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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC

…		…
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		11
		12	#define ECB_NO_THREADS 1
12	#include "ecb.h"	13	#include "ecb.h"
13		14
14	// compatibility with perl <5.18	15	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	16	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	17	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
97	};	98	};
98		99
99	#define F_SHRINK 0x00000001UL	100	#define F_SHRINK 0x00000001UL
100	#define F_ALLOW_UNKNOWN 0x00000002UL	101	#define F_ALLOW_UNKNOWN 0x00000002UL
101	#define F_ALLOW_SHARING 0x00000004UL	102	#define F_ALLOW_SHARING 0x00000004UL
		103	#define F_ALLOW_CYCLES 0x00000008UL
		104	#define F_FORBID_OBJECTS 0x00000010UL
102	#define F_PACK_STRINGS 0x00000008UL	105	#define F_PACK_STRINGS 0x00000020UL
		106	#define F_TEXT_KEYS 0x00000040UL
		107	#define F_TEXT_STRINGS 0x00000080UL
		108	#define F_VALIDATE_UTF8 0x00000100UL
103		109
104	#define INIT_SIZE 32 // initial scalar size to be allocated	110	#define INIT_SIZE 32 // initial scalar size to be allocated
105		111
106	#define SB do {	112	#define SB do {
107	#define SE } while (0)	113	#define SE } while (0)
…		…
126	typedef struct {	132	typedef struct {
127	U32 flags;	133	U32 flags;
128	U32 max_depth;	134	U32 max_depth;
129	STRLEN max_size;	135	STRLEN max_size;
130	SV *filter;	136	SV *filter;
		137
		138	// for the incremental parser
		139	STRLEN incr_pos; // the current offset into the text
		140	STRLEN incr_need; // minimum bytes needed to decode
		141	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131	} CBOR;	142	} CBOR;
132		143
133	ecb_inline void	144	ecb_inline void
134	cbor_init (CBOR *cbor)	145	cbor_init (CBOR *cbor)
135	{	146	{
…		…
139		150
140	ecb_inline void	151	ecb_inline void
141	cbor_free (CBOR *cbor)	152	cbor_free (CBOR *cbor)
142	{	153	{
143	SvREFCNT_dec (cbor->filter);	154	SvREFCNT_dec (cbor->filter);
		155	SvREFCNT_dec (cbor->incr_count);
144	}	156	}
145		157
146	/////////////////////////////////////////////////////////////////////////////	158	/////////////////////////////////////////////////////////////////////////////
147	// utility functions	159	// utility functions
148		160
…		…
228	{	240	{
229	need (enc, 9);	241	need (enc, 9);
230		242
231	if (ecb_expect_true (len < LENGTH_EXT1))	243	if (ecb_expect_true (len < LENGTH_EXT1))
232	*enc->cur++ = major \| len;	244	*enc->cur++ = major \| len;
233	else if (ecb_expect_true (len <= 0xff))	245	else if (ecb_expect_true (len <= 0xffU))
234	{	246	{
235	*enc->cur++ = major \| LENGTH_EXT1;	247	*enc->cur++ = major \| LENGTH_EXT1;
236	*enc->cur++ = len;	248	*enc->cur++ = len;
237	}	249	}
238	else if (len <= 0xffff)	250	else if (len <= 0xffffU)
239	{	251	{
240	*enc->cur++ = major \| LENGTH_EXT2;	252	*enc->cur++ = major \| LENGTH_EXT2;
241	*enc->cur++ = len >> 8;	253	*enc->cur++ = len >> 8;
242	*enc->cur++ = len;	254	*enc->cur++ = len;
243	}	255	}
244	else if (len <= 0xffffffff)	256	else if (len <= 0xffffffffU)
245	{	257	{
246	*enc->cur++ = major \| LENGTH_EXT4;	258	*enc->cur++ = major \| LENGTH_EXT4;
247	*enc->cur++ = len >> 24;	259	*enc->cur++ = len >> 24;
248	*enc->cur++ = len >> 16;	260	*enc->cur++ = len >> 16;
249	*enc->cur++ = len >> 8;	261	*enc->cur++ = len >> 8;
…		…
267	encode_tag (enc_t *enc, UV tag)	279	encode_tag (enc_t *enc, UV tag)
268	{	280	{
269	encode_uint (enc, MAJOR_TAG, tag);	281	encode_uint (enc, MAJOR_TAG, tag);
270	}	282	}
271		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
272	ecb_inline void	307	ecb_inline void
273	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)
274	{	309	{
		310	if (ecb_expect_false (upgrade_utf8))
		311	if (!utf8)
		312	{
		313	encode_str_utf8 (enc, utf8, str, len);
		314	return;
		315	}
		316
275	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);	317	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276	need (enc, len);	318	need (enc, len);
277	memcpy (enc->cur, str, len);	319	memcpy (enc->cur, str, len);
278	enc->cur += len;	320	enc->cur += len;
279	}	321	}
280		322
281	static void	323	ecb_inline void
282	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)
283	{	325	{
284	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))	326	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285	{	327	{
286	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	328	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287		329
…		…
298	sv_setuv (*svp, enc->stringref_idx);	340	sv_setuv (*svp, enc->stringref_idx);
299	++enc->stringref_idx;	341	++enc->stringref_idx;
300	}	342	}
301	}	343	}
302		344
303	encode_str (enc, utf8, str, len);	345	encode_str (enc, upgrade_utf8, utf8, str, len);
304	}	346	}
305		347
306	static void encode_sv (enc_t enc, SV sv);	348	static void encode_sv (enc_t enc, SV sv);
307		349
308	static void	350	static void
…		…
315		357
316	++enc->depth;	358	++enc->depth;
317		359
318	encode_uint (enc, MAJOR_ARRAY, len + 1);	360	encode_uint (enc, MAJOR_ARRAY, len + 1);
319		361
		362	if (SvMAGICAL (av))
320	for (i = 0; i <= len; ++i)	363	for (i = 0; i <= len; ++i)
321	{	364	{
322	SV **svp = av_fetch (av, i, 0);	365	SV **svp = av_fetch (av, i, 0);
323	encode_sv (enc, svp ? *svp : &PL_sv_undef);	366	encode_sv (enc, svp ? *svp : &PL_sv_undef);
324	}	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	}
325		374
326	--enc->depth;	375	--enc->depth;
327	}	376	}
328		377
329	static void	378	static void
…		…
347	while ((he = hv_iternext (hv)))	396	while ((he = hv_iternext (hv)))
348	{	397	{
349	if (HeKLEN (he) == HEf_SVKEY)	398	if (HeKLEN (he) == HEf_SVKEY)
350	encode_sv (enc, HeSVKEY (he));	399	encode_sv (enc, HeSVKEY (he));
351	else	400	else
352	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));
353		402
354	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));
355	}	404	}
356		405
357	if (mg)	406	if (mg)
…		…
429	if (ecb_expect_false (SvOBJECT (sv)))	478	if (ecb_expect_false (SvOBJECT (sv)))
430	{	479	{
431	HV *stash = SvSTASH (sv);	480	HV *stash = SvSTASH (sv);
432	GV *method;	481	GV *method;
433		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))));
434	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	486	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435	{	487	{
436	dSP;	488	dSP;
437		489
438	ENTER; SAVETMPS; PUSHMARK (SP);	490	ENTER; SAVETMPS;
		491	PUSHMARK (SP);
439	// 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
440	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	493	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441		494
442	PUTBACK;	495	PUTBACK;
443	// G_SCALAR ensures that return value is 1	496	// G_SCALAR ensures that return value is 1
…		…
456	}	509	}
457	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	510	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458	{	511	{
459	dSP;	512	dSP;
460		513
461	ENTER; SAVETMPS; PUSHMARK (SP);	514	ENTER; SAVETMPS;
		515	SAVESTACK_POS ();
		516	PUSHMARK (SP);
462	EXTEND (SP, 2);	517	EXTEND (SP, 2);
463	// 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
464	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	519	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465	PUSHs (sv_cbor);	520	PUSHs (sv_cbor);
466		521
…		…
472	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	527	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473	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));
474		529
475	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	530	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476	encode_uint (enc, MAJOR_ARRAY, count + 1);	531	encode_uint (enc, MAJOR_ARRAY, count + 1);
477	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	532	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478		533
479	while (count)	534	while (count)
480	encode_sv (enc, SP[1 - count--]);	535	encode_sv (enc, SP[1 - count--]);
481		536
482	PUTBACK;	537	PUTBACK;
…		…
541		596
542	if (SvPOKp (sv))	597	if (SvPOKp (sv))
543	{	598	{
544	STRLEN len;	599	STRLEN len;
545	char *str = SvPV (sv, len);	600	char *str = SvPV (sv, len);
546	encode_strref (enc, SvUTF8 (sv), str, len);	601	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547	}	602	}
548	else if (SvNOKp (sv))	603	else if (SvNOKp (sv))
549	encode_nv (enc, sv);	604	encode_nv (enc, sv);
550	else if (SvIOKp (sv))	605	else if (SvIOKp (sv))
551	{	606	{
…		…
568	}	623	}
569		624
570	static SV *	625	static SV *
571	encode_cbor (SV scalar, CBOR cbor)	626	encode_cbor (SV scalar, CBOR cbor)
572	{	627	{
573	enc_t enc = { };	628	enc_t enc = { 0 };
574		629
575	enc.cbor = *cbor;	630	enc.cbor = *cbor;
576	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	631	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577	enc.cur = SvPVX (enc.sv);	632	enc.cur = SvPVX (enc.sv);
578	enc.end = SvEND (enc.sv);	633	enc.end = SvEND (enc.sv);
…		…
610	U32 depth; // recursion depth	665	U32 depth; // recursion depth
611	U32 maxdepth; // recursion depth limit	666	U32 maxdepth; // recursion depth limit
612	AV *shareable;	667	AV *shareable;
613	AV *stringref;	668	AV *stringref;
614	SV *decode_tagged;	669	SV *decode_tagged;
		670	SV *err_sv; // optional sv for error, needs to be freed
615	} dec_t;	671	} dec_t;
616		672
617	#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);
618		680
619	#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;
620		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	ecb_cold static void
		704	err_nesting_exceeded (dec_t *dec)
		705	{
		706	err_set (dec, ERR_NESTING_EXCEEDED);
		707	}
		708
		709	#define ERR_DO(do) SB do; goto fail; SE
		710	#define ERR(reason) ERR_DO (err_set (dec, reason))
		711	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		712
		713	#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
		714
621	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	715	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622	#define DEC_DEC_DEPTH --dec->depth	716	#define DEC_DEC_DEPTH --dec->depth
623		717
624	static UV	718	static UV
625	decode_uint (dec_t *dec)	719	decode_uint (dec_t *dec)
626	{	720	{
627	U8 m = *dec->cur & MINOR_MASK;	721	U8 m = *dec->cur & MINOR_MASK;
628	++dec->cur;	722	++dec->cur;
629		723
630	if (ecb_expect_true (m < LENGTH_EXT1))	724	if (ecb_expect_true (m < LENGTH_EXT1))
631	return m;	725	return m;
632		726	else if (ecb_expect_true (m == LENGTH_EXT1))
633	switch (m)
634	{	727	{
635	case LENGTH_EXT1:
636	WANT (1);	728	WANT (1);
637	dec->cur += 1;	729	dec->cur += 1;
638	return dec->cur[-1];	730	return dec->cur[-1];
639		731	}
640	case LENGTH_EXT2:	732	else if (ecb_expect_true (m == LENGTH_EXT2))
		733	{
641	WANT (2);	734	WANT (2);
642	dec->cur += 2;	735	dec->cur += 2;
643	return (((UV)dec->cur[-2]) << 8)	736	return (((UV)dec->cur[-2]) << 8)
644	\| ((UV)dec->cur[-1]);	737	\| ((UV)dec->cur[-1]);
645		738	}
646	case LENGTH_EXT4:	739	else if (ecb_expect_true (m == LENGTH_EXT4))
		740	{
647	WANT (4);	741	WANT (4);
648	dec->cur += 4;	742	dec->cur += 4;
649	return (((UV)dec->cur[-4]) << 24)	743	return (((UV)dec->cur[-4]) << 24)
650	\| (((UV)dec->cur[-3]) << 16)	744	\| (((UV)dec->cur[-3]) << 16)
651	\| (((UV)dec->cur[-2]) << 8)	745	\| (((UV)dec->cur[-2]) << 8)
652	\| ((UV)dec->cur[-1]);	746	\| ((UV)dec->cur[-1]);
653		747	}
654	case LENGTH_EXT8:	748	else if (ecb_expect_true (m == LENGTH_EXT8))
		749	{
655	WANT (8);	750	WANT (8);
656	dec->cur += 8;	751	dec->cur += 8;
657		752
658	return	753	return
659	#if UVSIZE < 8	754	#if UVSIZE < 8
660	0	755	0
661	#else	756	#else
662	(((UV)dec->cur[-8]) << 56)	757	(((UV)dec->cur[-8]) << 56)
663	\| (((UV)dec->cur[-7]) << 48)	758	\| (((UV)dec->cur[-7]) << 48)
664	\| (((UV)dec->cur[-6]) << 40)	759	\| (((UV)dec->cur[-6]) << 40)
665	\| (((UV)dec->cur[-5]) << 32)	760	\| (((UV)dec->cur[-5]) << 32)
666	#endif	761	#endif
667	\| (((UV)dec->cur[-4]) << 24)	762	\| (((UV)dec->cur[-4]) << 24)
668	\| (((UV)dec->cur[-3]) << 16)	763	\| (((UV)dec->cur[-3]) << 16)
669	\| (((UV)dec->cur[-2]) << 8)	764	\| (((UV)dec->cur[-2]) << 8)
670	\| ((UV)dec->cur[-1]);	765	\| ((UV)dec->cur[-1]);
671		766	}
672	default:	767	else
673	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	768	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674	}
675		769
676	fail:	770	fail:
677	return 0;	771	return 0;
678	}	772	}
679		773
…		…
703	av_push (av, decode_sv (dec));	797	av_push (av, decode_sv (dec));
704	}	798	}
705	}	799	}
706	else	800	else
707	{	801	{
708	int i, len = decode_uint (dec);	802	UV i, len = decode_uint (dec);
709		803
		804	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
710	av_fill (av, len - 1);	805	av_fill (av, len - 1);
711		806
712	for (i = 0; i < len; ++i)	807	for (i = 0; i < len; ++i)
713	AvARRAY (av)[i] = decode_sv (dec);	808	AvARRAY (av)[i] = decode_sv (dec);
714	}	809	}
…		…
727	{	822	{
728	// for speed reasons, we specialcase single-string	823	// for speed reasons, we specialcase single-string
729	// byte or utf-8 strings as keys, but only when !stringref	824	// byte or utf-8 strings as keys, but only when !stringref
730		825
731	if (ecb_expect_true (!dec->stringref))	826	if (ecb_expect_true (!dec->stringref))
732	if ((*dec->cur - MAJOR_BYTES) <= 27)	827	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733	{	828	{
734	I32 len = decode_uint (dec);	829	STRLEN len = decode_uint (dec);
735	char key = (char )dec->cur;	830	char key = (char )dec->cur;
736		831
		832	WANT (len);
737	dec->cur += len;	833	dec->cur += len;
738		834
739	if (ecb_expect_false (dec->stringref))
740	av_push (dec->stringref, newSVpvn (key, len));
741
742	hv_store (hv, key, len, decode_sv (dec), 0);	835	hv_store (hv, key, len, decode_sv (dec), 0);
743		836
744	return;	837	return;
745	}	838	}
746	else if ((*dec->cur - MAJOR_TEXT) <= 27)	839	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747	{	840	{
748	I32 len = decode_uint (dec);	841	STRLEN len = decode_uint (dec);
749	char key = (char )dec->cur;	842	char key = (char )dec->cur;
750		843
		844	WANT (len);
751	dec->cur += len;	845	dec->cur += len;
752		846
753	if (ecb_expect_false (dec->stringref))	847	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	848	if (!is_utf8_string (key, len))
		849	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755		850
756	hv_store (hv, key, -len, decode_sv (dec), 0);	851	hv_store (hv, key, -len, decode_sv (dec), 0);
757		852
758	return;	853	return;
759	}	854	}
760		855
761	SV *k = decode_sv (dec);	856	SV *k = decode_sv (dec);
762	SV *v = decode_sv (dec);	857	SV *v = decode_sv (dec);
763		858
		859	// we leak memory if uncaught exceptions are thrown by random magical
		860	// methods, and this is hopefully the only place where it can happen,
		861	// so if there is a chance of an exception, take the very slow path.
		862	// since catching exceptions is "undocumented/internal/forbidden" by
		863	// the new p5p powers, we need to call out to a perl function :/
		864	if (ecb_expect_false (SvAMAGIC (k)))
		865	{
		866	dSP;
		867
		868	ENTER; SAVETMPS;
		869	PUSHMARK (SP);
		870	EXTEND (SP, 3);
		871	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		872	PUSHs (sv_2mortal (k));
		873	PUSHs (sv_2mortal (v));
		874
		875	PUTBACK;
		876	call_pv ("CBOR::XS::_hv_store", G_VOID \| G_DISCARD \| G_EVAL);
		877	SPAGAIN;
		878
		879	FREETMPS; LEAVE;
		880
		881	if (SvTRUE (ERRSV))
		882	ERR_ERRSV;
		883
		884	return;
		885	}
		886
764	hv_store_ent (hv, k, v, 0);	887	hv_store_ent (hv, k, v, 0);
765	SvREFCNT_dec (k);	888	SvREFCNT_dec (k);
		889
		890	fail:
		891	;
766	}	892	}
767		893
768	static SV *	894	static SV *
769	decode_hv (dec_t *dec)	895	decode_hv (dec_t *dec)
770	{	896	{
…		…
789	decode_he (dec, hv);	915	decode_he (dec, hv);
790	}	916	}
791	}	917	}
792	else	918	else
793	{	919	{
794	int pairs = decode_uint (dec);	920	UV pairs = decode_uint (dec);
		921
		922	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795		923
796	while (pairs--)	924	while (pairs--)
797	decode_he (dec, hv);	925	decode_he (dec, hv);
798	}	926	}
799		927
…		…
852	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	980	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	981	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854	}	982	}
855		983
856	if (utf8)	984	if (utf8)
		985	{
		986	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		987	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		988	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		989
857	SvUTF8_on (sv);	990	SvUTF8_on (sv);
		991	}
858		992
859	return sv;	993	return sv;
860		994
861	fail:	995	fail:
862	SvREFCNT_dec (sv);	996	SvREFCNT_dec (sv);
…		…
881	sv = newRV_noinc (decode_sv (dec));	1015	sv = newRV_noinc (decode_sv (dec));
882	break;	1016	break;
883		1017
884	case CBOR_TAG_STRINGREF_NAMESPACE:	1018	case CBOR_TAG_STRINGREF_NAMESPACE:
885	{	1019	{
		1020	// do nmot use SAVETMPS/FREETMPS, as these will
		1021	// erase mortalised caches, e.g. "shareable"
886	ENTER; SAVETMPS;	1022	ENTER;
887		1023
888	SAVESPTR (dec->stringref);	1024	SAVESPTR (dec->stringref);
889	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1025	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
890		1026
891	sv = decode_sv (dec);	1027	sv = decode_sv (dec);
892		1028
893	FREETMPS; LEAVE;	1029	LEAVE;
894	}	1030	}
895	break;	1031	break;
896		1032
897	case CBOR_TAG_STRINGREF:	1033	case CBOR_TAG_STRINGREF:
898	{	1034	{
…		…
911	case CBOR_TAG_VALUE_SHAREABLE:	1047	case CBOR_TAG_VALUE_SHAREABLE:
912	{	1048	{
913	if (ecb_expect_false (!dec->shareable))	1049	if (ecb_expect_false (!dec->shareable))
914	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1050	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
915		1051
		1052	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1053	{
916	sv = newSV (0);	1054	sv = newSV (0);
917	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1055	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918		1056
919	SV *osv = decode_sv (dec);	1057	SV *osv = decode_sv (dec);
920	sv_setsv (sv, osv);	1058	sv_setsv (sv, osv);
921	SvREFCNT_dec_NN (osv);	1059	SvREFCNT_dec_NN (osv);
		1060	}
		1061	else
		1062	{
		1063	av_push (dec->shareable, &PL_sv_undef);
		1064	int idx = AvFILLp (dec->shareable);
		1065	sv = decode_sv (dec);
		1066	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1067	}
922	}	1068	}
923	break;	1069	break;
924		1070
925	case CBOR_TAG_VALUE_SHAREDREF:	1071	case CBOR_TAG_VALUE_SHAREDREF:
926	{	1072	{
…		…
931		1077
932	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1078	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
933	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1079	ERR ("corrupted CBOR data (sharedref index out of bounds)");
934		1080
935	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1081	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1082
		1083	if (sv == &PL_sv_undef)
		1084	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
936	}	1085	}
937	break;	1086	break;
938		1087
939	case CBOR_TAG_PERL_OBJECT:	1088	case CBOR_TAG_PERL_OBJECT:
940	{	1089	{
		1090	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1091	goto filter;
		1092
941	sv = decode_sv (dec);	1093	sv = decode_sv (dec);
942		1094
943	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)	1095	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)
944	ERR ("corrupted CBOR data (non-array perl object)");	1096	ERR ("corrupted CBOR data (non-array perl object)");
945		1097
…		…
955	if (!method)	1107	if (!method)
956	ERR ("cannot decode perl-object (package does not have a THAW method)");	1108	ERR ("cannot decode perl-object (package does not have a THAW method)");
957		1109
958	dSP;	1110	dSP;
959		1111
960	ENTER; SAVETMPS; PUSHMARK (SP);	1112	ENTER; SAVETMPS;
		1113	PUSHMARK (SP);
961	EXTEND (SP, len + 1);	1114	EXTEND (SP, len + 1);
962	// we re-bless the reference to get overload and other niceties right	1115	// we re-bless the reference to get overload and other niceties right
963	PUSHs (*av_fetch (av, 0, 1));	1116	PUSHs (*av_fetch (av, 0, 1));
964	PUSHs (sv_cbor);	1117	PUSHs (sv_cbor);
965		1118
…		…
973	SPAGAIN;	1126	SPAGAIN;
974		1127
975	if (SvTRUE (ERRSV))	1128	if (SvTRUE (ERRSV))
976	{	1129	{
977	FREETMPS; LEAVE;	1130	FREETMPS; LEAVE;
978	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1131	ERR_ERRSV;
979	}	1132	}
980		1133
981	SvREFCNT_dec (sv);	1134	SvREFCNT_dec (sv);
982	sv = SvREFCNT_inc (POPs);	1135	sv = SvREFCNT_inc (POPs);
983		1136
…		…
986	FREETMPS; LEAVE;	1139	FREETMPS; LEAVE;
987	}	1140	}
988	break;	1141	break;
989		1142
990	default:	1143	default:
		1144	filter:
991	{	1145	{
		1146	SV *tag_sv = newSVuv (tag);
		1147
992	sv = decode_sv (dec);	1148	sv = decode_sv (dec);
993		1149
994	dSP;	1150	dSP;
995	ENTER; SAVETMPS; PUSHMARK (SP);	1151	ENTER; SAVETMPS;
		1152	SAVESTACK_POS ();
		1153	PUSHMARK (SP);
996	EXTEND (SP, 2);	1154	EXTEND (SP, 2);
997	PUSHs (newSVuv (tag));	1155	PUSHs (tag_sv);
998	PUSHs (sv);	1156	PUSHs (sv);
999		1157
1000	PUTBACK;	1158	PUTBACK;
1001	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);	1159	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
1002	SPAGAIN;	1160	SPAGAIN;
1003		1161
1004	if (SvTRUE (ERRSV))	1162	if (SvTRUE (ERRSV))
1005	{	1163	{
		1164	SvREFCNT_dec (tag_sv);
1006	FREETMPS; LEAVE;	1165	FREETMPS; LEAVE;
1007	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1166	ERR_ERRSV;
1008	}	1167	}
1009		1168
1010	if (count)	1169	if (count)
1011	{	1170	{
		1171	SvREFCNT_dec (tag_sv);
1012	SvREFCNT_dec (sv);	1172	SvREFCNT_dec (sv);
1013	sv = SvREFCNT_inc (POPs);	1173	sv = SvREFCNT_inc (POPs);
1014	}	1174	}
1015	else	1175	else
1016	{	1176	{
1017	AV *av = newAV ();	1177	AV *av = newAV ();
1018	av_push (av, newSVuv (tag));	1178	av_push (av, tag_sv);
1019	av_push (av, sv);	1179	av_push (av, sv);
1020		1180
1021	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1181	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
1022	? cbor_tagged_stash	1182	? cbor_tagged_stash
1023	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1183	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
1109		1269
1110	return newSVnv (ecb_binary64_to_double (fp));	1270	return newSVnv (ecb_binary64_to_double (fp));
1111	}	1271	}
1112		1272
1113	// 0..19 unassigned simple	1273	// 0..19 unassigned simple
1114	// 24 reserved + unassigned (reserved values are not encodable)	1274	// 24 reserved + unassigned simple (reserved values are not encodable)
		1275	// 28-30 unassigned misc
		1276	// 31 break code
1115	default:	1277	default:
1116	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1278	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117	}	1279	}
1118		1280
1119	break;	1281	break;
1120	}	1282	}
1121		1283
…		…
1124	}	1286	}
1125		1287
1126	static SV *	1288	static SV *
1127	decode_cbor (SV string, CBOR cbor, char **offset_return)	1289	decode_cbor (SV string, CBOR cbor, char **offset_return)
1128	{	1290	{
1129	dec_t dec = { };	1291	dec_t dec = { 0 };
1130	SV *sv;	1292	SV *sv;
1131	STRLEN len;	1293	STRLEN len;
1132	char *data = SvPVbyte (string, len);	1294	char *data = SvPVbyte (string, len);
1133		1295
1134	if (len > cbor->max_size && cbor->max_size)	1296	if (len > cbor->max_size && cbor->max_size)
…		…
1148	if (dec.cur != dec.end && !dec.err)	1310	if (dec.cur != dec.end && !dec.err)
1149	dec.err = "garbage after CBOR object";	1311	dec.err = "garbage after CBOR object";
1150		1312
1151	if (dec.err)	1313	if (dec.err)
1152	{	1314	{
		1315	if (dec.shareable)
		1316	{
		1317	// need to break cyclic links, which would all be in shareable
		1318	int i;
		1319	SV **svp;
		1320
		1321	for (i = av_len (dec.shareable) + 1; i--; )
		1322	if ((svp = av_fetch (dec.shareable, i, 0)))
		1323	sv_setsv (*svp, &PL_sv_undef);
		1324	}
		1325
1153	SvREFCNT_dec (sv);	1326	SvREFCNT_dec (sv);
		1327
		1328	if (dec.err_sv)
		1329	sv_2mortal (dec.err_sv);
		1330
1154	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1331	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1155	}	1332	}
1156		1333
1157	sv = sv_2mortal (sv);	1334	sv = sv_2mortal (sv);
1158		1335
1159	return sv;	1336	return sv;
1160	}	1337	}
1161		1338
		1339	/////////////////////////////////////////////////////////////////////////////
		1340	// incremental parser
		1341
		1342	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1343
		1344	// returns 0 for notyet, 1 for success or error
		1345	static int
		1346	incr_parse (CBOR self, SV cborstr)
		1347	{
		1348	STRLEN cur;
		1349	SvPV (cborstr, cur);
		1350
		1351	while (ecb_expect_true (self->incr_need <= cur))
		1352	{
		1353	// table of integer count bytes
		1354	static I8 incr_len[MINOR_MASK + 1] = {
		1355	0, 0, 0, 0, 0, 0, 0, 0,
		1356	0, 0, 0, 0, 0, 0, 0, 0,
		1357	0, 0, 0, 0, 0, 0, 0, 0,
		1358	1, 2, 4, 8,-1,-1,-1,-2
		1359	};
		1360
		1361	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1362	U8 m = *p & MINOR_MASK;
		1363	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1364	I8 ilen = incr_len[m];
		1365
		1366	self->incr_need = self->incr_pos + 1;
		1367
		1368	if (ecb_expect_false (ilen < 0))
		1369	{
		1370	if (m != MINOR_INDEF)
		1371	return 1; // error
		1372
		1373	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1374	{
		1375	if (count >= 0)
		1376	return 1; // error
		1377
		1378	count = 1;
		1379	}
		1380	else
		1381	{
		1382	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1383	count = -1;
		1384	}
		1385	}
		1386	else
		1387	{
		1388	self->incr_need += ilen;
		1389	if (ecb_expect_false (self->incr_need > cur))
		1390	return 0;
		1391
		1392	int major = *p >> MAJOR_SHIFT;
		1393
		1394	switch (major)
		1395	{
		1396	case MAJOR_TAG >> MAJOR_SHIFT:
		1397	++count; // tags merely prefix another value
		1398	break;
		1399
		1400	case MAJOR_BYTES >> MAJOR_SHIFT:
		1401	case MAJOR_TEXT >> MAJOR_SHIFT:
		1402	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1403	case MAJOR_MAP >> MAJOR_SHIFT:
		1404	{
		1405	UV len;
		1406
		1407	if (ecb_expect_false (ilen))
		1408	{
		1409	len = 0;
		1410
		1411	do {
		1412	len = (len << 8) \| *++p;
		1413	} while (--ilen);
		1414	}
		1415	else
		1416	len = m;
		1417
		1418	switch (major)
		1419	{
		1420	case MAJOR_BYTES >> MAJOR_SHIFT:
		1421	case MAJOR_TEXT >> MAJOR_SHIFT:
		1422	self->incr_need += len;
		1423	if (ecb_expect_false (self->incr_need > cur))
		1424	return 0;
		1425
		1426	break;
		1427
		1428	case MAJOR_MAP >> MAJOR_SHIFT:
		1429	len <<= 1;
		1430	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1431	if (len)
		1432	{
		1433	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1434	count = len + 1;
		1435	}
		1436	break;
		1437	}
		1438	}
		1439	}
		1440	}
		1441
		1442	self->incr_pos = self->incr_need;
		1443
		1444	if (count > 0)
		1445	{
		1446	while (!--count)
		1447	{
		1448	if (!AvFILLp (self->incr_count))
		1449	return 1; // done
		1450
		1451	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1452	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1453	}
		1454
		1455	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1456	}
		1457	}
		1458
		1459	return 0;
		1460	}
		1461
		1462
1162	/////////////////////////////////////////////////////////////////////////////	1463	/////////////////////////////////////////////////////////////////////////////
1163	// XS interface functions	1464	// XS interface functions
1164		1465
1165	MODULE = CBOR::XS PACKAGE = CBOR::XS	1466	MODULE = CBOR::XS PACKAGE = CBOR::XS
1166		1467
…		…
1178		1479
1179	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1480	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180		1481
1181	sv_cbor = newSVpv ("CBOR", 0);	1482	sv_cbor = newSVpv ("CBOR", 0);
1182	SvREADONLY_on (sv_cbor);	1483	SvREADONLY_on (sv_cbor);
		1484
		1485	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183	}	1486	}
1184		1487
1185	PROTOTYPES: DISABLE	1488	PROTOTYPES: DISABLE
1186		1489
1187	void CLONE (...)	1490	void CLONE (...)
…		…
1206	void shrink (CBOR *self, int enable = 1)	1509	void shrink (CBOR *self, int enable = 1)
1207	ALIAS:	1510	ALIAS:
1208	shrink = F_SHRINK	1511	shrink = F_SHRINK
1209	allow_unknown = F_ALLOW_UNKNOWN	1512	allow_unknown = F_ALLOW_UNKNOWN
1210	allow_sharing = F_ALLOW_SHARING	1513	allow_sharing = F_ALLOW_SHARING
		1514	allow_cycles = F_ALLOW_CYCLES
		1515	forbid_objects = F_FORBID_OBJECTS
1211	pack_strings = F_PACK_STRINGS	1516	pack_strings = F_PACK_STRINGS
		1517	text_keys = F_TEXT_KEYS
		1518	text_strings = F_TEXT_STRINGS
		1519	validate_utf8 = F_VALIDATE_UTF8
1212	PPCODE:	1520	PPCODE:
1213	{	1521	{
1214	if (enable)	1522	if (enable)
1215	self->flags \|= ix;	1523	self->flags \|= ix;
1216	else	1524	else
…		…
1222	void get_shrink (CBOR *self)	1530	void get_shrink (CBOR *self)
1223	ALIAS:	1531	ALIAS:
1224	get_shrink = F_SHRINK	1532	get_shrink = F_SHRINK
1225	get_allow_unknown = F_ALLOW_UNKNOWN	1533	get_allow_unknown = F_ALLOW_UNKNOWN
1226	get_allow_sharing = F_ALLOW_SHARING	1534	get_allow_sharing = F_ALLOW_SHARING
		1535	get_allow_cycles = F_ALLOW_CYCLES
		1536	get_forbid_objects = F_FORBID_OBJECTS
1227	get_pack_strings = F_PACK_STRINGS	1537	get_pack_strings = F_PACK_STRINGS
		1538	get_text_keys = F_TEXT_KEYS
		1539	get_text_strings = F_TEXT_STRINGS
		1540	get_validate_utf8 = F_VALIDATE_UTF8
1228	PPCODE:	1541	PPCODE:
1229	XPUSHs (boolSV (self->flags & ix));	1542	XPUSHs (boolSV (self->flags & ix));
1230		1543
1231	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1544	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232	PPCODE:	1545	PPCODE:
…		…
1281	EXTEND (SP, 2);	1594	EXTEND (SP, 2);
1282	PUSHs (sv);	1595	PUSHs (sv);
1283	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1596	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284	}	1597	}
1285		1598
		1599	void incr_parse (CBOR self, SV cborstr)
		1600	ALIAS:
		1601	incr_parse_multiple = 1
		1602	PPCODE:
		1603	{
		1604	if (SvUTF8 (cborstr))
		1605	sv_utf8_downgrade (cborstr, 0);
		1606
		1607	if (!self->incr_count)
		1608	{
		1609	self->incr_count = newAV ();
		1610	self->incr_pos = 0;
		1611	self->incr_need = 1;
		1612
		1613	av_push (self->incr_count, newSViv (1));
		1614	}
		1615
		1616	do
		1617	{
		1618	if (!incr_parse (self, cborstr))
		1619	{
		1620	if (self->incr_need > self->max_size && self->max_size)
		1621	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1622	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1623
		1624	break;
		1625	}
		1626
		1627	SV *sv;
		1628	char *offset;
		1629
		1630	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1631	XPUSHs (sv);
		1632
		1633	sv_chop (cborstr, offset);
		1634
		1635	av_clear (self->incr_count);
		1636	av_push (self->incr_count, newSViv (1));
		1637
		1638	self->incr_pos = 0;
		1639	self->incr_need = self->incr_pos + 1;
		1640	}
		1641	while (ix);
		1642	}
		1643
		1644	void incr_reset (CBOR *self)
		1645	CODE:
		1646	{
		1647	SvREFCNT_dec (self->incr_count);
		1648	self->incr_count = 0;
		1649	}
		1650
1286	void DESTROY (CBOR *self)	1651	void DESTROY (CBOR *self)
1287	PPCODE:	1652	PPCODE:
1288	cbor_free (self);	1653	cbor_free (self);
1289		1654
1290	PROTOTYPES: ENABLE	1655	PROTOTYPES: ENABLE
1291		1656
1292	void encode_cbor (SV *scalar)	1657	void encode_cbor (SV *scalar)
		1658	ALIAS:
		1659	encode_cbor = 0
		1660	encode_cbor_sharing = F_ALLOW_SHARING
1293	PPCODE:	1661	PPCODE:
1294	{	1662	{
1295	CBOR cbor;	1663	CBOR cbor;
1296	cbor_init (&cbor);	1664	cbor_init (&cbor);
		1665	cbor.flags \|= ix;
1297	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1666	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298	XPUSHs (scalar);	1667	XPUSHs (scalar);
1299	}	1668	}
1300		1669
1301	void decode_cbor (SV *cborstr)	1670	void decode_cbor (SV *cborstr)

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs. Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.60 by root, Sat Nov 26 00:47:02 2016 UTC