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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.23 by root, Wed Nov 20 16:19:47 2013 UTC

…		…
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
11		11
12	#include "ecb.h"	12	#include "ecb.h"
13		13
		14	// compatibility with perl <5.18
		15	#ifndef HvNAMELEN_get
		16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
		17	#endif
		18	#ifndef HvNAMELEN
		19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
		20	#endif
		21	#ifndef HvNAMEUTF8
		22	# define HvNAMEUTF8(hv) 0
		23	#endif
		24
		25	// known tags
		26	enum cbor_tag
		27	{
		28	// inofficial extensions (pending iana registration)
		29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object
		30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object
		31	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing
		32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing
		33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref
		34	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
		35	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
		36
		37	// rfc7049
		38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
		39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
		40	CBOR_TAG_POS_BIGNUM = 2, // byte string
		41	CBOR_TAG_NEG_BIGNUM = 3, // byte string
		42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
		43	CBOR_TAG_BIGFLOAT = 5, // array
		44
		45	CBOR_TAG_CONV_B64U = 21, // base64url, any
		46	CBOR_TAG_CONV_B64 = 22, // base64, any
		47	CBOR_TAG_CONV_HEX = 23, // base16, any
		48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
		49
		50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
		51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
		52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
		53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
		54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
		55
		56	CBOR_TAG_MAGIC = 55799 // self-describe cbor
		57	};
		58
14	#define F_SHRINK 0x00000200UL	59	#define F_SHRINK 0x00000001UL
15	#define F_ALLOW_UNKNOWN 0x00002000UL	60	#define F_ALLOW_UNKNOWN 0x00000002UL
		61	#define F_ALLOW_SHARING 0x00000004UL //TODO
		62	#define F_DEDUP_STRINGS 0x00000008UL //TODO
16		63
17	#define INIT_SIZE 32 // initial scalar size to be allocated	64	#define INIT_SIZE 32 // initial scalar size to be allocated
18		65
19	#define SB do {	66	#define SB do {
20	#define SE } while (0)	67	#define SE } while (0)
…		…
31	#else	78	#else
32	# define CBOR_SLOW 0	79	# define CBOR_SLOW 0
33	# define CBOR_STASH cbor_stash	80	# define CBOR_STASH cbor_stash
34	#endif	81	#endif
35		82
36	static HV cbor_stash, cbor_boolean_stash; // CBOR::XS::	83	static HV cbor_stash, types_boolean_stash, types_error_stash, cbor_tagged_stash; // CBOR::XS::
37	static SV cbor_true, cbor_false;	84	static SV types_true, types_false, types_error, sv_cbor;
38		85
39	typedef struct {	86	typedef struct {
40	U32 flags;	87	U32 flags;
41	U32 max_depth;	88	U32 max_depth;
42	STRLEN max_size;	89	STRLEN max_size;
43
44	SV *cb_object;
45	HV *cb_sk_object;
46	} CBOR;	90	} CBOR;
47		91
48	ecb_inline void	92	ecb_inline void
49	cbor_init (CBOR *cbor)	93	cbor_init (CBOR *cbor)
50	{	94	{
…		…
79	SvPV_renew (sv, SvCUR (sv) + 1);	123	SvPV_renew (sv, SvCUR (sv) + 1);
80	#endif	124	#endif
81	}	125	}
82	}	126	}
83		127
84	/////////////////////////////////////////////////////////////////////////////	128	// minimum length of a string to be registered for stringref
85	// fp hell	129	ecb_inline int
86		130	minimum_string_length (UV idx)
87	//TODO	131	{
		132	return idx > 23
		133	? idx > 0xffU
		134	? idx > 0xffffU
		135	? idx > 0xffffffffU
		136	? 7
		137	: 6
		138	: 5
		139	: 4
		140	: 3;
		141	}
88		142
89	/////////////////////////////////////////////////////////////////////////////	143	/////////////////////////////////////////////////////////////////////////////
90	// encoder	144	// encoder
91		145
92	// structure used for encoding CBOR	146	// structure used for encoding CBOR
…		…
95	char *cur; // SvPVX (sv) + current output position	149	char *cur; // SvPVX (sv) + current output position
96	char *end; // SvEND (sv)	150	char *end; // SvEND (sv)
97	SV *sv; // result scalar	151	SV *sv; // result scalar
98	CBOR cbor;	152	CBOR cbor;
99	U32 depth; // recursion level	153	U32 depth; // recursion level
		154	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		155	UV stringref_idx;
		156	HV *shareable; // ptr => index, or 0
		157	UV shareable_idx;
100	} enc_t;	158	} enc_t;
101		159
102	ecb_inline void	160	ecb_inline void
103	need (enc_t *enc, STRLEN len)	161	need (enc_t *enc, STRLEN len)
104	{	162	{
…		…
156	*enc->cur++ = len >> 8;	214	*enc->cur++ = len >> 8;
157	*enc->cur++ = len;	215	*enc->cur++ = len;
158	}	216	}
159	}	217	}
160		218
		219	ecb_inline void
		220	encode_tag (enc_t *enc, UV tag)
		221	{
		222	encode_uint (enc, 0xc0, tag);
		223	}
		224
161	static void	225	static void
162	encode_str (enc_t enc, int utf8, char str, STRLEN len)	226	encode_str (enc_t enc, int utf8, char str, STRLEN len)
163	{	227	{
		228	if (ecb_expect_false (enc->cbor.flags & F_DEDUP_STRINGS))
		229	{
		230	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		231
		232	if (SvOK (*svp))
		233	{
		234	// already registered, use stringref
		235	encode_tag (enc, CBOR_TAG_STRINGREF);
		236	encode_uint (enc, 0x00, SvUV (*svp));
		237	return;
		238	}
		239	else if (len >= minimum_string_length (enc->stringref_idx))
		240	{
		241	// register only
		242	sv_setuv (*svp, enc->stringref_idx);
		243	++enc->stringref_idx;
		244	}
		245	}
		246
164	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	247	encode_uint (enc, utf8 ? 0x60 : 0x40, len);
165	need (enc, len);	248	need (enc, len);
166	memcpy (enc->cur, str, len);	249	memcpy (enc->cur, str, len);
167	enc->cur += len;	250	enc->cur += len;
168	}	251	}
…		…
188	}	271	}
189		272
190	--enc->depth;	273	--enc->depth;
191	}	274	}
192		275
		276	ecb_inline void
		277	encode_he (enc_t enc, HE he)
		278	{
		279	}
		280
193	static void	281	static void
194	encode_hv (enc_t enc, HV hv)	282	encode_hv (enc_t enc, HV hv)
195	{	283	{
196	HE *he;	284	HE *he;
197		285
…		…
226		314
227	// encode objects, arrays and special \0=false and \1=true values.	315	// encode objects, arrays and special \0=false and \1=true values.
228	static void	316	static void
229	encode_rv (enc_t enc, SV sv)	317	encode_rv (enc_t enc, SV sv)
230	{	318	{
231	svtype svt;
232
233	SvGETMAGIC (sv);	319	SvGETMAGIC (sv);
		320
		321	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
		322	&& ecb_expect_false (SvREFCNT (sv) > 1))
		323	{
		324	if (!enc->shareable)
		325	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		326
		327	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		328
		329	if (SvOK (*svp))
		330	{
		331	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		332	encode_uint (enc, 0x00, SvUV (*svp));
		333	return;
		334	}
		335	else
		336	{
		337	sv_setuv (*svp, enc->shareable_idx);
		338	++enc->shareable_idx;
		339	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		340	}
		341	}
		342
234	svt = SvTYPE (sv);	343	svtype svt = SvTYPE (sv);
235		344
236	if (ecb_expect_false (SvOBJECT (sv)))	345	if (ecb_expect_false (SvOBJECT (sv)))
237	{	346	{
238	HV *stash = !CBOR_SLOW \|\| cbor_boolean_stash	347	HV *boolean_stash = !CBOR_SLOW \|\| types_boolean_stash
239	? cbor_boolean_stash	348	? types_boolean_stash
		349	: gv_stashpv ("Types::Serialiser::Boolean", 1);
		350	HV *error_stash = !CBOR_SLOW \|\| types_error_stash
		351	? types_error_stash
		352	: gv_stashpv ("Types::Serialiser::Error", 1);
		353	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
		354	? cbor_tagged_stash
240	: gv_stashpv ("CBOR::XS::Boolean", 1);	355	: gv_stashpv ("CBOR::XS::Tagged" , 1);
241		356
242	if (SvSTASH (sv) == stash)	357	HV *stash = SvSTASH (sv);
		358	GV *method;
		359
		360	if (stash == boolean_stash)
243	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	361	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);
		362	else if (stash == error_stash)
		363	encode_ch (enc, 0xe0 \| 23);
		364	else if (stash == tagged_stash)
		365	{
		366	if (svt != SVt_PVAV)
		367	croak ("encountered CBOR::XS::Tagged object that isn't an array");
		368
		369	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));
		370	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		371	}
		372	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
		373	{
		374	dSP;
		375
		376	ENTER; SAVETMPS; PUSHMARK (SP);
		377	// we re-bless the reference to get overload and other niceties right
		378	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
		379
		380	PUTBACK;
		381	// G_SCALAR ensures that return value is 1
		382	call_sv ((SV *)GvCV (method), G_SCALAR);
		383	SPAGAIN;
		384
		385	// catch this surprisingly common error
		386	if (SvROK (TOPs) && SvRV (TOPs) == sv)
		387	croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
		388
		389	encode_sv (enc, POPs);
		390
		391	PUTBACK;
		392
		393	FREETMPS; LEAVE;
		394	}
		395	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
		396	{
		397	dSP;
		398
		399	ENTER; SAVETMPS; PUSHMARK (SP);
		400	EXTEND (SP, 2);
		401	// we re-bless the reference to get overload and other niceties right
		402	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
		403	PUSHs (sv_cbor);
		404
		405	PUTBACK;
		406	int count = call_sv ((SV *)GvCV (method), G_ARRAY);
		407	SPAGAIN;
		408
		409	// catch this surprisingly common error
		410	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
		411	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
		412
		413	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
		414	encode_uint (enc, 0x80, count + 1);
		415	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
		416
		417	while (count)
		418	encode_sv (enc, SP[1 - count--]);
		419
		420	PUTBACK;
		421
		422	FREETMPS; LEAVE;
		423	}
244	else	424	else
245	{
246	#if 0 //TODO
247	if (enc->cbor.flags & F_CONV_BLESSED)
248	{
249	// we re-bless the reference to get overload and other niceties right
250	GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252	if (to_cbor)
253	{
254	dSP;
255
256	ENTER; SAVETMPS; PUSHMARK (SP);
257	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
258
259	// calling with G_SCALAR ensures that we always get a 1 return value
260	PUTBACK;
261	call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
262	SPAGAIN;
263
264	// catch this surprisingly common error
265	if (SvROK (TOPs) && SvRV (TOPs) == sv)
266	croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
267
268	sv = POPs;
269	PUTBACK;
270
271	encode_sv (enc, sv);
272
273	FREETMPS; LEAVE;
274	}
275	else if (enc->cbor.flags & F_ALLOW_BLESSED)
276	encode_str (enc, "null", 4, 0);
277	else
278	croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",	425	croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
279	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280	}
281	else if (enc->cbor.flags & F_ALLOW_BLESSED)
282	encode_str (enc, "null", 4, 0);
283	else
284	croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
285	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	426	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286	#endif
287	}
288	}	427	}
289	else if (svt == SVt_PVHV)	428	else if (svt == SVt_PVHV)
290	encode_hv (enc, (HV *)sv);	429	encode_hv (enc, (HV *)sv);
291	else if (svt == SVt_PVAV)	430	else if (svt == SVt_PVAV)
292	encode_av (enc, (AV *)sv);	431	encode_av (enc, (AV *)sv);
293	else if (svt < SVt_PVAV)
294	{
295	STRLEN len = 0;
296	char *pv = svt ? SvPV (sv, len) : 0;
297
298	if (len == 1 && *pv == '1')
299	encode_ch (enc, 0xe0 \| 21);
300	else if (len == 1 && *pv == '0')
301	encode_ch (enc, 0xe0 \| 20);
302	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303	encode_ch (enc, 0xe0 \| 23);
304	else
305	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307	}
308	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309	encode_ch (enc, 0xe0 \| 23);
310	else	432	else
311	croak ("encountered %s, but CBOR can only represent references to arrays or hashes",	433	{
312	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	434	encode_tag (enc, CBOR_TAG_INDIRECTION);
		435	encode_sv (enc, sv);
		436	}
313	}	437	}
314		438
315	static void	439	static void
316	encode_nv (enc_t enc, SV sv)	440	encode_nv (enc_t enc, SV sv)
317	{	441	{
…		…
382	}	506	}
383		507
384	static SV *	508	static SV *
385	encode_cbor (SV scalar, CBOR cbor)	509	encode_cbor (SV scalar, CBOR cbor)
386	{	510	{
387	enc_t enc;	511	enc_t enc = { };
388		512
389	enc.cbor = *cbor;	513	enc.cbor = *cbor;
390	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	514	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391	enc.cur = SvPVX (enc.sv);	515	enc.cur = SvPVX (enc.sv);
392	enc.end = SvEND (enc.sv);	516	enc.end = SvEND (enc.sv);
393	enc.depth = 0;
394		517
395	SvPOK_only (enc.sv);	518	SvPOK_only (enc.sv);
		519
		520	if (cbor->flags & F_DEDUP_STRINGS)
		521	{
		522	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		523	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
		524	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
		525	}
		526
396	encode_sv (&enc, scalar);	527	encode_sv (&enc, scalar);
397		528
398	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	529	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	530	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400		531
…		…
414	U8 *end; // end of input string	545	U8 *end; // end of input string
415	const char *err; // parse error, if != 0	546	const char *err; // parse error, if != 0
416	CBOR cbor;	547	CBOR cbor;
417	U32 depth; // recursion depth	548	U32 depth; // recursion depth
418	U32 maxdepth; // recursion depth limit	549	U32 maxdepth; // recursion depth limit
		550	AV *shareable;
		551	AV *stringref;
419	} dec_t;	552	} dec_t;
420		553
421	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	554	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422		555
423	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	556	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
…		…
517	SvREFCNT_dec (av);	650	SvREFCNT_dec (av);
518	DEC_DEC_DEPTH;	651	DEC_DEC_DEPTH;
519	return &PL_sv_undef;	652	return &PL_sv_undef;
520	}	653	}
521		654
		655	static void
		656	decode_he (dec_t dec, HV hv)
		657	{
		658	// for speed reasons, we specialcase single-string
		659	// byte or utf-8 strings as keys, but only when !stringref
		660
		661	if (ecb_expect_true (!dec->stringref))
		662	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)
		663	{
		664	I32 len = decode_uint (dec);
		665	char key = (char )dec->cur;
		666
		667	dec->cur += len;
		668
		669	if (ecb_expect_false (dec->stringref))
		670	av_push (dec->stringref, newSVpvn (key, len));
		671
		672	hv_store (hv, key, len, decode_sv (dec), 0);
		673
		674	return;
		675	}
		676	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)
		677	{
		678	I32 len = decode_uint (dec);
		679	char key = (char )dec->cur;
		680
		681	dec->cur += len;
		682
		683	if (ecb_expect_false (dec->stringref))
		684	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
		685
		686	hv_store (hv, key, -len, decode_sv (dec), 0);
		687
		688	return;
		689	}
		690
		691	SV *k = decode_sv (dec);
		692	SV *v = decode_sv (dec);
		693
		694	hv_store_ent (hv, k, v, 0);
		695	SvREFCNT_dec (k);
		696	}
		697
522	static SV *	698	static SV *
523	decode_hv (dec_t *dec)	699	decode_hv (dec_t *dec)
524	{	700	{
525	HV *hv = newHV ();	701	HV *hv = newHV ();
526		702
…		…
538	{	714	{
539	++dec->cur;	715	++dec->cur;
540	break;	716	break;
541	}	717	}
542		718
543	SV *k = decode_sv (dec);	719	decode_he (dec, hv);
544	SV *v = decode_sv (dec);
545
546	hv_store_ent (hv, k, v, 0);
547	}	720	}
548	}	721	}
549	else	722	else
550	{	723	{
551	int len = decode_uint (dec);	724	int pairs = decode_uint (dec);
552		725
553	while (len--)	726	while (pairs--)
554	{	727	decode_he (dec, hv);
555	SV *k = decode_sv (dec);
556	SV *v = decode_sv (dec);
557
558	hv_store_ent (hv, k, v, 0);
559	}
560	}	728	}
561		729
562	DEC_DEC_DEPTH;	730	DEC_DEC_DEPTH;
563	return newRV_noinc ((SV *)hv);	731	return newRV_noinc ((SV *)hv);
564
565	#if 0
566	SV *sv;
567	HV *hv = newHV ();
568
569	DEC_INC_DEPTH;
570	decode_ws (dec);
571
572	for (;;)
573	{
574	// heuristic: assume that
575	// a) decode_str + hv_store_ent are abysmally slow.
576	// b) most hash keys are short, simple ascii text.
577	// => try to "fast-match" such strings to avoid
578	// the overhead of decode_str + hv_store_ent.
579	{
580	SV *value;
581	char *p = dec->cur;
582	char *e = p + 24; // only try up to 24 bytes
583
584	for (;;)
585	{
586	// the >= 0x80 is false on most architectures
587	if (p == e \|\| p < 0x20 \|\| p >= 0x80 \|\| *p == '\\')
588	{
589	// slow path, back up and use decode_str
590	SV *key = decode_str (dec);
591	if (!key)
592	goto fail;
593
594	decode_ws (dec); EXPECT_CH (':');
595
596	decode_ws (dec);
597	value = decode_sv (dec);
598	if (!value)
599	{
600	SvREFCNT_dec (key);
601	goto fail;
602	}
603
604	hv_store_ent (hv, key, value, 0);
605	SvREFCNT_dec (key);
606
607	break;
608	}
609	else if (*p == '"')
610	{
611	// fast path, got a simple key
612	char *key = dec->cur;
613	int len = p - key;
614	dec->cur = p + 1;
615
616	decode_ws (dec); EXPECT_CH (':');
617
618	decode_ws (dec);
619	value = decode_sv (dec);
620	if (!value)
621	goto fail;
622
623	hv_store (hv, key, len, value, 0);
624
625	break;
626	}
627
628	++p;
629	}
630	}
631
632	decode_ws (dec);
633
634	if (*dec->cur == '}')
635	{
636	++dec->cur;
637	break;
638	}
639
640	if (*dec->cur != ',')
641	ERR (", or } expected while parsing object/hash");
642
643	++dec->cur;
644
645	decode_ws (dec);
646
647	if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648	{
649	++dec->cur;
650	break;
651	}
652	}
653
654	DEC_DEC_DEPTH;
655	sv = newRV_noinc ((SV *)hv);
656
657	// check filter callbacks
658	if (dec->cbor.flags & F_HOOK)
659	{
660	if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661	{
662	HE cb, he;
663
664	hv_iterinit (hv);
665	he = hv_iternext (hv);
666	hv_iterinit (hv);
667
668	// the next line creates a mortal sv each time its called.
669	// might want to optimise this for common cases.
670	cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672	if (cb)
673	{
674	dSP;
675	int count;
676
677	ENTER; SAVETMPS; PUSHMARK (SP);
678	XPUSHs (HeVAL (he));
679	sv_2mortal (sv);
680
681	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683	if (count == 1)
684	{
685	sv = newSVsv (POPs);
686	FREETMPS; LEAVE;
687	return sv;
688	}
689
690	SvREFCNT_inc (sv);
691	FREETMPS; LEAVE;
692	}
693	}
694
695	if (dec->cbor.cb_object)
696	{
697	dSP;
698	int count;
699
700	ENTER; SAVETMPS; PUSHMARK (SP);
701	XPUSHs (sv_2mortal (sv));
702
703	PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705	if (count == 1)
706	{
707	sv = newSVsv (POPs);
708	FREETMPS; LEAVE;
709	return sv;
710	}
711
712	SvREFCNT_inc (sv);
713	FREETMPS; LEAVE;
714	}
715	}
716
717	return sv;
718	#endif
719		732
720	fail:	733	fail:
721	SvREFCNT_dec (hv);	734	SvREFCNT_dec (hv);
722	DEC_DEC_DEPTH;	735	DEC_DEC_DEPTH;
723	return &PL_sv_undef;	736	return &PL_sv_undef;
724	}	737	}
725		738
726	static SV *	739	static SV *
727	decode_str (dec_t *dec, int utf8)	740	decode_str (dec_t *dec, int utf8)
728	{	741	{
729	SV *sv;	742	SV *sv = 0;
730		743
731	if ((*dec->cur & 31) == 31)	744	if ((*dec->cur & 31) == 31)
732	{	745	{
733	++dec->cur;	746	++dec->cur;
734		747
…		…
743	{	756	{
744	++dec->cur;	757	++dec->cur;
745	break;	758	break;
746	}	759	}
747		760
748	SV *sv2 = decode_sv (dec);
749	sv_catsv (sv, sv2);	761	sv_catsv (sv, decode_sv (dec));
750	}	762	}
751	}	763	}
752	else	764	else
753	{	765	{
754	STRLEN len = decode_uint (dec);	766	STRLEN len = decode_uint (dec);
…		…
759	}	771	}
760		772
761	if (utf8)	773	if (utf8)
762	SvUTF8_on (sv);	774	SvUTF8_on (sv);
763		775
		776	if (ecb_expect_false (dec->stringref)
		777	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		778	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
		779
764	return sv;	780	return sv;
765		781
766	fail:	782	fail:
		783	SvREFCNT_dec (sv);
767	return &PL_sv_undef;	784	return &PL_sv_undef;
768	}	785	}
769		786
770	static SV *	787	static SV *
771	decode_tagged (dec_t *dec)	788	decode_tagged (dec_t *dec)
772	{	789	{
		790	SV *sv = 0;
773	UV tag = decode_uint (dec);	791	UV tag = decode_uint (dec);
		792
		793	WANT (1);
		794
		795	switch (tag)
		796	{
		797	case CBOR_TAG_MAGIC:
		798	sv = decode_sv (dec);
		799	break;
		800
		801	case CBOR_TAG_INDIRECTION:
		802	sv = newRV_noinc (decode_sv (dec));
		803	break;
		804
		805	case CBOR_TAG_STRINGREF_NAMESPACE:
		806	{
		807	ENTER; SAVETMPS;
		808
		809	SAVESPTR (dec->stringref);
		810	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
		811
		812	sv = decode_sv (dec);
		813
		814	FREETMPS; LEAVE;
		815	}
		816	break;
		817
		818	case CBOR_TAG_STRINGREF:
		819	{
		820	if ((*dec->cur >> 5) != 0)
		821	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		822
		823	UV idx = decode_uint (dec);
		824
		825	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
		826	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		827
		828	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		829	}
		830	break;
		831
		832	case CBOR_TAG_VALUE_SHAREABLE:
		833	{
		834	if (ecb_expect_false (!dec->shareable))
		835	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
		836
		837	sv = newSV (0);
		838	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
		839
774	SV *sv = decode_sv (dec);	840	SV *osv = decode_sv (dec);
		841	sv_setsv (sv, osv);
		842	SvREFCNT_dec_NN (osv);
		843	}
		844	break;
775		845
776	if (tag == 55799) // 2.4.5 Self-Describe CBOR	846	case CBOR_TAG_VALUE_SHAREDREF:
		847	{
		848	if ((*dec->cur >> 5) != 0)
		849	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
		850
		851	UV idx = decode_uint (dec);
		852
		853	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
		854	ERR ("corrupted CBOR data (sharedref index out of bounds)");
		855
		856	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		857	}
		858	break;
		859
		860	case CBOR_TAG_PERL_OBJECT:
		861	{
		862	sv = decode_sv (dec);
		863
		864	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)
		865	ERR ("corrupted CBOR data (non-array perl object)");
		866
		867	AV av = (AV )SvRV (sv);
		868	int len = av_len (av) + 1;
		869	HV stash = gv_stashsv (av_fetch (av, 0, 1), 0);
		870
		871	if (!stash)
		872	ERR ("cannot decode perl-object (package does not exist)");
		873
		874	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
		875
		876	if (!method)
		877	ERR ("cannot decode perl-object (package does not have a THAW method)");
		878
		879	dSP;
		880
		881	ENTER; SAVETMPS; PUSHMARK (SP);
		882	EXTEND (SP, len + 1);
		883	// we re-bless the reference to get overload and other niceties right
		884	PUSHs (*av_fetch (av, 0, 1));
		885	PUSHs (sv_cbor);
		886
		887	int i;
		888
		889	for (i = 1; i < len; ++i)
		890	PUSHs (*av_fetch (av, i, 1));
		891
		892	PUTBACK;
		893	call_sv ((SV *)GvCV (method), G_SCALAR \| G_EVAL);
		894	SPAGAIN;
		895
		896	if (SvTRUE (ERRSV))
		897	{
		898	FREETMPS; LEAVE;
		899	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
		900	}
		901
		902	SvREFCNT_dec (sv);
		903	sv = SvREFCNT_inc (POPs);
		904
		905	PUTBACK;
		906
		907	FREETMPS; LEAVE;
		908	}
		909	break;
		910
		911	default:
		912	{
		913	sv = decode_sv (dec);
		914
		915	AV *av = newAV ();
		916	av_push (av, newSVuv (tag));
		917	av_push (av, sv);
		918
		919	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
		920	? cbor_tagged_stash
		921	: gv_stashpv ("CBOR::XS::Tagged" , 1);
		922
		923	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		924	}
		925	break;
		926	}
		927
777	return sv;	928	return sv;
778		929
779	AV *av = newAV ();	930	fail:
780	av_push (av, newSVuv (tag));	931	SvREFCNT_dec (sv);
781	av_push (av, sv);	932	return &PL_sv_undef;
782	return newRV_noinc ((SV *)av);
783	}	933	}
784		934
785	static SV *	935	static SV *
786	decode_sv (dec_t *dec)	936	decode_sv (dec_t *dec)
787	{	937	{
788	WANT (1);	938	WANT (1);
789		939
790	switch (*dec->cur >> 5)	940	switch (*dec->cur >> 5)
791	{	941	{
792	case 0: // unsigned int	942	case 0: // unsigned int
793	//TODO: 64 bit values on 3 2bit perls
794	return newSVuv (decode_uint (dec));	943	return newSVuv (decode_uint (dec));
795	case 1: // negative int	944	case 1: // negative int
796	return newSViv (-1 - (IV)decode_uint (dec));	945	return newSViv (-1 - (IV)decode_uint (dec));
797	case 2: // octet string	946	case 2: // octet string
798	return decode_str (dec, 0);	947	return decode_str (dec, 0);
…		…
807	case 7: // misc	956	case 7: // misc
808	switch (*dec->cur++ & 31)	957	switch (*dec->cur++ & 31)
809	{	958	{
810	case 20:	959	case 20:
811	#if CBOR_SLOW	960	#if CBOR_SLOW
812	cbor_false = get_bool ("CBOR::XS::false");	961	types_false = get_bool ("Types::Serialiser::false");
813	#endif	962	#endif
814	return newSVsv (cbor_false);	963	return newSVsv (types_false);
815	case 21:	964	case 21:
816	#if CBOR_SLOW	965	#if CBOR_SLOW
817	cbor_true = get_bool ("CBOR::XS::true");	966	types_true = get_bool ("Types::Serialiser::true");
818	#endif	967	#endif
819	return newSVsv (cbor_true);	968	return newSVsv (types_true);
820	case 22:	969	case 22:
821	return newSVsv (&PL_sv_undef);	970	return newSVsv (&PL_sv_undef);
		971	case 23:
		972	#if CBOR_SLOW
		973	types_error = get_bool ("Types::Serialiser::error");
		974	#endif
		975	return newSVsv (types_error);
822		976
823	case 25:	977	case 25:
824	{	978	{
825	WANT (2);	979	WANT (2);
826		980
…		…
862	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1016	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863	}	1017	}
864		1018
865	break;	1019	break;
866	}	1020	}
867	#if 0
868	switch (*dec->cur)
869	{
870	//case '"': ++dec->cur; return decode_str (dec);
871	case '[': ++dec->cur; return decode_av (dec);
872	case '{': ++dec->cur; return decode_hv (dec);
873
874	case '-':
875	case '0': case '1': case '2': case '3': case '4':
876	case '5': case '6': case '7': case '8': case '9':
877	//TODO return decode_num (dec);
878
879	case 't':
880	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881	{
882	dec->cur += 4;
883	#if CBOR_SLOW
884	cbor_true = get_bool ("CBOR::XS::true");
885	#endif
886	return newSVsv (cbor_true);
887	}
888	else
889	ERR ("'true' expected");
890
891	break;
892
893	case 'f':
894	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895	{
896	dec->cur += 5;
897	#if CBOR_SLOW
898	cbor_false = get_bool ("CBOR::XS::false");
899	#endif
900	return newSVsv (cbor_false);
901	}
902	else
903	ERR ("'false' expected");
904
905	break;
906
907	case 'n':
908	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909	{
910	dec->cur += 4;
911	return newSVsv (&PL_sv_undef);
912	}
913	else
914	ERR ("'null' expected");
915
916	break;
917
918	default:
919	ERR ("malformed CBOR string, neither array, object, number, string or atom");
920	break;
921	}
922	#endif
923		1021
924	fail:	1022	fail:
925	return &PL_sv_undef;	1023	return &PL_sv_undef;
926	}	1024	}
927		1025
928	static SV *	1026	static SV *
929	decode_cbor (SV string, CBOR cbor, char **offset_return)	1027	decode_cbor (SV string, CBOR cbor, char **offset_return)
930	{	1028	{
931	dec_t dec;	1029	dec_t dec = { };
932	SV *sv;	1030	SV *sv;
		1031	STRLEN len;
		1032	char *data = SvPVbyte (string, len);
933		1033
934	/* work around bugs in 5.10 where manipulating magic values
935	* makes perl ignore the magic in subsequent accesses.
936	* also make a copy of non-PV values, to get them into a clean
937	* state (SvPV should do that, but it's buggy, see below).
938	*/
939	/SvGETMAGIC (string);/
940	if (SvMAGICAL (string) \|\| !SvPOK (string))
941	string = sv_2mortal (newSVsv (string));
942
943	SvUPGRADE (string, SVt_PV);
944
945	/* work around a bug in perl 5.10, which causes SvCUR to fail an
946	* assertion with -DDEBUGGING, although SvCUR is documented to
947	* return the xpv_cur field which certainly exists after upgrading.
948	* according to nicholas clark, calling SvPOK fixes this.
949	* But it doesn't fix it, so try another workaround, call SvPV_nolen
950	* and hope for the best.
951	* Damnit, SvPV_nolen still trips over yet another assertion. This
952	* assertion business is seriously broken, try yet another workaround
953	* for the broken -DDEBUGGING.
954	*/
955	{
956	#ifdef DEBUGGING
957	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958	#else
959	STRLEN offset = SvCUR (string);
960	#endif
961
962	if (offset > cbor->max_size && cbor->max_size)	1034	if (len > cbor->max_size && cbor->max_size)
963	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",	1035	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964	(unsigned long)SvCUR (string), (unsigned long)cbor->max_size);	1036	(unsigned long)len, (unsigned long)cbor->max_size);
965	}
966
967	sv_utf8_downgrade (string, 0);
968		1037
969	dec.cbor = *cbor;	1038	dec.cbor = *cbor;
970	dec.cur = (U8 *)SvPVX (string);	1039	dec.cur = (U8 *)data;
971	dec.end = (U8 *)SvEND (string);	1040	dec.end = (U8 *)data + len;
972	dec.err = 0;
973	dec.depth = 0;
974
975	if (dec.cbor.cb_object \|\| dec.cbor.cb_sk_object)
976	;//TODO dec.cbor.flags \|= F_HOOK;
977		1041
978	sv = decode_sv (&dec);	1042	sv = decode_sv (&dec);
979		1043
980	if (offset_return)	1044	if (offset_return)
981	*offset_return = dec.cur;	1045	*offset_return = dec.cur;
…		…
985	dec.err = "garbage after CBOR object";	1049	dec.err = "garbage after CBOR object";
986		1050
987	if (dec.err)	1051	if (dec.err)
988	{	1052	{
989	SvREFCNT_dec (sv);	1053	SvREFCNT_dec (sv);
990	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )SvPVX (string), (int)(uint8_t)dec.cur);	1054	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
991	}	1055	}
992		1056
993	sv = sv_2mortal (sv);	1057	sv = sv_2mortal (sv);
994		1058
995	return sv;	1059	return sv;
…		…
1001	MODULE = CBOR::XS PACKAGE = CBOR::XS	1065	MODULE = CBOR::XS PACKAGE = CBOR::XS
1002		1066
1003	BOOT:	1067	BOOT:
1004	{	1068	{
1005	cbor_stash = gv_stashpv ("CBOR::XS" , 1);	1069	cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006	cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);	1070	cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007		1071
1008	cbor_true = get_bool ("CBOR::XS::true");	1072	types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009	cbor_false = get_bool ("CBOR::XS::false");	1073	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
		1074
		1075	types_true = get_bool ("Types::Serialiser::true" );
		1076	types_false = get_bool ("Types::Serialiser::false");
		1077	types_error = get_bool ("Types::Serialiser::error");
		1078
		1079	sv_cbor = newSVpv ("CBOR", 0);
		1080	SvREADONLY_on (sv_cbor);
1010	}	1081	}
1011		1082
1012	PROTOTYPES: DISABLE	1083	PROTOTYPES: DISABLE
1013		1084
1014	void CLONE (...)	1085	void CLONE (...)
1015	CODE:	1086	CODE:
1016	cbor_stash = 0;	1087	cbor_stash = 0;
		1088	cbor_tagged_stash = 0;
		1089	types_error_stash = 0;
1017	cbor_boolean_stash = 0;	1090	types_boolean_stash = 0;
1018		1091
1019	void new (char *klass)	1092	void new (char *klass)
1020	PPCODE:	1093	PPCODE:
1021	{	1094	{
1022	SV *pv = NEWSV (0, sizeof (CBOR));	1095	SV *pv = NEWSV (0, sizeof (CBOR));
…		…
1030		1103
1031	void shrink (CBOR *self, int enable = 1)	1104	void shrink (CBOR *self, int enable = 1)
1032	ALIAS:	1105	ALIAS:
1033	shrink = F_SHRINK	1106	shrink = F_SHRINK
1034	allow_unknown = F_ALLOW_UNKNOWN	1107	allow_unknown = F_ALLOW_UNKNOWN
		1108	allow_sharing = F_ALLOW_SHARING
		1109	dedup_strings = F_DEDUP_STRINGS
1035	PPCODE:	1110	PPCODE:
1036	{	1111	{
1037	if (enable)	1112	if (enable)
1038	self->flags \|= ix;	1113	self->flags \|= ix;
1039	else	1114	else
…		…
1044		1119
1045	void get_shrink (CBOR *self)	1120	void get_shrink (CBOR *self)
1046	ALIAS:	1121	ALIAS:
1047	get_shrink = F_SHRINK	1122	get_shrink = F_SHRINK
1048	get_allow_unknown = F_ALLOW_UNKNOWN	1123	get_allow_unknown = F_ALLOW_UNKNOWN
		1124	get_allow_sharing = F_ALLOW_SHARING
		1125	get_dedup_strings = F_DEDUP_STRINGS
1049	PPCODE:	1126	PPCODE:
1050	XPUSHs (boolSV (self->flags & ix));	1127	XPUSHs (boolSV (self->flags & ix));
1051		1128
1052	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1129	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053	PPCODE:	1130	PPCODE:
…		…
1068	int get_max_size (CBOR *self)	1145	int get_max_size (CBOR *self)
1069	CODE:	1146	CODE:
1070	RETVAL = self->max_size;	1147	RETVAL = self->max_size;
1071	OUTPUT:	1148	OUTPUT:
1072	RETVAL	1149	RETVAL
1073
1074	#if 0 //TODO
1075
1076	void filter_cbor_object (CBOR self, SV cb = &PL_sv_undef)
1077	PPCODE:
1078	{
1079	SvREFCNT_dec (self->cb_object);
1080	self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1081
1082	XPUSHs (ST (0));
1083	}
1084
1085	void filter_cbor_single_key_object (CBOR self, SV key, SV *cb = &PL_sv_undef)
1086	PPCODE:
1087	{
1088	if (!self->cb_sk_object)
1089	self->cb_sk_object = newHV ();
1090
1091	if (SvOK (cb))
1092	hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093	else
1094	{
1095	hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097	if (!HvKEYS (self->cb_sk_object))
1098	{
1099	SvREFCNT_dec (self->cb_sk_object);
1100	self->cb_sk_object = 0;
1101	}
1102	}
1103
1104	XPUSHs (ST (0));
1105	}
1106
1107	#endif
1108		1150
1109	void encode (CBOR self, SV scalar)	1151	void encode (CBOR self, SV scalar)
1110	PPCODE:	1152	PPCODE:
1111	PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;	1153	PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112	XPUSHs (scalar);	1154	XPUSHs (scalar);
…		…
1125	EXTEND (SP, 2);	1167	EXTEND (SP, 2);
1126	PUSHs (sv);	1168	PUSHs (sv);
1127	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1169	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128	}	1170	}
1129		1171
1130	void DESTROY (CBOR *self)
1131	CODE:
1132	SvREFCNT_dec (self->cb_sk_object);
1133	SvREFCNT_dec (self->cb_object);
1134
1135	PROTOTYPES: ENABLE	1172	PROTOTYPES: ENABLE
1136		1173
1137	void encode_cbor (SV *scalar)	1174	void encode_cbor (SV *scalar)
1138	PPCODE:	1175	PPCODE:
1139	{	1176	{

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs. Revision 1.23 by root, Wed Nov 20 16:19:47 2013 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.23 by root, Wed Nov 20 16:19:47 2013 UTC