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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.49 by root, Mon Feb 8 04:11:11 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	// inofficial extensions (pending iana registration)	69	// extensions
		70	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
32	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	71	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
33	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	72	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
34	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	73	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
35	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	74	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
36	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	75	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
37	CBOR_TAG_STRINGREF = 28, // 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_VALIDATE_UTF8 0x00000020UL
66		106
67	#define INIT_SIZE 32 // initial scalar size to be allocated	107	#define INIT_SIZE 32 // initial scalar size to be allocated
68		108
69	#define SB do {	109	#define SB do {
70	#define SE } while (0)	110	#define SE } while (0)
…		…
89	typedef struct {	129	typedef struct {
90	U32 flags;	130	U32 flags;
91	U32 max_depth;	131	U32 max_depth;
92	STRLEN max_size;	132	STRLEN max_size;
93	SV *filter;	133	SV *filter;
		134
		135	// for the incremental parser
		136	STRLEN incr_pos; // the current offset into the text
		137	STRLEN incr_need; // minimum bytes needed to decode
		138	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94	} CBOR;	139	} CBOR;
95		140
96	ecb_inline void	141	ecb_inline void
97	cbor_init (CBOR *cbor)	142	cbor_init (CBOR *cbor)
98	{	143	{
…		…
102		147
103	ecb_inline void	148	ecb_inline void
104	cbor_free (CBOR *cbor)	149	cbor_free (CBOR *cbor)
105	{	150	{
106	SvREFCNT_dec (cbor->filter);	151	SvREFCNT_dec (cbor->filter);
		152	SvREFCNT_dec (cbor->incr_count);
107	}	153	}
108		154
109	/////////////////////////////////////////////////////////////////////////////	155	/////////////////////////////////////////////////////////////////////////////
110	// utility functions	156	// utility functions
111		157
…		…
141	{	187	{
142	return idx > 23	188	return idx > 23
143	? idx > 0xffU	189	? idx > 0xffU
144	? idx > 0xffffU	190	? idx > 0xffffU
145	? idx > 0xffffffffU	191	? idx > 0xffffffffU
		192	? 11
146	? 7	193	: 7
147	: 6
148	: 5	194	: 5
149	: 4	195	: 4
150	: 3;	196	: 3;
151	}	197	}
152		198
…		…
189	static void	235	static void
190	encode_uint (enc_t *enc, int major, UV len)	236	encode_uint (enc_t *enc, int major, UV len)
191	{	237	{
192	need (enc, 9);	238	need (enc, 9);
193		239
194	if (len < 24)	240	if (ecb_expect_true (len < LENGTH_EXT1))
195	*enc->cur++ = major \| len;	241	*enc->cur++ = major \| len;
196	else if (len <= 0xff)	242	else if (ecb_expect_true (len <= 0xffU))
197	{	243	{
198	*enc->cur++ = major \| 24;	244	*enc->cur++ = major \| LENGTH_EXT1;
199	*enc->cur++ = len;	245	*enc->cur++ = len;
200	}	246	}
201	else if (len <= 0xffff)	247	else if (len <= 0xffffU)
202	{	248	{
203	*enc->cur++ = major \| 25;	249	*enc->cur++ = major \| LENGTH_EXT2;
204	*enc->cur++ = len >> 8;	250	*enc->cur++ = len >> 8;
205	*enc->cur++ = len;	251	*enc->cur++ = len;
206	}	252	}
207	else if (len <= 0xffffffff)	253	else if (len <= 0xffffffffU)
208	{	254	{
209	*enc->cur++ = major \| 26;	255	*enc->cur++ = major \| LENGTH_EXT4;
210	*enc->cur++ = len >> 24;	256	*enc->cur++ = len >> 24;
211	*enc->cur++ = len >> 16;	257	*enc->cur++ = len >> 16;
212	*enc->cur++ = len >> 8;	258	*enc->cur++ = len >> 8;
213	*enc->cur++ = len;	259	*enc->cur++ = len;
214	}	260	}
215	else	261	else
216	{	262	{
217	*enc->cur++ = major \| 27;	263	*enc->cur++ = major \| LENGTH_EXT8;
218	*enc->cur++ = len >> 56;	264	*enc->cur++ = len >> 56;
219	*enc->cur++ = len >> 48;	265	*enc->cur++ = len >> 48;
220	*enc->cur++ = len >> 40;	266	*enc->cur++ = len >> 40;
221	*enc->cur++ = len >> 32;	267	*enc->cur++ = len >> 32;
222	*enc->cur++ = len >> 24;	268	*enc->cur++ = len >> 24;
…		…
227	}	273	}
228		274
229	ecb_inline void	275	ecb_inline void
230	encode_tag (enc_t *enc, UV tag)	276	encode_tag (enc_t *enc, UV tag)
231	{	277	{
232	encode_uint (enc, 0xc0, tag);	278	encode_uint (enc, MAJOR_TAG, tag);
		279	}
		280
		281	ecb_inline void
		282	encode_str (enc_t enc, int utf8, char str, STRLEN len)
		283	{
		284	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		285	need (enc, len);
		286	memcpy (enc->cur, str, len);
		287	enc->cur += len;
233	}	288	}
234		289
235	static void	290	static void
236	encode_str (enc_t enc, int utf8, char str, STRLEN len)	291	encode_strref (enc_t enc, int utf8, char str, STRLEN len)
237	{	292	{
238	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	293	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239	{	294	{
240	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	295	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241		296
242	if (SvOK (*svp))	297	if (SvOK (*svp))
243	{	298	{
244	// already registered, use stringref	299	// already registered, use stringref
245	encode_tag (enc, CBOR_TAG_STRINGREF);	300	encode_tag (enc, CBOR_TAG_STRINGREF);
246	encode_uint (enc, 0x00, SvUV (*svp));	301	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247	return;	302	return;
248	}	303	}
249	else if (len >= minimum_string_length (enc->stringref_idx))	304	else if (len >= minimum_string_length (enc->stringref_idx))
250	{	305	{
251	// register only	306	// register only
252	sv_setuv (*svp, enc->stringref_idx);	307	sv_setuv (*svp, enc->stringref_idx);
253	++enc->stringref_idx;	308	++enc->stringref_idx;
254	}	309	}
255	}	310	}
256		311
257	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	312	encode_str (enc, utf8, str, len);
258	need (enc, len);
259	memcpy (enc->cur, str, len);
260	enc->cur += len;
261	}	313	}
262		314
263	static void encode_sv (enc_t enc, SV sv);	315	static void encode_sv (enc_t enc, SV sv);
264		316
265	static void	317	static void
…		…
270	if (enc->depth >= enc->cbor.max_depth)	322	if (enc->depth >= enc->cbor.max_depth)
271	croak (ERR_NESTING_EXCEEDED);	323	croak (ERR_NESTING_EXCEEDED);
272		324
273	++enc->depth;	325	++enc->depth;
274		326
275	encode_uint (enc, 0x80, len + 1);	327	encode_uint (enc, MAJOR_ARRAY, len + 1);
276		328
		329	if (SvMAGICAL (av))
277	for (i = 0; i <= len; ++i)	330	for (i = 0; i <= len; ++i)
278	{	331	{
279	SV **svp = av_fetch (av, i, 0);	332	SV **svp = av_fetch (av, i, 0);
280	encode_sv (enc, svp ? *svp : &PL_sv_undef);	333	encode_sv (enc, svp ? *svp : &PL_sv_undef);
281	}	334	}
		335	else
		336	for (i = 0; i <= len; ++i)
		337	{
		338	SV *sv = AvARRAY (av)[i];
		339	encode_sv (enc, sv ? sv : &PL_sv_undef);
		340	}
282		341
283	--enc->depth;	342	--enc->depth;
284	}	343	}
285		344
286	static void	345	static void
…		…
295		354
296	int pairs = hv_iterinit (hv);	355	int pairs = hv_iterinit (hv);
297	int mg = SvMAGICAL (hv);	356	int mg = SvMAGICAL (hv);
298		357
299	if (mg)	358	if (mg)
300	encode_ch (enc, 0xa0 \| 31);	359	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
301	else	360	else
302	encode_uint (enc, 0xa0, pairs);	361	encode_uint (enc, MAJOR_MAP, pairs);
303		362
304	while ((he = hv_iternext (hv)))	363	while ((he = hv_iternext (hv)))
305	{	364	{
306	if (HeKLEN (he) == HEf_SVKEY)	365	if (HeKLEN (he) == HEf_SVKEY)
307	encode_sv (enc, HeSVKEY (he));	366	encode_sv (enc, HeSVKEY (he));
308	else	367	else
309	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	368	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310		369
311	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	370	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312	}	371	}
313		372
314	if (mg)	373	if (mg)
315	encode_ch (enc, 0xe0 \| 31);	374	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
316		375
317	--enc->depth;	376	--enc->depth;
318	}	377	}
319		378
320	// encode objects, arrays and special \0=false and \1=true values.	379	// encode objects, arrays and special \0=false and \1=true values.
321	static void	380	static void
322	encode_rv (enc_t enc, SV sv)	381	encode_rv (enc_t enc, SV sv)
323	{	382	{
324	SvGETMAGIC (sv);	383	SvGETMAGIC (sv);
325
326	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327	&& ecb_expect_false (SvREFCNT (sv) > 1))
328	{
329	if (!enc->shareable)
330	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
331
332	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
333
334	if (SvOK (*svp))
335	{
336	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337	encode_uint (enc, 0x00, SvUV (*svp));
338	return;
339	}
340	else
341	{
342	sv_setuv (*svp, enc->shareable_idx);
343	++enc->shareable_idx;
344	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345	}
346	}
347		384
348	svtype svt = SvTYPE (sv);	385	svtype svt = SvTYPE (sv);
349		386
350	if (ecb_expect_false (SvOBJECT (sv)))	387	if (ecb_expect_false (SvOBJECT (sv)))
351	{	388	{
…		…
358	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	395	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
359	? cbor_tagged_stash	396	? cbor_tagged_stash
360	: gv_stashpv ("CBOR::XS::Tagged" , 1);	397	: gv_stashpv ("CBOR::XS::Tagged" , 1);
361		398
362	HV *stash = SvSTASH (sv);	399	HV *stash = SvSTASH (sv);
363	GV *method;
364		400
365	if (stash == boolean_stash)	401	if (stash == boolean_stash)
366	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	402	{
		403	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);
		404	return;
		405	}
367	else if (stash == error_stash)	406	else if (stash == error_stash)
368	encode_ch (enc, 0xe0 \| 23);	407	{
		408	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
		409	return;
		410	}
369	else if (stash == tagged_stash)	411	else if (stash == tagged_stash)
370	{	412	{
371	if (svt != SVt_PVAV)	413	if (svt != SVt_PVAV)
372	croak ("encountered CBOR::XS::Tagged object that isn't an array");	414	croak ("encountered CBOR::XS::Tagged object that isn't an array");
373		415
374	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	416	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
375	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	417	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		418
		419	return;
		420	}
		421	}
		422
		423	if (ecb_expect_false (SvREFCNT (sv) > 1)
		424	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		425	{
		426	if (!enc->shareable)
		427	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		428
		429	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		430
		431	if (SvOK (*svp))
376	}	432	{
		433	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		434	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		435	return;
		436	}
		437	else
		438	{
		439	sv_setuv (*svp, enc->shareable_idx);
		440	++enc->shareable_idx;
		441	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		442	}
		443	}
		444
		445	if (ecb_expect_false (SvOBJECT (sv)))
		446	{
		447	HV *stash = SvSTASH (sv);
		448	GV *method;
		449
377	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	450	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378	{	451	{
379	dSP;	452	dSP;
380		453
381	ENTER; SAVETMPS; PUSHMARK (SP);	454	ENTER; SAVETMPS; PUSHMARK (SP);
382	// we re-bless the reference to get overload and other niceties right	455	// we re-bless the reference to get overload and other niceties right
…		…
414	// catch this surprisingly common error	487	// catch this surprisingly common error
415	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	488	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	489	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417		490
418	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	491	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419	encode_uint (enc, 0x80, count + 1);	492	encode_uint (enc, MAJOR_ARRAY, count + 1);
420	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	493	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421		494
422	while (count)	495	while (count)
423	encode_sv (enc, SP[1 - count--]);	496	encode_sv (enc, SP[1 - count--]);
424		497
425	PUTBACK;	498	PUTBACK;
…		…
446	{	519	{
447	double nv = SvNVX (sv);	520	double nv = SvNVX (sv);
448		521
449	need (enc, 9);	522	need (enc, 9);
450		523
451	if (ecb_expect_false (nv == (U32)nv))	524	if (ecb_expect_false (nv == (NV)(U32)nv))
452	encode_uint (enc, 0x00, (U32)nv);	525	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453	//TODO: maybe I32?	526	//TODO: maybe I32?
454	else if (ecb_expect_false (nv == (float)nv))	527	else if (ecb_expect_false (nv == (float)nv))
455	{	528	{
456	uint32_t fp = ecb_float_to_binary32 (nv);	529	uint32_t fp = ecb_float_to_binary32 (nv);
457		530
458	*enc->cur++ = 0xe0 \| 26;	531	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
459		532
460	if (!ecb_big_endian ())	533	if (!ecb_big_endian ())
461	fp = ecb_bswap32 (fp);	534	fp = ecb_bswap32 (fp);
462		535
463	memcpy (enc->cur, &fp, 4);	536	memcpy (enc->cur, &fp, 4);
…		…
465	}	538	}
466	else	539	else
467	{	540	{
468	uint64_t fp = ecb_double_to_binary64 (nv);	541	uint64_t fp = ecb_double_to_binary64 (nv);
469		542
470	*enc->cur++ = 0xe0 \| 27;	543	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
471		544
472	if (!ecb_big_endian ())	545	if (!ecb_big_endian ())
473	fp = ecb_bswap64 (fp);	546	fp = ecb_bswap64 (fp);
474		547
475	memcpy (enc->cur, &fp, 8);	548	memcpy (enc->cur, &fp, 8);
…		…
484		557
485	if (SvPOKp (sv))	558	if (SvPOKp (sv))
486	{	559	{
487	STRLEN len;	560	STRLEN len;
488	char *str = SvPV (sv, len);	561	char *str = SvPV (sv, len);
489	encode_str (enc, SvUTF8 (sv), str, len);	562	encode_strref (enc, SvUTF8 (sv), str, len);
490	}	563	}
491	else if (SvNOKp (sv))	564	else if (SvNOKp (sv))
492	encode_nv (enc, sv);	565	encode_nv (enc, sv);
493	else if (SvIOKp (sv))	566	else if (SvIOKp (sv))
494	{	567	{
495	if (SvIsUV (sv))	568	if (SvIsUV (sv))
496	encode_uint (enc, 0x00, SvUVX (sv));	569	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497	else if (SvIVX (sv) >= 0)	570	else if (SvIVX (sv) >= 0)
498	encode_uint (enc, 0x00, SvIVX (sv));	571	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499	else	572	else
500	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	573	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501	}	574	}
502	else if (SvROK (sv))	575	else if (SvROK (sv))
503	encode_rv (enc, SvRV (sv));	576	encode_rv (enc, SvRV (sv));
504	else if (!SvOK (sv))	577	else if (!SvOK (sv))
505	encode_ch (enc, 0xe0 \| 22);	578	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
506	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	579	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507	encode_ch (enc, 0xe0 \| 23);	580	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
508	else	581	else
509	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	582	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	583	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511	}	584	}
512		585
513	static SV *	586	static SV *
514	encode_cbor (SV scalar, CBOR cbor)	587	encode_cbor (SV scalar, CBOR cbor)
515	{	588	{
516	enc_t enc = { };	589	enc_t enc = { 0 };
517		590
518	enc.cbor = *cbor;	591	enc.cbor = *cbor;
519	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	592	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520	enc.cur = SvPVX (enc.sv);	593	enc.cur = SvPVX (enc.sv);
521	enc.end = SvEND (enc.sv);	594	enc.end = SvEND (enc.sv);
522		595
523	SvPOK_only (enc.sv);	596	SvPOK_only (enc.sv);
524		597
525	if (cbor->flags & F_ALLOW_STRINGREF)	598	if (cbor->flags & F_PACK_STRINGS)
526	{	599	{
527	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	600	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());	601	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
529	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());	602	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
530	}	603	}
…		…
565	#define DEC_DEC_DEPTH --dec->depth	638	#define DEC_DEC_DEPTH --dec->depth
566		639
567	static UV	640	static UV
568	decode_uint (dec_t *dec)	641	decode_uint (dec_t *dec)
569	{	642	{
570	switch (*dec->cur & 31)	643	U8 m = *dec->cur & MINOR_MASK;
571	{	644	++dec->cur;
572	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575	return *dec->cur++ & 31;
576		645
577	case 24:	646	if (ecb_expect_true (m < LENGTH_EXT1))
		647	return m;
		648	else if (ecb_expect_true (m == LENGTH_EXT1))
		649	{
578	WANT (2);	650	WANT (1);
579	dec->cur += 2;	651	dec->cur += 1;
580	return dec->cur[-1];	652	return dec->cur[-1];
581		653	}
582	case 25:	654	else if (ecb_expect_true (m == LENGTH_EXT2))
		655	{
583	WANT (3);	656	WANT (2);
584	dec->cur += 3;	657	dec->cur += 2;
585	return (((UV)dec->cur[-2]) << 8)	658	return (((UV)dec->cur[-2]) << 8)
586	\| ((UV)dec->cur[-1]);	659	\| ((UV)dec->cur[-1]);
587		660	}
588	case 26:	661	else if (ecb_expect_true (m == LENGTH_EXT4))
		662	{
589	WANT (5);	663	WANT (4);
590	dec->cur += 5;	664	dec->cur += 4;
591	return (((UV)dec->cur[-4]) << 24)	665	return (((UV)dec->cur[-4]) << 24)
592	\| (((UV)dec->cur[-3]) << 16)	666	\| (((UV)dec->cur[-3]) << 16)
593	\| (((UV)dec->cur[-2]) << 8)	667	\| (((UV)dec->cur[-2]) << 8)
594	\| ((UV)dec->cur[-1]);	668	\| ((UV)dec->cur[-1]);
595		669	}
596	case 27:	670	else if (ecb_expect_true (m == LENGTH_EXT8))
		671	{
597	WANT (9);	672	WANT (8);
598	dec->cur += 9;	673	dec->cur += 8;
		674
		675	return
		676	#if UVSIZE < 8
		677	0
		678	#else
599	return (((UV)dec->cur[-8]) << 56)	679	(((UV)dec->cur[-8]) << 56)
600	\| (((UV)dec->cur[-7]) << 48)	680	\| (((UV)dec->cur[-7]) << 48)
601	\| (((UV)dec->cur[-6]) << 40)	681	\| (((UV)dec->cur[-6]) << 40)
602	\| (((UV)dec->cur[-5]) << 32)	682	\| (((UV)dec->cur[-5]) << 32)
		683	#endif
603	\| (((UV)dec->cur[-4]) << 24)	684	\| (((UV)dec->cur[-4]) << 24)
604	\| (((UV)dec->cur[-3]) << 16)	685	\| (((UV)dec->cur[-3]) << 16)
605	\| (((UV)dec->cur[-2]) << 8)	686	\| (((UV)dec->cur[-2]) << 8)
606	\| ((UV)dec->cur[-1]);	687	\| ((UV)dec->cur[-1]);
607		688	}
608	default:	689	else
609	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	690	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610	}
611		691
612	fail:	692	fail:
613	return 0;	693	return 0;
614	}	694	}
615		695
…		…
620	{	700	{
621	AV *av = newAV ();	701	AV *av = newAV ();
622		702
623	DEC_INC_DEPTH;	703	DEC_INC_DEPTH;
624		704
625	if ((*dec->cur & 31) == 31)	705	if (*dec->cur == (MAJOR_ARRAY \| MINOR_INDEF))
626	{	706	{
627	++dec->cur;	707	++dec->cur;
628		708
629	for (;;)	709	for (;;)
630	{	710	{
631	WANT (1);	711	WANT (1);
632		712
633	if (*dec->cur == (0xe0 \| 31))	713	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
634	{	714	{
635	++dec->cur;	715	++dec->cur;
636	break;	716	break;
637	}	717	}
638		718
…		…
641	}	721	}
642	else	722	else
643	{	723	{
644	int i, len = decode_uint (dec);	724	int i, len = decode_uint (dec);
645		725
		726	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646	av_fill (av, len - 1);	727	av_fill (av, len - 1);
647		728
648	for (i = 0; i < len; ++i)	729	for (i = 0; i < len; ++i)
649	AvARRAY (av)[i] = decode_sv (dec);	730	AvARRAY (av)[i] = decode_sv (dec);
650	}	731	}
…		…
663	{	744	{
664	// for speed reasons, we specialcase single-string	745	// for speed reasons, we specialcase single-string
665	// byte or utf-8 strings as keys, but only when !stringref	746	// byte or utf-8 strings as keys, but only when !stringref
666		747
667	if (ecb_expect_true (!dec->stringref))	748	if (ecb_expect_true (!dec->stringref))
668	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	749	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669	{	750	{
670	I32 len = decode_uint (dec);	751	I32 len = decode_uint (dec);
671	char key = (char )dec->cur;	752	char key = (char )dec->cur;
672		753
		754	WANT (len);
673	dec->cur += len;	755	dec->cur += len;
674		756
675	if (ecb_expect_false (dec->stringref))
676	av_push (dec->stringref, newSVpvn (key, len));
677
678	hv_store (hv, key, len, decode_sv (dec), 0);	757	hv_store (hv, key, len, decode_sv (dec), 0);
679		758
680	return;	759	return;
681	}	760	}
682	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	761	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683	{	762	{
684	I32 len = decode_uint (dec);	763	I32 len = decode_uint (dec);
685	char key = (char )dec->cur;	764	char key = (char )dec->cur;
686		765
		766	WANT (len);
687	dec->cur += len;	767	dec->cur += len;
688		768
689	if (ecb_expect_false (dec->stringref))	769	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	770	if (!is_utf8_string (key, len))
		771	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691		772
692	hv_store (hv, key, -len, decode_sv (dec), 0);	773	hv_store (hv, key, -len, decode_sv (dec), 0);
693		774
694	return;	775	return;
695	}	776	}
…		…
697	SV *k = decode_sv (dec);	778	SV *k = decode_sv (dec);
698	SV *v = decode_sv (dec);	779	SV *v = decode_sv (dec);
699		780
700	hv_store_ent (hv, k, v, 0);	781	hv_store_ent (hv, k, v, 0);
701	SvREFCNT_dec (k);	782	SvREFCNT_dec (k);
		783
		784	fail:
		785	;
702	}	786	}
703		787
704	static SV *	788	static SV *
705	decode_hv (dec_t *dec)	789	decode_hv (dec_t *dec)
706	{	790	{
707	HV *hv = newHV ();	791	HV *hv = newHV ();
708		792
709	DEC_INC_DEPTH;	793	DEC_INC_DEPTH;
710		794
711	if ((*dec->cur & 31) == 31)	795	if (*dec->cur == (MAJOR_MAP \| MINOR_INDEF))
712	{	796	{
713	++dec->cur;	797	++dec->cur;
714		798
715	for (;;)	799	for (;;)
716	{	800	{
717	WANT (1);	801	WANT (1);
718		802
719	if (*dec->cur == (0xe0 \| 31))	803	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
720	{	804	{
721	++dec->cur;	805	++dec->cur;
722	break;	806	break;
723	}	807	}
724		808
…		…
745	static SV *	829	static SV *
746	decode_str (dec_t *dec, int utf8)	830	decode_str (dec_t *dec, int utf8)
747	{	831	{
748	SV *sv = 0;	832	SV *sv = 0;
749		833
750	if ((*dec->cur & 31) == 31)	834	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751	{	835	{
		836	// indefinite length strings
752	++dec->cur;	837	++dec->cur;
753		838
		839	U8 major = *dec->cur & MAJOR_MISC;
		840
754	sv = newSVpvn ("", 0);	841	sv = newSVpvn ("", 0);
755		842
756	// not very fast, and certainly not robust against illegal input
757	for (;;)	843	for (;;)
758	{	844	{
759	WANT (1);	845	WANT (1);
760		846
761	if (*dec->cur == (0xe0 \| 31))	847	if ((*dec->cur - major) > LENGTH_EXT8)
		848	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
762	{	849	{
763	++dec->cur;	850	++dec->cur;
764	break;	851	break;
765	}	852	}
		853	else
		854	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766		855
767	sv_catsv (sv, decode_sv (dec));	856	STRLEN len = decode_uint (dec);
		857
		858	WANT (len);
		859	sv_catpvn (sv, dec->cur, len);
		860	dec->cur += len;
768	}	861	}
769	}	862	}
770	else	863	else
771	{	864	{
772	STRLEN len = decode_uint (dec);	865	STRLEN len = decode_uint (dec);
…		…
779	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	872	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	873	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781	}	874	}
782		875
783	if (utf8)	876	if (utf8)
		877	{
		878	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		879	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		880	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		881
784	SvUTF8_on (sv);	882	SvUTF8_on (sv);
		883	}
785		884
786	return sv;	885	return sv;
787		886
788	fail:	887	fail:
789	SvREFCNT_dec (sv);	888	SvREFCNT_dec (sv);
…		…
821	}	920	}
822	break;	921	break;
823		922
824	case CBOR_TAG_STRINGREF:	923	case CBOR_TAG_STRINGREF:
825	{	924	{
826	if ((*dec->cur >> 5) != 0)	925	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	926	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828		927
829	UV idx = decode_uint (dec);	928	UV idx = decode_uint (dec);
830		929
831	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	930	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
…		…
838	case CBOR_TAG_VALUE_SHAREABLE:	937	case CBOR_TAG_VALUE_SHAREABLE:
839	{	938	{
840	if (ecb_expect_false (!dec->shareable))	939	if (ecb_expect_false (!dec->shareable))
841	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	940	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
842		941
		942	if (dec->cbor.flags & F_ALLOW_CYCLES)
		943	{
843	sv = newSV (0);	944	sv = newSV (0);
844	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	945	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845		946
846	SV *osv = decode_sv (dec);	947	SV *osv = decode_sv (dec);
847	sv_setsv (sv, osv);	948	sv_setsv (sv, osv);
848	SvREFCNT_dec_NN (osv);	949	SvREFCNT_dec_NN (osv);
		950	}
		951	else
		952	{
		953	av_push (dec->shareable, &PL_sv_undef);
		954	int idx = AvFILLp (dec->shareable);
		955	sv = decode_sv (dec);
		956	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		957	}
849	}	958	}
850	break;	959	break;
851		960
852	case CBOR_TAG_VALUE_SHAREDREF:	961	case CBOR_TAG_VALUE_SHAREDREF:
853	{	962	{
854	if ((*dec->cur >> 5) != 0)	963	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	964	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856		965
857	UV idx = decode_uint (dec);	966	UV idx = decode_uint (dec);
858		967
859	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	968	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
860	ERR ("corrupted CBOR data (sharedref index out of bounds)");	969	ERR ("corrupted CBOR data (sharedref index out of bounds)");
861		970
862	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	971	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		972
		973	if (sv == &PL_sv_undef)
		974	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863	}	975	}
864	break;	976	break;
865		977
866	case CBOR_TAG_PERL_OBJECT:	978	case CBOR_TAG_PERL_OBJECT:
867	{	979	{
…		…
968	static SV *	1080	static SV *
969	decode_sv (dec_t *dec)	1081	decode_sv (dec_t *dec)
970	{	1082	{
971	WANT (1);	1083	WANT (1);
972		1084
973	switch (*dec->cur >> 5)	1085	switch (*dec->cur >> MAJOR_SHIFT)
974	{	1086	{
975	case 0: // unsigned int	1087	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976	return newSVuv (decode_uint (dec));	1088	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977	case 1: // negative int	1089	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978	return newSViv (-1 - (IV)decode_uint (dec));	1090	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979	case 2: // octet string	1091	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980	return decode_str (dec, 0);	1092	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981	case 3: // utf-8 string	1093	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982	return decode_str (dec, 1);	1094
983	case 4: // array	1095	case MAJOR_MISC >> MAJOR_SHIFT:
984	return decode_av (dec);
985	case 5: // map
986	return decode_hv (dec);
987	case 6: // tag
988	return decode_tagged (dec);
989	case 7: // misc
990	switch (*dec->cur++ & 31)	1096	switch (*dec->cur++ & MINOR_MASK)
991	{	1097	{
992	case 20:	1098	case SIMPLE_FALSE:
993	#if CBOR_SLOW	1099	#if CBOR_SLOW
994	types_false = get_bool ("Types::Serialiser::false");	1100	types_false = get_bool ("Types::Serialiser::false");
995	#endif	1101	#endif
996	return newSVsv (types_false);	1102	return newSVsv (types_false);
997	case 21:	1103	case SIMPLE_TRUE:
998	#if CBOR_SLOW	1104	#if CBOR_SLOW
999	types_true = get_bool ("Types::Serialiser::true");	1105	types_true = get_bool ("Types::Serialiser::true");
1000	#endif	1106	#endif
1001	return newSVsv (types_true);	1107	return newSVsv (types_true);
1002	case 22:	1108	case SIMPLE_NULL:
1003	return newSVsv (&PL_sv_undef);	1109	return newSVsv (&PL_sv_undef);
1004	case 23:	1110	case SIMPLE_UNDEF:
1005	#if CBOR_SLOW	1111	#if CBOR_SLOW
1006	types_error = get_bool ("Types::Serialiser::error");	1112	types_error = get_bool ("Types::Serialiser::error");
1007	#endif	1113	#endif
1008	return newSVsv (types_error);	1114	return newSVsv (types_error);
1009		1115
1010	case 25:	1116	case MISC_FLOAT16:
1011	{	1117	{
1012	WANT (2);	1118	WANT (2);
1013		1119
1014	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];	1120	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];
1015	dec->cur += 2;	1121	dec->cur += 2;
1016		1122
1017	return newSVnv (ecb_binary16_to_float (fp));	1123	return newSVnv (ecb_binary16_to_float (fp));
1018	}	1124	}
1019		1125
1020	case 26:	1126	case MISC_FLOAT32:
1021	{	1127	{
1022	uint32_t fp;	1128	uint32_t fp;
1023	WANT (4);	1129	WANT (4);
1024	memcpy (&fp, dec->cur, 4);	1130	memcpy (&fp, dec->cur, 4);
1025	dec->cur += 4;	1131	dec->cur += 4;
…		…
1028	fp = ecb_bswap32 (fp);	1134	fp = ecb_bswap32 (fp);
1029		1135
1030	return newSVnv (ecb_binary32_to_float (fp));	1136	return newSVnv (ecb_binary32_to_float (fp));
1031	}	1137	}
1032		1138
1033	case 27:	1139	case MISC_FLOAT64:
1034	{	1140	{
1035	uint64_t fp;	1141	uint64_t fp;
1036	WANT (8);	1142	WANT (8);
1037	memcpy (&fp, dec->cur, 8);	1143	memcpy (&fp, dec->cur, 8);
1038	dec->cur += 8;	1144	dec->cur += 8;
…		…
1041	fp = ecb_bswap64 (fp);	1147	fp = ecb_bswap64 (fp);
1042		1148
1043	return newSVnv (ecb_binary64_to_double (fp));	1149	return newSVnv (ecb_binary64_to_double (fp));
1044	}	1150	}
1045		1151
1046	// 0..19 unassigned	1152	// 0..19 unassigned simple
1047	// 24 reserved + unassigned (reserved values are not encodable)	1153	// 24 reserved + unassigned simple (reserved values are not encodable)
		1154	// 28-30 unassigned misc
		1155	// 31 break code
1048	default:	1156	default:
1049	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1157	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050	}	1158	}
1051		1159
1052	break;	1160	break;
1053	}	1161	}
1054		1162
…		…
1057	}	1165	}
1058		1166
1059	static SV *	1167	static SV *
1060	decode_cbor (SV string, CBOR cbor, char **offset_return)	1168	decode_cbor (SV string, CBOR cbor, char **offset_return)
1061	{	1169	{
1062	dec_t dec = { };	1170	dec_t dec = { 0 };
1063	SV *sv;	1171	SV *sv;
1064	STRLEN len;	1172	STRLEN len;
1065	char *data = SvPVbyte (string, len);	1173	char *data = SvPVbyte (string, len);
1066		1174
1067	if (len > cbor->max_size && cbor->max_size)	1175	if (len > cbor->max_size && cbor->max_size)
…		…
1081	if (dec.cur != dec.end && !dec.err)	1189	if (dec.cur != dec.end && !dec.err)
1082	dec.err = "garbage after CBOR object";	1190	dec.err = "garbage after CBOR object";
1083		1191
1084	if (dec.err)	1192	if (dec.err)
1085	{	1193	{
		1194	if (dec.shareable)
		1195	{
		1196	// need to break cyclic links, which whould all be in shareable
		1197	int i;
		1198	SV **svp;
		1199
		1200	for (i = av_len (dec.shareable) + 1; i--; )
		1201	if ((svp = av_fetch (dec.shareable, i, 0)))
		1202	sv_setsv (*svp, &PL_sv_undef);
		1203	}
		1204
1086	SvREFCNT_dec (sv);	1205	SvREFCNT_dec (sv);
1087	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1206	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1088	}	1207	}
1089		1208
1090	sv = sv_2mortal (sv);	1209	sv = sv_2mortal (sv);
1091		1210
1092	return sv;	1211	return sv;
1093	}	1212	}
1094		1213
		1214	/////////////////////////////////////////////////////////////////////////////
		1215	// incremental parser
		1216
		1217	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1218
		1219	// returns 0 for notyet, 1 for success or error
		1220	static int
		1221	incr_parse (CBOR self, SV cborstr)
		1222	{
		1223	STRLEN cur;
		1224	SvPV (cborstr, cur);
		1225
		1226	while (ecb_expect_true (self->incr_need <= cur))
		1227	{
		1228	// table of integer count bytes
		1229	static I8 incr_len[MINOR_MASK + 1] = {
		1230	0, 0, 0, 0, 0, 0, 0, 0,
		1231	0, 0, 0, 0, 0, 0, 0, 0,
		1232	0, 0, 0, 0, 0, 0, 0, 0,
		1233	1, 2, 4, 8,-1,-1,-1,-2
		1234	};
		1235
		1236	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1237	U8 m = *p & MINOR_MASK;
		1238	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1239	I8 ilen = incr_len[m];
		1240
		1241	self->incr_need = self->incr_pos + 1;
		1242
		1243	if (ecb_expect_false (ilen < 0))
		1244	{
		1245	if (m != MINOR_INDEF)
		1246	return 1; // error
		1247
		1248	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1249	{
		1250	if (count >= 0)
		1251	return 1; // error
		1252
		1253	count = 1;
		1254	}
		1255	else
		1256	{
		1257	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1258	count = -1;
		1259	}
		1260	}
		1261	else
		1262	{
		1263	self->incr_need += ilen;
		1264	if (ecb_expect_false (self->incr_need > cur))
		1265	return 0;
		1266
		1267	int major = *p >> MAJOR_SHIFT;
		1268
		1269	switch (major)
		1270	{
		1271	case MAJOR_TAG >> MAJOR_SHIFT:
		1272	++count; // tags merely prefix another value
		1273	break;
		1274
		1275	case MAJOR_BYTES >> MAJOR_SHIFT:
		1276	case MAJOR_TEXT >> MAJOR_SHIFT:
		1277	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1278	case MAJOR_MAP >> MAJOR_SHIFT:
		1279	{
		1280	UV len;
		1281
		1282	if (ecb_expect_false (ilen))
		1283	{
		1284	len = 0;
		1285
		1286	do {
		1287	len = (len << 8) \| *++p;
		1288	} while (--ilen);
		1289	}
		1290	else
		1291	len = m;
		1292
		1293	switch (major)
		1294	{
		1295	case MAJOR_BYTES >> MAJOR_SHIFT:
		1296	case MAJOR_TEXT >> MAJOR_SHIFT:
		1297	self->incr_need += len;
		1298	if (ecb_expect_false (self->incr_need > cur))
		1299	return 0;
		1300
		1301	break;
		1302
		1303	case MAJOR_MAP >> MAJOR_SHIFT:
		1304	len <<= 1;
		1305	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1306	if (len)
		1307	{
		1308	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1309	count = len + 1;
		1310	}
		1311	break;
		1312	}
		1313	}
		1314	}
		1315	}
		1316
		1317	self->incr_pos = self->incr_need;
		1318
		1319	if (count > 0)
		1320	{
		1321	while (!--count)
		1322	{
		1323	if (!AvFILLp (self->incr_count))
		1324	return 1; // done
		1325
		1326	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1327	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1328	}
		1329
		1330	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1331	}
		1332	}
		1333
		1334	return 0;
		1335	}
		1336
		1337
1095	/////////////////////////////////////////////////////////////////////////////	1338	/////////////////////////////////////////////////////////////////////////////
1096	// XS interface functions	1339	// XS interface functions
1097		1340
1098	MODULE = CBOR::XS PACKAGE = CBOR::XS	1341	MODULE = CBOR::XS PACKAGE = CBOR::XS
1099		1342
…		…
1139	void shrink (CBOR *self, int enable = 1)	1382	void shrink (CBOR *self, int enable = 1)
1140	ALIAS:	1383	ALIAS:
1141	shrink = F_SHRINK	1384	shrink = F_SHRINK
1142	allow_unknown = F_ALLOW_UNKNOWN	1385	allow_unknown = F_ALLOW_UNKNOWN
1143	allow_sharing = F_ALLOW_SHARING	1386	allow_sharing = F_ALLOW_SHARING
1144	allow_stringref = F_ALLOW_STRINGREF	1387	allow_cycles = F_ALLOW_CYCLES
		1388	pack_strings = F_PACK_STRINGS
		1389	validate_utf8 = F_VALIDATE_UTF8
1145	PPCODE:	1390	PPCODE:
1146	{	1391	{
1147	if (enable)	1392	if (enable)
1148	self->flags \|= ix;	1393	self->flags \|= ix;
1149	else	1394	else
…		…
1155	void get_shrink (CBOR *self)	1400	void get_shrink (CBOR *self)
1156	ALIAS:	1401	ALIAS:
1157	get_shrink = F_SHRINK	1402	get_shrink = F_SHRINK
1158	get_allow_unknown = F_ALLOW_UNKNOWN	1403	get_allow_unknown = F_ALLOW_UNKNOWN
1159	get_allow_sharing = F_ALLOW_SHARING	1404	get_allow_sharing = F_ALLOW_SHARING
1160	get_allow_stringref = F_ALLOW_STRINGREF	1405	get_allow_cycles = F_ALLOW_CYCLES
		1406	get_pack_strings = F_PACK_STRINGS
		1407	get_validate_utf8 = F_VALIDATE_UTF8
1161	PPCODE:	1408	PPCODE:
1162	XPUSHs (boolSV (self->flags & ix));	1409	XPUSHs (boolSV (self->flags & ix));
1163		1410
1164	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1411	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165	PPCODE:	1412	PPCODE:
…		…
1214	EXTEND (SP, 2);	1461	EXTEND (SP, 2);
1215	PUSHs (sv);	1462	PUSHs (sv);
1216	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1463	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217	}	1464	}
1218		1465
		1466	void incr_parse (CBOR self, SV cborstr)
		1467	ALIAS:
		1468	incr_parse_multiple = 1
		1469	PPCODE:
		1470	{
		1471	if (SvUTF8 (cborstr))
		1472	sv_utf8_downgrade (cborstr, 0);
		1473
		1474	if (!self->incr_count)
		1475	{
		1476	self->incr_count = newAV ();
		1477	self->incr_pos = 0;
		1478	self->incr_need = 1;
		1479
		1480	av_push (self->incr_count, newSViv (1));
		1481	}
		1482
		1483	do
		1484	{
		1485	if (!incr_parse (self, cborstr))
		1486	{
		1487	if (self->incr_need > self->max_size && self->max_size)
		1488	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1489	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1490
		1491	break;
		1492	}
		1493
		1494	SV *sv;
		1495	char *offset;
		1496
		1497	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1498	XPUSHs (sv);
		1499
		1500	sv_chop (cborstr, offset);
		1501
		1502	av_clear (self->incr_count);
		1503	av_push (self->incr_count, newSViv (1));
		1504
		1505	self->incr_pos = 0;
		1506	self->incr_need = self->incr_pos + 1;
		1507	}
		1508	while (ix);
		1509	}
		1510
		1511	void incr_reset (CBOR *self)
		1512	CODE:
		1513	{
		1514	SvREFCNT_dec (self->incr_count);
		1515	self->incr_count = 0;
		1516	}
		1517
1219	void DESTROY (CBOR *self)	1518	void DESTROY (CBOR *self)
1220	PPCODE:	1519	PPCODE:
1221	cbor_free (self);	1520	cbor_free (self);
1222		1521
1223	PROTOTYPES: ENABLE	1522	PROTOTYPES: ENABLE
1224		1523
1225	void encode_cbor (SV *scalar)	1524	void encode_cbor (SV *scalar)
		1525	ALIAS:
		1526	encode_cbor = 0
		1527	encode_cbor_sharing = F_ALLOW_SHARING
1226	PPCODE:	1528	PPCODE:
1227	{	1529	{
1228	CBOR cbor;	1530	CBOR cbor;
1229	cbor_init (&cbor);	1531	cbor_init (&cbor);
		1532	cbor.flags \|= ix;
1230	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1533	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231	XPUSHs (scalar);	1534	XPUSHs (scalar);
1232	}	1535	}
1233		1536
1234	void decode_cbor (SV *cborstr)	1537	void decode_cbor (SV *cborstr)

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs. Revision 1.49 by root, Mon Feb 8 04:11:11 2016 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.49 by root, Mon Feb 8 04:11:11 2016 UTC