… | |
… | |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
|
|
11 | #include <inttypes.h> |
11 | |
12 | |
12 | #define ECB_NO_THREADS 1 |
13 | #define ECB_NO_THREADS 1 |
13 | #include "ecb.h" |
14 | #include "ecb.h" |
14 | |
15 | |
15 | // compatibility with perl <5.18 |
16 | // compatibility with perl <5.18 |
… | |
… | |
20 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 | #endif |
22 | #endif |
22 | #ifndef HvNAMEUTF8 |
23 | #ifndef HvNAMEUTF8 |
23 | # define HvNAMEUTF8(hv) 0 |
24 | # define HvNAMEUTF8(hv) 0 |
24 | #endif |
25 | #endif |
|
|
26 | #ifndef SvREFCNT_inc_NN |
|
|
27 | # define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv) |
|
|
28 | #endif |
25 | #ifndef SvREFCNT_dec_NN |
29 | #ifndef SvREFCNT_dec_NN |
26 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
30 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
27 | #endif |
31 | #endif |
|
|
32 | |
|
|
33 | // perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0 |
|
|
34 | #define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len))) |
28 | |
35 | |
29 | // known major and minor types |
36 | // known major and minor types |
30 | enum cbor_type |
37 | enum cbor_type |
31 | { |
38 | { |
32 | MAJOR_SHIFT = 5, |
39 | MAJOR_SHIFT = 5, |
… | |
… | |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
102 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
96 | |
103 | |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
104 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
98 | }; |
105 | }; |
99 | |
106 | |
|
|
107 | // known forced types, also hardcoded in CBOR.pm |
|
|
108 | enum |
|
|
109 | { |
|
|
110 | AS_CBOR = 0, |
|
|
111 | AS_INT = 1, |
|
|
112 | AS_BYTES = 2, |
|
|
113 | AS_TEXT = 3, |
|
|
114 | AS_FLOAT16 = 4, |
|
|
115 | AS_FLOAT32 = 5, |
|
|
116 | AS_FLOAT64 = 6, |
|
|
117 | AS_MAP = 7, |
|
|
118 | // possibly future enhancements: (generic) float, (generic) string |
|
|
119 | }; |
|
|
120 | |
100 | #define F_SHRINK 0x00000001UL |
121 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
122 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
123 | #define F_ALLOW_SHARING 0x00000004UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
124 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
125 | #define F_ALLOW_WEAK_CYCLES 0x00000010UL |
|
|
126 | #define F_FORBID_OBJECTS 0x00000020UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
127 | #define F_PACK_STRINGS 0x00000040UL |
|
|
128 | #define F_TEXT_KEYS 0x00000080UL |
|
|
129 | #define F_TEXT_STRINGS 0x00000100UL |
105 | #define F_VALIDATE_UTF8 0x00000020UL |
130 | #define F_VALIDATE_UTF8 0x00000200UL |
106 | |
131 | |
107 | #define INIT_SIZE 32 // initial scalar size to be allocated |
132 | #define INIT_SIZE 32 // initial scalar size to be allocated |
108 | |
133 | |
109 | #define SB do { |
134 | #define SB do { |
110 | #define SE } while (0) |
135 | #define SE } while (0) |
… | |
… | |
180 | #endif |
205 | #endif |
181 | } |
206 | } |
182 | } |
207 | } |
183 | |
208 | |
184 | // minimum length of a string to be registered for stringref |
209 | // minimum length of a string to be registered for stringref |
185 | ecb_inline int |
210 | ecb_inline STRLEN |
186 | minimum_string_length (UV idx) |
211 | minimum_string_length (UV idx) |
187 | { |
212 | { |
188 | return idx > 23 |
213 | return idx <= 23 ? 3 |
189 | ? idx > 0xffU |
214 | : idx <= 0xffU ? 4 |
190 | ? idx > 0xffffU |
215 | : idx <= 0xffffU ? 5 |
191 | ? idx > 0xffffffffU |
216 | : idx <= 0xffffffffU ? 7 |
192 | ? 11 |
217 | : 11; |
193 | : 7 |
|
|
194 | : 5 |
|
|
195 | : 4 |
|
|
196 | : 3; |
|
|
197 | } |
218 | } |
198 | |
219 | |
199 | ///////////////////////////////////////////////////////////////////////////// |
220 | ///////////////////////////////////////////////////////////////////////////// |
200 | // encoder |
221 | // encoder |
201 | |
222 | |
… | |
… | |
214 | } enc_t; |
235 | } enc_t; |
215 | |
236 | |
216 | ecb_inline void |
237 | ecb_inline void |
217 | need (enc_t *enc, STRLEN len) |
238 | need (enc_t *enc, STRLEN len) |
218 | { |
239 | { |
219 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
240 | if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
220 | { |
241 | { |
221 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
242 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
222 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
243 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
223 | enc->cur = SvPVX (enc->sv) + cur; |
244 | enc->cur = SvPVX (enc->sv) + cur; |
224 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
245 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
225 | } |
246 | } |
226 | } |
247 | } |
227 | |
248 | |
|
|
249 | static void encode_sv (enc_t *enc, SV *sv); |
|
|
250 | |
228 | ecb_inline void |
251 | ecb_inline void |
229 | encode_ch (enc_t *enc, char ch) |
252 | encode_ch (enc_t *enc, char ch) |
230 | { |
253 | { |
231 | need (enc, 1); |
254 | need (enc, 1); |
232 | *enc->cur++ = ch; |
255 | *enc->cur++ = ch; |
233 | } |
256 | } |
234 | |
257 | |
|
|
258 | // used for tags, intregers, element counts and so on |
235 | static void |
259 | static void |
236 | encode_uint (enc_t *enc, int major, UV len) |
260 | encode_uint (enc_t *enc, int major, UV len) |
237 | { |
261 | { |
238 | need (enc, 9); |
262 | need (enc, 9); |
239 | |
263 | |
… | |
… | |
270 | *enc->cur++ = len >> 8; |
294 | *enc->cur++ = len >> 8; |
271 | *enc->cur++ = len; |
295 | *enc->cur++ = len; |
272 | } |
296 | } |
273 | } |
297 | } |
274 | |
298 | |
|
|
299 | // encodes a perl value into a CBOR integer |
|
|
300 | ecb_inline void |
|
|
301 | encode_int (enc_t *enc, SV *sv) |
|
|
302 | { |
|
|
303 | if (SvIsUV (sv)) |
|
|
304 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
|
|
305 | else if (SvIVX (sv) >= 0) |
|
|
306 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
|
|
307 | else |
|
|
308 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
|
|
309 | } |
|
|
310 | |
275 | ecb_inline void |
311 | ecb_inline void |
276 | encode_tag (enc_t *enc, UV tag) |
312 | encode_tag (enc_t *enc, UV tag) |
277 | { |
313 | { |
278 | encode_uint (enc, MAJOR_TAG, tag); |
314 | encode_uint (enc, MAJOR_TAG, tag); |
279 | } |
315 | } |
280 | |
316 | |
|
|
317 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
318 | ecb_noinline static void |
|
|
319 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
320 | { |
|
|
321 | STRLEN ulen = len; |
|
|
322 | U8 *p, *pend = (U8 *)str + len; |
|
|
323 | |
|
|
324 | for (p = (U8 *)str; p < pend; ++p) |
|
|
325 | ulen += *p >> 7; // count set high bits |
|
|
326 | |
|
|
327 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
328 | |
|
|
329 | need (enc, ulen); |
|
|
330 | for (p = (U8 *)str; p < pend; ++p) |
|
|
331 | if (*p < 0x80) |
|
|
332 | *enc->cur++ = *p; |
|
|
333 | else |
|
|
334 | { |
|
|
335 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
336 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
337 | } |
|
|
338 | } |
|
|
339 | |
281 | ecb_inline void |
340 | ecb_inline void |
282 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
341 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
283 | { |
342 | { |
|
|
343 | if (ecb_expect_false (upgrade_utf8)) |
|
|
344 | if (!utf8) |
|
|
345 | { |
|
|
346 | encode_str_utf8 (enc, utf8, str, len); |
|
|
347 | return; |
|
|
348 | } |
|
|
349 | |
284 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
350 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
285 | need (enc, len); |
351 | need (enc, len); |
286 | memcpy (enc->cur, str, len); |
352 | memcpy (enc->cur, str, len); |
287 | enc->cur += len; |
353 | enc->cur += len; |
288 | } |
354 | } |
289 | |
355 | |
290 | static void |
356 | ecb_inline void |
291 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
357 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
292 | { |
358 | { |
293 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
359 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
294 | { |
360 | { |
295 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
361 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
296 | |
362 | |
… | |
… | |
307 | sv_setuv (*svp, enc->stringref_idx); |
373 | sv_setuv (*svp, enc->stringref_idx); |
308 | ++enc->stringref_idx; |
374 | ++enc->stringref_idx; |
309 | } |
375 | } |
310 | } |
376 | } |
311 | |
377 | |
312 | encode_str (enc, utf8, str, len); |
378 | encode_str (enc, upgrade_utf8, utf8, str, len); |
313 | } |
379 | } |
314 | |
380 | |
315 | static void encode_sv (enc_t *enc, SV *sv); |
381 | ecb_inline void |
|
|
382 | encode_float16 (enc_t *enc, NV nv) |
|
|
383 | { |
|
|
384 | need (enc, 1+2); |
|
|
385 | |
|
|
386 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT16; |
|
|
387 | |
|
|
388 | uint16_t fp = ecb_float_to_binary16 (nv); |
|
|
389 | |
|
|
390 | if (!ecb_big_endian ()) |
|
|
391 | fp = ecb_bswap16 (fp); |
|
|
392 | |
|
|
393 | memcpy (enc->cur, &fp, 2); |
|
|
394 | enc->cur += 2; |
|
|
395 | } |
|
|
396 | |
|
|
397 | ecb_inline void |
|
|
398 | encode_float32 (enc_t *enc, NV nv) |
|
|
399 | { |
|
|
400 | need (enc, 1+4); |
|
|
401 | |
|
|
402 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
403 | |
|
|
404 | uint32_t fp = ecb_float_to_binary32 (nv); |
|
|
405 | |
|
|
406 | if (!ecb_big_endian ()) |
|
|
407 | fp = ecb_bswap32 (fp); |
|
|
408 | |
|
|
409 | memcpy (enc->cur, &fp, 4); |
|
|
410 | enc->cur += 4; |
|
|
411 | } |
|
|
412 | |
|
|
413 | ecb_inline void |
|
|
414 | encode_float64 (enc_t *enc, NV nv) |
|
|
415 | { |
|
|
416 | need (enc, 1+8); |
|
|
417 | |
|
|
418 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
419 | |
|
|
420 | uint64_t fp = ecb_double_to_binary64 (nv); |
|
|
421 | |
|
|
422 | if (!ecb_big_endian ()) |
|
|
423 | fp = ecb_bswap64 (fp); |
|
|
424 | |
|
|
425 | memcpy (enc->cur, &fp, 8); |
|
|
426 | enc->cur += 8; |
|
|
427 | } |
|
|
428 | |
|
|
429 | ecb_inline void |
|
|
430 | encode_bool (enc_t *enc, int istrue) |
|
|
431 | { |
|
|
432 | encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
433 | } |
|
|
434 | |
|
|
435 | // encodes an arrayref containing key-value pairs as CBOR map |
|
|
436 | ecb_inline void |
|
|
437 | encode_array_as_map (enc_t *enc, SV *sv) |
|
|
438 | { |
|
|
439 | if (enc->depth >= enc->cbor.max_depth) |
|
|
440 | croak (ERR_NESTING_EXCEEDED); |
|
|
441 | |
|
|
442 | ++enc->depth; |
|
|
443 | |
|
|
444 | // as_map does error checking for us, but we re-check in case |
|
|
445 | // things have changed. |
|
|
446 | |
|
|
447 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
|
|
448 | croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)"); |
|
|
449 | |
|
|
450 | AV *av = (AV *)SvRV (sv); |
|
|
451 | int i, len = av_len (av); |
|
|
452 | |
|
|
453 | if (!(len & 1)) |
|
|
454 | croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)"); |
|
|
455 | |
|
|
456 | encode_uint (enc, MAJOR_MAP, (len + 1) >> 1); |
|
|
457 | |
|
|
458 | for (i = 0; i <= len; ++i) |
|
|
459 | { |
|
|
460 | SV **svp = av_fetch (av, i, 0); |
|
|
461 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
|
|
462 | } |
|
|
463 | |
|
|
464 | --enc->depth; |
|
|
465 | } |
|
|
466 | |
|
|
467 | ecb_inline void |
|
|
468 | encode_forced (enc_t *enc, UV type, SV *sv) |
|
|
469 | { |
|
|
470 | switch (type) |
|
|
471 | { |
|
|
472 | case AS_CBOR: |
|
|
473 | { |
|
|
474 | STRLEN len; |
|
|
475 | char *str = SvPVbyte (sv, len); |
|
|
476 | |
|
|
477 | need (enc, len); |
|
|
478 | memcpy (enc->cur, str, len); |
|
|
479 | enc->cur += len; |
|
|
480 | } |
|
|
481 | break; |
|
|
482 | |
|
|
483 | case AS_BYTES: |
|
|
484 | { |
|
|
485 | STRLEN len; |
|
|
486 | char *str = SvPVbyte (sv, len); |
|
|
487 | encode_strref (enc, 0, 0, str, len); |
|
|
488 | } |
|
|
489 | break; |
|
|
490 | |
|
|
491 | case AS_TEXT: |
|
|
492 | { |
|
|
493 | STRLEN len; |
|
|
494 | char *str = SvPVutf8 (sv, len); |
|
|
495 | encode_strref (enc, 1, 1, str, len); |
|
|
496 | } |
|
|
497 | break; |
|
|
498 | |
|
|
499 | case AS_INT: encode_int (enc, sv); break; |
|
|
500 | |
|
|
501 | case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break; |
|
|
502 | case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break; |
|
|
503 | case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break; |
|
|
504 | |
|
|
505 | case AS_MAP: encode_array_as_map (enc, sv); break; |
|
|
506 | |
|
|
507 | default: |
|
|
508 | croak ("encountered malformed CBOR::XS::Tagged object"); |
|
|
509 | } |
|
|
510 | } |
316 | |
511 | |
317 | static void |
512 | static void |
318 | encode_av (enc_t *enc, AV *av) |
513 | encode_av (enc_t *enc, AV *av) |
319 | { |
514 | { |
320 | int i, len = av_len (av); |
515 | int i, len = av_len (av); |
… | |
… | |
324 | |
519 | |
325 | ++enc->depth; |
520 | ++enc->depth; |
326 | |
521 | |
327 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
522 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
328 | |
523 | |
|
|
524 | if (ecb_expect_false (SvMAGICAL (av))) |
329 | for (i = 0; i <= len; ++i) |
525 | for (i = 0; i <= len; ++i) |
330 | { |
526 | { |
331 | SV **svp = av_fetch (av, i, 0); |
527 | SV **svp = av_fetch (av, i, 0); |
332 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
528 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
333 | } |
529 | } |
|
|
530 | else |
|
|
531 | for (i = 0; i <= len; ++i) |
|
|
532 | { |
|
|
533 | SV *sv = AvARRAY (av)[i]; |
|
|
534 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
535 | } |
334 | |
536 | |
335 | --enc->depth; |
537 | --enc->depth; |
336 | } |
538 | } |
337 | |
539 | |
338 | static void |
540 | static void |
… | |
… | |
346 | ++enc->depth; |
548 | ++enc->depth; |
347 | |
549 | |
348 | int pairs = hv_iterinit (hv); |
550 | int pairs = hv_iterinit (hv); |
349 | int mg = SvMAGICAL (hv); |
551 | int mg = SvMAGICAL (hv); |
350 | |
552 | |
351 | if (mg) |
553 | if (ecb_expect_false (mg)) |
352 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
554 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
353 | else |
555 | else |
354 | encode_uint (enc, MAJOR_MAP, pairs); |
556 | encode_uint (enc, MAJOR_MAP, pairs); |
355 | |
557 | |
356 | while ((he = hv_iternext (hv))) |
558 | while ((he = hv_iternext (hv))) |
357 | { |
559 | { |
358 | if (HeKLEN (he) == HEf_SVKEY) |
560 | if (HeKLEN (he) == HEf_SVKEY) |
359 | encode_sv (enc, HeSVKEY (he)); |
561 | encode_sv (enc, HeSVKEY (he)); |
360 | else |
562 | else |
361 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
563 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
362 | |
564 | |
363 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
565 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
364 | } |
566 | } |
365 | |
567 | |
366 | if (mg) |
568 | if (ecb_expect_false (mg)) |
367 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
569 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
368 | |
570 | |
369 | --enc->depth; |
571 | --enc->depth; |
370 | } |
572 | } |
371 | |
573 | |
… | |
… | |
391 | |
593 | |
392 | HV *stash = SvSTASH (sv); |
594 | HV *stash = SvSTASH (sv); |
393 | |
595 | |
394 | if (stash == boolean_stash) |
596 | if (stash == boolean_stash) |
395 | { |
597 | { |
396 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
598 | encode_bool (enc, SvIV (sv)); |
397 | return; |
599 | return; |
398 | } |
600 | } |
399 | else if (stash == error_stash) |
601 | else if (stash == error_stash) |
400 | { |
602 | { |
401 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
603 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
… | |
… | |
404 | else if (stash == tagged_stash) |
606 | else if (stash == tagged_stash) |
405 | { |
607 | { |
406 | if (svt != SVt_PVAV) |
608 | if (svt != SVt_PVAV) |
407 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
609 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
408 | |
610 | |
|
|
611 | switch (av_len ((AV *)sv)) |
|
|
612 | { |
|
|
613 | case 2-1: |
|
|
614 | // actually a tagged value |
409 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
615 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
410 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
616 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
617 | break; |
|
|
618 | |
|
|
619 | case 3-1: |
|
|
620 | // a forced type [value, type, undef] |
|
|
621 | encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1)); |
|
|
622 | break; |
|
|
623 | |
|
|
624 | default: |
|
|
625 | croak ("encountered malformed CBOR::XS::Tagged object"); |
|
|
626 | } |
411 | |
627 | |
412 | return; |
628 | return; |
413 | } |
629 | } |
414 | } |
630 | } |
415 | |
631 | |
416 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
632 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
417 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
633 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
418 | { |
634 | { |
419 | if (!enc->shareable) |
635 | if (ecb_expect_false (!enc->shareable)) |
420 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
636 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
421 | |
637 | |
422 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
638 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
423 | |
639 | |
424 | if (SvOK (*svp)) |
640 | if (SvOK (*svp)) |
… | |
… | |
438 | if (ecb_expect_false (SvOBJECT (sv))) |
654 | if (ecb_expect_false (SvOBJECT (sv))) |
439 | { |
655 | { |
440 | HV *stash = SvSTASH (sv); |
656 | HV *stash = SvSTASH (sv); |
441 | GV *method; |
657 | GV *method; |
442 | |
658 | |
|
|
659 | if (enc->cbor.flags & F_FORBID_OBJECTS) |
|
|
660 | croak ("encountered object '%s', but forbid_objects is enabled", |
|
|
661 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
443 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
662 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
444 | { |
663 | { |
445 | dSP; |
664 | dSP; |
446 | |
665 | |
447 | ENTER; SAVETMPS; PUSHMARK (SP); |
666 | ENTER; SAVETMPS; |
|
|
667 | PUSHMARK (SP); |
448 | // we re-bless the reference to get overload and other niceties right |
668 | // we re-bless the reference to get overload and other niceties right |
449 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
669 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
450 | |
670 | |
451 | PUTBACK; |
671 | PUTBACK; |
452 | // G_SCALAR ensures that return value is 1 |
672 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
465 | } |
685 | } |
466 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
686 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
467 | { |
687 | { |
468 | dSP; |
688 | dSP; |
469 | |
689 | |
470 | ENTER; SAVETMPS; PUSHMARK (SP); |
690 | ENTER; SAVETMPS; |
|
|
691 | PUSHMARK (SP); |
471 | EXTEND (SP, 2); |
692 | EXTEND (SP, 2); |
472 | // we re-bless the reference to get overload and other niceties right |
693 | // we re-bless the reference to get overload and other niceties right |
473 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
694 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
474 | PUSHs (sv_cbor); |
695 | PUSHs (sv_cbor); |
475 | |
696 | |
… | |
… | |
481 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
702 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
482 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
703 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
483 | |
704 | |
484 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
705 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
485 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
706 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
486 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
707 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
487 | |
708 | |
488 | while (count) |
709 | { |
|
|
710 | int i; |
|
|
711 | |
|
|
712 | for (i = 0; i < count; ++i) |
489 | encode_sv (enc, SP[1 - count--]); |
713 | encode_sv (enc, SP[i + 1 - count]); |
|
|
714 | |
|
|
715 | SP -= count; |
|
|
716 | } |
490 | |
717 | |
491 | PUTBACK; |
718 | PUTBACK; |
492 | |
719 | |
493 | FREETMPS; LEAVE; |
720 | FREETMPS; LEAVE; |
494 | } |
721 | } |
… | |
… | |
516 | |
743 | |
517 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
744 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
518 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
745 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
519 | //TODO: maybe I32? |
746 | //TODO: maybe I32? |
520 | else if (ecb_expect_false (nv == (float)nv)) |
747 | else if (ecb_expect_false (nv == (float)nv)) |
521 | { |
748 | encode_float32 (enc, nv); |
522 | uint32_t fp = ecb_float_to_binary32 (nv); |
|
|
523 | |
|
|
524 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
525 | |
|
|
526 | if (!ecb_big_endian ()) |
|
|
527 | fp = ecb_bswap32 (fp); |
|
|
528 | |
|
|
529 | memcpy (enc->cur, &fp, 4); |
|
|
530 | enc->cur += 4; |
|
|
531 | } |
|
|
532 | else |
749 | else |
533 | { |
750 | encode_float64 (enc, nv); |
534 | uint64_t fp = ecb_double_to_binary64 (nv); |
|
|
535 | |
|
|
536 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
537 | |
|
|
538 | if (!ecb_big_endian ()) |
|
|
539 | fp = ecb_bswap64 (fp); |
|
|
540 | |
|
|
541 | memcpy (enc->cur, &fp, 8); |
|
|
542 | enc->cur += 8; |
|
|
543 | } |
|
|
544 | } |
751 | } |
545 | |
752 | |
546 | static void |
753 | static void |
547 | encode_sv (enc_t *enc, SV *sv) |
754 | encode_sv (enc_t *enc, SV *sv) |
548 | { |
755 | { |
… | |
… | |
550 | |
757 | |
551 | if (SvPOKp (sv)) |
758 | if (SvPOKp (sv)) |
552 | { |
759 | { |
553 | STRLEN len; |
760 | STRLEN len; |
554 | char *str = SvPV (sv, len); |
761 | char *str = SvPV (sv, len); |
555 | encode_strref (enc, SvUTF8 (sv), str, len); |
762 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
556 | } |
763 | } |
557 | else if (SvNOKp (sv)) |
764 | else if (SvNOKp (sv)) |
558 | encode_nv (enc, sv); |
765 | encode_nv (enc, sv); |
559 | else if (SvIOKp (sv)) |
766 | else if (SvIOKp (sv)) |
560 | { |
767 | encode_int (enc, sv); |
561 | if (SvIsUV (sv)) |
|
|
562 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
|
|
563 | else if (SvIVX (sv) >= 0) |
|
|
564 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
|
|
565 | else |
|
|
566 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
|
|
567 | } |
|
|
568 | else if (SvROK (sv)) |
768 | else if (SvROK (sv)) |
569 | encode_rv (enc, SvRV (sv)); |
769 | encode_rv (enc, SvRV (sv)); |
570 | else if (!SvOK (sv)) |
770 | else if (!SvOK (sv)) |
571 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
771 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
572 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
772 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
… | |
… | |
577 | } |
777 | } |
578 | |
778 | |
579 | static SV * |
779 | static SV * |
580 | encode_cbor (SV *scalar, CBOR *cbor) |
780 | encode_cbor (SV *scalar, CBOR *cbor) |
581 | { |
781 | { |
582 | enc_t enc = { }; |
782 | enc_t enc = { 0 }; |
583 | |
783 | |
584 | enc.cbor = *cbor; |
784 | enc.cbor = *cbor; |
585 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
785 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
586 | enc.cur = SvPVX (enc.sv); |
786 | enc.cur = SvPVX (enc.sv); |
587 | enc.end = SvEND (enc.sv); |
787 | enc.end = SvEND (enc.sv); |
588 | |
788 | |
589 | SvPOK_only (enc.sv); |
789 | SvPOK_only (enc.sv); |
590 | |
790 | |
591 | if (cbor->flags & F_PACK_STRINGS) |
791 | if (cbor->flags & F_PACK_STRINGS) |
592 | { |
792 | { |
… | |
… | |
619 | U32 depth; // recursion depth |
819 | U32 depth; // recursion depth |
620 | U32 maxdepth; // recursion depth limit |
820 | U32 maxdepth; // recursion depth limit |
621 | AV *shareable; |
821 | AV *shareable; |
622 | AV *stringref; |
822 | AV *stringref; |
623 | SV *decode_tagged; |
823 | SV *decode_tagged; |
|
|
824 | SV *err_sv; // optional sv for error, needs to be freed |
624 | } dec_t; |
825 | } dec_t; |
625 | |
826 | |
626 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
827 | // set dec->err to ERRSV |
|
|
828 | ecb_cold static void |
|
|
829 | err_errsv (dec_t *dec) |
|
|
830 | { |
|
|
831 | if (!dec->err) |
|
|
832 | { |
|
|
833 | dec->err_sv = newSVsv (ERRSV); |
627 | |
834 | |
628 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
835 | // chop off the trailing \n |
|
|
836 | SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1); |
|
|
837 | *SvEND (dec->err_sv) = 0; |
629 | |
838 | |
|
|
839 | dec->err = SvPVutf8_nolen (dec->err_sv); |
|
|
840 | } |
|
|
841 | } |
|
|
842 | |
|
|
843 | // the following functions are used to reduce code size and help the compiler to optimise |
|
|
844 | ecb_cold static void |
|
|
845 | err_set (dec_t *dec, const char *reason) |
|
|
846 | { |
|
|
847 | if (!dec->err) |
|
|
848 | dec->err = reason; |
|
|
849 | } |
|
|
850 | |
|
|
851 | ecb_cold static void |
|
|
852 | err_unexpected_end (dec_t *dec) |
|
|
853 | { |
|
|
854 | err_set (dec, "unexpected end of CBOR data"); |
|
|
855 | } |
|
|
856 | |
|
|
857 | #define ERR_DO(do) SB do; goto fail; SE |
|
|
858 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
|
|
859 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
|
|
860 | |
|
|
861 | #define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) |
|
|
862 | |
630 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
863 | #define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) |
631 | #define DEC_DEC_DEPTH --dec->depth |
864 | #define DEC_DEC_DEPTH --dec->depth |
632 | |
865 | |
633 | static UV |
866 | static UV |
634 | decode_uint (dec_t *dec) |
867 | decode_uint (dec_t *dec) |
635 | { |
868 | { |
… | |
… | |
701 | |
934 | |
702 | for (;;) |
935 | for (;;) |
703 | { |
936 | { |
704 | WANT (1); |
937 | WANT (1); |
705 | |
938 | |
706 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
939 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
707 | { |
940 | { |
708 | ++dec->cur; |
941 | ++dec->cur; |
709 | break; |
942 | break; |
710 | } |
943 | } |
711 | |
944 | |
712 | av_push (av, decode_sv (dec)); |
945 | av_push (av, decode_sv (dec)); |
713 | } |
946 | } |
714 | } |
947 | } |
715 | else |
948 | else |
716 | { |
949 | { |
717 | int i, len = decode_uint (dec); |
950 | UV i, len = decode_uint (dec); |
718 | |
951 | |
719 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
952 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
720 | av_fill (av, len - 1); |
953 | av_fill (av, len - 1); |
721 | |
954 | |
722 | for (i = 0; i < len; ++i) |
955 | for (i = 0; i < len; ++i) |
… | |
… | |
725 | |
958 | |
726 | DEC_DEC_DEPTH; |
959 | DEC_DEC_DEPTH; |
727 | return newRV_noinc ((SV *)av); |
960 | return newRV_noinc ((SV *)av); |
728 | |
961 | |
729 | fail: |
962 | fail: |
730 | SvREFCNT_dec (av); |
963 | SvREFCNT_dec_NN (av); |
731 | DEC_DEC_DEPTH; |
964 | DEC_DEC_DEPTH; |
732 | return &PL_sv_undef; |
965 | return &PL_sv_undef; |
733 | } |
966 | } |
734 | |
967 | |
735 | static void |
968 | static void |
… | |
… | |
739 | // byte or utf-8 strings as keys, but only when !stringref |
972 | // byte or utf-8 strings as keys, but only when !stringref |
740 | |
973 | |
741 | if (ecb_expect_true (!dec->stringref)) |
974 | if (ecb_expect_true (!dec->stringref)) |
742 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
975 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
743 | { |
976 | { |
744 | I32 len = decode_uint (dec); |
977 | STRLEN len = decode_uint (dec); |
745 | char *key = (char *)dec->cur; |
978 | char *key = (char *)dec->cur; |
746 | |
979 | |
|
|
980 | WANT (len); |
747 | dec->cur += len; |
981 | dec->cur += len; |
748 | |
982 | |
749 | hv_store (hv, key, len, decode_sv (dec), 0); |
983 | hv_store (hv, key, len, decode_sv (dec), 0); |
750 | |
984 | |
751 | return; |
985 | return; |
752 | } |
986 | } |
753 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
987 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
754 | { |
988 | { |
755 | I32 len = decode_uint (dec); |
989 | STRLEN len = decode_uint (dec); |
756 | char *key = (char *)dec->cur; |
990 | char *key = (char *)dec->cur; |
757 | |
991 | |
|
|
992 | WANT (len); |
758 | dec->cur += len; |
993 | dec->cur += len; |
759 | |
994 | |
760 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
995 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
761 | if (!is_utf8_string (key, len)) |
996 | if (!cbor_is_utf8_string ((U8 *)key, len)) |
762 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
997 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
763 | |
998 | |
764 | hv_store (hv, key, -len, decode_sv (dec), 0); |
999 | hv_store (hv, key, -len, decode_sv (dec), 0); |
765 | |
1000 | |
766 | return; |
1001 | return; |
767 | } |
1002 | } |
768 | |
1003 | |
769 | SV *k = decode_sv (dec); |
1004 | SV *k = decode_sv (dec); |
770 | SV *v = decode_sv (dec); |
1005 | SV *v = decode_sv (dec); |
771 | |
1006 | |
|
|
1007 | // we leak memory if uncaught exceptions are thrown by random magical |
|
|
1008 | // methods, and this is hopefully the only place where it can happen, |
|
|
1009 | // so if there is a chance of an exception, take the very slow path. |
|
|
1010 | // since catching exceptions is "undocumented/internal/forbidden" by |
|
|
1011 | // the new p5p powers, we need to call out to a perl function :/ |
|
|
1012 | if (ecb_expect_false (SvAMAGIC (k))) |
|
|
1013 | { |
|
|
1014 | dSP; |
|
|
1015 | |
|
|
1016 | ENTER; SAVETMPS; |
|
|
1017 | PUSHMARK (SP); |
|
|
1018 | EXTEND (SP, 3); |
|
|
1019 | PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); |
|
|
1020 | PUSHs (sv_2mortal (k)); |
|
|
1021 | PUSHs (sv_2mortal (v)); |
|
|
1022 | |
|
|
1023 | PUTBACK; |
|
|
1024 | call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); |
|
|
1025 | SPAGAIN; |
|
|
1026 | |
|
|
1027 | FREETMPS; LEAVE; |
|
|
1028 | |
|
|
1029 | if (SvTRUE (ERRSV)) |
|
|
1030 | ERR_ERRSV; |
|
|
1031 | |
|
|
1032 | return; |
|
|
1033 | } |
|
|
1034 | |
772 | hv_store_ent (hv, k, v, 0); |
1035 | hv_store_ent (hv, k, v, 0); |
773 | SvREFCNT_dec (k); |
1036 | SvREFCNT_dec_NN (k); |
774 | |
1037 | |
775 | fail: |
1038 | fail: |
776 | ; |
1039 | ; |
777 | } |
1040 | } |
778 | |
1041 | |
… | |
… | |
789 | |
1052 | |
790 | for (;;) |
1053 | for (;;) |
791 | { |
1054 | { |
792 | WANT (1); |
1055 | WANT (1); |
793 | |
1056 | |
794 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
1057 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) |
795 | { |
1058 | { |
796 | ++dec->cur; |
1059 | ++dec->cur; |
797 | break; |
1060 | break; |
798 | } |
1061 | } |
799 | |
1062 | |
800 | decode_he (dec, hv); |
1063 | decode_he (dec, hv); |
801 | } |
1064 | } |
802 | } |
1065 | } |
803 | else |
1066 | else |
804 | { |
1067 | { |
805 | int pairs = decode_uint (dec); |
1068 | UV pairs = decode_uint (dec); |
|
|
1069 | |
|
|
1070 | WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes |
806 | |
1071 | |
807 | while (pairs--) |
1072 | while (pairs--) |
808 | decode_he (dec, hv); |
1073 | decode_he (dec, hv); |
809 | } |
1074 | } |
810 | |
1075 | |
811 | DEC_DEC_DEPTH; |
1076 | DEC_DEC_DEPTH; |
812 | return newRV_noinc ((SV *)hv); |
1077 | return newRV_noinc ((SV *)hv); |
813 | |
1078 | |
814 | fail: |
1079 | fail: |
815 | SvREFCNT_dec (hv); |
1080 | SvREFCNT_dec_NN (hv); |
816 | DEC_DEC_DEPTH; |
1081 | DEC_DEC_DEPTH; |
817 | return &PL_sv_undef; |
1082 | return &PL_sv_undef; |
818 | } |
1083 | } |
819 | |
1084 | |
820 | static SV * |
1085 | static SV * |
821 | decode_str (dec_t *dec, int utf8) |
1086 | decode_str (dec_t *dec, int utf8) |
822 | { |
1087 | { |
823 | SV *sv = 0; |
1088 | SV *sv = 0; |
824 | |
1089 | |
825 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
1090 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
826 | { |
1091 | { |
827 | // indefinite length strings |
1092 | // indefinite length strings |
828 | ++dec->cur; |
1093 | ++dec->cur; |
829 | |
1094 | |
830 | U8 major = *dec->cur & MAJOR_MISC; |
1095 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
865 | } |
1130 | } |
866 | |
1131 | |
867 | if (utf8) |
1132 | if (utf8) |
868 | { |
1133 | { |
869 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
1134 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
870 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
1135 | if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv))) |
871 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
1136 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
872 | |
1137 | |
873 | SvUTF8_on (sv); |
1138 | SvUTF8_on (sv); |
874 | } |
1139 | } |
875 | |
1140 | |
… | |
… | |
898 | sv = newRV_noinc (decode_sv (dec)); |
1163 | sv = newRV_noinc (decode_sv (dec)); |
899 | break; |
1164 | break; |
900 | |
1165 | |
901 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1166 | case CBOR_TAG_STRINGREF_NAMESPACE: |
902 | { |
1167 | { |
|
|
1168 | // do not use SAVETMPS/FREETMPS, as these will |
|
|
1169 | // erase mortalised caches, e.g. "shareable" |
903 | ENTER; SAVETMPS; |
1170 | ENTER; |
904 | |
1171 | |
905 | SAVESPTR (dec->stringref); |
1172 | SAVESPTR (dec->stringref); |
906 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1173 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
907 | |
1174 | |
908 | sv = decode_sv (dec); |
1175 | sv = decode_sv (dec); |
909 | |
1176 | |
910 | FREETMPS; LEAVE; |
1177 | LEAVE; |
911 | } |
1178 | } |
912 | break; |
1179 | break; |
913 | |
1180 | |
914 | case CBOR_TAG_STRINGREF: |
1181 | case CBOR_TAG_STRINGREF: |
915 | { |
1182 | { |
916 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1183 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
917 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
1184 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
918 | |
1185 | |
919 | UV idx = decode_uint (dec); |
1186 | UV idx = decode_uint (dec); |
920 | |
1187 | |
921 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
1188 | if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref))) |
922 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
1189 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
923 | |
1190 | |
924 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
1191 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
925 | } |
1192 | } |
926 | break; |
1193 | break; |
… | |
… | |
928 | case CBOR_TAG_VALUE_SHAREABLE: |
1195 | case CBOR_TAG_VALUE_SHAREABLE: |
929 | { |
1196 | { |
930 | if (ecb_expect_false (!dec->shareable)) |
1197 | if (ecb_expect_false (!dec->shareable)) |
931 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
1198 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
932 | |
1199 | |
933 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
1200 | if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES | F_ALLOW_WEAK_CYCLES))) |
934 | { |
1201 | { |
|
|
1202 | // if cycles are allowed, then we store an AV as value |
|
|
1203 | // while it is being decoded, and gather unresolved |
|
|
1204 | // references in it, to be re4solved after decoding. |
|
|
1205 | int idx, i; |
935 | sv = newSV (0); |
1206 | AV *av = newAV (); |
936 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
1207 | av_push (dec->shareable, (SV *)av); |
|
|
1208 | idx = AvFILLp (dec->shareable); |
937 | |
1209 | |
938 | SV *osv = decode_sv (dec); |
1210 | sv = decode_sv (dec); |
939 | sv_setsv (sv, osv); |
1211 | |
|
|
1212 | // the AV now contains \undef for all unresolved references, |
|
|
1213 | // so we fix them up here. |
|
|
1214 | for (i = 0; i <= AvFILLp (av); ++i) |
|
|
1215 | SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv))); |
|
|
1216 | |
|
|
1217 | // weaken all recursive references |
|
|
1218 | if (dec->cbor.flags & F_ALLOW_WEAK_CYCLES) |
|
|
1219 | for (i = 0; i <= AvFILLp (av); ++i) |
|
|
1220 | sv_rvweaken (AvARRAY (av)[i]); |
|
|
1221 | |
|
|
1222 | // now replace the AV by a reference to the completed value |
940 | SvREFCNT_dec_NN (osv); |
1223 | SvREFCNT_dec_NN ((SV *)av); |
|
|
1224 | AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv); |
941 | } |
1225 | } |
942 | else |
1226 | else |
943 | { |
1227 | { |
944 | av_push (dec->shareable, &PL_sv_undef); |
1228 | av_push (dec->shareable, &PL_sv_undef); |
945 | int idx = AvFILLp (dec->shareable); |
1229 | int idx = AvFILLp (dec->shareable); |
946 | sv = decode_sv (dec); |
1230 | sv = decode_sv (dec); |
947 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
1231 | AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv); |
948 | } |
1232 | } |
949 | } |
1233 | } |
950 | break; |
1234 | break; |
951 | |
1235 | |
952 | case CBOR_TAG_VALUE_SHAREDREF: |
1236 | case CBOR_TAG_VALUE_SHAREDREF: |
… | |
… | |
954 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
1238 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
955 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
1239 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
956 | |
1240 | |
957 | UV idx = decode_uint (dec); |
1241 | UV idx = decode_uint (dec); |
958 | |
1242 | |
959 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1243 | if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable))) |
960 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1244 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
961 | |
1245 | |
962 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1246 | sv = AvARRAY (dec->shareable)[idx]; |
963 | |
1247 | |
964 | if (sv == &PL_sv_undef) |
1248 | // reference to cycle, we create a new \undef and use that, and also |
|
|
1249 | // registerr it in the AV for later fixing |
|
|
1250 | if (ecb_expect_false (SvTYPE (sv) == SVt_PVAV)) |
|
|
1251 | { |
|
|
1252 | AV *av = (AV *)sv; |
|
|
1253 | sv = newRV_noinc (&PL_sv_undef); |
|
|
1254 | av_push (av, SvREFCNT_inc_NN (sv)); |
|
|
1255 | } |
|
|
1256 | else if (ecb_expect_false (sv == &PL_sv_undef)) // not yet decoded, but cycles not allowed |
965 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
1257 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
|
|
1258 | else // we decoded the object earlier, no cycle |
|
|
1259 | sv = newSVsv (sv); |
966 | } |
1260 | } |
967 | break; |
1261 | break; |
968 | |
1262 | |
969 | case CBOR_TAG_PERL_OBJECT: |
1263 | case CBOR_TAG_PERL_OBJECT: |
970 | { |
1264 | { |
|
|
1265 | if (dec->cbor.flags & F_FORBID_OBJECTS) |
|
|
1266 | goto filter; |
|
|
1267 | |
971 | sv = decode_sv (dec); |
1268 | sv = decode_sv (dec); |
972 | |
1269 | |
973 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1270 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
974 | ERR ("corrupted CBOR data (non-array perl object)"); |
1271 | ERR ("corrupted CBOR data (non-array perl object)"); |
975 | |
1272 | |
… | |
… | |
985 | if (!method) |
1282 | if (!method) |
986 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1283 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
987 | |
1284 | |
988 | dSP; |
1285 | dSP; |
989 | |
1286 | |
990 | ENTER; SAVETMPS; PUSHMARK (SP); |
1287 | ENTER; SAVETMPS; |
|
|
1288 | PUSHMARK (SP); |
991 | EXTEND (SP, len + 1); |
1289 | EXTEND (SP, len + 1); |
992 | // we re-bless the reference to get overload and other niceties right |
1290 | // we re-bless the reference to get overload and other niceties right |
993 | PUSHs (*av_fetch (av, 0, 1)); |
1291 | PUSHs (*av_fetch (av, 0, 1)); |
994 | PUSHs (sv_cbor); |
1292 | PUSHs (sv_cbor); |
995 | |
1293 | |
… | |
… | |
1003 | SPAGAIN; |
1301 | SPAGAIN; |
1004 | |
1302 | |
1005 | if (SvTRUE (ERRSV)) |
1303 | if (SvTRUE (ERRSV)) |
1006 | { |
1304 | { |
1007 | FREETMPS; LEAVE; |
1305 | FREETMPS; LEAVE; |
1008 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1306 | ERR_ERRSV; |
1009 | } |
1307 | } |
1010 | |
1308 | |
1011 | SvREFCNT_dec (sv); |
1309 | SvREFCNT_dec_NN (sv); |
1012 | sv = SvREFCNT_inc (POPs); |
1310 | sv = SvREFCNT_inc (POPs); |
1013 | |
1311 | |
1014 | PUTBACK; |
1312 | PUTBACK; |
1015 | |
1313 | |
1016 | FREETMPS; LEAVE; |
1314 | FREETMPS; LEAVE; |
1017 | } |
1315 | } |
1018 | break; |
1316 | break; |
1019 | |
1317 | |
1020 | default: |
1318 | default: |
|
|
1319 | filter: |
1021 | { |
1320 | { |
|
|
1321 | SV *tag_sv = newSVuv (tag); |
|
|
1322 | |
1022 | sv = decode_sv (dec); |
1323 | sv = decode_sv (dec); |
1023 | |
1324 | |
1024 | dSP; |
1325 | dSP; |
1025 | ENTER; SAVETMPS; PUSHMARK (SP); |
1326 | ENTER; SAVETMPS; |
|
|
1327 | PUSHMARK (SP); |
1026 | EXTEND (SP, 2); |
1328 | EXTEND (SP, 2); |
1027 | PUSHs (newSVuv (tag)); |
1329 | PUSHs (tag_sv); |
1028 | PUSHs (sv); |
1330 | PUSHs (sv); |
1029 | |
1331 | |
1030 | PUTBACK; |
1332 | PUTBACK; |
1031 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1333 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1032 | SPAGAIN; |
1334 | SPAGAIN; |
1033 | |
1335 | |
1034 | if (SvTRUE (ERRSV)) |
1336 | if (SvTRUE (ERRSV)) |
1035 | { |
1337 | { |
|
|
1338 | SvREFCNT_dec_NN (tag_sv); |
1036 | FREETMPS; LEAVE; |
1339 | FREETMPS; LEAVE; |
1037 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1340 | ERR_ERRSV; |
1038 | } |
1341 | } |
1039 | |
1342 | |
1040 | if (count) |
1343 | if (count) |
1041 | { |
1344 | { |
|
|
1345 | SvREFCNT_dec_NN (tag_sv); |
1042 | SvREFCNT_dec (sv); |
1346 | SvREFCNT_dec_NN (sv); |
1043 | sv = SvREFCNT_inc (POPs); |
1347 | sv = SvREFCNT_inc_NN (TOPs); |
|
|
1348 | SP -= count; |
1044 | } |
1349 | } |
1045 | else |
1350 | else |
1046 | { |
1351 | { |
1047 | AV *av = newAV (); |
1352 | AV *av = newAV (); |
1048 | av_push (av, newSVuv (tag)); |
1353 | av_push (av, tag_sv); |
1049 | av_push (av, sv); |
1354 | av_push (av, sv); |
1050 | |
1355 | |
1051 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1356 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1052 | ? cbor_tagged_stash |
1357 | ? cbor_tagged_stash |
1053 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1358 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
… | |
… | |
1156 | } |
1461 | } |
1157 | |
1462 | |
1158 | static SV * |
1463 | static SV * |
1159 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1464 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1160 | { |
1465 | { |
1161 | dec_t dec = { }; |
1466 | dec_t dec = { 0 }; |
1162 | SV *sv; |
1467 | SV *sv; |
1163 | STRLEN len; |
1468 | STRLEN len; |
1164 | char *data = SvPVbyte (string, len); |
1469 | char *data = SvPVbyte (string, len); |
1165 | |
1470 | |
1166 | if (len > cbor->max_size && cbor->max_size) |
1471 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
1182 | |
1487 | |
1183 | if (dec.err) |
1488 | if (dec.err) |
1184 | { |
1489 | { |
1185 | if (dec.shareable) |
1490 | if (dec.shareable) |
1186 | { |
1491 | { |
1187 | // need to break cyclic links, which whould all be in shareable |
1492 | // need to break cyclic links, which would all be in shareable |
1188 | int i; |
1493 | int i; |
1189 | SV **svp; |
1494 | SV **svp; |
1190 | |
1495 | |
1191 | for (i = av_len (dec.shareable) + 1; i--; ) |
1496 | for (i = av_len (dec.shareable) + 1; i--; ) |
1192 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1497 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1193 | sv_setsv (*svp, &PL_sv_undef); |
1498 | sv_setsv (*svp, &PL_sv_undef); |
1194 | } |
1499 | } |
1195 | |
1500 | |
1196 | SvREFCNT_dec (sv); |
1501 | SvREFCNT_dec_NN (sv); |
|
|
1502 | |
|
|
1503 | if (dec.err_sv) |
|
|
1504 | sv_2mortal (dec.err_sv); |
|
|
1505 | |
1197 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1506 | croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur); |
1198 | } |
1507 | } |
1199 | |
1508 | |
1200 | sv = sv_2mortal (sv); |
1509 | sv = sv_2mortal (sv); |
1201 | |
1510 | |
1202 | return sv; |
1511 | return sv; |
… | |
… | |
1257 | |
1566 | |
1258 | int major = *p >> MAJOR_SHIFT; |
1567 | int major = *p >> MAJOR_SHIFT; |
1259 | |
1568 | |
1260 | switch (major) |
1569 | switch (major) |
1261 | { |
1570 | { |
|
|
1571 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1572 | ++count; // tags merely prefix another value |
|
|
1573 | break; |
|
|
1574 | |
1262 | case MAJOR_BYTES >> MAJOR_SHIFT: |
1575 | case MAJOR_BYTES >> MAJOR_SHIFT: |
1263 | case MAJOR_TEXT >> MAJOR_SHIFT: |
1576 | case MAJOR_TEXT >> MAJOR_SHIFT: |
1264 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1577 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1265 | case MAJOR_MAP >> MAJOR_SHIFT: |
1578 | case MAJOR_MAP >> MAJOR_SHIFT: |
1266 | { |
1579 | { |
… | |
… | |
1287 | |
1600 | |
1288 | break; |
1601 | break; |
1289 | |
1602 | |
1290 | case MAJOR_MAP >> MAJOR_SHIFT: |
1603 | case MAJOR_MAP >> MAJOR_SHIFT: |
1291 | len <<= 1; |
1604 | len <<= 1; |
|
|
1605 | /* FALLTHROUGH */ |
1292 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1606 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1293 | if (len) |
1607 | if (len) |
1294 | { |
1608 | { |
1295 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1609 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1296 | count = len + 1; |
1610 | count = len + 1; |
… | |
… | |
1341 | |
1655 | |
1342 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1656 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1343 | |
1657 | |
1344 | sv_cbor = newSVpv ("CBOR", 0); |
1658 | sv_cbor = newSVpv ("CBOR", 0); |
1345 | SvREADONLY_on (sv_cbor); |
1659 | SvREADONLY_on (sv_cbor); |
|
|
1660 | |
|
|
1661 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
1346 | } |
1662 | } |
1347 | |
1663 | |
1348 | PROTOTYPES: DISABLE |
1664 | PROTOTYPES: DISABLE |
1349 | |
1665 | |
1350 | void CLONE (...) |
1666 | void CLONE (...) |
… | |
… | |
1366 | ))); |
1682 | ))); |
1367 | } |
1683 | } |
1368 | |
1684 | |
1369 | void shrink (CBOR *self, int enable = 1) |
1685 | void shrink (CBOR *self, int enable = 1) |
1370 | ALIAS: |
1686 | ALIAS: |
1371 | shrink = F_SHRINK |
1687 | shrink = F_SHRINK |
1372 | allow_unknown = F_ALLOW_UNKNOWN |
1688 | allow_unknown = F_ALLOW_UNKNOWN |
1373 | allow_sharing = F_ALLOW_SHARING |
1689 | allow_sharing = F_ALLOW_SHARING |
1374 | allow_cycles = F_ALLOW_CYCLES |
1690 | allow_cycles = F_ALLOW_CYCLES |
|
|
1691 | allow_weak_cycles = F_ALLOW_WEAK_CYCLES |
|
|
1692 | forbid_objects = F_FORBID_OBJECTS |
1375 | pack_strings = F_PACK_STRINGS |
1693 | pack_strings = F_PACK_STRINGS |
|
|
1694 | text_keys = F_TEXT_KEYS |
|
|
1695 | text_strings = F_TEXT_STRINGS |
1376 | validate_utf8 = F_VALIDATE_UTF8 |
1696 | validate_utf8 = F_VALIDATE_UTF8 |
1377 | PPCODE: |
1697 | PPCODE: |
1378 | { |
1698 | { |
1379 | if (enable) |
1699 | if (enable) |
1380 | self->flags |= ix; |
1700 | self->flags |= ix; |
1381 | else |
1701 | else |
… | |
… | |
1384 | XPUSHs (ST (0)); |
1704 | XPUSHs (ST (0)); |
1385 | } |
1705 | } |
1386 | |
1706 | |
1387 | void get_shrink (CBOR *self) |
1707 | void get_shrink (CBOR *self) |
1388 | ALIAS: |
1708 | ALIAS: |
1389 | get_shrink = F_SHRINK |
1709 | get_shrink = F_SHRINK |
1390 | get_allow_unknown = F_ALLOW_UNKNOWN |
1710 | get_allow_unknown = F_ALLOW_UNKNOWN |
1391 | get_allow_sharing = F_ALLOW_SHARING |
1711 | get_allow_sharing = F_ALLOW_SHARING |
1392 | get_allow_cycles = F_ALLOW_CYCLES |
1712 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1713 | get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES |
|
|
1714 | get_forbid_objects = F_FORBID_OBJECTS |
1393 | get_pack_strings = F_PACK_STRINGS |
1715 | get_pack_strings = F_PACK_STRINGS |
|
|
1716 | get_text_keys = F_TEXT_KEYS |
|
|
1717 | get_text_strings = F_TEXT_STRINGS |
1394 | get_validate_utf8 = F_VALIDATE_UTF8 |
1718 | get_validate_utf8 = F_VALIDATE_UTF8 |
1395 | PPCODE: |
1719 | PPCODE: |
1396 | XPUSHs (boolSV (self->flags & ix)); |
1720 | XPUSHs (boolSV (self->flags & ix)); |
1397 | |
1721 | |
1398 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1722 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1399 | PPCODE: |
1723 | PPCODE: |
… | |
… | |
1528 | cbor_init (&cbor); |
1852 | cbor_init (&cbor); |
1529 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1853 | PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1530 | XPUSHs (cborstr); |
1854 | XPUSHs (cborstr); |
1531 | } |
1855 | } |
1532 | |
1856 | |
|
|
1857 | #ifdef __AFL_COMPILER |
|
|
1858 | |
|
|
1859 | void |
|
|
1860 | afl_init () |
|
|
1861 | CODE: |
|
|
1862 | __AFL_INIT (); |
|
|
1863 | |
|
|
1864 | int |
|
|
1865 | afl_loop (unsigned int count = 10000) |
|
|
1866 | CODE: |
|
|
1867 | RETVAL = __AFL_LOOP (count); |
|
|
1868 | OUTPUT: |
|
|
1869 | RETVAL |
|
|
1870 | |
|
|
1871 | #endif |
|
|
1872 | |