ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.67 by root, Sun Nov 29 21:35:06 2020 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
		102	};
		103
		104	// known forced types, also hardcoded in CBOR.pm
		105	enum
		106	{
		107	AS_CBOR = 0,
		108	AS_BYTES = 1,
		109	AS_TEXT = 2,
		110	AS_INT = 3,
		111	AS_FLOAT16 = 4,
		112	AS_FLOAT32 = 5,
		113	AS_FLOAT64 = 6,
		114	// possibly future enhancements: (generic) float, (generic) string
57	};	115	};
58		116
59	#define F_SHRINK 0x00000001UL	117	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	118	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	119	#define F_ALLOW_SHARING 0x00000004UL
62	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	120	#define F_ALLOW_CYCLES 0x00000008UL
		121	#define F_FORBID_OBJECTS 0x00000010UL
		122	#define F_PACK_STRINGS 0x00000020UL
		123	#define F_TEXT_KEYS 0x00000040UL
		124	#define F_TEXT_STRINGS 0x00000080UL
		125	#define F_VALIDATE_UTF8 0x00000100UL
63		126
64	#define INIT_SIZE 32 // initial scalar size to be allocated	127	#define INIT_SIZE 32 // initial scalar size to be allocated
65		128
66	#define SB do {	129	#define SB do {
67	#define SE } while (0)	130	#define SE } while (0)
…		…
86	typedef struct {	149	typedef struct {
87	U32 flags;	150	U32 flags;
88	U32 max_depth;	151	U32 max_depth;
89	STRLEN max_size;	152	STRLEN max_size;
90	SV *filter;	153	SV *filter;
		154
		155	// for the incremental parser
		156	STRLEN incr_pos; // the current offset into the text
		157	STRLEN incr_need; // minimum bytes needed to decode
		158	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	159	} CBOR;
92		160
93	ecb_inline void	161	ecb_inline void
94	cbor_init (CBOR *cbor)	162	cbor_init (CBOR *cbor)
95	{	163	{
…		…
99		167
100	ecb_inline void	168	ecb_inline void
101	cbor_free (CBOR *cbor)	169	cbor_free (CBOR *cbor)
102	{	170	{
103	SvREFCNT_dec (cbor->filter);	171	SvREFCNT_dec (cbor->filter);
		172	SvREFCNT_dec (cbor->incr_count);
104	}	173	}
105		174
106	/////////////////////////////////////////////////////////////////////////////	175	/////////////////////////////////////////////////////////////////////////////
107	// utility functions	176	// utility functions
108		177
…		…
134		203
135	// minimum length of a string to be registered for stringref	204	// minimum length of a string to be registered for stringref
136	ecb_inline int	205	ecb_inline int
137	minimum_string_length (UV idx)	206	minimum_string_length (UV idx)
138	{	207	{
139	return idx > 23	208	return idx <= 23 ? 3
140	? idx > 0xffU	209	: idx <= 0xffU ? 4
141	? idx > 0xffffU	210	: idx <= 0xffffU ? 5
142	? idx > 0xffffffffU	211	: idx <= 0xffffffffU ? 7
143	? 7	212	: 11;
144	: 6
145	: 5
146	: 4
147	: 3;
148	}	213	}
149		214
150	/////////////////////////////////////////////////////////////////////////////	215	/////////////////////////////////////////////////////////////////////////////
151	// encoder	216	// encoder
152		217
…		…
165	} enc_t;	230	} enc_t;
166		231
167	ecb_inline void	232	ecb_inline void
168	need (enc_t *enc, STRLEN len)	233	need (enc_t *enc, STRLEN len)
169	{	234	{
170	if (ecb_expect_false (enc->cur + len >= enc->end))	235	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
171	{	236	{
172	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	237	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
173	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	238	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
174	enc->cur = SvPVX (enc->sv) + cur;	239	enc->cur = SvPVX (enc->sv) + cur;
175	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	240	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
181	{	246	{
182	need (enc, 1);	247	need (enc, 1);
183	*enc->cur++ = ch;	248	*enc->cur++ = ch;
184	}	249	}
185		250
		251	// used for tags, intregers, element counts and so on
186	static void	252	static void
187	encode_uint (enc_t *enc, int major, UV len)	253	encode_uint (enc_t *enc, int major, UV len)
188	{	254	{
189	need (enc, 9);	255	need (enc, 9);
190		256
191	if (len < 24)	257	if (ecb_expect_true (len < LENGTH_EXT1))
192	*enc->cur++ = major | len;	258	*enc->cur++ = major | len;
193	else if (len <= 0xff)	259	else if (ecb_expect_true (len <= 0xffU))
194	{	260	{
195	*enc->cur++ = major | 24;	261	*enc->cur++ = major | LENGTH_EXT1;
196	*enc->cur++ = len;	262	*enc->cur++ = len;
197	}	263	}
198	else if (len <= 0xffff)	264	else if (len <= 0xffffU)
199	{	265	{
200	*enc->cur++ = major | 25;	266	*enc->cur++ = major | LENGTH_EXT2;
201	*enc->cur++ = len >> 8;	267	*enc->cur++ = len >> 8;
202	*enc->cur++ = len;	268	*enc->cur++ = len;
203	}	269	}
204	else if (len <= 0xffffffff)	270	else if (len <= 0xffffffffU)
205	{	271	{
206	*enc->cur++ = major | 26;	272	*enc->cur++ = major | LENGTH_EXT4;
207	*enc->cur++ = len >> 24;	273	*enc->cur++ = len >> 24;
208	*enc->cur++ = len >> 16;	274	*enc->cur++ = len >> 16;
209	*enc->cur++ = len >> 8;	275	*enc->cur++ = len >> 8;
210	*enc->cur++ = len;	276	*enc->cur++ = len;
211	}	277	}
212	else	278	else
213	{	279	{
214	*enc->cur++ = major | 27;	280	*enc->cur++ = major | LENGTH_EXT8;
215	*enc->cur++ = len >> 56;	281	*enc->cur++ = len >> 56;
216	*enc->cur++ = len >> 48;	282	*enc->cur++ = len >> 48;
217	*enc->cur++ = len >> 40;	283	*enc->cur++ = len >> 40;
218	*enc->cur++ = len >> 32;	284	*enc->cur++ = len >> 32;
219	*enc->cur++ = len >> 24;	285	*enc->cur++ = len >> 24;
…		…
221	*enc->cur++ = len >> 8;	287	*enc->cur++ = len >> 8;
222	*enc->cur++ = len;	288	*enc->cur++ = len;
223	}	289	}
224	}	290	}
225		291
		292	// encodes a perl value into a CBOR integer
		293	ecb_inline void
		294	encode_int (enc_t *enc, SV *sv)
		295	{
		296	if (SvIsUV (sv))
		297	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		298	else if (SvIVX (sv) >= 0)
		299	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		300	else
		301	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		302	}
		303
226	ecb_inline void	304	ecb_inline void
227	encode_tag (enc_t *enc, UV tag)	305	encode_tag (enc_t *enc, UV tag)
228	{	306	{
229	encode_uint (enc, 0xc0, tag);	307	encode_uint (enc, MAJOR_TAG, tag);
230	}	308	}
231		309
232	static void	310	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		311	ecb_noinline static void
233	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	312	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
234	{	313	{
		314	STRLEN ulen = len;
		315	U8 *p, *pend = (U8 *)str + len;
		316
		317	for (p = (U8 *)str; p < pend; ++p)
		318	ulen += *p >> 7; // count set high bits
		319
		320	encode_uint (enc, MAJOR_TEXT, ulen);
		321
		322	need (enc, ulen);
		323	for (p = (U8 *)str; p < pend; ++p)
		324	if (*p < 0x80)
		325	*enc->cur++ = *p;
		326	else
		327	{
		328	*enc->cur++ = 0xc0 + (*p >> 6);
		329	*enc->cur++ = 0x80 + (*p & 63);
		330	}
		331	}
		332
		333	ecb_inline void
		334	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		335	{
		336	if (ecb_expect_false (upgrade_utf8))
		337	if (!utf8)
		338	{
		339	encode_str_utf8 (enc, utf8, str, len);
		340	return;
		341	}
		342
		343	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		344	need (enc, len);
		345	memcpy (enc->cur, str, len);
		346	enc->cur += len;
		347	}
		348
		349	ecb_inline void
		350	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		351	{
235	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	352	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236	{	353	{
237	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	354	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238		355
239	if (SvOK (*svp))	356	if (SvOK (*svp))
240	{	357	{
241	// already registered, use stringref	358	// already registered, use stringref
242	encode_tag (enc, CBOR_TAG_STRINGREF);	359	encode_tag (enc, CBOR_TAG_STRINGREF);
243	encode_uint (enc, 0x00, SvUV (*svp));	360	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244	return;	361	return;
245	}	362	}
246	else if (len >= minimum_string_length (enc->stringref_idx))	363	else if (len >= minimum_string_length (enc->stringref_idx))
247	{	364	{
248	// register only	365	// register only
249	sv_setuv (*svp, enc->stringref_idx);	366	sv_setuv (*svp, enc->stringref_idx);
250	++enc->stringref_idx;	367	++enc->stringref_idx;
251	}	368	}
252	}	369	}
253		370
254	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	371	encode_str (enc, upgrade_utf8, utf8, str, len);
		372	}
		373
		374	ecb_inline void
		375	encode_float16 (enc_t *enc, NV nv)
		376	{
255	need (enc, len);	377	need (enc, 1+2);
		378
		379	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		380
		381	uint16_t fp = ecb_float_to_binary16 (nv);
		382
		383	if (!ecb_big_endian ())
		384	fp = ecb_bswap16 (fp);
		385
256	memcpy (enc->cur, str, len);	386	memcpy (enc->cur, &fp, 2);
257	enc->cur += len;	387	enc->cur += 2;
		388	}
		389
		390	ecb_inline void
		391	encode_float32 (enc_t *enc, NV nv)
		392	{
		393	need (enc, 1+4);
		394
		395	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		396
		397	uint32_t fp = ecb_float_to_binary32 (nv);
		398
		399	if (!ecb_big_endian ())
		400	fp = ecb_bswap32 (fp);
		401
		402	memcpy (enc->cur, &fp, 4);
		403	enc->cur += 4;
		404	}
		405
		406	ecb_inline void
		407	encode_float64 (enc_t *enc, NV nv)
		408	{
		409	need (enc, 1+8);
		410
		411	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		412
		413	uint64_t fp = ecb_double_to_binary64 (nv);
		414
		415	if (!ecb_big_endian ())
		416	fp = ecb_bswap64 (fp);
		417
		418	memcpy (enc->cur, &fp, 8);
		419	enc->cur += 8;
		420	}
		421
		422	ecb_inline void
		423	encode_forced (enc_t *enc, UV type, SV *sv)
		424	{
		425	switch (type)
		426	{
		427	case AS_CBOR:
		428	{
		429	STRLEN len;
		430	char *str = SvPVbyte (sv, len);
		431
		432	need (enc, len);
		433	memcpy (enc->cur, str, len);
		434	enc->cur += len;
		435	}
		436	break;
		437
		438	case AS_BYTES:
		439	{
		440	STRLEN len;
		441	char *str = SvPVbyte (sv, len);
		442	encode_strref (enc, 0, 0, str, len);
		443	}
		444	break;
		445
		446	case AS_TEXT:
		447	{
		448	STRLEN len;
		449	char *str = SvPVutf8 (sv, len);
		450	encode_strref (enc, 1, 1, str, len);
		451	}
		452	break;
		453
		454	case AS_INT: encode_int (enc, sv); break;
		455	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		456	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		457	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		458
		459	default:
		460	croak ("encountered malformed CBOR::XS::Tagged object");
		461	}
258	}	462	}
259		463
260	static void encode_sv (enc_t *enc, SV *sv);	464	static void encode_sv (enc_t *enc, SV *sv);
261		465
262	static void	466	static void
…		…
267	if (enc->depth >= enc->cbor.max_depth)	471	if (enc->depth >= enc->cbor.max_depth)
268	croak (ERR_NESTING_EXCEEDED);	472	croak (ERR_NESTING_EXCEEDED);
269		473
270	++enc->depth;	474	++enc->depth;
271		475
272	encode_uint (enc, 0x80, len + 1);	476	encode_uint (enc, MAJOR_ARRAY, len + 1);
273		477
		478	if (ecb_expect_false (SvMAGICAL (av)))
274	for (i = 0; i <= len; ++i)	479	for (i = 0; i <= len; ++i)
275	{	480	{
276	SV **svp = av_fetch (av, i, 0);	481	SV **svp = av_fetch (av, i, 0);
277	encode_sv (enc, svp ? *svp : &PL_sv_undef);	482	encode_sv (enc, svp ? *svp : &PL_sv_undef);
278	}	483	}
		484	else
		485	for (i = 0; i <= len; ++i)
		486	{
		487	SV *sv = AvARRAY (av)[i];
		488	encode_sv (enc, sv ? sv : &PL_sv_undef);
		489	}
279		490
280	--enc->depth;	491	--enc->depth;
281	}	492	}
282		493
283	static void	494	static void
…		…
291	++enc->depth;	502	++enc->depth;
292		503
293	int pairs = hv_iterinit (hv);	504	int pairs = hv_iterinit (hv);
294	int mg = SvMAGICAL (hv);	505	int mg = SvMAGICAL (hv);
295		506
296	if (mg)	507	if (ecb_expect_false (mg))
297	encode_ch (enc, 0xa0 | 31);	508	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298	else	509	else
299	encode_uint (enc, 0xa0, pairs);	510	encode_uint (enc, MAJOR_MAP, pairs);
300		511
301	while ((he = hv_iternext (hv)))	512	while ((he = hv_iternext (hv)))
302	{	513	{
303	if (HeKLEN (he) == HEf_SVKEY)	514	if (HeKLEN (he) == HEf_SVKEY)
304	encode_sv (enc, HeSVKEY (he));	515	encode_sv (enc, HeSVKEY (he));
305	else	516	else
306	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	517	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307		518
308	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	519	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309	}	520	}
310		521
311	if (mg)	522	if (ecb_expect_false (mg))
312	encode_ch (enc, 0xe0 | 31);	523	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313		524
314	--enc->depth;	525	--enc->depth;
315	}	526	}
316		527
317	// encode objects, arrays and special \0=false and \1=true values.	528	// encode objects, arrays and special \0=false and \1=true values.
318	static void	529	static void
319	encode_rv (enc_t *enc, SV *sv)	530	encode_rv (enc_t *enc, SV *sv)
320	{	531	{
321	SvGETMAGIC (sv);	532	SvGETMAGIC (sv);
322
323	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324	&& ecb_expect_false (SvREFCNT (sv) > 1))
325	{
326	if (!enc->shareable)
327	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
328
329	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
330
331	if (SvOK (*svp))
332	{
333	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334	encode_uint (enc, 0x00, SvUV (*svp));
335	return;
336	}
337	else
338	{
339	sv_setuv (*svp, enc->shareable_idx);
340	++enc->shareable_idx;
341	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342	}
343	}
344		533
345	svtype svt = SvTYPE (sv);	534	svtype svt = SvTYPE (sv);
346		535
347	if (ecb_expect_false (SvOBJECT (sv)))	536	if (ecb_expect_false (SvOBJECT (sv)))
348	{	537	{
…		…
355	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	544	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356	? cbor_tagged_stash	545	? cbor_tagged_stash
357	: gv_stashpv ("CBOR::XS::Tagged" , 1);	546	: gv_stashpv ("CBOR::XS::Tagged" , 1);
358		547
359	HV *stash = SvSTASH (sv);	548	HV *stash = SvSTASH (sv);
360	GV *method;
361		549
362	if (stash == boolean_stash)	550	if (stash == boolean_stash)
363	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	551	{
		552	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		553	return;
		554	}
364	else if (stash == error_stash)	555	else if (stash == error_stash)
365	encode_ch (enc, 0xe0 | 23);	556	{
		557	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		558	return;
		559	}
366	else if (stash == tagged_stash)	560	else if (stash == tagged_stash)
367	{	561	{
368	if (svt != SVt_PVAV)	562	if (svt != SVt_PVAV)
369	croak ("encountered CBOR::XS::Tagged object that isn't an array");	563	croak ("encountered CBOR::XS::Tagged object that isn't an array");
370		564
		565	switch (av_len ((AV *)sv))
		566	{
		567	case 2-1:
		568	// actually a tagged value
371	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	569	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	570	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		571	break;
		572
		573	case 3-1:
		574	// a forced type [value, type, undef]
		575	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		576	break;
		577
		578	default:
		579	croak ("encountered malformed CBOR::XS::Tagged object");
		580	}
		581
		582	return;
		583	}
		584	}
		585
		586	if (ecb_expect_false (SvREFCNT (sv) > 1)
		587	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		588	{
		589	if (ecb_expect_false (!enc->shareable))
		590	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		591
		592	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		593
		594	if (SvOK (*svp))
373	}	595	{
		596	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		597	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		598	return;
		599	}
		600	else
		601	{
		602	sv_setuv (*svp, enc->shareable_idx);
		603	++enc->shareable_idx;
		604	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		605	}
		606	}
		607
		608	if (ecb_expect_false (SvOBJECT (sv)))
		609	{
		610	HV *stash = SvSTASH (sv);
		611	GV *method;
		612
		613	if (enc->cbor.flags & F_FORBID_OBJECTS)
		614	croak ("encountered object '%s', but forbid_objects is enabled",
		615	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	616	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375	{	617	{
376	dSP;	618	dSP;
377		619
378	ENTER; SAVETMPS; PUSHMARK (SP);	620	ENTER; SAVETMPS;
		621	PUSHMARK (SP);
379	// we re-bless the reference to get overload and other niceties right	622	// we re-bless the reference to get overload and other niceties right
380	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	623	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381		624
382	PUTBACK;	625	PUTBACK;
383	// G_SCALAR ensures that return value is 1	626	// G_SCALAR ensures that return value is 1
…		…
396	}	639	}
397	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	640	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398	{	641	{
399	dSP;	642	dSP;
400		643
401	ENTER; SAVETMPS; PUSHMARK (SP);	644	ENTER; SAVETMPS;
		645	PUSHMARK (SP);
402	EXTEND (SP, 2);	646	EXTEND (SP, 2);
403	// we re-bless the reference to get overload and other niceties right	647	// we re-bless the reference to get overload and other niceties right
404	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	648	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405	PUSHs (sv_cbor);	649	PUSHs (sv_cbor);
406		650
…		…
411	// catch this surprisingly common error	655	// catch this surprisingly common error
412	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	656	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	657	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414		658
415	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	659	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416	encode_uint (enc, 0x80, count + 1);	660	encode_uint (enc, MAJOR_ARRAY, count + 1);
417	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	661	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418		662
419	while (count)	663	{
		664	int i;
		665
		666	for (i = 0; i < count; ++i)
420	encode_sv (enc, SP[1 - count--]);	667	encode_sv (enc, SP[i + 1 - count]);
		668
		669	SP -= count;
		670	}
421		671
422	PUTBACK;	672	PUTBACK;
423		673
424	FREETMPS; LEAVE;	674	FREETMPS; LEAVE;
425	}	675	}
…		…
443	{	693	{
444	double nv = SvNVX (sv);	694	double nv = SvNVX (sv);
445		695
446	need (enc, 9);	696	need (enc, 9);
447		697
448	if (ecb_expect_false (nv == (U32)nv))	698	if (ecb_expect_false (nv == (NV)(U32)nv))
449	encode_uint (enc, 0x00, (U32)nv);	699	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450	//TODO: maybe I32?	700	//TODO: maybe I32?
451	else if (ecb_expect_false (nv == (float)nv))	701	else if (ecb_expect_false (nv == (float)nv))
452	{	702	encode_float32 (enc, nv);
453	uint32_t fp = ecb_float_to_binary32 (nv);
454
455	*enc->cur++ = 0xe0 | 26;
456
457	if (!ecb_big_endian ())
458	fp = ecb_bswap32 (fp);
459
460	memcpy (enc->cur, &fp, 4);
461	enc->cur += 4;
462	}
463	else	703	else
464	{	704	encode_float64 (enc, nv);
465	uint64_t fp = ecb_double_to_binary64 (nv);
466
467	*enc->cur++ = 0xe0 | 27;
468
469	if (!ecb_big_endian ())
470	fp = ecb_bswap64 (fp);
471
472	memcpy (enc->cur, &fp, 8);
473	enc->cur += 8;
474	}
475	}	705	}
476		706
477	static void	707	static void
478	encode_sv (enc_t *enc, SV *sv)	708	encode_sv (enc_t *enc, SV *sv)
479	{	709	{
…		…
481		711
482	if (SvPOKp (sv))	712	if (SvPOKp (sv))
483	{	713	{
484	STRLEN len;	714	STRLEN len;
485	char *str = SvPV (sv, len);	715	char *str = SvPV (sv, len);
486	encode_str (enc, SvUTF8 (sv), str, len);	716	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
487	}	717	}
488	else if (SvNOKp (sv))	718	else if (SvNOKp (sv))
489	encode_nv (enc, sv);	719	encode_nv (enc, sv);
490	else if (SvIOKp (sv))	720	else if (SvIOKp (sv))
491	{	721	encode_int (enc, sv);
492	if (SvIsUV (sv))
493	encode_uint (enc, 0x00, SvUVX (sv));
494	else if (SvIVX (sv) >= 0)
495	encode_uint (enc, 0x00, SvIVX (sv));
496	else
497	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
498	}
499	else if (SvROK (sv))	722	else if (SvROK (sv))
500	encode_rv (enc, SvRV (sv));	723	encode_rv (enc, SvRV (sv));
501	else if (!SvOK (sv))	724	else if (!SvOK (sv))
502	encode_ch (enc, 0xe0 | 22);	725	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	726	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504	encode_ch (enc, 0xe0 | 23);	727	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505	else	728	else
506	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	729	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	730	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508	}	731	}
509		732
510	static SV *	733	static SV *
511	encode_cbor (SV *scalar, CBOR *cbor)	734	encode_cbor (SV *scalar, CBOR *cbor)
512	{	735	{
513	enc_t enc = { };	736	enc_t enc = { 0 };
514		737
515	enc.cbor = *cbor;	738	enc.cbor = *cbor;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	739	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	740	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	741	enc.end = SvEND (enc.sv);
519		742
520	SvPOK_only (enc.sv);	743	SvPOK_only (enc.sv);
521		744
522	if (cbor->flags & F_ALLOW_STRINGREF)	745	if (cbor->flags & F_PACK_STRINGS)
523	{	746	{
524	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	747	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	748	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	749	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527	}	750	}
…		…
550	U32 depth; // recursion depth	773	U32 depth; // recursion depth
551	U32 maxdepth; // recursion depth limit	774	U32 maxdepth; // recursion depth limit
552	AV *shareable;	775	AV *shareable;
553	AV *stringref;	776	AV *stringref;
554	SV *decode_tagged;	777	SV *decode_tagged;
		778	SV *err_sv; // optional sv for error, needs to be freed
555	} dec_t;	779	} dec_t;
556		780
557	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	781	// set dec->err to ERRSV
		782	ecb_cold static void
		783	err_errsv (dec_t *dec)
		784	{
		785	if (!dec->err)
		786	{
		787	dec->err_sv = newSVsv (ERRSV);
558		788
559	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	789	// chop off the trailing \n
		790	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		791	*SvEND (dec->err_sv) = 0;
560		792
		793	dec->err = SvPVutf8_nolen (dec->err_sv);
		794	}
		795	}
		796
		797	// the following functions are used to reduce code size and help the compiler to optimise
		798	ecb_cold static void
		799	err_set (dec_t *dec, const char *reason)
		800	{
		801	if (!dec->err)
		802	dec->err = reason;
		803	}
		804
		805	ecb_cold static void
		806	err_unexpected_end (dec_t *dec)
		807	{
		808	err_set (dec, "unexpected end of CBOR data");
		809	}
		810
		811	#define ERR_DO(do) SB do; goto fail; SE
		812	#define ERR(reason) ERR_DO (err_set (dec, reason))
		813	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		814
		815	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		816
561	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	817	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
562	#define DEC_DEC_DEPTH --dec->depth	818	#define DEC_DEC_DEPTH --dec->depth
563		819
564	static UV	820	static UV
565	decode_uint (dec_t *dec)	821	decode_uint (dec_t *dec)
566	{	822	{
567	switch (*dec->cur & 31)	823	U8 m = *dec->cur & MINOR_MASK;
568	{	824	++dec->cur;
569	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572	return *dec->cur++ & 31;
573		825
574	case 24:	826	if (ecb_expect_true (m < LENGTH_EXT1))
		827	return m;
		828	else if (ecb_expect_true (m == LENGTH_EXT1))
		829	{
575	WANT (2);	830	WANT (1);
576	dec->cur += 2;	831	dec->cur += 1;
577	return dec->cur[-1];	832	return dec->cur[-1];
578		833	}
579	case 25:	834	else if (ecb_expect_true (m == LENGTH_EXT2))
		835	{
580	WANT (3);	836	WANT (2);
581	dec->cur += 3;	837	dec->cur += 2;
582	return (((UV)dec->cur[-2]) << 8)	838	return (((UV)dec->cur[-2]) << 8)
583	| ((UV)dec->cur[-1]);	839	| ((UV)dec->cur[-1]);
584		840	}
585	case 26:	841	else if (ecb_expect_true (m == LENGTH_EXT4))
		842	{
586	WANT (5);	843	WANT (4);
587	dec->cur += 5;	844	dec->cur += 4;
588	return (((UV)dec->cur[-4]) << 24)	845	return (((UV)dec->cur[-4]) << 24)
589	| (((UV)dec->cur[-3]) << 16)	846	| (((UV)dec->cur[-3]) << 16)
590	| (((UV)dec->cur[-2]) << 8)	847	| (((UV)dec->cur[-2]) << 8)
591	| ((UV)dec->cur[-1]);	848	| ((UV)dec->cur[-1]);
592		849	}
593	case 27:	850	else if (ecb_expect_true (m == LENGTH_EXT8))
		851	{
594	WANT (9);	852	WANT (8);
595	dec->cur += 9;	853	dec->cur += 8;
		854
		855	return
		856	#if UVSIZE < 8
		857	0
		858	#else
596	return (((UV)dec->cur[-8]) << 56)	859	(((UV)dec->cur[-8]) << 56)
597	| (((UV)dec->cur[-7]) << 48)	860	| (((UV)dec->cur[-7]) << 48)
598	| (((UV)dec->cur[-6]) << 40)	861	| (((UV)dec->cur[-6]) << 40)
599	| (((UV)dec->cur[-5]) << 32)	862	| (((UV)dec->cur[-5]) << 32)
		863	#endif
600	| (((UV)dec->cur[-4]) << 24)	864	| (((UV)dec->cur[-4]) << 24)
601	| (((UV)dec->cur[-3]) << 16)	865	| (((UV)dec->cur[-3]) << 16)
602	| (((UV)dec->cur[-2]) << 8)	866	| (((UV)dec->cur[-2]) << 8)
603	| ((UV)dec->cur[-1]);	867	| ((UV)dec->cur[-1]);
604		868	}
605	default:	869	else
606	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	870	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607	}
608		871
609	fail:	872	fail:
610	return 0;	873	return 0;
611	}	874	}
612		875
…		…
617	{	880	{
618	AV *av = newAV ();	881	AV *av = newAV ();
619		882
620	DEC_INC_DEPTH;	883	DEC_INC_DEPTH;
621		884
622	if ((*dec->cur & 31) == 31)	885	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623	{	886	{
624	++dec->cur;	887	++dec->cur;
625		888
626	for (;;)	889	for (;;)
627	{	890	{
628	WANT (1);	891	WANT (1);
629		892
630	if (*dec->cur == (0xe0 | 31))	893	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
631	{	894	{
632	++dec->cur;	895	++dec->cur;
633	break;	896	break;
634	}	897	}
635		898
636	av_push (av, decode_sv (dec));	899	av_push (av, decode_sv (dec));
637	}	900	}
638	}	901	}
639	else	902	else
640	{	903	{
641	int i, len = decode_uint (dec);	904	UV i, len = decode_uint (dec);
642		905
		906	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643	av_fill (av, len - 1);	907	av_fill (av, len - 1);
644		908
645	for (i = 0; i < len; ++i)	909	for (i = 0; i < len; ++i)
646	AvARRAY (av)[i] = decode_sv (dec);	910	AvARRAY (av)[i] = decode_sv (dec);
647	}	911	}
648		912
649	DEC_DEC_DEPTH;	913	DEC_DEC_DEPTH;
650	return newRV_noinc ((SV *)av);	914	return newRV_noinc ((SV *)av);
651		915
652	fail:	916	fail:
653	SvREFCNT_dec (av);	917	SvREFCNT_dec_NN (av);
654	DEC_DEC_DEPTH;	918	DEC_DEC_DEPTH;
655	return &PL_sv_undef;	919	return &PL_sv_undef;
656	}	920	}
657		921
658	static void	922	static void
…		…
660	{	924	{
661	// for speed reasons, we specialcase single-string	925	// for speed reasons, we specialcase single-string
662	// byte or utf-8 strings as keys, but only when !stringref	926	// byte or utf-8 strings as keys, but only when !stringref
663		927
664	if (ecb_expect_true (!dec->stringref))	928	if (ecb_expect_true (!dec->stringref))
665	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	929	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666	{	930	{
667	I32 len = decode_uint (dec);	931	STRLEN len = decode_uint (dec);
668	char *key = (char *)dec->cur;	932	char *key = (char *)dec->cur;
669		933
		934	WANT (len);
670	dec->cur += len;	935	dec->cur += len;
671		936
672	if (ecb_expect_false (dec->stringref))
673	av_push (dec->stringref, newSVpvn (key, len));
674
675	hv_store (hv, key, len, decode_sv (dec), 0);	937	hv_store (hv, key, len, decode_sv (dec), 0);
676		938
677	return;	939	return;
678	}	940	}
679	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	941	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680	{	942	{
681	I32 len = decode_uint (dec);	943	STRLEN len = decode_uint (dec);
682	char *key = (char *)dec->cur;	944	char *key = (char *)dec->cur;
683		945
		946	WANT (len);
684	dec->cur += len;	947	dec->cur += len;
685		948
686	if (ecb_expect_false (dec->stringref))	949	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	950	if (!is_utf8_string (key, len))
		951	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688		952
689	hv_store (hv, key, -len, decode_sv (dec), 0);	953	hv_store (hv, key, -len, decode_sv (dec), 0);
690		954
691	return;	955	return;
692	}	956	}
693		957
694	SV *k = decode_sv (dec);	958	SV *k = decode_sv (dec);
695	SV *v = decode_sv (dec);	959	SV *v = decode_sv (dec);
696		960
		961	// we leak memory if uncaught exceptions are thrown by random magical
		962	// methods, and this is hopefully the only place where it can happen,
		963	// so if there is a chance of an exception, take the very slow path.
		964	// since catching exceptions is "undocumented/internal/forbidden" by
		965	// the new p5p powers, we need to call out to a perl function :/
		966	if (ecb_expect_false (SvAMAGIC (k)))
		967	{
		968	dSP;
		969
		970	ENTER; SAVETMPS;
		971	PUSHMARK (SP);
		972	EXTEND (SP, 3);
		973	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		974	PUSHs (sv_2mortal (k));
		975	PUSHs (sv_2mortal (v));
		976
		977	PUTBACK;
		978	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		979	SPAGAIN;
		980
		981	FREETMPS; LEAVE;
		982
		983	if (SvTRUE (ERRSV))
		984	ERR_ERRSV;
		985
		986	return;
		987	}
		988
697	hv_store_ent (hv, k, v, 0);	989	hv_store_ent (hv, k, v, 0);
698	SvREFCNT_dec (k);	990	SvREFCNT_dec_NN (k);
		991
		992	fail:
		993	;
699	}	994	}
700		995
701	static SV *	996	static SV *
702	decode_hv (dec_t *dec)	997	decode_hv (dec_t *dec)
703	{	998	{
704	HV *hv = newHV ();	999	HV *hv = newHV ();
705		1000
706	DEC_INC_DEPTH;	1001	DEC_INC_DEPTH;
707		1002
708	if ((*dec->cur & 31) == 31)	1003	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709	{	1004	{
710	++dec->cur;	1005	++dec->cur;
711		1006
712	for (;;)	1007	for (;;)
713	{	1008	{
714	WANT (1);	1009	WANT (1);
715		1010
716	if (*dec->cur == (0xe0 | 31))	1011	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
717	{	1012	{
718	++dec->cur;	1013	++dec->cur;
719	break;	1014	break;
720	}	1015	}
721		1016
722	decode_he (dec, hv);	1017	decode_he (dec, hv);
723	}	1018	}
724	}	1019	}
725	else	1020	else
726	{	1021	{
727	int pairs = decode_uint (dec);	1022	UV pairs = decode_uint (dec);
		1023
		1024	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
728		1025
729	while (pairs--)	1026	while (pairs--)
730	decode_he (dec, hv);	1027	decode_he (dec, hv);
731	}	1028	}
732		1029
733	DEC_DEC_DEPTH;	1030	DEC_DEC_DEPTH;
734	return newRV_noinc ((SV *)hv);	1031	return newRV_noinc ((SV *)hv);
735		1032
736	fail:	1033	fail:
737	SvREFCNT_dec (hv);	1034	SvREFCNT_dec_NN (hv);
738	DEC_DEC_DEPTH;	1035	DEC_DEC_DEPTH;
739	return &PL_sv_undef;	1036	return &PL_sv_undef;
740	}	1037	}
741		1038
742	static SV *	1039	static SV *
743	decode_str (dec_t *dec, int utf8)	1040	decode_str (dec_t *dec, int utf8)
744	{	1041	{
745	SV *sv = 0;	1042	SV *sv = 0;
746		1043
747	if ((*dec->cur & 31) == 31)	1044	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
748	{	1045	{
		1046	// indefinite length strings
749	++dec->cur;	1047	++dec->cur;
750		1048
		1049	U8 major = *dec->cur & MAJOR_MISC;
		1050
751	sv = newSVpvn ("", 0);	1051	sv = newSVpvn ("", 0);
752		1052
753	// not very fast, and certainly not robust against illegal input
754	for (;;)	1053	for (;;)
755	{	1054	{
756	WANT (1);	1055	WANT (1);
757		1056
758	if (*dec->cur == (0xe0 | 31))	1057	if ((*dec->cur - major) > LENGTH_EXT8)
		1058	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759	{	1059	{
760	++dec->cur;	1060	++dec->cur;
761	break;	1061	break;
762	}	1062	}
		1063	else
		1064	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763		1065
764	sv_catsv (sv, decode_sv (dec));	1066	STRLEN len = decode_uint (dec);
		1067
		1068	WANT (len);
		1069	sv_catpvn (sv, dec->cur, len);
		1070	dec->cur += len;
765	}	1071	}
766	}	1072	}
767	else	1073	else
768	{	1074	{
769	STRLEN len = decode_uint (dec);	1075	STRLEN len = decode_uint (dec);
…		…
776	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1082	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1083	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778	}	1084	}
779		1085
780	if (utf8)	1086	if (utf8)
		1087	{
		1088	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1089	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1090	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1091
781	SvUTF8_on (sv);	1092	SvUTF8_on (sv);
		1093	}
782		1094
783	return sv;	1095	return sv;
784		1096
785	fail:	1097	fail:
786	SvREFCNT_dec (sv);	1098	SvREFCNT_dec (sv);
…		…
805	sv = newRV_noinc (decode_sv (dec));	1117	sv = newRV_noinc (decode_sv (dec));
806	break;	1118	break;
807		1119
808	case CBOR_TAG_STRINGREF_NAMESPACE:	1120	case CBOR_TAG_STRINGREF_NAMESPACE:
809	{	1121	{
		1122	// do not use SAVETMPS/FREETMPS, as these will
		1123	// erase mortalised caches, e.g. "shareable"
810	ENTER; SAVETMPS;	1124	ENTER;
811		1125
812	SAVESPTR (dec->stringref);	1126	SAVESPTR (dec->stringref);
813	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1127	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
814		1128
815	sv = decode_sv (dec);	1129	sv = decode_sv (dec);
816		1130
817	FREETMPS; LEAVE;	1131	LEAVE;
818	}	1132	}
819	break;	1133	break;
820		1134
821	case CBOR_TAG_STRINGREF:	1135	case CBOR_TAG_STRINGREF:
822	{	1136	{
823	if ((*dec->cur >> 5) != 0)	1137	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1138	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825		1139
826	UV idx = decode_uint (dec);	1140	UV idx = decode_uint (dec);
827		1141
828	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1142	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
829	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1143	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
830		1144
831	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1145	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
832	}	1146	}
833	break;	1147	break;
…		…
835	case CBOR_TAG_VALUE_SHAREABLE:	1149	case CBOR_TAG_VALUE_SHAREABLE:
836	{	1150	{
837	if (ecb_expect_false (!dec->shareable))	1151	if (ecb_expect_false (!dec->shareable))
838	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1152	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839		1153
		1154	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1155	{
840	sv = newSV (0);	1156	sv = newSV (0);
841	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1157	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842		1158
843	SV *osv = decode_sv (dec);	1159	SV *osv = decode_sv (dec);
844	sv_setsv (sv, osv);	1160	sv_setsv (sv, osv);
845	SvREFCNT_dec_NN (osv);	1161	SvREFCNT_dec_NN (osv);
		1162	}
		1163	else
		1164	{
		1165	av_push (dec->shareable, &PL_sv_undef);
		1166	int idx = AvFILLp (dec->shareable);
		1167	sv = decode_sv (dec);
		1168	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1169	}
846	}	1170	}
847	break;	1171	break;
848		1172
849	case CBOR_TAG_VALUE_SHAREDREF:	1173	case CBOR_TAG_VALUE_SHAREDREF:
850	{	1174	{
851	if ((*dec->cur >> 5) != 0)	1175	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1176	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853		1177
854	UV idx = decode_uint (dec);	1178	UV idx = decode_uint (dec);
855		1179
856	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1180	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
857	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1181	ERR ("corrupted CBOR data (sharedref index out of bounds)");
858		1182
859	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1183	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1184
		1185	if (sv == &PL_sv_undef)
		1186	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860	}	1187	}
861	break;	1188	break;
862		1189
863	case CBOR_TAG_PERL_OBJECT:	1190	case CBOR_TAG_PERL_OBJECT:
864	{	1191	{
		1192	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1193	goto filter;
		1194
865	sv = decode_sv (dec);	1195	sv = decode_sv (dec);
866		1196
867	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1197	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
868	ERR ("corrupted CBOR data (non-array perl object)");	1198	ERR ("corrupted CBOR data (non-array perl object)");
869		1199
…		…
879	if (!method)	1209	if (!method)
880	ERR ("cannot decode perl-object (package does not have a THAW method)");	1210	ERR ("cannot decode perl-object (package does not have a THAW method)");
881		1211
882	dSP;	1212	dSP;
883		1213
884	ENTER; SAVETMPS; PUSHMARK (SP);	1214	ENTER; SAVETMPS;
		1215	PUSHMARK (SP);
885	EXTEND (SP, len + 1);	1216	EXTEND (SP, len + 1);
886	// we re-bless the reference to get overload and other niceties right	1217	// we re-bless the reference to get overload and other niceties right
887	PUSHs (*av_fetch (av, 0, 1));	1218	PUSHs (*av_fetch (av, 0, 1));
888	PUSHs (sv_cbor);	1219	PUSHs (sv_cbor);
889		1220
…		…
897	SPAGAIN;	1228	SPAGAIN;
898		1229
899	if (SvTRUE (ERRSV))	1230	if (SvTRUE (ERRSV))
900	{	1231	{
901	FREETMPS; LEAVE;	1232	FREETMPS; LEAVE;
902	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1233	ERR_ERRSV;
903	}	1234	}
904		1235
905	SvREFCNT_dec (sv);	1236	SvREFCNT_dec_NN (sv);
906	sv = SvREFCNT_inc (POPs);	1237	sv = SvREFCNT_inc (POPs);
907		1238
908	PUTBACK;	1239	PUTBACK;
909		1240
910	FREETMPS; LEAVE;	1241	FREETMPS; LEAVE;
911	}	1242	}
912	break;	1243	break;
913		1244
914	default:	1245	default:
		1246	filter:
915	{	1247	{
		1248	SV *tag_sv = newSVuv (tag);
		1249
916	sv = decode_sv (dec);	1250	sv = decode_sv (dec);
917		1251
918	dSP;	1252	dSP;
919	ENTER; SAVETMPS; PUSHMARK (SP);	1253	ENTER; SAVETMPS;
		1254	PUSHMARK (SP);
920	EXTEND (SP, 2);	1255	EXTEND (SP, 2);
921	PUSHs (newSVuv (tag));	1256	PUSHs (tag_sv);
922	PUSHs (sv);	1257	PUSHs (sv);
923		1258
924	PUTBACK;	1259	PUTBACK;
925	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1260	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
926	SPAGAIN;	1261	SPAGAIN;
927		1262
928	if (SvTRUE (ERRSV))	1263	if (SvTRUE (ERRSV))
929	{	1264	{
		1265	SvREFCNT_dec_NN (tag_sv);
930	FREETMPS; LEAVE;	1266	FREETMPS; LEAVE;
931	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1267	ERR_ERRSV;
932	}	1268	}
933		1269
934	if (count)	1270	if (count)
935	{	1271	{
		1272	SvREFCNT_dec_NN (tag_sv);
936	SvREFCNT_dec (sv);	1273	SvREFCNT_dec_NN (sv);
937	sv = SvREFCNT_inc (POPs);	1274	sv = SvREFCNT_inc_NN (TOPs);
		1275	SP -= count;
938	}	1276	}
939	else	1277	else
940	{	1278	{
941	AV *av = newAV ();	1279	AV *av = newAV ();
942	av_push (av, newSVuv (tag));	1280	av_push (av, tag_sv);
943	av_push (av, sv);	1281	av_push (av, sv);
944		1282
945	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1283	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
946	? cbor_tagged_stash	1284	? cbor_tagged_stash
947	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1285	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
965	static SV *	1303	static SV *
966	decode_sv (dec_t *dec)	1304	decode_sv (dec_t *dec)
967	{	1305	{
968	WANT (1);	1306	WANT (1);
969		1307
970	switch (*dec->cur >> 5)	1308	switch (*dec->cur >> MAJOR_SHIFT)
971	{	1309	{
972	case 0: // unsigned int	1310	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973	return newSVuv (decode_uint (dec));	1311	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974	case 1: // negative int	1312	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975	return newSViv (-1 - (IV)decode_uint (dec));	1313	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976	case 2: // octet string	1314	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977	return decode_str (dec, 0);	1315	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978	case 3: // utf-8 string	1316	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979	return decode_str (dec, 1);	1317
980	case 4: // array	1318	case MAJOR_MISC >> MAJOR_SHIFT:
981	return decode_av (dec);
982	case 5: // map
983	return decode_hv (dec);
984	case 6: // tag
985	return decode_tagged (dec);
986	case 7: // misc
987	switch (*dec->cur++ & 31)	1319	switch (*dec->cur++ & MINOR_MASK)
988	{	1320	{
989	case 20:	1321	case SIMPLE_FALSE:
990	#if CBOR_SLOW	1322	#if CBOR_SLOW
991	types_false = get_bool ("Types::Serialiser::false");	1323	types_false = get_bool ("Types::Serialiser::false");
992	#endif	1324	#endif
993	return newSVsv (types_false);	1325	return newSVsv (types_false);
994	case 21:	1326	case SIMPLE_TRUE:
995	#if CBOR_SLOW	1327	#if CBOR_SLOW
996	types_true = get_bool ("Types::Serialiser::true");	1328	types_true = get_bool ("Types::Serialiser::true");
997	#endif	1329	#endif
998	return newSVsv (types_true);	1330	return newSVsv (types_true);
999	case 22:	1331	case SIMPLE_NULL:
1000	return newSVsv (&PL_sv_undef);	1332	return newSVsv (&PL_sv_undef);
1001	case 23:	1333	case SIMPLE_UNDEF:
1002	#if CBOR_SLOW	1334	#if CBOR_SLOW
1003	types_error = get_bool ("Types::Serialiser::error");	1335	types_error = get_bool ("Types::Serialiser::error");
1004	#endif	1336	#endif
1005	return newSVsv (types_error);	1337	return newSVsv (types_error);
1006		1338
1007	case 25:	1339	case MISC_FLOAT16:
1008	{	1340	{
1009	WANT (2);	1341	WANT (2);
1010		1342
1011	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1343	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012	dec->cur += 2;	1344	dec->cur += 2;
1013		1345
1014	return newSVnv (ecb_binary16_to_float (fp));	1346	return newSVnv (ecb_binary16_to_float (fp));
1015	}	1347	}
1016		1348
1017	case 26:	1349	case MISC_FLOAT32:
1018	{	1350	{
1019	uint32_t fp;	1351	uint32_t fp;
1020	WANT (4);	1352	WANT (4);
1021	memcpy (&fp, dec->cur, 4);	1353	memcpy (&fp, dec->cur, 4);
1022	dec->cur += 4;	1354	dec->cur += 4;
…		…
1025	fp = ecb_bswap32 (fp);	1357	fp = ecb_bswap32 (fp);
1026		1358
1027	return newSVnv (ecb_binary32_to_float (fp));	1359	return newSVnv (ecb_binary32_to_float (fp));
1028	}	1360	}
1029		1361
1030	case 27:	1362	case MISC_FLOAT64:
1031	{	1363	{
1032	uint64_t fp;	1364	uint64_t fp;
1033	WANT (8);	1365	WANT (8);
1034	memcpy (&fp, dec->cur, 8);	1366	memcpy (&fp, dec->cur, 8);
1035	dec->cur += 8;	1367	dec->cur += 8;
…		…
1038	fp = ecb_bswap64 (fp);	1370	fp = ecb_bswap64 (fp);
1039		1371
1040	return newSVnv (ecb_binary64_to_double (fp));	1372	return newSVnv (ecb_binary64_to_double (fp));
1041	}	1373	}
1042		1374
1043	// 0..19 unassigned	1375	// 0..19 unassigned simple
1044	// 24 reserved + unassigned (reserved values are not encodable)	1376	// 24 reserved + unassigned simple (reserved values are not encodable)
		1377	// 28-30 unassigned misc
		1378	// 31 break code
1045	default:	1379	default:
1046	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1380	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047	}	1381	}
1048		1382
1049	break;	1383	break;
1050	}	1384	}
1051		1385
…		…
1054	}	1388	}
1055		1389
1056	static SV *	1390	static SV *
1057	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1391	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058	{	1392	{
1059	dec_t dec = { };	1393	dec_t dec = { 0 };
1060	SV *sv;	1394	SV *sv;
1061	STRLEN len;	1395	STRLEN len;
1062	char *data = SvPVbyte (string, len);	1396	char *data = SvPVbyte (string, len);
1063		1397
1064	if (len > cbor->max_size && cbor->max_size)	1398	if (len > cbor->max_size && cbor->max_size)
…		…
1078	if (dec.cur != dec.end && !dec.err)	1412	if (dec.cur != dec.end && !dec.err)
1079	dec.err = "garbage after CBOR object";	1413	dec.err = "garbage after CBOR object";
1080		1414
1081	if (dec.err)	1415	if (dec.err)
1082	{	1416	{
		1417	if (dec.shareable)
		1418	{
		1419	// need to break cyclic links, which would all be in shareable
		1420	int i;
		1421	SV **svp;
		1422
		1423	for (i = av_len (dec.shareable) + 1; i--; )
		1424	if ((svp = av_fetch (dec.shareable, i, 0)))
		1425	sv_setsv (*svp, &PL_sv_undef);
		1426	}
		1427
1083	SvREFCNT_dec (sv);	1428	SvREFCNT_dec_NN (sv);
		1429
		1430	if (dec.err_sv)
		1431	sv_2mortal (dec.err_sv);
		1432
1084	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1433	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1085	}	1434	}
1086		1435
1087	sv = sv_2mortal (sv);	1436	sv = sv_2mortal (sv);
1088		1437
1089	return sv;	1438	return sv;
1090	}	1439	}
1091		1440
		1441	/////////////////////////////////////////////////////////////////////////////
		1442	// incremental parser
		1443
		1444	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1445
		1446	// returns 0 for notyet, 1 for success or error
		1447	static int
		1448	incr_parse (CBOR *self, SV *cborstr)
		1449	{
		1450	STRLEN cur;
		1451	SvPV (cborstr, cur);
		1452
		1453	while (ecb_expect_true (self->incr_need <= cur))
		1454	{
		1455	// table of integer count bytes
		1456	static I8 incr_len[MINOR_MASK + 1] = {
		1457	0, 0, 0, 0, 0, 0, 0, 0,
		1458	0, 0, 0, 0, 0, 0, 0, 0,
		1459	0, 0, 0, 0, 0, 0, 0, 0,
		1460	1, 2, 4, 8,-1,-1,-1,-2
		1461	};
		1462
		1463	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1464	U8 m = *p & MINOR_MASK;
		1465	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1466	I8 ilen = incr_len[m];
		1467
		1468	self->incr_need = self->incr_pos + 1;
		1469
		1470	if (ecb_expect_false (ilen < 0))
		1471	{
		1472	if (m != MINOR_INDEF)
		1473	return 1; // error
		1474
		1475	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1476	{
		1477	if (count >= 0)
		1478	return 1; // error
		1479
		1480	count = 1;
		1481	}
		1482	else
		1483	{
		1484	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1485	count = -1;
		1486	}
		1487	}
		1488	else
		1489	{
		1490	self->incr_need += ilen;
		1491	if (ecb_expect_false (self->incr_need > cur))
		1492	return 0;
		1493
		1494	int major = *p >> MAJOR_SHIFT;
		1495
		1496	switch (major)
		1497	{
		1498	case MAJOR_TAG >> MAJOR_SHIFT:
		1499	++count; // tags merely prefix another value
		1500	break;
		1501
		1502	case MAJOR_BYTES >> MAJOR_SHIFT:
		1503	case MAJOR_TEXT >> MAJOR_SHIFT:
		1504	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1505	case MAJOR_MAP >> MAJOR_SHIFT:
		1506	{
		1507	UV len;
		1508
		1509	if (ecb_expect_false (ilen))
		1510	{
		1511	len = 0;
		1512
		1513	do {
		1514	len = (len << 8) | *++p;
		1515	} while (--ilen);
		1516	}
		1517	else
		1518	len = m;
		1519
		1520	switch (major)
		1521	{
		1522	case MAJOR_BYTES >> MAJOR_SHIFT:
		1523	case MAJOR_TEXT >> MAJOR_SHIFT:
		1524	self->incr_need += len;
		1525	if (ecb_expect_false (self->incr_need > cur))
		1526	return 0;
		1527
		1528	break;
		1529
		1530	case MAJOR_MAP >> MAJOR_SHIFT:
		1531	len <<= 1;
		1532	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1533	if (len)
		1534	{
		1535	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1536	count = len + 1;
		1537	}
		1538	break;
		1539	}
		1540	}
		1541	}
		1542	}
		1543
		1544	self->incr_pos = self->incr_need;
		1545
		1546	if (count > 0)
		1547	{
		1548	while (!--count)
		1549	{
		1550	if (!AvFILLp (self->incr_count))
		1551	return 1; // done
		1552
		1553	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1554	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1555	}
		1556
		1557	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1558	}
		1559	}
		1560
		1561	return 0;
		1562	}
		1563
		1564
1092	/////////////////////////////////////////////////////////////////////////////	1565	/////////////////////////////////////////////////////////////////////////////
1093	// XS interface functions	1566	// XS interface functions
1094		1567
1095	MODULE = CBOR::XS PACKAGE = CBOR::XS	1568	MODULE = CBOR::XS PACKAGE = CBOR::XS
1096		1569
…		…
1108		1581
1109	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1582	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1110		1583
1111	sv_cbor = newSVpv ("CBOR", 0);	1584	sv_cbor = newSVpv ("CBOR", 0);
1112	SvREADONLY_on (sv_cbor);	1585	SvREADONLY_on (sv_cbor);
		1586
		1587	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1113	}	1588	}
1114		1589
1115	PROTOTYPES: DISABLE	1590	PROTOTYPES: DISABLE
1116		1591
1117	void CLONE (...)	1592	void CLONE (...)
…		…
1136	void shrink (CBOR *self, int enable = 1)	1611	void shrink (CBOR *self, int enable = 1)
1137	ALIAS:	1612	ALIAS:
1138	shrink = F_SHRINK	1613	shrink = F_SHRINK
1139	allow_unknown = F_ALLOW_UNKNOWN	1614	allow_unknown = F_ALLOW_UNKNOWN
1140	allow_sharing = F_ALLOW_SHARING	1615	allow_sharing = F_ALLOW_SHARING
1141	allow_stringref = F_ALLOW_STRINGREF	1616	allow_cycles = F_ALLOW_CYCLES
		1617	forbid_objects = F_FORBID_OBJECTS
		1618	pack_strings = F_PACK_STRINGS
		1619	text_keys = F_TEXT_KEYS
		1620	text_strings = F_TEXT_STRINGS
		1621	validate_utf8 = F_VALIDATE_UTF8
1142	PPCODE:	1622	PPCODE:
1143	{	1623	{
1144	if (enable)	1624	if (enable)
1145	self->flags |= ix;	1625	self->flags |= ix;
1146	else	1626	else
…		…
1152	void get_shrink (CBOR *self)	1632	void get_shrink (CBOR *self)
1153	ALIAS:	1633	ALIAS:
1154	get_shrink = F_SHRINK	1634	get_shrink = F_SHRINK
1155	get_allow_unknown = F_ALLOW_UNKNOWN	1635	get_allow_unknown = F_ALLOW_UNKNOWN
1156	get_allow_sharing = F_ALLOW_SHARING	1636	get_allow_sharing = F_ALLOW_SHARING
1157	get_allow_stringref = F_ALLOW_STRINGREF	1637	get_allow_cycles = F_ALLOW_CYCLES
		1638	get_forbid_objects = F_FORBID_OBJECTS
		1639	get_pack_strings = F_PACK_STRINGS
		1640	get_text_keys = F_TEXT_KEYS
		1641	get_text_strings = F_TEXT_STRINGS
		1642	get_validate_utf8 = F_VALIDATE_UTF8
1158	PPCODE:	1643	PPCODE:
1159	XPUSHs (boolSV (self->flags & ix));	1644	XPUSHs (boolSV (self->flags & ix));
1160		1645
1161	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1646	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162	PPCODE:	1647	PPCODE:
…		…
1211	EXTEND (SP, 2);	1696	EXTEND (SP, 2);
1212	PUSHs (sv);	1697	PUSHs (sv);
1213	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1698	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214	}	1699	}
1215		1700
		1701	void incr_parse (CBOR *self, SV *cborstr)
		1702	ALIAS:
		1703	incr_parse_multiple = 1
		1704	PPCODE:
		1705	{
		1706	if (SvUTF8 (cborstr))
		1707	sv_utf8_downgrade (cborstr, 0);
		1708
		1709	if (!self->incr_count)
		1710	{
		1711	self->incr_count = newAV ();
		1712	self->incr_pos = 0;
		1713	self->incr_need = 1;
		1714
		1715	av_push (self->incr_count, newSViv (1));
		1716	}
		1717
		1718	do
		1719	{
		1720	if (!incr_parse (self, cborstr))
		1721	{
		1722	if (self->incr_need > self->max_size && self->max_size)
		1723	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1724	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1725
		1726	break;
		1727	}
		1728
		1729	SV *sv;
		1730	char *offset;
		1731
		1732	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1733	XPUSHs (sv);
		1734
		1735	sv_chop (cborstr, offset);
		1736
		1737	av_clear (self->incr_count);
		1738	av_push (self->incr_count, newSViv (1));
		1739
		1740	self->incr_pos = 0;
		1741	self->incr_need = self->incr_pos + 1;
		1742	}
		1743	while (ix);
		1744	}
		1745
		1746	void incr_reset (CBOR *self)
		1747	CODE:
		1748	{
		1749	SvREFCNT_dec (self->incr_count);
		1750	self->incr_count = 0;
		1751	}
		1752
1216	void DESTROY (CBOR *self)	1753	void DESTROY (CBOR *self)
1217	PPCODE:	1754	PPCODE:
1218	cbor_free (self);	1755	cbor_free (self);
1219		1756
1220	PROTOTYPES: ENABLE	1757	PROTOTYPES: ENABLE
1221		1758
1222	void encode_cbor (SV *scalar)	1759	void encode_cbor (SV *scalar)
		1760	ALIAS:
		1761	encode_cbor = 0
		1762	encode_cbor_sharing = F_ALLOW_SHARING
1223	PPCODE:	1763	PPCODE:
1224	{	1764	{
1225	CBOR cbor;	1765	CBOR cbor;
1226	cbor_init (&cbor);	1766	cbor_init (&cbor);
		1767	cbor.flags |= ix;
1227	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1768	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228	XPUSHs (scalar);	1769	XPUSHs (scalar);
1229	}	1770	}
1230		1771
1231	void decode_cbor (SV *cborstr)	1772	void decode_cbor (SV *cborstr)
…		…
1235	cbor_init (&cbor);	1776	cbor_init (&cbor);
1236	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1777	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1237	XPUSHs (cborstr);	1778	XPUSHs (cborstr);
1238	}	1779	}
1239		1780
		1781	#ifdef __AFL_COMPILER
		1782
		1783	void
		1784	afl_init ()
		1785	CODE:
		1786	__AFL_INIT ();
		1787
		1788	int
		1789	afl_loop (unsigned int count = 10000)
		1790	CODE:
		1791	RETVAL = __AFL_LOOP (count);
		1792	OUTPUT:
		1793	RETVAL
		1794
		1795	#endif
		1796

Diff Legend

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