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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.31 by root, Thu Nov 28 11:43:26 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 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))
…		…
23	#endif	24	#endif
24	#ifndef SvREFCNT_dec_NN	25	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	26	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	27	#endif
27		28
		29	// known major and minor types
		30	enum cbor_type
		31	{
		32	MAJOR_SHIFT = 5,
		33	MINOR_MASK = 0x1f,
		34
		35	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		36	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		37	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		38	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		39	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		40	MAJOR_MAP = 5 << MAJOR_SHIFT,
		41	MAJOR_TAG = 6 << MAJOR_SHIFT,
		42	MAJOR_MISC = 7 << MAJOR_SHIFT,
		43
		44	// INT/STRING/ARRAY/MAP subtypes
		45	LENGTH_EXT1 = 24,
		46	LENGTH_EXT2 = 25,
		47	LENGTH_EXT4 = 26,
		48	LENGTH_EXT8 = 27,
		49
		50	// SIMPLE types (effectively MISC subtypes)
		51	SIMPLE_FALSE = 20,
		52	SIMPLE_TRUE = 21,
		53	SIMPLE_NULL = 22,
		54	SIMPLE_UNDEF = 23,
		55
		56	// MISC subtype (unused)
		57	MISC_EXT1 = 24,
		58	MISC_FLOAT16 = 25,
		59	MISC_FLOAT32 = 26,
		60	MISC_FLOAT64 = 27,
		61
		62	// BYTES/TEXT/ARRAY/MAP
		63	MINOR_INDEF = 31,
		64	};
		65
28	// known tags	66	// known tags
29	enum cbor_tag	67	enum cbor_tag
30	{	68	{
31	// extensions	69	// extensions
32	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref	70	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object	71	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object	72	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing	73	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing	74	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref	75	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	76	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39		77
40	// rfc7049	78	// rfc7049
41	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	79	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	80	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43	CBOR_TAG_POS_BIGNUM = 2, // byte string	81	CBOR_TAG_POS_BIGNUM = 2, // byte string
44	CBOR_TAG_NEG_BIGNUM = 3, // byte string	82	CBOR_TAG_NEG_BIGNUM = 3, // byte string
45	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	83	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46	CBOR_TAG_BIGFLOAT = 5, // array	84	CBOR_TAG_BIGFLOAT = 5, // array
47		85
48	CBOR_TAG_CONV_B64U = 21, // base64url, any	86	CBOR_TAG_CONV_B64U = 21, // base64url, any
49	CBOR_TAG_CONV_B64 = 22, // base64, any	87	CBOR_TAG_CONV_B64 = 22, // base64, any
50	CBOR_TAG_CONV_HEX = 23, // base16, any	88	CBOR_TAG_CONV_HEX = 23, // base16, any
51	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	89	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52		90
53	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	91	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	92	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	93	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	94	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	95	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58		96
59	CBOR_TAG_MAGIC = 55799 // self-describe cbor	97	CBOR_TAG_MAGIC = 55799, // self-describe cbor
60	};	98	};
61		99
62	#define F_SHRINK 0x00000001UL	100	#define F_SHRINK 0x00000001UL
63	#define F_ALLOW_UNKNOWN 0x00000002UL	101	#define F_ALLOW_UNKNOWN 0x00000002UL
64	#define F_ALLOW_SHARING 0x00000004UL //TODO	102	#define F_ALLOW_SHARING 0x00000004UL
65	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	103	#define F_ALLOW_CYCLES 0x00000008UL
		104	#define F_PACK_STRINGS 0x00000010UL
		105	#define F_TEXT_KEYS 0x00000020UL
		106	#define F_TEXT_STRINGS 0x00000040UL
		107	#define F_VALIDATE_UTF8 0x00000080UL
66		108
67	#define INIT_SIZE 32 // initial scalar size to be allocated	109	#define INIT_SIZE 32 // initial scalar size to be allocated
68		110
69	#define SB do {	111	#define SB do {
70	#define SE } while (0)	112	#define SE } while (0)
…		…
89	typedef struct {	131	typedef struct {
90	U32 flags;	132	U32 flags;
91	U32 max_depth;	133	U32 max_depth;
92	STRLEN max_size;	134	STRLEN max_size;
93	SV *filter;	135	SV *filter;
		136
		137	// for the incremental parser
		138	STRLEN incr_pos; // the current offset into the text
		139	STRLEN incr_need; // minimum bytes needed to decode
		140	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94	} CBOR;	141	} CBOR;
95		142
96	ecb_inline void	143	ecb_inline void
97	cbor_init (CBOR *cbor)	144	cbor_init (CBOR *cbor)
98	{	145	{
…		…
102		149
103	ecb_inline void	150	ecb_inline void
104	cbor_free (CBOR *cbor)	151	cbor_free (CBOR *cbor)
105	{	152	{
106	SvREFCNT_dec (cbor->filter);	153	SvREFCNT_dec (cbor->filter);
		154	SvREFCNT_dec (cbor->incr_count);
107	}	155	}
108		156
109	/////////////////////////////////////////////////////////////////////////////	157	/////////////////////////////////////////////////////////////////////////////
110	// utility functions	158	// utility functions
111		159
…		…
189	static void	237	static void
190	encode_uint (enc_t *enc, int major, UV len)	238	encode_uint (enc_t *enc, int major, UV len)
191	{	239	{
192	need (enc, 9);	240	need (enc, 9);
193		241
194	if (ecb_expect_true (len < 24))	242	if (ecb_expect_true (len < LENGTH_EXT1))
195	*enc->cur++ = major \| len;	243	*enc->cur++ = major \| len;
196	else if (ecb_expect_true (len <= 0xff))	244	else if (ecb_expect_true (len <= 0xffU))
197	{	245	{
198	*enc->cur++ = major \| 24;	246	*enc->cur++ = major \| LENGTH_EXT1;
199	*enc->cur++ = len;	247	*enc->cur++ = len;
200	}	248	}
201	else if (len <= 0xffff)	249	else if (len <= 0xffffU)
202	{	250	{
203	*enc->cur++ = major \| 25;	251	*enc->cur++ = major \| LENGTH_EXT2;
204	*enc->cur++ = len >> 8;	252	*enc->cur++ = len >> 8;
205	*enc->cur++ = len;	253	*enc->cur++ = len;
206	}	254	}
207	else if (len <= 0xffffffff)	255	else if (len <= 0xffffffffU)
208	{	256	{
209	*enc->cur++ = major \| 26;	257	*enc->cur++ = major \| LENGTH_EXT4;
210	*enc->cur++ = len >> 24;	258	*enc->cur++ = len >> 24;
211	*enc->cur++ = len >> 16;	259	*enc->cur++ = len >> 16;
212	*enc->cur++ = len >> 8;	260	*enc->cur++ = len >> 8;
213	*enc->cur++ = len;	261	*enc->cur++ = len;
214	}	262	}
215	else	263	else
216	{	264	{
217	*enc->cur++ = major \| 27;	265	*enc->cur++ = major \| LENGTH_EXT8;
218	*enc->cur++ = len >> 56;	266	*enc->cur++ = len >> 56;
219	*enc->cur++ = len >> 48;	267	*enc->cur++ = len >> 48;
220	*enc->cur++ = len >> 40;	268	*enc->cur++ = len >> 40;
221	*enc->cur++ = len >> 32;	269	*enc->cur++ = len >> 32;
222	*enc->cur++ = len >> 24;	270	*enc->cur++ = len >> 24;
…		…
227	}	275	}
228		276
229	ecb_inline void	277	ecb_inline void
230	encode_tag (enc_t *enc, UV tag)	278	encode_tag (enc_t *enc, UV tag)
231	{	279	{
232	encode_uint (enc, 0xc0, tag);	280	encode_uint (enc, MAJOR_TAG, tag);
		281	}
		282
		283	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		284	ecb_noinline static void
		285	encode_str_utf8 (enc_t enc, int utf8, char str, STRLEN len)
		286	{
		287	STRLEN ulen = len;
		288	U8 p, pend = (U8 *)str + len;
		289
		290	for (p = (U8 *)str; p < pend; ++p)
		291	ulen += *p >> 7; // count set high bits
		292
		293	encode_uint (enc, MAJOR_TEXT, ulen);
		294
		295	need (enc, ulen);
		296	for (p = (U8 *)str; p < pend; ++p)
		297	if (*p < 0x80)
		298	enc->cur++ = p;
		299	else
		300	{
		301	enc->cur++ = 0xc0 + (p >> 6);
		302	enc->cur++ = 0x80 + (p & 63);
		303	}
233	}	304	}
234		305
235	ecb_inline void	306	ecb_inline void
236	encode_str (enc_t enc, int utf8, char str, STRLEN len)	307	encode_str (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
237	{	308	{
238	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	309	if (ecb_expect_false (upgrade_utf8))
		310	if (!utf8)
		311	{
		312	encode_str_utf8 (enc, utf8, str, len);
		313	return;
		314	}
		315
		316	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239	need (enc, len);	317	need (enc, len);
240	memcpy (enc->cur, str, len);	318	memcpy (enc->cur, str, len);
241	enc->cur += len;	319	enc->cur += len;
242	}	320	}
243		321
244	static void	322	ecb_inline void
245	encode_strref (enc_t enc, int utf8, char str, STRLEN len)	323	encode_strref (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
246	{	324	{
247	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	325	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248	{	326	{
249	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	327	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250		328
251	if (SvOK (*svp))	329	if (SvOK (*svp))
252	{	330	{
253	// already registered, use stringref	331	// already registered, use stringref
254	encode_tag (enc, CBOR_TAG_STRINGREF);	332	encode_tag (enc, CBOR_TAG_STRINGREF);
255	encode_uint (enc, 0x00, SvUV (*svp));	333	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256	return;	334	return;
257	}	335	}
258	else if (len >= minimum_string_length (enc->stringref_idx))	336	else if (len >= minimum_string_length (enc->stringref_idx))
259	{	337	{
260	// register only	338	// register only
261	sv_setuv (*svp, enc->stringref_idx);	339	sv_setuv (*svp, enc->stringref_idx);
262	++enc->stringref_idx;	340	++enc->stringref_idx;
263	}	341	}
264	}	342	}
265		343
266	encode_str (enc, utf8, str, len);	344	encode_str (enc, upgrade_utf8, utf8, str, len);
267	}	345	}
268		346
269	static void encode_sv (enc_t enc, SV sv);	347	static void encode_sv (enc_t enc, SV sv);
270		348
271	static void	349	static void
…		…
276	if (enc->depth >= enc->cbor.max_depth)	354	if (enc->depth >= enc->cbor.max_depth)
277	croak (ERR_NESTING_EXCEEDED);	355	croak (ERR_NESTING_EXCEEDED);
278		356
279	++enc->depth;	357	++enc->depth;
280		358
281	encode_uint (enc, 0x80, len + 1);	359	encode_uint (enc, MAJOR_ARRAY, len + 1);
282		360
		361	if (SvMAGICAL (av))
283	for (i = 0; i <= len; ++i)	362	for (i = 0; i <= len; ++i)
284	{	363	{
285	SV **svp = av_fetch (av, i, 0);	364	SV **svp = av_fetch (av, i, 0);
286	encode_sv (enc, svp ? *svp : &PL_sv_undef);	365	encode_sv (enc, svp ? *svp : &PL_sv_undef);
287	}	366	}
		367	else
		368	for (i = 0; i <= len; ++i)
		369	{
		370	SV *sv = AvARRAY (av)[i];
		371	encode_sv (enc, sv ? sv : &PL_sv_undef);
		372	}
288		373
289	--enc->depth;	374	--enc->depth;
290	}	375	}
291		376
292	static void	377	static void
…		…
301		386
302	int pairs = hv_iterinit (hv);	387	int pairs = hv_iterinit (hv);
303	int mg = SvMAGICAL (hv);	388	int mg = SvMAGICAL (hv);
304		389
305	if (mg)	390	if (mg)
306	encode_ch (enc, 0xa0 \| 31);	391	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
307	else	392	else
308	encode_uint (enc, 0xa0, pairs);	393	encode_uint (enc, MAJOR_MAP, pairs);
309		394
310	while ((he = hv_iternext (hv)))	395	while ((he = hv_iternext (hv)))
311	{	396	{
312	if (HeKLEN (he) == HEf_SVKEY)	397	if (HeKLEN (he) == HEf_SVKEY)
313	encode_sv (enc, HeSVKEY (he));	398	encode_sv (enc, HeSVKEY (he));
314	else	399	else
315	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	400	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS \| F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
316		401
317	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	402	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318	}	403	}
319		404
320	if (mg)	405	if (mg)
321	encode_ch (enc, 0xe0 \| 31);	406	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
322		407
323	--enc->depth;	408	--enc->depth;
324	}	409	}
325		410
326	// encode objects, arrays and special \0=false and \1=true values.	411	// encode objects, arrays and special \0=false and \1=true values.
327	static void	412	static void
328	encode_rv (enc_t enc, SV sv)	413	encode_rv (enc_t enc, SV sv)
329	{	414	{
330	SvGETMAGIC (sv);	415	SvGETMAGIC (sv);
331
332	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
333	&& ecb_expect_false (SvREFCNT (sv) > 1))
334	{
335	if (!enc->shareable)
336	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
337
338	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
339
340	if (SvOK (*svp))
341	{
342	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
343	encode_uint (enc, 0x00, SvUV (*svp));
344	return;
345	}
346	else
347	{
348	sv_setuv (*svp, enc->shareable_idx);
349	++enc->shareable_idx;
350	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
351	}
352	}
353		416
354	svtype svt = SvTYPE (sv);	417	svtype svt = SvTYPE (sv);
355		418
356	if (ecb_expect_false (SvOBJECT (sv)))	419	if (ecb_expect_false (SvOBJECT (sv)))
357	{	420	{
…		…
364	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	427	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
365	? cbor_tagged_stash	428	? cbor_tagged_stash
366	: gv_stashpv ("CBOR::XS::Tagged" , 1);	429	: gv_stashpv ("CBOR::XS::Tagged" , 1);
367		430
368	HV *stash = SvSTASH (sv);	431	HV *stash = SvSTASH (sv);
369	GV *method;
370		432
371	if (stash == boolean_stash)	433	if (stash == boolean_stash)
372	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	434	{
		435	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);
		436	return;
		437	}
373	else if (stash == error_stash)	438	else if (stash == error_stash)
374	encode_ch (enc, 0xe0 \| 23);	439	{
		440	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
		441	return;
		442	}
375	else if (stash == tagged_stash)	443	else if (stash == tagged_stash)
376	{	444	{
377	if (svt != SVt_PVAV)	445	if (svt != SVt_PVAV)
378	croak ("encountered CBOR::XS::Tagged object that isn't an array");	446	croak ("encountered CBOR::XS::Tagged object that isn't an array");
379		447
380	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	448	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
381	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	449	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		450
		451	return;
		452	}
		453	}
		454
		455	if (ecb_expect_false (SvREFCNT (sv) > 1)
		456	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		457	{
		458	if (!enc->shareable)
		459	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		460
		461	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		462
		463	if (SvOK (*svp))
382	}	464	{
		465	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		466	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		467	return;
		468	}
		469	else
		470	{
		471	sv_setuv (*svp, enc->shareable_idx);
		472	++enc->shareable_idx;
		473	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		474	}
		475	}
		476
		477	if (ecb_expect_false (SvOBJECT (sv)))
		478	{
		479	HV *stash = SvSTASH (sv);
		480	GV *method;
		481
383	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	482	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384	{	483	{
385	dSP;	484	dSP;
386		485
387	ENTER; SAVETMPS; PUSHMARK (SP);	486	ENTER; SAVETMPS;
		487	PUSHMARK (SP);
388	// we re-bless the reference to get overload and other niceties right	488	// we re-bless the reference to get overload and other niceties right
389	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	489	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
390		490
391	PUTBACK;	491	PUTBACK;
392	// G_SCALAR ensures that return value is 1	492	// G_SCALAR ensures that return value is 1
…		…
405	}	505	}
406	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	506	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
407	{	507	{
408	dSP;	508	dSP;
409		509
410	ENTER; SAVETMPS; PUSHMARK (SP);	510	ENTER; SAVETMPS;
		511	SAVESTACK_POS ();
		512	PUSHMARK (SP);
411	EXTEND (SP, 2);	513	EXTEND (SP, 2);
412	// we re-bless the reference to get overload and other niceties right	514	// we re-bless the reference to get overload and other niceties right
413	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	515	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
414	PUSHs (sv_cbor);	516	PUSHs (sv_cbor);
415		517
…		…
420	// catch this surprisingly common error	522	// catch this surprisingly common error
421	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	523	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
422	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	524	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423		525
424	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	526	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425	encode_uint (enc, 0x80, count + 1);	527	encode_uint (enc, MAJOR_ARRAY, count + 1);
426	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	528	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427		529
428	while (count)	530	while (count)
429	encode_sv (enc, SP[1 - count--]);	531	encode_sv (enc, SP[1 - count--]);
430		532
431	PUTBACK;	533	PUTBACK;
…		…
452	{	554	{
453	double nv = SvNVX (sv);	555	double nv = SvNVX (sv);
454		556
455	need (enc, 9);	557	need (enc, 9);
456		558
457	if (ecb_expect_false (nv == (U32)nv))	559	if (ecb_expect_false (nv == (NV)(U32)nv))
458	encode_uint (enc, 0x00, (U32)nv);	560	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459	//TODO: maybe I32?	561	//TODO: maybe I32?
460	else if (ecb_expect_false (nv == (float)nv))	562	else if (ecb_expect_false (nv == (float)nv))
461	{	563	{
462	uint32_t fp = ecb_float_to_binary32 (nv);	564	uint32_t fp = ecb_float_to_binary32 (nv);
463		565
464	*enc->cur++ = 0xe0 \| 26;	566	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
465		567
466	if (!ecb_big_endian ())	568	if (!ecb_big_endian ())
467	fp = ecb_bswap32 (fp);	569	fp = ecb_bswap32 (fp);
468		570
469	memcpy (enc->cur, &fp, 4);	571	memcpy (enc->cur, &fp, 4);
…		…
471	}	573	}
472	else	574	else
473	{	575	{
474	uint64_t fp = ecb_double_to_binary64 (nv);	576	uint64_t fp = ecb_double_to_binary64 (nv);
475		577
476	*enc->cur++ = 0xe0 \| 27;	578	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
477		579
478	if (!ecb_big_endian ())	580	if (!ecb_big_endian ())
479	fp = ecb_bswap64 (fp);	581	fp = ecb_bswap64 (fp);
480		582
481	memcpy (enc->cur, &fp, 8);	583	memcpy (enc->cur, &fp, 8);
…		…
490		592
491	if (SvPOKp (sv))	593	if (SvPOKp (sv))
492	{	594	{
493	STRLEN len;	595	STRLEN len;
494	char *str = SvPV (sv, len);	596	char *str = SvPV (sv, len);
495	encode_strref (enc, SvUTF8 (sv), str, len);	597	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
496	}	598	}
497	else if (SvNOKp (sv))	599	else if (SvNOKp (sv))
498	encode_nv (enc, sv);	600	encode_nv (enc, sv);
499	else if (SvIOKp (sv))	601	else if (SvIOKp (sv))
500	{	602	{
501	if (SvIsUV (sv))	603	if (SvIsUV (sv))
502	encode_uint (enc, 0x00, SvUVX (sv));	604	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
503	else if (SvIVX (sv) >= 0)	605	else if (SvIVX (sv) >= 0)
504	encode_uint (enc, 0x00, SvIVX (sv));	606	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
505	else	607	else
506	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	608	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
507	}	609	}
508	else if (SvROK (sv))	610	else if (SvROK (sv))
509	encode_rv (enc, SvRV (sv));	611	encode_rv (enc, SvRV (sv));
510	else if (!SvOK (sv))	612	else if (!SvOK (sv))
511	encode_ch (enc, 0xe0 \| 22);	613	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
512	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	614	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513	encode_ch (enc, 0xe0 \| 23);	615	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
514	else	616	else
515	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	617	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
516	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	618	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517	}	619	}
518		620
519	static SV *	621	static SV *
520	encode_cbor (SV scalar, CBOR cbor)	622	encode_cbor (SV scalar, CBOR cbor)
521	{	623	{
522	enc_t enc = { };	624	enc_t enc = { 0 };
523		625
524	enc.cbor = *cbor;	626	enc.cbor = *cbor;
525	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	627	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
526	enc.cur = SvPVX (enc.sv);	628	enc.cur = SvPVX (enc.sv);
527	enc.end = SvEND (enc.sv);	629	enc.end = SvEND (enc.sv);
528		630
529	SvPOK_only (enc.sv);	631	SvPOK_only (enc.sv);
530		632
531	if (cbor->flags & F_ALLOW_STRINGREF)	633	if (cbor->flags & F_PACK_STRINGS)
532	{	634	{
533	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	635	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
534	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());	636	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
535	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());	637	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
536	}	638	}
…		…
559	U32 depth; // recursion depth	661	U32 depth; // recursion depth
560	U32 maxdepth; // recursion depth limit	662	U32 maxdepth; // recursion depth limit
561	AV *shareable;	663	AV *shareable;
562	AV *stringref;	664	AV *stringref;
563	SV *decode_tagged;	665	SV *decode_tagged;
		666	SV *err_sv; // optional sv for error, needs to be freed
564	} dec_t;	667	} dec_t;
565		668
566	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	669	// set dec->err to ERRSV
		670	ecb_cold static void
		671	err_errsv (dec_t *dec)
		672	{
		673	if (!dec->err)
		674	{
		675	dec->err_sv = newSVsv (ERRSV);
567		676
568	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	677	// chop off the trailing \n
		678	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		679	*SvEND (dec->err_sv) = 0;
569		680
		681	dec->err = SvPVutf8_nolen (dec->err_sv);
		682	}
		683	}
		684
		685	// the following functions are used to reduce code size and help the compiler to optimise
		686	ecb_cold static void
		687	err_set (dec_t dec, const char reason)
		688	{
		689	if (!dec->err)
		690	dec->err = reason;
		691	}
		692
		693	ecb_cold static void
		694	err_unexpected_end (dec_t *dec)
		695	{
		696	err_set (dec, "unexpected end of CBOR data");
		697	}
		698
		699	ecb_cold static void
		700	err_nesting_exceeded (dec_t *dec)
		701	{
		702	err_set (dec, ERR_NESTING_EXCEEDED);
		703	}
		704
		705	#define ERR_DO(do) SB do; goto fail; SE
		706	#define ERR(reason) ERR_DO (err_set (dec, reason))
		707	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		708
		709	#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
		710
570	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	711	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
571	#define DEC_DEC_DEPTH --dec->depth	712	#define DEC_DEC_DEPTH --dec->depth
572		713
573	static UV	714	static UV
574	decode_uint (dec_t *dec)	715	decode_uint (dec_t *dec)
575	{	716	{
576	switch (*dec->cur & 31)	717	U8 m = *dec->cur & MINOR_MASK;
577	{	718	++dec->cur;
578	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
579	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
580	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
581	return *dec->cur++ & 31;
582		719
583	case 24:	720	if (ecb_expect_true (m < LENGTH_EXT1))
		721	return m;
		722	else if (ecb_expect_true (m == LENGTH_EXT1))
		723	{
584	WANT (2);	724	WANT (1);
585	dec->cur += 2;	725	dec->cur += 1;
586	return dec->cur[-1];	726	return dec->cur[-1];
587		727	}
588	case 25:	728	else if (ecb_expect_true (m == LENGTH_EXT2))
		729	{
589	WANT (3);	730	WANT (2);
590	dec->cur += 3;	731	dec->cur += 2;
591	return (((UV)dec->cur[-2]) << 8)	732	return (((UV)dec->cur[-2]) << 8)
592	\| ((UV)dec->cur[-1]);	733	\| ((UV)dec->cur[-1]);
593		734	}
594	case 26:	735	else if (ecb_expect_true (m == LENGTH_EXT4))
		736	{
595	WANT (5);	737	WANT (4);
596	dec->cur += 5;	738	dec->cur += 4;
597	return (((UV)dec->cur[-4]) << 24)	739	return (((UV)dec->cur[-4]) << 24)
598	\| (((UV)dec->cur[-3]) << 16)	740	\| (((UV)dec->cur[-3]) << 16)
599	\| (((UV)dec->cur[-2]) << 8)	741	\| (((UV)dec->cur[-2]) << 8)
600	\| ((UV)dec->cur[-1]);	742	\| ((UV)dec->cur[-1]);
601		743	}
602	case 27:	744	else if (ecb_expect_true (m == LENGTH_EXT8))
		745	{
603	WANT (9);	746	WANT (8);
604	dec->cur += 9;	747	dec->cur += 8;
		748
		749	return
		750	#if UVSIZE < 8
		751	0
		752	#else
605	return (((UV)dec->cur[-8]) << 56)	753	(((UV)dec->cur[-8]) << 56)
606	\| (((UV)dec->cur[-7]) << 48)	754	\| (((UV)dec->cur[-7]) << 48)
607	\| (((UV)dec->cur[-6]) << 40)	755	\| (((UV)dec->cur[-6]) << 40)
608	\| (((UV)dec->cur[-5]) << 32)	756	\| (((UV)dec->cur[-5]) << 32)
		757	#endif
609	\| (((UV)dec->cur[-4]) << 24)	758	\| (((UV)dec->cur[-4]) << 24)
610	\| (((UV)dec->cur[-3]) << 16)	759	\| (((UV)dec->cur[-3]) << 16)
611	\| (((UV)dec->cur[-2]) << 8)	760	\| (((UV)dec->cur[-2]) << 8)
612	\| ((UV)dec->cur[-1]);	761	\| ((UV)dec->cur[-1]);
613		762	}
614	default:	763	else
615	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	764	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616	}
617		765
618	fail:	766	fail:
619	return 0;	767	return 0;
620	}	768	}
621		769
…		…
626	{	774	{
627	AV *av = newAV ();	775	AV *av = newAV ();
628		776
629	DEC_INC_DEPTH;	777	DEC_INC_DEPTH;
630		778
631	if ((*dec->cur & 31) == 31)	779	if (*dec->cur == (MAJOR_ARRAY \| MINOR_INDEF))
632	{	780	{
633	++dec->cur;	781	++dec->cur;
634		782
635	for (;;)	783	for (;;)
636	{	784	{
637	WANT (1);	785	WANT (1);
638		786
639	if (*dec->cur == (0xe0 \| 31))	787	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
640	{	788	{
641	++dec->cur;	789	++dec->cur;
642	break;	790	break;
643	}	791	}
644		792
645	av_push (av, decode_sv (dec));	793	av_push (av, decode_sv (dec));
646	}	794	}
647	}	795	}
648	else	796	else
649	{	797	{
650	int i, len = decode_uint (dec);	798	UV i, len = decode_uint (dec);
651		799
		800	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
652	av_fill (av, len - 1);	801	av_fill (av, len - 1);
653		802
654	for (i = 0; i < len; ++i)	803	for (i = 0; i < len; ++i)
655	AvARRAY (av)[i] = decode_sv (dec);	804	AvARRAY (av)[i] = decode_sv (dec);
656	}	805	}
…		…
669	{	818	{
670	// for speed reasons, we specialcase single-string	819	// for speed reasons, we specialcase single-string
671	// byte or utf-8 strings as keys, but only when !stringref	820	// byte or utf-8 strings as keys, but only when !stringref
672		821
673	if (ecb_expect_true (!dec->stringref))	822	if (ecb_expect_true (!dec->stringref))
674	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	823	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675	{	824	{
676	I32 len = decode_uint (dec);	825	STRLEN len = decode_uint (dec);
677	char key = (char )dec->cur;	826	char key = (char )dec->cur;
678		827
		828	WANT (len);
679	dec->cur += len;	829	dec->cur += len;
680		830
681	if (ecb_expect_false (dec->stringref))
682	av_push (dec->stringref, newSVpvn (key, len));
683
684	hv_store (hv, key, len, decode_sv (dec), 0);	831	hv_store (hv, key, len, decode_sv (dec), 0);
685		832
686	return;	833	return;
687	}	834	}
688	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	835	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689	{	836	{
690	I32 len = decode_uint (dec);	837	STRLEN len = decode_uint (dec);
691	char key = (char )dec->cur;	838	char key = (char )dec->cur;
692		839
		840	WANT (len);
693	dec->cur += len;	841	dec->cur += len;
694		842
695	if (ecb_expect_false (dec->stringref))	843	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
696	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	844	if (!is_utf8_string (key, len))
		845	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
697		846
698	hv_store (hv, key, -len, decode_sv (dec), 0);	847	hv_store (hv, key, -len, decode_sv (dec), 0);
699		848
700	return;	849	return;
701	}	850	}
702		851
703	SV *k = decode_sv (dec);	852	SV *k = decode_sv (dec);
704	SV *v = decode_sv (dec);	853	SV *v = decode_sv (dec);
705		854
		855	// we leak memory if uncaught exceptions are thrown by random magical
		856	// methods, and this is hopefully the only place where it can happen,
		857	// so if there is a chance of an exception, take the very slow path.
		858	// since catching exceptions is "undocumented/internal/forbidden" by
		859	// the new p5p powers, we need to call out to a perl function :/
		860	if (ecb_expect_false (SvAMAGIC (k)))
		861	{
		862	dSP;
		863
		864	ENTER; SAVETMPS;
		865	PUSHMARK (SP);
		866	EXTEND (SP, 3);
		867	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		868	PUSHs (sv_2mortal (k));
		869	PUSHs (sv_2mortal (v));
		870
		871	PUTBACK;
		872	call_pv ("CBOR::XS::_hv_store", G_VOID \| G_DISCARD \| G_EVAL);
		873	SPAGAIN;
		874
		875	FREETMPS; LEAVE;
		876
		877	if (SvTRUE (ERRSV))
		878	ERR_ERRSV;
		879
		880	return;
		881	}
		882
706	hv_store_ent (hv, k, v, 0);	883	hv_store_ent (hv, k, v, 0);
707	SvREFCNT_dec (k);	884	SvREFCNT_dec (k);
		885
		886	fail:
		887	;
708	}	888	}
709		889
710	static SV *	890	static SV *
711	decode_hv (dec_t *dec)	891	decode_hv (dec_t *dec)
712	{	892	{
713	HV *hv = newHV ();	893	HV *hv = newHV ();
714		894
715	DEC_INC_DEPTH;	895	DEC_INC_DEPTH;
716		896
717	if ((*dec->cur & 31) == 31)	897	if (*dec->cur == (MAJOR_MAP \| MINOR_INDEF))
718	{	898	{
719	++dec->cur;	899	++dec->cur;
720		900
721	for (;;)	901	for (;;)
722	{	902	{
723	WANT (1);	903	WANT (1);
724		904
725	if (*dec->cur == (0xe0 \| 31))	905	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
726	{	906	{
727	++dec->cur;	907	++dec->cur;
728	break;	908	break;
729	}	909	}
730		910
731	decode_he (dec, hv);	911	decode_he (dec, hv);
732	}	912	}
733	}	913	}
734	else	914	else
735	{	915	{
736	int pairs = decode_uint (dec);	916	UV pairs = decode_uint (dec);
		917
		918	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
737		919
738	while (pairs--)	920	while (pairs--)
739	decode_he (dec, hv);	921	decode_he (dec, hv);
740	}	922	}
741		923
…		…
751	static SV *	933	static SV *
752	decode_str (dec_t *dec, int utf8)	934	decode_str (dec_t *dec, int utf8)
753	{	935	{
754	SV *sv = 0;	936	SV *sv = 0;
755		937
756	if ((*dec->cur & 31) == 31)	938	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
757	{	939	{
		940	// indefinite length strings
758	++dec->cur;	941	++dec->cur;
759		942
		943	U8 major = *dec->cur & MAJOR_MISC;
		944
760	sv = newSVpvn ("", 0);	945	sv = newSVpvn ("", 0);
761		946
762	// not very fast, and certainly not robust against illegal input
763	for (;;)	947	for (;;)
764	{	948	{
765	WANT (1);	949	WANT (1);
766		950
767	if (*dec->cur == (0xe0 \| 31))	951	if ((*dec->cur - major) > LENGTH_EXT8)
		952	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
768	{	953	{
769	++dec->cur;	954	++dec->cur;
770	break;	955	break;
771	}	956	}
		957	else
		958	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772		959
773	sv_catsv (sv, decode_sv (dec));	960	STRLEN len = decode_uint (dec);
		961
		962	WANT (len);
		963	sv_catpvn (sv, dec->cur, len);
		964	dec->cur += len;
774	}	965	}
775	}	966	}
776	else	967	else
777	{	968	{
778	STRLEN len = decode_uint (dec);	969	STRLEN len = decode_uint (dec);
…		…
785	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	976	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
786	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	977	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
787	}	978	}
788		979
789	if (utf8)	980	if (utf8)
		981	{
		982	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		983	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		984	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		985
790	SvUTF8_on (sv);	986	SvUTF8_on (sv);
		987	}
791		988
792	return sv;	989	return sv;
793		990
794	fail:	991	fail:
795	SvREFCNT_dec (sv);	992	SvREFCNT_dec (sv);
…		…
814	sv = newRV_noinc (decode_sv (dec));	1011	sv = newRV_noinc (decode_sv (dec));
815	break;	1012	break;
816		1013
817	case CBOR_TAG_STRINGREF_NAMESPACE:	1014	case CBOR_TAG_STRINGREF_NAMESPACE:
818	{	1015	{
		1016	// do nmot use SAVETMPS/FREETMPS, as these will
		1017	// erase mortalised caches, e.g. "shareable"
819	ENTER; SAVETMPS;	1018	ENTER;
820		1019
821	SAVESPTR (dec->stringref);	1020	SAVESPTR (dec->stringref);
822	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1021	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
823		1022
824	sv = decode_sv (dec);	1023	sv = decode_sv (dec);
825		1024
826	FREETMPS; LEAVE;	1025	LEAVE;
827	}	1026	}
828	break;	1027	break;
829		1028
830	case CBOR_TAG_STRINGREF:	1029	case CBOR_TAG_STRINGREF:
831	{	1030	{
832	if ((*dec->cur >> 5) != 0)	1031	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1032	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834		1033
835	UV idx = decode_uint (dec);	1034	UV idx = decode_uint (dec);
836		1035
837	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	1036	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
…		…
844	case CBOR_TAG_VALUE_SHAREABLE:	1043	case CBOR_TAG_VALUE_SHAREABLE:
845	{	1044	{
846	if (ecb_expect_false (!dec->shareable))	1045	if (ecb_expect_false (!dec->shareable))
847	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1046	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
848		1047
		1048	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1049	{
849	sv = newSV (0);	1050	sv = newSV (0);
850	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1051	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851		1052
852	SV *osv = decode_sv (dec);	1053	SV *osv = decode_sv (dec);
853	sv_setsv (sv, osv);	1054	sv_setsv (sv, osv);
854	SvREFCNT_dec_NN (osv);	1055	SvREFCNT_dec_NN (osv);
		1056	}
		1057	else
		1058	{
		1059	av_push (dec->shareable, &PL_sv_undef);
		1060	int idx = AvFILLp (dec->shareable);
		1061	sv = decode_sv (dec);
		1062	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1063	}
855	}	1064	}
856	break;	1065	break;
857		1066
858	case CBOR_TAG_VALUE_SHAREDREF:	1067	case CBOR_TAG_VALUE_SHAREDREF:
859	{	1068	{
860	if ((*dec->cur >> 5) != 0)	1069	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1070	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862		1071
863	UV idx = decode_uint (dec);	1072	UV idx = decode_uint (dec);
864		1073
865	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1074	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
866	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1075	ERR ("corrupted CBOR data (sharedref index out of bounds)");
867		1076
868	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1077	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1078
		1079	if (sv == &PL_sv_undef)
		1080	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
869	}	1081	}
870	break;	1082	break;
871		1083
872	case CBOR_TAG_PERL_OBJECT:	1084	case CBOR_TAG_PERL_OBJECT:
873	{	1085	{
…		…
888	if (!method)	1100	if (!method)
889	ERR ("cannot decode perl-object (package does not have a THAW method)");	1101	ERR ("cannot decode perl-object (package does not have a THAW method)");
890		1102
891	dSP;	1103	dSP;
892		1104
893	ENTER; SAVETMPS; PUSHMARK (SP);	1105	ENTER; SAVETMPS;
		1106	PUSHMARK (SP);
894	EXTEND (SP, len + 1);	1107	EXTEND (SP, len + 1);
895	// we re-bless the reference to get overload and other niceties right	1108	// we re-bless the reference to get overload and other niceties right
896	PUSHs (*av_fetch (av, 0, 1));	1109	PUSHs (*av_fetch (av, 0, 1));
897	PUSHs (sv_cbor);	1110	PUSHs (sv_cbor);
898		1111
…		…
906	SPAGAIN;	1119	SPAGAIN;
907		1120
908	if (SvTRUE (ERRSV))	1121	if (SvTRUE (ERRSV))
909	{	1122	{
910	FREETMPS; LEAVE;	1123	FREETMPS; LEAVE;
911	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1124	ERR_ERRSV;
912	}	1125	}
913		1126
914	SvREFCNT_dec (sv);	1127	SvREFCNT_dec (sv);
915	sv = SvREFCNT_inc (POPs);	1128	sv = SvREFCNT_inc (POPs);
916		1129
…		…
920	}	1133	}
921	break;	1134	break;
922		1135
923	default:	1136	default:
924	{	1137	{
		1138	SV *tag_sv = newSVuv (tag);
		1139
925	sv = decode_sv (dec);	1140	sv = decode_sv (dec);
926		1141
927	dSP;	1142	dSP;
928	ENTER; SAVETMPS; PUSHMARK (SP);	1143	ENTER; SAVETMPS;
		1144	SAVESTACK_POS ();
		1145	PUSHMARK (SP);
929	EXTEND (SP, 2);	1146	EXTEND (SP, 2);
930	PUSHs (newSVuv (tag));	1147	PUSHs (tag_sv);
931	PUSHs (sv);	1148	PUSHs (sv);
932		1149
933	PUTBACK;	1150	PUTBACK;
934	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);	1151	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
935	SPAGAIN;	1152	SPAGAIN;
936		1153
937	if (SvTRUE (ERRSV))	1154	if (SvTRUE (ERRSV))
938	{	1155	{
		1156	SvREFCNT_dec (tag_sv);
939	FREETMPS; LEAVE;	1157	FREETMPS; LEAVE;
940	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1158	ERR_ERRSV;
941	}	1159	}
942		1160
943	if (count)	1161	if (count)
944	{	1162	{
		1163	SvREFCNT_dec (tag_sv);
945	SvREFCNT_dec (sv);	1164	SvREFCNT_dec (sv);
946	sv = SvREFCNT_inc (POPs);	1165	sv = SvREFCNT_inc (POPs);
947	}	1166	}
948	else	1167	else
949	{	1168	{
950	AV *av = newAV ();	1169	AV *av = newAV ();
951	av_push (av, newSVuv (tag));	1170	av_push (av, tag_sv);
952	av_push (av, sv);	1171	av_push (av, sv);
953		1172
954	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1173	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
955	? cbor_tagged_stash	1174	? cbor_tagged_stash
956	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1175	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
974	static SV *	1193	static SV *
975	decode_sv (dec_t *dec)	1194	decode_sv (dec_t *dec)
976	{	1195	{
977	WANT (1);	1196	WANT (1);
978		1197
979	switch (*dec->cur >> 5)	1198	switch (*dec->cur >> MAJOR_SHIFT)
980	{	1199	{
981	case 0: // unsigned int	1200	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982	return newSVuv (decode_uint (dec));	1201	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983	case 1: // negative int	1202	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984	return newSViv (-1 - (IV)decode_uint (dec));	1203	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985	case 2: // octet string	1204	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986	return decode_str (dec, 0);	1205	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987	case 3: // utf-8 string	1206	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988	return decode_str (dec, 1);	1207
989	case 4: // array	1208	case MAJOR_MISC >> MAJOR_SHIFT:
990	return decode_av (dec);
991	case 5: // map
992	return decode_hv (dec);
993	case 6: // tag
994	return decode_tagged (dec);
995	case 7: // misc
996	switch (*dec->cur++ & 31)	1209	switch (*dec->cur++ & MINOR_MASK)
997	{	1210	{
998	case 20:	1211	case SIMPLE_FALSE:
999	#if CBOR_SLOW	1212	#if CBOR_SLOW
1000	types_false = get_bool ("Types::Serialiser::false");	1213	types_false = get_bool ("Types::Serialiser::false");
1001	#endif	1214	#endif
1002	return newSVsv (types_false);	1215	return newSVsv (types_false);
1003	case 21:	1216	case SIMPLE_TRUE:
1004	#if CBOR_SLOW	1217	#if CBOR_SLOW
1005	types_true = get_bool ("Types::Serialiser::true");	1218	types_true = get_bool ("Types::Serialiser::true");
1006	#endif	1219	#endif
1007	return newSVsv (types_true);	1220	return newSVsv (types_true);
1008	case 22:	1221	case SIMPLE_NULL:
1009	return newSVsv (&PL_sv_undef);	1222	return newSVsv (&PL_sv_undef);
1010	case 23:	1223	case SIMPLE_UNDEF:
1011	#if CBOR_SLOW	1224	#if CBOR_SLOW
1012	types_error = get_bool ("Types::Serialiser::error");	1225	types_error = get_bool ("Types::Serialiser::error");
1013	#endif	1226	#endif
1014	return newSVsv (types_error);	1227	return newSVsv (types_error);
1015		1228
1016	case 25:	1229	case MISC_FLOAT16:
1017	{	1230	{
1018	WANT (2);	1231	WANT (2);
1019		1232
1020	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];	1233	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];
1021	dec->cur += 2;	1234	dec->cur += 2;
1022		1235
1023	return newSVnv (ecb_binary16_to_float (fp));	1236	return newSVnv (ecb_binary16_to_float (fp));
1024	}	1237	}
1025		1238
1026	case 26:	1239	case MISC_FLOAT32:
1027	{	1240	{
1028	uint32_t fp;	1241	uint32_t fp;
1029	WANT (4);	1242	WANT (4);
1030	memcpy (&fp, dec->cur, 4);	1243	memcpy (&fp, dec->cur, 4);
1031	dec->cur += 4;	1244	dec->cur += 4;
…		…
1034	fp = ecb_bswap32 (fp);	1247	fp = ecb_bswap32 (fp);
1035		1248
1036	return newSVnv (ecb_binary32_to_float (fp));	1249	return newSVnv (ecb_binary32_to_float (fp));
1037	}	1250	}
1038		1251
1039	case 27:	1252	case MISC_FLOAT64:
1040	{	1253	{
1041	uint64_t fp;	1254	uint64_t fp;
1042	WANT (8);	1255	WANT (8);
1043	memcpy (&fp, dec->cur, 8);	1256	memcpy (&fp, dec->cur, 8);
1044	dec->cur += 8;	1257	dec->cur += 8;
…		…
1047	fp = ecb_bswap64 (fp);	1260	fp = ecb_bswap64 (fp);
1048		1261
1049	return newSVnv (ecb_binary64_to_double (fp));	1262	return newSVnv (ecb_binary64_to_double (fp));
1050	}	1263	}
1051		1264
1052	// 0..19 unassigned	1265	// 0..19 unassigned simple
1053	// 24 reserved + unassigned (reserved values are not encodable)	1266	// 24 reserved + unassigned simple (reserved values are not encodable)
		1267	// 28-30 unassigned misc
		1268	// 31 break code
1054	default:	1269	default:
1055	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1270	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1056	}	1271	}
1057		1272
1058	break;	1273	break;
1059	}	1274	}
1060		1275
…		…
1063	}	1278	}
1064		1279
1065	static SV *	1280	static SV *
1066	decode_cbor (SV string, CBOR cbor, char **offset_return)	1281	decode_cbor (SV string, CBOR cbor, char **offset_return)
1067	{	1282	{
1068	dec_t dec = { };	1283	dec_t dec = { 0 };
1069	SV *sv;	1284	SV *sv;
1070	STRLEN len;	1285	STRLEN len;
1071	char *data = SvPVbyte (string, len);	1286	char *data = SvPVbyte (string, len);
1072		1287
1073	if (len > cbor->max_size && cbor->max_size)	1288	if (len > cbor->max_size && cbor->max_size)
…		…
1087	if (dec.cur != dec.end && !dec.err)	1302	if (dec.cur != dec.end && !dec.err)
1088	dec.err = "garbage after CBOR object";	1303	dec.err = "garbage after CBOR object";
1089		1304
1090	if (dec.err)	1305	if (dec.err)
1091	{	1306	{
		1307	if (dec.shareable)
		1308	{
		1309	// need to break cyclic links, which would all be in shareable
		1310	int i;
		1311	SV **svp;
		1312
		1313	for (i = av_len (dec.shareable) + 1; i--; )
		1314	if ((svp = av_fetch (dec.shareable, i, 0)))
		1315	sv_setsv (*svp, &PL_sv_undef);
		1316	}
		1317
1092	SvREFCNT_dec (sv);	1318	SvREFCNT_dec (sv);
		1319
		1320	if (dec.err_sv)
		1321	sv_2mortal (dec.err_sv);
		1322
1093	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1323	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1094	}	1324	}
1095		1325
1096	sv = sv_2mortal (sv);	1326	sv = sv_2mortal (sv);
1097		1327
1098	return sv;	1328	return sv;
1099	}	1329	}
1100		1330
		1331	/////////////////////////////////////////////////////////////////////////////
		1332	// incremental parser
		1333
		1334	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1335
		1336	// returns 0 for notyet, 1 for success or error
		1337	static int
		1338	incr_parse (CBOR self, SV cborstr)
		1339	{
		1340	STRLEN cur;
		1341	SvPV (cborstr, cur);
		1342
		1343	while (ecb_expect_true (self->incr_need <= cur))
		1344	{
		1345	// table of integer count bytes
		1346	static I8 incr_len[MINOR_MASK + 1] = {
		1347	0, 0, 0, 0, 0, 0, 0, 0,
		1348	0, 0, 0, 0, 0, 0, 0, 0,
		1349	0, 0, 0, 0, 0, 0, 0, 0,
		1350	1, 2, 4, 8,-1,-1,-1,-2
		1351	};
		1352
		1353	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1354	U8 m = *p & MINOR_MASK;
		1355	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1356	I8 ilen = incr_len[m];
		1357
		1358	self->incr_need = self->incr_pos + 1;
		1359
		1360	if (ecb_expect_false (ilen < 0))
		1361	{
		1362	if (m != MINOR_INDEF)
		1363	return 1; // error
		1364
		1365	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1366	{
		1367	if (count >= 0)
		1368	return 1; // error
		1369
		1370	count = 1;
		1371	}
		1372	else
		1373	{
		1374	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1375	count = -1;
		1376	}
		1377	}
		1378	else
		1379	{
		1380	self->incr_need += ilen;
		1381	if (ecb_expect_false (self->incr_need > cur))
		1382	return 0;
		1383
		1384	int major = *p >> MAJOR_SHIFT;
		1385
		1386	switch (major)
		1387	{
		1388	case MAJOR_TAG >> MAJOR_SHIFT:
		1389	++count; // tags merely prefix another value
		1390	break;
		1391
		1392	case MAJOR_BYTES >> MAJOR_SHIFT:
		1393	case MAJOR_TEXT >> MAJOR_SHIFT:
		1394	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1395	case MAJOR_MAP >> MAJOR_SHIFT:
		1396	{
		1397	UV len;
		1398
		1399	if (ecb_expect_false (ilen))
		1400	{
		1401	len = 0;
		1402
		1403	do {
		1404	len = (len << 8) \| *++p;
		1405	} while (--ilen);
		1406	}
		1407	else
		1408	len = m;
		1409
		1410	switch (major)
		1411	{
		1412	case MAJOR_BYTES >> MAJOR_SHIFT:
		1413	case MAJOR_TEXT >> MAJOR_SHIFT:
		1414	self->incr_need += len;
		1415	if (ecb_expect_false (self->incr_need > cur))
		1416	return 0;
		1417
		1418	break;
		1419
		1420	case MAJOR_MAP >> MAJOR_SHIFT:
		1421	len <<= 1;
		1422	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1423	if (len)
		1424	{
		1425	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1426	count = len + 1;
		1427	}
		1428	break;
		1429	}
		1430	}
		1431	}
		1432	}
		1433
		1434	self->incr_pos = self->incr_need;
		1435
		1436	if (count > 0)
		1437	{
		1438	while (!--count)
		1439	{
		1440	if (!AvFILLp (self->incr_count))
		1441	return 1; // done
		1442
		1443	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1444	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1445	}
		1446
		1447	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1448	}
		1449	}
		1450
		1451	return 0;
		1452	}
		1453
		1454
1101	/////////////////////////////////////////////////////////////////////////////	1455	/////////////////////////////////////////////////////////////////////////////
1102	// XS interface functions	1456	// XS interface functions
1103		1457
1104	MODULE = CBOR::XS PACKAGE = CBOR::XS	1458	MODULE = CBOR::XS PACKAGE = CBOR::XS
1105		1459
…		…
1117		1471
1118	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1472	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1119		1473
1120	sv_cbor = newSVpv ("CBOR", 0);	1474	sv_cbor = newSVpv ("CBOR", 0);
1121	SvREADONLY_on (sv_cbor);	1475	SvREADONLY_on (sv_cbor);
		1476
		1477	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1122	}	1478	}
1123		1479
1124	PROTOTYPES: DISABLE	1480	PROTOTYPES: DISABLE
1125		1481
1126	void CLONE (...)	1482	void CLONE (...)
…		…
1145	void shrink (CBOR *self, int enable = 1)	1501	void shrink (CBOR *self, int enable = 1)
1146	ALIAS:	1502	ALIAS:
1147	shrink = F_SHRINK	1503	shrink = F_SHRINK
1148	allow_unknown = F_ALLOW_UNKNOWN	1504	allow_unknown = F_ALLOW_UNKNOWN
1149	allow_sharing = F_ALLOW_SHARING	1505	allow_sharing = F_ALLOW_SHARING
1150	allow_stringref = F_ALLOW_STRINGREF	1506	allow_cycles = F_ALLOW_CYCLES
		1507	pack_strings = F_PACK_STRINGS
		1508	text_keys = F_TEXT_KEYS
		1509	text_strings = F_TEXT_STRINGS
		1510	validate_utf8 = F_VALIDATE_UTF8
1151	PPCODE:	1511	PPCODE:
1152	{	1512	{
1153	if (enable)	1513	if (enable)
1154	self->flags \|= ix;	1514	self->flags \|= ix;
1155	else	1515	else
…		…
1161	void get_shrink (CBOR *self)	1521	void get_shrink (CBOR *self)
1162	ALIAS:	1522	ALIAS:
1163	get_shrink = F_SHRINK	1523	get_shrink = F_SHRINK
1164	get_allow_unknown = F_ALLOW_UNKNOWN	1524	get_allow_unknown = F_ALLOW_UNKNOWN
1165	get_allow_sharing = F_ALLOW_SHARING	1525	get_allow_sharing = F_ALLOW_SHARING
1166	get_allow_stringref = F_ALLOW_STRINGREF	1526	get_allow_cycles = F_ALLOW_CYCLES
		1527	get_pack_strings = F_PACK_STRINGS
		1528	get_text_keys = F_TEXT_KEYS
		1529	get_text_strings = F_TEXT_STRINGS
		1530	get_validate_utf8 = F_VALIDATE_UTF8
1167	PPCODE:	1531	PPCODE:
1168	XPUSHs (boolSV (self->flags & ix));	1532	XPUSHs (boolSV (self->flags & ix));
1169		1533
1170	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1534	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1171	PPCODE:	1535	PPCODE:
…		…
1220	EXTEND (SP, 2);	1584	EXTEND (SP, 2);
1221	PUSHs (sv);	1585	PUSHs (sv);
1222	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1586	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1223	}	1587	}
1224		1588
		1589	void incr_parse (CBOR self, SV cborstr)
		1590	ALIAS:
		1591	incr_parse_multiple = 1
		1592	PPCODE:
		1593	{
		1594	if (SvUTF8 (cborstr))
		1595	sv_utf8_downgrade (cborstr, 0);
		1596
		1597	if (!self->incr_count)
		1598	{
		1599	self->incr_count = newAV ();
		1600	self->incr_pos = 0;
		1601	self->incr_need = 1;
		1602
		1603	av_push (self->incr_count, newSViv (1));
		1604	}
		1605
		1606	do
		1607	{
		1608	if (!incr_parse (self, cborstr))
		1609	{
		1610	if (self->incr_need > self->max_size && self->max_size)
		1611	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1612	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1613
		1614	break;
		1615	}
		1616
		1617	SV *sv;
		1618	char *offset;
		1619
		1620	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1621	XPUSHs (sv);
		1622
		1623	sv_chop (cborstr, offset);
		1624
		1625	av_clear (self->incr_count);
		1626	av_push (self->incr_count, newSViv (1));
		1627
		1628	self->incr_pos = 0;
		1629	self->incr_need = self->incr_pos + 1;
		1630	}
		1631	while (ix);
		1632	}
		1633
		1634	void incr_reset (CBOR *self)
		1635	CODE:
		1636	{
		1637	SvREFCNT_dec (self->incr_count);
		1638	self->incr_count = 0;
		1639	}
		1640
1225	void DESTROY (CBOR *self)	1641	void DESTROY (CBOR *self)
1226	PPCODE:	1642	PPCODE:
1227	cbor_free (self);	1643	cbor_free (self);
1228		1644
1229	PROTOTYPES: ENABLE	1645	PROTOTYPES: ENABLE
1230		1646
1231	void encode_cbor (SV *scalar)	1647	void encode_cbor (SV *scalar)
		1648	ALIAS:
		1649	encode_cbor = 0
		1650	encode_cbor_sharing = F_ALLOW_SHARING
1232	PPCODE:	1651	PPCODE:
1233	{	1652	{
1234	CBOR cbor;	1653	CBOR cbor;
1235	cbor_init (&cbor);	1654	cbor_init (&cbor);
		1655	cbor.flags \|= ix;
1236	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1656	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237	XPUSHs (scalar);	1657	XPUSHs (scalar);
1238	}	1658	}
1239		1659
1240	void decode_cbor (SV *cborstr)	1660	void decode_cbor (SV *cborstr)

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.31 by root, Thu Nov 28 11:43:26 2013 UTC vs. Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.31 by root, Thu Nov 28 11:43:26 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC