1 | /* Copyright (C) 2012-2015 Mark Nunberg.
|
---|
2 | *
|
---|
3 | * See included LICENSE file for license details.
|
---|
4 | */
|
---|
5 | /* copy from LICENSE file
|
---|
6 | Copyright (c) 2012-2015 M. Nunberg, mnunberg@haskalah.org
|
---|
7 |
|
---|
8 | Permission is hereby granted, free of charge, to any person obtaining
|
---|
9 | a copy of this software and associated documentation files (the
|
---|
10 | "Software"), to deal in the Software without restriction, including
|
---|
11 | without limitation the rights to use, copy, modify, merge, publish,
|
---|
12 | distribute, sublicense, and/or sell copies of the Software, and to
|
---|
13 | permit persons to whom the Software is furnished to do so, subject to
|
---|
14 | the following conditions:
|
---|
15 |
|
---|
16 | The above copyright notice and this permission notice shall be
|
---|
17 | included in all copies or substantial portions of the Software.
|
---|
18 |
|
---|
19 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
---|
20 | EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
---|
21 | MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
---|
22 | NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
|
---|
23 | LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
|
---|
24 | OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
|
---|
25 | WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
---|
26 | */
|
---|
27 | #include "jsonsl.h"
|
---|
28 | //#include <assert.h>
|
---|
29 | #include <limits.h>
|
---|
30 | #include <ctype.h>
|
---|
31 |
|
---|
32 | #ifdef JSONSL_USE_METRICS
|
---|
33 | #define XMETRICS \
|
---|
34 | X(STRINGY_INSIGNIFICANT) \
|
---|
35 | X(STRINGY_SLOWPATH) \
|
---|
36 | X(ALLOWED_WHITESPACE) \
|
---|
37 | X(QUOTE_FASTPATH) \
|
---|
38 | X(SPECIAL_FASTPATH) \
|
---|
39 | X(SPECIAL_WSPOP) \
|
---|
40 | X(SPECIAL_SLOWPATH) \
|
---|
41 | X(GENERIC) \
|
---|
42 | X(STRUCTURAL_TOKEN) \
|
---|
43 | X(SPECIAL_SWITCHFIRST) \
|
---|
44 | X(STRINGY_CATCH) \
|
---|
45 | X(ESCAPES) \
|
---|
46 | X(TOTAL) \
|
---|
47 |
|
---|
48 | struct jsonsl_metrics_st {
|
---|
49 | #define X(m) \
|
---|
50 | unsigned long metric_##m;
|
---|
51 | XMETRICS
|
---|
52 | #undef X
|
---|
53 | };
|
---|
54 |
|
---|
55 | static struct jsonsl_metrics_st GlobalMetrics = { 0 };
|
---|
56 | static unsigned long GenericCounter[0x100] = { 0 };
|
---|
57 | static unsigned long StringyCatchCounter[0x100] = { 0 };
|
---|
58 |
|
---|
59 | #define INCR_METRIC(m) \
|
---|
60 | GlobalMetrics.metric_##m++;
|
---|
61 |
|
---|
62 | #define INCR_GENERIC(c) \
|
---|
63 | INCR_METRIC(GENERIC); \
|
---|
64 | GenericCounter[c]++; \
|
---|
65 |
|
---|
66 | #define INCR_STRINGY_CATCH(c) \
|
---|
67 | INCR_METRIC(STRINGY_CATCH); \
|
---|
68 | StringyCatchCounter[c]++;
|
---|
69 |
|
---|
70 | JSONSL_API
|
---|
71 | void jsonsl_dump_global_metrics(void)
|
---|
72 | {
|
---|
73 | int ii;
|
---|
74 | printf("JSONSL Metrics:\n");
|
---|
75 | #define X(m) \
|
---|
76 | printf("\t%-30s %20lu (%0.2f%%)\n", #m, GlobalMetrics.metric_##m, \
|
---|
77 | (float)((float)(GlobalMetrics.metric_##m/(float)GlobalMetrics.metric_TOTAL)) * 100);
|
---|
78 | XMETRICS
|
---|
79 | #undef X
|
---|
80 | printf("Generic Characters:\n");
|
---|
81 | for (ii = 0; ii < 0xff; ii++) {
|
---|
82 | if (GenericCounter[ii]) {
|
---|
83 | printf("\t[ %c ] %lu\n", ii, GenericCounter[ii]);
|
---|
84 | }
|
---|
85 | }
|
---|
86 | printf("Weird string loop\n");
|
---|
87 | for (ii = 0; ii < 0xff; ii++) {
|
---|
88 | if (StringyCatchCounter[ii]) {
|
---|
89 | printf("\t[ %c ] %lu\n", ii, StringyCatchCounter[ii]);
|
---|
90 | }
|
---|
91 | }
|
---|
92 | }
|
---|
93 |
|
---|
94 | #else
|
---|
95 | #define INCR_METRIC(m)
|
---|
96 | #define INCR_GENERIC(c)
|
---|
97 | #define INCR_STRINGY_CATCH(c)
|
---|
98 | JSONSL_API
|
---|
99 | void jsonsl_dump_global_metrics(void) { }
|
---|
100 | #endif /* JSONSL_USE_METRICS */
|
---|
101 |
|
---|
102 | #define CASE_DIGITS \
|
---|
103 | case '1': \
|
---|
104 | case '2': \
|
---|
105 | case '3': \
|
---|
106 | case '4': \
|
---|
107 | case '5': \
|
---|
108 | case '6': \
|
---|
109 | case '7': \
|
---|
110 | case '8': \
|
---|
111 | case '9': \
|
---|
112 | case '0':
|
---|
113 |
|
---|
114 | static unsigned extract_special(unsigned);
|
---|
115 | static int is_special_end(unsigned);
|
---|
116 | static int is_allowed_whitespace(unsigned);
|
---|
117 | static int is_allowed_escape(unsigned);
|
---|
118 | static char get_escape_equiv(unsigned);
|
---|
119 |
|
---|
120 | JSONSL_API
|
---|
121 | jsonsl_t jsonsl_new(struct jsonsl_st *jsn, int nlevels)
|
---|
122 | {
|
---|
123 | jsn->levels_max = nlevels;
|
---|
124 | jsn->max_callback_level = -1;
|
---|
125 | jsonsl_reset(jsn);
|
---|
126 | return jsn;
|
---|
127 | }
|
---|
128 |
|
---|
129 | JSONSL_API
|
---|
130 | void jsonsl_reset(jsonsl_t jsn)
|
---|
131 | {
|
---|
132 | unsigned int ii;
|
---|
133 | jsn->tok_last = 0;
|
---|
134 | jsn->can_insert = 1;
|
---|
135 | jsn->pos = 0;
|
---|
136 | jsn->level = 0;
|
---|
137 | jsn->stopfl = 0;
|
---|
138 | jsn->in_escape = 0;
|
---|
139 | jsn->expecting = 0;
|
---|
140 |
|
---|
141 | memset(jsn->stack, 0, (jsn->levels_max * sizeof (struct jsonsl_state_st)));
|
---|
142 |
|
---|
143 | for (ii = 0; ii < jsn->levels_max; ii++) {
|
---|
144 | jsn->stack[ii].level = ii;
|
---|
145 | }
|
---|
146 | }
|
---|
147 |
|
---|
148 | JSONSL_API
|
---|
149 | void jsonsl_destroy(jsonsl_t jsn)
|
---|
150 | {
|
---|
151 | }
|
---|
152 |
|
---|
153 | JSONSL_API
|
---|
154 | void
|
---|
155 | jsonsl_feed(jsonsl_t jsn, const jsonsl_char_t *bytes, size_t nbytes)
|
---|
156 | {
|
---|
157 |
|
---|
158 | #define INVOKE_ERROR(eb) \
|
---|
159 | if (jsn->error_callback(jsn, JSONSL_ERROR_##eb, state, (char*)c)) { \
|
---|
160 | goto GT_AGAIN; \
|
---|
161 | } \
|
---|
162 | return;
|
---|
163 |
|
---|
164 | #define STACK_PUSH \
|
---|
165 | if (jsn->level >= (levels_max-1)) { \
|
---|
166 | jsn->error_callback(jsn, JSONSL_ERROR_LEVELS_EXCEEDED, state, (char*)c); \
|
---|
167 | return; \
|
---|
168 | } \
|
---|
169 | state = jsn->stack + (++jsn->level); \
|
---|
170 | state->ignore_callback = jsn->stack[jsn->level-1].ignore_callback; \
|
---|
171 | state->pos_begin = jsn->pos;
|
---|
172 |
|
---|
173 | #define STACK_POP_NOPOS \
|
---|
174 | state->pos_cur = jsn->pos; \
|
---|
175 | state = jsn->stack + (--jsn->level);
|
---|
176 |
|
---|
177 |
|
---|
178 | #define STACK_POP \
|
---|
179 | STACK_POP_NOPOS; \
|
---|
180 | state->pos_cur = jsn->pos;
|
---|
181 |
|
---|
182 | #define CALLBACK_AND_POP_NOPOS(T) \
|
---|
183 | state->pos_cur = jsn->pos; \
|
---|
184 | DO_CALLBACK(T, POP); \
|
---|
185 | state->nescapes = 0; \
|
---|
186 | state = jsn->stack + (--jsn->level);
|
---|
187 |
|
---|
188 | #define CALLBACK_AND_POP(T) \
|
---|
189 | CALLBACK_AND_POP_NOPOS(T); \
|
---|
190 | state->pos_cur = jsn->pos;
|
---|
191 |
|
---|
192 | #define SPECIAL_POP \
|
---|
193 | CALLBACK_AND_POP(SPECIAL); \
|
---|
194 | jsn->expecting = 0; \
|
---|
195 | jsn->tok_last = 0; \
|
---|
196 |
|
---|
197 | #define CUR_CHAR (*(jsonsl_uchar_t*)c)
|
---|
198 |
|
---|
199 | #define DO_CALLBACK(T, action) \
|
---|
200 | if (jsn->call_##T && \
|
---|
201 | jsn->max_callback_level > state->level && \
|
---|
202 | state->ignore_callback == 0) { \
|
---|
203 | \
|
---|
204 | if (jsn->action_callback_##action) { \
|
---|
205 | jsn->action_callback_##action(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
|
---|
206 | } else if (jsn->action_callback) { \
|
---|
207 | jsn->action_callback(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
|
---|
208 | } \
|
---|
209 | if (jsn->stopfl) { return; } \
|
---|
210 | }
|
---|
211 |
|
---|
212 | /**
|
---|
213 | * Verifies that we are able to insert the (non-string) item into a hash.
|
---|
214 | */
|
---|
215 | #define ENSURE_HVAL \
|
---|
216 | if (state->nelem % 2 == 0 && state->type == JSONSL_T_OBJECT) { \
|
---|
217 | INVOKE_ERROR(HKEY_EXPECTED); \
|
---|
218 | }
|
---|
219 |
|
---|
220 | #define VERIFY_SPECIAL(lit) \
|
---|
221 | if (CUR_CHAR != (lit)[jsn->pos - state->pos_begin]) { \
|
---|
222 | INVOKE_ERROR(SPECIAL_EXPECTED); \
|
---|
223 | }
|
---|
224 |
|
---|
225 | const jsonsl_uchar_t *c = (jsonsl_uchar_t*)bytes;
|
---|
226 | size_t levels_max = jsn->levels_max;
|
---|
227 | struct jsonsl_state_st *state = jsn->stack + jsn->level;
|
---|
228 | static int chrt_string_nopass[0x100] = { JSONSL_CHARTABLE_string_nopass };
|
---|
229 | jsn->base = bytes;
|
---|
230 |
|
---|
231 | for (; nbytes; nbytes--, jsn->pos++, c++) {
|
---|
232 | register unsigned state_type;
|
---|
233 | INCR_METRIC(TOTAL);
|
---|
234 | /* Special escape handling for some stuff */
|
---|
235 | if (jsn->in_escape) {
|
---|
236 | jsn->in_escape = 0;
|
---|
237 | if (!is_allowed_escape(CUR_CHAR)) {
|
---|
238 | INVOKE_ERROR(ESCAPE_INVALID);
|
---|
239 | } else if (CUR_CHAR == 'u') {
|
---|
240 | DO_CALLBACK(UESCAPE, UESCAPE);
|
---|
241 | if (jsn->return_UESCAPE) {
|
---|
242 | return;
|
---|
243 | }
|
---|
244 | }
|
---|
245 | goto GT_NEXT;
|
---|
246 | }
|
---|
247 | GT_AGAIN:
|
---|
248 | /**
|
---|
249 | * Several fast-tracks for common cases:
|
---|
250 | */
|
---|
251 | state_type = state->type;
|
---|
252 | if (state_type & JSONSL_Tf_STRINGY) {
|
---|
253 | /* check if our character cannot ever change our current string state
|
---|
254 | * or throw an error
|
---|
255 | */
|
---|
256 | if (
|
---|
257 | #ifdef JSONSL_USE_WCHAR
|
---|
258 | CUR_CHAR >= 0x100 ||
|
---|
259 | #endif /* JSONSL_USE_WCHAR */
|
---|
260 | (!chrt_string_nopass[CUR_CHAR & 0xff])) {
|
---|
261 | INCR_METRIC(STRINGY_INSIGNIFICANT);
|
---|
262 | goto GT_NEXT;
|
---|
263 | } else if (CUR_CHAR == '"') {
|
---|
264 | goto GT_QUOTE;
|
---|
265 | } else if (CUR_CHAR == '\\') {
|
---|
266 | goto GT_ESCAPE;
|
---|
267 | } else {
|
---|
268 | INVOKE_ERROR(WEIRD_WHITESPACE);
|
---|
269 | }
|
---|
270 | INCR_METRIC(STRINGY_SLOWPATH);
|
---|
271 |
|
---|
272 | } else if (state_type == JSONSL_T_SPECIAL) {
|
---|
273 | if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
|
---|
274 | switch (CUR_CHAR) {
|
---|
275 | CASE_DIGITS
|
---|
276 | state->nelem = (state->nelem*10) + (CUR_CHAR-0x30);
|
---|
277 | goto GT_NEXT;
|
---|
278 |
|
---|
279 | case 'e':
|
---|
280 | case 'E':
|
---|
281 | case '-':
|
---|
282 | case '+':
|
---|
283 | state->special_flags |= JSONSL_SPECIALf_EXPONENT;
|
---|
284 | goto GT_NEXT;
|
---|
285 | case '.':
|
---|
286 | state->special_flags |= JSONSL_SPECIALf_FLOAT;
|
---|
287 | goto GT_NEXT;
|
---|
288 | default:
|
---|
289 | if (is_special_end(CUR_CHAR)) {
|
---|
290 | goto GT_SPECIAL_POP;
|
---|
291 | }
|
---|
292 | INVOKE_ERROR(INVALID_NUMBER);
|
---|
293 | break;
|
---|
294 | }
|
---|
295 | }
|
---|
296 | /* else if (!NUMERIC) */
|
---|
297 | if (!is_special_end(CUR_CHAR)) {
|
---|
298 | /* Verify TRUE, FALSE, NULL */
|
---|
299 | if (state->special_flags == JSONSL_SPECIALf_TRUE) {
|
---|
300 | VERIFY_SPECIAL("true");
|
---|
301 | } else if (state->special_flags == JSONSL_SPECIALf_FALSE) {
|
---|
302 | VERIFY_SPECIAL("false");
|
---|
303 | } else if (state->special_flags == JSONSL_SPECIALf_NULL) {
|
---|
304 | VERIFY_SPECIAL("null");
|
---|
305 | }
|
---|
306 | INCR_METRIC(SPECIAL_FASTPATH);
|
---|
307 | goto GT_NEXT;
|
---|
308 | }
|
---|
309 |
|
---|
310 | GT_SPECIAL_POP:
|
---|
311 | SPECIAL_POP;
|
---|
312 | jsn->expecting = ',';
|
---|
313 | if (is_allowed_whitespace(CUR_CHAR)) {
|
---|
314 | goto GT_NEXT;
|
---|
315 | }
|
---|
316 | /**
|
---|
317 | * This works because we have a non-whitespace token
|
---|
318 | * which is not a special token. If this is a structural
|
---|
319 | * character then it will be gracefully handled by the
|
---|
320 | * switch statement. Otherwise it will default to the 'special'
|
---|
321 | * state again,
|
---|
322 | */
|
---|
323 | goto GT_STRUCTURAL_TOKEN;
|
---|
324 | } else if (is_allowed_whitespace(CUR_CHAR)) {
|
---|
325 | INCR_METRIC(ALLOWED_WHITESPACE);
|
---|
326 | /* So we're not special. Harmless insignificant whitespace
|
---|
327 | * passthrough
|
---|
328 | */
|
---|
329 | goto GT_NEXT;
|
---|
330 | } else if (extract_special(CUR_CHAR)) {
|
---|
331 | /* not a string, whitespace, or structural token. must be special */
|
---|
332 | goto GT_SPECIAL_BEGIN;
|
---|
333 | }
|
---|
334 |
|
---|
335 | INCR_GENERIC(CUR_CHAR);
|
---|
336 |
|
---|
337 | if (CUR_CHAR == '"') {
|
---|
338 | GT_QUOTE:
|
---|
339 | jsn->can_insert = 0;
|
---|
340 | switch (state_type) {
|
---|
341 |
|
---|
342 | /* the end of a string or hash key */
|
---|
343 | case JSONSL_T_STRING:
|
---|
344 | CALLBACK_AND_POP(STRING);
|
---|
345 | goto GT_NEXT;
|
---|
346 | case JSONSL_T_HKEY:
|
---|
347 | CALLBACK_AND_POP(HKEY);
|
---|
348 | goto GT_NEXT;
|
---|
349 |
|
---|
350 | case JSONSL_T_OBJECT:
|
---|
351 | state->nelem++;
|
---|
352 | if ( (state->nelem-1) % 2 ) {
|
---|
353 | /* Odd, this must be a hash value */
|
---|
354 | if (jsn->tok_last != ':') {
|
---|
355 | INVOKE_ERROR(MISSING_TOKEN);
|
---|
356 | }
|
---|
357 | jsn->expecting = ','; /* Can't figure out what to expect next */
|
---|
358 | jsn->tok_last = 0;
|
---|
359 |
|
---|
360 | STACK_PUSH;
|
---|
361 | state->type = JSONSL_T_STRING;
|
---|
362 | DO_CALLBACK(STRING, PUSH);
|
---|
363 |
|
---|
364 | } else {
|
---|
365 | /* hash key */
|
---|
366 | if (jsn->expecting != '"') {
|
---|
367 | INVOKE_ERROR(STRAY_TOKEN);
|
---|
368 | }
|
---|
369 | jsn->tok_last = 0;
|
---|
370 | jsn->expecting = ':';
|
---|
371 |
|
---|
372 | STACK_PUSH;
|
---|
373 | state->type = JSONSL_T_HKEY;
|
---|
374 | DO_CALLBACK(HKEY, PUSH);
|
---|
375 | }
|
---|
376 | goto GT_NEXT;
|
---|
377 |
|
---|
378 | case JSONSL_T_LIST:
|
---|
379 | state->nelem++;
|
---|
380 | STACK_PUSH;
|
---|
381 | state->type = JSONSL_T_STRING;
|
---|
382 | jsn->expecting = ',';
|
---|
383 | jsn->tok_last = 0;
|
---|
384 | DO_CALLBACK(STRING, PUSH);
|
---|
385 | goto GT_NEXT;
|
---|
386 |
|
---|
387 | case JSONSL_T_SPECIAL:
|
---|
388 | INVOKE_ERROR(STRAY_TOKEN);
|
---|
389 | break;
|
---|
390 |
|
---|
391 | default:
|
---|
392 | INVOKE_ERROR(STRING_OUTSIDE_CONTAINER);
|
---|
393 | break;
|
---|
394 | } /* switch(state->type) */
|
---|
395 | } else if (CUR_CHAR == '\\') {
|
---|
396 | GT_ESCAPE:
|
---|
397 | INCR_METRIC(ESCAPES);
|
---|
398 | /* Escape */
|
---|
399 | if ( (state->type & JSONSL_Tf_STRINGY) == 0 ) {
|
---|
400 | INVOKE_ERROR(ESCAPE_OUTSIDE_STRING);
|
---|
401 | }
|
---|
402 | state->nescapes++;
|
---|
403 | jsn->in_escape = 1;
|
---|
404 | goto GT_NEXT;
|
---|
405 | } /* " or \ */
|
---|
406 |
|
---|
407 | GT_STRUCTURAL_TOKEN:
|
---|
408 | switch (CUR_CHAR) {
|
---|
409 | case ':':
|
---|
410 | INCR_METRIC(STRUCTURAL_TOKEN);
|
---|
411 | if (jsn->expecting != CUR_CHAR) {
|
---|
412 | INVOKE_ERROR(STRAY_TOKEN);
|
---|
413 | }
|
---|
414 | jsn->tok_last = ':';
|
---|
415 | jsn->can_insert = 1;
|
---|
416 | jsn->expecting = '"';
|
---|
417 | goto GT_NEXT;
|
---|
418 |
|
---|
419 | case ',':
|
---|
420 | INCR_METRIC(STRUCTURAL_TOKEN);
|
---|
421 | /**
|
---|
422 | * The comma is one of the more generic tokens.
|
---|
423 | * In the context of an OBJECT, the can_insert flag
|
---|
424 | * should never be set, and no other action is
|
---|
425 | * necessary.
|
---|
426 | */
|
---|
427 | if (jsn->expecting != CUR_CHAR) {
|
---|
428 | /* make this branch execute only when we haven't manually
|
---|
429 | * just placed the ',' in the expecting register.
|
---|
430 | */
|
---|
431 | INVOKE_ERROR(STRAY_TOKEN);
|
---|
432 | }
|
---|
433 |
|
---|
434 | if (state->type == JSONSL_T_OBJECT) {
|
---|
435 | /* end of hash value, expect a string as a hash key */
|
---|
436 | jsn->expecting = '"';
|
---|
437 | } else {
|
---|
438 | jsn->can_insert = 1;
|
---|
439 | }
|
---|
440 |
|
---|
441 | jsn->tok_last = ',';
|
---|
442 | jsn->expecting = '"';
|
---|
443 | goto GT_NEXT;
|
---|
444 |
|
---|
445 | /* new list or object */
|
---|
446 | /* hashes are more common */
|
---|
447 | case '{':
|
---|
448 | case '[':
|
---|
449 | INCR_METRIC(STRUCTURAL_TOKEN);
|
---|
450 | if (!jsn->can_insert) {
|
---|
451 | INVOKE_ERROR(CANT_INSERT);
|
---|
452 | }
|
---|
453 |
|
---|
454 | ENSURE_HVAL;
|
---|
455 | state->nelem++;
|
---|
456 |
|
---|
457 | STACK_PUSH;
|
---|
458 | /* because the constants match the opening delimiters, we can do this: */
|
---|
459 | state->type = CUR_CHAR;
|
---|
460 | state->nelem = 0;
|
---|
461 | jsn->can_insert = 1;
|
---|
462 | if (CUR_CHAR == '{') {
|
---|
463 | /* If we're a hash, we expect a key first, which is quouted */
|
---|
464 | jsn->expecting = '"';
|
---|
465 | }
|
---|
466 | if (CUR_CHAR == JSONSL_T_OBJECT) {
|
---|
467 | DO_CALLBACK(OBJECT, PUSH);
|
---|
468 | } else {
|
---|
469 | DO_CALLBACK(LIST, PUSH);
|
---|
470 | }
|
---|
471 | jsn->tok_last = 0;
|
---|
472 | goto GT_NEXT;
|
---|
473 |
|
---|
474 | /* closing of list or object */
|
---|
475 | case '}':
|
---|
476 | case ']':
|
---|
477 | INCR_METRIC(STRUCTURAL_TOKEN);
|
---|
478 | if (jsn->tok_last == ',' && jsn->options.allow_trailing_comma == 0) {
|
---|
479 | INVOKE_ERROR(TRAILING_COMMA);
|
---|
480 | }
|
---|
481 |
|
---|
482 | jsn->can_insert = 0;
|
---|
483 | jsn->level--;
|
---|
484 | jsn->expecting = ',';
|
---|
485 | jsn->tok_last = 0;
|
---|
486 | if (CUR_CHAR == ']') {
|
---|
487 | if (state->type != '[') {
|
---|
488 | INVOKE_ERROR(BRACKET_MISMATCH);
|
---|
489 | }
|
---|
490 | DO_CALLBACK(LIST, POP);
|
---|
491 | } else {
|
---|
492 | if (state->type != '{') {
|
---|
493 | INVOKE_ERROR(BRACKET_MISMATCH);
|
---|
494 | }
|
---|
495 | DO_CALLBACK(OBJECT, POP);
|
---|
496 | }
|
---|
497 | state = jsn->stack + jsn->level;
|
---|
498 | state->pos_cur = jsn->pos;
|
---|
499 | goto GT_NEXT;
|
---|
500 |
|
---|
501 | default:
|
---|
502 | GT_SPECIAL_BEGIN:
|
---|
503 | /**
|
---|
504 | * Not a string, not a structural token, and not benign whitespace.
|
---|
505 | * Technically we should iterate over the character always, but since
|
---|
506 | * we are not doing full numerical/value decoding anyway (but only hinting),
|
---|
507 | * we only check upon entry.
|
---|
508 | */
|
---|
509 | if (state->type != JSONSL_T_SPECIAL) {
|
---|
510 | int special_flags = extract_special(CUR_CHAR);
|
---|
511 | if (!special_flags) {
|
---|
512 | /**
|
---|
513 | * Try to do some heuristics here anyway to figure out what kind of
|
---|
514 | * error this is. The 'special' case is a fallback scenario anyway.
|
---|
515 | */
|
---|
516 | if (CUR_CHAR == '\0') {
|
---|
517 | INVOKE_ERROR(FOUND_NULL_BYTE);
|
---|
518 | } else if (CUR_CHAR < 0x20) {
|
---|
519 | INVOKE_ERROR(WEIRD_WHITESPACE);
|
---|
520 | } else {
|
---|
521 | INVOKE_ERROR(SPECIAL_EXPECTED);
|
---|
522 | }
|
---|
523 | }
|
---|
524 | ENSURE_HVAL;
|
---|
525 | state->nelem++;
|
---|
526 | if (!jsn->can_insert) {
|
---|
527 | INVOKE_ERROR(CANT_INSERT);
|
---|
528 | }
|
---|
529 | STACK_PUSH;
|
---|
530 | state->type = JSONSL_T_SPECIAL;
|
---|
531 | state->special_flags = special_flags;
|
---|
532 | if (special_flags == JSONSL_SPECIALf_UNSIGNED) {
|
---|
533 | state->nelem = CUR_CHAR - 0x30;
|
---|
534 | } else {
|
---|
535 | state->nelem = 0;
|
---|
536 | }
|
---|
537 | DO_CALLBACK(SPECIAL, PUSH);
|
---|
538 | }
|
---|
539 | goto GT_NEXT;
|
---|
540 | }
|
---|
541 |
|
---|
542 | GT_NEXT:
|
---|
543 | continue;
|
---|
544 | }
|
---|
545 | }
|
---|
546 |
|
---|
547 | JSONSL_API
|
---|
548 | const char* jsonsl_strerror(jsonsl_error_t err)
|
---|
549 | {
|
---|
550 | if (err == JSONSL_ERROR_SUCCESS) {
|
---|
551 | return "SUCCESS";
|
---|
552 | }
|
---|
553 | #define X(t) \
|
---|
554 | if (err == JSONSL_ERROR_##t) \
|
---|
555 | return #t;
|
---|
556 | JSONSL_XERR;
|
---|
557 | #undef X
|
---|
558 | return "<UNKNOWN_ERROR>";
|
---|
559 | }
|
---|
560 |
|
---|
561 | JSONSL_API
|
---|
562 | const char *jsonsl_strtype(jsonsl_type_t type)
|
---|
563 | {
|
---|
564 | #define X(o,c) \
|
---|
565 | if (type == JSONSL_T_##o) \
|
---|
566 | return #o;
|
---|
567 | JSONSL_XTYPE
|
---|
568 | #undef X
|
---|
569 | return "UNKNOWN TYPE";
|
---|
570 |
|
---|
571 | }
|
---|
572 |
|
---|
573 | /*
|
---|
574 | *
|
---|
575 | * JPR/JSONPointer functions
|
---|
576 | *
|
---|
577 | *
|
---|
578 | */
|
---|
579 | #ifndef JSONSL_NO_JPR
|
---|
580 | static
|
---|
581 | jsonsl_jpr_type_t
|
---|
582 | populate_component(char *in,
|
---|
583 | struct jsonsl_jpr_component_st *component,
|
---|
584 | char **next,
|
---|
585 | jsonsl_error_t *errp)
|
---|
586 | {
|
---|
587 | unsigned long pctval;
|
---|
588 | char *c = NULL, *outp = NULL, *end = NULL;
|
---|
589 | size_t input_len;
|
---|
590 | jsonsl_jpr_type_t ret = JSONSL_PATH_NONE;
|
---|
591 |
|
---|
592 | if (*next == NULL || *(*next) == '\0') {
|
---|
593 | return JSONSL_PATH_NONE;
|
---|
594 | }
|
---|
595 |
|
---|
596 | /* Replace the next / with a NULL */
|
---|
597 | *next = strstr(in, "/");
|
---|
598 | if (*next != NULL) {
|
---|
599 | *(*next) = '\0'; /* drop the forward slash */
|
---|
600 | input_len = *next - in;
|
---|
601 | end = *next;
|
---|
602 | *next += 1; /* next character after the '/' */
|
---|
603 | } else {
|
---|
604 | input_len = strlen(in);
|
---|
605 | end = in + input_len + 1;
|
---|
606 | }
|
---|
607 |
|
---|
608 | component->pstr = in;
|
---|
609 |
|
---|
610 | /* Check for special components of interest */
|
---|
611 | if (*in == JSONSL_PATH_WILDCARD_CHAR && input_len == 1) {
|
---|
612 | /* Lone wildcard */
|
---|
613 | ret = JSONSL_PATH_WILDCARD;
|
---|
614 | goto GT_RET;
|
---|
615 | } else if (isdigit(*in)) {
|
---|
616 | /* ASCII Numeric */
|
---|
617 | char *endptr;
|
---|
618 | component->idx = strtoul(in, &endptr, 10);
|
---|
619 | if (endptr && *endptr == '\0') {
|
---|
620 | ret = JSONSL_PATH_NUMERIC;
|
---|
621 | goto GT_RET;
|
---|
622 | }
|
---|
623 | }
|
---|
624 |
|
---|
625 | /* Default, it's a string */
|
---|
626 | ret = JSONSL_PATH_STRING;
|
---|
627 | for (c = outp = in; c < end; c++, outp++) {
|
---|
628 | char origc;
|
---|
629 | if (*c != '%') {
|
---|
630 | goto GT_ASSIGN;
|
---|
631 | }
|
---|
632 | /*
|
---|
633 | * c = { [+0] = '%', [+1] = 'b', [+2] = 'e', [+3] = '\0' }
|
---|
634 | */
|
---|
635 |
|
---|
636 | /* Need %XX */
|
---|
637 | if (c+2 >= end) {
|
---|
638 | *errp = JSONSL_ERROR_PERCENT_BADHEX;
|
---|
639 | return JSONSL_PATH_INVALID;
|
---|
640 | }
|
---|
641 | if (! (isxdigit(*(c+1)) && isxdigit(*(c+2))) ) {
|
---|
642 | *errp = JSONSL_ERROR_PERCENT_BADHEX;
|
---|
643 | return JSONSL_PATH_INVALID;
|
---|
644 | }
|
---|
645 |
|
---|
646 | /* Temporarily null-terminate the characters */
|
---|
647 | origc = *(c+3);
|
---|
648 | *(c+3) = '\0';
|
---|
649 | pctval = strtoul(c+1, NULL, 16);
|
---|
650 | *(c+3) = origc;
|
---|
651 |
|
---|
652 | *outp = (char) pctval;
|
---|
653 | c += 2;
|
---|
654 | continue;
|
---|
655 |
|
---|
656 | GT_ASSIGN:
|
---|
657 | *outp = *c;
|
---|
658 | }
|
---|
659 | /* Null-terminate the string */
|
---|
660 | for (; outp < c; outp++) {
|
---|
661 | *outp = '\0';
|
---|
662 | }
|
---|
663 |
|
---|
664 | GT_RET:
|
---|
665 | component->ptype = ret;
|
---|
666 | if (ret != JSONSL_PATH_WILDCARD) {
|
---|
667 | component->len = strlen(component->pstr);
|
---|
668 | }
|
---|
669 | return ret;
|
---|
670 | }
|
---|
671 |
|
---|
672 | JSONSL_API
|
---|
673 | jsonsl_jpr_t
|
---|
674 | jsonsl_jpr_new(const char *path, jsonsl_error_t *errp)
|
---|
675 | {
|
---|
676 | char *my_copy = NULL;
|
---|
677 | int count, curidx, len;
|
---|
678 | struct jsonsl_jpr_st *ret = NULL;
|
---|
679 | struct jsonsl_jpr_component_st *components = NULL;
|
---|
680 | size_t origlen;
|
---|
681 | jsonsl_error_t errstacked;
|
---|
682 |
|
---|
683 | #define JPR_BAIL(err) *errp = err; goto GT_ERROR;
|
---|
684 |
|
---|
685 | if (errp == NULL) {
|
---|
686 | errp = &errstacked;
|
---|
687 | }
|
---|
688 |
|
---|
689 | if (path == NULL || *path != '/') {
|
---|
690 | JPR_BAIL(JSONSL_ERROR_JPR_NOROOT);
|
---|
691 | return NULL;
|
---|
692 | }
|
---|
693 |
|
---|
694 | count = 1;
|
---|
695 | path++;
|
---|
696 | {
|
---|
697 | const char *c = path;
|
---|
698 | for (; *c; c++) {
|
---|
699 | if (*c == '/') {
|
---|
700 | count++;
|
---|
701 | if (*(c+1) == '/') {
|
---|
702 | JPR_BAIL(JSONSL_ERROR_JPR_DUPSLASH);
|
---|
703 | }
|
---|
704 | }
|
---|
705 | }
|
---|
706 | }
|
---|
707 | if(*path) {
|
---|
708 | count++;
|
---|
709 | }
|
---|
710 |
|
---|
711 | components = (struct jsonsl_jpr_component_st *)
|
---|
712 | malloc(sizeof(*components) * count);
|
---|
713 | if (!components) {
|
---|
714 | JPR_BAIL(JSONSL_ERROR_ENOMEM);
|
---|
715 | }
|
---|
716 |
|
---|
717 | len = strlen(path) + 1;
|
---|
718 | my_copy = (char *)malloc(len);
|
---|
719 | if (!my_copy) {
|
---|
720 | JPR_BAIL(JSONSL_ERROR_ENOMEM);
|
---|
721 | }
|
---|
722 |
|
---|
723 | strcpy_s(my_copy, len, path);
|
---|
724 |
|
---|
725 | components[0].ptype = JSONSL_PATH_ROOT;
|
---|
726 |
|
---|
727 | if (*my_copy) {
|
---|
728 | char *cur = my_copy;
|
---|
729 | int pathret = JSONSL_PATH_STRING;
|
---|
730 | curidx = 1;
|
---|
731 | while (pathret > 0 && curidx < count) {
|
---|
732 | pathret = populate_component(cur, components + curidx, &cur, errp);
|
---|
733 | if (pathret > 0) {
|
---|
734 | curidx++;
|
---|
735 | } else {
|
---|
736 | break;
|
---|
737 | }
|
---|
738 | }
|
---|
739 |
|
---|
740 | if (pathret == JSONSL_PATH_INVALID) {
|
---|
741 | JPR_BAIL(JSONSL_ERROR_JPR_BADPATH);
|
---|
742 | }
|
---|
743 | } else {
|
---|
744 | curidx = 1;
|
---|
745 | }
|
---|
746 |
|
---|
747 | path--; /*revert path to leading '/' */
|
---|
748 | origlen = strlen(path) + 1;
|
---|
749 | ret = (struct jsonsl_jpr_st *)malloc(sizeof(*ret));
|
---|
750 | if (!ret) {
|
---|
751 | JPR_BAIL(JSONSL_ERROR_ENOMEM);
|
---|
752 | }
|
---|
753 | ret->orig = (char *)malloc(origlen);
|
---|
754 | if (!ret->orig) {
|
---|
755 | JPR_BAIL(JSONSL_ERROR_ENOMEM);
|
---|
756 | }
|
---|
757 | ret->components = components;
|
---|
758 | ret->ncomponents = curidx;
|
---|
759 | ret->basestr = my_copy;
|
---|
760 | ret->norig = origlen-1;
|
---|
761 | strcpy_s(ret->orig, origlen, path);
|
---|
762 |
|
---|
763 | return ret;
|
---|
764 |
|
---|
765 | GT_ERROR:
|
---|
766 | free(my_copy);
|
---|
767 | free(components);
|
---|
768 | if (ret) {
|
---|
769 | free(ret->orig);
|
---|
770 | }
|
---|
771 | free(ret);
|
---|
772 | return NULL;
|
---|
773 | #undef JPR_BAIL
|
---|
774 | }
|
---|
775 |
|
---|
776 | void jsonsl_jpr_destroy(jsonsl_jpr_t jpr)
|
---|
777 | {
|
---|
778 | free(jpr->components);
|
---|
779 | free(jpr->basestr);
|
---|
780 | free(jpr->orig);
|
---|
781 | free(jpr);
|
---|
782 | }
|
---|
783 |
|
---|
784 | JSONSL_API
|
---|
785 | jsonsl_jpr_match_t
|
---|
786 | jsonsl_jpr_match(jsonsl_jpr_t jpr,
|
---|
787 | unsigned int parent_type,
|
---|
788 | unsigned int parent_level,
|
---|
789 | const char *key,
|
---|
790 | size_t nkey)
|
---|
791 | {
|
---|
792 | /* find our current component. This is the child level */
|
---|
793 | int cmpret;
|
---|
794 | struct jsonsl_jpr_component_st *p_component;
|
---|
795 | p_component = jpr->components + parent_level;
|
---|
796 |
|
---|
797 | if (parent_level >= jpr->ncomponents) {
|
---|
798 | return JSONSL_MATCH_NOMATCH;
|
---|
799 | }
|
---|
800 |
|
---|
801 | /* Lone query for 'root' element. Always matches */
|
---|
802 | if (parent_level == 0) {
|
---|
803 | if (jpr->ncomponents == 1) {
|
---|
804 | return JSONSL_MATCH_COMPLETE;
|
---|
805 | } else {
|
---|
806 | return JSONSL_MATCH_POSSIBLE;
|
---|
807 | }
|
---|
808 | }
|
---|
809 |
|
---|
810 | /* Wildcard, always matches */
|
---|
811 | if (p_component->ptype == JSONSL_PATH_WILDCARD) {
|
---|
812 | if (parent_level == jpr->ncomponents-1) {
|
---|
813 | return JSONSL_MATCH_COMPLETE;
|
---|
814 | } else {
|
---|
815 | return JSONSL_MATCH_POSSIBLE;
|
---|
816 | }
|
---|
817 | }
|
---|
818 |
|
---|
819 | /* Check numeric array index. This gets its special block so we can avoid
|
---|
820 | * string comparisons */
|
---|
821 | if (p_component->ptype == JSONSL_PATH_NUMERIC) {
|
---|
822 | if (parent_type == JSONSL_T_LIST) {
|
---|
823 | if (p_component->idx != nkey) {
|
---|
824 | /* Wrong index */
|
---|
825 | return JSONSL_MATCH_NOMATCH;
|
---|
826 | } else {
|
---|
827 | if (parent_level == jpr->ncomponents-1) {
|
---|
828 | /* This is the last element of the path */
|
---|
829 | return JSONSL_MATCH_COMPLETE;
|
---|
830 | } else {
|
---|
831 | /* Intermediate element */
|
---|
832 | return JSONSL_MATCH_POSSIBLE;
|
---|
833 | }
|
---|
834 | }
|
---|
835 | } else if (p_component->is_arridx) {
|
---|
836 | /* Numeric and an array index (set explicitly by user). But not
|
---|
837 | * a list for a parent */
|
---|
838 | return JSONSL_MATCH_TYPE_MISMATCH;
|
---|
839 | }
|
---|
840 | }
|
---|
841 |
|
---|
842 | /* Check lengths */
|
---|
843 | if (p_component->len != nkey) {
|
---|
844 | return JSONSL_MATCH_NOMATCH;
|
---|
845 | }
|
---|
846 |
|
---|
847 | /* Check string comparison */
|
---|
848 | cmpret = strncmp(p_component->pstr, key, nkey);
|
---|
849 | if (cmpret == 0) {
|
---|
850 | if (parent_level == jpr->ncomponents-1) {
|
---|
851 | return JSONSL_MATCH_COMPLETE;
|
---|
852 | } else {
|
---|
853 | return JSONSL_MATCH_POSSIBLE;
|
---|
854 | }
|
---|
855 | }
|
---|
856 |
|
---|
857 | return JSONSL_MATCH_NOMATCH;
|
---|
858 | }
|
---|
859 |
|
---|
860 | JSONSL_API
|
---|
861 | void jsonsl_jpr_match_state_init(jsonsl_t jsn,
|
---|
862 | jsonsl_jpr_t *jprs,
|
---|
863 | size_t njprs)
|
---|
864 | {
|
---|
865 | size_t ii, *firstjmp;
|
---|
866 | if (njprs == 0) {
|
---|
867 | return;
|
---|
868 | }
|
---|
869 | jsn->jprs = (jsonsl_jpr_t *)malloc(sizeof(jsonsl_jpr_t) * njprs);
|
---|
870 | jsn->jpr_count = njprs;
|
---|
871 | jsn->jpr_root = (size_t*)calloc(1, sizeof(size_t) * njprs * jsn->levels_max);
|
---|
872 | memcpy(jsn->jprs, jprs, sizeof(jsonsl_jpr_t) * njprs);
|
---|
873 | /* Set the initial jump table values */
|
---|
874 |
|
---|
875 | firstjmp = jsn->jpr_root;
|
---|
876 | for (ii = 0; ii < njprs; ii++) {
|
---|
877 | firstjmp[ii] = ii+1;
|
---|
878 | }
|
---|
879 | }
|
---|
880 |
|
---|
881 | JSONSL_API
|
---|
882 | void jsonsl_jpr_match_state_cleanup(jsonsl_t jsn)
|
---|
883 | {
|
---|
884 | if (jsn->jpr_count == 0) {
|
---|
885 | return;
|
---|
886 | }
|
---|
887 |
|
---|
888 | free(jsn->jpr_root);
|
---|
889 | free(jsn->jprs);
|
---|
890 | jsn->jprs = NULL;
|
---|
891 | jsn->jpr_root = NULL;
|
---|
892 | jsn->jpr_count = 0;
|
---|
893 | }
|
---|
894 |
|
---|
895 | /**
|
---|
896 | * This function should be called exactly once on each element...
|
---|
897 | * This should also be called in recursive order, since we rely
|
---|
898 | * on the parent having been initalized for a match.
|
---|
899 | *
|
---|
900 | * Since the parent is checked for a match as well, we maintain a 'serial' counter.
|
---|
901 | * Whenever we traverse an element, we expect the serial to be the same as a global
|
---|
902 | * integer. If they do not match, we re-initialize the context, and set the serial.
|
---|
903 | *
|
---|
904 | * This ensures a type of consistency without having a proactive reset by the
|
---|
905 | * main lexer itself.
|
---|
906 | *
|
---|
907 | */
|
---|
908 | JSONSL_API
|
---|
909 | jsonsl_jpr_t jsonsl_jpr_match_state(jsonsl_t jsn,
|
---|
910 | struct jsonsl_state_st *state,
|
---|
911 | const char *key,
|
---|
912 | size_t nkey,
|
---|
913 | jsonsl_jpr_match_t *out)
|
---|
914 | {
|
---|
915 | struct jsonsl_state_st *parent_state;
|
---|
916 | jsonsl_jpr_t ret = NULL;
|
---|
917 |
|
---|
918 | /* Jump and JPR tables for our own state and the parent state */
|
---|
919 | size_t *jmptable, *pjmptable;
|
---|
920 | size_t jmp_cur, ii, ourjmpidx;
|
---|
921 |
|
---|
922 | if (!jsn->jpr_root) {
|
---|
923 | *out = JSONSL_MATCH_NOMATCH;
|
---|
924 | return NULL;
|
---|
925 | }
|
---|
926 |
|
---|
927 | pjmptable = jsn->jpr_root + (jsn->jpr_count * (state->level-1));
|
---|
928 | jmptable = pjmptable + jsn->jpr_count;
|
---|
929 |
|
---|
930 | /* If the parent cannot match, then invalidate it */
|
---|
931 | if (*pjmptable == 0) {
|
---|
932 | *jmptable = 0;
|
---|
933 | *out = JSONSL_MATCH_NOMATCH;
|
---|
934 | return NULL;
|
---|
935 | }
|
---|
936 |
|
---|
937 | parent_state = jsn->stack + state->level - 1;
|
---|
938 |
|
---|
939 | if (parent_state->type == JSONSL_T_LIST) {
|
---|
940 | nkey = (size_t) parent_state->nelem;
|
---|
941 | }
|
---|
942 |
|
---|
943 | *jmptable = 0;
|
---|
944 | ourjmpidx = 0;
|
---|
945 | memset(jmptable, 0, sizeof(int) * jsn->jpr_count);
|
---|
946 |
|
---|
947 | for (ii = 0; ii < jsn->jpr_count; ii++) {
|
---|
948 | jmp_cur = pjmptable[ii];
|
---|
949 | if (jmp_cur) {
|
---|
950 | jsonsl_jpr_t jpr = jsn->jprs[jmp_cur-1];
|
---|
951 | *out = jsonsl_jpr_match(jpr,
|
---|
952 | parent_state->type,
|
---|
953 | parent_state->level,
|
---|
954 | key, nkey);
|
---|
955 | if (*out == JSONSL_MATCH_COMPLETE) {
|
---|
956 | ret = jpr;
|
---|
957 | *jmptable = 0;
|
---|
958 | return ret;
|
---|
959 | } else if (*out == JSONSL_MATCH_POSSIBLE) {
|
---|
960 | jmptable[ourjmpidx] = ii+1;
|
---|
961 | ourjmpidx++;
|
---|
962 | }
|
---|
963 | } else {
|
---|
964 | break;
|
---|
965 | }
|
---|
966 | }
|
---|
967 | if (!*jmptable) {
|
---|
968 | *out = JSONSL_MATCH_NOMATCH;
|
---|
969 | }
|
---|
970 | return NULL;
|
---|
971 | }
|
---|
972 |
|
---|
973 | JSONSL_API
|
---|
974 | const char *jsonsl_strmatchtype(jsonsl_jpr_match_t match)
|
---|
975 | {
|
---|
976 | #define X(T,v) \
|
---|
977 | if ( match == JSONSL_MATCH_##T ) \
|
---|
978 | return #T;
|
---|
979 | JSONSL_XMATCH
|
---|
980 | #undef X
|
---|
981 | return "<UNKNOWN>";
|
---|
982 | }
|
---|
983 |
|
---|
984 | #endif /* JSONSL_WITH_JPR */
|
---|
985 |
|
---|
986 | /**
|
---|
987 | * Utility function to convert escape sequences
|
---|
988 | */
|
---|
989 | JSONSL_API
|
---|
990 | size_t jsonsl_util_unescape_ex(const char *in,
|
---|
991 | char *out,
|
---|
992 | size_t len,
|
---|
993 | const int toEscape[128],
|
---|
994 | unsigned *oflags,
|
---|
995 | jsonsl_error_t *err,
|
---|
996 | const char **errat)
|
---|
997 | {
|
---|
998 | const unsigned char *c = (const unsigned char*)in;
|
---|
999 | int in_escape = 0;
|
---|
1000 | size_t origlen = len;
|
---|
1001 | /* difference between the length of the input buffer and the output buffer */
|
---|
1002 | size_t ndiff = 0;
|
---|
1003 | if (oflags) {
|
---|
1004 | *oflags = 0;
|
---|
1005 | }
|
---|
1006 | #define UNESCAPE_BAIL(e,offset) \
|
---|
1007 | *err = JSONSL_ERROR_##e; \
|
---|
1008 | if (errat) { \
|
---|
1009 | *errat = (const char*)(c+ (ptrdiff_t)(offset)); \
|
---|
1010 | } \
|
---|
1011 | return 0;
|
---|
1012 |
|
---|
1013 | for (; len; len--, c++, out++) {
|
---|
1014 | unsigned int uesc_val[2];
|
---|
1015 | if (in_escape) {
|
---|
1016 | /* inside a previously ignored escape. Ignore */
|
---|
1017 | in_escape = 0;
|
---|
1018 | goto GT_ASSIGN;
|
---|
1019 | }
|
---|
1020 |
|
---|
1021 | if (*c != '\\') {
|
---|
1022 | /* Not an escape, so we don't care about this */
|
---|
1023 | goto GT_ASSIGN;
|
---|
1024 | }
|
---|
1025 |
|
---|
1026 | if (len < 2) {
|
---|
1027 | UNESCAPE_BAIL(ESCAPE_INVALID, 0);
|
---|
1028 | }
|
---|
1029 | if (!is_allowed_escape(c[1])) {
|
---|
1030 | UNESCAPE_BAIL(ESCAPE_INVALID, 1)
|
---|
1031 | }
|
---|
1032 | if ((toEscape[(unsigned char)c[1] & 0x7f] == 0 &&
|
---|
1033 | c[1] != '\\' && c[1] != '"')) {
|
---|
1034 | /* if we don't want to unescape this string, just continue with
|
---|
1035 | * the escape flag set
|
---|
1036 | */
|
---|
1037 | in_escape = 1;
|
---|
1038 | goto GT_ASSIGN;
|
---|
1039 | }
|
---|
1040 |
|
---|
1041 | if (c[1] != 'u') {
|
---|
1042 | /* simple skip-and-replace using pre-defined maps.
|
---|
1043 | * TODO: should the maps actually reflect the desired
|
---|
1044 | * replacement character in toEscape?
|
---|
1045 | */
|
---|
1046 | char esctmp = get_escape_equiv(c[1]);
|
---|
1047 | if (esctmp) {
|
---|
1048 | /* Check if there is a corresponding replacement */
|
---|
1049 | *out = esctmp;
|
---|
1050 | } else {
|
---|
1051 | /* Just gobble up the 'reverse-solidus' */
|
---|
1052 | *out = c[1];
|
---|
1053 | }
|
---|
1054 | len--;
|
---|
1055 | ndiff++;
|
---|
1056 | c++;
|
---|
1057 | /* do not assign, just continue */
|
---|
1058 | continue;
|
---|
1059 | }
|
---|
1060 |
|
---|
1061 | /* next == 'u' */
|
---|
1062 | if (len < 6) {
|
---|
1063 | /* Need at least six characters:
|
---|
1064 | * { [0] = '\\', [1] = 'u', [2] = 'f', [3] = 'f', [4] = 'f', [5] = 'f' }
|
---|
1065 | */
|
---|
1066 | UNESCAPE_BAIL(UESCAPE_TOOSHORT, -1);
|
---|
1067 | }
|
---|
1068 |
|
---|
1069 | if (sscanf_s((const char*)(c+2), "%02x%02x", uesc_val, uesc_val+1) != 2) {
|
---|
1070 | /* We treat the sequence as two octets */
|
---|
1071 | UNESCAPE_BAIL(UESCAPE_TOOSHORT, -1);
|
---|
1072 | }
|
---|
1073 |
|
---|
1074 | /* By now, we gobble up all the six bytes (current implied + 5 next
|
---|
1075 | * characters), and have at least four missing bytes from the output
|
---|
1076 | * buffer.
|
---|
1077 | */
|
---|
1078 | len -= 5;
|
---|
1079 | c += 5;
|
---|
1080 |
|
---|
1081 | ndiff += 4;
|
---|
1082 | if (uesc_val[0] == 0) {
|
---|
1083 | /* only one byte is extracted from the two
|
---|
1084 | * possible octets. Increment the diff counter by one.
|
---|
1085 | */
|
---|
1086 | *out = uesc_val[1];
|
---|
1087 | if (oflags && *(unsigned char*)out > 0x7f) {
|
---|
1088 | *oflags |= JSONSL_SPECIALf_NONASCII;
|
---|
1089 | }
|
---|
1090 | ndiff++;
|
---|
1091 | } else {
|
---|
1092 | *(out++) = uesc_val[0];
|
---|
1093 | *out = uesc_val[1];
|
---|
1094 | if (oflags && (uesc_val[0] > 0x7f || uesc_val[1] > 0x7f)) {
|
---|
1095 | *oflags |= JSONSL_SPECIALf_NONASCII;
|
---|
1096 | }
|
---|
1097 | }
|
---|
1098 | continue;
|
---|
1099 |
|
---|
1100 | /* Only reached by previous branches */
|
---|
1101 | GT_ASSIGN:
|
---|
1102 | *out = *c;
|
---|
1103 | }
|
---|
1104 | *err = JSONSL_ERROR_SUCCESS;
|
---|
1105 | return origlen - ndiff;
|
---|
1106 | }
|
---|
1107 |
|
---|
1108 | /**
|
---|
1109 | * Character Table definitions.
|
---|
1110 | * These were all generated via srcutil/genchartables.pl
|
---|
1111 | */
|
---|
1112 |
|
---|
1113 | /**
|
---|
1114 | * This table contains the beginnings of non-string
|
---|
1115 | * allowable (bareword) values.
|
---|
1116 | */
|
---|
1117 | static unsigned short Special_Table[0x100] = {
|
---|
1118 | /* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
|
---|
1119 | /* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2c */
|
---|
1120 | /* 0x2d */ JSONSL_SPECIALf_SIGNED /* - */, /* 0x2d */
|
---|
1121 | /* 0x2e */ 0,0, /* 0x2f */
|
---|
1122 | /* 0x30 */ JSONSL_SPECIALf_UNSIGNED /* 0 */, /* 0x30 */
|
---|
1123 | /* 0x31 */ JSONSL_SPECIALf_UNSIGNED /* 1 */, /* 0x31 */
|
---|
1124 | /* 0x32 */ JSONSL_SPECIALf_UNSIGNED /* 2 */, /* 0x32 */
|
---|
1125 | /* 0x33 */ JSONSL_SPECIALf_UNSIGNED /* 3 */, /* 0x33 */
|
---|
1126 | /* 0x34 */ JSONSL_SPECIALf_UNSIGNED /* 4 */, /* 0x34 */
|
---|
1127 | /* 0x35 */ JSONSL_SPECIALf_UNSIGNED /* 5 */, /* 0x35 */
|
---|
1128 | /* 0x36 */ JSONSL_SPECIALf_UNSIGNED /* 6 */, /* 0x36 */
|
---|
1129 | /* 0x37 */ JSONSL_SPECIALf_UNSIGNED /* 7 */, /* 0x37 */
|
---|
1130 | /* 0x38 */ JSONSL_SPECIALf_UNSIGNED /* 8 */, /* 0x38 */
|
---|
1131 | /* 0x39 */ JSONSL_SPECIALf_UNSIGNED /* 9 */, /* 0x39 */
|
---|
1132 | /* 0x3a */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x59 */
|
---|
1133 | /* 0x5a */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x65 */
|
---|
1134 | /* 0x66 */ JSONSL_SPECIALf_FALSE /* f */, /* 0x66 */
|
---|
1135 | /* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
|
---|
1136 | /* 0x6e */ JSONSL_SPECIALf_NULL /* n */, /* 0x6e */
|
---|
1137 | /* 0x6f */ 0,0,0,0,0, /* 0x73 */
|
---|
1138 | /* 0x74 */ JSONSL_SPECIALf_TRUE /* t */, /* 0x74 */
|
---|
1139 | /* 0x75 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x94 */
|
---|
1140 | /* 0x95 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb4 */
|
---|
1141 | /* 0xb5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd4 */
|
---|
1142 | /* 0xd5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf4 */
|
---|
1143 | /* 0xf5 */ 0,0,0,0,0,0,0,0,0,0 /* 0xfe */
|
---|
1144 | };
|
---|
1145 |
|
---|
1146 | /**
|
---|
1147 | * Contains characters which signal the termination of any of the 'special' bareword
|
---|
1148 | * values.
|
---|
1149 | */
|
---|
1150 | static int Special_Endings[0x100] = {
|
---|
1151 | /* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
|
---|
1152 | /* 0x09 */ 1 /* <TAB> */, /* 0x09 */
|
---|
1153 | /* 0x0a */ 1 /* <LF> */, /* 0x0a */
|
---|
1154 | /* 0x0b */ 0,0, /* 0x0c */
|
---|
1155 | /* 0x0d */ 1 /* <CR> */, /* 0x0d */
|
---|
1156 | /* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
|
---|
1157 | /* 0x20 */ 1 /* <SP> */, /* 0x20 */
|
---|
1158 | /* 0x21 */ 0, /* 0x21 */
|
---|
1159 | /* 0x22 */ 1 /* " */, /* 0x22 */
|
---|
1160 | /* 0x23 */ 0,0,0,0,0,0,0,0,0, /* 0x2b */
|
---|
1161 | /* 0x2c */ 1 /* , */, /* 0x2c */
|
---|
1162 | /* 0x2d */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x39 */
|
---|
1163 | /* 0x3a */ 1 /* : */, /* 0x3a */
|
---|
1164 | /* 0x3b */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5a */
|
---|
1165 | /* 0x5b */ 1 /* [ */, /* 0x5b */
|
---|
1166 | /* 0x5c */ 1 /* \ */, /* 0x5c */
|
---|
1167 | /* 0x5d */ 1 /* ] */, /* 0x5d */
|
---|
1168 | /* 0x5e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x7a */
|
---|
1169 | /* 0x7b */ 1 /* { */, /* 0x7b */
|
---|
1170 | /* 0x7c */ 0, /* 0x7c */
|
---|
1171 | /* 0x7d */ 1 /* } */, /* 0x7d */
|
---|
1172 | /* 0x7e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x9d */
|
---|
1173 | /* 0x9e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xbd */
|
---|
1174 | /* 0xbe */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xdd */
|
---|
1175 | /* 0xde */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xfd */
|
---|
1176 | /* 0xfe */ 0 /* 0xfe */
|
---|
1177 | };
|
---|
1178 |
|
---|
1179 | /**
|
---|
1180 | * This table contains entries for the allowed whitespace as per RFC 4627
|
---|
1181 | */
|
---|
1182 | static int Allowed_Whitespace[0x100] = {
|
---|
1183 | /* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
|
---|
1184 | /* 0x09 */ 1 /* <TAB> */, /* 0x09 */
|
---|
1185 | /* 0x0a */ 1 /* <LF> */, /* 0x0a */
|
---|
1186 | /* 0x0b */ 0,0, /* 0x0c */
|
---|
1187 | /* 0x0d */ 1 /* <CR> */, /* 0x0d */
|
---|
1188 | /* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
|
---|
1189 | /* 0x20 */ 1 /* <SP> */, /* 0x20 */
|
---|
1190 | /* 0x21 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x40 */
|
---|
1191 | /* 0x41 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x60 */
|
---|
1192 | /* 0x61 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x80 */
|
---|
1193 | /* 0x81 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xa0 */
|
---|
1194 | /* 0xa1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xc0 */
|
---|
1195 | /* 0xc1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xe0 */
|
---|
1196 | /* 0xe1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 /* 0xfe */
|
---|
1197 | };
|
---|
1198 |
|
---|
1199 | /**
|
---|
1200 | * Allowable two-character 'common' escapes:
|
---|
1201 | */
|
---|
1202 | static int Allowed_Escapes[0x100] = {
|
---|
1203 | /* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
|
---|
1204 | /* 0x20 */ 0,0, /* 0x21 */
|
---|
1205 | /* 0x22 */ 1 /* <"> */, /* 0x22 */
|
---|
1206 | /* 0x23 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2e */
|
---|
1207 | /* 0x2f */ 1 /* </> */, /* 0x2f */
|
---|
1208 | /* 0x30 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x4f */
|
---|
1209 | /* 0x50 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5b */
|
---|
1210 | /* 0x5c */ 1 /* <\> */, /* 0x5c */
|
---|
1211 | /* 0x5d */ 0,0,0,0,0, /* 0x61 */
|
---|
1212 | /* 0x62 */ 1 /* <b> */, /* 0x62 */
|
---|
1213 | /* 0x63 */ 0,0,0, /* 0x65 */
|
---|
1214 | /* 0x66 */ 1 /* <f> */, /* 0x66 */
|
---|
1215 | /* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
|
---|
1216 | /* 0x6e */ 1 /* <n> */, /* 0x6e */
|
---|
1217 | /* 0x6f */ 0,0,0, /* 0x71 */
|
---|
1218 | /* 0x72 */ 1 /* <r> */, /* 0x72 */
|
---|
1219 | /* 0x73 */ 0, /* 0x73 */
|
---|
1220 | /* 0x74 */ 1 /* <t> */, /* 0x74 */
|
---|
1221 | /* 0x75 */ 1 /* <u> */, /* 0x75 */
|
---|
1222 | /* 0x76 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x95 */
|
---|
1223 | /* 0x96 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb5 */
|
---|
1224 | /* 0xb6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd5 */
|
---|
1225 | /* 0xd6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf5 */
|
---|
1226 | /* 0xf6 */ 0,0,0,0,0,0,0,0,0, /* 0xfe */
|
---|
1227 | };
|
---|
1228 |
|
---|
1229 | /**
|
---|
1230 | * This table contains the _values_ for a given (single) escaped character.
|
---|
1231 | */
|
---|
1232 | static unsigned char Escape_Equivs[0x100] = {
|
---|
1233 | /* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
|
---|
1234 | /* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x3f */
|
---|
1235 | /* 0x40 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5f */
|
---|
1236 | /* 0x60 */ 0,0, /* 0x61 */
|
---|
1237 | /* 0x62 */ 8 /* <b> */, /* 0x62 */
|
---|
1238 | /* 0x63 */ 0,0,0, /* 0x65 */
|
---|
1239 | /* 0x66 */ 12 /* <f> */, /* 0x66 */
|
---|
1240 | /* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
|
---|
1241 | /* 0x6e */ 10 /* <n> */, /* 0x6e */
|
---|
1242 | /* 0x6f */ 0,0,0, /* 0x71 */
|
---|
1243 | /* 0x72 */ 13 /* <r> */, /* 0x72 */
|
---|
1244 | /* 0x73 */ 0, /* 0x73 */
|
---|
1245 | /* 0x74 */ 9 /* <t> */, /* 0x74 */
|
---|
1246 | /* 0x75 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x94 */
|
---|
1247 | /* 0x95 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb4 */
|
---|
1248 | /* 0xb5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd4 */
|
---|
1249 | /* 0xd5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf4 */
|
---|
1250 | /* 0xf5 */ 0,0,0,0,0,0,0,0,0,0 /* 0xfe */
|
---|
1251 | };
|
---|
1252 |
|
---|
1253 | /* Definitions of above-declared static functions */
|
---|
1254 | static char get_escape_equiv(unsigned c) {
|
---|
1255 | return Escape_Equivs[c & 0xff];
|
---|
1256 | }
|
---|
1257 | static unsigned extract_special(unsigned c) {
|
---|
1258 | return Special_Table[c & 0xff];
|
---|
1259 | }
|
---|
1260 | static int is_special_end(unsigned c) {
|
---|
1261 | return Special_Endings[c & 0xff];
|
---|
1262 | }
|
---|
1263 | static int is_allowed_whitespace(unsigned c) {
|
---|
1264 | return c == ' ' || Allowed_Whitespace[c & 0xff];
|
---|
1265 | }
|
---|
1266 | static int is_allowed_escape(unsigned c) {
|
---|
1267 | return Allowed_Escapes[c & 0xff];
|
---|
1268 | }
|
---|