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 1.69 by root, Mon Nov 30 20:38:25 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_INT = 1,
		109	AS_BYTES = 2,
		110	AS_TEXT = 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_bool (enc_t *enc, int istrue)
		424	{
		425	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		426	}
		427
		428	ecb_inline void
		429	encode_forced (enc_t *enc, UV type, SV *sv)
		430	{
		431	switch (type)
		432	{
		433	case AS_CBOR:
		434	{
		435	STRLEN len;
		436	char *str = SvPVbyte (sv, len);
		437
		438	need (enc, len);
		439	memcpy (enc->cur, str, len);
		440	enc->cur += len;
		441	}
		442	break;
		443
		444	case AS_BYTES:
		445	{
		446	STRLEN len;
		447	char *str = SvPVbyte (sv, len);
		448	encode_strref (enc, 0, 0, str, len);
		449	}
		450	break;
		451
		452	case AS_TEXT:
		453	{
		454	STRLEN len;
		455	char *str = SvPVutf8 (sv, len);
		456	encode_strref (enc, 1, 1, str, len);
		457	}
		458	break;
		459
		460	case AS_INT: encode_int (enc, sv); break;
		461
		462	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		463	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		464	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		465
		466	default:
		467	croak ("encountered malformed CBOR::XS::Tagged object");
		468	}
258	}	469	}
259		470
260	static void encode_sv (enc_t *enc, SV *sv);	471	static void encode_sv (enc_t *enc, SV *sv);
261		472
262	static void	473	static void
…		…
267	if (enc->depth >= enc->cbor.max_depth)	478	if (enc->depth >= enc->cbor.max_depth)
268	croak (ERR_NESTING_EXCEEDED);	479	croak (ERR_NESTING_EXCEEDED);
269		480
270	++enc->depth;	481	++enc->depth;
271		482
272	encode_uint (enc, 0x80, len + 1);	483	encode_uint (enc, MAJOR_ARRAY, len + 1);
273		484
		485	if (ecb_expect_false (SvMAGICAL (av)))
274	for (i = 0; i <= len; ++i)	486	for (i = 0; i <= len; ++i)
275	{	487	{
276	SV **svp = av_fetch (av, i, 0);	488	SV **svp = av_fetch (av, i, 0);
277	encode_sv (enc, svp ? *svp : &PL_sv_undef);	489	encode_sv (enc, svp ? *svp : &PL_sv_undef);
278	}	490	}
		491	else
		492	for (i = 0; i <= len; ++i)
		493	{
		494	SV *sv = AvARRAY (av)[i];
		495	encode_sv (enc, sv ? sv : &PL_sv_undef);
		496	}
279		497
280	--enc->depth;	498	--enc->depth;
281	}	499	}
282		500
283	static void	501	static void
…		…
291	++enc->depth;	509	++enc->depth;
292		510
293	int pairs = hv_iterinit (hv);	511	int pairs = hv_iterinit (hv);
294	int mg = SvMAGICAL (hv);	512	int mg = SvMAGICAL (hv);
295		513
296	if (mg)	514	if (ecb_expect_false (mg))
297	encode_ch (enc, 0xa0 | 31);	515	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298	else	516	else
299	encode_uint (enc, 0xa0, pairs);	517	encode_uint (enc, MAJOR_MAP, pairs);
300		518
301	while ((he = hv_iternext (hv)))	519	while ((he = hv_iternext (hv)))
302	{	520	{
303	if (HeKLEN (he) == HEf_SVKEY)	521	if (HeKLEN (he) == HEf_SVKEY)
304	encode_sv (enc, HeSVKEY (he));	522	encode_sv (enc, HeSVKEY (he));
305	else	523	else
306	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	524	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307		525
308	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	526	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309	}	527	}
310		528
311	if (mg)	529	if (ecb_expect_false (mg))
312	encode_ch (enc, 0xe0 | 31);	530	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313		531
314	--enc->depth;	532	--enc->depth;
315	}	533	}
316		534
317	// encode objects, arrays and special \0=false and \1=true values.	535	// encode objects, arrays and special \0=false and \1=true values.
318	static void	536	static void
319	encode_rv (enc_t *enc, SV *sv)	537	encode_rv (enc_t *enc, SV *sv)
320	{	538	{
321	SvGETMAGIC (sv);	539	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		540
345	svtype svt = SvTYPE (sv);	541	svtype svt = SvTYPE (sv);
346		542
347	if (ecb_expect_false (SvOBJECT (sv)))	543	if (ecb_expect_false (SvOBJECT (sv)))
348	{	544	{
…		…
355	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	551	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356	? cbor_tagged_stash	552	? cbor_tagged_stash
357	: gv_stashpv ("CBOR::XS::Tagged" , 1);	553	: gv_stashpv ("CBOR::XS::Tagged" , 1);
358		554
359	HV *stash = SvSTASH (sv);	555	HV *stash = SvSTASH (sv);
360	GV *method;
361		556
362	if (stash == boolean_stash)	557	if (stash == boolean_stash)
363	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	558	{
		559	encode_bool (enc, SvIV (sv));
		560	return;
		561	}
364	else if (stash == error_stash)	562	else if (stash == error_stash)
365	encode_ch (enc, 0xe0 | 23);	563	{
		564	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		565	return;
		566	}
366	else if (stash == tagged_stash)	567	else if (stash == tagged_stash)
367	{	568	{
368	if (svt != SVt_PVAV)	569	if (svt != SVt_PVAV)
369	croak ("encountered CBOR::XS::Tagged object that isn't an array");	570	croak ("encountered CBOR::XS::Tagged object that isn't an array");
370		571
		572	switch (av_len ((AV *)sv))
		573	{
		574	case 2-1:
		575	// actually a tagged value
371	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	576	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	577	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		578	break;
		579
		580	case 3-1:
		581	// a forced type [value, type, undef]
		582	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		583	break;
		584
		585	default:
		586	croak ("encountered malformed CBOR::XS::Tagged object");
		587	}
		588
		589	return;
		590	}
		591	}
		592
		593	if (ecb_expect_false (SvREFCNT (sv) > 1)
		594	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		595	{
		596	if (ecb_expect_false (!enc->shareable))
		597	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		598
		599	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		600
		601	if (SvOK (*svp))
373	}	602	{
		603	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		604	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		605	return;
		606	}
		607	else
		608	{
		609	sv_setuv (*svp, enc->shareable_idx);
		610	++enc->shareable_idx;
		611	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		612	}
		613	}
		614
		615	if (ecb_expect_false (SvOBJECT (sv)))
		616	{
		617	HV *stash = SvSTASH (sv);
		618	GV *method;
		619
		620	if (enc->cbor.flags & F_FORBID_OBJECTS)
		621	croak ("encountered object '%s', but forbid_objects is enabled",
		622	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	623	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375	{	624	{
376	dSP;	625	dSP;
377		626
378	ENTER; SAVETMPS; PUSHMARK (SP);	627	ENTER; SAVETMPS;
		628	PUSHMARK (SP);
379	// we re-bless the reference to get overload and other niceties right	629	// we re-bless the reference to get overload and other niceties right
380	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	630	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381		631
382	PUTBACK;	632	PUTBACK;
383	// G_SCALAR ensures that return value is 1	633	// G_SCALAR ensures that return value is 1
…		…
396	}	646	}
397	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	647	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398	{	648	{
399	dSP;	649	dSP;
400		650
401	ENTER; SAVETMPS; PUSHMARK (SP);	651	ENTER; SAVETMPS;
		652	PUSHMARK (SP);
402	EXTEND (SP, 2);	653	EXTEND (SP, 2);
403	// we re-bless the reference to get overload and other niceties right	654	// we re-bless the reference to get overload and other niceties right
404	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	655	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405	PUSHs (sv_cbor);	656	PUSHs (sv_cbor);
406		657
…		…
411	// catch this surprisingly common error	662	// catch this surprisingly common error
412	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	663	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));	664	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414		665
415	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	666	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416	encode_uint (enc, 0x80, count + 1);	667	encode_uint (enc, MAJOR_ARRAY, count + 1);
417	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	668	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418		669
419	while (count)	670	{
		671	int i;
		672
		673	for (i = 0; i < count; ++i)
420	encode_sv (enc, SP[1 - count--]);	674	encode_sv (enc, SP[i + 1 - count]);
		675
		676	SP -= count;
		677	}
421		678
422	PUTBACK;	679	PUTBACK;
423		680
424	FREETMPS; LEAVE;	681	FREETMPS; LEAVE;
425	}	682	}
…		…
443	{	700	{
444	double nv = SvNVX (sv);	701	double nv = SvNVX (sv);
445		702
446	need (enc, 9);	703	need (enc, 9);
447		704
448	if (ecb_expect_false (nv == (U32)nv))	705	if (ecb_expect_false (nv == (NV)(U32)nv))
449	encode_uint (enc, 0x00, (U32)nv);	706	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450	//TODO: maybe I32?	707	//TODO: maybe I32?
451	else if (ecb_expect_false (nv == (float)nv))	708	else if (ecb_expect_false (nv == (float)nv))
452	{	709	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	710	else
464	{	711	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	}	712	}
476		713
477	static void	714	static void
478	encode_sv (enc_t *enc, SV *sv)	715	encode_sv (enc_t *enc, SV *sv)
479	{	716	{
…		…
481		718
482	if (SvPOKp (sv))	719	if (SvPOKp (sv))
483	{	720	{
484	STRLEN len;	721	STRLEN len;
485	char *str = SvPV (sv, len);	722	char *str = SvPV (sv, len);
486	encode_str (enc, SvUTF8 (sv), str, len);	723	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
487	}	724	}
488	else if (SvNOKp (sv))	725	else if (SvNOKp (sv))
489	encode_nv (enc, sv);	726	encode_nv (enc, sv);
490	else if (SvIOKp (sv))	727	else if (SvIOKp (sv))
491	{	728	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))	729	else if (SvROK (sv))
500	encode_rv (enc, SvRV (sv));	730	encode_rv (enc, SvRV (sv));
501	else if (!SvOK (sv))	731	else if (!SvOK (sv))
502	encode_ch (enc, 0xe0 | 22);	732	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	733	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504	encode_ch (enc, 0xe0 | 23);	734	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505	else	735	else
506	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	736	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	737	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508	}	738	}
509		739
510	static SV *	740	static SV *
511	encode_cbor (SV *scalar, CBOR *cbor)	741	encode_cbor (SV *scalar, CBOR *cbor)
512	{	742	{
513	enc_t enc = { };	743	enc_t enc = { 0 };
514		744
515	enc.cbor = *cbor;	745	enc.cbor = *cbor;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	746	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	747	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	748	enc.end = SvEND (enc.sv);
519		749
520	SvPOK_only (enc.sv);	750	SvPOK_only (enc.sv);
521		751
522	if (cbor->flags & F_ALLOW_STRINGREF)	752	if (cbor->flags & F_PACK_STRINGS)
523	{	753	{
524	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	754	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	755	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	756	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527	}	757	}
…		…
550	U32 depth; // recursion depth	780	U32 depth; // recursion depth
551	U32 maxdepth; // recursion depth limit	781	U32 maxdepth; // recursion depth limit
552	AV *shareable;	782	AV *shareable;
553	AV *stringref;	783	AV *stringref;
554	SV *decode_tagged;	784	SV *decode_tagged;
		785	SV *err_sv; // optional sv for error, needs to be freed
555	} dec_t;	786	} dec_t;
556		787
557	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	788	// set dec->err to ERRSV
		789	ecb_cold static void
		790	err_errsv (dec_t *dec)
		791	{
		792	if (!dec->err)
		793	{
		794	dec->err_sv = newSVsv (ERRSV);
558		795
559	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	796	// chop off the trailing \n
		797	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		798	*SvEND (dec->err_sv) = 0;
560		799
		800	dec->err = SvPVutf8_nolen (dec->err_sv);
		801	}
		802	}
		803
		804	// the following functions are used to reduce code size and help the compiler to optimise
		805	ecb_cold static void
		806	err_set (dec_t *dec, const char *reason)
		807	{
		808	if (!dec->err)
		809	dec->err = reason;
		810	}
		811
		812	ecb_cold static void
		813	err_unexpected_end (dec_t *dec)
		814	{
		815	err_set (dec, "unexpected end of CBOR data");
		816	}
		817
		818	#define ERR_DO(do) SB do; goto fail; SE
		819	#define ERR(reason) ERR_DO (err_set (dec, reason))
		820	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		821
		822	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		823
561	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	824	#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	825	#define DEC_DEC_DEPTH --dec->depth
563		826
564	static UV	827	static UV
565	decode_uint (dec_t *dec)	828	decode_uint (dec_t *dec)
566	{	829	{
567	switch (*dec->cur & 31)	830	U8 m = *dec->cur & MINOR_MASK;
568	{	831	++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		832
574	case 24:	833	if (ecb_expect_true (m < LENGTH_EXT1))
		834	return m;
		835	else if (ecb_expect_true (m == LENGTH_EXT1))
		836	{
575	WANT (2);	837	WANT (1);
576	dec->cur += 2;	838	dec->cur += 1;
577	return dec->cur[-1];	839	return dec->cur[-1];
578		840	}
579	case 25:	841	else if (ecb_expect_true (m == LENGTH_EXT2))
		842	{
580	WANT (3);	843	WANT (2);
581	dec->cur += 3;	844	dec->cur += 2;
582	return (((UV)dec->cur[-2]) << 8)	845	return (((UV)dec->cur[-2]) << 8)
583	| ((UV)dec->cur[-1]);	846	| ((UV)dec->cur[-1]);
584		847	}
585	case 26:	848	else if (ecb_expect_true (m == LENGTH_EXT4))
		849	{
586	WANT (5);	850	WANT (4);
587	dec->cur += 5;	851	dec->cur += 4;
588	return (((UV)dec->cur[-4]) << 24)	852	return (((UV)dec->cur[-4]) << 24)
589	| (((UV)dec->cur[-3]) << 16)	853	| (((UV)dec->cur[-3]) << 16)
590	| (((UV)dec->cur[-2]) << 8)	854	| (((UV)dec->cur[-2]) << 8)
591	| ((UV)dec->cur[-1]);	855	| ((UV)dec->cur[-1]);
592		856	}
593	case 27:	857	else if (ecb_expect_true (m == LENGTH_EXT8))
		858	{
594	WANT (9);	859	WANT (8);
595	dec->cur += 9;	860	dec->cur += 8;
		861
		862	return
		863	#if UVSIZE < 8
		864	0
		865	#else
596	return (((UV)dec->cur[-8]) << 56)	866	(((UV)dec->cur[-8]) << 56)
597	| (((UV)dec->cur[-7]) << 48)	867	| (((UV)dec->cur[-7]) << 48)
598	| (((UV)dec->cur[-6]) << 40)	868	| (((UV)dec->cur[-6]) << 40)
599	| (((UV)dec->cur[-5]) << 32)	869	| (((UV)dec->cur[-5]) << 32)
		870	#endif
600	| (((UV)dec->cur[-4]) << 24)	871	| (((UV)dec->cur[-4]) << 24)
601	| (((UV)dec->cur[-3]) << 16)	872	| (((UV)dec->cur[-3]) << 16)
602	| (((UV)dec->cur[-2]) << 8)	873	| (((UV)dec->cur[-2]) << 8)
603	| ((UV)dec->cur[-1]);	874	| ((UV)dec->cur[-1]);
604		875	}
605	default:	876	else
606	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	877	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607	}
608		878
609	fail:	879	fail:
610	return 0;	880	return 0;
611	}	881	}
612		882
…		…
617	{	887	{
618	AV *av = newAV ();	888	AV *av = newAV ();
619		889
620	DEC_INC_DEPTH;	890	DEC_INC_DEPTH;
621		891
622	if ((*dec->cur & 31) == 31)	892	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623	{	893	{
624	++dec->cur;	894	++dec->cur;
625		895
626	for (;;)	896	for (;;)
627	{	897	{
628	WANT (1);	898	WANT (1);
629		899
630	if (*dec->cur == (0xe0 | 31))	900	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
631	{	901	{
632	++dec->cur;	902	++dec->cur;
633	break;	903	break;
634	}	904	}
635		905
636	av_push (av, decode_sv (dec));	906	av_push (av, decode_sv (dec));
637	}	907	}
638	}	908	}
639	else	909	else
640	{	910	{
641	int i, len = decode_uint (dec);	911	UV i, len = decode_uint (dec);
642		912
		913	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);	914	av_fill (av, len - 1);
644		915
645	for (i = 0; i < len; ++i)	916	for (i = 0; i < len; ++i)
646	AvARRAY (av)[i] = decode_sv (dec);	917	AvARRAY (av)[i] = decode_sv (dec);
647	}	918	}
648		919
649	DEC_DEC_DEPTH;	920	DEC_DEC_DEPTH;
650	return newRV_noinc ((SV *)av);	921	return newRV_noinc ((SV *)av);
651		922
652	fail:	923	fail:
653	SvREFCNT_dec (av);	924	SvREFCNT_dec_NN (av);
654	DEC_DEC_DEPTH;	925	DEC_DEC_DEPTH;
655	return &PL_sv_undef;	926	return &PL_sv_undef;
656	}	927	}
657		928
658	static void	929	static void
…		…
660	{	931	{
661	// for speed reasons, we specialcase single-string	932	// for speed reasons, we specialcase single-string
662	// byte or utf-8 strings as keys, but only when !stringref	933	// byte or utf-8 strings as keys, but only when !stringref
663		934
664	if (ecb_expect_true (!dec->stringref))	935	if (ecb_expect_true (!dec->stringref))
665	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	936	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666	{	937	{
667	I32 len = decode_uint (dec);	938	STRLEN len = decode_uint (dec);
668	char *key = (char *)dec->cur;	939	char *key = (char *)dec->cur;
669		940
		941	WANT (len);
670	dec->cur += len;	942	dec->cur += len;
671		943
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);	944	hv_store (hv, key, len, decode_sv (dec), 0);
676		945
677	return;	946	return;
678	}	947	}
679	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	948	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680	{	949	{
681	I32 len = decode_uint (dec);	950	STRLEN len = decode_uint (dec);
682	char *key = (char *)dec->cur;	951	char *key = (char *)dec->cur;
683		952
		953	WANT (len);
684	dec->cur += len;	954	dec->cur += len;
685		955
686	if (ecb_expect_false (dec->stringref))	956	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	957	if (!is_utf8_string (key, len))
		958	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688		959
689	hv_store (hv, key, -len, decode_sv (dec), 0);	960	hv_store (hv, key, -len, decode_sv (dec), 0);
690		961
691	return;	962	return;
692	}	963	}
693		964
694	SV *k = decode_sv (dec);	965	SV *k = decode_sv (dec);
695	SV *v = decode_sv (dec);	966	SV *v = decode_sv (dec);
696		967
		968	// we leak memory if uncaught exceptions are thrown by random magical
		969	// methods, and this is hopefully the only place where it can happen,
		970	// so if there is a chance of an exception, take the very slow path.
		971	// since catching exceptions is "undocumented/internal/forbidden" by
		972	// the new p5p powers, we need to call out to a perl function :/
		973	if (ecb_expect_false (SvAMAGIC (k)))
		974	{
		975	dSP;
		976
		977	ENTER; SAVETMPS;
		978	PUSHMARK (SP);
		979	EXTEND (SP, 3);
		980	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		981	PUSHs (sv_2mortal (k));
		982	PUSHs (sv_2mortal (v));
		983
		984	PUTBACK;
		985	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		986	SPAGAIN;
		987
		988	FREETMPS; LEAVE;
		989
		990	if (SvTRUE (ERRSV))
		991	ERR_ERRSV;
		992
		993	return;
		994	}
		995
697	hv_store_ent (hv, k, v, 0);	996	hv_store_ent (hv, k, v, 0);
698	SvREFCNT_dec (k);	997	SvREFCNT_dec_NN (k);
		998
		999	fail:
		1000	;
699	}	1001	}
700		1002
701	static SV *	1003	static SV *
702	decode_hv (dec_t *dec)	1004	decode_hv (dec_t *dec)
703	{	1005	{
704	HV *hv = newHV ();	1006	HV *hv = newHV ();
705		1007
706	DEC_INC_DEPTH;	1008	DEC_INC_DEPTH;
707		1009
708	if ((*dec->cur & 31) == 31)	1010	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709	{	1011	{
710	++dec->cur;	1012	++dec->cur;
711		1013
712	for (;;)	1014	for (;;)
713	{	1015	{
714	WANT (1);	1016	WANT (1);
715		1017
716	if (*dec->cur == (0xe0 | 31))	1018	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
717	{	1019	{
718	++dec->cur;	1020	++dec->cur;
719	break;	1021	break;
720	}	1022	}
721		1023
722	decode_he (dec, hv);	1024	decode_he (dec, hv);
723	}	1025	}
724	}	1026	}
725	else	1027	else
726	{	1028	{
727	int pairs = decode_uint (dec);	1029	UV pairs = decode_uint (dec);
		1030
		1031	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
728		1032
729	while (pairs--)	1033	while (pairs--)
730	decode_he (dec, hv);	1034	decode_he (dec, hv);
731	}	1035	}
732		1036
733	DEC_DEC_DEPTH;	1037	DEC_DEC_DEPTH;
734	return newRV_noinc ((SV *)hv);	1038	return newRV_noinc ((SV *)hv);
735		1039
736	fail:	1040	fail:
737	SvREFCNT_dec (hv);	1041	SvREFCNT_dec_NN (hv);
738	DEC_DEC_DEPTH;	1042	DEC_DEC_DEPTH;
739	return &PL_sv_undef;	1043	return &PL_sv_undef;
740	}	1044	}
741		1045
742	static SV *	1046	static SV *
743	decode_str (dec_t *dec, int utf8)	1047	decode_str (dec_t *dec, int utf8)
744	{	1048	{
745	SV *sv = 0;	1049	SV *sv = 0;
746		1050
747	if ((*dec->cur & 31) == 31)	1051	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
748	{	1052	{
		1053	// indefinite length strings
749	++dec->cur;	1054	++dec->cur;
750		1055
		1056	U8 major = *dec->cur & MAJOR_MISC;
		1057
751	sv = newSVpvn ("", 0);	1058	sv = newSVpvn ("", 0);
752		1059
753	// not very fast, and certainly not robust against illegal input
754	for (;;)	1060	for (;;)
755	{	1061	{
756	WANT (1);	1062	WANT (1);
757		1063
758	if (*dec->cur == (0xe0 | 31))	1064	if ((*dec->cur - major) > LENGTH_EXT8)
		1065	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759	{	1066	{
760	++dec->cur;	1067	++dec->cur;
761	break;	1068	break;
762	}	1069	}
		1070	else
		1071	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763		1072
764	sv_catsv (sv, decode_sv (dec));	1073	STRLEN len = decode_uint (dec);
		1074
		1075	WANT (len);
		1076	sv_catpvn (sv, dec->cur, len);
		1077	dec->cur += len;
765	}	1078	}
766	}	1079	}
767	else	1080	else
768	{	1081	{
769	STRLEN len = decode_uint (dec);	1082	STRLEN len = decode_uint (dec);
…		…
776	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1089	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1090	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778	}	1091	}
779		1092
780	if (utf8)	1093	if (utf8)
		1094	{
		1095	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1096	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1097	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1098
781	SvUTF8_on (sv);	1099	SvUTF8_on (sv);
		1100	}
782		1101
783	return sv;	1102	return sv;
784		1103
785	fail:	1104	fail:
786	SvREFCNT_dec (sv);	1105	SvREFCNT_dec (sv);
…		…
805	sv = newRV_noinc (decode_sv (dec));	1124	sv = newRV_noinc (decode_sv (dec));
806	break;	1125	break;
807		1126
808	case CBOR_TAG_STRINGREF_NAMESPACE:	1127	case CBOR_TAG_STRINGREF_NAMESPACE:
809	{	1128	{
		1129	// do not use SAVETMPS/FREETMPS, as these will
		1130	// erase mortalised caches, e.g. "shareable"
810	ENTER; SAVETMPS;	1131	ENTER;
811		1132
812	SAVESPTR (dec->stringref);	1133	SAVESPTR (dec->stringref);
813	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1134	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
814		1135
815	sv = decode_sv (dec);	1136	sv = decode_sv (dec);
816		1137
817	FREETMPS; LEAVE;	1138	LEAVE;
818	}	1139	}
819	break;	1140	break;
820		1141
821	case CBOR_TAG_STRINGREF:	1142	case CBOR_TAG_STRINGREF:
822	{	1143	{
823	if ((*dec->cur >> 5) != 0)	1144	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1145	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825		1146
826	UV idx = decode_uint (dec);	1147	UV idx = decode_uint (dec);
827		1148
828	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1149	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
829	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1150	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
830		1151
831	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1152	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
832	}	1153	}
833	break;	1154	break;
…		…
835	case CBOR_TAG_VALUE_SHAREABLE:	1156	case CBOR_TAG_VALUE_SHAREABLE:
836	{	1157	{
837	if (ecb_expect_false (!dec->shareable))	1158	if (ecb_expect_false (!dec->shareable))
838	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1159	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839		1160
		1161	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1162	{
840	sv = newSV (0);	1163	sv = newSV (0);
841	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1164	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842		1165
843	SV *osv = decode_sv (dec);	1166	SV *osv = decode_sv (dec);
844	sv_setsv (sv, osv);	1167	sv_setsv (sv, osv);
845	SvREFCNT_dec_NN (osv);	1168	SvREFCNT_dec_NN (osv);
		1169	}
		1170	else
		1171	{
		1172	av_push (dec->shareable, &PL_sv_undef);
		1173	int idx = AvFILLp (dec->shareable);
		1174	sv = decode_sv (dec);
		1175	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1176	}
846	}	1177	}
847	break;	1178	break;
848		1179
849	case CBOR_TAG_VALUE_SHAREDREF:	1180	case CBOR_TAG_VALUE_SHAREDREF:
850	{	1181	{
851	if ((*dec->cur >> 5) != 0)	1182	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1183	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853		1184
854	UV idx = decode_uint (dec);	1185	UV idx = decode_uint (dec);
855		1186
856	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1187	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
857	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1188	ERR ("corrupted CBOR data (sharedref index out of bounds)");
858		1189
859	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1190	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1191
		1192	if (sv == &PL_sv_undef)
		1193	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860	}	1194	}
861	break;	1195	break;
862		1196
863	case CBOR_TAG_PERL_OBJECT:	1197	case CBOR_TAG_PERL_OBJECT:
864	{	1198	{
		1199	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1200	goto filter;
		1201
865	sv = decode_sv (dec);	1202	sv = decode_sv (dec);
866		1203
867	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1204	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
868	ERR ("corrupted CBOR data (non-array perl object)");	1205	ERR ("corrupted CBOR data (non-array perl object)");
869		1206
…		…
879	if (!method)	1216	if (!method)
880	ERR ("cannot decode perl-object (package does not have a THAW method)");	1217	ERR ("cannot decode perl-object (package does not have a THAW method)");
881		1218
882	dSP;	1219	dSP;
883		1220
884	ENTER; SAVETMPS; PUSHMARK (SP);	1221	ENTER; SAVETMPS;
		1222	PUSHMARK (SP);
885	EXTEND (SP, len + 1);	1223	EXTEND (SP, len + 1);
886	// we re-bless the reference to get overload and other niceties right	1224	// we re-bless the reference to get overload and other niceties right
887	PUSHs (*av_fetch (av, 0, 1));	1225	PUSHs (*av_fetch (av, 0, 1));
888	PUSHs (sv_cbor);	1226	PUSHs (sv_cbor);
889		1227
…		…
897	SPAGAIN;	1235	SPAGAIN;
898		1236
899	if (SvTRUE (ERRSV))	1237	if (SvTRUE (ERRSV))
900	{	1238	{
901	FREETMPS; LEAVE;	1239	FREETMPS; LEAVE;
902	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1240	ERR_ERRSV;
903	}	1241	}
904		1242
905	SvREFCNT_dec (sv);	1243	SvREFCNT_dec_NN (sv);
906	sv = SvREFCNT_inc (POPs);	1244	sv = SvREFCNT_inc (POPs);
907		1245
908	PUTBACK;	1246	PUTBACK;
909		1247
910	FREETMPS; LEAVE;	1248	FREETMPS; LEAVE;
911	}	1249	}
912	break;	1250	break;
913		1251
914	default:	1252	default:
		1253	filter:
915	{	1254	{
		1255	SV *tag_sv = newSVuv (tag);
		1256
916	sv = decode_sv (dec);	1257	sv = decode_sv (dec);
917		1258
918	dSP;	1259	dSP;
919	ENTER; SAVETMPS; PUSHMARK (SP);	1260	ENTER; SAVETMPS;
		1261	PUSHMARK (SP);
920	EXTEND (SP, 2);	1262	EXTEND (SP, 2);
921	PUSHs (newSVuv (tag));	1263	PUSHs (tag_sv);
922	PUSHs (sv);	1264	PUSHs (sv);
923		1265
924	PUTBACK;	1266	PUTBACK;
925	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1267	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
926	SPAGAIN;	1268	SPAGAIN;
927		1269
928	if (SvTRUE (ERRSV))	1270	if (SvTRUE (ERRSV))
929	{	1271	{
		1272	SvREFCNT_dec_NN (tag_sv);
930	FREETMPS; LEAVE;	1273	FREETMPS; LEAVE;
931	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1274	ERR_ERRSV;
932	}	1275	}
933		1276
934	if (count)	1277	if (count)
935	{	1278	{
		1279	SvREFCNT_dec_NN (tag_sv);
936	SvREFCNT_dec (sv);	1280	SvREFCNT_dec_NN (sv);
937	sv = SvREFCNT_inc (POPs);	1281	sv = SvREFCNT_inc_NN (TOPs);
		1282	SP -= count;
938	}	1283	}
939	else	1284	else
940	{	1285	{
941	AV *av = newAV ();	1286	AV *av = newAV ();
942	av_push (av, newSVuv (tag));	1287	av_push (av, tag_sv);
943	av_push (av, sv);	1288	av_push (av, sv);
944		1289
945	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1290	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
946	? cbor_tagged_stash	1291	? cbor_tagged_stash
947	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1292	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
965	static SV *	1310	static SV *
966	decode_sv (dec_t *dec)	1311	decode_sv (dec_t *dec)
967	{	1312	{
968	WANT (1);	1313	WANT (1);
969		1314
970	switch (*dec->cur >> 5)	1315	switch (*dec->cur >> MAJOR_SHIFT)
971	{	1316	{
972	case 0: // unsigned int	1317	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973	return newSVuv (decode_uint (dec));	1318	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974	case 1: // negative int	1319	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975	return newSViv (-1 - (IV)decode_uint (dec));	1320	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976	case 2: // octet string	1321	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977	return decode_str (dec, 0);	1322	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978	case 3: // utf-8 string	1323	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979	return decode_str (dec, 1);	1324
980	case 4: // array	1325	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)	1326	switch (*dec->cur++ & MINOR_MASK)
988	{	1327	{
989	case 20:	1328	case SIMPLE_FALSE:
990	#if CBOR_SLOW	1329	#if CBOR_SLOW
991	types_false = get_bool ("Types::Serialiser::false");	1330	types_false = get_bool ("Types::Serialiser::false");
992	#endif	1331	#endif
993	return newSVsv (types_false);	1332	return newSVsv (types_false);
994	case 21:	1333	case SIMPLE_TRUE:
995	#if CBOR_SLOW	1334	#if CBOR_SLOW
996	types_true = get_bool ("Types::Serialiser::true");	1335	types_true = get_bool ("Types::Serialiser::true");
997	#endif	1336	#endif
998	return newSVsv (types_true);	1337	return newSVsv (types_true);
999	case 22:	1338	case SIMPLE_NULL:
1000	return newSVsv (&PL_sv_undef);	1339	return newSVsv (&PL_sv_undef);
1001	case 23:	1340	case SIMPLE_UNDEF:
1002	#if CBOR_SLOW	1341	#if CBOR_SLOW
1003	types_error = get_bool ("Types::Serialiser::error");	1342	types_error = get_bool ("Types::Serialiser::error");
1004	#endif	1343	#endif
1005	return newSVsv (types_error);	1344	return newSVsv (types_error);
1006		1345
1007	case 25:	1346	case MISC_FLOAT16:
1008	{	1347	{
1009	WANT (2);	1348	WANT (2);
1010		1349
1011	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1350	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012	dec->cur += 2;	1351	dec->cur += 2;
1013		1352
1014	return newSVnv (ecb_binary16_to_float (fp));	1353	return newSVnv (ecb_binary16_to_float (fp));
1015	}	1354	}
1016		1355
1017	case 26:	1356	case MISC_FLOAT32:
1018	{	1357	{
1019	uint32_t fp;	1358	uint32_t fp;
1020	WANT (4);	1359	WANT (4);
1021	memcpy (&fp, dec->cur, 4);	1360	memcpy (&fp, dec->cur, 4);
1022	dec->cur += 4;	1361	dec->cur += 4;
…		…
1025	fp = ecb_bswap32 (fp);	1364	fp = ecb_bswap32 (fp);
1026		1365
1027	return newSVnv (ecb_binary32_to_float (fp));	1366	return newSVnv (ecb_binary32_to_float (fp));
1028	}	1367	}
1029		1368
1030	case 27:	1369	case MISC_FLOAT64:
1031	{	1370	{
1032	uint64_t fp;	1371	uint64_t fp;
1033	WANT (8);	1372	WANT (8);
1034	memcpy (&fp, dec->cur, 8);	1373	memcpy (&fp, dec->cur, 8);
1035	dec->cur += 8;	1374	dec->cur += 8;
…		…
1038	fp = ecb_bswap64 (fp);	1377	fp = ecb_bswap64 (fp);
1039		1378
1040	return newSVnv (ecb_binary64_to_double (fp));	1379	return newSVnv (ecb_binary64_to_double (fp));
1041	}	1380	}
1042		1381
1043	// 0..19 unassigned	1382	// 0..19 unassigned simple
1044	// 24 reserved + unassigned (reserved values are not encodable)	1383	// 24 reserved + unassigned simple (reserved values are not encodable)
		1384	// 28-30 unassigned misc
		1385	// 31 break code
1045	default:	1386	default:
1046	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1387	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047	}	1388	}
1048		1389
1049	break;	1390	break;
1050	}	1391	}
1051		1392
…		…
1054	}	1395	}
1055		1396
1056	static SV *	1397	static SV *
1057	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1398	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058	{	1399	{
1059	dec_t dec = { };	1400	dec_t dec = { 0 };
1060	SV *sv;	1401	SV *sv;
1061	STRLEN len;	1402	STRLEN len;
1062	char *data = SvPVbyte (string, len);	1403	char *data = SvPVbyte (string, len);
1063		1404
1064	if (len > cbor->max_size && cbor->max_size)	1405	if (len > cbor->max_size && cbor->max_size)
…		…
1078	if (dec.cur != dec.end && !dec.err)	1419	if (dec.cur != dec.end && !dec.err)
1079	dec.err = "garbage after CBOR object";	1420	dec.err = "garbage after CBOR object";
1080		1421
1081	if (dec.err)	1422	if (dec.err)
1082	{	1423	{
		1424	if (dec.shareable)
		1425	{
		1426	// need to break cyclic links, which would all be in shareable
		1427	int i;
		1428	SV **svp;
		1429
		1430	for (i = av_len (dec.shareable) + 1; i--; )
		1431	if ((svp = av_fetch (dec.shareable, i, 0)))
		1432	sv_setsv (*svp, &PL_sv_undef);
		1433	}
		1434
1083	SvREFCNT_dec (sv);	1435	SvREFCNT_dec_NN (sv);
		1436
		1437	if (dec.err_sv)
		1438	sv_2mortal (dec.err_sv);
		1439
1084	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1440	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1085	}	1441	}
1086		1442
1087	sv = sv_2mortal (sv);	1443	sv = sv_2mortal (sv);
1088		1444
1089	return sv;	1445	return sv;
1090	}	1446	}
1091		1447
		1448	/////////////////////////////////////////////////////////////////////////////
		1449	// incremental parser
		1450
		1451	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1452
		1453	// returns 0 for notyet, 1 for success or error
		1454	static int
		1455	incr_parse (CBOR *self, SV *cborstr)
		1456	{
		1457	STRLEN cur;
		1458	SvPV (cborstr, cur);
		1459
		1460	while (ecb_expect_true (self->incr_need <= cur))
		1461	{
		1462	// table of integer count bytes
		1463	static I8 incr_len[MINOR_MASK + 1] = {
		1464	0, 0, 0, 0, 0, 0, 0, 0,
		1465	0, 0, 0, 0, 0, 0, 0, 0,
		1466	0, 0, 0, 0, 0, 0, 0, 0,
		1467	1, 2, 4, 8,-1,-1,-1,-2
		1468	};
		1469
		1470	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1471	U8 m = *p & MINOR_MASK;
		1472	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1473	I8 ilen = incr_len[m];
		1474
		1475	self->incr_need = self->incr_pos + 1;
		1476
		1477	if (ecb_expect_false (ilen < 0))
		1478	{
		1479	if (m != MINOR_INDEF)
		1480	return 1; // error
		1481
		1482	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1483	{
		1484	if (count >= 0)
		1485	return 1; // error
		1486
		1487	count = 1;
		1488	}
		1489	else
		1490	{
		1491	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1492	count = -1;
		1493	}
		1494	}
		1495	else
		1496	{
		1497	self->incr_need += ilen;
		1498	if (ecb_expect_false (self->incr_need > cur))
		1499	return 0;
		1500
		1501	int major = *p >> MAJOR_SHIFT;
		1502
		1503	switch (major)
		1504	{
		1505	case MAJOR_TAG >> MAJOR_SHIFT:
		1506	++count; // tags merely prefix another value
		1507	break;
		1508
		1509	case MAJOR_BYTES >> MAJOR_SHIFT:
		1510	case MAJOR_TEXT >> MAJOR_SHIFT:
		1511	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1512	case MAJOR_MAP >> MAJOR_SHIFT:
		1513	{
		1514	UV len;
		1515
		1516	if (ecb_expect_false (ilen))
		1517	{
		1518	len = 0;
		1519
		1520	do {
		1521	len = (len << 8) | *++p;
		1522	} while (--ilen);
		1523	}
		1524	else
		1525	len = m;
		1526
		1527	switch (major)
		1528	{
		1529	case MAJOR_BYTES >> MAJOR_SHIFT:
		1530	case MAJOR_TEXT >> MAJOR_SHIFT:
		1531	self->incr_need += len;
		1532	if (ecb_expect_false (self->incr_need > cur))
		1533	return 0;
		1534
		1535	break;
		1536
		1537	case MAJOR_MAP >> MAJOR_SHIFT:
		1538	len <<= 1;
		1539	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1540	if (len)
		1541	{
		1542	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1543	count = len + 1;
		1544	}
		1545	break;
		1546	}
		1547	}
		1548	}
		1549	}
		1550
		1551	self->incr_pos = self->incr_need;
		1552
		1553	if (count > 0)
		1554	{
		1555	while (!--count)
		1556	{
		1557	if (!AvFILLp (self->incr_count))
		1558	return 1; // done
		1559
		1560	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1561	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1562	}
		1563
		1564	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1565	}
		1566	}
		1567
		1568	return 0;
		1569	}
		1570
		1571
1092	/////////////////////////////////////////////////////////////////////////////	1572	/////////////////////////////////////////////////////////////////////////////
1093	// XS interface functions	1573	// XS interface functions
1094		1574
1095	MODULE = CBOR::XS PACKAGE = CBOR::XS	1575	MODULE = CBOR::XS PACKAGE = CBOR::XS
1096		1576
…		…
1108		1588
1109	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1589	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1110		1590
1111	sv_cbor = newSVpv ("CBOR", 0);	1591	sv_cbor = newSVpv ("CBOR", 0);
1112	SvREADONLY_on (sv_cbor);	1592	SvREADONLY_on (sv_cbor);
		1593
		1594	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1113	}	1595	}
1114		1596
1115	PROTOTYPES: DISABLE	1597	PROTOTYPES: DISABLE
1116		1598
1117	void CLONE (...)	1599	void CLONE (...)
…		…
1136	void shrink (CBOR *self, int enable = 1)	1618	void shrink (CBOR *self, int enable = 1)
1137	ALIAS:	1619	ALIAS:
1138	shrink = F_SHRINK	1620	shrink = F_SHRINK
1139	allow_unknown = F_ALLOW_UNKNOWN	1621	allow_unknown = F_ALLOW_UNKNOWN
1140	allow_sharing = F_ALLOW_SHARING	1622	allow_sharing = F_ALLOW_SHARING
1141	allow_stringref = F_ALLOW_STRINGREF	1623	allow_cycles = F_ALLOW_CYCLES
		1624	forbid_objects = F_FORBID_OBJECTS
		1625	pack_strings = F_PACK_STRINGS
		1626	text_keys = F_TEXT_KEYS
		1627	text_strings = F_TEXT_STRINGS
		1628	validate_utf8 = F_VALIDATE_UTF8
1142	PPCODE:	1629	PPCODE:
1143	{	1630	{
1144	if (enable)	1631	if (enable)
1145	self->flags |= ix;	1632	self->flags |= ix;
1146	else	1633	else
…		…
1152	void get_shrink (CBOR *self)	1639	void get_shrink (CBOR *self)
1153	ALIAS:	1640	ALIAS:
1154	get_shrink = F_SHRINK	1641	get_shrink = F_SHRINK
1155	get_allow_unknown = F_ALLOW_UNKNOWN	1642	get_allow_unknown = F_ALLOW_UNKNOWN
1156	get_allow_sharing = F_ALLOW_SHARING	1643	get_allow_sharing = F_ALLOW_SHARING
1157	get_allow_stringref = F_ALLOW_STRINGREF	1644	get_allow_cycles = F_ALLOW_CYCLES
		1645	get_forbid_objects = F_FORBID_OBJECTS
		1646	get_pack_strings = F_PACK_STRINGS
		1647	get_text_keys = F_TEXT_KEYS
		1648	get_text_strings = F_TEXT_STRINGS
		1649	get_validate_utf8 = F_VALIDATE_UTF8
1158	PPCODE:	1650	PPCODE:
1159	XPUSHs (boolSV (self->flags & ix));	1651	XPUSHs (boolSV (self->flags & ix));
1160		1652
1161	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1653	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162	PPCODE:	1654	PPCODE:
…		…
1211	EXTEND (SP, 2);	1703	EXTEND (SP, 2);
1212	PUSHs (sv);	1704	PUSHs (sv);
1213	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1705	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214	}	1706	}
1215		1707
		1708	void incr_parse (CBOR *self, SV *cborstr)
		1709	ALIAS:
		1710	incr_parse_multiple = 1
		1711	PPCODE:
		1712	{
		1713	if (SvUTF8 (cborstr))
		1714	sv_utf8_downgrade (cborstr, 0);
		1715
		1716	if (!self->incr_count)
		1717	{
		1718	self->incr_count = newAV ();
		1719	self->incr_pos = 0;
		1720	self->incr_need = 1;
		1721
		1722	av_push (self->incr_count, newSViv (1));
		1723	}
		1724
		1725	do
		1726	{
		1727	if (!incr_parse (self, cborstr))
		1728	{
		1729	if (self->incr_need > self->max_size && self->max_size)
		1730	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1731	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1732
		1733	break;
		1734	}
		1735
		1736	SV *sv;
		1737	char *offset;
		1738
		1739	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1740	XPUSHs (sv);
		1741
		1742	sv_chop (cborstr, offset);
		1743
		1744	av_clear (self->incr_count);
		1745	av_push (self->incr_count, newSViv (1));
		1746
		1747	self->incr_pos = 0;
		1748	self->incr_need = self->incr_pos + 1;
		1749	}
		1750	while (ix);
		1751	}
		1752
		1753	void incr_reset (CBOR *self)
		1754	CODE:
		1755	{
		1756	SvREFCNT_dec (self->incr_count);
		1757	self->incr_count = 0;
		1758	}
		1759
1216	void DESTROY (CBOR *self)	1760	void DESTROY (CBOR *self)
1217	PPCODE:	1761	PPCODE:
1218	cbor_free (self);	1762	cbor_free (self);
1219		1763
1220	PROTOTYPES: ENABLE	1764	PROTOTYPES: ENABLE
1221		1765
1222	void encode_cbor (SV *scalar)	1766	void encode_cbor (SV *scalar)
		1767	ALIAS:
		1768	encode_cbor = 0
		1769	encode_cbor_sharing = F_ALLOW_SHARING
1223	PPCODE:	1770	PPCODE:
1224	{	1771	{
1225	CBOR cbor;	1772	CBOR cbor;
1226	cbor_init (&cbor);	1773	cbor_init (&cbor);
		1774	cbor.flags |= ix;
1227	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1775	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228	XPUSHs (scalar);	1776	XPUSHs (scalar);
1229	}	1777	}
1230		1778
1231	void decode_cbor (SV *cborstr)	1779	void decode_cbor (SV *cborstr)
…		…
1235	cbor_init (&cbor);	1783	cbor_init (&cbor);
1236	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1784	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1237	XPUSHs (cborstr);	1785	XPUSHs (cborstr);
1238	}	1786	}
1239		1787
		1788	#ifdef __AFL_COMPILER
		1789
		1790	void
		1791	afl_init ()
		1792	CODE:
		1793	__AFL_INIT ();
		1794
		1795	int
		1796	afl_loop (unsigned int count = 10000)
		1797	CODE:
		1798	RETVAL = __AFL_LOOP (count);
		1799	OUTPUT:
		1800	RETVAL
		1801
		1802	#endif
		1803

Diff Legend

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