yeptris 0.6.15.2 → 0.6.16.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -27,6 +27,8 @@
27
27
  #include <math.h>
28
28
  #include <string.h>
29
29
 
30
+ #include <stdlib.h>
31
+
30
32
  #include <yeptris/cbor.h>
31
33
  #include <yeptris/resolve.h>
32
34
 
@@ -37,29 +39,41 @@
37
39
  #include "../memory/allocator.h"
38
40
  #include "../parse/events.h"
39
41
  #include "doc.h"
42
+ #include "sink.h"
40
43
 
41
44
  #define YEP_CBOR_INDEF UINT64_MAX
42
45
  #define YEP_CBOR_MAX_ARG_BYTES 8
43
46
 
44
47
  typedef struct {
45
- uint64_t rem; /* items left; maps count 2*pairs; INDEF = until break */
48
+ uint64_t rem; /* items left; maps count 2*pairs; INDEF = until break */
49
+ uint64_t consumed; /* items landed (the key/value parity) */
46
50
  uint8_t is_map;
47
51
  } yep_cframe;
48
52
 
53
+ #define YEP_CBOR_TAGBUF \
54
+ 512 /* a semantic-tag chain over ~23 links is \
55
+ * pathological; the arena form had no cap, \
56
+ * the stack form does (documented) */
57
+
49
58
  typedef struct {
50
- yep_dom* d;
59
+ yep_dom* d; /* the DOM sink's ctx (NULL for pure sinks) */
51
60
  const unsigned char* p;
52
61
  size_t len;
53
62
  size_t i;
54
63
  int strict;
55
64
  YeptrisStatus status;
56
- /* pending tag chain: tag numbers arrive before their content and
57
- * render into the arena as "N N N" (outermost first — the encoder
58
- * splits it back); the span attaches to the next data item */
59
- yep_sview pending_tag;
60
- int have_pending;
65
+ /* the pending semantic-tag chain, text form ("N N N", outermost
66
+ * first — the next item's writer consumes it) */
67
+ char tagbuf[YEP_CBOR_TAGBUF];
68
+ uint32_t taglen;
69
+ /* indefinite-string accumulation (chunk framing is the grammar's;
70
+ * the finished span goes to one str/bytes callback) */
71
+ unsigned char* ibuf;
72
+ size_t icap;
73
+ uint32_t ilen;
74
+ const yep_cbor_sink* s;
61
75
  yep_cframe frame[YEP_DOM_MAX_DEPTH];
62
- int depth; /* == d->depth by construction */
76
+ int depth;
63
77
  } yep_cdec;
64
78
 
65
79
  static void cbor_fail(yep_cdec* c, yep_err_code code, const char* msg) {
@@ -146,116 +160,74 @@ static double cbor_half(uint16_t h) {
146
160
  return (h & 0x8000u) != 0 ? -val : val;
147
161
  }
148
162
 
149
- /* Scalar from arena text (numbers, diagnostics, literals). */
150
- static uint32_t cbor_text_node(yep_cdec* c, const char* text, uint32_t n, uint8_t tag_id) {
151
- yep_view tag_v;
152
- const yep_view* tag = NULL;
153
- if (c->have_pending) {
154
- /* the pending chain lives in the DOM arena (offsets, not
155
- * pointers — the arena reallocs as it grows) */
156
- tag_v.p = c->d->str + (c->pending_tag.off & YEP_SV_OFF);
157
- tag_v.len = c->pending_tag.len;
158
- tag = &tag_v;
159
- c->have_pending = 0;
160
- }
161
- uint32_t id = dom_open_node(c->d, YEP_DOM_SCALAR, tag, NULL, 0, YEP_STYLE_PLAIN, 1, 0);
162
- if (id == UINT32_MAX) {
163
- c->status = YEPTRIS_ERROR_MEMORY;
164
- return UINT32_MAX;
165
- }
166
- yep_dnode* node = &c->d->nodes[id];
167
- char* dst = yep_dom_str_tail(c->d, n);
168
- if (dst == NULL) {
169
- c->status = YEPTRIS_ERROR_MEMORY;
170
- return UINT32_MAX;
171
- }
172
- memcpy(dst, text, n);
173
- node->value = yep_dom_str_commit(c->d, n);
174
- node->tag_id = tag_id;
175
- if (dom_place(c->d, id) != 0) {
176
- c->status = YEPTRIS_ERROR_MEMORY;
177
- return UINT32_MAX;
163
+ /* The pending semantic-tag chain, consumed by the next writer. */
164
+ static const char* cbor_take_tag(yep_cdec* c, uint32_t* len) {
165
+ if (c->taglen == 0) {
166
+ *len = 0;
167
+ return NULL;
178
168
  }
179
- return id;
169
+ *len = c->taglen;
170
+ c->taglen = 0;
171
+ return c->tagbuf;
180
172
  }
181
173
 
182
- /* Scalar borrowing the input (definite strings, bignum byte strings). */
183
- static uint32_t cbor_span_node(yep_cdec* c, const unsigned char* p, uint32_t n) {
184
- yep_view tag_v;
185
- const yep_view* tag = NULL;
186
- if (c->have_pending) {
187
- tag_v.p = c->d->str + (c->pending_tag.off & YEP_SV_OFF);
188
- tag_v.len = c->pending_tag.len;
189
- tag = &tag_v;
190
- c->have_pending = 0;
191
- }
192
- uint32_t id = dom_open_node(c->d, YEP_DOM_SCALAR, tag, NULL, 0, YEP_STYLE_DOUBLE_QUOTED, 0, 0);
193
- if (id == UINT32_MAX) {
174
+ /* Rendered text (numbers-as-text, literals, diagnostics). */
175
+ static uint32_t cbor_text_node(yep_cdec* c, const char* text, uint32_t n, uint8_t tag_id,
176
+ int is_key) {
177
+ uint32_t tag_len;
178
+ const char* tag = cbor_take_tag(c, &tag_len);
179
+ if (!c->s->text(c->s->ctx, text, n, tag_id, is_key, tag, tag_len)) {
194
180
  c->status = YEPTRIS_ERROR_MEMORY;
195
181
  return UINT32_MAX;
196
182
  }
197
- yep_dnode* node = &c->d->nodes[id];
198
- node->value.off = (uint32_t)((const char*)p - c->d->input_base);
199
- node->value.len = n;
200
- if (dom_place(c->d, id) != 0) {
183
+ return 0; /* item identity is the sink's business */
184
+ }
185
+
186
+ /* Definite string (mt2 -> bytes_val, mt3 -> str_val), borrowing the input. */
187
+ static uint32_t cbor_span_node(yep_cdec* c, const unsigned char* p, uint32_t n, uint8_t mt,
188
+ int is_key) {
189
+ uint32_t tag_len;
190
+ const char* tag = cbor_take_tag(c, &tag_len);
191
+ int ok = (mt == 2) ? c->s->bytes_val(c->s->ctx, p, n, 1, is_key, tag, tag_len)
192
+ : c->s->str_val(c->s->ctx, p, n, 1, is_key, tag, tag_len);
193
+ if (!ok) {
201
194
  c->status = YEPTRIS_ERROR_MEMORY;
202
195
  return UINT32_MAX;
203
196
  }
204
- return id;
197
+ return 0;
205
198
  }
206
199
 
207
- static int cbor_open(yep_cdec* c, int is_map) {
200
+ static int cbor_open(yep_cdec* c, int is_map, uint64_t cap) {
208
201
  if (c->depth >= YEP_DOM_MAX_DEPTH) {
209
202
  yep_error_set(yep_error_tls(), YEP_ERR_UNEXPECTED, 0, 0, c->i,
210
203
  "CBOR nesting exceeds the depth guard at byte %zu", c->i);
211
204
  c->status = YEPTRIS_ERROR_DEPTH;
212
205
  return 0;
213
206
  }
214
- yep_view tag_v;
215
- const yep_view* tag = NULL;
216
- if (c->have_pending) {
217
- tag_v.p = c->d->str + (c->pending_tag.off & YEP_SV_OFF);
218
- tag_v.len = c->pending_tag.len;
219
- tag = &tag_v;
220
- c->have_pending = 0;
221
- }
222
- uint32_t id =
223
- dom_open_node(c->d, is_map ? YEP_DOM_MAPPING : YEP_DOM_SEQUENCE, tag, NULL, 0, 0, 0, 1);
224
- if (id == UINT32_MAX) {
207
+ uint32_t tag_len;
208
+ const char* tag = cbor_take_tag(c, &tag_len);
209
+ if (!c->s->open(c->s->ctx, is_map, cap, tag, tag_len)) {
225
210
  c->status = YEPTRIS_ERROR_MEMORY;
226
211
  return 0;
227
212
  }
228
- if (dom_place(c->d, id) != 0) {
229
- c->status = YEPTRIS_ERROR_MEMORY;
230
- return 0;
231
- }
232
- c->d->map_pending_key[c->d->depth] = 0;
233
- c->d->stack[c->d->depth++] = id;
234
- c->depth++; /* the frame mirrors the DOM's stack one-to-one */
213
+ /* a sibling at this depth may have left state behind */
214
+ c->frame[c->depth].consumed = 0;
215
+ c->depth++;
235
216
  return 1;
236
217
  }
237
218
 
238
219
  static int cbor_note_tag(yep_cdec* c, uint64_t tag) {
239
- char buf[24];
240
- uint32_t n = 0;
241
- if (c->have_pending) {
242
- buf[n++] = ' '; /* separator between chain links */
243
- }
244
- n += cbor_u64dec(tag, buf + n);
245
- char* dst = yep_dom_str_tail(c->d, n);
246
- if (dst == NULL) {
247
- c->status = YEPTRIS_ERROR_MEMORY;
220
+ if (c->taglen + 1 + 20 > sizeof c->tagbuf) {
221
+ yep_error_set(yep_error_tls(), YEP_ERR_UNEXPECTED, 0, 0, c->i,
222
+ "semantic-tag chain exceeds %u bytes at byte %zu", (unsigned)sizeof c->tagbuf,
223
+ c->i);
224
+ c->status = YEPTRIS_ERROR_PARSE;
248
225
  return 0;
249
226
  }
250
- yep_sview piece = yep_dom_str_commit(c->d, n);
251
- if (!c->have_pending) {
252
- c->pending_tag = piece;
253
- c->pending_tag.len = 0;
254
- c->have_pending = 1;
227
+ if (c->taglen > 0) {
228
+ c->tagbuf[c->taglen++] = ' '; /* separator between chain links */
255
229
  }
256
- memcpy(dst, buf, n);
257
- /* piece-by-piece commit keeps the chain contiguous in the arena */
258
- c->pending_tag.len += n;
230
+ c->taglen += cbor_u64dec(tag, c->tagbuf + c->taglen);
259
231
  return 1;
260
232
  }
261
233
 
@@ -263,7 +235,6 @@ static int cbor_note_tag(yep_cdec* c, uint64_t tag) {
263
235
  * arena; text chunks validate per chunk — RFC 8949 s3.2.3's code-point
264
236
  * boundary rule makes per-chunk validity imply whole-string validity). */
265
237
  static int cbor_indef_string(yep_cdec* c, int is_text) {
266
- uint32_t first = 0;
267
238
  uint32_t total = 0;
268
239
  for (;;) {
269
240
  if (c->i >= c->len) {
@@ -291,75 +262,55 @@ static int cbor_indef_string(yep_cdec* c, int is_text) {
291
262
  return 0;
292
263
  }
293
264
  if (n > 0) { /* zero-length chunks are legal, just empty */
294
- char* dst = yep_dom_str_tail(c->d, (uint32_t)n);
295
- if (dst == NULL) {
296
- c->status = YEPTRIS_ERROR_MEMORY;
297
- return 0;
298
- }
299
- yep_sview piece = yep_dom_str_commit(c->d, (uint32_t)n);
300
- memcpy(dst, c->p + c->i, (size_t)n);
301
- if (total == 0) {
302
- first = piece.off;
265
+ if (total + n > c->icap) {
266
+ size_t ncap = c->icap ? c->icap * 2 : 256;
267
+ while (ncap < total + n) {
268
+ ncap *= 2;
269
+ }
270
+ unsigned char* nib = realloc(c->ibuf, ncap);
271
+ if (nib == NULL) {
272
+ c->status = YEPTRIS_ERROR_MEMORY;
273
+ return 0;
274
+ }
275
+ c->ibuf = nib;
276
+ c->icap = ncap;
303
277
  }
278
+ memcpy(c->ibuf + total, c->p + c->i, (size_t)n);
304
279
  total += (uint32_t)n;
305
280
  }
306
281
  c->i += (size_t)n;
307
282
  }
308
- /* node over the concatenated arena span */
309
- yep_view tag_v;
310
- const yep_view* tag = NULL;
311
- if (c->have_pending) {
312
- tag_v.p = c->d->str + (c->pending_tag.off & YEP_SV_OFF);
313
- tag_v.len = c->pending_tag.len;
314
- tag = &tag_v;
315
- c->have_pending = 0;
316
- }
317
- uint32_t id = dom_open_node(c->d, YEP_DOM_SCALAR, tag, NULL, 0, YEP_STYLE_DOUBLE_QUOTED, 0, 0);
318
- if (id == UINT32_MAX) {
319
- c->status = YEPTRIS_ERROR_MEMORY;
320
- return 0;
321
- }
322
- yep_dnode* node = &c->d->nodes[id];
323
- node->value.off = first;
324
- node->value.len = total;
325
- if (dom_place(c->d, id) != 0) {
283
+ /* one value over the accumulated span */
284
+ uint32_t tag_len;
285
+ const char* tag = cbor_take_tag(c, &tag_len);
286
+ int ok =
287
+ (is_text ? c->s->str_val : c->s->bytes_val)(c->s->ctx, c->ibuf, total, 0, 0, tag, tag_len);
288
+ if (!ok) {
326
289
  c->status = YEPTRIS_ERROR_MEMORY;
327
290
  return 0;
328
291
  }
329
292
  return 1;
330
293
  }
331
294
 
332
- /* yep_json_number-style inline render: one node for a known integer. */
333
- static int cbor_int_node(yep_cdec* c, int negative, uint64_t mag) {
334
- char buf[24];
335
- uint32_t n = 0;
336
- uint8_t tag_id;
337
- if (negative) {
338
- buf[n++] = '-';
295
+ /* Known integer: value = negative ? -mag : mag, mag <= 2^63. */
296
+ static int cbor_int_node(yep_cdec* c, int negative, uint64_t mag, int is_key) {
297
+ uint32_t tag_len;
298
+ const char* tag = cbor_take_tag(c, &tag_len);
299
+ if (!c->s->int_val(c->s->ctx, negative, mag, is_key, tag, tag_len)) {
300
+ c->status = YEPTRIS_ERROR_MEMORY;
301
+ return 0;
339
302
  }
340
- n += cbor_u64dec(mag, buf + n);
341
- tag_id = YEPTRIS_TAG_INT;
342
- return cbor_text_node(c, buf, n, tag_id) != UINT32_MAX;
303
+ return 1;
343
304
  }
344
305
 
345
- static int cbor_float_node(yep_cdec* c, double dv) {
346
- char buf[48];
347
- uint32_t n;
348
- uint8_t tag_id = YEPTRIS_TAG_FLOAT;
349
- if (dv != dv) {
350
- memcpy(buf, "NaN", 3);
351
- n = 3;
352
- } else if (dv > 1.7976931348623157e308) {
353
- memcpy(buf, "Infinity", 8);
354
- n = 8;
355
- } else if (dv < -1.7976931348623157e308) {
356
- memcpy(buf, "-Infinity", 9);
357
- n = 9;
358
- } else {
359
- n = (uint32_t)yep_d2s_shortest(dv, buf);
306
+ static int cbor_float_node(yep_cdec* c, double dv, int is_key) {
307
+ uint32_t tag_len;
308
+ const char* tag = cbor_take_tag(c, &tag_len);
309
+ if (!c->s->float_val(c->s->ctx, dv, is_key, tag, tag_len)) {
310
+ c->status = YEPTRIS_ERROR_MEMORY;
311
+ return 0;
360
312
  }
361
- (void)tag_id;
362
- return cbor_text_node(c, buf, n, YEPTRIS_TAG_FLOAT) != UINT32_MAX;
313
+ return 1;
363
314
  }
364
315
 
365
316
  /* The diagnostic renderer for lenient-mode non-string SCALAR map keys
@@ -408,9 +359,9 @@ static int cbor_key_diag(yep_cdec* c, uint8_t mt, uint64_t val, double dv) {
408
359
  }
409
360
  break;
410
361
  default:
411
- return cbor_float_node(c, dv); /* floats: the shortest text */
362
+ return cbor_float_node(c, dv, 1); /* floats: the shortest text */
412
363
  }
413
- return cbor_text_node(c, buf, n, 0) != UINT32_MAX;
364
+ return cbor_text_node(c, buf, n, 0, 1) != UINT32_MAX;
414
365
  }
415
366
 
416
367
  /* Decodes ONE data item (scalar: creates+places the node; container:
@@ -425,9 +376,10 @@ static int cbor_item(yep_cdec* c) {
425
376
  uint8_t ai = ib & 0x1F;
426
377
 
427
378
  /* the map-key position decides representation before the item is
428
- * built (the DOM pairs whatever node lands there) */
379
+ * built: even items consumed = key, odd = value (the frame's own
380
+ * parity — the DOM's placement machine tracks its own copy) */
429
381
  int key_slot = 0;
430
- if (c->depth > 0 && c->frame[c->depth - 1].is_map && !c->d->map_pending_key[c->d->depth - 1]) {
382
+ if (c->depth > 0 && c->frame[c->depth - 1].is_map && c->frame[c->depth - 1].consumed % 2 == 0) {
431
383
  key_slot = 1;
432
384
  }
433
385
 
@@ -444,7 +396,7 @@ static int cbor_item(yep_cdec* c) {
444
396
  if (key_slot) {
445
397
  CBOR_REJECT("structure as map key");
446
398
  }
447
- if (!cbor_open(c, 0)) {
399
+ if (!cbor_open(c, 0, YEP_CBOR_INDEF)) {
448
400
  return 0;
449
401
  }
450
402
  c->frame[c->depth - 1].rem = YEP_CBOR_INDEF;
@@ -454,7 +406,7 @@ static int cbor_item(yep_cdec* c) {
454
406
  if (key_slot) {
455
407
  CBOR_REJECT("structure as map key");
456
408
  }
457
- if (!cbor_open(c, 1)) {
409
+ if (!cbor_open(c, 1, YEP_CBOR_INDEF)) {
458
410
  return 0;
459
411
  }
460
412
  c->frame[c->depth - 1].rem = YEP_CBOR_INDEF;
@@ -482,9 +434,9 @@ static int cbor_item(yep_cdec* c) {
482
434
  return cbor_key_diag(c, 0, arg, 0) ? 1 : 0;
483
435
  }
484
436
  if (arg <= (uint64_t)INT64_MAX) {
485
- return cbor_int_node(c, 0, arg) ? 1 : 0;
437
+ return cbor_int_node(c, 0, arg, key_slot) ? 1 : 0;
486
438
  }
487
- return cbor_float_node(c, (double)arg) ? 1 : 0;
439
+ return cbor_float_node(c, (double)arg, key_slot) ? 1 : 0;
488
440
  }
489
441
  case 1: {
490
442
  if (key_slot && c->strict) {
@@ -495,9 +447,9 @@ static int cbor_item(yep_cdec* c) {
495
447
  }
496
448
  if (arg <= (uint64_t)INT64_MAX + 1) { /* -1-2^63 == INT64_MIN */
497
449
  uint64_t mag = arg + 1;
498
- return cbor_int_node(c, 1, mag) ? 1 : 0;
450
+ return cbor_int_node(c, 1, mag, key_slot) ? 1 : 0;
499
451
  }
500
- return cbor_float_node(c, -((double)arg + 1.0)) ? 1 : 0;
452
+ return cbor_float_node(c, -((double)arg + 1.0), key_slot) ? 1 : 0;
501
453
  }
502
454
  case 2:
503
455
  case 3: {
@@ -518,7 +470,7 @@ static int cbor_item(yep_cdec* c) {
518
470
  c->i += (size_t)arg;
519
471
  return ok ? 1 : 0;
520
472
  }
521
- uint32_t id = cbor_span_node(c, c->p + c->i, (uint32_t)arg);
473
+ uint32_t id = cbor_span_node(c, c->p + c->i, (uint32_t)arg, (uint8_t)mt, key_slot);
522
474
  c->i += (size_t)arg;
523
475
  return id != UINT32_MAX ? 1 : 0;
524
476
  }
@@ -529,7 +481,7 @@ static int cbor_item(yep_cdec* c) {
529
481
  if (arg > c->len - c->i) { /* every item consumes >= 1 byte */
530
482
  CBOR_REJECT("array overruns the input");
531
483
  }
532
- if (!cbor_open(c, 0)) {
484
+ if (!cbor_open(c, 0, arg)) {
533
485
  return 0;
534
486
  }
535
487
  c->frame[c->depth - 1].rem = arg;
@@ -543,7 +495,7 @@ static int cbor_item(yep_cdec* c) {
543
495
  if (arg > (c->len - c->i) / 2) { /* pairs need 2 items each */
544
496
  CBOR_REJECT("map overruns the input");
545
497
  }
546
- if (!cbor_open(c, 1)) {
498
+ if (!cbor_open(c, 1, arg * 2)) {
547
499
  return 0;
548
500
  }
549
501
  c->frame[c->depth - 1].rem = arg * 2;
@@ -566,22 +518,22 @@ static int cbor_item(yep_cdec* c) {
566
518
  if (key_slot) {
567
519
  return cbor_key_diag(c, 7, 20, 0) ? 1 : 0;
568
520
  }
569
- return cbor_text_node(c, "false", 5, YEPTRIS_TAG_BOOL) != UINT32_MAX ? 1 : 0;
521
+ return cbor_text_node(c, "false", 5, YEPTRIS_TAG_BOOL, key_slot) != UINT32_MAX ? 1 : 0;
570
522
  case 21:
571
523
  if (key_slot) {
572
524
  return cbor_key_diag(c, 7, 21, 0) ? 1 : 0;
573
525
  }
574
- return cbor_text_node(c, "true", 4, YEPTRIS_TAG_BOOL) != UINT32_MAX ? 1 : 0;
526
+ return cbor_text_node(c, "true", 4, YEPTRIS_TAG_BOOL, key_slot) != UINT32_MAX ? 1 : 0;
575
527
  case 22:
576
528
  if (key_slot) {
577
529
  return cbor_key_diag(c, 7, 22, 0) ? 1 : 0;
578
530
  }
579
- return cbor_text_node(c, "null", 4, YEPTRIS_TAG_NULL) != UINT32_MAX ? 1 : 0;
531
+ return cbor_text_node(c, "null", 4, YEPTRIS_TAG_NULL, key_slot) != UINT32_MAX ? 1 : 0;
580
532
  case 23:
581
533
  if (key_slot) {
582
534
  return cbor_key_diag(c, 7, 23, 0) ? 1 : 0;
583
535
  }
584
- return cbor_text_node(c, "undefined", 9, 0) != UINT32_MAX ? 1 : 0;
536
+ return cbor_text_node(c, "undefined", 9, 0, key_slot) != UINT32_MAX ? 1 : 0;
585
537
  case 24: {
586
538
  if (c->len - c->i < 1) {
587
539
  CBOR_REJECT("truncated simple value");
@@ -597,7 +549,7 @@ static int cbor_item(yep_cdec* c) {
597
549
  memcpy(buf, "simple(", 7);
598
550
  uint32_t n = 7 + cbor_u64dec(v, buf + 7);
599
551
  buf[n++] = ')';
600
- return cbor_text_node(c, buf, n, 0) != UINT32_MAX ? 1 : 0;
552
+ return cbor_text_node(c, buf, n, 0, key_slot) != UINT32_MAX ? 1 : 0;
601
553
  }
602
554
  case 25: {
603
555
  if (c->len - c->i < 2) {
@@ -641,7 +593,7 @@ static int cbor_item(yep_cdec* c) {
641
593
  memcpy(buf, "simple(", 7);
642
594
  uint32_t n = 7 + cbor_u64dec(ai, buf + 7);
643
595
  buf[n++] = ')';
644
- return cbor_text_node(c, buf, n, 0) != UINT32_MAX ? 1 : 0;
596
+ return cbor_text_node(c, buf, n, 0, key_slot) != UINT32_MAX ? 1 : 0;
645
597
  }
646
598
  CBOR_REJECT("reserved additional information");
647
599
  }
@@ -652,34 +604,191 @@ static int cbor_item(yep_cdec* c) {
652
604
  if (key_slot) {
653
605
  return cbor_key_diag(c, 8, 0, dv) ? 1 : 0;
654
606
  }
655
- return cbor_float_node(c, dv) ? 1 : 0;
607
+ return cbor_float_node(c, dv, key_slot) ? 1 : 0;
656
608
  }
657
609
 
658
- int yep_cbor_decode_dom(yep_dom* d, const unsigned char* p, size_t len, int strict,
659
- size_t* consumed) {
610
+ /* The DOM sink: the yeptris_cbor_decode public path. Each impl is the
611
+ * pre-sink writer, byte for byte (the differential suite pins it). */
612
+
613
+ static int dom_sink_text(void* dp, const char* s, uint32_t n, uint8_t tag_id, int is_key,
614
+ const char* tag, uint32_t tag_len) {
615
+ yep_dom* d = dp;
616
+ (void)is_key;
617
+ yep_view tag_v;
618
+ const yep_view* tagp = NULL;
619
+ if (tag != NULL) {
620
+ char* dst = yep_dom_str_tail(d, tag_len);
621
+ if (dst == NULL) {
622
+ return 0;
623
+ }
624
+ memcpy(dst, tag, tag_len);
625
+ yep_sview piece = yep_dom_str_commit(d, tag_len);
626
+ tag_v.p = d->str + (piece.off & YEP_SV_OFF);
627
+ tag_v.len = tag_len;
628
+ tagp = &tag_v;
629
+ }
630
+ uint32_t id = dom_open_node(d, YEP_DOM_SCALAR, tagp, NULL, 0, YEP_STYLE_PLAIN, 1, 0);
631
+ if (id == UINT32_MAX) {
632
+ return 0;
633
+ }
634
+ yep_dnode* node = &d->nodes[id];
635
+ char* dst = yep_dom_str_tail(d, n);
636
+ if (dst == NULL) {
637
+ return 0;
638
+ }
639
+ memcpy(dst, s, n);
640
+ node->value = yep_dom_str_commit(d, n);
641
+ node->tag_id = tag_id;
642
+ return dom_place(d, id) == 0;
643
+ }
644
+
645
+ static int dom_sink_span(void* dp, const unsigned char* p, uint32_t n, int borrowed, int is_key,
646
+ const char* tag, uint32_t tag_len) {
647
+ yep_dom* d = dp;
648
+ (void)is_key;
649
+ (void)borrowed; /* 0 rides the scratch buffer: still copied to the arena */
650
+ yep_view tag_v;
651
+ const yep_view* tagp = NULL;
652
+ if (tag != NULL) {
653
+ char* dst = yep_dom_str_tail(d, tag_len);
654
+ if (dst == NULL) {
655
+ return 0;
656
+ }
657
+ memcpy(dst, tag, tag_len);
658
+ yep_sview piece = yep_dom_str_commit(d, tag_len);
659
+ tag_v.p = d->str + (piece.off & YEP_SV_OFF);
660
+ tag_v.len = tag_len;
661
+ tagp = &tag_v;
662
+ }
663
+ uint32_t id = dom_open_node(d, YEP_DOM_SCALAR, tagp, NULL, 0, YEP_STYLE_DOUBLE_QUOTED, 0, 0);
664
+ if (id == UINT32_MAX) {
665
+ return 0;
666
+ }
667
+ yep_dnode* node = &d->nodes[id];
668
+ if (borrowed) {
669
+ node->value.off = (uint32_t)((const char*)p - d->input_base);
670
+ node->value.len = n;
671
+ } else if (n > 0) { /* the empty indefinite string needs no arena */
672
+ char* dst = yep_dom_str_tail(d, n);
673
+ if (dst == NULL) {
674
+ return 0;
675
+ }
676
+ memcpy(dst, p, n);
677
+ node->value = yep_dom_str_commit(d, n);
678
+ }
679
+ return dom_place(d, id) == 0;
680
+ }
681
+
682
+ static int dom_sink_str(void* dp, const unsigned char* p, uint32_t n, int borrowed, int is_key,
683
+ const char* tag, uint32_t tag_len) {
684
+ return dom_sink_span(dp, p, n, borrowed, is_key, tag, tag_len);
685
+ }
686
+
687
+ static int dom_sink_bytes(void* dp, const unsigned char* p, uint32_t n, int borrowed, int is_key,
688
+ const char* tag, uint32_t tag_len) {
689
+ return dom_sink_span(dp, p, n, borrowed, is_key, tag, tag_len);
690
+ }
691
+
692
+ static int dom_sink_open(void* dp, int is_map, uint64_t cap, const char* tag, uint32_t tag_len) {
693
+ yep_dom* d = dp;
694
+ (void)cap;
695
+ yep_view tag_v;
696
+ const yep_view* tagp = NULL;
697
+ if (tag != NULL) {
698
+ char* dst = yep_dom_str_tail(d, tag_len);
699
+ if (dst == NULL) {
700
+ return 0;
701
+ }
702
+ memcpy(dst, tag, tag_len);
703
+ yep_sview piece = yep_dom_str_commit(d, tag_len);
704
+ tag_v.p = d->str + (piece.off & YEP_SV_OFF);
705
+ tag_v.len = tag_len;
706
+ tagp = &tag_v;
707
+ }
708
+ uint32_t id =
709
+ dom_open_node(d, is_map ? YEP_DOM_MAPPING : YEP_DOM_SEQUENCE, tagp, NULL, 0, 0, 0, 1);
710
+ if (id == UINT32_MAX) {
711
+ return 0;
712
+ }
713
+ if (dom_place(d, id) != 0) {
714
+ return 0;
715
+ }
716
+ d->map_pending_key[d->depth] = 0;
717
+ d->stack[d->depth++] = id;
718
+ return 1;
719
+ }
720
+
721
+ static int dom_sink_close(void* dp, int is_map) {
722
+ yep_dom* d = dp;
723
+ (void)is_map;
724
+ d->depth--;
725
+ return 1;
726
+ }
727
+
728
+ static int dom_sink_int(void* dp, int negative, uint64_t mag, int is_key, const char* tag,
729
+ uint32_t tag_len) {
730
+ char buf[24];
731
+ uint32_t n = 0;
732
+ if (negative) {
733
+ buf[n++] = '-';
734
+ }
735
+ n += cbor_u64dec(mag, buf + n);
736
+ return dom_sink_text(dp, buf, n, YEPTRIS_TAG_INT, is_key, tag, tag_len);
737
+ }
738
+
739
+ static int dom_sink_float(void* dp, double dv, int is_key, const char* tag, uint32_t tag_len) {
740
+ char buf[48];
741
+ (void)is_key;
742
+ uint32_t n;
743
+ if (dv != dv) {
744
+ memcpy(buf, "NaN", 3);
745
+ n = 3;
746
+ } else if (dv > 1.7976931348623157e308) {
747
+ memcpy(buf, "Infinity", 8);
748
+ n = 8;
749
+ } else if (dv < -1.7976931348623157e308) {
750
+ memcpy(buf, "-Infinity", 9);
751
+ n = 9;
752
+ } else {
753
+ n = (uint32_t)yep_d2s_shortest(dv, buf);
754
+ }
755
+ return dom_sink_text(dp, buf, n, YEPTRIS_TAG_FLOAT, is_key, tag, tag_len);
756
+ }
757
+
758
+ static const yep_cbor_sink yep_cbor_dom_sink_impl = {
759
+ NULL, dom_sink_text, dom_sink_int, dom_sink_float,
760
+ dom_sink_str, dom_sink_bytes, dom_sink_open, dom_sink_close,
761
+ };
762
+
763
+ int yep_cbor_decode_gen(const unsigned char* p, size_t len, int strict, size_t* consumed,
764
+ const yep_cbor_sink* sink) {
660
765
  yep_cdec c;
661
766
  memset(&c, 0, sizeof(c));
662
- c.d = d;
767
+ c.s = sink;
663
768
  c.p = p;
664
769
  c.len = len;
665
770
  c.strict = strict;
666
771
  c.status = YEPTRIS_OK;
667
772
  int done_top = 0; /* the ONE top-level item is done (tags never count) */
773
+ int rc;
668
774
  for (;;) {
669
775
  if (c.depth == 0) {
670
776
  if (done_top) { /* the root closed; one data item per call */
671
777
  if (c.i != len && consumed == NULL) {
672
778
  cbor_fail(&c, YEP_ERR_UNEXPECTED, "trailing bytes after the data item");
673
- return YEPTRIS_ERROR_PARSE;
779
+ rc = YEPTRIS_ERROR_PARSE;
780
+ goto done;
674
781
  }
675
782
  if (consumed != NULL) {
676
783
  *consumed = c.i;
677
784
  }
678
- return YEPTRIS_OK;
785
+ rc = YEPTRIS_OK;
786
+ goto done;
679
787
  }
680
788
  int st = cbor_item(&c);
681
789
  if (st == 0) {
682
- return (int)c.status;
790
+ rc = (int)c.status;
791
+ goto done;
683
792
  }
684
793
  if (st == 2) {
685
794
  continue; /* tag chain: the content item follows */
@@ -689,19 +798,26 @@ int yep_cbor_decode_dom(yep_dom* d, const unsigned char* p, size_t len, int stri
689
798
  }
690
799
  yep_cframe* fr = &c.frame[c.depth - 1];
691
800
  if (fr->rem == 0) { /* definite container complete */
692
- c.d->depth--;
801
+ if (!c.s->close(c.s->ctx, fr->is_map)) {
802
+ rc = YEPTRIS_ERROR_MEMORY;
803
+ goto done;
804
+ }
693
805
  c.depth--;
694
806
  continue;
695
807
  }
696
808
  if (fr->rem == YEP_CBOR_INDEF) {
697
809
  if (c.i < c.len && c.p[c.i] == 0xFF) {
698
- if (fr->is_map && c.d->map_pending_key[c.d->depth - 1]) {
810
+ if (fr->is_map && fr->consumed % 2 == 1) {
699
811
  cbor_fail(&c, YEP_ERR_UNEXPECTED,
700
812
  "break in a map value position (odd item count)");
701
- return YEPTRIS_ERROR_PARSE;
813
+ rc = YEPTRIS_ERROR_PARSE;
814
+ goto done;
702
815
  }
703
816
  c.i++;
704
- c.d->depth--;
817
+ if (!c.s->close(c.s->ctx, fr->is_map)) {
818
+ rc = YEPTRIS_ERROR_MEMORY;
819
+ goto done;
820
+ }
705
821
  c.depth--;
706
822
  continue;
707
823
  }
@@ -709,16 +825,33 @@ int yep_cbor_decode_dom(yep_dom* d, const unsigned char* p, size_t len, int stri
709
825
  size_t parent = (size_t)(c.depth - 1);
710
826
  int st = cbor_item(&c);
711
827
  if (st == 0) {
712
- return (int)c.status;
828
+ rc = (int)c.status;
829
+ goto done;
713
830
  }
714
831
  if (st == 2) {
715
832
  continue; /* tags never consume a container slot */
716
833
  }
834
+ c.frame[parent].consumed++;
717
835
  if (c.frame[parent].rem != YEP_CBOR_INDEF) {
718
836
  c.frame[parent].rem--;
719
837
  /* one item (scalar, or a container now open) consumed a parent slot */
720
838
  }
721
839
  }
840
+ done:
841
+ free(c.ibuf);
842
+ return rc;
843
+ }
844
+
845
+ int yep_cbor_decode_dom_sink(void* dp, const unsigned char* p, size_t len, int strict,
846
+ size_t* consumed) {
847
+ yep_cbor_sink s = yep_cbor_dom_sink_impl;
848
+ s.ctx = dp;
849
+ return yep_cbor_decode_gen(p, len, strict, consumed, &s);
850
+ }
851
+
852
+ int yep_cbor_decode_dom(yep_dom* d, const unsigned char* p, size_t len, int strict,
853
+ size_t* consumed) {
854
+ return yep_cbor_decode_dom_sink(d, p, len, strict, consumed);
722
855
  }
723
856
 
724
857
  static YeptrisDocument cbor_wrap(yep_dom* dom, const void* buf, const yep_allocator* sys) {