) {
490 obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
491 } else if (token_is_escape(token, "%ld")) {
492 obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
2c0d4b36
RT
493 } else if (token_is_escape(token, "%lld") ||
494 token_is_escape(token, "%I64d")) {
4a5fcab7
AL
495 obj = QOBJECT(qint_from_int(va_arg(*ap, long long)));
496 } else if (token_is_escape(token, "%s")) {
497 obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
498 } else if (token_is_escape(token, "%f")) {
499 obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
500 } else {
501 goto out;
502 }
503
504 qobject_decref(token);
505 QDECREF(*tokens);
506 *tokens = working;
507
508 return obj;
509
510out:
511 qobject_decref(token);
512 QDECREF(working);
513
514 return NULL;
515}
516
517static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
518{
519 QObject *token, *obj;
520 QList *working = qlist_copy(*tokens);
521
522 token = qlist_pop(working);
523 switch (token_get_type(token)) {
524 case JSON_STRING:
525 obj = QOBJECT(qstring_from_escaped_str(ctxt, token));
526 break;
527 case JSON_INTEGER:
528 obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10)));
529 break;
530 case JSON_FLOAT:
531 /* FIXME dependent on locale */
532 obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
533 break;
534 default:
535 goto out;
536 }
537
538 qobject_decref(token);
539 QDECREF(*tokens);
540 *tokens = working;
541
542 return obj;
543
544out:
545 qobject_decref(token);
546 QDECREF(working);
547
548 return NULL;
549}
550
551static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
552{
553 QObject *obj;
554
555 obj = parse_object(ctxt, tokens, ap);
556 if (obj == NULL) {
557 obj = parse_array(ctxt, tokens, ap);
558 }
559 if (obj == NULL) {
560 obj = parse_escape(ctxt, tokens, ap);
561 }
562 if (obj == NULL) {
563 obj = parse_keyword(ctxt, tokens);
564 }
565 if (obj == NULL) {
566 obj = parse_literal(ctxt, tokens);
567 }
568
569 return obj;
570}
571
572QObject *json_parser_parse(QList *tokens, va_list *ap)
ef749d07
AL
573{
574 return json_parser_parse_err(tokens, ap, NULL);
575}
576
577QObject *json_parser_parse_err(QList *tokens, va_list *ap, Error **errp)
4a5fcab7
AL
578{
579 JSONParserContext ctxt = {};
580 QList *working = qlist_copy(tokens);
581 QObject *result;
582
583 result = parse_value(&ctxt, &working, ap);
584
585 QDECREF(working);
586
ef749d07
AL
587 error_propagate(errp, ctxt.err);
588
4a5fcab7
AL
589 return result;
590}