Mercurial > hg > dbutils / file comparison

comparison: ucx/json.c

ucx/json.c

changeset 11
0aa8cbd7912e
child 16
04c9f8d8f03b
equal deleted inserted replaced
10:80f9d007cb52 11:0aa8cbd7912e
1 /*
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
3 *
4 * Copyright 2024 Mike Becker, Olaf Wintermann All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions are met:
8 *
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 *
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
17 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
20 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
21 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
22 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
23 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
24 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
25 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
26 * POSSIBILITY OF SUCH DAMAGE.
27 */
28
29 #include "cx/json.h"
30 #include "cx/compare.h"
31
32 #include <string.h>
33 #include <ctype.h>
34 #include <assert.h>
35 #include <stdio.h>
36 #include <errno.h>
37 #include <inttypes.h>
38
39 /*
40 * RFC 8259
41 * https://tools.ietf.org/html/rfc8259
42 */
43
44 static CxJsonValue cx_json_value_nothing = {.type = CX_JSON_NOTHING};
45
46 static int json_cmp_objvalue(const void *l, const void *r) {
47 const CxJsonObjValue *left = l;
48 const CxJsonObjValue *right = r;
49 return cx_strcmp(cx_strcast(left->name), cx_strcast(right->name));
50 }
51
52 static CxJsonObjValue *json_find_objvalue(const CxJsonValue *obj, cxstring name) {
53 assert(obj->type == CX_JSON_OBJECT);
54 CxJsonObjValue kv_dummy;
55 kv_dummy.name = cx_mutstrn((char*) name.ptr, name.length);
56 size_t index = cx_array_binary_search(
57 obj->value.object.values,
58 obj->value.object.values_size,
59 sizeof(CxJsonObjValue),
60 &kv_dummy,
61 json_cmp_objvalue
62 );
63 if (index == obj->value.object.values_size) {
64 return NULL;
65 } else {
66 return &obj->value.object.values[index];
67 }
68 }
69
70 static int json_add_objvalue(CxJsonValue *objv, CxJsonObjValue member) {
71 assert(objv->type == CX_JSON_OBJECT);
72 const CxAllocator * const al = objv->allocator;
73 CxJsonObject *obj = &(objv->value.object);
74
75 // determine the index where we need to insert the new member
76 size_t index = cx_array_binary_search_sup(
77 obj->values,
78 obj->values_size,
79 sizeof(CxJsonObjValue),
80 &member, json_cmp_objvalue
81 );
82
83 // is the name already present?
84 if (index < obj->values_size && 0 == json_cmp_objvalue(&member, &obj->values[index])) {
85 // free the original value
86 cx_strfree_a(al, &obj->values[index].name);
87 cxJsonValueFree(obj->values[index].value);
88 // replace the item
89 obj->values[index] = member;
90
91 // nothing more to do
92 return 0;
93 }
94
95 // determine the old capacity and reserve for one more element
96 CxArrayReallocator arealloc = cx_array_reallocator(al, NULL);
97 size_t oldcap = obj->values_capacity;
98 if (cx_array_simple_reserve_a(&arealloc, obj->values, 1)) return 1;
99
100 // check the new capacity, if we need to realloc the index array
101 size_t newcap = obj->values_capacity;
102 if (newcap > oldcap) {
103 if (cxReallocateArray(al, &obj->indices, newcap, sizeof(size_t))) {
104 return 1;
105 }
106 }
107
108 // check if append or insert
109 if (index < obj->values_size) {
110 // move the other elements
111 memmove(
112 &obj->values[index+1],
113 &obj->values[index],
114 (obj->values_size - index) * sizeof(CxJsonObjValue)
115 );
116 // increase indices for the moved elements
117 for (size_t i = 0; i < obj->values_size ; i++) {
118 if (obj->indices[i] >= index) {
119 obj->indices[i]++;
120 }
121 }
122 }
123
124 // insert the element and set the index
125 obj->values[index] = member;
126 obj->indices[obj->values_size] = index;
127 obj->values_size++;
128
129 return 0;
130 }
131
132 static void token_destroy(CxJsonToken *token) {
133 if (token->allocated) {
134 cx_strfree(&token->content);
135 }
136 }
137
138 static int num_isexp(const char *content, size_t length, size_t pos) {
139 if (pos >= length) {
140 return 0;
141 }
142
143 int ok = 0;
144 for (size_t i = pos; i < length; i++) {
145 char c = content[i];
146 if (isdigit(c)) {
147 ok = 1;
148 } else if (i == pos) {
149 if (!(c == '+' || c == '-')) {
150 return 0;
151 }
152 } else {
153 return 0;
154 }
155 }
156
157 return ok;
158 }
159
160 static CxJsonTokenType token_numbertype(const char *content, size_t length) {
161 if (length == 0) return CX_JSON_TOKEN_ERROR;
162
163 if (content[0] != '-' && !isdigit(content[0])) {
164 return CX_JSON_TOKEN_ERROR;
165 }
166
167 CxJsonTokenType type = CX_JSON_TOKEN_INTEGER;
168 for (size_t i = 1; i < length; i++) {
169 if (content[i] == '.') {
170 if (type == CX_JSON_TOKEN_NUMBER) {
171 return CX_JSON_TOKEN_ERROR; // more than one decimal separator
172 }
173 type = CX_JSON_TOKEN_NUMBER;
174 } else if (content[i] == 'e' || content[i] == 'E') {
175 return num_isexp(content, length, i + 1) ? CX_JSON_TOKEN_NUMBER : CX_JSON_TOKEN_ERROR;
176 } else if (!isdigit(content[i])) {
177 return CX_JSON_TOKEN_ERROR; // char is not a digit, decimal separator or exponent sep
178 }
179 }
180
181 return type;
182 }
183
184 static CxJsonToken token_create(CxJson *json, bool isstring, size_t start, size_t end) {
185 cxmutstr str = cx_mutstrn(json->buffer.space + start, end - start);
186 bool allocated = false;
187 if (json->uncompleted.tokentype != CX_JSON_NO_TOKEN) {
188 allocated = true;
189 str = cx_strcat_m(json->uncompleted.content, 1, str);
190 if (str.ptr == NULL) { // LCOV_EXCL_START
191 return (CxJsonToken){CX_JSON_NO_TOKEN, false, {NULL, 0}};
192 } // LCOV_EXCL_STOP
193 }
194 json->uncompleted = (CxJsonToken){0};
195 CxJsonTokenType ttype;
196 if (isstring) {
197 ttype = CX_JSON_TOKEN_STRING;
198 } else {
199 cxstring s = cx_strcast(str);
200 if (!cx_strcmp(s, CX_STR("true")) || !cx_strcmp(s, CX_STR("false"))
201 || !cx_strcmp(s, CX_STR("null"))) {
202 ttype = CX_JSON_TOKEN_LITERAL;
203 } else {
204 ttype = token_numbertype(str.ptr, str.length);
205 }
206 }
207 if (ttype == CX_JSON_TOKEN_ERROR) {
208 if (allocated) {
209 cx_strfree(&str);
210 }
211 return (CxJsonToken){CX_JSON_TOKEN_ERROR, false, {NULL, 0}};
212 }
213 return (CxJsonToken){ttype, allocated, str};
214 }
215
216 static CxJsonTokenType char2ttype(char c) {
217 switch (c) {
218 case '[': {
219 return CX_JSON_TOKEN_BEGIN_ARRAY;
220 }
221 case '{': {
222 return CX_JSON_TOKEN_BEGIN_OBJECT;
223 }
224 case ']': {
225 return CX_JSON_TOKEN_END_ARRAY;
226 }
227 case '}': {
228 return CX_JSON_TOKEN_END_OBJECT;
229 }
230 case ':': {
231 return CX_JSON_TOKEN_NAME_SEPARATOR;
232 }
233 case ',': {
234 return CX_JSON_TOKEN_VALUE_SEPARATOR;
235 }
236 case '"': {
237 return CX_JSON_TOKEN_STRING;
238 }
239 default: {
240 if (isspace(c)) {
241 return CX_JSON_TOKEN_SPACE;
242 }
243 }
244 }
245 return CX_JSON_NO_TOKEN;
246 }
247
248 static enum cx_json_status token_parse_next(CxJson *json, CxJsonToken *result) {
249 // check if there is data in the buffer
250 if (cxBufferEof(&json->buffer)) {
251 return json->uncompleted.tokentype == CX_JSON_NO_TOKEN ?
252 CX_JSON_NO_DATA : CX_JSON_INCOMPLETE_DATA;
253 }
254
255 // current token type and start index
256 CxJsonTokenType ttype = json->uncompleted.tokentype;
257 size_t token_start = json->buffer.pos;
258
259 for (size_t i = json->buffer.pos; i < json->buffer.size; i++) {
260 char c = json->buffer.space[i];
261 if (ttype != CX_JSON_TOKEN_STRING) {
262 // currently non-string token
263 CxJsonTokenType ctype = char2ttype(c); // start of new token?
264 if (ttype == CX_JSON_NO_TOKEN) {
265 if (ctype == CX_JSON_TOKEN_SPACE) {
266 json->buffer.pos++;
267 continue;
268 } else if (ctype == CX_JSON_TOKEN_STRING) {
269 // begin string
270 ttype = CX_JSON_TOKEN_STRING;
271 token_start = i;
272 } else if (ctype != CX_JSON_NO_TOKEN) {
273 // single-char token
274 json->buffer.pos = i + 1;
275 *result = (CxJsonToken){ctype, false, {NULL, 0}};
276 return CX_JSON_NO_ERROR;
277 } else {
278 ttype = CX_JSON_TOKEN_LITERAL; // number or literal
279 token_start = i;
280 }
281 } else {
282 // finish token
283 if (ctype != CX_JSON_NO_TOKEN) {
284 *result = token_create(json, false, token_start, i);
285 if (result->tokentype == CX_JSON_NO_TOKEN) {
286 return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
287 }
288 if (result->tokentype == CX_JSON_TOKEN_ERROR) {
289 return CX_JSON_FORMAT_ERROR_NUMBER;
290 }
291 json->buffer.pos = i;
292 return CX_JSON_NO_ERROR;
293 }
294 }
295 } else {
296 // currently inside a string
297 if (json->tokenizer_escape) {
298 json->tokenizer_escape = false;
299 } else {
300 if (c == '"') {
301 *result = token_create(json, true, token_start, i + 1);
302 if (result->tokentype == CX_JSON_NO_TOKEN) {
303 return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
304 }
305 json->buffer.pos = i + 1;
306 return CX_JSON_NO_ERROR;
307 } else if (c == '\\') {
308 json->tokenizer_escape = true;
309 }
310 }
311 }
312 }
313
314 if (ttype != CX_JSON_NO_TOKEN) {
315 // uncompleted token
316 size_t uncompleted_len = json->buffer.size - token_start;
317 if (json->uncompleted.tokentype == CX_JSON_NO_TOKEN) {
318 // current token is uncompleted
319 // save current token content
320 CxJsonToken uncompleted = {
321 ttype, true,
322 cx_strdup(cx_strn(json->buffer.space + token_start, uncompleted_len))
323 };
324 if (uncompleted.content.ptr == NULL) {
325 return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
326 }
327 json->uncompleted = uncompleted;
328 } else {
329 // previously we also had an uncompleted token
330 // combine the uncompleted token with the current token
331 assert(json->uncompleted.allocated);
332 cxmutstr str = cx_strcat_m(json->uncompleted.content, 1,
333 cx_strn(json->buffer.space + token_start, uncompleted_len));
334 if (str.ptr == NULL) {
335 return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
336 }
337 json->uncompleted.content = str;
338 }
339 // advance the buffer position - we saved the stuff in the uncompleted token
340 json->buffer.pos += uncompleted_len;
341 }
342
343 return CX_JSON_INCOMPLETE_DATA;
344 }
345
346 static cxmutstr unescape_string(const CxAllocator *a, cxmutstr str) {
347 // TODO: support more escape sequences
348 // we know that the unescaped string will be shorter by at least 2 chars
349 cxmutstr result;
350 result.length = 0;
351 result.ptr = cxMalloc(a, str.length - 1);
352 if (result.ptr == NULL) return result; // LCOV_EXCL_LINE
353
354 bool u = false;
355 for (size_t i = 1; i < str.length - 1; i++) {
356 char c = str.ptr[i];
357 if (u) {
358 u = false;
359 if (c == 'n') {
360 c = '\n';
361 } else if (c == 't') {
362 c = '\t';
363 }
364 result.ptr[result.length++] = c;
365 } else {
366 if (c == '\\') {
367 u = true;
368 } else {
369 result.ptr[result.length++] = c;
370 }
371 }
372 }
373 result.ptr[result.length] = 0;
374
375 return result;
376 }
377
378 static CxJsonValue* create_json_value(CxJson *json, CxJsonValueType type) {
379 CxJsonValue *v = cxCalloc(json->allocator, 1, sizeof(CxJsonValue));
380 if (v == NULL) return NULL; // LCOV_EXCL_LINE
381
382 // initialize the value
383 v->type = type;
384 v->allocator = json->allocator;
385 if (type == CX_JSON_ARRAY) {
386 cx_array_initialize_a(json->allocator, v->value.array.array, 16);
387 if (v->value.array.array == NULL) goto create_json_value_exit_error; // LCOV_EXCL_LINE
388 } else if (type == CX_JSON_OBJECT) {
389 cx_array_initialize_a(json->allocator, v->value.object.values, 16);
390 v->value.object.indices = cxCalloc(json->allocator, 16, sizeof(size_t));
391 if (v->value.object.values == NULL ||
392 v->value.object.indices == NULL)
393 goto create_json_value_exit_error; // LCOV_EXCL_LINE
394 }
395
396 // add the new value to a possible parent
397 if (json->vbuf_size > 0) {
398 CxJsonValue *parent = json->vbuf[json->vbuf_size - 1];
399 assert(parent != NULL);
400 if (parent->type == CX_JSON_ARRAY) {
401 CxArrayReallocator value_realloc = cx_array_reallocator(json->allocator, NULL);
402 if (cx_array_simple_add_a(&value_realloc, parent->value.array.array, v)) {
403 goto create_json_value_exit_error; // LCOV_EXCL_LINE
404 }
405 } else if (parent->type == CX_JSON_OBJECT) {
406 // the member was already created after parsing the name
407 assert(json->uncompleted_member.name.ptr != NULL);
408 json->uncompleted_member.value = v;
409 if (json_add_objvalue(parent, json->uncompleted_member)) {
410 goto create_json_value_exit_error; // LCOV_EXCL_LINE
411 }
412 json->uncompleted_member.name = (cxmutstr) {NULL, 0};
413 } else {
414 assert(false); // LCOV_EXCL_LINE
415 }
416 }
417
418 // add the new value to the stack, if it is an array or object
419 if (type == CX_JSON_ARRAY || type == CX_JSON_OBJECT) {
420 CxArrayReallocator vbuf_realloc = cx_array_reallocator(NULL, json->vbuf_internal);
421 if (cx_array_simple_add_a(&vbuf_realloc, json->vbuf, v)) {
422 goto create_json_value_exit_error; // LCOV_EXCL_LINE
423 }
424 }
425
426 // if currently no value is parsed, this is now the value of interest
427 if (json->parsed == NULL) {
428 json->parsed = v;
429 }
430
431 return v;
432 // LCOV_EXCL_START
433 create_json_value_exit_error:
434 cxJsonValueFree(v);
435 return NULL;
436 // LCOV_EXCL_STOP
437 }
438
439 #define JP_STATE_VALUE_BEGIN 0
440 #define JP_STATE_VALUE_END 10
441 #define JP_STATE_VALUE_BEGIN_OBJ 1
442 #define JP_STATE_OBJ_SEP_OR_CLOSE 11
443 #define JP_STATE_VALUE_BEGIN_AR 2
444 #define JP_STATE_ARRAY_SEP_OR_CLOSE 12
445 #define JP_STATE_OBJ_NAME_OR_CLOSE 5
446 #define JP_STATE_OBJ_NAME 6
447 #define JP_STATE_OBJ_COLON 7
448
449 void cxJsonInit(CxJson *json, const CxAllocator *allocator) {
450 if (allocator == NULL) {
451 allocator = cxDefaultAllocator;
452 }
453
454 memset(json, 0, sizeof(CxJson));
455 json->allocator = allocator;
456
457 json->states = json->states_internal;
458 json->states_capacity = cx_nmemb(json->states_internal);
459 json->states[0] = JP_STATE_VALUE_BEGIN;
460 json->states_size = 1;
461
462 json->vbuf = json->vbuf_internal;
463 json->vbuf_capacity = cx_nmemb(json->vbuf_internal);
464 }
465
466 void cxJsonDestroy(CxJson *json) {
467 cxBufferDestroy(&json->buffer);
468 if (json->states != json->states_internal) {
469 free(json->states);
470 }
471 if (json->vbuf != json->vbuf_internal) {
472 free(json->vbuf);
473 }
474 cxJsonValueFree(json->parsed);
475 json->parsed = NULL;
476 if (json->uncompleted_member.name.ptr != NULL) {
477 cx_strfree_a(json->allocator, &json->uncompleted_member.name);
478 json->uncompleted_member = (CxJsonObjValue){{NULL, 0}, NULL};
479 }
480 }
481
482 int cxJsonFilln(CxJson *json, const char *buf, size_t size) {
483 if (cxBufferEof(&json->buffer)) {
484 // reinitialize the buffer
485 cxBufferDestroy(&json->buffer);
486 cxBufferInit(&json->buffer, (char*) buf, size,
487 NULL, CX_BUFFER_AUTO_EXTEND | CX_BUFFER_COPY_ON_WRITE);
488 json->buffer.size = size;
489 return 0;
490 } else {
491 return size != cxBufferAppend(buf, 1, size, &json->buffer);
492 }
493 }
494
495 static void json_add_state(CxJson *json, int state) {
496 // we have guaranteed the necessary space with cx_array_simple_reserve()
497 // therefore, we can safely add the state in the simplest way possible
498 json->states[json->states_size++] = state;
499 }
500
501 #define return_rec(code) \
502 token_destroy(&token); \
503 return code
504
505 static enum cx_json_status json_parse(CxJson *json) {
506 // Reserve a pointer for a possibly read value
507 CxJsonValue *vbuf = NULL;
508
509 // grab the next token
510 CxJsonToken token;
511 {
512 enum cx_json_status ret = token_parse_next(json, &token);
513 if (ret != CX_JSON_NO_ERROR) {
514 return ret;
515 }
516 }
517
518 // pop the current state
519 assert(json->states_size > 0);
520 int state = json->states[--json->states_size];
521
522 // guarantee that at least two more states fit on the stack
523 CxArrayReallocator state_realloc = cx_array_reallocator(NULL, json->states_internal);
524 if (cx_array_simple_reserve_a(&state_realloc, json->states, 2)) {
525 return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
526 }
527
528
529 // 0 JP_STATE_VALUE_BEGIN value begin
530 // 10 JP_STATE_VALUE_END expect value end
531 // 1 JP_STATE_VALUE_BEGIN_OBJ value begin (inside object)
532 // 11 JP_STATE_OBJ_SEP_OR_CLOSE object, expect separator, objclose
533 // 2 JP_STATE_VALUE_BEGIN_AR value begin (inside array)
534 // 12 JP_STATE_ARRAY_SEP_OR_CLOSE array, expect separator or arrayclose
535 // 5 JP_STATE_OBJ_NAME_OR_CLOSE object, expect name or objclose
536 // 6 JP_STATE_OBJ_NAME object, expect name
537 // 7 JP_STATE_OBJ_COLON object, expect ':'
538
539 if (state < 3) {
540 // push expected end state to the stack
541 json_add_state(json, 10 + state);
542 switch (token.tokentype) {
543 case CX_JSON_TOKEN_BEGIN_ARRAY: {
544 if (create_json_value(json, CX_JSON_ARRAY) == NULL) {
545 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
546 }
547 json_add_state(json, JP_STATE_VALUE_BEGIN_AR);
548 return_rec(CX_JSON_NO_ERROR);
549 }
550 case CX_JSON_TOKEN_BEGIN_OBJECT: {
551 if (create_json_value(json, CX_JSON_OBJECT) == NULL) {
552 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
553 }
554 json_add_state(json, JP_STATE_OBJ_NAME_OR_CLOSE);
555 return_rec(CX_JSON_NO_ERROR);
556 }
557 case CX_JSON_TOKEN_STRING: {
558 if ((vbuf = create_json_value(json, CX_JSON_STRING)) == NULL) {
559 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
560 }
561 cxmutstr str = unescape_string(json->allocator, token.content);
562 if (str.ptr == NULL) {
563 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
564 }
565 vbuf->value.string = str;
566 return_rec(CX_JSON_NO_ERROR);
567 }
568 case CX_JSON_TOKEN_INTEGER:
569 case CX_JSON_TOKEN_NUMBER: {
570 int type = token.tokentype == CX_JSON_TOKEN_INTEGER ? CX_JSON_INTEGER : CX_JSON_NUMBER;
571 if (NULL == (vbuf = create_json_value(json, type))) {
572 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
573 }
574 if (type == CX_JSON_INTEGER) {
575 if (cx_strtoi64(token.content, &vbuf->value.integer, 10)) {
576 return_rec(CX_JSON_FORMAT_ERROR_NUMBER);
577 }
578 } else {
579 if (cx_strtod(token.content, &vbuf->value.number)) {
580 return_rec(CX_JSON_FORMAT_ERROR_NUMBER);
581 }
582 }
583 return_rec(CX_JSON_NO_ERROR);
584 }
585 case CX_JSON_TOKEN_LITERAL: {
586 if ((vbuf = create_json_value(json, CX_JSON_LITERAL)) == NULL) {
587 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
588 }
589 if (0 == cx_strcmp(cx_strcast(token.content), cx_str("true"))) {
590 vbuf->value.literal = CX_JSON_TRUE;
591 } else if (0 == cx_strcmp(cx_strcast(token.content), cx_str("false"))) {
592 vbuf->value.literal = CX_JSON_FALSE;
593 } else {
594 vbuf->value.literal = CX_JSON_NULL;
595 }
596 return_rec(CX_JSON_NO_ERROR);
597 }
598 default: {
599 return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
600 }
601 }
602 } else if (state == JP_STATE_ARRAY_SEP_OR_CLOSE) {
603 // expect ',' or ']'
604 if (token.tokentype == CX_JSON_TOKEN_VALUE_SEPARATOR) {
605 json_add_state(json, JP_STATE_VALUE_BEGIN_AR);
606 return_rec(CX_JSON_NO_ERROR);
607 } else if (token.tokentype == CX_JSON_TOKEN_END_ARRAY) {
608 // discard the array from the value buffer
609 json->vbuf_size--;
610 return_rec(CX_JSON_NO_ERROR);
611 } else {
612 return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
613 }
614 } else if (state == JP_STATE_OBJ_NAME_OR_CLOSE || state == JP_STATE_OBJ_NAME) {
615 if (state == JP_STATE_OBJ_NAME_OR_CLOSE && token.tokentype == CX_JSON_TOKEN_END_OBJECT) {
616 // discard the obj from the value buffer
617 json->vbuf_size--;
618 return_rec(CX_JSON_NO_ERROR);
619 } else {
620 // expect string
621 if (token.tokentype != CX_JSON_TOKEN_STRING) {
622 return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
623 }
624
625 // add new entry
626 cxmutstr name = unescape_string(json->allocator, token.content);
627 if (name.ptr == NULL) {
628 return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
629 }
630 assert(json->uncompleted_member.name.ptr == NULL);
631 json->uncompleted_member.name = name;
632 assert(json->vbuf_size > 0);
633
634 // next state
635 json_add_state(json, JP_STATE_OBJ_COLON);
636 return_rec(CX_JSON_NO_ERROR);
637 }
638 } else if (state == JP_STATE_OBJ_COLON) {
639 // expect ':'
640 if (token.tokentype != CX_JSON_TOKEN_NAME_SEPARATOR) {
641 return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
642 }
643 // next state
644 json_add_state(json, JP_STATE_VALUE_BEGIN_OBJ);
645 return_rec(CX_JSON_NO_ERROR);
646 } else if (state == JP_STATE_OBJ_SEP_OR_CLOSE) {
647 // expect ',' or '}'
648 if (token.tokentype == CX_JSON_TOKEN_VALUE_SEPARATOR) {
649 json_add_state(json, JP_STATE_OBJ_NAME);
650 return_rec(CX_JSON_NO_ERROR);
651 } else if (token.tokentype == CX_JSON_TOKEN_END_OBJECT) {
652 // discard the obj from the value buffer
653 json->vbuf_size--;
654 return_rec(CX_JSON_NO_ERROR);
655 } else {
656 return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
657 }
658 } else {
659 // should be unreachable
660 assert(false);
661 return_rec(-1);
662 }
663 }
664
665 CxJsonStatus cxJsonNext(CxJson *json, CxJsonValue **value) {
666 // check if buffer has been filled
667 if (json->buffer.space == NULL) {
668 return CX_JSON_NULL_DATA;
669 }
670
671 // initialize output value
672 *value = &cx_json_value_nothing;
673
674 // parse data
675 CxJsonStatus result;
676 do {
677 result = json_parse(json);
678 if (result == CX_JSON_NO_ERROR && json->states_size == 1) {
679 // final state reached
680 assert(json->states[0] == JP_STATE_VALUE_END);
681 assert(json->vbuf_size == 0);
682
683 // write output value
684 *value = json->parsed;
685 json->parsed = NULL;
686
687 // re-initialize state machine
688 json->states[0] = JP_STATE_VALUE_BEGIN;
689
690 return CX_JSON_NO_ERROR;
691 }
692 } while (result == CX_JSON_NO_ERROR);
693
694 // the parser might think there is no data
695 // but when we did not reach the final state,
696 // we know that there must be more to come
697 if (result == CX_JSON_NO_DATA && json->states_size > 1) {
698 return CX_JSON_INCOMPLETE_DATA;
699 }
700
701 return result;
702 }
703
704 void cxJsonValueFree(CxJsonValue *value) {
705 if (value == NULL || value->type == CX_JSON_NOTHING) return;
706 switch (value->type) {
707 case CX_JSON_OBJECT: {
708 CxJsonObject obj = value->value.object;
709 for (size_t i = 0; i < obj.values_size; i++) {
710 cxJsonValueFree(obj.values[i].value);
711 cx_strfree_a(value->allocator, &obj.values[i].name);
712 }
713 cxFree(value->allocator, obj.values);
714 cxFree(value->allocator, obj.indices);
715 break;
716 }
717 case CX_JSON_ARRAY: {
718 CxJsonArray array = value->value.array;
719 for (size_t i = 0; i < array.array_size; i++) {
720 cxJsonValueFree(array.array[i]);
721 }
722 cxFree(value->allocator, array.array);
723 break;
724 }
725 case CX_JSON_STRING: {
726 cxFree(value->allocator, value->value.string.ptr);
727 break;
728 }
729 default: {
730 break;
731 }
732 }
733 cxFree(value->allocator, value);
734 }
735
736 CxJsonValue* cxJsonCreateObj(const CxAllocator* allocator) {
737 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
738 if (v == NULL) return NULL;
739 v->allocator = allocator;
740 v->type = CX_JSON_OBJECT;
741 cx_array_initialize_a(allocator, v->value.object.values, 16);
742 if (v->value.object.values == NULL) { // LCOV_EXCL_START
743 cxFree(allocator, v);
744 return NULL;
745 // LCOV_EXCL_STOP
746 }
747 v->value.object.indices = cxCalloc(allocator, 16, sizeof(size_t));
748 if (v->value.object.indices == NULL) { // LCOV_EXCL_START
749 cxFree(allocator, v->value.object.values);
750 cxFree(allocator, v);
751 return NULL;
752 // LCOV_EXCL_STOP
753 }
754 return v;
755 }
756
757 CxJsonValue* cxJsonCreateArr(const CxAllocator* allocator) {
758 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
759 if (v == NULL) return NULL;
760 v->allocator = allocator;
761 v->type = CX_JSON_ARRAY;
762 cx_array_initialize_a(allocator, v->value.array.array, 16);
763 if (v->value.array.array == NULL) { cxFree(allocator, v); return NULL; }
764 return v;
765 }
766
767 CxJsonValue* cxJsonCreateNumber(const CxAllocator* allocator, double num) {
768 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
769 if (v == NULL) return NULL;
770 v->allocator = allocator;
771 v->type = CX_JSON_NUMBER;
772 v->value.number = num;
773 return v;
774 }
775
776 CxJsonValue* cxJsonCreateInteger(const CxAllocator* allocator, int64_t num) {
777 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
778 if (v == NULL) return NULL;
779 v->allocator = allocator;
780 v->type = CX_JSON_INTEGER;
781 v->value.integer = num;
782 return v;
783 }
784
785 CxJsonValue* cxJsonCreateString(const CxAllocator* allocator, const char* str) {
786 return cxJsonCreateCxString(allocator, cx_str(str));
787 }
788
789 CxJsonValue* cxJsonCreateCxString(const CxAllocator* allocator, cxstring str) {
790 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
791 if (v == NULL) return NULL;
792 v->allocator = allocator;
793 v->type = CX_JSON_STRING;
794 cxmutstr s = cx_strdup_a(allocator, str);
795 if (s.ptr == NULL) { cxFree(allocator, v); return NULL; }
796 v->value.string = s;
797 return v;
798 }
799
800 CxJsonValue* cxJsonCreateLiteral(const CxAllocator* allocator, CxJsonLiteral lit) {
801 CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
802 if (v == NULL) return NULL;
803 v->allocator = allocator;
804 v->type = CX_JSON_LITERAL;
805 v->value.literal = lit;
806 return v;
807 }
808
809 // LCOV_EXCL_START
810 // never called as long as malloc() does not return NULL
811 static void cx_json_arr_free_temp(CxJsonValue** values, size_t count) {
812 for (size_t i = 0; i < count; i++) {
813 if (values[i] == NULL) break;
814 cxJsonValueFree(values[i]);
815 }
816 free(values);
817 }
818 // LCOV_EXCL_STOP
819
820 int cxJsonArrAddNumbers(CxJsonValue* arr, const double* num, size_t count) {
821 CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
822 if (values == NULL) return -1;
823 for (size_t i = 0; i < count; i++) {
824 values[i] = cxJsonCreateNumber(arr->allocator, num[i]);
825 if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
826 }
827 int ret = cxJsonArrAddValues(arr, values, count);
828 free(values);
829 return ret;
830 }
831
832 int cxJsonArrAddIntegers(CxJsonValue* arr, const int64_t* num, size_t count) {
833 CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
834 if (values == NULL) return -1;
835 for (size_t i = 0; i < count; i++) {
836 values[i] = cxJsonCreateInteger(arr->allocator, num[i]);
837 if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
838 }
839 int ret = cxJsonArrAddValues(arr, values, count);
840 free(values);
841 return ret;
842 }
843
844 int cxJsonArrAddStrings(CxJsonValue* arr, const char* const* str, size_t count) {
845 CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
846 if (values == NULL) return -1;
847 for (size_t i = 0; i < count; i++) {
848 values[i] = cxJsonCreateString(arr->allocator, str[i]);
849 if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
850 }
851 int ret = cxJsonArrAddValues(arr, values, count);
852 free(values);
853 return ret;
854 }
855
856 int cxJsonArrAddCxStrings(CxJsonValue* arr, const cxstring* str, size_t count) {
857 CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
858 if (values == NULL) return -1;
859 for (size_t i = 0; i < count; i++) {
860 values[i] = cxJsonCreateCxString(arr->allocator, str[i]);
861 if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
862 }
863 int ret = cxJsonArrAddValues(arr, values, count);
864 free(values);
865 return ret;
866 }
867
868 int cxJsonArrAddLiterals(CxJsonValue* arr, const CxJsonLiteral* lit, size_t count) {
869 CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
870 if (values == NULL) return -1;
871 for (size_t i = 0; i < count; i++) {
872 values[i] = cxJsonCreateLiteral(arr->allocator, lit[i]);
873 if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
874 }
875 int ret = cxJsonArrAddValues(arr, values, count);
876 free(values);
877 return ret;
878 }
879
880 int cxJsonArrAddValues(CxJsonValue* arr, CxJsonValue* const* val, size_t count) {
881 CxArrayReallocator value_realloc = cx_array_reallocator(arr->allocator, NULL);
882 assert(arr->type == CX_JSON_ARRAY);
883 return cx_array_simple_copy_a(&value_realloc,
884 arr->value.array.array,
885 arr->value.array.array_size,
886 val, count
887 );
888 }
889
890 int cxJsonObjPut(CxJsonValue* obj, cxstring name, CxJsonValue* child) {
891 cxmutstr k = cx_strdup_a(obj->allocator, name);
892 if (k.ptr == NULL) return -1;
893 CxJsonObjValue kv = {k, child};
894 if (json_add_objvalue(obj, kv)) {
895 cx_strfree_a(obj->allocator, &k);
896 return 1;
897 } else {
898 return 0;
899 }
900 }
901
902 CxJsonValue* cxJsonObjPutObj(CxJsonValue* obj, cxstring name) {
903 CxJsonValue* v = cxJsonCreateObj(obj->allocator);
904 if (v == NULL) return NULL;
905 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
906 return v;
907 }
908
909 CxJsonValue* cxJsonObjPutArr(CxJsonValue* obj, cxstring name) {
910 CxJsonValue* v = cxJsonCreateArr(obj->allocator);
911 if (v == NULL) return NULL;
912 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
913 return v;
914 }
915
916 CxJsonValue* cxJsonObjPutNumber(CxJsonValue* obj, cxstring name, double num) {
917 CxJsonValue* v = cxJsonCreateNumber(obj->allocator, num);
918 if (v == NULL) return NULL;
919 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
920 return v;
921 }
922
923 CxJsonValue* cxJsonObjPutInteger(CxJsonValue* obj, cxstring name, int64_t num) {
924 CxJsonValue* v = cxJsonCreateInteger(obj->allocator, num);
925 if (v == NULL) return NULL;
926 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
927 return v;
928 }
929
930 CxJsonValue* cxJsonObjPutString(CxJsonValue* obj, cxstring name, const char* str) {
931 CxJsonValue* v = cxJsonCreateString(obj->allocator, str);
932 if (v == NULL) return NULL;
933 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
934 return v;
935 }
936
937 CxJsonValue* cxJsonObjPutCxString(CxJsonValue* obj, cxstring name, cxstring str) {
938 CxJsonValue* v = cxJsonCreateCxString(obj->allocator, str);
939 if (v == NULL) return NULL;
940 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
941 return v;
942 }
943
944 CxJsonValue* cxJsonObjPutLiteral(CxJsonValue* obj, cxstring name, CxJsonLiteral lit) {
945 CxJsonValue* v = cxJsonCreateLiteral(obj->allocator, lit);
946 if (v == NULL) return NULL;
947 if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL;}
948 return v;
949 }
950
951 CxJsonValue *cxJsonArrGet(const CxJsonValue *value, size_t index) {
952 if (index >= value->value.array.array_size) {
953 return &cx_json_value_nothing;
954 }
955 return value->value.array.array[index];
956 }
957
958 CxIterator cxJsonArrIter(const CxJsonValue *value) {
959 return cxIteratorPtr(
960 value->value.array.array,
961 value->value.array.array_size
962 );
963 }
964
965 CxIterator cxJsonObjIter(const CxJsonValue *value) {
966 return cxIterator(
967 value->value.object.values,
968 sizeof(CxJsonObjValue),
969 value->value.object.values_size
970 );
971 }
972
973 CxJsonValue *cx_json_obj_get_cxstr(const CxJsonValue *value, cxstring name) {
974 CxJsonObjValue *member = json_find_objvalue(value, name);
975 if (member == NULL) {
976 return &cx_json_value_nothing;
977 } else {
978 return member->value;
979 }
980 }
981
982 static const CxJsonWriter cx_json_writer_default = {
983 false,
984 true,
985 255,
986 false,
987 4
988 };
989
990 CxJsonWriter cxJsonWriterCompact(void) {
991 return cx_json_writer_default;
992 }
993
994 CxJsonWriter cxJsonWriterPretty(bool use_spaces) {
995 return (CxJsonWriter) {
996 true,
997 true,
998 255,
999 use_spaces,
1000 4
1001 };
1002 }
1003
1004 static int cx_json_writer_indent(
1005 void *target,
1006 cx_write_func wfunc,
1007 const CxJsonWriter *settings,
1008 unsigned int depth
1009 ) {
1010 if (depth == 0) return 0;
1011
1012 // determine the width and characters to use
1013 const char* indent; // for 32 prepared chars
1014 size_t width = depth;
1015 if (settings->indent_space) {
1016 if (settings->indent == 0) return 0;
1017 width *= settings->indent;
1018 indent = " ";
1019 } else {
1020 indent = "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t";
1021 }
1022
1023 // calculate the number of write calls and write
1024 size_t full = width / 32;
1025 size_t remaining = width % 32;
1026 for (size_t i = 0; i < full; i++) {
1027 if (32 != wfunc(indent, 1, 32, target)) return 1;
1028 }
1029 if (remaining != wfunc(indent, 1, remaining, target)) return 1;
1030
1031 return 0;
1032 }
1033
1034
1035 int cx_json_write_rec(
1036 void *target,
1037 const CxJsonValue *value,
1038 cx_write_func wfunc,
1039 const CxJsonWriter *settings,
1040 unsigned int depth
1041 ) {
1042 // keep track of written items
1043 // the idea is to reduce the number of jumps for error checking
1044 size_t actual = 0, expected = 0;
1045
1046 // small buffer for number to string conversions
1047 char numbuf[32];
1048
1049 // recursively write the values
1050 switch (value->type) {
1051 case CX_JSON_OBJECT: {
1052 const char *begin_obj = "{\n";
1053 if (settings->pretty) {
1054 actual += wfunc(begin_obj, 1, 2, target);
1055 expected += 2;
1056 } else {
1057 actual += wfunc(begin_obj, 1, 1, target);
1058 expected++;
1059 }
1060 depth++;
1061 size_t elem_count = value->value.object.values_size;
1062 for (size_t look_idx = 0; look_idx < elem_count; look_idx++) {
1063 // get the member either via index array or directly
1064 size_t elem_idx = settings->sort_members
1065 ? look_idx
1066 : value->value.object.indices[look_idx];
1067 CxJsonObjValue *member = &value->value.object.values[elem_idx];
1068 if (settings->sort_members) {
1069 depth++;depth--;
1070 }
1071
1072 // possible indentation
1073 if (settings->pretty) {
1074 if (cx_json_writer_indent(target, wfunc, settings, depth)) {
1075 return 1; // LCOV_EXCL_LINE
1076 }
1077 }
1078
1079 // the name
1080 actual += wfunc("\"", 1, 1, target);
1081 // TODO: escape the string
1082 actual += wfunc(member->name.ptr, 1,
1083 member->name.length, target);
1084 actual += wfunc("\"", 1, 1, target);
1085 const char *obj_name_sep = ": ";
1086 if (settings->pretty) {
1087 actual += wfunc(obj_name_sep, 1, 2, target);
1088 expected += 4 + member->name.length;
1089 } else {
1090 actual += wfunc(obj_name_sep, 1, 1, target);
1091 expected += 3 + member->name.length;
1092 }
1093
1094 // the value
1095 if (cx_json_write_rec(target, member->value, wfunc, settings, depth)) return 1;
1096
1097 // end of object-value
1098 if (look_idx < elem_count - 1) {
1099 const char *obj_value_sep = ",\n";
1100 if (settings->pretty) {
1101 actual += wfunc(obj_value_sep, 1, 2, target);
1102 expected += 2;
1103 } else {
1104 actual += wfunc(obj_value_sep, 1, 1, target);
1105 expected++;
1106 }
1107 } else {
1108 if (settings->pretty) {
1109 actual += wfunc("\n", 1, 1, target);
1110 expected ++;
1111 }
1112 }
1113 }
1114 depth--;
1115 if (settings->pretty) {
1116 if (cx_json_writer_indent(target, wfunc, settings, depth)) return 1;
1117 }
1118 actual += wfunc("}", 1, 1, target);
1119 expected++;
1120 break;
1121 }
1122 case CX_JSON_ARRAY: {
1123 actual += wfunc("[", 1, 1, target);
1124 expected++;
1125 CxIterator iter = cxJsonArrIter(value);
1126 cx_foreach(CxJsonValue*, element, iter) {
1127 if (cx_json_write_rec(
1128 target, element,
1129 wfunc, settings, depth)
1130 ) return 1;
1131
1132 if (iter.index < iter.elem_count - 1) {
1133 const char *arr_value_sep = ", ";
1134 if (settings->pretty) {
1135 actual += wfunc(arr_value_sep, 1, 2, target);
1136 expected += 2;
1137 } else {
1138 actual += wfunc(arr_value_sep, 1, 1, target);
1139 expected++;
1140 }
1141 }
1142 }
1143 actual += wfunc("]", 1, 1, target);
1144 expected++;
1145 break;
1146 }
1147 case CX_JSON_STRING: {
1148 actual += wfunc("\"", 1, 1, target);
1149 // TODO: escape the string
1150 actual += wfunc(value->value.string.ptr, 1,
1151 value->value.string.length, target);
1152 actual += wfunc("\"", 1, 1, target);
1153 expected += 2 + value->value.string.length;
1154 break;
1155 }
1156 case CX_JSON_NUMBER: {
1157 // TODO: locale bullshit
1158 // TODO: formatting settings
1159 snprintf(numbuf, 32, "%g", value->value.number);
1160 size_t len = strlen(numbuf);
1161 actual += wfunc(numbuf, 1, len, target);
1162 expected += len;
1163 break;
1164 }
1165 case CX_JSON_INTEGER: {
1166 snprintf(numbuf, 32, "%" PRIi64, value->value.integer);
1167 size_t len = strlen(numbuf);
1168 actual += wfunc(numbuf, 1, len, target);
1169 expected += len;
1170 break;
1171 }
1172 case CX_JSON_LITERAL: {
1173 if (value->value.literal == CX_JSON_TRUE) {
1174 actual += wfunc("true", 1, 4, target);
1175 expected += 4;
1176 } else if (value->value.literal == CX_JSON_FALSE) {
1177 actual += wfunc("false", 1, 5, target);
1178 expected += 5;
1179 } else {
1180 actual += wfunc("null", 1, 4, target);
1181 expected += 4;
1182 }
1183 break;
1184 }
1185 case CX_JSON_NOTHING: {
1186 // deliberately supported as an empty string!
1187 // users might want to just write the result
1188 // of a get operation without testing the value
1189 // and therefore this should not blow up
1190 break;
1191 }
1192 default: assert(false); // LCOV_EXCL_LINE
1193 }
1194
1195 return expected != actual;
1196 }
1197
1198 int cxJsonWrite(
1199 void *target,
1200 const CxJsonValue *value,
1201 cx_write_func wfunc,
1202 const CxJsonWriter *settings
1203 ) {
1204 if (settings == NULL) {
1205 settings = &cx_json_writer_default;
1206 }
1207 assert(target != NULL);
1208 assert(value != NULL);
1209 assert(wfunc != NULL);
1210
1211 return cx_json_write_rec(target, value, wfunc, settings, 0);
1212 }

Mercurial Repository: dbutils

mercurial