ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.44 by root, Tue Feb 18 22:12:12 2014 UTC vs.
Revision 1.53 by root, Sun Apr 24 19:51:41 2016 UTC

… …
100#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL 104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_UTF8_KEYS 0x00000040UL
105#define F_VALIDATE_UTF8 0x00000020UL 107#define F_VALIDATE_UTF8 0x00000080UL
106 108
107#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
108 110
109#define SB do { 111#define SB do {
110#define SE } while (0) 112#define SE } while (0)
… …
276encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
277{ 279{
278 encode_uint (enc, MAJOR_TAG, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
279} 281}
280 282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
286{
287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
281ecb_inline void 306ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len); 317 need (enc, len);
286 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
287 enc->cur += len; 319 enc->cur += len;
288} 320}
289 321
290static void 322ecb_inline
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
292{ 324{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 { 326 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296 328
… …
307 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx; 340 ++enc->stringref_idx;
309 } 341 }
310 } 342 }
311 343
312 encode_str (enc, utf8, str, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
313} 345}
314 346
315static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
316 348
317static void 349static void
… …
324 356
325 ++enc->depth; 357 ++enc->depth;
326 358
327 encode_uint (enc, MAJOR_ARRAY, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
328 360
361 if (SvMAGICAL (av))
329 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
330 { 363 {
331 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
332 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
333 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
334 373
335 --enc->depth; 374 --enc->depth;
336} 375}
337 376
338static void 377static void
… …
356 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
357 { 396 {
358 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
359 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
360 else 399 else
361 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
362 401
363 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
364 } 403 }
365 404
366 if (mg) 405 if (mg)
… …
442 481
443 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
444 { 483 {
445 dSP; 484 dSP;
446 485
447 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
448 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
449 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
450 490
451 PUTBACK; 491 PUTBACK;
452 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
… …
465 } 505 }
466 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
467 { 507 {
468 dSP; 508 dSP;
469 509
470 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
471 EXTEND (SP, 2); 513 EXTEND (SP, 2);
472 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
473 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
474 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
475 517
… …
481 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 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)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
483 525
484 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
485 encode_uint (enc, MAJOR_ARRAY, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
486 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
487 529
488 while (count) 530 while (count)
489 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
490 532
491 PUTBACK; 533 PUTBACK;
… …
550 592
551 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
552 { 594 {
553 STRLEN len; 595 STRLEN len;
554 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
555 encode_strref (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len);
556 } 598 }
557 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
558 encode_nv (enc, sv); 600 encode_nv (enc, sv);
559 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
560 { 602 {
… …
577} 619}
578 620
579static SV * 621static SV *
580encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
581{ 623{
582 enc_t enc = { }; 624 enc_t enc = { 0 };
583 625
584 enc.cbor = *cbor; 626 enc.cbor = *cbor;
585 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
586 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
587 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
… …
742 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
743 { 785 {
744 I32 len = decode_uint (dec); 786 I32 len = decode_uint (dec);
745 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
746 788
789 WANT (len);
747 dec->cur += len; 790 dec->cur += len;
748 791
749 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
750 793
751 return; 794 return;
752 } 795 }
753 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
754 { 797 {
755 I32 len = decode_uint (dec); 798 I32 len = decode_uint (dec);
756 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
757 800
801 WANT (len);
758 dec->cur += len; 802 dec->cur += len;
759 803
760 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
761 if (!is_utf8_string (key, len)) 805 if (!is_utf8_string (key, len))
762 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
985 if (!method) 1029 if (!method)
986 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1030 ERR ("cannot decode perl-object (package does not have a THAW method)");
987 1031
988 dSP; 1032 dSP;
989 1033
990 ENTER; SAVETMPS; PUSHMARK (SP); 1034 ENTER; SAVETMPS;
1035 PUSHMARK (SP);
991 EXTEND (SP, len + 1); 1036 EXTEND (SP, len + 1);
992 // we re-bless the reference to get overload and other niceties right 1037 // we re-bless the reference to get overload and other niceties right
993 PUSHs (*av_fetch (av, 0, 1)); 1038 PUSHs (*av_fetch (av, 0, 1));
994 PUSHs (sv_cbor); 1039 PUSHs (sv_cbor);
995 1040
… …
1020 default: 1065 default:
1021 { 1066 {
1022 sv = decode_sv (dec); 1067 sv = decode_sv (dec);
1023 1068
1024 dSP; 1069 dSP;
1025 ENTER; SAVETMPS; PUSHMARK (SP); 1070 ENTER; SAVETMPS;
1071 SAVESTACK_POS ();
1072 PUSHMARK (SP);
1026 EXTEND (SP, 2); 1073 EXTEND (SP, 2);
1027 PUSHs (newSVuv (tag)); 1074 PUSHs (newSVuv (tag));
1028 PUSHs (sv); 1075 PUSHs (sv);
1029 1076
1030 PUTBACK; 1077 PUTBACK;
… …
1156} 1203}
1157 1204
1158static SV * 1205static SV *
1159decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1206decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1160{ 1207{
1161 dec_t dec = { }; 1208 dec_t dec = { 0 };
1162 SV *sv; 1209 SV *sv;
1163 STRLEN len; 1210 STRLEN len;
1164 char *data = SvPVbyte (string, len); 1211 char *data = SvPVbyte (string, len);
1165 1212
1166 if (len > cbor->max_size && cbor->max_size) 1213 if (len > cbor->max_size && cbor->max_size)
… …
1257 1304
1258 int major = *p >> MAJOR_SHIFT; 1305 int major = *p >> MAJOR_SHIFT;
1259 1306
1260 switch (major) 1307 switch (major)
1261 { 1308 {
1309 case MAJOR_TAG >> MAJOR_SHIFT:
1310 ++count; // tags merely prefix another value
1311 break;
1312
1262 case MAJOR_BYTES >> MAJOR_SHIFT: 1313 case MAJOR_BYTES >> MAJOR_SHIFT:
1263 case MAJOR_TEXT >> MAJOR_SHIFT: 1314 case MAJOR_TEXT >> MAJOR_SHIFT:
1264 case MAJOR_ARRAY >> MAJOR_SHIFT: 1315 case MAJOR_ARRAY >> MAJOR_SHIFT:
1265 case MAJOR_MAP >> MAJOR_SHIFT: 1316 case MAJOR_MAP >> MAJOR_SHIFT:
1266 { 1317 {
… …
1371 shrink = F_SHRINK 1422 shrink = F_SHRINK
1372 allow_unknown = F_ALLOW_UNKNOWN 1423 allow_unknown = F_ALLOW_UNKNOWN
1373 allow_sharing = F_ALLOW_SHARING 1424 allow_sharing = F_ALLOW_SHARING
1374 allow_cycles = F_ALLOW_CYCLES 1425 allow_cycles = F_ALLOW_CYCLES
1375 pack_strings = F_PACK_STRINGS 1426 pack_strings = F_PACK_STRINGS
1427 utf8_strings = F_UTF8_STRINGS
1376 validate_utf8 = F_VALIDATE_UTF8 1428 validate_utf8 = F_VALIDATE_UTF8
1377 PPCODE: 1429 PPCODE:
1378{ 1430{
1379 if (enable) 1431 if (enable)
1380 self->flags |= ix; 1432 self->flags |= ix;

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines