ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 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
57	};	102	};
58		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_FLOAT16 = 3,
		111	AS_FLOAT32 = 4,
		112	AS_FLOAT64 = 5,
		113	// possibly future enhancements: float, integer
		114	};
		115
59	#define F_SHRINK 0x00000001UL	116	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	117	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	118	#define F_ALLOW_SHARING 0x00000004UL
		119	#define F_ALLOW_CYCLES 0x00000008UL
		120	#define F_FORBID_OBJECTS 0x00000010UL
62	#define F_DEDUP_STRINGS 0x00000008UL //TODO	121	#define F_PACK_STRINGS 0x00000020UL
63	#define F_DEDUP_KEYS 0x00000010UL //TODO	122	#define F_TEXT_KEYS 0x00000040UL
		123	#define F_TEXT_STRINGS 0x00000080UL
		124	#define F_VALIDATE_UTF8 0x00000100UL
64		125
65	#define INIT_SIZE 32 // initial scalar size to be allocated	126	#define INIT_SIZE 32 // initial scalar size to be allocated
66		127
67	#define SB do {	128	#define SB do {
68	#define SE } while (0)	129	#define SE } while (0)
…		…
80	# define CBOR_SLOW 0	141	# define CBOR_SLOW 0
81	# define CBOR_STASH cbor_stash	142	# define CBOR_STASH cbor_stash
82	#endif	143	#endif
83		144
84	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::	145	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85	static SV *types_true, *types_false, *types_error, *sv_cbor;	146	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86		147
87	typedef struct {	148	typedef struct {
88	U32 flags;	149	U32 flags;
89	U32 max_depth;	150	U32 max_depth;
90	STRLEN max_size;	151	STRLEN max_size;
		152	SV *filter;
		153
		154	// for the incremental parser
		155	STRLEN incr_pos; // the current offset into the text
		156	STRLEN incr_need; // minimum bytes needed to decode
		157	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	158	} CBOR;
92		159
93	ecb_inline void	160	ecb_inline void
94	cbor_init (CBOR *cbor)	161	cbor_init (CBOR *cbor)
95	{	162	{
96	Zero (cbor, 1, CBOR);	163	Zero (cbor, 1, CBOR);
97	cbor->max_depth = 512;	164	cbor->max_depth = 512;
		165	}
		166
		167	ecb_inline void
		168	cbor_free (CBOR *cbor)
		169	{
		170	SvREFCNT_dec (cbor->filter);
		171	SvREFCNT_dec (cbor->incr_count);
98	}	172	}
99		173
100	/////////////////////////////////////////////////////////////////////////////	174	/////////////////////////////////////////////////////////////////////////////
101	// utility functions	175	// utility functions
102		176
…		…
122	SvPV_shrink_to_cur (sv);	196	SvPV_shrink_to_cur (sv);
123	#elif defined (SvPV_renew)	197	#elif defined (SvPV_renew)
124	SvPV_renew (sv, SvCUR (sv) + 1);	198	SvPV_renew (sv, SvCUR (sv) + 1);
125	#endif	199	#endif
126	}	200	}
		201	}
		202
		203	// minimum length of a string to be registered for stringref
		204	ecb_inline int
		205	minimum_string_length (UV idx)
		206	{
		207	return idx <= 23 ? 3
		208	: idx <= 0xffU ? 4
		209	: idx <= 0xffffU ? 5
		210	: idx <= 0xffffffffU ? 7
		211	: 11;
127	}	212	}
128		213
129	/////////////////////////////////////////////////////////////////////////////	214	/////////////////////////////////////////////////////////////////////////////
130	// encoder	215	// encoder
131		216
…		…
144	} enc_t;	229	} enc_t;
145		230
146	ecb_inline void	231	ecb_inline void
147	need (enc_t *enc, STRLEN len)	232	need (enc_t *enc, STRLEN len)
148	{	233	{
149	if (ecb_expect_false (enc->cur + len >= enc->end))	234	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150	{	235	{
151	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	236	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	237	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153	enc->cur = SvPVX (enc->sv) + cur;	238	enc->cur = SvPVX (enc->sv) + cur;
154	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	239	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
160	{	245	{
161	need (enc, 1);	246	need (enc, 1);
162	*enc->cur++ = ch;	247	*enc->cur++ = ch;
163	}	248	}
164		249
		250	// used for tags, intregers, element counts and so on
165	static void	251	static void
166	encode_uint (enc_t *enc, int major, UV len)	252	encode_uint (enc_t *enc, int major, UV len)
167	{	253	{
168	need (enc, 9);	254	need (enc, 9);
169		255
170	if (len < 24)	256	if (ecb_expect_true (len < LENGTH_EXT1))
171	*enc->cur++ = major | len;	257	*enc->cur++ = major | len;
172	else if (len <= 0xff)	258	else if (ecb_expect_true (len <= 0xffU))
173	{	259	{
174	*enc->cur++ = major | 24;	260	*enc->cur++ = major | LENGTH_EXT1;
175	*enc->cur++ = len;	261	*enc->cur++ = len;
176	}	262	}
177	else if (len <= 0xffff)	263	else if (len <= 0xffffU)
178	{	264	{
179	*enc->cur++ = major | 25;	265	*enc->cur++ = major | LENGTH_EXT2;
180	*enc->cur++ = len >> 8;	266	*enc->cur++ = len >> 8;
181	*enc->cur++ = len;	267	*enc->cur++ = len;
182	}	268	}
183	else if (len <= 0xffffffff)	269	else if (len <= 0xffffffffU)
184	{	270	{
185	*enc->cur++ = major | 26;	271	*enc->cur++ = major | LENGTH_EXT4;
186	*enc->cur++ = len >> 24;	272	*enc->cur++ = len >> 24;
187	*enc->cur++ = len >> 16;	273	*enc->cur++ = len >> 16;
188	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
189	*enc->cur++ = len;	275	*enc->cur++ = len;
190	}	276	}
191	else	277	else
192	{	278	{
193	*enc->cur++ = major | 27;	279	*enc->cur++ = major | LENGTH_EXT8;
194	*enc->cur++ = len >> 56;	280	*enc->cur++ = len >> 56;
195	*enc->cur++ = len >> 48;	281	*enc->cur++ = len >> 48;
196	*enc->cur++ = len >> 40;	282	*enc->cur++ = len >> 40;
197	*enc->cur++ = len >> 32;	283	*enc->cur++ = len >> 32;
198	*enc->cur++ = len >> 24;	284	*enc->cur++ = len >> 24;
…		…
200	*enc->cur++ = len >> 8;	286	*enc->cur++ = len >> 8;
201	*enc->cur++ = len;	287	*enc->cur++ = len;
202	}	288	}
203	}	289	}
204		290
205	static void	291	// encodes a perl value into a CBOR integer
		292	ecb_inline
		293	encode_int (enc_t *enc, SV *sv)
		294	{
		295	if (SvIsUV (sv))
		296	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		297	else if (SvIVX (sv) >= 0)
		298	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		299	else
		300	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		301	}
		302
		303	ecb_inline void
		304	encode_tag (enc_t *enc, UV tag)
		305	{
		306	encode_uint (enc, MAJOR_TAG, tag);
		307	}
		308
		309	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		310	ecb_noinline static void
206	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	311	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207	{	312	{
208	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	313	STRLEN ulen = len;
		314	U8 *p, *pend = (U8 *)str + len;
		315
		316	for (p = (U8 *)str; p < pend; ++p)
		317	ulen += *p >> 7; // count set high bits
		318
		319	encode_uint (enc, MAJOR_TEXT, ulen);
		320
		321	need (enc, ulen);
		322	for (p = (U8 *)str; p < pend; ++p)
		323	if (*p < 0x80)
		324	*enc->cur++ = *p;
		325	else
		326	{
		327	*enc->cur++ = 0xc0 + (*p >> 6);
		328	*enc->cur++ = 0x80 + (*p & 63);
		329	}
		330	}
		331
		332	ecb_inline void
		333	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		334	{
		335	if (ecb_expect_false (upgrade_utf8))
		336	if (!utf8)
		337	{
		338	encode_str_utf8 (enc, utf8, str, len);
		339	return;
		340	}
		341
		342	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
209	need (enc, len);	343	need (enc, len);
210	memcpy (enc->cur, str, len);	344	memcpy (enc->cur, str, len);
211	enc->cur += len;	345	enc->cur += len;
212	}	346	}
213		347
214	ecb_inline void	348	ecb_inline void
		349	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		350	{
		351	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		352	{
		353	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		354
		355	if (SvOK (*svp))
		356	{
		357	// already registered, use stringref
		358	encode_tag (enc, CBOR_TAG_STRINGREF);
		359	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		360	return;
		361	}
		362	else if (len >= minimum_string_length (enc->stringref_idx))
		363	{
		364	// register only
		365	sv_setuv (*svp, enc->stringref_idx);
		366	++enc->stringref_idx;
		367	}
		368	}
		369
		370	encode_str (enc, upgrade_utf8, utf8, str, len);
		371	}
		372
		373	ecb_inline void
215	encode_tag (enc_t *enc, UV tag)	374	encode_float16 (enc_t *enc, NV nv)
216	{	375	{
217	encode_uint (enc, 0xc0, tag);	376	need (enc, 1+2);
		377
		378	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		379
		380	uint16_t fp = ecb_float_to_binary16 (nv);
		381
		382	if (!ecb_big_endian ())
		383	fp = ecb_bswap16 (fp);
		384
		385	memcpy (enc->cur, &fp, 2);
		386	enc->cur += 2;
		387	}
		388
		389	ecb_inline void
		390	encode_float32 (enc_t *enc, NV nv)
		391	{
		392	need (enc, 1+4);
		393
		394	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		395
		396	uint32_t fp = ecb_float_to_binary32 (nv);
		397
		398	if (!ecb_big_endian ())
		399	fp = ecb_bswap32 (fp);
		400
		401	memcpy (enc->cur, &fp, 4);
		402	enc->cur += 4;
		403	}
		404
		405	ecb_inline void
		406	encode_float64 (enc_t *enc, NV nv)
		407	{
		408	need (enc, 1+8);
		409
		410	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		411
		412	uint64_t fp = ecb_double_to_binary64 (nv);
		413
		414	if (!ecb_big_endian ())
		415	fp = ecb_bswap64 (fp);
		416
		417	memcpy (enc->cur, &fp, 8);
		418	enc->cur += 8;
		419	}
		420
		421	ecb_inline void
		422	encode_forced (enc_t *enc, UV type, SV *sv)
		423	{
		424	switch (type)
		425	{
		426	case AS_CBOR:
		427	{
		428	STRLEN len;
		429	char *str = SvPVbyte (sv, len);
		430
		431	need (enc, len);
		432	memcpy (enc->cur, str, len);
		433	enc->cur += len;
		434	}
		435	break;
		436
		437	case AS_BYTES:
		438	{
		439	STRLEN len;
		440	char *str = SvPVbyte (sv, len);
		441	encode_strref (enc, 0, 0, str, len);
		442	}
		443	break;
		444
		445	case AS_TEXT:
		446	{
		447	STRLEN len;
		448	char *str = SvPVutf8 (sv, len);
		449	encode_strref (enc, 1, 1, str, len);
		450	}
		451	break;
		452
		453	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		454	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		455	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		456
		457	default:
		458	croak ("encountered malformed CBOR::XS::Tagged object");
		459	}
218	}	460	}
219		461
220	static void encode_sv (enc_t *enc, SV *sv);	462	static void encode_sv (enc_t *enc, SV *sv);
221		463
222	static void	464	static void
…		…
227	if (enc->depth >= enc->cbor.max_depth)	469	if (enc->depth >= enc->cbor.max_depth)
228	croak (ERR_NESTING_EXCEEDED);	470	croak (ERR_NESTING_EXCEEDED);
229		471
230	++enc->depth;	472	++enc->depth;
231		473
232	encode_uint (enc, 0x80, len + 1);	474	encode_uint (enc, MAJOR_ARRAY, len + 1);
233		475
		476	if (ecb_expect_false (SvMAGICAL (av)))
234	for (i = 0; i <= len; ++i)	477	for (i = 0; i <= len; ++i)
235	{	478	{
236	SV **svp = av_fetch (av, i, 0);	479	SV **svp = av_fetch (av, i, 0);
237	encode_sv (enc, svp ? *svp : &PL_sv_undef);	480	encode_sv (enc, svp ? *svp : &PL_sv_undef);
238	}	481	}
		482	else
		483	for (i = 0; i <= len; ++i)
		484	{
		485	SV *sv = AvARRAY (av)[i];
		486	encode_sv (enc, sv ? sv : &PL_sv_undef);
		487	}
239		488
240	--enc->depth;	489	--enc->depth;
241	}
242
243	ecb_inline void
244	encode_he (enc_t *enc, HE *he)
245	{
246	if (HeKLEN (he) == HEf_SVKEY)
247	encode_sv (enc, HeSVKEY (he));
248	else
249	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250	}	490	}
251		491
252	static void	492	static void
253	encode_hv (enc_t *enc, HV *hv)	493	encode_hv (enc_t *enc, HV *hv)
254	{	494	{
…		…
260	++enc->depth;	500	++enc->depth;
261		501
262	int pairs = hv_iterinit (hv);	502	int pairs = hv_iterinit (hv);
263	int mg = SvMAGICAL (hv);	503	int mg = SvMAGICAL (hv);
264		504
265	if (mg)	505	if (ecb_expect_false (mg))
266	encode_ch (enc, 0xa0 | 31);	506	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267	else	507	else
268	encode_uint (enc, 0xa0, pairs);	508	encode_uint (enc, MAJOR_MAP, pairs);
269		509
270	while ((he = hv_iternext (hv)))	510	while ((he = hv_iternext (hv)))
271	{	511	{
272	if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273	{
274	SV **svp;
275
276	if (HeKLEN (he) == HEf_SVKEY)	512	if (HeKLEN (he) == HEf_SVKEY)
277	svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/	513	encode_sv (enc, HeSVKEY (he));
278	else
279	svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281	if (SvOK (*svp))
282	{
283	encode_tag (enc, CBOR_TAG_STRINGREF);
284	encode_uint (enc, 0x00, SvUV (*svp));
285	}
286	else
287	{
288	sv_setuv (*svp, enc->stringref_idx);
289	++enc->stringref_idx;
290	encode_he (enc, he);
291	}
292	}
293	else	514	else
294	encode_he (enc, he);	515	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295		516
296	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	517	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
297	}	518	}
298		519
299	if (mg)	520	if (ecb_expect_false (mg))
300	encode_ch (enc, 0xe0 | 31);	521	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301		522
302	--enc->depth;	523	--enc->depth;
303	}	524	}
304		525
305	// encode objects, arrays and special \0=false and \1=true values.	526	// encode objects, arrays and special \0=false and \1=true values.
306	static void	527	static void
307	encode_rv (enc_t *enc, SV *sv)	528	encode_rv (enc_t *enc, SV *sv)
308	{	529	{
309	SvGETMAGIC (sv);	530	SvGETMAGIC (sv);
310
311	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312	&& ecb_expect_false (SvREFCNT (sv) > 1))
313	{
314	if (!enc->shareable)
315	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319	if (SvOK (*svp))
320	{
321	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322	encode_uint (enc, 0x00, SvUV (*svp));
323	return;
324	}
325	else
326	{
327	sv_setuv (*svp, enc->shareable_idx);
328	++enc->shareable_idx;
329	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330	}
331	}
332		531
333	svtype svt = SvTYPE (sv);	532	svtype svt = SvTYPE (sv);
334		533
335	if (ecb_expect_false (SvOBJECT (sv)))	534	if (ecb_expect_false (SvOBJECT (sv)))
336	{	535	{
…		…
343	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	542	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344	? cbor_tagged_stash	543	? cbor_tagged_stash
345	: gv_stashpv ("CBOR::XS::Tagged" , 1);	544	: gv_stashpv ("CBOR::XS::Tagged" , 1);
346		545
347	HV *stash = SvSTASH (sv);	546	HV *stash = SvSTASH (sv);
348	GV *method;
349		547
350	if (stash == boolean_stash)	548	if (stash == boolean_stash)
351	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	549	{
		550	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		551	return;
		552	}
352	else if (stash == error_stash)	553	else if (stash == error_stash)
353	encode_ch (enc, 0xe0 | 23);	554	{
		555	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		556	return;
		557	}
354	else if (stash == tagged_stash)	558	else if (stash == tagged_stash)
355	{	559	{
356	if (svt != SVt_PVAV)	560	if (svt != SVt_PVAV)
357	croak ("encountered CBOR::XS::Tagged object that isn't an array");	561	croak ("encountered CBOR::XS::Tagged object that isn't an array");
358		562
		563	switch (av_len ((AV *)sv))
		564	{
		565	case 2-1:
		566	// actually a tagged value
359	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	567	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	568	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		569	break;
		570
		571	case 3-1:
		572	// a forced type [value, type, undef]
		573	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		574	break;
		575
		576	default:
		577	croak ("encountered malformed CBOR::XS::Tagged object");
		578	}
		579
		580	return;
		581	}
		582	}
		583
		584	if (ecb_expect_false (SvREFCNT (sv) > 1)
		585	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		586	{
		587	if (ecb_expect_false (!enc->shareable))
		588	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		589
		590	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		591
		592	if (SvOK (*svp))
361	}	593	{
		594	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		595	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		596	return;
		597	}
		598	else
		599	{
		600	sv_setuv (*svp, enc->shareable_idx);
		601	++enc->shareable_idx;
		602	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		603	}
		604	}
		605
		606	if (ecb_expect_false (SvOBJECT (sv)))
		607	{
		608	HV *stash = SvSTASH (sv);
		609	GV *method;
		610
		611	if (enc->cbor.flags & F_FORBID_OBJECTS)
		612	croak ("encountered object '%s', but forbid_objects is enabled",
		613	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
362	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	614	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363	{	615	{
364	dSP;	616	dSP;
365		617
366	ENTER; SAVETMPS; PUSHMARK (SP);	618	ENTER; SAVETMPS;
		619	PUSHMARK (SP);
367	// we re-bless the reference to get overload and other niceties right	620	// we re-bless the reference to get overload and other niceties right
368	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	621	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369		622
370	PUTBACK;	623	PUTBACK;
371	// G_SCALAR ensures that return value is 1	624	// G_SCALAR ensures that return value is 1
…		…
384	}	637	}
385	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	638	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386	{	639	{
387	dSP;	640	dSP;
388		641
389	ENTER; SAVETMPS; PUSHMARK (SP);	642	ENTER; SAVETMPS;
		643	PUSHMARK (SP);
390	EXTEND (SP, 2);	644	EXTEND (SP, 2);
391	// we re-bless the reference to get overload and other niceties right	645	// we re-bless the reference to get overload and other niceties right
392	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	646	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393	PUSHs (sv_cbor);	647	PUSHs (sv_cbor);
394		648
…		…
399	// catch this surprisingly common error	653	// catch this surprisingly common error
400	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	654	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	655	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402		656
403	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	657	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404	encode_uint (enc, 0x80, count + 1);	658	encode_uint (enc, MAJOR_ARRAY, count + 1);
405	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	659	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406		660
407	while (count)	661	{
		662	int i;
		663
		664	for (i = 0; i < count; ++i)
408	encode_sv (enc, SP[1 - count--]);	665	encode_sv (enc, SP[i + 1 - count]);
		666
		667	SP -= count;
		668	}
409		669
410	PUTBACK;	670	PUTBACK;
411		671
412	FREETMPS; LEAVE;	672	FREETMPS; LEAVE;
413	}	673	}
…		…
431	{	691	{
432	double nv = SvNVX (sv);	692	double nv = SvNVX (sv);
433		693
434	need (enc, 9);	694	need (enc, 9);
435		695
436	if (ecb_expect_false (nv == (U32)nv))	696	if (ecb_expect_false (nv == (NV)(U32)nv))
437	encode_uint (enc, 0x00, (U32)nv);	697	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438	//TODO: maybe I32?	698	//TODO: maybe I32?
439	else if (ecb_expect_false (nv == (float)nv))	699	else if (ecb_expect_false (nv == (float)nv))
440	{	700	encode_float32 (enc, nv);
441	uint32_t fp = ecb_float_to_binary32 (nv);
442
443	*enc->cur++ = 0xe0 | 26;
444
445	if (!ecb_big_endian ())
446	fp = ecb_bswap32 (fp);
447
448	memcpy (enc->cur, &fp, 4);
449	enc->cur += 4;
450	}
451	else	701	else
452	{	702	encode_float64 (enc, nv);
453	uint64_t fp = ecb_double_to_binary64 (nv);
454
455	*enc->cur++ = 0xe0 | 27;
456
457	if (!ecb_big_endian ())
458	fp = ecb_bswap64 (fp);
459
460	memcpy (enc->cur, &fp, 8);
461	enc->cur += 8;
462	}
463	}	703	}
464		704
465	static void	705	static void
466	encode_sv (enc_t *enc, SV *sv)	706	encode_sv (enc_t *enc, SV *sv)
467	{	707	{
…		…
469		709
470	if (SvPOKp (sv))	710	if (SvPOKp (sv))
471	{	711	{
472	STRLEN len;	712	STRLEN len;
473	char *str = SvPV (sv, len);	713	char *str = SvPV (sv, len);
474	encode_str (enc, SvUTF8 (sv), str, len);	714	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475	}	715	}
476	else if (SvNOKp (sv))	716	else if (SvNOKp (sv))
477	encode_nv (enc, sv);	717	encode_nv (enc, sv);
478	else if (SvIOKp (sv))	718	else if (SvIOKp (sv))
479	{	719	encode_int (e,v sv);
480	if (SvIsUV (sv))
481	encode_uint (enc, 0x00, SvUVX (sv));
482	else if (SvIVX (sv) >= 0)
483	encode_uint (enc, 0x00, SvIVX (sv));
484	else
485	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
486	}
487	else if (SvROK (sv))	720	else if (SvROK (sv))
488	encode_rv (enc, SvRV (sv));	721	encode_rv (enc, SvRV (sv));
489	else if (!SvOK (sv))	722	else if (!SvOK (sv))
490	encode_ch (enc, 0xe0 | 22);	723	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	724	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492	encode_ch (enc, 0xe0 | 23);	725	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493	else	726	else
494	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	727	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
495	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	728	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496	}	729	}
497		730
498	static SV *	731	static SV *
499	encode_cbor (SV *scalar, CBOR *cbor)	732	encode_cbor (SV *scalar, CBOR *cbor)
500	{	733	{
501	enc_t enc = { };	734	enc_t enc = { 0 };
502		735
503	enc.cbor = *cbor;	736	enc.cbor = *cbor;
504	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	737	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505	enc.cur = SvPVX (enc.sv);	738	enc.cur = SvPVX (enc.sv);
506	enc.end = SvEND (enc.sv);	739	enc.end = SvEND (enc.sv);
507		740
508	SvPOK_only (enc.sv);	741	SvPOK_only (enc.sv);
509		742
510	if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))	743	if (cbor->flags & F_PACK_STRINGS)
511	{	744	{
512	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	745	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	746	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	747	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515	}	748	}
…		…
537	CBOR cbor;	770	CBOR cbor;
538	U32 depth; // recursion depth	771	U32 depth; // recursion depth
539	U32 maxdepth; // recursion depth limit	772	U32 maxdepth; // recursion depth limit
540	AV *shareable;	773	AV *shareable;
541	AV *stringref;	774	AV *stringref;
		775	SV *decode_tagged;
		776	SV *err_sv; // optional sv for error, needs to be freed
542	} dec_t;	777	} dec_t;
543		778
544	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	779	// set dec->err to ERRSV
		780	ecb_cold static void
		781	err_errsv (dec_t *dec)
		782	{
		783	if (!dec->err)
		784	{
		785	dec->err_sv = newSVsv (ERRSV);
545		786
546	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	787	// chop off the trailing \n
		788	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		789	*SvEND (dec->err_sv) = 0;
547		790
		791	dec->err = SvPVutf8_nolen (dec->err_sv);
		792	}
		793	}
		794
		795	// the following functions are used to reduce code size and help the compiler to optimise
		796	ecb_cold static void
		797	err_set (dec_t *dec, const char *reason)
		798	{
		799	if (!dec->err)
		800	dec->err = reason;
		801	}
		802
		803	ecb_cold static void
		804	err_unexpected_end (dec_t *dec)
		805	{
		806	err_set (dec, "unexpected end of CBOR data");
		807	}
		808
		809	#define ERR_DO(do) SB do; goto fail; SE
		810	#define ERR(reason) ERR_DO (err_set (dec, reason))
		811	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		812
		813	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		814
548	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	815	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
549	#define DEC_DEC_DEPTH --dec->depth	816	#define DEC_DEC_DEPTH --dec->depth
550		817
551	static UV	818	static UV
552	decode_uint (dec_t *dec)	819	decode_uint (dec_t *dec)
553	{	820	{
554	switch (*dec->cur & 31)	821	U8 m = *dec->cur & MINOR_MASK;
555	{	822	++dec->cur;
556	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559	return *dec->cur++ & 31;
560		823
561	case 24:	824	if (ecb_expect_true (m < LENGTH_EXT1))
		825	return m;
		826	else if (ecb_expect_true (m == LENGTH_EXT1))
		827	{
562	WANT (2);	828	WANT (1);
563	dec->cur += 2;	829	dec->cur += 1;
564	return dec->cur[-1];	830	return dec->cur[-1];
565		831	}
566	case 25:	832	else if (ecb_expect_true (m == LENGTH_EXT2))
		833	{
567	WANT (3);	834	WANT (2);
568	dec->cur += 3;	835	dec->cur += 2;
569	return (((UV)dec->cur[-2]) << 8)	836	return (((UV)dec->cur[-2]) << 8)
570	| ((UV)dec->cur[-1]);	837	| ((UV)dec->cur[-1]);
571		838	}
572	case 26:	839	else if (ecb_expect_true (m == LENGTH_EXT4))
		840	{
573	WANT (5);	841	WANT (4);
574	dec->cur += 5;	842	dec->cur += 4;
575	return (((UV)dec->cur[-4]) << 24)	843	return (((UV)dec->cur[-4]) << 24)
576	| (((UV)dec->cur[-3]) << 16)	844	| (((UV)dec->cur[-3]) << 16)
577	| (((UV)dec->cur[-2]) << 8)	845	| (((UV)dec->cur[-2]) << 8)
578	| ((UV)dec->cur[-1]);	846	| ((UV)dec->cur[-1]);
579		847	}
580	case 27:	848	else if (ecb_expect_true (m == LENGTH_EXT8))
		849	{
581	WANT (9);	850	WANT (8);
582	dec->cur += 9;	851	dec->cur += 8;
		852
		853	return
		854	#if UVSIZE < 8
		855	0
		856	#else
583	return (((UV)dec->cur[-8]) << 56)	857	(((UV)dec->cur[-8]) << 56)
584	| (((UV)dec->cur[-7]) << 48)	858	| (((UV)dec->cur[-7]) << 48)
585	| (((UV)dec->cur[-6]) << 40)	859	| (((UV)dec->cur[-6]) << 40)
586	| (((UV)dec->cur[-5]) << 32)	860	| (((UV)dec->cur[-5]) << 32)
		861	#endif
587	| (((UV)dec->cur[-4]) << 24)	862	| (((UV)dec->cur[-4]) << 24)
588	| (((UV)dec->cur[-3]) << 16)	863	| (((UV)dec->cur[-3]) << 16)
589	| (((UV)dec->cur[-2]) << 8)	864	| (((UV)dec->cur[-2]) << 8)
590	| ((UV)dec->cur[-1]);	865	| ((UV)dec->cur[-1]);
591		866	}
592	default:	867	else
593	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	868	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594	}
595		869
596	fail:	870	fail:
597	return 0;	871	return 0;
598	}	872	}
599		873
…		…
604	{	878	{
605	AV *av = newAV ();	879	AV *av = newAV ();
606		880
607	DEC_INC_DEPTH;	881	DEC_INC_DEPTH;
608		882
609	if ((*dec->cur & 31) == 31)	883	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610	{	884	{
611	++dec->cur;	885	++dec->cur;
612		886
613	for (;;)	887	for (;;)
614	{	888	{
615	WANT (1);	889	WANT (1);
616		890
617	if (*dec->cur == (0xe0 | 31))	891	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
618	{	892	{
619	++dec->cur;	893	++dec->cur;
620	break;	894	break;
621	}	895	}
622		896
623	av_push (av, decode_sv (dec));	897	av_push (av, decode_sv (dec));
624	}	898	}
625	}	899	}
626	else	900	else
627	{	901	{
628	int i, len = decode_uint (dec);	902	UV i, len = decode_uint (dec);
629		903
		904	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630	av_fill (av, len - 1);	905	av_fill (av, len - 1);
631		906
632	for (i = 0; i < len; ++i)	907	for (i = 0; i < len; ++i)
633	AvARRAY (av)[i] = decode_sv (dec);	908	AvARRAY (av)[i] = decode_sv (dec);
634	}	909	}
635		910
636	DEC_DEC_DEPTH;	911	DEC_DEC_DEPTH;
637	return newRV_noinc ((SV *)av);	912	return newRV_noinc ((SV *)av);
638		913
639	fail:	914	fail:
640	SvREFCNT_dec (av);	915	SvREFCNT_dec_NN (av);
641	DEC_DEC_DEPTH;	916	DEC_DEC_DEPTH;
642	return &PL_sv_undef;	917	return &PL_sv_undef;
643	}	918	}
644		919
645	static void	920	static void
646	decode_he (dec_t *dec, HV *hv)	921	decode_he (dec_t *dec, HV *hv)
647	{	922	{
648	// for speed reasons, we specialcase single-string	923	// for speed reasons, we specialcase single-string
649	// byte or utf-8 strings as keys.	924	// byte or utf-8 strings as keys, but only when !stringref
650		925
651	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	926	if (ecb_expect_true (!dec->stringref))
		927	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652	{	928	{
653	I32 len = decode_uint (dec);	929	STRLEN len = decode_uint (dec);
654	char *key = (char *)dec->cur;	930	char *key = (char *)dec->cur;
655		931
		932	WANT (len);
656	dec->cur += len;	933	dec->cur += len;
657		934
658	if (ecb_expect_false (dec->stringref))
659	av_push (dec->stringref, newSVpvn (key, len));
660
661	hv_store (hv, key, len, decode_sv (dec), 0);	935	hv_store (hv, key, len, decode_sv (dec), 0);
		936
		937	return;
662	}	938	}
663	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	939	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664	{	940	{
665	I32 len = decode_uint (dec);	941	STRLEN len = decode_uint (dec);
666	char *key = (char *)dec->cur;	942	char *key = (char *)dec->cur;
667		943
		944	WANT (len);
668	dec->cur += len;	945	dec->cur += len;
669		946
670	if (ecb_expect_false (dec->stringref))	947	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	948	if (!is_utf8_string (key, len))
		949	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672		950
673	hv_store (hv, key, -len, decode_sv (dec), 0);	951	hv_store (hv, key, -len, decode_sv (dec), 0);
		952
		953	return;
674	}	954	}
675	else	955
676	{
677	SV *k = decode_sv (dec);	956	SV *k = decode_sv (dec);
678	SV *v = decode_sv (dec);	957	SV *v = decode_sv (dec);
679		958
		959	// we leak memory if uncaught exceptions are thrown by random magical
		960	// methods, and this is hopefully the only place where it can happen,
		961	// so if there is a chance of an exception, take the very slow path.
		962	// since catching exceptions is "undocumented/internal/forbidden" by
		963	// the new p5p powers, we need to call out to a perl function :/
		964	if (ecb_expect_false (SvAMAGIC (k)))
		965	{
		966	dSP;
		967
		968	ENTER; SAVETMPS;
		969	PUSHMARK (SP);
		970	EXTEND (SP, 3);
		971	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		972	PUSHs (sv_2mortal (k));
		973	PUSHs (sv_2mortal (v));
		974
		975	PUTBACK;
		976	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		977	SPAGAIN;
		978
		979	FREETMPS; LEAVE;
		980
		981	if (SvTRUE (ERRSV))
		982	ERR_ERRSV;
		983
		984	return;
		985	}
		986
680	hv_store_ent (hv, k, v, 0);	987	hv_store_ent (hv, k, v, 0);
681	SvREFCNT_dec (k);	988	SvREFCNT_dec_NN (k);
682	}	989
		990	fail:
		991	;
683	}	992	}
684		993
685	static SV *	994	static SV *
686	decode_hv (dec_t *dec)	995	decode_hv (dec_t *dec)
687	{	996	{
688	HV *hv = newHV ();	997	HV *hv = newHV ();
689		998
690	DEC_INC_DEPTH;	999	DEC_INC_DEPTH;
691		1000
692	if ((*dec->cur & 31) == 31)	1001	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693	{	1002	{
694	++dec->cur;	1003	++dec->cur;
695		1004
696	for (;;)	1005	for (;;)
697	{	1006	{
698	WANT (1);	1007	WANT (1);
699		1008
700	if (*dec->cur == (0xe0 | 31))	1009	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
701	{	1010	{
702	++dec->cur;	1011	++dec->cur;
703	break;	1012	break;
704	}	1013	}
705		1014
706	decode_he (dec, hv);	1015	decode_he (dec, hv);
707	}	1016	}
708	}	1017	}
709	else	1018	else
710	{	1019	{
711	int pairs = decode_uint (dec);	1020	UV pairs = decode_uint (dec);
		1021
		1022	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
712		1023
713	while (pairs--)	1024	while (pairs--)
714	decode_he (dec, hv);	1025	decode_he (dec, hv);
715	}	1026	}
716		1027
717	DEC_DEC_DEPTH;	1028	DEC_DEC_DEPTH;
718	return newRV_noinc ((SV *)hv);	1029	return newRV_noinc ((SV *)hv);
719		1030
720	fail:	1031	fail:
721	SvREFCNT_dec (hv);	1032	SvREFCNT_dec_NN (hv);
722	DEC_DEC_DEPTH;	1033	DEC_DEC_DEPTH;
723	return &PL_sv_undef;	1034	return &PL_sv_undef;
724	}	1035	}
725		1036
726	static SV *	1037	static SV *
727	decode_str (dec_t *dec, int utf8)	1038	decode_str (dec_t *dec, int utf8)
728	{	1039	{
729	SV *sv = 0;	1040	SV *sv = 0;
730		1041
731	if ((*dec->cur & 31) == 31)	1042	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
732	{	1043	{
		1044	// indefinite length strings
733	++dec->cur;	1045	++dec->cur;
734		1046
		1047	U8 major = *dec->cur & MAJOR_MISC;
		1048
735	sv = newSVpvn ("", 0);	1049	sv = newSVpvn ("", 0);
736		1050
737	// not very fast, and certainly not robust against illegal input
738	for (;;)	1051	for (;;)
739	{	1052	{
740	WANT (1);	1053	WANT (1);
741		1054
742	if (*dec->cur == (0xe0 | 31))	1055	if ((*dec->cur - major) > LENGTH_EXT8)
		1056	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743	{	1057	{
744	++dec->cur;	1058	++dec->cur;
745	break;	1059	break;
746	}	1060	}
		1061	else
		1062	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747		1063
748	sv_catsv (sv, decode_sv (dec));	1064	STRLEN len = decode_uint (dec);
		1065
		1066	WANT (len);
		1067	sv_catpvn (sv, dec->cur, len);
		1068	dec->cur += len;
749	}	1069	}
750	}	1070	}
751	else	1071	else
752	{	1072	{
753	STRLEN len = decode_uint (dec);	1073	STRLEN len = decode_uint (dec);
754		1074
755	WANT (len);	1075	WANT (len);
756	sv = newSVpvn (dec->cur, len);	1076	sv = newSVpvn (dec->cur, len);
757	dec->cur += len;	1077	dec->cur += len;
		1078
		1079	if (ecb_expect_false (dec->stringref)
		1080	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		1081	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
758	}	1082	}
759		1083
760	if (utf8)	1084	if (utf8)
		1085	{
		1086	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1087	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1088	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1089
761	SvUTF8_on (sv);	1090	SvUTF8_on (sv);
762		1091	}
763	if (ecb_expect_false (dec->stringref))
764	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765		1092
766	return sv;	1093	return sv;
767		1094
768	fail:	1095	fail:
769	SvREFCNT_dec (sv);	1096	SvREFCNT_dec (sv);
…		…
788	sv = newRV_noinc (decode_sv (dec));	1115	sv = newRV_noinc (decode_sv (dec));
789	break;	1116	break;
790		1117
791	case CBOR_TAG_STRINGREF_NAMESPACE:	1118	case CBOR_TAG_STRINGREF_NAMESPACE:
792	{	1119	{
		1120	// do not use SAVETMPS/FREETMPS, as these will
		1121	// erase mortalised caches, e.g. "shareable"
793	ENTER; SAVETMPS;	1122	ENTER;
794		1123
795	SAVESPTR (dec->stringref);	1124	SAVESPTR (dec->stringref);
796	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1125	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797		1126
798	sv = decode_sv (dec);	1127	sv = decode_sv (dec);
799		1128
800	FREETMPS; LEAVE;	1129	LEAVE;
801	}	1130	}
802	break;	1131	break;
803		1132
804	case CBOR_TAG_STRINGREF:	1133	case CBOR_TAG_STRINGREF:
805	{	1134	{
806	if ((*dec->cur >> 5) != 0)	1135	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1136	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808		1137
809	UV idx = decode_uint (dec);	1138	UV idx = decode_uint (dec);
810		1139
811	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1140	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
812	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1141	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813		1142
814	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1143	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815	}	1144	}
816	break;	1145	break;
…		…
818	case CBOR_TAG_VALUE_SHAREABLE:	1147	case CBOR_TAG_VALUE_SHAREABLE:
819	{	1148	{
820	if (ecb_expect_false (!dec->shareable))	1149	if (ecb_expect_false (!dec->shareable))
821	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1150	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822		1151
		1152	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1153	{
823	sv = newSV (0);	1154	sv = newSV (0);
824	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1155	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825		1156
826	SV *osv = decode_sv (dec);	1157	SV *osv = decode_sv (dec);
827	sv_setsv (sv, osv);	1158	sv_setsv (sv, osv);
828	SvREFCNT_dec_NN (osv);	1159	SvREFCNT_dec_NN (osv);
		1160	}
		1161	else
		1162	{
		1163	av_push (dec->shareable, &PL_sv_undef);
		1164	int idx = AvFILLp (dec->shareable);
		1165	sv = decode_sv (dec);
		1166	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1167	}
829	}	1168	}
830	break;	1169	break;
831		1170
832	case CBOR_TAG_VALUE_SHAREDREF:	1171	case CBOR_TAG_VALUE_SHAREDREF:
833	{	1172	{
834	if ((*dec->cur >> 5) != 0)	1173	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1174	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836		1175
837	UV idx = decode_uint (dec);	1176	UV idx = decode_uint (dec);
838		1177
839	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1178	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
840	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1179	ERR ("corrupted CBOR data (sharedref index out of bounds)");
841		1180
842	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1181	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1182
		1183	if (sv == &PL_sv_undef)
		1184	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
843	}	1185	}
844	break;	1186	break;
845		1187
846	case CBOR_TAG_PERL_OBJECT:	1188	case CBOR_TAG_PERL_OBJECT:
847	{	1189	{
		1190	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1191	goto filter;
		1192
848	sv = decode_sv (dec);	1193	sv = decode_sv (dec);
849		1194
850	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1195	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851	ERR ("corrupted CBOR data (non-array perl object)");	1196	ERR ("corrupted CBOR data (non-array perl object)");
852		1197
…		…
862	if (!method)	1207	if (!method)
863	ERR ("cannot decode perl-object (package does not have a THAW method)");	1208	ERR ("cannot decode perl-object (package does not have a THAW method)");
864		1209
865	dSP;	1210	dSP;
866		1211
867	ENTER; SAVETMPS; PUSHMARK (SP);	1212	ENTER; SAVETMPS;
		1213	PUSHMARK (SP);
868	EXTEND (SP, len + 1);	1214	EXTEND (SP, len + 1);
869	// we re-bless the reference to get overload and other niceties right	1215	// we re-bless the reference to get overload and other niceties right
870	PUSHs (*av_fetch (av, 0, 1));	1216	PUSHs (*av_fetch (av, 0, 1));
871	PUSHs (sv_cbor);	1217	PUSHs (sv_cbor);
872		1218
…		…
880	SPAGAIN;	1226	SPAGAIN;
881		1227
882	if (SvTRUE (ERRSV))	1228	if (SvTRUE (ERRSV))
883	{	1229	{
884	FREETMPS; LEAVE;	1230	FREETMPS; LEAVE;
885	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1231	ERR_ERRSV;
886	}	1232	}
887		1233
888	SvREFCNT_dec (sv);	1234	SvREFCNT_dec_NN (sv);
889	sv = SvREFCNT_inc (POPs);	1235	sv = SvREFCNT_inc (POPs);
890		1236
891	PUTBACK;	1237	PUTBACK;
892		1238
893	FREETMPS; LEAVE;	1239	FREETMPS; LEAVE;
894	}	1240	}
895	break;	1241	break;
896		1242
897	default:	1243	default:
		1244	filter:
898	{	1245	{
		1246	SV *tag_sv = newSVuv (tag);
		1247
899	sv = decode_sv (dec);	1248	sv = decode_sv (dec);
900		1249
		1250	dSP;
		1251	ENTER; SAVETMPS;
		1252	PUSHMARK (SP);
		1253	EXTEND (SP, 2);
		1254	PUSHs (tag_sv);
		1255	PUSHs (sv);
		1256
		1257	PUTBACK;
		1258	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1259	SPAGAIN;
		1260
		1261	if (SvTRUE (ERRSV))
		1262	{
		1263	SvREFCNT_dec_NN (tag_sv);
		1264	FREETMPS; LEAVE;
		1265	ERR_ERRSV;
		1266	}
		1267
		1268	if (count)
		1269	{
		1270	SvREFCNT_dec_NN (tag_sv);
		1271	SvREFCNT_dec_NN (sv);
		1272	sv = SvREFCNT_inc_NN (TOPs);
		1273	SP -= count;
		1274	}
		1275	else
		1276	{
901	AV *av = newAV ();	1277	AV *av = newAV ();
902	av_push (av, newSVuv (tag));	1278	av_push (av, tag_sv);
903	av_push (av, sv);	1279	av_push (av, sv);
904		1280
905	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1281	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906	? cbor_tagged_stash	1282	? cbor_tagged_stash
907	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1283	: gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1284	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1285	}
		1286
		1287	PUTBACK;
		1288
		1289	FREETMPS; LEAVE;
910	}	1290	}
911	break;	1291	break;
912	}	1292	}
913		1293
914	return sv;	1294	return sv;
…		…
921	static SV *	1301	static SV *
922	decode_sv (dec_t *dec)	1302	decode_sv (dec_t *dec)
923	{	1303	{
924	WANT (1);	1304	WANT (1);
925		1305
926	switch (*dec->cur >> 5)	1306	switch (*dec->cur >> MAJOR_SHIFT)
927	{	1307	{
928	case 0: // unsigned int	1308	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929	return newSVuv (decode_uint (dec));	1309	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930	case 1: // negative int	1310	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931	return newSViv (-1 - (IV)decode_uint (dec));	1311	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932	case 2: // octet string	1312	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933	return decode_str (dec, 0);	1313	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934	case 3: // utf-8 string	1314	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935	return decode_str (dec, 1);	1315
936	case 4: // array	1316	case MAJOR_MISC >> MAJOR_SHIFT:
937	return decode_av (dec);
938	case 5: // map
939	return decode_hv (dec);
940	case 6: // tag
941	return decode_tagged (dec);
942	case 7: // misc
943	switch (*dec->cur++ & 31)	1317	switch (*dec->cur++ & MINOR_MASK)
944	{	1318	{
945	case 20:	1319	case SIMPLE_FALSE:
946	#if CBOR_SLOW	1320	#if CBOR_SLOW
947	types_false = get_bool ("Types::Serialiser::false");	1321	types_false = get_bool ("Types::Serialiser::false");
948	#endif	1322	#endif
949	return newSVsv (types_false);	1323	return newSVsv (types_false);
950	case 21:	1324	case SIMPLE_TRUE:
951	#if CBOR_SLOW	1325	#if CBOR_SLOW
952	types_true = get_bool ("Types::Serialiser::true");	1326	types_true = get_bool ("Types::Serialiser::true");
953	#endif	1327	#endif
954	return newSVsv (types_true);	1328	return newSVsv (types_true);
955	case 22:	1329	case SIMPLE_NULL:
956	return newSVsv (&PL_sv_undef);	1330	return newSVsv (&PL_sv_undef);
957	case 23:	1331	case SIMPLE_UNDEF:
958	#if CBOR_SLOW	1332	#if CBOR_SLOW
959	types_error = get_bool ("Types::Serialiser::error");	1333	types_error = get_bool ("Types::Serialiser::error");
960	#endif	1334	#endif
961	return newSVsv (types_error);	1335	return newSVsv (types_error);
962		1336
963	case 25:	1337	case MISC_FLOAT16:
964	{	1338	{
965	WANT (2);	1339	WANT (2);
966		1340
967	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1341	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968	dec->cur += 2;	1342	dec->cur += 2;
969		1343
970	return newSVnv (ecb_binary16_to_float (fp));	1344	return newSVnv (ecb_binary16_to_float (fp));
971	}	1345	}
972		1346
973	case 26:	1347	case MISC_FLOAT32:
974	{	1348	{
975	uint32_t fp;	1349	uint32_t fp;
976	WANT (4);	1350	WANT (4);
977	memcpy (&fp, dec->cur, 4);	1351	memcpy (&fp, dec->cur, 4);
978	dec->cur += 4;	1352	dec->cur += 4;
…		…
981	fp = ecb_bswap32 (fp);	1355	fp = ecb_bswap32 (fp);
982		1356
983	return newSVnv (ecb_binary32_to_float (fp));	1357	return newSVnv (ecb_binary32_to_float (fp));
984	}	1358	}
985		1359
986	case 27:	1360	case MISC_FLOAT64:
987	{	1361	{
988	uint64_t fp;	1362	uint64_t fp;
989	WANT (8);	1363	WANT (8);
990	memcpy (&fp, dec->cur, 8);	1364	memcpy (&fp, dec->cur, 8);
991	dec->cur += 8;	1365	dec->cur += 8;
…		…
994	fp = ecb_bswap64 (fp);	1368	fp = ecb_bswap64 (fp);
995		1369
996	return newSVnv (ecb_binary64_to_double (fp));	1370	return newSVnv (ecb_binary64_to_double (fp));
997	}	1371	}
998		1372
999	// 0..19 unassigned	1373	// 0..19 unassigned simple
1000	// 24 reserved + unassigned (reserved values are not encodable)	1374	// 24 reserved + unassigned simple (reserved values are not encodable)
		1375	// 28-30 unassigned misc
		1376	// 31 break code
1001	default:	1377	default:
1002	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1378	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003	}	1379	}
1004		1380
1005	break;	1381	break;
1006	}	1382	}
1007		1383
…		…
1010	}	1386	}
1011		1387
1012	static SV *	1388	static SV *
1013	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1389	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014	{	1390	{
1015	dec_t dec = { };	1391	dec_t dec = { 0 };
1016	SV *sv;	1392	SV *sv;
1017	STRLEN len;	1393	STRLEN len;
1018	char *data = SvPVbyte (string, len);	1394	char *data = SvPVbyte (string, len);
1019		1395
1020	if (len > cbor->max_size && cbor->max_size)	1396	if (len > cbor->max_size && cbor->max_size)
…		…
1034	if (dec.cur != dec.end && !dec.err)	1410	if (dec.cur != dec.end && !dec.err)
1035	dec.err = "garbage after CBOR object";	1411	dec.err = "garbage after CBOR object";
1036		1412
1037	if (dec.err)	1413	if (dec.err)
1038	{	1414	{
		1415	if (dec.shareable)
		1416	{
		1417	// need to break cyclic links, which would all be in shareable
		1418	int i;
		1419	SV **svp;
		1420
		1421	for (i = av_len (dec.shareable) + 1; i--; )
		1422	if ((svp = av_fetch (dec.shareable, i, 0)))
		1423	sv_setsv (*svp, &PL_sv_undef);
		1424	}
		1425
1039	SvREFCNT_dec (sv);	1426	SvREFCNT_dec_NN (sv);
		1427
		1428	if (dec.err_sv)
		1429	sv_2mortal (dec.err_sv);
		1430
1040	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1431	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1041	}	1432	}
1042		1433
1043	sv = sv_2mortal (sv);	1434	sv = sv_2mortal (sv);
1044		1435
1045	return sv;	1436	return sv;
1046	}	1437	}
1047		1438
		1439	/////////////////////////////////////////////////////////////////////////////
		1440	// incremental parser
		1441
		1442	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1443
		1444	// returns 0 for notyet, 1 for success or error
		1445	static int
		1446	incr_parse (CBOR *self, SV *cborstr)
		1447	{
		1448	STRLEN cur;
		1449	SvPV (cborstr, cur);
		1450
		1451	while (ecb_expect_true (self->incr_need <= cur))
		1452	{
		1453	// table of integer count bytes
		1454	static I8 incr_len[MINOR_MASK + 1] = {
		1455	0, 0, 0, 0, 0, 0, 0, 0,
		1456	0, 0, 0, 0, 0, 0, 0, 0,
		1457	0, 0, 0, 0, 0, 0, 0, 0,
		1458	1, 2, 4, 8,-1,-1,-1,-2
		1459	};
		1460
		1461	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1462	U8 m = *p & MINOR_MASK;
		1463	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1464	I8 ilen = incr_len[m];
		1465
		1466	self->incr_need = self->incr_pos + 1;
		1467
		1468	if (ecb_expect_false (ilen < 0))
		1469	{
		1470	if (m != MINOR_INDEF)
		1471	return 1; // error
		1472
		1473	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1474	{
		1475	if (count >= 0)
		1476	return 1; // error
		1477
		1478	count = 1;
		1479	}
		1480	else
		1481	{
		1482	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1483	count = -1;
		1484	}
		1485	}
		1486	else
		1487	{
		1488	self->incr_need += ilen;
		1489	if (ecb_expect_false (self->incr_need > cur))
		1490	return 0;
		1491
		1492	int major = *p >> MAJOR_SHIFT;
		1493
		1494	switch (major)
		1495	{
		1496	case MAJOR_TAG >> MAJOR_SHIFT:
		1497	++count; // tags merely prefix another value
		1498	break;
		1499
		1500	case MAJOR_BYTES >> MAJOR_SHIFT:
		1501	case MAJOR_TEXT >> MAJOR_SHIFT:
		1502	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1503	case MAJOR_MAP >> MAJOR_SHIFT:
		1504	{
		1505	UV len;
		1506
		1507	if (ecb_expect_false (ilen))
		1508	{
		1509	len = 0;
		1510
		1511	do {
		1512	len = (len << 8) | *++p;
		1513	} while (--ilen);
		1514	}
		1515	else
		1516	len = m;
		1517
		1518	switch (major)
		1519	{
		1520	case MAJOR_BYTES >> MAJOR_SHIFT:
		1521	case MAJOR_TEXT >> MAJOR_SHIFT:
		1522	self->incr_need += len;
		1523	if (ecb_expect_false (self->incr_need > cur))
		1524	return 0;
		1525
		1526	break;
		1527
		1528	case MAJOR_MAP >> MAJOR_SHIFT:
		1529	len <<= 1;
		1530	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1531	if (len)
		1532	{
		1533	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1534	count = len + 1;
		1535	}
		1536	break;
		1537	}
		1538	}
		1539	}
		1540	}
		1541
		1542	self->incr_pos = self->incr_need;
		1543
		1544	if (count > 0)
		1545	{
		1546	while (!--count)
		1547	{
		1548	if (!AvFILLp (self->incr_count))
		1549	return 1; // done
		1550
		1551	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1552	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1553	}
		1554
		1555	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1556	}
		1557	}
		1558
		1559	return 0;
		1560	}
		1561
		1562
1048	/////////////////////////////////////////////////////////////////////////////	1563	/////////////////////////////////////////////////////////////////////////////
1049	// XS interface functions	1564	// XS interface functions
1050		1565
1051	MODULE = CBOR::XS PACKAGE = CBOR::XS	1566	MODULE = CBOR::XS PACKAGE = CBOR::XS
1052		1567
…		…
1060		1575
1061	types_true = get_bool ("Types::Serialiser::true" );	1576	types_true = get_bool ("Types::Serialiser::true" );
1062	types_false = get_bool ("Types::Serialiser::false");	1577	types_false = get_bool ("Types::Serialiser::false");
1063	types_error = get_bool ("Types::Serialiser::error");	1578	types_error = get_bool ("Types::Serialiser::error");
1064		1579
		1580	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1581
1065	sv_cbor = newSVpv ("CBOR", 0);	1582	sv_cbor = newSVpv ("CBOR", 0);
1066	SvREADONLY_on (sv_cbor);	1583	SvREADONLY_on (sv_cbor);
		1584
		1585	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067	}	1586	}
1068		1587
1069	PROTOTYPES: DISABLE	1588	PROTOTYPES: DISABLE
1070		1589
1071	void CLONE (...)	1590	void CLONE (...)
…		…
1090	void shrink (CBOR *self, int enable = 1)	1609	void shrink (CBOR *self, int enable = 1)
1091	ALIAS:	1610	ALIAS:
1092	shrink = F_SHRINK	1611	shrink = F_SHRINK
1093	allow_unknown = F_ALLOW_UNKNOWN	1612	allow_unknown = F_ALLOW_UNKNOWN
1094	allow_sharing = F_ALLOW_SHARING	1613	allow_sharing = F_ALLOW_SHARING
1095	dedup_keys = F_DEDUP_KEYS	1614	allow_cycles = F_ALLOW_CYCLES
		1615	forbid_objects = F_FORBID_OBJECTS
1096	dedup_strings = F_DEDUP_STRINGS	1616	pack_strings = F_PACK_STRINGS
		1617	text_keys = F_TEXT_KEYS
		1618	text_strings = F_TEXT_STRINGS
		1619	validate_utf8 = F_VALIDATE_UTF8
1097	PPCODE:	1620	PPCODE:
1098	{	1621	{
1099	if (enable)	1622	if (enable)
1100	self->flags |= ix;	1623	self->flags |= ix;
1101	else	1624	else
…		…
1107	void get_shrink (CBOR *self)	1630	void get_shrink (CBOR *self)
1108	ALIAS:	1631	ALIAS:
1109	get_shrink = F_SHRINK	1632	get_shrink = F_SHRINK
1110	get_allow_unknown = F_ALLOW_UNKNOWN	1633	get_allow_unknown = F_ALLOW_UNKNOWN
1111	get_allow_sharing = F_ALLOW_SHARING	1634	get_allow_sharing = F_ALLOW_SHARING
1112	get_dedup_keys = F_DEDUP_KEYS	1635	get_allow_cycles = F_ALLOW_CYCLES
		1636	get_forbid_objects = F_FORBID_OBJECTS
1113	get_dedup_strings = F_DEDUP_STRINGS	1637	get_pack_strings = F_PACK_STRINGS
		1638	get_text_keys = F_TEXT_KEYS
		1639	get_text_strings = F_TEXT_STRINGS
		1640	get_validate_utf8 = F_VALIDATE_UTF8
1114	PPCODE:	1641	PPCODE:
1115	XPUSHs (boolSV (self->flags & ix));	1642	XPUSHs (boolSV (self->flags & ix));
1116		1643
1117	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1644	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118	PPCODE:	1645	PPCODE:
…		…
1131	XPUSHs (ST (0));	1658	XPUSHs (ST (0));
1132		1659
1133	int get_max_size (CBOR *self)	1660	int get_max_size (CBOR *self)
1134	CODE:	1661	CODE:
1135	RETVAL = self->max_size;	1662	RETVAL = self->max_size;
		1663	OUTPUT:
		1664	RETVAL
		1665
		1666	void filter (CBOR *self, SV *filter = 0)
		1667	PPCODE:
		1668	SvREFCNT_dec (self->filter);
		1669	self->filter = filter ? newSVsv (filter) : filter;
		1670	XPUSHs (ST (0));
		1671
		1672	SV *get_filter (CBOR *self)
		1673	CODE:
		1674	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136	OUTPUT:	1675	OUTPUT:
1137	RETVAL	1676	RETVAL
1138		1677
1139	void encode (CBOR *self, SV *scalar)	1678	void encode (CBOR *self, SV *scalar)
1140	PPCODE:	1679	PPCODE:
…		…
1155	EXTEND (SP, 2);	1694	EXTEND (SP, 2);
1156	PUSHs (sv);	1695	PUSHs (sv);
1157	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1696	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158	}	1697	}
1159		1698
		1699	void incr_parse (CBOR *self, SV *cborstr)
		1700	ALIAS:
		1701	incr_parse_multiple = 1
		1702	PPCODE:
		1703	{
		1704	if (SvUTF8 (cborstr))
		1705	sv_utf8_downgrade (cborstr, 0);
		1706
		1707	if (!self->incr_count)
		1708	{
		1709	self->incr_count = newAV ();
		1710	self->incr_pos = 0;
		1711	self->incr_need = 1;
		1712
		1713	av_push (self->incr_count, newSViv (1));
		1714	}
		1715
		1716	do
		1717	{
		1718	if (!incr_parse (self, cborstr))
		1719	{
		1720	if (self->incr_need > self->max_size && self->max_size)
		1721	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1722	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1723
		1724	break;
		1725	}
		1726
		1727	SV *sv;
		1728	char *offset;
		1729
		1730	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1731	XPUSHs (sv);
		1732
		1733	sv_chop (cborstr, offset);
		1734
		1735	av_clear (self->incr_count);
		1736	av_push (self->incr_count, newSViv (1));
		1737
		1738	self->incr_pos = 0;
		1739	self->incr_need = self->incr_pos + 1;
		1740	}
		1741	while (ix);
		1742	}
		1743
		1744	void incr_reset (CBOR *self)
		1745	CODE:
		1746	{
		1747	SvREFCNT_dec (self->incr_count);
		1748	self->incr_count = 0;
		1749	}
		1750
		1751	void DESTROY (CBOR *self)
		1752	PPCODE:
		1753	cbor_free (self);
		1754
1160	PROTOTYPES: ENABLE	1755	PROTOTYPES: ENABLE
1161		1756
1162	void encode_cbor (SV *scalar)	1757	void encode_cbor (SV *scalar)
		1758	ALIAS:
		1759	encode_cbor = 0
		1760	encode_cbor_sharing = F_ALLOW_SHARING
1163	PPCODE:	1761	PPCODE:
1164	{	1762	{
1165	CBOR cbor;	1763	CBOR cbor;
1166	cbor_init (&cbor);	1764	cbor_init (&cbor);
		1765	cbor.flags |= ix;
1167	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1766	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168	XPUSHs (scalar);	1767	XPUSHs (scalar);
1169	}	1768	}
1170		1769
1171	void decode_cbor (SV *cborstr)	1770	void decode_cbor (SV *cborstr)
…		…
1175	cbor_init (&cbor);	1774	cbor_init (&cbor);
1176	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1775	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1177	XPUSHs (cborstr);	1776	XPUSHs (cborstr);
1178	}	1777	}
1179		1778
		1779	#ifdef __AFL_COMPILER
		1780
		1781	void
		1782	afl_init ()
		1783	CODE:
		1784	__AFL_INIT ();
		1785
		1786	int
		1787	afl_loop (unsigned int count = 10000)
		1788	CODE:
		1789	RETVAL = __AFL_LOOP (count);
		1790	OUTPUT:
		1791	RETVAL
		1792
		1793	#endif
		1794

Diff Legend

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