ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.65 by root, Thu Nov 15 19:52:41 2018 UTC

…		…
6	#include <string.h>	6	#include <string.h>
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	#include <inttypes.h>
11		12
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
		29	#ifndef SvREFCNT_dec_NN
		30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		31	#endif
		32
		33	// known major and minor types
		34	enum cbor_type
		35	{
		36	MAJOR_SHIFT = 5,
		37	MINOR_MASK = 0x1f,
		38
		39	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		40	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		41	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		42	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		43	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		44	MAJOR_MAP = 5 << MAJOR_SHIFT,
		45	MAJOR_TAG = 6 << MAJOR_SHIFT,
		46	MAJOR_MISC = 7 << MAJOR_SHIFT,
		47
		48	// INT/STRING/ARRAY/MAP subtypes
		49	LENGTH_EXT1 = 24,
		50	LENGTH_EXT2 = 25,
		51	LENGTH_EXT4 = 26,
		52	LENGTH_EXT8 = 27,
		53
		54	// SIMPLE types (effectively MISC subtypes)
		55	SIMPLE_FALSE = 20,
		56	SIMPLE_TRUE = 21,
		57	SIMPLE_NULL = 22,
		58	SIMPLE_UNDEF = 23,
		59
		60	// MISC subtype (unused)
		61	MISC_EXT1 = 24,
		62	MISC_FLOAT16 = 25,
		63	MISC_FLOAT32 = 26,
		64	MISC_FLOAT64 = 27,
		65
		66	// BYTES/TEXT/ARRAY/MAP
		67	MINOR_INDEF = 31,
		68	};
24		69
25	// known tags	70	// known tags
26	enum cbor_tag	71	enum cbor_tag
27	{	72	{
28	// inofficial extensions (pending iana registration)	73	// extensions
		74	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	75	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	76	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	77	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	78	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	79	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	80	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36		81
37	// rfc7049	82	// rfc7049
38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	83	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	84	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40	CBOR_TAG_POS_BIGNUM = 2, // byte string	85	CBOR_TAG_POS_BIGNUM = 2, // byte string
41	CBOR_TAG_NEG_BIGNUM = 3, // byte string	86	CBOR_TAG_NEG_BIGNUM = 3, // byte string
42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	87	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43	CBOR_TAG_BIGFLOAT = 5, // array	88	CBOR_TAG_BIGFLOAT = 5, // array
44		89
45	CBOR_TAG_CONV_B64U = 21, // base64url, any	90	CBOR_TAG_CONV_B64U = 21, // base64url, any
46	CBOR_TAG_CONV_B64 = 22, // base64, any	91	CBOR_TAG_CONV_B64 = 22, // base64, any
47	CBOR_TAG_CONV_HEX = 23, // base16, any	92	CBOR_TAG_CONV_HEX = 23, // base16, any
48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	93	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49		94
50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	95	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	96	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	97	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	98	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55		100
56	CBOR_TAG_MAGIC = 55799 // self-describe cbor	101	CBOR_TAG_MAGIC = 55799, // self-describe cbor
57	};	102	};
58		103
59	#define F_SHRINK 0x00000001UL	104	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	105	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	106	#define F_ALLOW_SHARING 0x00000004UL
		107	#define F_ALLOW_CYCLES 0x00000008UL
		108	#define F_FORBID_OBJECTS 0x00000010UL
62	#define F_DEDUP_STRINGS 0x00000008UL //TODO	109	#define F_PACK_STRINGS 0x00000020UL
63	#define F_DEDUP_KEYS 0x00000010UL //TODO	110	#define F_TEXT_KEYS 0x00000040UL
		111	#define F_TEXT_STRINGS 0x00000080UL
		112	#define F_VALIDATE_UTF8 0x00000100UL
64		113
65	#define INIT_SIZE 32 // initial scalar size to be allocated	114	#define INIT_SIZE 32 // initial scalar size to be allocated
66		115
67	#define SB do {	116	#define SB do {
68	#define SE } while (0)	117	#define SE } while (0)
…		…
80	# define CBOR_SLOW 0	129	# define CBOR_SLOW 0
81	# define CBOR_STASH cbor_stash	130	# define CBOR_STASH cbor_stash
82	#endif	131	#endif
83		132
84	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::	133	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85	static SV *types_true, *types_false, *types_error, *sv_cbor;	134	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86		135
87	typedef struct {	136	typedef struct {
88	U32 flags;	137	U32 flags;
89	U32 max_depth;	138	U32 max_depth;
90	STRLEN max_size;	139	STRLEN max_size;
		140	SV *filter;
		141
		142	// for the incremental parser
		143	STRLEN incr_pos; // the current offset into the text
		144	STRLEN incr_need; // minimum bytes needed to decode
		145	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	146	} CBOR;
92		147
93	ecb_inline void	148	ecb_inline void
94	cbor_init (CBOR *cbor)	149	cbor_init (CBOR *cbor)
95	{	150	{
96	Zero (cbor, 1, CBOR);	151	Zero (cbor, 1, CBOR);
97	cbor->max_depth = 512;	152	cbor->max_depth = 512;
		153	}
		154
		155	ecb_inline void
		156	cbor_free (CBOR *cbor)
		157	{
		158	SvREFCNT_dec (cbor->filter);
		159	SvREFCNT_dec (cbor->incr_count);
98	}	160	}
99		161
100	/////////////////////////////////////////////////////////////////////////////	162	/////////////////////////////////////////////////////////////////////////////
101	// utility functions	163	// utility functions
102		164
…		…
122	SvPV_shrink_to_cur (sv);	184	SvPV_shrink_to_cur (sv);
123	#elif defined (SvPV_renew)	185	#elif defined (SvPV_renew)
124	SvPV_renew (sv, SvCUR (sv) + 1);	186	SvPV_renew (sv, SvCUR (sv) + 1);
125	#endif	187	#endif
126	}	188	}
		189	}
		190
		191	// minimum length of a string to be registered for stringref
		192	ecb_inline int
		193	minimum_string_length (UV idx)
		194	{
		195	return idx <= 23 ? 3
		196	: idx <= 0xffU ? 4
		197	: idx <= 0xffffU ? 5
		198	: idx <= 0xffffffffU ? 7
		199	: 11;
127	}	200	}
128		201
129	/////////////////////////////////////////////////////////////////////////////	202	/////////////////////////////////////////////////////////////////////////////
130	// encoder	203	// encoder
131		204
…		…
144	} enc_t;	217	} enc_t;
145		218
146	ecb_inline void	219	ecb_inline void
147	need (enc_t *enc, STRLEN len)	220	need (enc_t *enc, STRLEN len)
148	{	221	{
149	if (ecb_expect_false (enc->cur + len >= enc->end))	222	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150	{	223	{
151	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	224	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	225	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153	enc->cur = SvPVX (enc->sv) + cur;	226	enc->cur = SvPVX (enc->sv) + cur;
154	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	227	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
165	static void	238	static void
166	encode_uint (enc_t *enc, int major, UV len)	239	encode_uint (enc_t *enc, int major, UV len)
167	{	240	{
168	need (enc, 9);	241	need (enc, 9);
169		242
170	if (len < 24)	243	if (ecb_expect_true (len < LENGTH_EXT1))
171	*enc->cur++ = major | len;	244	*enc->cur++ = major | len;
172	else if (len <= 0xff)	245	else if (ecb_expect_true (len <= 0xffU))
173	{	246	{
174	*enc->cur++ = major | 24;	247	*enc->cur++ = major | LENGTH_EXT1;
175	*enc->cur++ = len;	248	*enc->cur++ = len;
176	}	249	}
177	else if (len <= 0xffff)	250	else if (len <= 0xffffU)
178	{	251	{
179	*enc->cur++ = major | 25;	252	*enc->cur++ = major | LENGTH_EXT2;
180	*enc->cur++ = len >> 8;	253	*enc->cur++ = len >> 8;
181	*enc->cur++ = len;	254	*enc->cur++ = len;
182	}	255	}
183	else if (len <= 0xffffffff)	256	else if (len <= 0xffffffffU)
184	{	257	{
185	*enc->cur++ = major | 26;	258	*enc->cur++ = major | LENGTH_EXT4;
186	*enc->cur++ = len >> 24;	259	*enc->cur++ = len >> 24;
187	*enc->cur++ = len >> 16;	260	*enc->cur++ = len >> 16;
188	*enc->cur++ = len >> 8;	261	*enc->cur++ = len >> 8;
189	*enc->cur++ = len;	262	*enc->cur++ = len;
190	}	263	}
191	else	264	else
192	{	265	{
193	*enc->cur++ = major | 27;	266	*enc->cur++ = major | LENGTH_EXT8;
194	*enc->cur++ = len >> 56;	267	*enc->cur++ = len >> 56;
195	*enc->cur++ = len >> 48;	268	*enc->cur++ = len >> 48;
196	*enc->cur++ = len >> 40;	269	*enc->cur++ = len >> 40;
197	*enc->cur++ = len >> 32;	270	*enc->cur++ = len >> 32;
198	*enc->cur++ = len >> 24;	271	*enc->cur++ = len >> 24;
…		…
200	*enc->cur++ = len >> 8;	273	*enc->cur++ = len >> 8;
201	*enc->cur++ = len;	274	*enc->cur++ = len;
202	}	275	}
203	}	276	}
204		277
205	static void	278	ecb_inline void
		279	encode_tag (enc_t *enc, UV tag)
		280	{
		281	encode_uint (enc, MAJOR_TAG, tag);
		282	}
		283
		284	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		285	ecb_noinline static void
206	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	286	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207	{	287	{
208	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	288	STRLEN ulen = len;
		289	U8 *p, *pend = (U8 *)str + len;
		290
		291	for (p = (U8 *)str; p < pend; ++p)
		292	ulen += *p >> 7; // count set high bits
		293
		294	encode_uint (enc, MAJOR_TEXT, ulen);
		295
		296	need (enc, ulen);
		297	for (p = (U8 *)str; p < pend; ++p)
		298	if (*p < 0x80)
		299	*enc->cur++ = *p;
		300	else
		301	{
		302	*enc->cur++ = 0xc0 + (*p >> 6);
		303	*enc->cur++ = 0x80 + (*p & 63);
		304	}
		305	}
		306
		307	ecb_inline void
		308	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		309	{
		310	if (ecb_expect_false (upgrade_utf8))
		311	if (!utf8)
		312	{
		313	encode_str_utf8 (enc, utf8, str, len);
		314	return;
		315	}
		316
		317	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
209	need (enc, len);	318	need (enc, len);
210	memcpy (enc->cur, str, len);	319	memcpy (enc->cur, str, len);
211	enc->cur += len;	320	enc->cur += len;
212	}	321	}
213		322
214	ecb_inline void	323	ecb_inline void
215	encode_tag (enc_t *enc, UV tag)	324	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
216	{	325	{
217	encode_uint (enc, 0xc0, tag);	326	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		327	{
		328	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		329
		330	if (SvOK (*svp))
		331	{
		332	// already registered, use stringref
		333	encode_tag (enc, CBOR_TAG_STRINGREF);
		334	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		335	return;
		336	}
		337	else if (len >= minimum_string_length (enc->stringref_idx))
		338	{
		339	// register only
		340	sv_setuv (*svp, enc->stringref_idx);
		341	++enc->stringref_idx;
		342	}
		343	}
		344
		345	encode_str (enc, upgrade_utf8, utf8, str, len);
218	}	346	}
219		347
220	static void encode_sv (enc_t *enc, SV *sv);	348	static void encode_sv (enc_t *enc, SV *sv);
221		349
222	static void	350	static void
…		…
227	if (enc->depth >= enc->cbor.max_depth)	355	if (enc->depth >= enc->cbor.max_depth)
228	croak (ERR_NESTING_EXCEEDED);	356	croak (ERR_NESTING_EXCEEDED);
229		357
230	++enc->depth;	358	++enc->depth;
231		359
232	encode_uint (enc, 0x80, len + 1);	360	encode_uint (enc, MAJOR_ARRAY, len + 1);
233		361
		362	if (ecb_expect_false (SvMAGICAL (av)))
234	for (i = 0; i <= len; ++i)	363	for (i = 0; i <= len; ++i)
235	{	364	{
236	SV **svp = av_fetch (av, i, 0);	365	SV **svp = av_fetch (av, i, 0);
237	encode_sv (enc, svp ? *svp : &PL_sv_undef);	366	encode_sv (enc, svp ? *svp : &PL_sv_undef);
238	}	367	}
		368	else
		369	for (i = 0; i <= len; ++i)
		370	{
		371	SV *sv = AvARRAY (av)[i];
		372	encode_sv (enc, sv ? sv : &PL_sv_undef);
		373	}
239		374
240	--enc->depth;	375	--enc->depth;
241	}
242
243	ecb_inline void
244	encode_he (enc_t *enc, HE *he)
245	{
246	if (HeKLEN (he) == HEf_SVKEY)
247	encode_sv (enc, HeSVKEY (he));
248	else
249	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250	}	376	}
251		377
252	static void	378	static void
253	encode_hv (enc_t *enc, HV *hv)	379	encode_hv (enc_t *enc, HV *hv)
254	{	380	{
…		…
260	++enc->depth;	386	++enc->depth;
261		387
262	int pairs = hv_iterinit (hv);	388	int pairs = hv_iterinit (hv);
263	int mg = SvMAGICAL (hv);	389	int mg = SvMAGICAL (hv);
264		390
265	if (mg)	391	if (ecb_expect_false (mg))
266	encode_ch (enc, 0xa0 | 31);	392	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267	else	393	else
268	encode_uint (enc, 0xa0, pairs);	394	encode_uint (enc, MAJOR_MAP, pairs);
269		395
270	while ((he = hv_iternext (hv)))	396	while ((he = hv_iternext (hv)))
271	{	397	{
272	if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273	{
274	SV **svp;
275
276	if (HeKLEN (he) == HEf_SVKEY)	398	if (HeKLEN (he) == HEf_SVKEY)
277	svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/	399	encode_sv (enc, HeSVKEY (he));
278	else
279	svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281	if (SvOK (*svp))
282	{
283	encode_tag (enc, CBOR_TAG_STRINGREF);
284	encode_uint (enc, 0x00, SvUV (*svp));
285	}
286	else
287	{
288	sv_setuv (*svp, enc->stringref_idx);
289	++enc->stringref_idx;
290	encode_he (enc, he);
291	}
292	}
293	else	400	else
294	encode_he (enc, he);	401	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295		402
296	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	403	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
297	}	404	}
298		405
299	if (mg)	406	if (ecb_expect_false (mg))
300	encode_ch (enc, 0xe0 | 31);	407	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301		408
302	--enc->depth;	409	--enc->depth;
303	}	410	}
304		411
305	// encode objects, arrays and special \0=false and \1=true values.	412	// encode objects, arrays and special \0=false and \1=true values.
306	static void	413	static void
307	encode_rv (enc_t *enc, SV *sv)	414	encode_rv (enc_t *enc, SV *sv)
308	{	415	{
309	SvGETMAGIC (sv);	416	SvGETMAGIC (sv);
310
311	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312	&& ecb_expect_false (SvREFCNT (sv) > 1))
313	{
314	if (!enc->shareable)
315	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319	if (SvOK (*svp))
320	{
321	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322	encode_uint (enc, 0x00, SvUV (*svp));
323	return;
324	}
325	else
326	{
327	sv_setuv (*svp, enc->shareable_idx);
328	++enc->shareable_idx;
329	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330	}
331	}
332		417
333	svtype svt = SvTYPE (sv);	418	svtype svt = SvTYPE (sv);
334		419
335	if (ecb_expect_false (SvOBJECT (sv)))	420	if (ecb_expect_false (SvOBJECT (sv)))
336	{	421	{
…		…
343	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	428	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344	? cbor_tagged_stash	429	? cbor_tagged_stash
345	: gv_stashpv ("CBOR::XS::Tagged" , 1);	430	: gv_stashpv ("CBOR::XS::Tagged" , 1);
346		431
347	HV *stash = SvSTASH (sv);	432	HV *stash = SvSTASH (sv);
348	GV *method;
349		433
350	if (stash == boolean_stash)	434	if (stash == boolean_stash)
351	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	435	{
		436	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		437	return;
		438	}
352	else if (stash == error_stash)	439	else if (stash == error_stash)
353	encode_ch (enc, 0xe0 | 23);	440	{
		441	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		442	return;
		443	}
354	else if (stash == tagged_stash)	444	else if (stash == tagged_stash)
355	{	445	{
356	if (svt != SVt_PVAV)	446	if (svt != SVt_PVAV)
357	croak ("encountered CBOR::XS::Tagged object that isn't an array");	447	croak ("encountered CBOR::XS::Tagged object that isn't an array");
358		448
359	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	449	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	450	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		451
		452	return;
		453	}
		454	}
		455
		456	if (ecb_expect_false (SvREFCNT (sv) > 1)
		457	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		458	{
		459	if (ecb_expect_false (!enc->shareable))
		460	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		461
		462	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		463
		464	if (SvOK (*svp))
361	}	465	{
		466	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		467	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		468	return;
		469	}
		470	else
		471	{
		472	sv_setuv (*svp, enc->shareable_idx);
		473	++enc->shareable_idx;
		474	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		475	}
		476	}
		477
		478	if (ecb_expect_false (SvOBJECT (sv)))
		479	{
		480	HV *stash = SvSTASH (sv);
		481	GV *method;
		482
		483	if (enc->cbor.flags & F_FORBID_OBJECTS)
		484	croak ("encountered object '%s', but forbid_objects is enabled",
		485	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
362	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	486	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363	{	487	{
364	dSP;	488	dSP;
365		489
366	ENTER; SAVETMPS; PUSHMARK (SP);	490	ENTER; SAVETMPS;
		491	PUSHMARK (SP);
367	// we re-bless the reference to get overload and other niceties right	492	// we re-bless the reference to get overload and other niceties right
368	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	493	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369		494
370	PUTBACK;	495	PUTBACK;
371	// G_SCALAR ensures that return value is 1	496	// G_SCALAR ensures that return value is 1
…		…
384	}	509	}
385	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	510	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386	{	511	{
387	dSP;	512	dSP;
388		513
389	ENTER; SAVETMPS; PUSHMARK (SP);	514	ENTER; SAVETMPS;
		515	PUSHMARK (SP);
390	EXTEND (SP, 2);	516	EXTEND (SP, 2);
391	// we re-bless the reference to get overload and other niceties right	517	// we re-bless the reference to get overload and other niceties right
392	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	518	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393	PUSHs (sv_cbor);	519	PUSHs (sv_cbor);
394		520
…		…
399	// catch this surprisingly common error	525	// catch this surprisingly common error
400	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	526	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	527	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402		528
403	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	529	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404	encode_uint (enc, 0x80, count + 1);	530	encode_uint (enc, MAJOR_ARRAY, count + 1);
405	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	531	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406		532
407	while (count)	533	{
		534	int i;
		535
		536	for (i = 0; i < count; ++i)
408	encode_sv (enc, SP[1 - count--]);	537	encode_sv (enc, SP[i + 1 - count]);
		538
		539	SP -= count;
		540	}
409		541
410	PUTBACK;	542	PUTBACK;
411		543
412	FREETMPS; LEAVE;	544	FREETMPS; LEAVE;
413	}	545	}
…		…
431	{	563	{
432	double nv = SvNVX (sv);	564	double nv = SvNVX (sv);
433		565
434	need (enc, 9);	566	need (enc, 9);
435		567
436	if (ecb_expect_false (nv == (U32)nv))	568	if (ecb_expect_false (nv == (NV)(U32)nv))
437	encode_uint (enc, 0x00, (U32)nv);	569	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438	//TODO: maybe I32?	570	//TODO: maybe I32?
439	else if (ecb_expect_false (nv == (float)nv))	571	else if (ecb_expect_false (nv == (float)nv))
440	{	572	{
		573	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		574
441	uint32_t fp = ecb_float_to_binary32 (nv);	575	uint32_t fp = ecb_float_to_binary32 (nv);
442
443	*enc->cur++ = 0xe0 | 26;
444		576
445	if (!ecb_big_endian ())	577	if (!ecb_big_endian ())
446	fp = ecb_bswap32 (fp);	578	fp = ecb_bswap32 (fp);
447		579
448	memcpy (enc->cur, &fp, 4);	580	memcpy (enc->cur, &fp, 4);
449	enc->cur += 4;	581	enc->cur += 4;
450	}	582	}
451	else	583	else
452	{	584	{
		585	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		586
453	uint64_t fp = ecb_double_to_binary64 (nv);	587	uint64_t fp = ecb_double_to_binary64 (nv);
454
455	*enc->cur++ = 0xe0 | 27;
456		588
457	if (!ecb_big_endian ())	589	if (!ecb_big_endian ())
458	fp = ecb_bswap64 (fp);	590	fp = ecb_bswap64 (fp);
459		591
460	memcpy (enc->cur, &fp, 8);	592	memcpy (enc->cur, &fp, 8);
…		…
469		601
470	if (SvPOKp (sv))	602	if (SvPOKp (sv))
471	{	603	{
472	STRLEN len;	604	STRLEN len;
473	char *str = SvPV (sv, len);	605	char *str = SvPV (sv, len);
474	encode_str (enc, SvUTF8 (sv), str, len);	606	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475	}	607	}
476	else if (SvNOKp (sv))	608	else if (SvNOKp (sv))
477	encode_nv (enc, sv);	609	encode_nv (enc, sv);
478	else if (SvIOKp (sv))	610	else if (SvIOKp (sv))
479	{	611	{
480	if (SvIsUV (sv))	612	if (SvIsUV (sv))
481	encode_uint (enc, 0x00, SvUVX (sv));	613	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
482	else if (SvIVX (sv) >= 0)	614	else if (SvIVX (sv) >= 0)
483	encode_uint (enc, 0x00, SvIVX (sv));	615	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
484	else	616	else
485	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	617	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
486	}	618	}
487	else if (SvROK (sv))	619	else if (SvROK (sv))
488	encode_rv (enc, SvRV (sv));	620	encode_rv (enc, SvRV (sv));
489	else if (!SvOK (sv))	621	else if (!SvOK (sv))
490	encode_ch (enc, 0xe0 | 22);	622	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	623	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492	encode_ch (enc, 0xe0 | 23);	624	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493	else	625	else
494	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	626	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
495	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	627	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496	}	628	}
497		629
498	static SV *	630	static SV *
499	encode_cbor (SV *scalar, CBOR *cbor)	631	encode_cbor (SV *scalar, CBOR *cbor)
500	{	632	{
501	enc_t enc = { };	633	enc_t enc = { 0 };
502		634
503	enc.cbor = *cbor;	635	enc.cbor = *cbor;
504	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	636	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505	enc.cur = SvPVX (enc.sv);	637	enc.cur = SvPVX (enc.sv);
506	enc.end = SvEND (enc.sv);	638	enc.end = SvEND (enc.sv);
507		639
508	SvPOK_only (enc.sv);	640	SvPOK_only (enc.sv);
509		641
510	if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))	642	if (cbor->flags & F_PACK_STRINGS)
511	{	643	{
512	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	644	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	645	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	646	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515	}	647	}
…		…
537	CBOR cbor;	669	CBOR cbor;
538	U32 depth; // recursion depth	670	U32 depth; // recursion depth
539	U32 maxdepth; // recursion depth limit	671	U32 maxdepth; // recursion depth limit
540	AV *shareable;	672	AV *shareable;
541	AV *stringref;	673	AV *stringref;
		674	SV *decode_tagged;
		675	SV *err_sv; // optional sv for error, needs to be freed
542	} dec_t;	676	} dec_t;
543		677
544	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	678	// set dec->err to ERRSV
		679	ecb_cold static void
		680	err_errsv (dec_t *dec)
		681	{
		682	if (!dec->err)
		683	{
		684	dec->err_sv = newSVsv (ERRSV);
545		685
546	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	686	// chop off the trailing \n
		687	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		688	*SvEND (dec->err_sv) = 0;
547		689
		690	dec->err = SvPVutf8_nolen (dec->err_sv);
		691	}
		692	}
		693
		694	// the following functions are used to reduce code size and help the compiler to optimise
		695	ecb_cold static void
		696	err_set (dec_t *dec, const char *reason)
		697	{
		698	if (!dec->err)
		699	dec->err = reason;
		700	}
		701
		702	ecb_cold static void
		703	err_unexpected_end (dec_t *dec)
		704	{
		705	err_set (dec, "unexpected end of CBOR data");
		706	}
		707
		708	#define ERR_DO(do) SB do; goto fail; SE
		709	#define ERR(reason) ERR_DO (err_set (dec, reason))
		710	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		711
		712	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		713
548	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	714	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
549	#define DEC_DEC_DEPTH --dec->depth	715	#define DEC_DEC_DEPTH --dec->depth
550		716
551	static UV	717	static UV
552	decode_uint (dec_t *dec)	718	decode_uint (dec_t *dec)
553	{	719	{
554	switch (*dec->cur & 31)	720	U8 m = *dec->cur & MINOR_MASK;
555	{	721	++dec->cur;
556	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559	return *dec->cur++ & 31;
560		722
561	case 24:	723	if (ecb_expect_true (m < LENGTH_EXT1))
		724	return m;
		725	else if (ecb_expect_true (m == LENGTH_EXT1))
		726	{
562	WANT (2);	727	WANT (1);
563	dec->cur += 2;	728	dec->cur += 1;
564	return dec->cur[-1];	729	return dec->cur[-1];
565		730	}
566	case 25:	731	else if (ecb_expect_true (m == LENGTH_EXT2))
		732	{
567	WANT (3);	733	WANT (2);
568	dec->cur += 3;	734	dec->cur += 2;
569	return (((UV)dec->cur[-2]) << 8)	735	return (((UV)dec->cur[-2]) << 8)
570	| ((UV)dec->cur[-1]);	736	| ((UV)dec->cur[-1]);
571		737	}
572	case 26:	738	else if (ecb_expect_true (m == LENGTH_EXT4))
		739	{
573	WANT (5);	740	WANT (4);
574	dec->cur += 5;	741	dec->cur += 4;
575	return (((UV)dec->cur[-4]) << 24)	742	return (((UV)dec->cur[-4]) << 24)
576	| (((UV)dec->cur[-3]) << 16)	743	| (((UV)dec->cur[-3]) << 16)
577	| (((UV)dec->cur[-2]) << 8)	744	| (((UV)dec->cur[-2]) << 8)
578	| ((UV)dec->cur[-1]);	745	| ((UV)dec->cur[-1]);
579		746	}
580	case 27:	747	else if (ecb_expect_true (m == LENGTH_EXT8))
		748	{
581	WANT (9);	749	WANT (8);
582	dec->cur += 9;	750	dec->cur += 8;
		751
		752	return
		753	#if UVSIZE < 8
		754	0
		755	#else
583	return (((UV)dec->cur[-8]) << 56)	756	(((UV)dec->cur[-8]) << 56)
584	| (((UV)dec->cur[-7]) << 48)	757	| (((UV)dec->cur[-7]) << 48)
585	| (((UV)dec->cur[-6]) << 40)	758	| (((UV)dec->cur[-6]) << 40)
586	| (((UV)dec->cur[-5]) << 32)	759	| (((UV)dec->cur[-5]) << 32)
		760	#endif
587	| (((UV)dec->cur[-4]) << 24)	761	| (((UV)dec->cur[-4]) << 24)
588	| (((UV)dec->cur[-3]) << 16)	762	| (((UV)dec->cur[-3]) << 16)
589	| (((UV)dec->cur[-2]) << 8)	763	| (((UV)dec->cur[-2]) << 8)
590	| ((UV)dec->cur[-1]);	764	| ((UV)dec->cur[-1]);
591		765	}
592	default:	766	else
593	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	767	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594	}
595		768
596	fail:	769	fail:
597	return 0;	770	return 0;
598	}	771	}
599		772
…		…
604	{	777	{
605	AV *av = newAV ();	778	AV *av = newAV ();
606		779
607	DEC_INC_DEPTH;	780	DEC_INC_DEPTH;
608		781
609	if ((*dec->cur & 31) == 31)	782	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610	{	783	{
611	++dec->cur;	784	++dec->cur;
612		785
613	for (;;)	786	for (;;)
614	{	787	{
615	WANT (1);	788	WANT (1);
616		789
617	if (*dec->cur == (0xe0 | 31))	790	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
618	{	791	{
619	++dec->cur;	792	++dec->cur;
620	break;	793	break;
621	}	794	}
622		795
623	av_push (av, decode_sv (dec));	796	av_push (av, decode_sv (dec));
624	}	797	}
625	}	798	}
626	else	799	else
627	{	800	{
628	int i, len = decode_uint (dec);	801	UV i, len = decode_uint (dec);
629		802
		803	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630	av_fill (av, len - 1);	804	av_fill (av, len - 1);
631		805
632	for (i = 0; i < len; ++i)	806	for (i = 0; i < len; ++i)
633	AvARRAY (av)[i] = decode_sv (dec);	807	AvARRAY (av)[i] = decode_sv (dec);
634	}	808	}
635		809
636	DEC_DEC_DEPTH;	810	DEC_DEC_DEPTH;
637	return newRV_noinc ((SV *)av);	811	return newRV_noinc ((SV *)av);
638		812
639	fail:	813	fail:
640	SvREFCNT_dec (av);	814	SvREFCNT_dec_NN (av);
641	DEC_DEC_DEPTH;	815	DEC_DEC_DEPTH;
642	return &PL_sv_undef;	816	return &PL_sv_undef;
643	}	817	}
644		818
645	static void	819	static void
646	decode_he (dec_t *dec, HV *hv)	820	decode_he (dec_t *dec, HV *hv)
647	{	821	{
648	// for speed reasons, we specialcase single-string	822	// for speed reasons, we specialcase single-string
649	// byte or utf-8 strings as keys.	823	// byte or utf-8 strings as keys, but only when !stringref
650		824
651	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	825	if (ecb_expect_true (!dec->stringref))
		826	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652	{	827	{
653	I32 len = decode_uint (dec);	828	STRLEN len = decode_uint (dec);
654	char *key = (char *)dec->cur;	829	char *key = (char *)dec->cur;
655		830
		831	WANT (len);
656	dec->cur += len;	832	dec->cur += len;
657		833
658	if (ecb_expect_false (dec->stringref))
659	av_push (dec->stringref, newSVpvn (key, len));
660
661	hv_store (hv, key, len, decode_sv (dec), 0);	834	hv_store (hv, key, len, decode_sv (dec), 0);
		835
		836	return;
662	}	837	}
663	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	838	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664	{	839	{
665	I32 len = decode_uint (dec);	840	STRLEN len = decode_uint (dec);
666	char *key = (char *)dec->cur;	841	char *key = (char *)dec->cur;
667		842
		843	WANT (len);
668	dec->cur += len;	844	dec->cur += len;
669		845
670	if (ecb_expect_false (dec->stringref))	846	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	847	if (!is_utf8_string (key, len))
		848	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672		849
673	hv_store (hv, key, -len, decode_sv (dec), 0);	850	hv_store (hv, key, -len, decode_sv (dec), 0);
		851
		852	return;
674	}	853	}
675	else	854
676	{
677	SV *k = decode_sv (dec);	855	SV *k = decode_sv (dec);
678	SV *v = decode_sv (dec);	856	SV *v = decode_sv (dec);
679		857
		858	// we leak memory if uncaught exceptions are thrown by random magical
		859	// methods, and this is hopefully the only place where it can happen,
		860	// so if there is a chance of an exception, take the very slow path.
		861	// since catching exceptions is "undocumented/internal/forbidden" by
		862	// the new p5p powers, we need to call out to a perl function :/
		863	if (ecb_expect_false (SvAMAGIC (k)))
		864	{
		865	dSP;
		866
		867	ENTER; SAVETMPS;
		868	PUSHMARK (SP);
		869	EXTEND (SP, 3);
		870	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		871	PUSHs (sv_2mortal (k));
		872	PUSHs (sv_2mortal (v));
		873
		874	PUTBACK;
		875	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		876	SPAGAIN;
		877
		878	FREETMPS; LEAVE;
		879
		880	if (SvTRUE (ERRSV))
		881	ERR_ERRSV;
		882
		883	return;
		884	}
		885
680	hv_store_ent (hv, k, v, 0);	886	hv_store_ent (hv, k, v, 0);
681	SvREFCNT_dec (k);	887	SvREFCNT_dec_NN (k);
682	}	888
		889	fail:
		890	;
683	}	891	}
684		892
685	static SV *	893	static SV *
686	decode_hv (dec_t *dec)	894	decode_hv (dec_t *dec)
687	{	895	{
688	HV *hv = newHV ();	896	HV *hv = newHV ();
689		897
690	DEC_INC_DEPTH;	898	DEC_INC_DEPTH;
691		899
692	if ((*dec->cur & 31) == 31)	900	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693	{	901	{
694	++dec->cur;	902	++dec->cur;
695		903
696	for (;;)	904	for (;;)
697	{	905	{
698	WANT (1);	906	WANT (1);
699		907
700	if (*dec->cur == (0xe0 | 31))	908	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
701	{	909	{
702	++dec->cur;	910	++dec->cur;
703	break;	911	break;
704	}	912	}
705		913
706	decode_he (dec, hv);	914	decode_he (dec, hv);
707	}	915	}
708	}	916	}
709	else	917	else
710	{	918	{
711	int pairs = decode_uint (dec);	919	UV pairs = decode_uint (dec);
		920
		921	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
712		922
713	while (pairs--)	923	while (pairs--)
714	decode_he (dec, hv);	924	decode_he (dec, hv);
715	}	925	}
716		926
717	DEC_DEC_DEPTH;	927	DEC_DEC_DEPTH;
718	return newRV_noinc ((SV *)hv);	928	return newRV_noinc ((SV *)hv);
719		929
720	fail:	930	fail:
721	SvREFCNT_dec (hv);	931	SvREFCNT_dec_NN (hv);
722	DEC_DEC_DEPTH;	932	DEC_DEC_DEPTH;
723	return &PL_sv_undef;	933	return &PL_sv_undef;
724	}	934	}
725		935
726	static SV *	936	static SV *
727	decode_str (dec_t *dec, int utf8)	937	decode_str (dec_t *dec, int utf8)
728	{	938	{
729	SV *sv = 0;	939	SV *sv = 0;
730		940
731	if ((*dec->cur & 31) == 31)	941	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
732	{	942	{
		943	// indefinite length strings
733	++dec->cur;	944	++dec->cur;
734		945
		946	U8 major = *dec->cur & MAJOR_MISC;
		947
735	sv = newSVpvn ("", 0);	948	sv = newSVpvn ("", 0);
736		949
737	// not very fast, and certainly not robust against illegal input
738	for (;;)	950	for (;;)
739	{	951	{
740	WANT (1);	952	WANT (1);
741		953
742	if (*dec->cur == (0xe0 | 31))	954	if ((*dec->cur - major) > LENGTH_EXT8)
		955	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743	{	956	{
744	++dec->cur;	957	++dec->cur;
745	break;	958	break;
746	}	959	}
		960	else
		961	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747		962
748	sv_catsv (sv, decode_sv (dec));	963	STRLEN len = decode_uint (dec);
		964
		965	WANT (len);
		966	sv_catpvn (sv, dec->cur, len);
		967	dec->cur += len;
749	}	968	}
750	}	969	}
751	else	970	else
752	{	971	{
753	STRLEN len = decode_uint (dec);	972	STRLEN len = decode_uint (dec);
754		973
755	WANT (len);	974	WANT (len);
756	sv = newSVpvn (dec->cur, len);	975	sv = newSVpvn (dec->cur, len);
757	dec->cur += len;	976	dec->cur += len;
		977
		978	if (ecb_expect_false (dec->stringref)
		979	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		980	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
758	}	981	}
759		982
760	if (utf8)	983	if (utf8)
		984	{
		985	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		986	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		987	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		988
761	SvUTF8_on (sv);	989	SvUTF8_on (sv);
762		990	}
763	if (ecb_expect_false (dec->stringref))
764	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765		991
766	return sv;	992	return sv;
767		993
768	fail:	994	fail:
769	SvREFCNT_dec (sv);	995	SvREFCNT_dec (sv);
…		…
788	sv = newRV_noinc (decode_sv (dec));	1014	sv = newRV_noinc (decode_sv (dec));
789	break;	1015	break;
790		1016
791	case CBOR_TAG_STRINGREF_NAMESPACE:	1017	case CBOR_TAG_STRINGREF_NAMESPACE:
792	{	1018	{
		1019	// do not use SAVETMPS/FREETMPS, as these will
		1020	// erase mortalised caches, e.g. "shareable"
793	ENTER; SAVETMPS;	1021	ENTER;
794		1022
795	SAVESPTR (dec->stringref);	1023	SAVESPTR (dec->stringref);
796	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1024	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797		1025
798	sv = decode_sv (dec);	1026	sv = decode_sv (dec);
799		1027
800	FREETMPS; LEAVE;	1028	LEAVE;
801	}	1029	}
802	break;	1030	break;
803		1031
804	case CBOR_TAG_STRINGREF:	1032	case CBOR_TAG_STRINGREF:
805	{	1033	{
806	if ((*dec->cur >> 5) != 0)	1034	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1035	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808		1036
809	UV idx = decode_uint (dec);	1037	UV idx = decode_uint (dec);
810		1038
811	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1039	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
812	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1040	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813		1041
814	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1042	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815	}	1043	}
816	break;	1044	break;
…		…
818	case CBOR_TAG_VALUE_SHAREABLE:	1046	case CBOR_TAG_VALUE_SHAREABLE:
819	{	1047	{
820	if (ecb_expect_false (!dec->shareable))	1048	if (ecb_expect_false (!dec->shareable))
821	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1049	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822		1050
		1051	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1052	{
823	sv = newSV (0);	1053	sv = newSV (0);
824	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1054	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825		1055
826	SV *osv = decode_sv (dec);	1056	SV *osv = decode_sv (dec);
827	sv_setsv (sv, osv);	1057	sv_setsv (sv, osv);
828	SvREFCNT_dec_NN (osv);	1058	SvREFCNT_dec_NN (osv);
		1059	}
		1060	else
		1061	{
		1062	av_push (dec->shareable, &PL_sv_undef);
		1063	int idx = AvFILLp (dec->shareable);
		1064	sv = decode_sv (dec);
		1065	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1066	}
829	}	1067	}
830	break;	1068	break;
831		1069
832	case CBOR_TAG_VALUE_SHAREDREF:	1070	case CBOR_TAG_VALUE_SHAREDREF:
833	{	1071	{
834	if ((*dec->cur >> 5) != 0)	1072	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1073	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836		1074
837	UV idx = decode_uint (dec);	1075	UV idx = decode_uint (dec);
838		1076
839	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1077	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
840	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1078	ERR ("corrupted CBOR data (sharedref index out of bounds)");
841		1079
842	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1080	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1081
		1082	if (sv == &PL_sv_undef)
		1083	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
843	}	1084	}
844	break;	1085	break;
845		1086
846	case CBOR_TAG_PERL_OBJECT:	1087	case CBOR_TAG_PERL_OBJECT:
847	{	1088	{
		1089	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1090	goto filter;
		1091
848	sv = decode_sv (dec);	1092	sv = decode_sv (dec);
849		1093
850	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1094	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851	ERR ("corrupted CBOR data (non-array perl object)");	1095	ERR ("corrupted CBOR data (non-array perl object)");
852		1096
…		…
862	if (!method)	1106	if (!method)
863	ERR ("cannot decode perl-object (package does not have a THAW method)");	1107	ERR ("cannot decode perl-object (package does not have a THAW method)");
864		1108
865	dSP;	1109	dSP;
866		1110
867	ENTER; SAVETMPS; PUSHMARK (SP);	1111	ENTER; SAVETMPS;
		1112	PUSHMARK (SP);
868	EXTEND (SP, len + 1);	1113	EXTEND (SP, len + 1);
869	// we re-bless the reference to get overload and other niceties right	1114	// we re-bless the reference to get overload and other niceties right
870	PUSHs (*av_fetch (av, 0, 1));	1115	PUSHs (*av_fetch (av, 0, 1));
871	PUSHs (sv_cbor);	1116	PUSHs (sv_cbor);
872		1117
…		…
880	SPAGAIN;	1125	SPAGAIN;
881		1126
882	if (SvTRUE (ERRSV))	1127	if (SvTRUE (ERRSV))
883	{	1128	{
884	FREETMPS; LEAVE;	1129	FREETMPS; LEAVE;
885	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1130	ERR_ERRSV;
886	}	1131	}
887		1132
888	SvREFCNT_dec (sv);	1133	SvREFCNT_dec_NN (sv);
889	sv = SvREFCNT_inc (POPs);	1134	sv = SvREFCNT_inc (POPs);
890		1135
891	PUTBACK;	1136	PUTBACK;
892		1137
893	FREETMPS; LEAVE;	1138	FREETMPS; LEAVE;
894	}	1139	}
895	break;	1140	break;
896		1141
897	default:	1142	default:
		1143	filter:
898	{	1144	{
		1145	SV *tag_sv = newSVuv (tag);
		1146
899	sv = decode_sv (dec);	1147	sv = decode_sv (dec);
900		1148
		1149	dSP;
		1150	ENTER; SAVETMPS;
		1151	PUSHMARK (SP);
		1152	EXTEND (SP, 2);
		1153	PUSHs (tag_sv);
		1154	PUSHs (sv);
		1155
		1156	PUTBACK;
		1157	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1158	SPAGAIN;
		1159
		1160	if (SvTRUE (ERRSV))
		1161	{
		1162	SvREFCNT_dec_NN (tag_sv);
		1163	FREETMPS; LEAVE;
		1164	ERR_ERRSV;
		1165	}
		1166
		1167	if (count)
		1168	{
		1169	SvREFCNT_dec_NN (tag_sv);
		1170	SvREFCNT_dec_NN (sv);
		1171	sv = SvREFCNT_inc_NN (TOPs);
		1172	SP -= count;
		1173	}
		1174	else
		1175	{
901	AV *av = newAV ();	1176	AV *av = newAV ();
902	av_push (av, newSVuv (tag));	1177	av_push (av, tag_sv);
903	av_push (av, sv);	1178	av_push (av, sv);
904		1179
905	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1180	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906	? cbor_tagged_stash	1181	? cbor_tagged_stash
907	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1182	: gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1183	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1184	}
		1185
		1186	PUTBACK;
		1187
		1188	FREETMPS; LEAVE;
910	}	1189	}
911	break;	1190	break;
912	}	1191	}
913		1192
914	return sv;	1193	return sv;
…		…
921	static SV *	1200	static SV *
922	decode_sv (dec_t *dec)	1201	decode_sv (dec_t *dec)
923	{	1202	{
924	WANT (1);	1203	WANT (1);
925		1204
926	switch (*dec->cur >> 5)	1205	switch (*dec->cur >> MAJOR_SHIFT)
927	{	1206	{
928	case 0: // unsigned int	1207	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929	return newSVuv (decode_uint (dec));	1208	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930	case 1: // negative int	1209	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931	return newSViv (-1 - (IV)decode_uint (dec));	1210	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932	case 2: // octet string	1211	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933	return decode_str (dec, 0);	1212	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934	case 3: // utf-8 string	1213	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935	return decode_str (dec, 1);	1214
936	case 4: // array	1215	case MAJOR_MISC >> MAJOR_SHIFT:
937	return decode_av (dec);
938	case 5: // map
939	return decode_hv (dec);
940	case 6: // tag
941	return decode_tagged (dec);
942	case 7: // misc
943	switch (*dec->cur++ & 31)	1216	switch (*dec->cur++ & MINOR_MASK)
944	{	1217	{
945	case 20:	1218	case SIMPLE_FALSE:
946	#if CBOR_SLOW	1219	#if CBOR_SLOW
947	types_false = get_bool ("Types::Serialiser::false");	1220	types_false = get_bool ("Types::Serialiser::false");
948	#endif	1221	#endif
949	return newSVsv (types_false);	1222	return newSVsv (types_false);
950	case 21:	1223	case SIMPLE_TRUE:
951	#if CBOR_SLOW	1224	#if CBOR_SLOW
952	types_true = get_bool ("Types::Serialiser::true");	1225	types_true = get_bool ("Types::Serialiser::true");
953	#endif	1226	#endif
954	return newSVsv (types_true);	1227	return newSVsv (types_true);
955	case 22:	1228	case SIMPLE_NULL:
956	return newSVsv (&PL_sv_undef);	1229	return newSVsv (&PL_sv_undef);
957	case 23:	1230	case SIMPLE_UNDEF:
958	#if CBOR_SLOW	1231	#if CBOR_SLOW
959	types_error = get_bool ("Types::Serialiser::error");	1232	types_error = get_bool ("Types::Serialiser::error");
960	#endif	1233	#endif
961	return newSVsv (types_error);	1234	return newSVsv (types_error);
962		1235
963	case 25:	1236	case MISC_FLOAT16:
964	{	1237	{
965	WANT (2);	1238	WANT (2);
966		1239
967	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1240	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968	dec->cur += 2;	1241	dec->cur += 2;
969		1242
970	return newSVnv (ecb_binary16_to_float (fp));	1243	return newSVnv (ecb_binary16_to_float (fp));
971	}	1244	}
972		1245
973	case 26:	1246	case MISC_FLOAT32:
974	{	1247	{
975	uint32_t fp;	1248	uint32_t fp;
976	WANT (4);	1249	WANT (4);
977	memcpy (&fp, dec->cur, 4);	1250	memcpy (&fp, dec->cur, 4);
978	dec->cur += 4;	1251	dec->cur += 4;
…		…
981	fp = ecb_bswap32 (fp);	1254	fp = ecb_bswap32 (fp);
982		1255
983	return newSVnv (ecb_binary32_to_float (fp));	1256	return newSVnv (ecb_binary32_to_float (fp));
984	}	1257	}
985		1258
986	case 27:	1259	case MISC_FLOAT64:
987	{	1260	{
988	uint64_t fp;	1261	uint64_t fp;
989	WANT (8);	1262	WANT (8);
990	memcpy (&fp, dec->cur, 8);	1263	memcpy (&fp, dec->cur, 8);
991	dec->cur += 8;	1264	dec->cur += 8;
…		…
994	fp = ecb_bswap64 (fp);	1267	fp = ecb_bswap64 (fp);
995		1268
996	return newSVnv (ecb_binary64_to_double (fp));	1269	return newSVnv (ecb_binary64_to_double (fp));
997	}	1270	}
998		1271
999	// 0..19 unassigned	1272	// 0..19 unassigned simple
1000	// 24 reserved + unassigned (reserved values are not encodable)	1273	// 24 reserved + unassigned simple (reserved values are not encodable)
		1274	// 28-30 unassigned misc
		1275	// 31 break code
1001	default:	1276	default:
1002	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1277	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003	}	1278	}
1004		1279
1005	break;	1280	break;
1006	}	1281	}
1007		1282
…		…
1010	}	1285	}
1011		1286
1012	static SV *	1287	static SV *
1013	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1288	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014	{	1289	{
1015	dec_t dec = { };	1290	dec_t dec = { 0 };
1016	SV *sv;	1291	SV *sv;
1017	STRLEN len;	1292	STRLEN len;
1018	char *data = SvPVbyte (string, len);	1293	char *data = SvPVbyte (string, len);
1019		1294
1020	if (len > cbor->max_size && cbor->max_size)	1295	if (len > cbor->max_size && cbor->max_size)
…		…
1034	if (dec.cur != dec.end && !dec.err)	1309	if (dec.cur != dec.end && !dec.err)
1035	dec.err = "garbage after CBOR object";	1310	dec.err = "garbage after CBOR object";
1036		1311
1037	if (dec.err)	1312	if (dec.err)
1038	{	1313	{
		1314	if (dec.shareable)
		1315	{
		1316	// need to break cyclic links, which would all be in shareable
		1317	int i;
		1318	SV **svp;
		1319
		1320	for (i = av_len (dec.shareable) + 1; i--; )
		1321	if ((svp = av_fetch (dec.shareable, i, 0)))
		1322	sv_setsv (*svp, &PL_sv_undef);
		1323	}
		1324
1039	SvREFCNT_dec (sv);	1325	SvREFCNT_dec_NN (sv);
		1326
		1327	if (dec.err_sv)
		1328	sv_2mortal (dec.err_sv);
		1329
1040	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1330	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1041	}	1331	}
1042		1332
1043	sv = sv_2mortal (sv);	1333	sv = sv_2mortal (sv);
1044		1334
1045	return sv;	1335	return sv;
1046	}	1336	}
1047		1337
		1338	/////////////////////////////////////////////////////////////////////////////
		1339	// incremental parser
		1340
		1341	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1342
		1343	// returns 0 for notyet, 1 for success or error
		1344	static int
		1345	incr_parse (CBOR *self, SV *cborstr)
		1346	{
		1347	STRLEN cur;
		1348	SvPV (cborstr, cur);
		1349
		1350	while (ecb_expect_true (self->incr_need <= cur))
		1351	{
		1352	// table of integer count bytes
		1353	static I8 incr_len[MINOR_MASK + 1] = {
		1354	0, 0, 0, 0, 0, 0, 0, 0,
		1355	0, 0, 0, 0, 0, 0, 0, 0,
		1356	0, 0, 0, 0, 0, 0, 0, 0,
		1357	1, 2, 4, 8,-1,-1,-1,-2
		1358	};
		1359
		1360	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1361	U8 m = *p & MINOR_MASK;
		1362	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1363	I8 ilen = incr_len[m];
		1364
		1365	self->incr_need = self->incr_pos + 1;
		1366
		1367	if (ecb_expect_false (ilen < 0))
		1368	{
		1369	if (m != MINOR_INDEF)
		1370	return 1; // error
		1371
		1372	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1373	{
		1374	if (count >= 0)
		1375	return 1; // error
		1376
		1377	count = 1;
		1378	}
		1379	else
		1380	{
		1381	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1382	count = -1;
		1383	}
		1384	}
		1385	else
		1386	{
		1387	self->incr_need += ilen;
		1388	if (ecb_expect_false (self->incr_need > cur))
		1389	return 0;
		1390
		1391	int major = *p >> MAJOR_SHIFT;
		1392
		1393	switch (major)
		1394	{
		1395	case MAJOR_TAG >> MAJOR_SHIFT:
		1396	++count; // tags merely prefix another value
		1397	break;
		1398
		1399	case MAJOR_BYTES >> MAJOR_SHIFT:
		1400	case MAJOR_TEXT >> MAJOR_SHIFT:
		1401	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1402	case MAJOR_MAP >> MAJOR_SHIFT:
		1403	{
		1404	UV len;
		1405
		1406	if (ecb_expect_false (ilen))
		1407	{
		1408	len = 0;
		1409
		1410	do {
		1411	len = (len << 8) | *++p;
		1412	} while (--ilen);
		1413	}
		1414	else
		1415	len = m;
		1416
		1417	switch (major)
		1418	{
		1419	case MAJOR_BYTES >> MAJOR_SHIFT:
		1420	case MAJOR_TEXT >> MAJOR_SHIFT:
		1421	self->incr_need += len;
		1422	if (ecb_expect_false (self->incr_need > cur))
		1423	return 0;
		1424
		1425	break;
		1426
		1427	case MAJOR_MAP >> MAJOR_SHIFT:
		1428	len <<= 1;
		1429	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1430	if (len)
		1431	{
		1432	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1433	count = len + 1;
		1434	}
		1435	break;
		1436	}
		1437	}
		1438	}
		1439	}
		1440
		1441	self->incr_pos = self->incr_need;
		1442
		1443	if (count > 0)
		1444	{
		1445	while (!--count)
		1446	{
		1447	if (!AvFILLp (self->incr_count))
		1448	return 1; // done
		1449
		1450	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1451	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1452	}
		1453
		1454	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1455	}
		1456	}
		1457
		1458	return 0;
		1459	}
		1460
		1461
1048	/////////////////////////////////////////////////////////////////////////////	1462	/////////////////////////////////////////////////////////////////////////////
1049	// XS interface functions	1463	// XS interface functions
1050		1464
1051	MODULE = CBOR::XS PACKAGE = CBOR::XS	1465	MODULE = CBOR::XS PACKAGE = CBOR::XS
1052		1466
…		…
1060		1474
1061	types_true = get_bool ("Types::Serialiser::true" );	1475	types_true = get_bool ("Types::Serialiser::true" );
1062	types_false = get_bool ("Types::Serialiser::false");	1476	types_false = get_bool ("Types::Serialiser::false");
1063	types_error = get_bool ("Types::Serialiser::error");	1477	types_error = get_bool ("Types::Serialiser::error");
1064		1478
		1479	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1480
1065	sv_cbor = newSVpv ("CBOR", 0);	1481	sv_cbor = newSVpv ("CBOR", 0);
1066	SvREADONLY_on (sv_cbor);	1482	SvREADONLY_on (sv_cbor);
		1483
		1484	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067	}	1485	}
1068		1486
1069	PROTOTYPES: DISABLE	1487	PROTOTYPES: DISABLE
1070		1488
1071	void CLONE (...)	1489	void CLONE (...)
…		…
1090	void shrink (CBOR *self, int enable = 1)	1508	void shrink (CBOR *self, int enable = 1)
1091	ALIAS:	1509	ALIAS:
1092	shrink = F_SHRINK	1510	shrink = F_SHRINK
1093	allow_unknown = F_ALLOW_UNKNOWN	1511	allow_unknown = F_ALLOW_UNKNOWN
1094	allow_sharing = F_ALLOW_SHARING	1512	allow_sharing = F_ALLOW_SHARING
1095	dedup_keys = F_DEDUP_KEYS	1513	allow_cycles = F_ALLOW_CYCLES
		1514	forbid_objects = F_FORBID_OBJECTS
1096	dedup_strings = F_DEDUP_STRINGS	1515	pack_strings = F_PACK_STRINGS
		1516	text_keys = F_TEXT_KEYS
		1517	text_strings = F_TEXT_STRINGS
		1518	validate_utf8 = F_VALIDATE_UTF8
1097	PPCODE:	1519	PPCODE:
1098	{	1520	{
1099	if (enable)	1521	if (enable)
1100	self->flags |= ix;	1522	self->flags |= ix;
1101	else	1523	else
…		…
1107	void get_shrink (CBOR *self)	1529	void get_shrink (CBOR *self)
1108	ALIAS:	1530	ALIAS:
1109	get_shrink = F_SHRINK	1531	get_shrink = F_SHRINK
1110	get_allow_unknown = F_ALLOW_UNKNOWN	1532	get_allow_unknown = F_ALLOW_UNKNOWN
1111	get_allow_sharing = F_ALLOW_SHARING	1533	get_allow_sharing = F_ALLOW_SHARING
1112	get_dedup_keys = F_DEDUP_KEYS	1534	get_allow_cycles = F_ALLOW_CYCLES
		1535	get_forbid_objects = F_FORBID_OBJECTS
1113	get_dedup_strings = F_DEDUP_STRINGS	1536	get_pack_strings = F_PACK_STRINGS
		1537	get_text_keys = F_TEXT_KEYS
		1538	get_text_strings = F_TEXT_STRINGS
		1539	get_validate_utf8 = F_VALIDATE_UTF8
1114	PPCODE:	1540	PPCODE:
1115	XPUSHs (boolSV (self->flags & ix));	1541	XPUSHs (boolSV (self->flags & ix));
1116		1542
1117	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1543	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118	PPCODE:	1544	PPCODE:
…		…
1131	XPUSHs (ST (0));	1557	XPUSHs (ST (0));
1132		1558
1133	int get_max_size (CBOR *self)	1559	int get_max_size (CBOR *self)
1134	CODE:	1560	CODE:
1135	RETVAL = self->max_size;	1561	RETVAL = self->max_size;
		1562	OUTPUT:
		1563	RETVAL
		1564
		1565	void filter (CBOR *self, SV *filter = 0)
		1566	PPCODE:
		1567	SvREFCNT_dec (self->filter);
		1568	self->filter = filter ? newSVsv (filter) : filter;
		1569	XPUSHs (ST (0));
		1570
		1571	SV *get_filter (CBOR *self)
		1572	CODE:
		1573	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136	OUTPUT:	1574	OUTPUT:
1137	RETVAL	1575	RETVAL
1138		1576
1139	void encode (CBOR *self, SV *scalar)	1577	void encode (CBOR *self, SV *scalar)
1140	PPCODE:	1578	PPCODE:
…		…
1155	EXTEND (SP, 2);	1593	EXTEND (SP, 2);
1156	PUSHs (sv);	1594	PUSHs (sv);
1157	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1595	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158	}	1596	}
1159		1597
		1598	void incr_parse (CBOR *self, SV *cborstr)
		1599	ALIAS:
		1600	incr_parse_multiple = 1
		1601	PPCODE:
		1602	{
		1603	if (SvUTF8 (cborstr))
		1604	sv_utf8_downgrade (cborstr, 0);
		1605
		1606	if (!self->incr_count)
		1607	{
		1608	self->incr_count = newAV ();
		1609	self->incr_pos = 0;
		1610	self->incr_need = 1;
		1611
		1612	av_push (self->incr_count, newSViv (1));
		1613	}
		1614
		1615	do
		1616	{
		1617	if (!incr_parse (self, cborstr))
		1618	{
		1619	if (self->incr_need > self->max_size && self->max_size)
		1620	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1621	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1622
		1623	break;
		1624	}
		1625
		1626	SV *sv;
		1627	char *offset;
		1628
		1629	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1630	XPUSHs (sv);
		1631
		1632	sv_chop (cborstr, offset);
		1633
		1634	av_clear (self->incr_count);
		1635	av_push (self->incr_count, newSViv (1));
		1636
		1637	self->incr_pos = 0;
		1638	self->incr_need = self->incr_pos + 1;
		1639	}
		1640	while (ix);
		1641	}
		1642
		1643	void incr_reset (CBOR *self)
		1644	CODE:
		1645	{
		1646	SvREFCNT_dec (self->incr_count);
		1647	self->incr_count = 0;
		1648	}
		1649
		1650	void DESTROY (CBOR *self)
		1651	PPCODE:
		1652	cbor_free (self);
		1653
1160	PROTOTYPES: ENABLE	1654	PROTOTYPES: ENABLE
1161		1655
1162	void encode_cbor (SV *scalar)	1656	void encode_cbor (SV *scalar)
		1657	ALIAS:
		1658	encode_cbor = 0
		1659	encode_cbor_sharing = F_ALLOW_SHARING
1163	PPCODE:	1660	PPCODE:
1164	{	1661	{
1165	CBOR cbor;	1662	CBOR cbor;
1166	cbor_init (&cbor);	1663	cbor_init (&cbor);
		1664	cbor.flags |= ix;
1167	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1665	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168	XPUSHs (scalar);	1666	XPUSHs (scalar);
1169	}	1667	}
1170		1668
1171	void decode_cbor (SV *cborstr)	1669	void decode_cbor (SV *cborstr)
…		…
1175	cbor_init (&cbor);	1673	cbor_init (&cbor);
1176	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1674	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1177	XPUSHs (cborstr);	1675	XPUSHs (cborstr);
1178	}	1676	}
1179		1677
		1678	#ifdef __AFL_COMPILER
		1679
		1680	void
		1681	afl_init ()
		1682	CODE:
		1683	__AFL_INIT ();
		1684
		1685	int
		1686	afl_loop (unsigned int count = 10000)
		1687	CODE:
		1688	RETVAL = __AFL_LOOP (count);
		1689	OUTPUT:
		1690	RETVAL
		1691
		1692	#endif
		1693

Diff Legend

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