root / json-parser.c @ 11165820
History | View | Annotate | Download (12.9 kB)
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 <stdarg.h> |
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" |
25 |
|
26 |
typedef struct JSONParserContext |
27 |
{ |
28 |
} JSONParserContext; |
29 |
|
30 |
#define BUG_ON(cond) assert(!(cond))
|
31 |
|
32 |
/**
|
33 |
* TODO
|
34 |
*
|
35 |
* 0) make errors meaningful again
|
36 |
* 1) add geometry information to tokens
|
37 |
* 3) should we return a parsed size?
|
38 |
* 4) deal with premature EOI
|
39 |
*/
|
40 |
|
41 |
static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap);
|
42 |
|
43 |
/**
|
44 |
* Token manipulators
|
45 |
*
|
46 |
* tokens are dictionaries that contain a type, a string value, and geometry information
|
47 |
* about a token identified by the lexer. These are routines that make working with
|
48 |
* these objects a bit easier.
|
49 |
*/
|
50 |
static const char *token_get_value(QObject *obj) |
51 |
{ |
52 |
return qdict_get_str(qobject_to_qdict(obj), "token"); |
53 |
} |
54 |
|
55 |
static JSONTokenType token_get_type(QObject *obj)
|
56 |
{ |
57 |
return qdict_get_int(qobject_to_qdict(obj), "type"); |
58 |
} |
59 |
|
60 |
static int token_is_operator(QObject *obj, char op) |
61 |
{ |
62 |
const char *val; |
63 |
|
64 |
if (token_get_type(obj) != JSON_OPERATOR) {
|
65 |
return 0; |
66 |
} |
67 |
|
68 |
val = token_get_value(obj); |
69 |
|
70 |
return (val[0] == op) && (val[1] == 0); |
71 |
} |
72 |
|
73 |
static int token_is_keyword(QObject *obj, const char *value) |
74 |
{ |
75 |
if (token_get_type(obj) != JSON_KEYWORD) {
|
76 |
return 0; |
77 |
} |
78 |
|
79 |
return strcmp(token_get_value(obj), value) == 0; |
80 |
} |
81 |
|
82 |
static int token_is_escape(QObject *obj, const char *value) |
83 |
{ |
84 |
if (token_get_type(obj) != JSON_ESCAPE) {
|
85 |
return 0; |
86 |
} |
87 |
|
88 |
return (strcmp(token_get_value(obj), value) == 0); |
89 |
} |
90 |
|
91 |
/**
|
92 |
* Error handler
|
93 |
*/
|
94 |
static void parse_error(JSONParserContext *ctxt, QObject *token, const char *msg, ...) |
95 |
{ |
96 |
va_list ap; |
97 |
va_start(ap, msg); |
98 |
fprintf(stderr, "parse error: ");
|
99 |
vfprintf(stderr, msg, ap); |
100 |
fprintf(stderr, "\n");
|
101 |
va_end(ap); |
102 |
} |
103 |
|
104 |
/**
|
105 |
* String helpers
|
106 |
*
|
107 |
* These helpers are used to unescape strings.
|
108 |
*/
|
109 |
static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length) |
110 |
{ |
111 |
if (wchar <= 0x007F) { |
112 |
BUG_ON(buffer_length < 2);
|
113 |
|
114 |
buffer[0] = wchar & 0x7F; |
115 |
buffer[1] = 0; |
116 |
} else if (wchar <= 0x07FF) { |
117 |
BUG_ON(buffer_length < 3);
|
118 |
|
119 |
buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F); |
120 |
buffer[1] = 0x80 | (wchar & 0x3F); |
121 |
buffer[2] = 0; |
122 |
} else {
|
123 |
BUG_ON(buffer_length < 4);
|
124 |
|
125 |
buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F); |
126 |
buffer[1] = 0x80 | ((wchar >> 6) & 0x3F); |
127 |
buffer[2] = 0x80 | (wchar & 0x3F); |
128 |
buffer[3] = 0; |
129 |
} |
130 |
} |
131 |
|
132 |
static int hex2decimal(char ch) |
133 |
{ |
134 |
if (ch >= '0' && ch <= '9') { |
135 |
return (ch - '0'); |
136 |
} else if (ch >= 'a' && ch <= 'f') { |
137 |
return 10 + (ch - 'a'); |
138 |
} else if (ch >= 'A' && ch <= 'F') { |
139 |
return 10 + (ch - 'A'); |
140 |
} |
141 |
|
142 |
return -1; |
143 |
} |
144 |
|
145 |
/**
|
146 |
* parse_string(): Parse a json string and return a QObject
|
147 |
*
|
148 |
* string
|
149 |
* ""
|
150 |
* " chars "
|
151 |
* chars
|
152 |
* char
|
153 |
* char chars
|
154 |
* char
|
155 |
* any-Unicode-character-
|
156 |
* except-"-or-\-or-
|
157 |
* control-character
|
158 |
* \"
|
159 |
* \\
|
160 |
* \/
|
161 |
* \b
|
162 |
* \f
|
163 |
* \n
|
164 |
* \r
|
165 |
* \t
|
166 |
* \u four-hex-digits
|
167 |
*/
|
168 |
static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token)
|
169 |
{ |
170 |
const char *ptr = token_get_value(token); |
171 |
QString *str; |
172 |
int double_quote = 1; |
173 |
|
174 |
if (*ptr == '"') { |
175 |
double_quote = 1;
|
176 |
} else {
|
177 |
double_quote = 0;
|
178 |
} |
179 |
ptr++; |
180 |
|
181 |
str = qstring_new(); |
182 |
while (*ptr &&
|
183 |
((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) { |
184 |
if (*ptr == '\\') { |
185 |
ptr++; |
186 |
|
187 |
switch (*ptr) {
|
188 |
case '"': |
189 |
qstring_append(str, "\"");
|
190 |
ptr++; |
191 |
break;
|
192 |
case '\'': |
193 |
qstring_append(str, "'");
|
194 |
ptr++; |
195 |
break;
|
196 |
case '\\': |
197 |
qstring_append(str, "\\");
|
198 |
ptr++; |
199 |
break;
|
200 |
case '/': |
201 |
qstring_append(str, "/");
|
202 |
ptr++; |
203 |
break;
|
204 |
case 'b': |
205 |
qstring_append(str, "\b");
|
206 |
ptr++; |
207 |
break;
|
208 |
case 'n': |
209 |
qstring_append(str, "\n");
|
210 |
ptr++; |
211 |
break;
|
212 |
case 'r': |
213 |
qstring_append(str, "\r");
|
214 |
ptr++; |
215 |
break;
|
216 |
case 't': |
217 |
qstring_append(str, "\t");
|
218 |
ptr++; |
219 |
break;
|
220 |
case 'u': { |
221 |
uint16_t unicode_char = 0;
|
222 |
char utf8_char[4]; |
223 |
int i = 0; |
224 |
|
225 |
ptr++; |
226 |
|
227 |
for (i = 0; i < 4; i++) { |
228 |
if (qemu_isxdigit(*ptr)) {
|
229 |
unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4); |
230 |
} else {
|
231 |
parse_error(ctxt, token, |
232 |
"invalid hex escape sequence in string");
|
233 |
goto out;
|
234 |
} |
235 |
ptr++; |
236 |
} |
237 |
|
238 |
wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
|
239 |
qstring_append(str, utf8_char); |
240 |
} break;
|
241 |
default:
|
242 |
parse_error(ctxt, token, "invalid escape sequence in string");
|
243 |
goto out;
|
244 |
} |
245 |
} else {
|
246 |
char dummy[2]; |
247 |
|
248 |
dummy[0] = *ptr++;
|
249 |
dummy[1] = 0; |
250 |
|
251 |
qstring_append(str, dummy); |
252 |
} |
253 |
} |
254 |
|
255 |
return str;
|
256 |
|
257 |
out:
|
258 |
QDECREF(str); |
259 |
return NULL; |
260 |
} |
261 |
|
262 |
/**
|
263 |
* Parsing rules
|
264 |
*/
|
265 |
static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap) |
266 |
{ |
267 |
QObject *key, *token = NULL, *value, *peek;
|
268 |
QList *working = qlist_copy(*tokens); |
269 |
|
270 |
peek = qlist_peek(working); |
271 |
key = parse_value(ctxt, &working, ap); |
272 |
if (!key || qobject_type(key) != QTYPE_QSTRING) {
|
273 |
parse_error(ctxt, peek, "key is not a string in object");
|
274 |
goto out;
|
275 |
} |
276 |
|
277 |
token = qlist_pop(working); |
278 |
if (!token_is_operator(token, ':')) { |
279 |
parse_error(ctxt, token, "missing : in object pair");
|
280 |
goto out;
|
281 |
} |
282 |
|
283 |
value = parse_value(ctxt, &working, ap); |
284 |
if (value == NULL) { |
285 |
parse_error(ctxt, token, "Missing value in dict");
|
286 |
goto out;
|
287 |
} |
288 |
|
289 |
qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value); |
290 |
|
291 |
qobject_decref(token); |
292 |
qobject_decref(key); |
293 |
QDECREF(*tokens); |
294 |
*tokens = working; |
295 |
|
296 |
return 0; |
297 |
|
298 |
out:
|
299 |
qobject_decref(token); |
300 |
qobject_decref(key); |
301 |
QDECREF(working); |
302 |
|
303 |
return -1; |
304 |
} |
305 |
|
306 |
static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
|
307 |
{ |
308 |
QDict *dict = NULL;
|
309 |
QObject *token, *peek; |
310 |
QList *working = qlist_copy(*tokens); |
311 |
|
312 |
token = qlist_pop(working); |
313 |
if (!token_is_operator(token, '{')) { |
314 |
goto out;
|
315 |
} |
316 |
qobject_decref(token); |
317 |
token = NULL;
|
318 |
|
319 |
dict = qdict_new(); |
320 |
|
321 |
peek = qlist_peek(working); |
322 |
if (!token_is_operator(peek, '}')) { |
323 |
if (parse_pair(ctxt, dict, &working, ap) == -1) { |
324 |
goto out;
|
325 |
} |
326 |
|
327 |
token = qlist_pop(working); |
328 |
while (!token_is_operator(token, '}')) { |
329 |
if (!token_is_operator(token, ',')) { |
330 |
parse_error(ctxt, token, "expected separator in dict");
|
331 |
goto out;
|
332 |
} |
333 |
qobject_decref(token); |
334 |
token = NULL;
|
335 |
|
336 |
if (parse_pair(ctxt, dict, &working, ap) == -1) { |
337 |
goto out;
|
338 |
} |
339 |
|
340 |
token = qlist_pop(working); |
341 |
} |
342 |
qobject_decref(token); |
343 |
token = NULL;
|
344 |
} else {
|
345 |
token = qlist_pop(working); |
346 |
qobject_decref(token); |
347 |
token = NULL;
|
348 |
} |
349 |
|
350 |
QDECREF(*tokens); |
351 |
*tokens = working; |
352 |
|
353 |
return QOBJECT(dict);
|
354 |
|
355 |
out:
|
356 |
qobject_decref(token); |
357 |
QDECREF(working); |
358 |
QDECREF(dict); |
359 |
return NULL; |
360 |
} |
361 |
|
362 |
static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
|
363 |
{ |
364 |
QList *list = NULL;
|
365 |
QObject *token, *peek; |
366 |
QList *working = qlist_copy(*tokens); |
367 |
|
368 |
token = qlist_pop(working); |
369 |
if (!token_is_operator(token, '[')) { |
370 |
goto out;
|
371 |
} |
372 |
qobject_decref(token); |
373 |
token = NULL;
|
374 |
|
375 |
list = qlist_new(); |
376 |
|
377 |
peek = qlist_peek(working); |
378 |
if (!token_is_operator(peek, ']')) { |
379 |
QObject *obj; |
380 |
|
381 |
obj = parse_value(ctxt, &working, ap); |
382 |
if (obj == NULL) { |
383 |
parse_error(ctxt, token, "expecting value");
|
384 |
goto out;
|
385 |
} |
386 |
|
387 |
qlist_append_obj(list, obj); |
388 |
|
389 |
token = qlist_pop(working); |
390 |
while (!token_is_operator(token, ']')) { |
391 |
if (!token_is_operator(token, ',')) { |
392 |
parse_error(ctxt, token, "expected separator in list");
|
393 |
goto out;
|
394 |
} |
395 |
|
396 |
qobject_decref(token); |
397 |
token = NULL;
|
398 |
|
399 |
obj = parse_value(ctxt, &working, ap); |
400 |
if (obj == NULL) { |
401 |
parse_error(ctxt, token, "expecting value");
|
402 |
goto out;
|
403 |
} |
404 |
|
405 |
qlist_append_obj(list, obj); |
406 |
|
407 |
token = qlist_pop(working); |
408 |
} |
409 |
|
410 |
qobject_decref(token); |
411 |
token = NULL;
|
412 |
} else {
|
413 |
token = qlist_pop(working); |
414 |
qobject_decref(token); |
415 |
token = NULL;
|
416 |
} |
417 |
|
418 |
QDECREF(*tokens); |
419 |
*tokens = working; |
420 |
|
421 |
return QOBJECT(list);
|
422 |
|
423 |
out:
|
424 |
qobject_decref(token); |
425 |
QDECREF(working); |
426 |
QDECREF(list); |
427 |
return NULL; |
428 |
} |
429 |
|
430 |
static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
|
431 |
{ |
432 |
QObject *token, *ret; |
433 |
QList *working = qlist_copy(*tokens); |
434 |
|
435 |
token = qlist_pop(working); |
436 |
|
437 |
if (token_get_type(token) != JSON_KEYWORD) {
|
438 |
goto out;
|
439 |
} |
440 |
|
441 |
if (token_is_keyword(token, "true")) { |
442 |
ret = QOBJECT(qbool_from_int(true));
|
443 |
} else if (token_is_keyword(token, "false")) { |
444 |
ret = QOBJECT(qbool_from_int(false));
|
445 |
} else {
|
446 |
parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
|
447 |
goto out;
|
448 |
} |
449 |
|
450 |
qobject_decref(token); |
451 |
QDECREF(*tokens); |
452 |
*tokens = working; |
453 |
|
454 |
return ret;
|
455 |
|
456 |
out:
|
457 |
qobject_decref(token); |
458 |
QDECREF(working); |
459 |
|
460 |
return NULL; |
461 |
} |
462 |
|
463 |
static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
|
464 |
{ |
465 |
QObject *token = NULL, *obj;
|
466 |
QList *working = qlist_copy(*tokens); |
467 |
|
468 |
if (ap == NULL) { |
469 |
goto out;
|
470 |
} |
471 |
|
472 |
token = qlist_pop(working); |
473 |
|
474 |
if (token_is_escape(token, "%p")) { |
475 |
obj = va_arg(*ap, QObject *); |
476 |
} else if (token_is_escape(token, "%i")) { |
477 |
obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
|
478 |
} else if (token_is_escape(token, "%d")) { |
479 |
obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
|
480 |
} else if (token_is_escape(token, "%ld")) { |
481 |
obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
|
482 |
} else if (token_is_escape(token, "%lld") || |
483 |
token_is_escape(token, "%I64d")) {
|
484 |
obj = QOBJECT(qint_from_int(va_arg(*ap, long long))); |
485 |
} else if (token_is_escape(token, "%s")) { |
486 |
obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *))); |
487 |
} else if (token_is_escape(token, "%f")) { |
488 |
obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
|
489 |
} else {
|
490 |
goto out;
|
491 |
} |
492 |
|
493 |
qobject_decref(token); |
494 |
QDECREF(*tokens); |
495 |
*tokens = working; |
496 |
|
497 |
return obj;
|
498 |
|
499 |
out:
|
500 |
qobject_decref(token); |
501 |
QDECREF(working); |
502 |
|
503 |
return NULL; |
504 |
} |
505 |
|
506 |
static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
|
507 |
{ |
508 |
QObject *token, *obj; |
509 |
QList *working = qlist_copy(*tokens); |
510 |
|
511 |
token = qlist_pop(working); |
512 |
switch (token_get_type(token)) {
|
513 |
case JSON_STRING:
|
514 |
obj = QOBJECT(qstring_from_escaped_str(ctxt, token)); |
515 |
break;
|
516 |
case JSON_INTEGER:
|
517 |
obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10))); |
518 |
break;
|
519 |
case JSON_FLOAT:
|
520 |
/* FIXME dependent on locale */
|
521 |
obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
|
522 |
break;
|
523 |
default:
|
524 |
goto out;
|
525 |
} |
526 |
|
527 |
qobject_decref(token); |
528 |
QDECREF(*tokens); |
529 |
*tokens = working; |
530 |
|
531 |
return obj;
|
532 |
|
533 |
out:
|
534 |
qobject_decref(token); |
535 |
QDECREF(working); |
536 |
|
537 |
return NULL; |
538 |
} |
539 |
|
540 |
static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
|
541 |
{ |
542 |
QObject *obj; |
543 |
|
544 |
obj = parse_object(ctxt, tokens, ap); |
545 |
if (obj == NULL) { |
546 |
obj = parse_array(ctxt, tokens, ap); |
547 |
} |
548 |
if (obj == NULL) { |
549 |
obj = parse_escape(ctxt, tokens, ap); |
550 |
} |
551 |
if (obj == NULL) { |
552 |
obj = parse_keyword(ctxt, tokens); |
553 |
} |
554 |
if (obj == NULL) { |
555 |
obj = parse_literal(ctxt, tokens); |
556 |
} |
557 |
|
558 |
return obj;
|
559 |
} |
560 |
|
561 |
QObject *json_parser_parse(QList *tokens, va_list *ap) |
562 |
{ |
563 |
JSONParserContext ctxt = {}; |
564 |
QList *working = qlist_copy(tokens); |
565 |
QObject *result; |
566 |
|
567 |
result = parse_value(&ctxt, &working, ap); |
568 |
|
569 |
QDECREF(working); |
570 |
|
571 |
return result;
|
572 |
} |