… | |
… | |
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 |
28 | |
32 | |
29 | // known major and minor types |
33 | // known major and minor types |
… | |
… | |
99 | |
103 | |
100 | #define F_SHRINK 0x00000001UL |
104 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
105 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
106 | #define F_ALLOW_SHARING 0x00000004UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
107 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
108 | #define F_FORBID_OBJECTS 0x00000010UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
109 | #define F_PACK_STRINGS 0x00000020UL |
|
|
110 | #define F_TEXT_KEYS 0x00000040UL |
|
|
111 | #define F_TEXT_STRINGS 0x00000080UL |
105 | #define F_VALIDATE_UTF8 0x00000020UL |
112 | #define F_VALIDATE_UTF8 0x00000100UL |
106 | |
113 | |
107 | #define INIT_SIZE 32 // initial scalar size to be allocated |
114 | #define INIT_SIZE 32 // initial scalar size to be allocated |
108 | |
115 | |
109 | #define SB do { |
116 | #define SB do { |
110 | #define SE } while (0) |
117 | #define SE } while (0) |
… | |
… | |
183 | |
190 | |
184 | // minimum length of a string to be registered for stringref |
191 | // minimum length of a string to be registered for stringref |
185 | ecb_inline int |
192 | ecb_inline int |
186 | minimum_string_length (UV idx) |
193 | minimum_string_length (UV idx) |
187 | { |
194 | { |
188 | return idx > 23 |
195 | return idx <= 23 ? 3 |
189 | ? idx > 0xffU |
196 | : idx <= 0xffU ? 4 |
190 | ? idx > 0xffffU |
197 | : idx <= 0xffffU ? 5 |
191 | ? idx > 0xffffffffU |
198 | : idx <= 0xffffffffU ? 7 |
192 | ? 11 |
199 | : 11; |
193 | : 7 |
|
|
194 | : 5 |
|
|
195 | : 4 |
|
|
196 | : 3; |
|
|
197 | } |
200 | } |
198 | |
201 | |
199 | ///////////////////////////////////////////////////////////////////////////// |
202 | ///////////////////////////////////////////////////////////////////////////// |
200 | // encoder |
203 | // encoder |
201 | |
204 | |
… | |
… | |
214 | } enc_t; |
217 | } enc_t; |
215 | |
218 | |
216 | ecb_inline void |
219 | ecb_inline void |
217 | need (enc_t *enc, STRLEN len) |
220 | need (enc_t *enc, STRLEN len) |
218 | { |
221 | { |
219 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
222 | if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
220 | { |
223 | { |
221 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
224 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
222 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
225 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
223 | enc->cur = SvPVX (enc->sv) + cur; |
226 | enc->cur = SvPVX (enc->sv) + cur; |
224 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
227 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
… | |
… | |
276 | encode_tag (enc_t *enc, UV tag) |
279 | encode_tag (enc_t *enc, UV tag) |
277 | { |
280 | { |
278 | encode_uint (enc, MAJOR_TAG, tag); |
281 | encode_uint (enc, MAJOR_TAG, tag); |
279 | } |
282 | } |
280 | |
283 | |
|
|
284 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
285 | ecb_noinline static void |
|
|
286 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
287 | { |
|
|
288 | STRLEN ulen = len; |
|
|
289 | U8 *p, *pend = (U8 *)str + len; |
|
|
290 | |
|
|
291 | for (p = (U8 *)str; p < pend; ++p) |
|
|
292 | ulen += *p >> 7; // count set high bits |
|
|
293 | |
|
|
294 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
295 | |
|
|
296 | need (enc, ulen); |
|
|
297 | for (p = (U8 *)str; p < pend; ++p) |
|
|
298 | if (*p < 0x80) |
|
|
299 | *enc->cur++ = *p; |
|
|
300 | else |
|
|
301 | { |
|
|
302 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
303 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
304 | } |
|
|
305 | } |
|
|
306 | |
281 | ecb_inline void |
307 | ecb_inline void |
282 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
308 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
283 | { |
309 | { |
|
|
310 | if (ecb_expect_false (upgrade_utf8)) |
|
|
311 | if (!utf8) |
|
|
312 | { |
|
|
313 | encode_str_utf8 (enc, utf8, str, len); |
|
|
314 | return; |
|
|
315 | } |
|
|
316 | |
284 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
317 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
285 | need (enc, len); |
318 | need (enc, len); |
286 | memcpy (enc->cur, str, len); |
319 | memcpy (enc->cur, str, len); |
287 | enc->cur += len; |
320 | enc->cur += len; |
288 | } |
321 | } |
289 | |
322 | |
290 | static void |
323 | ecb_inline void |
291 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
324 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
292 | { |
325 | { |
293 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
326 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
294 | { |
327 | { |
295 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
328 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
296 | |
329 | |
… | |
… | |
307 | sv_setuv (*svp, enc->stringref_idx); |
340 | sv_setuv (*svp, enc->stringref_idx); |
308 | ++enc->stringref_idx; |
341 | ++enc->stringref_idx; |
309 | } |
342 | } |
310 | } |
343 | } |
311 | |
344 | |
312 | encode_str (enc, utf8, str, len); |
345 | encode_str (enc, upgrade_utf8, utf8, str, len); |
313 | } |
346 | } |
314 | |
347 | |
315 | static void encode_sv (enc_t *enc, SV *sv); |
348 | static void encode_sv (enc_t *enc, SV *sv); |
316 | |
349 | |
317 | static void |
350 | static void |
… | |
… | |
324 | |
357 | |
325 | ++enc->depth; |
358 | ++enc->depth; |
326 | |
359 | |
327 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
360 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
328 | |
361 | |
|
|
362 | if (ecb_expect_false (SvMAGICAL (av))) |
329 | for (i = 0; i <= len; ++i) |
363 | for (i = 0; i <= len; ++i) |
330 | { |
364 | { |
331 | SV **svp = av_fetch (av, i, 0); |
365 | SV **svp = av_fetch (av, i, 0); |
332 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
366 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
333 | } |
367 | } |
|
|
368 | else |
|
|
369 | for (i = 0; i <= len; ++i) |
|
|
370 | { |
|
|
371 | SV *sv = AvARRAY (av)[i]; |
|
|
372 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
373 | } |
334 | |
374 | |
335 | --enc->depth; |
375 | --enc->depth; |
336 | } |
376 | } |
337 | |
377 | |
338 | static void |
378 | static void |
… | |
… | |
346 | ++enc->depth; |
386 | ++enc->depth; |
347 | |
387 | |
348 | int pairs = hv_iterinit (hv); |
388 | int pairs = hv_iterinit (hv); |
349 | int mg = SvMAGICAL (hv); |
389 | int mg = SvMAGICAL (hv); |
350 | |
390 | |
351 | if (mg) |
391 | if (ecb_expect_false (mg)) |
352 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
392 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
353 | else |
393 | else |
354 | encode_uint (enc, MAJOR_MAP, pairs); |
394 | encode_uint (enc, MAJOR_MAP, pairs); |
355 | |
395 | |
356 | while ((he = hv_iternext (hv))) |
396 | while ((he = hv_iternext (hv))) |
357 | { |
397 | { |
358 | if (HeKLEN (he) == HEf_SVKEY) |
398 | if (HeKLEN (he) == HEf_SVKEY) |
359 | encode_sv (enc, HeSVKEY (he)); |
399 | encode_sv (enc, HeSVKEY (he)); |
360 | else |
400 | else |
361 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
401 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
362 | |
402 | |
363 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
403 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
364 | } |
404 | } |
365 | |
405 | |
366 | if (mg) |
406 | if (ecb_expect_false (mg)) |
367 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
407 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
368 | |
408 | |
369 | --enc->depth; |
409 | --enc->depth; |
370 | } |
410 | } |
371 | |
411 | |
… | |
… | |
414 | } |
454 | } |
415 | |
455 | |
416 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
456 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
417 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
457 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
418 | { |
458 | { |
419 | if (!enc->shareable) |
459 | if (ecb_expect_false (!enc->shareable)) |
420 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
460 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
421 | |
461 | |
422 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
462 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
423 | |
463 | |
424 | if (SvOK (*svp)) |
464 | if (SvOK (*svp)) |
… | |
… | |
438 | if (ecb_expect_false (SvOBJECT (sv))) |
478 | if (ecb_expect_false (SvOBJECT (sv))) |
439 | { |
479 | { |
440 | HV *stash = SvSTASH (sv); |
480 | HV *stash = SvSTASH (sv); |
441 | GV *method; |
481 | GV *method; |
442 | |
482 | |
|
|
483 | if (enc->cbor.flags & F_FORBID_OBJECTS) |
|
|
484 | croak ("encountered object '%s', but forbid_objects is enabled", |
|
|
485 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
443 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
486 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
444 | { |
487 | { |
445 | dSP; |
488 | dSP; |
446 | |
489 | |
447 | ENTER; SAVETMPS; PUSHMARK (SP); |
490 | ENTER; SAVETMPS; |
|
|
491 | PUSHMARK (SP); |
448 | // we re-bless the reference to get overload and other niceties right |
492 | // we re-bless the reference to get overload and other niceties right |
449 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
493 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
450 | |
494 | |
451 | PUTBACK; |
495 | PUTBACK; |
452 | // G_SCALAR ensures that return value is 1 |
496 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
465 | } |
509 | } |
466 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
510 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
467 | { |
511 | { |
468 | dSP; |
512 | dSP; |
469 | |
513 | |
470 | ENTER; SAVETMPS; PUSHMARK (SP); |
514 | ENTER; SAVETMPS; |
|
|
515 | SAVESTACK_POS (); |
|
|
516 | PUSHMARK (SP); |
471 | EXTEND (SP, 2); |
517 | EXTEND (SP, 2); |
472 | // we re-bless the reference to get overload and other niceties right |
518 | // we re-bless the reference to get overload and other niceties right |
473 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
519 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
474 | PUSHs (sv_cbor); |
520 | PUSHs (sv_cbor); |
475 | |
521 | |
… | |
… | |
481 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
527 | 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)); |
528 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
483 | |
529 | |
484 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
530 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
485 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
531 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
486 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
532 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
487 | |
533 | |
488 | while (count) |
534 | while (count) |
489 | encode_sv (enc, SP[1 - count--]); |
535 | encode_sv (enc, SP[1 - count--]); |
490 | |
536 | |
491 | PUTBACK; |
537 | PUTBACK; |
… | |
… | |
517 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
563 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
518 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
564 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
519 | //TODO: maybe I32? |
565 | //TODO: maybe I32? |
520 | else if (ecb_expect_false (nv == (float)nv)) |
566 | else if (ecb_expect_false (nv == (float)nv)) |
521 | { |
567 | { |
|
|
568 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
569 | |
522 | uint32_t fp = ecb_float_to_binary32 (nv); |
570 | uint32_t fp = ecb_float_to_binary32 (nv); |
523 | |
|
|
524 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
|
|
525 | |
571 | |
526 | if (!ecb_big_endian ()) |
572 | if (!ecb_big_endian ()) |
527 | fp = ecb_bswap32 (fp); |
573 | fp = ecb_bswap32 (fp); |
528 | |
574 | |
529 | memcpy (enc->cur, &fp, 4); |
575 | memcpy (enc->cur, &fp, 4); |
530 | enc->cur += 4; |
576 | enc->cur += 4; |
531 | } |
577 | } |
532 | else |
578 | else |
533 | { |
579 | { |
|
|
580 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
581 | |
534 | uint64_t fp = ecb_double_to_binary64 (nv); |
582 | uint64_t fp = ecb_double_to_binary64 (nv); |
535 | |
|
|
536 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
|
|
537 | |
583 | |
538 | if (!ecb_big_endian ()) |
584 | if (!ecb_big_endian ()) |
539 | fp = ecb_bswap64 (fp); |
585 | fp = ecb_bswap64 (fp); |
540 | |
586 | |
541 | memcpy (enc->cur, &fp, 8); |
587 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
550 | |
596 | |
551 | if (SvPOKp (sv)) |
597 | if (SvPOKp (sv)) |
552 | { |
598 | { |
553 | STRLEN len; |
599 | STRLEN len; |
554 | char *str = SvPV (sv, len); |
600 | char *str = SvPV (sv, len); |
555 | encode_strref (enc, SvUTF8 (sv), str, len); |
601 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
556 | } |
602 | } |
557 | else if (SvNOKp (sv)) |
603 | else if (SvNOKp (sv)) |
558 | encode_nv (enc, sv); |
604 | encode_nv (enc, sv); |
559 | else if (SvIOKp (sv)) |
605 | else if (SvIOKp (sv)) |
560 | { |
606 | { |
… | |
… | |
577 | } |
623 | } |
578 | |
624 | |
579 | static SV * |
625 | static SV * |
580 | encode_cbor (SV *scalar, CBOR *cbor) |
626 | encode_cbor (SV *scalar, CBOR *cbor) |
581 | { |
627 | { |
582 | enc_t enc = { }; |
628 | enc_t enc = { 0 }; |
583 | |
629 | |
584 | enc.cbor = *cbor; |
630 | enc.cbor = *cbor; |
585 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
631 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
586 | enc.cur = SvPVX (enc.sv); |
632 | enc.cur = SvPVX (enc.sv); |
587 | enc.end = SvEND (enc.sv); |
633 | enc.end = SvEND (enc.sv); |
588 | |
634 | |
589 | SvPOK_only (enc.sv); |
635 | SvPOK_only (enc.sv); |
590 | |
636 | |
591 | if (cbor->flags & F_PACK_STRINGS) |
637 | if (cbor->flags & F_PACK_STRINGS) |
592 | { |
638 | { |
… | |
… | |
619 | U32 depth; // recursion depth |
665 | U32 depth; // recursion depth |
620 | U32 maxdepth; // recursion depth limit |
666 | U32 maxdepth; // recursion depth limit |
621 | AV *shareable; |
667 | AV *shareable; |
622 | AV *stringref; |
668 | AV *stringref; |
623 | SV *decode_tagged; |
669 | SV *decode_tagged; |
|
|
670 | SV *err_sv; // optional sv for error, needs to be freed |
624 | } dec_t; |
671 | } dec_t; |
625 | |
672 | |
626 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
673 | // set dec->err to ERRSV |
|
|
674 | ecb_cold static void |
|
|
675 | err_errsv (dec_t *dec) |
|
|
676 | { |
|
|
677 | if (!dec->err) |
|
|
678 | { |
|
|
679 | dec->err_sv = newSVsv (ERRSV); |
627 | |
680 | |
628 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
681 | // chop off the trailing \n |
|
|
682 | SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1); |
|
|
683 | *SvEND (dec->err_sv) = 0; |
629 | |
684 | |
|
|
685 | dec->err = SvPVutf8_nolen (dec->err_sv); |
|
|
686 | } |
|
|
687 | } |
|
|
688 | |
|
|
689 | // the following functions are used to reduce code size and help the compiler to optimise |
|
|
690 | ecb_cold static void |
|
|
691 | err_set (dec_t *dec, const char *reason) |
|
|
692 | { |
|
|
693 | if (!dec->err) |
|
|
694 | dec->err = reason; |
|
|
695 | } |
|
|
696 | |
|
|
697 | ecb_cold static void |
|
|
698 | err_unexpected_end (dec_t *dec) |
|
|
699 | { |
|
|
700 | err_set (dec, "unexpected end of CBOR data"); |
|
|
701 | } |
|
|
702 | |
|
|
703 | #define ERR_DO(do) SB do; goto fail; SE |
|
|
704 | #define ERR(reason) ERR_DO (err_set (dec, reason)) |
|
|
705 | #define ERR_ERRSV ERR_DO (err_errsv (dec)) |
|
|
706 | |
|
|
707 | #define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) |
|
|
708 | |
630 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
709 | #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 |
710 | #define DEC_DEC_DEPTH --dec->depth |
632 | |
711 | |
633 | static UV |
712 | static UV |
634 | decode_uint (dec_t *dec) |
713 | decode_uint (dec_t *dec) |
635 | { |
714 | { |
… | |
… | |
712 | av_push (av, decode_sv (dec)); |
791 | av_push (av, decode_sv (dec)); |
713 | } |
792 | } |
714 | } |
793 | } |
715 | else |
794 | else |
716 | { |
795 | { |
717 | int i, len = decode_uint (dec); |
796 | UV i, len = decode_uint (dec); |
718 | |
797 | |
719 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
798 | 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); |
799 | av_fill (av, len - 1); |
721 | |
800 | |
722 | for (i = 0; i < len; ++i) |
801 | for (i = 0; i < len; ++i) |
… | |
… | |
725 | |
804 | |
726 | DEC_DEC_DEPTH; |
805 | DEC_DEC_DEPTH; |
727 | return newRV_noinc ((SV *)av); |
806 | return newRV_noinc ((SV *)av); |
728 | |
807 | |
729 | fail: |
808 | fail: |
730 | SvREFCNT_dec (av); |
809 | SvREFCNT_dec_NN (av); |
731 | DEC_DEC_DEPTH; |
810 | DEC_DEC_DEPTH; |
732 | return &PL_sv_undef; |
811 | return &PL_sv_undef; |
733 | } |
812 | } |
734 | |
813 | |
735 | static void |
814 | static void |
… | |
… | |
739 | // byte or utf-8 strings as keys, but only when !stringref |
818 | // byte or utf-8 strings as keys, but only when !stringref |
740 | |
819 | |
741 | if (ecb_expect_true (!dec->stringref)) |
820 | if (ecb_expect_true (!dec->stringref)) |
742 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
821 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
743 | { |
822 | { |
744 | I32 len = decode_uint (dec); |
823 | STRLEN len = decode_uint (dec); |
745 | char *key = (char *)dec->cur; |
824 | char *key = (char *)dec->cur; |
746 | |
825 | |
|
|
826 | WANT (len); |
747 | dec->cur += len; |
827 | dec->cur += len; |
748 | |
828 | |
749 | hv_store (hv, key, len, decode_sv (dec), 0); |
829 | hv_store (hv, key, len, decode_sv (dec), 0); |
750 | |
830 | |
751 | return; |
831 | return; |
752 | } |
832 | } |
753 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
833 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
754 | { |
834 | { |
755 | I32 len = decode_uint (dec); |
835 | STRLEN len = decode_uint (dec); |
756 | char *key = (char *)dec->cur; |
836 | char *key = (char *)dec->cur; |
757 | |
837 | |
|
|
838 | WANT (len); |
758 | dec->cur += len; |
839 | dec->cur += len; |
759 | |
840 | |
760 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
841 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
761 | if (!is_utf8_string (key, len)) |
842 | if (!is_utf8_string (key, len)) |
762 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
843 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
… | |
… | |
767 | } |
848 | } |
768 | |
849 | |
769 | SV *k = decode_sv (dec); |
850 | SV *k = decode_sv (dec); |
770 | SV *v = decode_sv (dec); |
851 | SV *v = decode_sv (dec); |
771 | |
852 | |
|
|
853 | // we leak memory if uncaught exceptions are thrown by random magical |
|
|
854 | // methods, and this is hopefully the only place where it can happen, |
|
|
855 | // so if there is a chance of an exception, take the very slow path. |
|
|
856 | // since catching exceptions is "undocumented/internal/forbidden" by |
|
|
857 | // the new p5p powers, we need to call out to a perl function :/ |
|
|
858 | if (ecb_expect_false (SvAMAGIC (k))) |
|
|
859 | { |
|
|
860 | dSP; |
|
|
861 | |
|
|
862 | ENTER; SAVETMPS; |
|
|
863 | PUSHMARK (SP); |
|
|
864 | EXTEND (SP, 3); |
|
|
865 | PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); |
|
|
866 | PUSHs (sv_2mortal (k)); |
|
|
867 | PUSHs (sv_2mortal (v)); |
|
|
868 | |
|
|
869 | PUTBACK; |
|
|
870 | call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); |
|
|
871 | SPAGAIN; |
|
|
872 | |
|
|
873 | FREETMPS; LEAVE; |
|
|
874 | |
|
|
875 | if (SvTRUE (ERRSV)) |
|
|
876 | ERR_ERRSV; |
|
|
877 | |
|
|
878 | return; |
|
|
879 | } |
|
|
880 | |
772 | hv_store_ent (hv, k, v, 0); |
881 | hv_store_ent (hv, k, v, 0); |
773 | SvREFCNT_dec (k); |
882 | SvREFCNT_dec_NN (k); |
774 | |
883 | |
775 | fail: |
884 | fail: |
776 | ; |
885 | ; |
777 | } |
886 | } |
778 | |
887 | |
… | |
… | |
800 | decode_he (dec, hv); |
909 | decode_he (dec, hv); |
801 | } |
910 | } |
802 | } |
911 | } |
803 | else |
912 | else |
804 | { |
913 | { |
805 | int pairs = decode_uint (dec); |
914 | UV pairs = decode_uint (dec); |
|
|
915 | |
|
|
916 | WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes |
806 | |
917 | |
807 | while (pairs--) |
918 | while (pairs--) |
808 | decode_he (dec, hv); |
919 | decode_he (dec, hv); |
809 | } |
920 | } |
810 | |
921 | |
811 | DEC_DEC_DEPTH; |
922 | DEC_DEC_DEPTH; |
812 | return newRV_noinc ((SV *)hv); |
923 | return newRV_noinc ((SV *)hv); |
813 | |
924 | |
814 | fail: |
925 | fail: |
815 | SvREFCNT_dec (hv); |
926 | SvREFCNT_dec_NN (hv); |
816 | DEC_DEC_DEPTH; |
927 | DEC_DEC_DEPTH; |
817 | return &PL_sv_undef; |
928 | return &PL_sv_undef; |
818 | } |
929 | } |
819 | |
930 | |
820 | static SV * |
931 | static SV * |
821 | decode_str (dec_t *dec, int utf8) |
932 | decode_str (dec_t *dec, int utf8) |
822 | { |
933 | { |
823 | SV *sv = 0; |
934 | SV *sv = 0; |
824 | |
935 | |
825 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
936 | if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) |
826 | { |
937 | { |
827 | // indefinite length strings |
938 | // indefinite length strings |
828 | ++dec->cur; |
939 | ++dec->cur; |
829 | |
940 | |
830 | U8 major = *dec->cur & MAJOR_MISC; |
941 | U8 major = *dec->cur & MAJOR_MISC; |
… | |
… | |
898 | sv = newRV_noinc (decode_sv (dec)); |
1009 | sv = newRV_noinc (decode_sv (dec)); |
899 | break; |
1010 | break; |
900 | |
1011 | |
901 | case CBOR_TAG_STRINGREF_NAMESPACE: |
1012 | case CBOR_TAG_STRINGREF_NAMESPACE: |
902 | { |
1013 | { |
|
|
1014 | // do not use SAVETMPS/FREETMPS, as these will |
|
|
1015 | // erase mortalised caches, e.g. "shareable" |
903 | ENTER; SAVETMPS; |
1016 | ENTER; |
904 | |
1017 | |
905 | SAVESPTR (dec->stringref); |
1018 | SAVESPTR (dec->stringref); |
906 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
1019 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
907 | |
1020 | |
908 | sv = decode_sv (dec); |
1021 | sv = decode_sv (dec); |
909 | |
1022 | |
910 | FREETMPS; LEAVE; |
1023 | LEAVE; |
911 | } |
1024 | } |
912 | break; |
1025 | break; |
913 | |
1026 | |
914 | case CBOR_TAG_STRINGREF: |
1027 | case CBOR_TAG_STRINGREF: |
915 | { |
1028 | { |
… | |
… | |
966 | } |
1079 | } |
967 | break; |
1080 | break; |
968 | |
1081 | |
969 | case CBOR_TAG_PERL_OBJECT: |
1082 | case CBOR_TAG_PERL_OBJECT: |
970 | { |
1083 | { |
|
|
1084 | if (dec->cbor.flags & F_FORBID_OBJECTS) |
|
|
1085 | goto filter; |
|
|
1086 | |
971 | sv = decode_sv (dec); |
1087 | sv = decode_sv (dec); |
972 | |
1088 | |
973 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
1089 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
974 | ERR ("corrupted CBOR data (non-array perl object)"); |
1090 | ERR ("corrupted CBOR data (non-array perl object)"); |
975 | |
1091 | |
… | |
… | |
985 | if (!method) |
1101 | if (!method) |
986 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1102 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
987 | |
1103 | |
988 | dSP; |
1104 | dSP; |
989 | |
1105 | |
990 | ENTER; SAVETMPS; PUSHMARK (SP); |
1106 | ENTER; SAVETMPS; |
|
|
1107 | PUSHMARK (SP); |
991 | EXTEND (SP, len + 1); |
1108 | EXTEND (SP, len + 1); |
992 | // we re-bless the reference to get overload and other niceties right |
1109 | // we re-bless the reference to get overload and other niceties right |
993 | PUSHs (*av_fetch (av, 0, 1)); |
1110 | PUSHs (*av_fetch (av, 0, 1)); |
994 | PUSHs (sv_cbor); |
1111 | PUSHs (sv_cbor); |
995 | |
1112 | |
… | |
… | |
1003 | SPAGAIN; |
1120 | SPAGAIN; |
1004 | |
1121 | |
1005 | if (SvTRUE (ERRSV)) |
1122 | if (SvTRUE (ERRSV)) |
1006 | { |
1123 | { |
1007 | FREETMPS; LEAVE; |
1124 | FREETMPS; LEAVE; |
1008 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1125 | ERR_ERRSV; |
1009 | } |
1126 | } |
1010 | |
1127 | |
1011 | SvREFCNT_dec (sv); |
1128 | SvREFCNT_dec_NN (sv); |
1012 | sv = SvREFCNT_inc (POPs); |
1129 | sv = SvREFCNT_inc (POPs); |
1013 | |
1130 | |
1014 | PUTBACK; |
1131 | PUTBACK; |
1015 | |
1132 | |
1016 | FREETMPS; LEAVE; |
1133 | FREETMPS; LEAVE; |
1017 | } |
1134 | } |
1018 | break; |
1135 | break; |
1019 | |
1136 | |
1020 | default: |
1137 | default: |
|
|
1138 | filter: |
1021 | { |
1139 | { |
|
|
1140 | SV *tag_sv = newSVuv (tag); |
|
|
1141 | |
1022 | sv = decode_sv (dec); |
1142 | sv = decode_sv (dec); |
1023 | |
1143 | |
1024 | dSP; |
1144 | dSP; |
1025 | ENTER; SAVETMPS; PUSHMARK (SP); |
1145 | ENTER; SAVETMPS; |
|
|
1146 | SAVESTACK_POS (); |
|
|
1147 | PUSHMARK (SP); |
1026 | EXTEND (SP, 2); |
1148 | EXTEND (SP, 2); |
1027 | PUSHs (newSVuv (tag)); |
1149 | PUSHs (tag_sv); |
1028 | PUSHs (sv); |
1150 | PUSHs (sv); |
1029 | |
1151 | |
1030 | PUTBACK; |
1152 | PUTBACK; |
1031 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1153 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1032 | SPAGAIN; |
1154 | SPAGAIN; |
1033 | |
1155 | |
1034 | if (SvTRUE (ERRSV)) |
1156 | if (SvTRUE (ERRSV)) |
1035 | { |
1157 | { |
|
|
1158 | SvREFCNT_dec_NN (tag_sv); |
1036 | FREETMPS; LEAVE; |
1159 | FREETMPS; LEAVE; |
1037 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1160 | ERR_ERRSV; |
1038 | } |
1161 | } |
1039 | |
1162 | |
1040 | if (count) |
1163 | if (count) |
1041 | { |
1164 | { |
|
|
1165 | SvREFCNT_dec_NN (tag_sv); |
1042 | SvREFCNT_dec (sv); |
1166 | SvREFCNT_dec_NN (sv); |
1043 | sv = SvREFCNT_inc (POPs); |
1167 | sv = SvREFCNT_inc_NN (POPs); |
1044 | } |
1168 | } |
1045 | else |
1169 | else |
1046 | { |
1170 | { |
1047 | AV *av = newAV (); |
1171 | AV *av = newAV (); |
1048 | av_push (av, newSVuv (tag)); |
1172 | av_push (av, tag_sv); |
1049 | av_push (av, sv); |
1173 | av_push (av, sv); |
1050 | |
1174 | |
1051 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1175 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1052 | ? cbor_tagged_stash |
1176 | ? cbor_tagged_stash |
1053 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1177 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
… | |
… | |
1156 | } |
1280 | } |
1157 | |
1281 | |
1158 | static SV * |
1282 | static SV * |
1159 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1283 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1160 | { |
1284 | { |
1161 | dec_t dec = { }; |
1285 | dec_t dec = { 0 }; |
1162 | SV *sv; |
1286 | SV *sv; |
1163 | STRLEN len; |
1287 | STRLEN len; |
1164 | char *data = SvPVbyte (string, len); |
1288 | char *data = SvPVbyte (string, len); |
1165 | |
1289 | |
1166 | if (len > cbor->max_size && cbor->max_size) |
1290 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
1182 | |
1306 | |
1183 | if (dec.err) |
1307 | if (dec.err) |
1184 | { |
1308 | { |
1185 | if (dec.shareable) |
1309 | if (dec.shareable) |
1186 | { |
1310 | { |
1187 | // need to break cyclic links, which whould all be in shareable |
1311 | // need to break cyclic links, which would all be in shareable |
1188 | int i; |
1312 | int i; |
1189 | SV **svp; |
1313 | SV **svp; |
1190 | |
1314 | |
1191 | for (i = av_len (dec.shareable) + 1; i--; ) |
1315 | for (i = av_len (dec.shareable) + 1; i--; ) |
1192 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1316 | if ((svp = av_fetch (dec.shareable, i, 0))) |
1193 | sv_setsv (*svp, &PL_sv_undef); |
1317 | sv_setsv (*svp, &PL_sv_undef); |
1194 | } |
1318 | } |
1195 | |
1319 | |
1196 | SvREFCNT_dec (sv); |
1320 | SvREFCNT_dec_NN (sv); |
|
|
1321 | |
|
|
1322 | if (dec.err_sv) |
|
|
1323 | sv_2mortal (dec.err_sv); |
|
|
1324 | |
1197 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1325 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1198 | } |
1326 | } |
1199 | |
1327 | |
1200 | sv = sv_2mortal (sv); |
1328 | sv = sv_2mortal (sv); |
1201 | |
1329 | |
… | |
… | |
1257 | |
1385 | |
1258 | int major = *p >> MAJOR_SHIFT; |
1386 | int major = *p >> MAJOR_SHIFT; |
1259 | |
1387 | |
1260 | switch (major) |
1388 | switch (major) |
1261 | { |
1389 | { |
|
|
1390 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1391 | ++count; // tags merely prefix another value |
|
|
1392 | break; |
|
|
1393 | |
1262 | case MAJOR_BYTES >> MAJOR_SHIFT: |
1394 | case MAJOR_BYTES >> MAJOR_SHIFT: |
1263 | case MAJOR_TEXT >> MAJOR_SHIFT: |
1395 | case MAJOR_TEXT >> MAJOR_SHIFT: |
1264 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1396 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
1265 | case MAJOR_MAP >> MAJOR_SHIFT: |
1397 | case MAJOR_MAP >> MAJOR_SHIFT: |
1266 | { |
1398 | { |
… | |
… | |
1341 | |
1473 | |
1342 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1474 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1343 | |
1475 | |
1344 | sv_cbor = newSVpv ("CBOR", 0); |
1476 | sv_cbor = newSVpv ("CBOR", 0); |
1345 | SvREADONLY_on (sv_cbor); |
1477 | SvREADONLY_on (sv_cbor); |
|
|
1478 | |
|
|
1479 | assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); |
1346 | } |
1480 | } |
1347 | |
1481 | |
1348 | PROTOTYPES: DISABLE |
1482 | PROTOTYPES: DISABLE |
1349 | |
1483 | |
1350 | void CLONE (...) |
1484 | void CLONE (...) |
… | |
… | |
1370 | ALIAS: |
1504 | ALIAS: |
1371 | shrink = F_SHRINK |
1505 | shrink = F_SHRINK |
1372 | allow_unknown = F_ALLOW_UNKNOWN |
1506 | allow_unknown = F_ALLOW_UNKNOWN |
1373 | allow_sharing = F_ALLOW_SHARING |
1507 | allow_sharing = F_ALLOW_SHARING |
1374 | allow_cycles = F_ALLOW_CYCLES |
1508 | allow_cycles = F_ALLOW_CYCLES |
|
|
1509 | forbid_objects = F_FORBID_OBJECTS |
1375 | pack_strings = F_PACK_STRINGS |
1510 | pack_strings = F_PACK_STRINGS |
|
|
1511 | text_keys = F_TEXT_KEYS |
|
|
1512 | text_strings = F_TEXT_STRINGS |
1376 | validate_utf8 = F_VALIDATE_UTF8 |
1513 | validate_utf8 = F_VALIDATE_UTF8 |
1377 | PPCODE: |
1514 | PPCODE: |
1378 | { |
1515 | { |
1379 | if (enable) |
1516 | if (enable) |
1380 | self->flags |= ix; |
1517 | self->flags |= ix; |
… | |
… | |
1388 | ALIAS: |
1525 | ALIAS: |
1389 | get_shrink = F_SHRINK |
1526 | get_shrink = F_SHRINK |
1390 | get_allow_unknown = F_ALLOW_UNKNOWN |
1527 | get_allow_unknown = F_ALLOW_UNKNOWN |
1391 | get_allow_sharing = F_ALLOW_SHARING |
1528 | get_allow_sharing = F_ALLOW_SHARING |
1392 | get_allow_cycles = F_ALLOW_CYCLES |
1529 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1530 | get_forbid_objects = F_FORBID_OBJECTS |
1393 | get_pack_strings = F_PACK_STRINGS |
1531 | get_pack_strings = F_PACK_STRINGS |
|
|
1532 | get_text_keys = F_TEXT_KEYS |
|
|
1533 | get_text_strings = F_TEXT_STRINGS |
1394 | get_validate_utf8 = F_VALIDATE_UTF8 |
1534 | get_validate_utf8 = F_VALIDATE_UTF8 |
1395 | PPCODE: |
1535 | PPCODE: |
1396 | XPUSHs (boolSV (self->flags & ix)); |
1536 | XPUSHs (boolSV (self->flags & ix)); |
1397 | |
1537 | |
1398 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1538 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |