[ViewVC] Diff of: cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.62 by root, Sat Nov 26 02:11:52 2016 UTC vs.
Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC

…		…
21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22	#endif	22	#endif
23	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
24	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
25	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
26	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
27	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
28	#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)))
29		35
30	// known major and minor types	36	// known major and minor types
31	enum cbor_type	37	enum cbor_type
32	{	38	{
33	MAJOR_SHIFT = 5,	39	MAJOR_SHIFT = 5,
…		…
96	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	102	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
97		103
98	CBOR_TAG_MAGIC = 55799, // self-describe cbor	104	CBOR_TAG_MAGIC = 55799, // self-describe cbor
99	};	105	};
100		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
101	#define F_SHRINK 0x00000001UL	121	#define F_SHRINK 0x00000001UL
102	#define F_ALLOW_UNKNOWN 0x00000002UL	122	#define F_ALLOW_UNKNOWN 0x00000002UL
103	#define F_ALLOW_SHARING 0x00000004UL	123	#define F_ALLOW_SHARING 0x00000004UL
104	#define F_ALLOW_CYCLES 0x00000008UL	124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_ALLOW_WEAK_CYCLES 0x00000010UL
105	#define F_FORBID_OBJECTS 0x00000010UL	126	#define F_FORBID_OBJECTS 0x00000020UL
106	#define F_PACK_STRINGS 0x00000020UL	127	#define F_PACK_STRINGS 0x00000040UL
107	#define F_TEXT_KEYS 0x00000040UL	128	#define F_TEXT_KEYS 0x00000080UL
108	#define F_TEXT_STRINGS 0x00000080UL	129	#define F_TEXT_STRINGS 0x00000100UL
109	#define F_VALIDATE_UTF8 0x00000100UL	130	#define F_VALIDATE_UTF8 0x00000200UL
110		131
111	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
112		133
113	#define SB do {	134	#define SB do {
114	#define SE } while (0)	135	#define SE } while (0)
…		…
184	#endif	205	#endif
185	}	206	}
186	}	207	}
187		208
188	// minimum length of a string to be registered for stringref	209	// minimum length of a string to be registered for stringref
189	ecb_inline int	210	ecb_inline STRLEN
190	minimum_string_length (UV idx)	211	minimum_string_length (UV idx)
191	{	212	{
192	return idx <= 23 ? 3	213	return idx <= 23 ? 3
193	: idx <= 0xffU ? 4	214	: idx <= 0xffU ? 4
194	: idx <= 0xffffU ? 5	215	: idx <= 0xffffU ? 5
…		…
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	}
…		…
340	}	376	}
341		377
342	encode_str (enc, upgrade_utf8, utf8, str, len);	378	encode_str (enc, upgrade_utf8, utf8, str, len);
343	}	379	}
344		380
345	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	}
346		511
347	static void	512	static void
348	encode_av (enc_t enc, AV av)	513	encode_av (enc_t enc, AV av)
349	{	514	{
350	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
354		519
355	++enc->depth;	520	++enc->depth;
356		521
357	encode_uint (enc, MAJOR_ARRAY, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
358		523
359	if (SvMAGICAL (av))	524	if (ecb_expect_false (SvMAGICAL (av)))
360	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
361	{	526	{
362	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
363	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
364	}	529	}
…		…
383	++enc->depth;	548	++enc->depth;
384		549
385	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
386	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
387		552
388	if (mg)	553	if (ecb_expect_false (mg))
389	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);	554	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
390	else	555	else
391	encode_uint (enc, MAJOR_MAP, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
392		557
393	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
…		…
398	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS \| F_TEXT_STRINGS), 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));
399		564
400	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));
401	}	566	}
402		567
403	if (mg)	568	if (ecb_expect_false (mg))
404	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);	569	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
405		570
406	--enc->depth;	571	--enc->depth;
407	}	572	}
408		573
…		…
428		593
429	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
430		595
431	if (stash == boolean_stash)	596	if (stash == boolean_stash)
432	{	597	{
433	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);	598	encode_bool (enc, SvIV (sv));
434	return;	599	return;
435	}	600	}
436	else if (stash == error_stash)	601	else if (stash == error_stash)
437	{	602	{
438	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);	603	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
…		…
441	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
442	{	607	{
443	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
444	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
445		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
446	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)));
447	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	}
448		627
449	return;	628	return;
450	}	629	}
451	}	630	}
452		631
453	if (ecb_expect_false (SvREFCNT (sv) > 1)	632	if (ecb_expect_false (SvREFCNT (sv) > 1)
454	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))	633	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
455	{	634	{
456	if (!enc->shareable)	635	if (ecb_expect_false (!enc->shareable))
457	enc->shareable = (HV )sv_2mortal ((SV )newHV ());	636	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
458		637
459	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);	638	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
460		639
461	if (SvOK (*svp))	640	if (SvOK (*svp))
…		…
507	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
508	{	687	{
509	dSP;	688	dSP;
510		689
511	ENTER; SAVETMPS;	690	ENTER; SAVETMPS;
512	SAVESTACK_POS ();
513	PUSHMARK (SP);	691	PUSHMARK (SP);
514	EXTEND (SP, 2);	692	EXTEND (SP, 2);
515	// 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
516	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
517	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
…		…
526		704
527	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
528	encode_uint (enc, MAJOR_ARRAY, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
529	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
530		708
531	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
532	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
533		717
534	PUTBACK;	718	PUTBACK;
535		719
536	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
537	}	721	}
…		…
559		743
560	if (ecb_expect_false (nv == (NV)(U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
561	encode_uint (enc, MAJOR_POS_INT, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
562	//TODO: maybe I32?	746	//TODO: maybe I32?
563	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
564	{	748	encode_float32 (enc, nv);
565	uint32_t fp = ecb_float_to_binary32 (nv);
566
567	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
568
569	if (!ecb_big_endian ())
570	fp = ecb_bswap32 (fp);
571
572	memcpy (enc->cur, &fp, 4);
573	enc->cur += 4;
574	}
575	else	749	else
576	{	750	encode_float64 (enc, nv);
577	uint64_t fp = ecb_double_to_binary64 (nv);
578
579	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
580
581	if (!ecb_big_endian ())
582	fp = ecb_bswap64 (fp);
583
584	memcpy (enc->cur, &fp, 8);
585	enc->cur += 8;
586	}
587	}	751	}
588		752
589	static void	753	static void
590	encode_sv (enc_t enc, SV sv)	754	encode_sv (enc_t enc, SV sv)
591	{	755	{
…		…
598	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
599	}	763	}
600	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
601	encode_nv (enc, sv);	765	encode_nv (enc, sv);
602	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
603	{	767	encode_int (enc, sv);
604	if (SvIsUV (sv))
605	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
606	else if (SvIVX (sv) >= 0)
607	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
608	else
609	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
610	}
611	else if (SvROK (sv))	768	else if (SvROK (sv))
612	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
613	else if (!SvOK (sv))	770	else if (!SvOK (sv))
614	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);	771	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
615	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
…		…
622	static SV *	779	static SV *
623	encode_cbor (SV scalar, CBOR cbor)	780	encode_cbor (SV scalar, CBOR cbor)
624	{	781	{
625	enc_t enc = { 0 };	782	enc_t enc = { 0 };
626		783
627	enc.cbor = *cbor;	784	enc.cbor = *cbor;
628	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
629	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
630	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
631		788
632	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
633		790
634	if (cbor->flags & F_PACK_STRINGS)	791	if (cbor->flags & F_PACK_STRINGS)
635	{	792	{
…		…
693		850
694	ecb_cold static void	851	ecb_cold static void
695	err_unexpected_end (dec_t *dec)	852	err_unexpected_end (dec_t *dec)
696	{	853	{
697	err_set (dec, "unexpected end of CBOR data");	854	err_set (dec, "unexpected end of CBOR data");
698	}
699
700	ecb_cold static void
701	err_nesting_exceeded (dec_t *dec)
702	{
703	err_set (dec, ERR_NESTING_EXCEEDED);
704	}	855	}
705		856
706	#define ERR_DO(do) SB do; goto fail; SE	857	#define ERR_DO(do) SB do; goto fail; SE
707	#define ERR(reason) ERR_DO (err_set (dec, reason))	858	#define ERR(reason) ERR_DO (err_set (dec, reason))
708	#define ERR_ERRSV ERR_DO (err_errsv (dec))	859	#define ERR_ERRSV ERR_DO (err_errsv (dec))
…		…
783		934
784	for (;;)	935	for (;;)
785	{	936	{
786	WANT (1);	937	WANT (1);
787		938
788	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	939	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
789	{	940	{
790	++dec->cur;	941	++dec->cur;
791	break;	942	break;
792	}	943	}
793		944
…		…
807		958
808	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
809	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
810		961
811	fail:	962	fail:
812	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
813	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
814	return &PL_sv_undef;	965	return &PL_sv_undef;
815	}	966	}
816		967
817	static void	968	static void
…		…
840		991
841	WANT (len);	992	WANT (len);
842	dec->cur += len;	993	dec->cur += len;
843		994
844	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))	995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
845	if (!is_utf8_string (key, len))	996	if (!cbor_is_utf8_string ((U8 *)key, len))
846	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");	997	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
847		998
848	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
849		1000
850	return;	1001	return;
…		…
880		1031
881	return;	1032	return;
882	}	1033	}
883		1034
884	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
885	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
886		1037
887	fail:	1038	fail:
888	;	1039	;
889	}	1040	}
890		1041
…		…
901		1052
902	for (;;)	1053	for (;;)
903	{	1054	{
904	WANT (1);	1055	WANT (1);
905		1056
906	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	1057	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
907	{	1058	{
908	++dec->cur;	1059	++dec->cur;
909	break;	1060	break;
910	}	1061	}
911		1062
…		…
924		1075
925	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
926	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
927		1078
928	fail:	1079	fail:
929	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
930	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
931	return &PL_sv_undef;	1082	return &PL_sv_undef;
932	}	1083	}
933		1084
934	static SV *	1085	static SV *
935	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
936	{	1087	{
937	SV *sv = 0;	1088	SV *sv = 0;
938		1089
939	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
940	{	1091	{
941	// indefinite length strings	1092	// indefinite length strings
942	++dec->cur;	1093	++dec->cur;
943		1094
944	U8 major = *dec->cur & MAJOR_MISC;	1095	U8 major = *dec->cur & MAJOR_MISC;
…		…
979	}	1130	}
980		1131
981	if (utf8)	1132	if (utf8)
982	{	1133	{
983	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))	1134	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
984	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))	1135	if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
985	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");	1136	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
986		1137
987	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
988	}	1139	}
989		1140
…		…
1012	sv = newRV_noinc (decode_sv (dec));	1163	sv = newRV_noinc (decode_sv (dec));
1013	break;	1164	break;
1014		1165
1015	case CBOR_TAG_STRINGREF_NAMESPACE:	1166	case CBOR_TAG_STRINGREF_NAMESPACE:
1016	{	1167	{
1017	// do nmot use SAVETMPS/FREETMPS, as these will	1168	// do not use SAVETMPS/FREETMPS, as these will
1018	// erase mortalised caches, e.g. "shareable"	1169	// erase mortalised caches, e.g. "shareable"
1019	ENTER;	1170	ENTER;
1020		1171
1021	SAVESPTR (dec->stringref);	1172	SAVESPTR (dec->stringref);
1022	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1173	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
…		…
1032	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1033	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1034		1185
1035	UV idx = decode_uint (dec);	1186	UV idx = decode_uint (dec);
1036		1187
1037	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	1188	if (!dec->stringref \|\| idx >= (UV)(1 + AvFILLp (dec->stringref)))
1038	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)");
1039		1190
1040	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1041	}	1192	}
1042	break;	1193	break;
…		…
1044	case CBOR_TAG_VALUE_SHAREABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
1045	{	1196	{
1046	if (ecb_expect_false (!dec->shareable))	1197	if (ecb_expect_false (!dec->shareable))
1047	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1198	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
1048		1199
1049	if (dec->cbor.flags & F_ALLOW_CYCLES)	1200	if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES \| F_ALLOW_WEAK_CYCLES)))
1050	{	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;
1051	sv = newSV (0);	1206	AV *av = newAV ();
1052	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
1053		1209
1054	SV *osv = decode_sv (dec);	1210	sv = decode_sv (dec);
1055	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
1056	SvREFCNT_dec_NN (osv);	1223	SvREFCNT_dec_NN ((SV *)av);
		1224	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1057	}	1225	}
1058	else	1226	else
1059	{	1227	{
1060	av_push (dec->shareable, &PL_sv_undef);	1228	av_push (dec->shareable, &PL_sv_undef);
1061	int idx = AvFILLp (dec->shareable);	1229	int idx = AvFILLp (dec->shareable);
1062	sv = decode_sv (dec);	1230	sv = decode_sv (dec);
1063	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));	1231	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1064	}	1232	}
1065	}	1233	}
1066	break;	1234	break;
1067		1235
1068	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
…		…
1070	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1071	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1072		1240
1073	UV idx = decode_uint (dec);	1241	UV idx = decode_uint (dec);
1074		1242
1075	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1243	if (!dec->shareable \|\| idx >= (UV)(1 + AvFILLp (dec->shareable)))
1076	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
1077		1245
1078	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1246	sv = AvARRAY (dec->shareable)[idx];
1079		1247
1080	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
1081	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);
1082	}	1260	}
1083	break;	1261	break;
1084		1262
1085	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
1086	{	1264	{
…		…
1126	{	1304	{
1127	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
1128	ERR_ERRSV;	1306	ERR_ERRSV;
1129	}	1307	}
1130		1308
1131	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
1132	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
1133		1311
1134	PUTBACK;	1312	PUTBACK;
1135		1313
1136	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
…		…
1144		1322
1145	sv = decode_sv (dec);	1323	sv = decode_sv (dec);
1146		1324
1147	dSP;	1325	dSP;
1148	ENTER; SAVETMPS;	1326	ENTER; SAVETMPS;
1149	SAVESTACK_POS ();
1150	PUSHMARK (SP);	1327	PUSHMARK (SP);
1151	EXTEND (SP, 2);	1328	EXTEND (SP, 2);
1152	PUSHs (tag_sv);	1329	PUSHs (tag_sv);
1153	PUSHs (sv);	1330	PUSHs (sv);
1154		1331
…		…
1156	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);
1157	SPAGAIN;	1334	SPAGAIN;
1158		1335
1159	if (SvTRUE (ERRSV))	1336	if (SvTRUE (ERRSV))
1160	{	1337	{
1161	SvREFCNT_dec (tag_sv);	1338	SvREFCNT_dec_NN (tag_sv);
1162	FREETMPS; LEAVE;	1339	FREETMPS; LEAVE;
1163	ERR_ERRSV;	1340	ERR_ERRSV;
1164	}	1341	}
1165		1342
1166	if (count)	1343	if (count)
1167	{	1344	{
1168	SvREFCNT_dec (tag_sv);	1345	SvREFCNT_dec_NN (tag_sv);
1169	SvREFCNT_dec (sv);	1346	SvREFCNT_dec_NN (sv);
1170	sv = SvREFCNT_inc (POPs);	1347	sv = SvREFCNT_inc_NN (TOPs);
		1348	SP -= count;
1171	}	1349	}
1172	else	1350	else
1173	{	1351	{
1174	AV *av = newAV ();	1352	AV *av = newAV ();
1175	av_push (av, tag_sv);	1353	av_push (av, tag_sv);
…		…
1318	for (i = av_len (dec.shareable) + 1; i--; )	1496	for (i = av_len (dec.shareable) + 1; i--; )
1319	if ((svp = av_fetch (dec.shareable, i, 0)))	1497	if ((svp = av_fetch (dec.shareable, i, 0)))
1320	sv_setsv (*svp, &PL_sv_undef);	1498	sv_setsv (*svp, &PL_sv_undef);
1321	}	1499	}
1322		1500
1323	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
1324		1502
1325	if (dec.err_sv)	1503	if (dec.err_sv)
1326	sv_2mortal (dec.err_sv);	1504	sv_2mortal (dec.err_sv);
1327		1505
1328	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);
1329	}	1507	}
1330		1508
1331	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
1332		1510
1333	return sv;	1511	return sv;
…		…
1422		1600
1423	break;	1601	break;
1424		1602
1425	case MAJOR_MAP >> MAJOR_SHIFT:	1603	case MAJOR_MAP >> MAJOR_SHIFT:
1426	len <<= 1;	1604	len <<= 1;
		1605	/* FALLTHROUGH */
1427	case MAJOR_ARRAY >> MAJOR_SHIFT:	1606	case MAJOR_ARRAY >> MAJOR_SHIFT:
1428	if (len)	1607	if (len)
1429	{	1608	{
1430	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest	1609	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1431	count = len + 1;	1610	count = len + 1;
…		…
1503	)));	1682	)));
1504	}	1683	}
1505		1684
1506	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1507	ALIAS:	1686	ALIAS:
1508	shrink = F_SHRINK	1687	shrink = F_SHRINK
1509	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1510	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
1511	allow_cycles = F_ALLOW_CYCLES	1690	allow_cycles = F_ALLOW_CYCLES
		1691	allow_weak_cycles = F_ALLOW_WEAK_CYCLES
1512	forbid_objects = F_FORBID_OBJECTS	1692	forbid_objects = F_FORBID_OBJECTS
1513	pack_strings = F_PACK_STRINGS	1693	pack_strings = F_PACK_STRINGS
1514	text_keys = F_TEXT_KEYS	1694	text_keys = F_TEXT_KEYS
1515	text_strings = F_TEXT_STRINGS	1695	text_strings = F_TEXT_STRINGS
1516	validate_utf8 = F_VALIDATE_UTF8	1696	validate_utf8 = F_VALIDATE_UTF8
1517	PPCODE:	1697	PPCODE:
1518	{	1698	{
1519	if (enable)	1699	if (enable)
1520	self->flags \|= ix;	1700	self->flags \|= ix;
1521	else	1701	else
…		…
1524	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1525	}	1705	}
1526		1706
1527	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1528	ALIAS:	1708	ALIAS:
1529	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1530	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1531	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
1532	get_allow_cycles = F_ALLOW_CYCLES	1712	get_allow_cycles = F_ALLOW_CYCLES
		1713	get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
1533	get_forbid_objects = F_FORBID_OBJECTS	1714	get_forbid_objects = F_FORBID_OBJECTS
1534	get_pack_strings = F_PACK_STRINGS	1715	get_pack_strings = F_PACK_STRINGS
1535	get_text_keys = F_TEXT_KEYS	1716	get_text_keys = F_TEXT_KEYS
1536	get_text_strings = F_TEXT_STRINGS	1717	get_text_strings = F_TEXT_STRINGS
1537	get_validate_utf8 = F_VALIDATE_UTF8	1718	get_validate_utf8 = F_VALIDATE_UTF8
1538	PPCODE:	1719	PPCODE:
1539	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1540		1721
1541	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1542	PPCODE:	1723	PPCODE:
…		…
1671	cbor_init (&cbor);	1852	cbor_init (&cbor);
1672	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1673	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1674	}	1855	}
1675		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

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.62 by root, Sat Nov 26 02:11:52 2016 UTC vs. Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.62 by root, Sat Nov 26 02:11:52 2016 UTC vs.
Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC