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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.16 by root, Tue Oct 29 20:59:16 2013 UTC vs.
Revision 1.34 by root, Sat Nov 30 16:19:59 2013 UTC

…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	20	#endif
21	#ifndef HvNAMEUTF8	21	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	22	# define HvNAMEUTF8(hv) 0
23	#endif	23	#endif
		24	#ifndef SvREFCNT_dec_NN
		25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		26	#endif
24		27
25	// known tags	28	// known tags
26	enum cbor_tag	29	enum cbor_tag
27	{	30	{
28	// inofficial extensions (pending iana registration)	31	// extensions
29	CBOR_TAG_PERL_OBJECT = 256,	32	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30	CBOR_TAG_GENERIC_OBJECT = 257,	33	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
		34	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
		35	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
		36	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
		37	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
		38	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31		39
32	// rfc7049	40	// rfc7049
33	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	41	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	42	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35	CBOR_TAG_POS_BIGNUM = 2, // byte string	43	CBOR_TAG_POS_BIGNUM = 2, // byte string
…		…
49	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	57	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50		58
51	CBOR_TAG_MAGIC = 55799 // self-describe cbor	59	CBOR_TAG_MAGIC = 55799 // self-describe cbor
52	};	60	};
53		61
54	#define F_SHRINK 0x00000200UL	62	#define F_SHRINK 0x00000001UL
55	#define F_ALLOW_UNKNOWN 0x00002000UL	63	#define F_ALLOW_UNKNOWN 0x00000002UL
		64	#define F_ALLOW_SHARING 0x00000004UL
		65	#define F_PACK_STRINGS 0x00000008UL
56		66
57	#define INIT_SIZE 32 // initial scalar size to be allocated	67	#define INIT_SIZE 32 // initial scalar size to be allocated
58		68
59	#define SB do {	69	#define SB do {
60	#define SE } while (0)	70	#define SE } while (0)
…		…
72	# define CBOR_SLOW 0	82	# define CBOR_SLOW 0
73	# define CBOR_STASH cbor_stash	83	# define CBOR_STASH cbor_stash
74	#endif	84	#endif
75		85
76	static HV cbor_stash, types_boolean_stash, types_error_stash, cbor_tagged_stash; // CBOR::XS::	86	static HV cbor_stash, types_boolean_stash, types_error_stash, cbor_tagged_stash; // CBOR::XS::
77	static SV types_true, types_false, types_error, sv_cbor;	87	static SV types_true, types_false, types_error, sv_cbor, *default_filter;
78		88
79	typedef struct {	89	typedef struct {
80	U32 flags;	90	U32 flags;
81	U32 max_depth;	91	U32 max_depth;
82	STRLEN max_size;	92	STRLEN max_size;
		93	SV *filter;
83	} CBOR;	94	} CBOR;
84		95
85	ecb_inline void	96	ecb_inline void
86	cbor_init (CBOR *cbor)	97	cbor_init (CBOR *cbor)
87	{	98	{
88	Zero (cbor, 1, CBOR);	99	Zero (cbor, 1, CBOR);
89	cbor->max_depth = 512;	100	cbor->max_depth = 512;
		101	}
		102
		103	ecb_inline void
		104	cbor_free (CBOR *cbor)
		105	{
		106	SvREFCNT_dec (cbor->filter);
90	}	107	}
91		108
92	/////////////////////////////////////////////////////////////////////////////	109	/////////////////////////////////////////////////////////////////////////////
93	// utility functions	110	// utility functions
94		111
…		…
116	SvPV_renew (sv, SvCUR (sv) + 1);	133	SvPV_renew (sv, SvCUR (sv) + 1);
117	#endif	134	#endif
118	}	135	}
119	}	136	}
120		137
121	/////////////////////////////////////////////////////////////////////////////	138	// minimum length of a string to be registered for stringref
122	// fp hell	139	ecb_inline int
123		140	minimum_string_length (UV idx)
124	//TODO	141	{
		142	return idx > 23
		143	? idx > 0xffU
		144	? idx > 0xffffU
		145	? idx > 0xffffffffU
		146	? 11
		147	: 7
		148	: 5
		149	: 4
		150	: 3;
		151	}
125		152
126	/////////////////////////////////////////////////////////////////////////////	153	/////////////////////////////////////////////////////////////////////////////
127	// encoder	154	// encoder
128		155
129	// structure used for encoding CBOR	156	// structure used for encoding CBOR
…		…
132	char *cur; // SvPVX (sv) + current output position	159	char *cur; // SvPVX (sv) + current output position
133	char *end; // SvEND (sv)	160	char *end; // SvEND (sv)
134	SV *sv; // result scalar	161	SV *sv; // result scalar
135	CBOR cbor;	162	CBOR cbor;
136	U32 depth; // recursion level	163	U32 depth; // recursion level
		164	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		165	UV stringref_idx;
		166	HV *shareable; // ptr => index, or 0
		167	UV shareable_idx;
137	} enc_t;	168	} enc_t;
138		169
139	ecb_inline void	170	ecb_inline void
140	need (enc_t *enc, STRLEN len)	171	need (enc_t *enc, STRLEN len)
141	{	172	{
…		…
158	static void	189	static void
159	encode_uint (enc_t *enc, int major, UV len)	190	encode_uint (enc_t *enc, int major, UV len)
160	{	191	{
161	need (enc, 9);	192	need (enc, 9);
162		193
163	if (len < 24)	194	if (ecb_expect_true (len < 24))
164	*enc->cur++ = major \| len;	195	*enc->cur++ = major \| len;
165	else if (len <= 0xff)	196	else if (ecb_expect_true (len <= 0xff))
166	{	197	{
167	*enc->cur++ = major \| 24;	198	*enc->cur++ = major \| 24;
168	*enc->cur++ = len;	199	*enc->cur++ = len;
169	}	200	}
170	else if (len <= 0xffff)	201	else if (len <= 0xffff)
…		…
193	*enc->cur++ = len >> 8;	224	*enc->cur++ = len >> 8;
194	*enc->cur++ = len;	225	*enc->cur++ = len;
195	}	226	}
196	}	227	}
197		228
198	static void	229	ecb_inline void
		230	encode_tag (enc_t *enc, UV tag)
		231	{
		232	encode_uint (enc, 0xc0, tag);
		233	}
		234
		235	ecb_inline void
199	encode_str (enc_t enc, int utf8, char str, STRLEN len)	236	encode_str (enc_t enc, int utf8, char str, STRLEN len)
200	{	237	{
201	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	238	encode_uint (enc, utf8 ? 0x60 : 0x40, len);
202	need (enc, len);	239	need (enc, len);
203	memcpy (enc->cur, str, len);	240	memcpy (enc->cur, str, len);
204	enc->cur += len;	241	enc->cur += len;
205	}	242	}
206		243
		244	static void
		245	encode_strref (enc_t enc, int utf8, char str, STRLEN len)
		246	{
		247	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		248	{
		249	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		250
		251	if (SvOK (*svp))
		252	{
		253	// already registered, use stringref
		254	encode_tag (enc, CBOR_TAG_STRINGREF);
		255	encode_uint (enc, 0x00, SvUV (*svp));
		256	return;
		257	}
		258	else if (len >= minimum_string_length (enc->stringref_idx))
		259	{
		260	// register only
		261	sv_setuv (*svp, enc->stringref_idx);
		262	++enc->stringref_idx;
		263	}
		264	}
		265
		266	encode_str (enc, utf8, str, len);
		267	}
		268
207	static void encode_sv (enc_t enc, SV sv);	269	static void encode_sv (enc_t enc, SV sv);
208		270
209	static void	271	static void
210	encode_av (enc_t enc, AV av)	272	encode_av (enc_t enc, AV av)
211	{	273	{
…		…
248	while ((he = hv_iternext (hv)))	310	while ((he = hv_iternext (hv)))
249	{	311	{
250	if (HeKLEN (he) == HEf_SVKEY)	312	if (HeKLEN (he) == HEf_SVKEY)
251	encode_sv (enc, HeSVKEY (he));	313	encode_sv (enc, HeSVKEY (he));
252	else	314	else
253	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	315	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254		316
255	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	317	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
256	}	318	}
257		319
258	if (mg)	320	if (mg)
…		…
263		325
264	// encode objects, arrays and special \0=false and \1=true values.	326	// encode objects, arrays and special \0=false and \1=true values.
265	static void	327	static void
266	encode_rv (enc_t enc, SV sv)	328	encode_rv (enc_t enc, SV sv)
267	{	329	{
268	svtype svt;
269
270	SvGETMAGIC (sv);	330	SvGETMAGIC (sv);
		331
271	svt = SvTYPE (sv);	332	svtype svt = SvTYPE (sv);
272		333
273	if (ecb_expect_false (SvOBJECT (sv)))	334	if (ecb_expect_false (SvOBJECT (sv)))
274	{	335	{
275	HV *boolean_stash = !CBOR_SLOW \|\| types_boolean_stash	336	HV *boolean_stash = !CBOR_SLOW \|\| types_boolean_stash
276	? types_boolean_stash	337	? types_boolean_stash
…		…
281	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	342	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
282	? cbor_tagged_stash	343	? cbor_tagged_stash
283	: gv_stashpv ("CBOR::XS::Tagged" , 1);	344	: gv_stashpv ("CBOR::XS::Tagged" , 1);
284		345
285	HV *stash = SvSTASH (sv);	346	HV *stash = SvSTASH (sv);
286	GV *method;
287		347
288	if (stash == boolean_stash)	348	if (stash == boolean_stash)
		349	{
289	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	350	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);
		351	return;
		352	}
290	else if (stash == error_stash)	353	else if (stash == error_stash)
		354	{
291	encode_ch (enc, 0xe0 \| 23);	355	encode_ch (enc, 0xe0 \| 23);
		356	return;
		357	}
292	else if (stash == tagged_stash)	358	else if (stash == tagged_stash)
293	{	359	{
294	if (svt != SVt_PVAV)	360	if (svt != SVt_PVAV)
295	croak ("encountered CBOR::XS::Tagged object that isn't an array");	361	croak ("encountered CBOR::XS::Tagged object that isn't an array");
296		362
297	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	363	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));
298	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	364	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		365
		366	return;
		367	}
		368	}
		369
		370	if (ecb_expect_false (SvREFCNT (sv) > 1)
		371	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		372	{
		373	if (!enc->shareable)
		374	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		375
		376	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		377
		378	if (SvOK (*svp))
299	}	379	{
		380	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		381	encode_uint (enc, 0x00, SvUV (*svp));
		382	return;
		383	}
		384	else
		385	{
		386	sv_setuv (*svp, enc->shareable_idx);
		387	++enc->shareable_idx;
		388	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		389	}
		390	}
		391
		392	if (ecb_expect_false (SvOBJECT (sv)))
		393	{
		394	HV *stash = SvSTASH (sv);
		395	GV *method;
		396
300	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	397	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301	{	398	{
302	dSP;	399	dSP;
303		400
304	ENTER; SAVETMPS; PUSHMARK (SP);	401	ENTER; SAVETMPS; PUSHMARK (SP);
305	// we re-bless the reference to get overload and other niceties right	402	// we re-bless the reference to get overload and other niceties right
…		…
336		433
337	// catch this surprisingly common error	434	// catch this surprisingly common error
338	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	435	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	436	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340		437
341	encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT);	438	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342	encode_uint (enc, 0x80, count + 1);	439	encode_uint (enc, 0x80, count + 1);
343	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	440	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344		441
345	while (count)	442	while (count)
346	encode_sv (enc, SP[1 - count--]);	443	encode_sv (enc, SP[1 - count--]);
347		444
348	PUTBACK;	445	PUTBACK;
…		…
355	}	452	}
356	else if (svt == SVt_PVHV)	453	else if (svt == SVt_PVHV)
357	encode_hv (enc, (HV *)sv);	454	encode_hv (enc, (HV *)sv);
358	else if (svt == SVt_PVAV)	455	else if (svt == SVt_PVAV)
359	encode_av (enc, (AV *)sv);	456	encode_av (enc, (AV *)sv);
360	else if (svt < SVt_PVAV)
361	{
362	STRLEN len = 0;
363	char *pv = svt ? SvPV (sv, len) : 0;
364
365	if (len == 1 && *pv == '1')
366	encode_ch (enc, 0xe0 \| 21);
367	else if (len == 1 && *pv == '0')
368	encode_ch (enc, 0xe0 \| 20);
369	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370	encode_ch (enc, 0xe0 \| 23);
371	else
372	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374	}
375	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376	encode_ch (enc, 0xe0 \| 23);
377	else	457	else
378	croak ("encountered %s, but CBOR can only represent references to arrays or hashes",	458	{
379	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	459	encode_tag (enc, CBOR_TAG_INDIRECTION);
		460	encode_sv (enc, sv);
		461	}
380	}	462	}
381		463
382	static void	464	static void
383	encode_nv (enc_t enc, SV sv)	465	encode_nv (enc_t enc, SV sv)
384	{	466	{
…		…
422		504
423	if (SvPOKp (sv))	505	if (SvPOKp (sv))
424	{	506	{
425	STRLEN len;	507	STRLEN len;
426	char *str = SvPV (sv, len);	508	char *str = SvPV (sv, len);
427	encode_str (enc, SvUTF8 (sv), str, len);	509	encode_strref (enc, SvUTF8 (sv), str, len);
428	}	510	}
429	else if (SvNOKp (sv))	511	else if (SvNOKp (sv))
430	encode_nv (enc, sv);	512	encode_nv (enc, sv);
431	else if (SvIOKp (sv))	513	else if (SvIOKp (sv))
432	{	514	{
…		…
449	}	531	}
450		532
451	static SV *	533	static SV *
452	encode_cbor (SV scalar, CBOR cbor)	534	encode_cbor (SV scalar, CBOR cbor)
453	{	535	{
454	enc_t enc;	536	enc_t enc = { };
455		537
456	enc.cbor = *cbor;	538	enc.cbor = *cbor;
457	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	539	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458	enc.cur = SvPVX (enc.sv);	540	enc.cur = SvPVX (enc.sv);
459	enc.end = SvEND (enc.sv);	541	enc.end = SvEND (enc.sv);
460	enc.depth = 0;
461		542
462	SvPOK_only (enc.sv);	543	SvPOK_only (enc.sv);
		544
		545	if (cbor->flags & F_PACK_STRINGS)
		546	{
		547	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		548	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
		549	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
		550	}
		551
463	encode_sv (&enc, scalar);	552	encode_sv (&enc, scalar);
464		553
465	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	554	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	555	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467		556
…		…
481	U8 *end; // end of input string	570	U8 *end; // end of input string
482	const char *err; // parse error, if != 0	571	const char *err; // parse error, if != 0
483	CBOR cbor;	572	CBOR cbor;
484	U32 depth; // recursion depth	573	U32 depth; // recursion depth
485	U32 maxdepth; // recursion depth limit	574	U32 maxdepth; // recursion depth limit
		575	AV *shareable;
		576	AV *stringref;
		577	SV *decode_tagged;
486	} dec_t;	578	} dec_t;
487		579
488	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	580	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
489		581
490	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	582	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
…		…
522	\| ((UV)dec->cur[-1]);	614	\| ((UV)dec->cur[-1]);
523		615
524	case 27:	616	case 27:
525	WANT (9);	617	WANT (9);
526	dec->cur += 9;	618	dec->cur += 9;
		619
		620	return
		621	#if UVSIZE < 8
		622	0
		623	#else
527	return (((UV)dec->cur[-8]) << 56)	624	(((UV)dec->cur[-8]) << 56)
528	\| (((UV)dec->cur[-7]) << 48)	625	\| (((UV)dec->cur[-7]) << 48)
529	\| (((UV)dec->cur[-6]) << 40)	626	\| (((UV)dec->cur[-6]) << 40)
530	\| (((UV)dec->cur[-5]) << 32)	627	\| (((UV)dec->cur[-5]) << 32)
		628	#endif
531	\| (((UV)dec->cur[-4]) << 24)	629	\| (((UV)dec->cur[-4]) << 24)
532	\| (((UV)dec->cur[-3]) << 16)	630	\| (((UV)dec->cur[-3]) << 16)
533	\| (((UV)dec->cur[-2]) << 8)	631	\| (((UV)dec->cur[-2]) << 8)
534	\| ((UV)dec->cur[-1]);	632	\| ((UV)dec->cur[-1]);
535		633
…		…
588		686
589	static void	687	static void
590	decode_he (dec_t dec, HV hv)	688	decode_he (dec_t dec, HV hv)
591	{	689	{
592	// for speed reasons, we specialcase single-string	690	// for speed reasons, we specialcase single-string
593	// byte or utf-8 strings as keys.	691	// byte or utf-8 strings as keys, but only when !stringref
594		692
		693	if (ecb_expect_true (!dec->stringref))
595	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	694	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)
596	{	695	{
597	I32 len = decode_uint (dec);	696	I32 len = decode_uint (dec);
598	char key = (char )dec->cur;	697	char key = (char )dec->cur;
599		698
600	dec->cur += len;	699	dec->cur += len;
601		700
		701	if (ecb_expect_false (dec->stringref))
		702	av_push (dec->stringref, newSVpvn (key, len));
		703
602	hv_store (hv, key, len, decode_sv (dec), 0);	704	hv_store (hv, key, len, decode_sv (dec), 0);
		705
		706	return;
603	}	707	}
604	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	708	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)
605	{	709	{
606	I32 len = decode_uint (dec);	710	I32 len = decode_uint (dec);
607	char key = (char )dec->cur;	711	char key = (char )dec->cur;
608		712
609	dec->cur += len;	713	dec->cur += len;
610		714
		715	if (ecb_expect_false (dec->stringref))
		716	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
		717
611	hv_store (hv, key, -len, decode_sv (dec), 0);	718	hv_store (hv, key, -len, decode_sv (dec), 0);
		719
		720	return;
612	}	721	}
613	else	722
614	{
615	SV *k = decode_sv (dec);	723	SV *k = decode_sv (dec);
616	SV *v = decode_sv (dec);	724	SV *v = decode_sv (dec);
617		725
618	hv_store_ent (hv, k, v, 0);	726	hv_store_ent (hv, k, v, 0);
619	SvREFCNT_dec (k);	727	SvREFCNT_dec (k);
620	}
621	}	728	}
622		729
623	static SV *	730	static SV *
624	decode_hv (dec_t *dec)	731	decode_hv (dec_t *dec)
625	{	732	{
…		…
666	{	773	{
667	SV *sv = 0;	774	SV *sv = 0;
668		775
669	if ((*dec->cur & 31) == 31)	776	if ((*dec->cur & 31) == 31)
670	{	777	{
		778	// indefinite length strings
671	++dec->cur;	779	++dec->cur;
672		780
		781	unsigned char major = *dec->cur & 0xe0;
		782
673	sv = newSVpvn ("", 0);	783	sv = newSVpvn ("", 0);
674		784
675	// not very fast, and certainly not robust against illegal input
676	for (;;)	785	for (;;)
677	{	786	{
678	WANT (1);	787	WANT (1);
679		788
		789	if ((*dec->cur ^ major) >= 31)
680	if (*dec->cur == (0xe0 \| 31))	790	if (*dec->cur == (0xe0 \| 31))
681	{	791	{
682	++dec->cur;	792	++dec->cur;
683	break;	793	break;
684	}	794	}
		795	else
		796	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685		797
686	sv_catsv (sv, decode_sv (dec));	798	STRLEN len = decode_uint (dec);
		799
		800	WANT (len);
		801	sv_catpvn (sv, dec->cur, len);
		802	dec->cur += len;
687	}	803	}
688	}	804	}
689	else	805	else
690	{	806	{
691	STRLEN len = decode_uint (dec);	807	STRLEN len = decode_uint (dec);
692		808
693	WANT (len);	809	WANT (len);
694	sv = newSVpvn (dec->cur, len);	810	sv = newSVpvn (dec->cur, len);
695	dec->cur += len;	811	dec->cur += len;
		812
		813	if (ecb_expect_false (dec->stringref)
		814	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		815	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
696	}	816	}
697		817
698	if (utf8)	818	if (utf8)
699	SvUTF8_on (sv);	819	SvUTF8_on (sv);
700		820
…		…
706	}	826	}
707		827
708	static SV *	828	static SV *
709	decode_tagged (dec_t *dec)	829	decode_tagged (dec_t *dec)
710	{	830	{
		831	SV *sv = 0;
711	UV tag = decode_uint (dec);	832	UV tag = decode_uint (dec);
		833
		834	WANT (1);
		835
		836	switch (tag)
		837	{
		838	case CBOR_TAG_MAGIC:
		839	sv = decode_sv (dec);
		840	break;
		841
		842	case CBOR_TAG_INDIRECTION:
		843	sv = newRV_noinc (decode_sv (dec));
		844	break;
		845
		846	case CBOR_TAG_STRINGREF_NAMESPACE:
		847	{
		848	ENTER; SAVETMPS;
		849
		850	SAVESPTR (dec->stringref);
		851	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
		852
		853	sv = decode_sv (dec);
		854
		855	FREETMPS; LEAVE;
		856	}
		857	break;
		858
		859	case CBOR_TAG_STRINGREF:
		860	{
		861	if ((*dec->cur >> 5) != 0)
		862	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		863
		864	UV idx = decode_uint (dec);
		865
		866	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
		867	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		868
		869	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		870	}
		871	break;
		872
		873	case CBOR_TAG_VALUE_SHAREABLE:
		874	{
		875	if (ecb_expect_false (!dec->shareable))
		876	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
		877
		878	sv = newSV (0);
		879	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
		880
712	SV *sv = decode_sv (dec);	881	SV *osv = decode_sv (dec);
		882	sv_setsv (sv, osv);
		883	SvREFCNT_dec_NN (osv);
		884	}
		885	break;
713		886
714	if (tag == CBOR_TAG_MAGIC)	887	case CBOR_TAG_VALUE_SHAREDREF:
715	return sv;	888	{
716	else if (tag == CBOR_TAG_PERL_OBJECT)	889	if ((*dec->cur >> 5) != 0)
717	{	890	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
		891
		892	UV idx = decode_uint (dec);
		893
		894	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
		895	ERR ("corrupted CBOR data (sharedref index out of bounds)");
		896
		897	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		898	}
		899	break;
		900
		901	case CBOR_TAG_PERL_OBJECT:
		902	{
		903	sv = decode_sv (dec);
		904
718	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)	905	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)
719	ERR ("corrupted CBOR data (non-array perl object)");	906	ERR ("corrupted CBOR data (non-array perl object)");
720		907
721	AV av = (AV )SvRV (sv);	908	AV av = (AV )SvRV (sv);
722	int len = av_len (av) + 1;	909	int len = av_len (av) + 1;
723	HV stash = gv_stashsv (av_fetch (av, 0, 1), 0);	910	HV stash = gv_stashsv (av_fetch (av, 0, 1), 0);
724		911
725	if (!stash)	912	if (!stash)
726	ERR ("cannot decode perl-object (package does not exist)");	913	ERR ("cannot decode perl-object (package does not exist)");
727		914
728	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);	915	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729		916
730	if (!method)	917	if (!method)
731	ERR ("cannot decode perl-object (package does not have a THAW method)");	918	ERR ("cannot decode perl-object (package does not have a THAW method)");
732		919
733	dSP;	920	dSP;
734		921
735	ENTER; SAVETMPS; PUSHMARK (SP);	922	ENTER; SAVETMPS; PUSHMARK (SP);
736	EXTEND (SP, len + 1);	923	EXTEND (SP, len + 1);
737	// we re-bless the reference to get overload and other niceties right	924	// we re-bless the reference to get overload and other niceties right
738	PUSHs (*av_fetch (av, 0, 1));	925	PUSHs (*av_fetch (av, 0, 1));
739	PUSHs (sv_cbor);	926	PUSHs (sv_cbor);
740		927
741	int i;	928	int i;
742		929
743	for (i = 1; i < len; ++i)	930	for (i = 1; i < len; ++i)
744	PUSHs (*av_fetch (av, i, 1));	931	PUSHs (*av_fetch (av, i, 1));
745		932
746	PUTBACK;	933	PUTBACK;
747	call_sv ((SV *)GvCV (method), G_SCALAR);	934	call_sv ((SV *)GvCV (method), G_SCALAR \| G_EVAL);
748	SPAGAIN;	935	SPAGAIN;
749		936
		937	if (SvTRUE (ERRSV))
		938	{
		939	FREETMPS; LEAVE;
		940	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
		941	}
		942
750	SvREFCNT_dec (sv);	943	SvREFCNT_dec (sv);
751	sv = SvREFCNT_inc (POPs);	944	sv = SvREFCNT_inc (POPs);
752		945
753	PUTBACK;	946	PUTBACK;
754		947
755	FREETMPS; LEAVE;	948	FREETMPS; LEAVE;
		949	}
		950	break;
756		951
757	return sv;	952	default:
758	}	953	{
759	else	954	sv = decode_sv (dec);
760	{	955
		956	dSP;
		957	ENTER; SAVETMPS; PUSHMARK (SP);
		958	EXTEND (SP, 2);
		959	PUSHs (newSVuv (tag));
		960	PUSHs (sv);
		961
		962	PUTBACK;
		963	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
		964	SPAGAIN;
		965
		966	if (SvTRUE (ERRSV))
		967	{
		968	FREETMPS; LEAVE;
		969	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
		970	}
		971
		972	if (count)
		973	{
		974	SvREFCNT_dec (sv);
		975	sv = SvREFCNT_inc (POPs);
		976	}
		977	else
		978	{
761	AV *av = newAV ();	979	AV *av = newAV ();
762	av_push (av, newSVuv (tag));	980	av_push (av, newSVuv (tag));
763	av_push (av, sv);	981	av_push (av, sv);
764		982
765	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	983	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
766	? cbor_tagged_stash	984	? cbor_tagged_stash
767	: gv_stashpv ("CBOR::XS::Tagged" , 1);	985	: gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769	return sv_bless (newRV_noinc ((SV *)av), tagged_stash);	986	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		987	}
		988
		989	PUTBACK;
		990
		991	FREETMPS; LEAVE;
		992	}
		993	break;
770	}	994	}
		995
		996	return sv;
771		997
772	fail:	998	fail:
773	SvREFCNT_dec (sv);	999	SvREFCNT_dec (sv);
774	return &PL_sv_undef;	1000	return &PL_sv_undef;
775	}	1001	}
…		…
866	}	1092	}
867		1093
868	static SV *	1094	static SV *
869	decode_cbor (SV string, CBOR cbor, char **offset_return)	1095	decode_cbor (SV string, CBOR cbor, char **offset_return)
870	{	1096	{
871	dec_t dec;	1097	dec_t dec = { };
872	SV *sv;	1098	SV *sv;
873	STRLEN len;	1099	STRLEN len;
874	char *data = SvPVbyte (string, len);	1100	char *data = SvPVbyte (string, len);
875		1101
876	if (len > cbor->max_size && cbor->max_size)	1102	if (len > cbor->max_size && cbor->max_size)
…		…
878	(unsigned long)len, (unsigned long)cbor->max_size);	1104	(unsigned long)len, (unsigned long)cbor->max_size);
879		1105
880	dec.cbor = *cbor;	1106	dec.cbor = *cbor;
881	dec.cur = (U8 *)data;	1107	dec.cur = (U8 *)data;
882	dec.end = (U8 *)data + len;	1108	dec.end = (U8 *)data + len;
883	dec.err = 0;
884	dec.depth = 0;
885		1109
886	sv = decode_sv (&dec);	1110	sv = decode_sv (&dec);
887		1111
888	if (offset_return)	1112	if (offset_return)
889	*offset_return = dec.cur;	1113	*offset_return = dec.cur;
…		…
917	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);	1141	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
918		1142
919	types_true = get_bool ("Types::Serialiser::true" );	1143	types_true = get_bool ("Types::Serialiser::true" );
920	types_false = get_bool ("Types::Serialiser::false");	1144	types_false = get_bool ("Types::Serialiser::false");
921	types_error = get_bool ("Types::Serialiser::error");	1145	types_error = get_bool ("Types::Serialiser::error");
		1146
		1147	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
922		1148
923	sv_cbor = newSVpv ("CBOR", 0);	1149	sv_cbor = newSVpv ("CBOR", 0);
924	SvREADONLY_on (sv_cbor);	1150	SvREADONLY_on (sv_cbor);
925	}	1151	}
926		1152
…		…
947		1173
948	void shrink (CBOR *self, int enable = 1)	1174	void shrink (CBOR *self, int enable = 1)
949	ALIAS:	1175	ALIAS:
950	shrink = F_SHRINK	1176	shrink = F_SHRINK
951	allow_unknown = F_ALLOW_UNKNOWN	1177	allow_unknown = F_ALLOW_UNKNOWN
		1178	allow_sharing = F_ALLOW_SHARING
		1179	pack_strings = F_PACK_STRINGS
952	PPCODE:	1180	PPCODE:
953	{	1181	{
954	if (enable)	1182	if (enable)
955	self->flags \|= ix;	1183	self->flags \|= ix;
956	else	1184	else
…		…
961		1189
962	void get_shrink (CBOR *self)	1190	void get_shrink (CBOR *self)
963	ALIAS:	1191	ALIAS:
964	get_shrink = F_SHRINK	1192	get_shrink = F_SHRINK
965	get_allow_unknown = F_ALLOW_UNKNOWN	1193	get_allow_unknown = F_ALLOW_UNKNOWN
		1194	get_allow_sharing = F_ALLOW_SHARING
		1195	get_pack_strings = F_PACK_STRINGS
966	PPCODE:	1196	PPCODE:
967	XPUSHs (boolSV (self->flags & ix));	1197	XPUSHs (boolSV (self->flags & ix));
968		1198
969	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1199	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
970	PPCODE:	1200	PPCODE:
…		…
983	XPUSHs (ST (0));	1213	XPUSHs (ST (0));
984		1214
985	int get_max_size (CBOR *self)	1215	int get_max_size (CBOR *self)
986	CODE:	1216	CODE:
987	RETVAL = self->max_size;	1217	RETVAL = self->max_size;
		1218	OUTPUT:
		1219	RETVAL
		1220
		1221	void filter (CBOR self, SV filter = 0)
		1222	PPCODE:
		1223	SvREFCNT_dec (self->filter);
		1224	self->filter = filter ? newSVsv (filter) : filter;
		1225	XPUSHs (ST (0));
		1226
		1227	SV get_filter (CBOR self)
		1228	CODE:
		1229	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
988	OUTPUT:	1230	OUTPUT:
989	RETVAL	1231	RETVAL
990		1232
991	void encode (CBOR self, SV scalar)	1233	void encode (CBOR self, SV scalar)
992	PPCODE:	1234	PPCODE:
…		…
1007	EXTEND (SP, 2);	1249	EXTEND (SP, 2);
1008	PUSHs (sv);	1250	PUSHs (sv);
1009	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1251	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1010	}	1252	}
1011		1253
		1254	void DESTROY (CBOR *self)
		1255	PPCODE:
		1256	cbor_free (self);
		1257
1012	PROTOTYPES: ENABLE	1258	PROTOTYPES: ENABLE
1013		1259
1014	void encode_cbor (SV *scalar)	1260	void encode_cbor (SV *scalar)
1015	PPCODE:	1261	PPCODE:
1016	{	1262	{

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.16 by root, Tue Oct 29 20:59:16 2013 UTC vs. Revision 1.34 by root, Sat Nov 30 16:19:59 2013 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.16 by root, Tue Oct 29 20:59:16 2013 UTC vs.
Revision 1.34 by root, Sat Nov 30 16:19:59 2013 UTC