ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.30 by root, Thu Nov 28 11:36:53 2013 UTC vs.
Revision 1.71 by root, Fri Mar 19 17:30:27 2021 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
24	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	31	#endif
27		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	};
		69
28	// known tags	70	// known tags
29	enum cbor_tag	71	enum cbor_tag
30	{	72	{
31	// extensions	73	// extensions
32	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref	74	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object	75	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object	76	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing	77	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing	78	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref	79	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	80	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39		81
40	// rfc7049	82	// rfc7049
41	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	83	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	84	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43	CBOR_TAG_POS_BIGNUM = 2, // byte string	85	CBOR_TAG_POS_BIGNUM = 2, // byte string
44	CBOR_TAG_NEG_BIGNUM = 3, // byte string	86	CBOR_TAG_NEG_BIGNUM = 3, // byte string
45	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	87	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46	CBOR_TAG_BIGFLOAT = 5, // array	88	CBOR_TAG_BIGFLOAT = 5, // array
47		89
48	CBOR_TAG_CONV_B64U = 21, // base64url, any	90	CBOR_TAG_CONV_B64U = 21, // base64url, any
49	CBOR_TAG_CONV_B64 = 22, // base64, any	91	CBOR_TAG_CONV_B64 = 22, // base64, any
50	CBOR_TAG_CONV_HEX = 23, // base16, any	92	CBOR_TAG_CONV_HEX = 23, // base16, any
51	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	93	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52		94
53	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	95	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	96	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	97	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	98	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58		100
59	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	AS_MAP = 7,
		115	// possibly future enhancements: (generic) float, (generic) string
60	};	116	};
61		117
62	#define F_SHRINK 0x00000001UL	118	#define F_SHRINK 0x00000001UL
63	#define F_ALLOW_UNKNOWN 0x00000002UL	119	#define F_ALLOW_UNKNOWN 0x00000002UL
64	#define F_ALLOW_SHARING 0x00000004UL //TODO	120	#define F_ALLOW_SHARING 0x00000004UL
65	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	121	#define F_ALLOW_CYCLES 0x00000008UL
		122	#define F_FORBID_OBJECTS 0x00000010UL
		123	#define F_PACK_STRINGS 0x00000020UL
		124	#define F_TEXT_KEYS 0x00000040UL
		125	#define F_TEXT_STRINGS 0x00000080UL
		126	#define F_VALIDATE_UTF8 0x00000100UL
66		127
67	#define INIT_SIZE 32 // initial scalar size to be allocated	128	#define INIT_SIZE 32 // initial scalar size to be allocated
68		129
69	#define SB do {	130	#define SB do {
70	#define SE } while (0)	131	#define SE } while (0)
…		…
89	typedef struct {	150	typedef struct {
90	U32 flags;	151	U32 flags;
91	U32 max_depth;	152	U32 max_depth;
92	STRLEN max_size;	153	STRLEN max_size;
93	SV *filter;	154	SV *filter;
		155
		156	// for the incremental parser
		157	STRLEN incr_pos; // the current offset into the text
		158	STRLEN incr_need; // minimum bytes needed to decode
		159	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94	} CBOR;	160	} CBOR;
95		161
96	ecb_inline void	162	ecb_inline void
97	cbor_init (CBOR *cbor)	163	cbor_init (CBOR *cbor)
98	{	164	{
…		…
102		168
103	ecb_inline void	169	ecb_inline void
104	cbor_free (CBOR *cbor)	170	cbor_free (CBOR *cbor)
105	{	171	{
106	SvREFCNT_dec (cbor->filter);	172	SvREFCNT_dec (cbor->filter);
		173	SvREFCNT_dec (cbor->incr_count);
107	}	174	}
108		175
109	/////////////////////////////////////////////////////////////////////////////	176	/////////////////////////////////////////////////////////////////////////////
110	// utility functions	177	// utility functions
111		178
…		…
134	#endif	201	#endif
135	}	202	}
136	}	203	}
137		204
138	// minimum length of a string to be registered for stringref	205	// minimum length of a string to be registered for stringref
139	ecb_inline int	206	ecb_inline STRLEN
140	minimum_string_length (UV idx)	207	minimum_string_length (UV idx)
141	{	208	{
142	return idx > 23	209	return idx <= 23 ? 3
143	? idx > 0xffU	210	: idx <= 0xffU ? 4
144	? idx > 0xffffU	211	: idx <= 0xffffU ? 5
145	? idx > 0xffffffffU	212	: idx <= 0xffffffffU ? 7
146	? 11	213	: 11;
147	: 7
148	: 5
149	: 4
150	: 3;
151	}	214	}
152		215
153	/////////////////////////////////////////////////////////////////////////////	216	/////////////////////////////////////////////////////////////////////////////
154	// encoder	217	// encoder
155		218
…		…
168	} enc_t;	231	} enc_t;
169		232
170	ecb_inline void	233	ecb_inline void
171	need (enc_t *enc, STRLEN len)	234	need (enc_t *enc, STRLEN len)
172	{	235	{
173	if (ecb_expect_false (enc->cur + len >= enc->end))	236	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
174	{	237	{
175	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	238	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	239	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177	enc->cur = SvPVX (enc->sv) + cur;	240	enc->cur = SvPVX (enc->sv) + cur;
178	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	241	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
179	}	242	}
180	}	243	}
181		244
		245	static void encode_sv (enc_t *enc, SV *sv);
		246
182	ecb_inline void	247	ecb_inline void
183	encode_ch (enc_t *enc, char ch)	248	encode_ch (enc_t *enc, char ch)
184	{	249	{
185	need (enc, 1);	250	need (enc, 1);
186	*enc->cur++ = ch;	251	*enc->cur++ = ch;
187	}	252	}
188		253
		254	// used for tags, intregers, element counts and so on
189	static void	255	static void
190	encode_uint (enc_t *enc, int major, UV len)	256	encode_uint (enc_t *enc, int major, UV len)
191	{	257	{
192	need (enc, 9);	258	need (enc, 9);
193		259
194	if (len < 24)	260	if (ecb_expect_true (len < LENGTH_EXT1))
195	*enc->cur++ = major | len;	261	*enc->cur++ = major | len;
196	else if (len <= 0xff)	262	else if (ecb_expect_true (len <= 0xffU))
197	{	263	{
198	*enc->cur++ = major | 24;	264	*enc->cur++ = major | LENGTH_EXT1;
199	*enc->cur++ = len;	265	*enc->cur++ = len;
200	}	266	}
201	else if (len <= 0xffff)	267	else if (len <= 0xffffU)
202	{	268	{
203	*enc->cur++ = major | 25;	269	*enc->cur++ = major | LENGTH_EXT2;
204	*enc->cur++ = len >> 8;	270	*enc->cur++ = len >> 8;
205	*enc->cur++ = len;	271	*enc->cur++ = len;
206	}	272	}
207	else if (len <= 0xffffffff)	273	else if (len <= 0xffffffffU)
208	{	274	{
209	*enc->cur++ = major | 26;	275	*enc->cur++ = major | LENGTH_EXT4;
210	*enc->cur++ = len >> 24;	276	*enc->cur++ = len >> 24;
211	*enc->cur++ = len >> 16;	277	*enc->cur++ = len >> 16;
212	*enc->cur++ = len >> 8;	278	*enc->cur++ = len >> 8;
213	*enc->cur++ = len;	279	*enc->cur++ = len;
214	}	280	}
215	else	281	else
216	{	282	{
217	*enc->cur++ = major | 27;	283	*enc->cur++ = major | LENGTH_EXT8;
218	*enc->cur++ = len >> 56;	284	*enc->cur++ = len >> 56;
219	*enc->cur++ = len >> 48;	285	*enc->cur++ = len >> 48;
220	*enc->cur++ = len >> 40;	286	*enc->cur++ = len >> 40;
221	*enc->cur++ = len >> 32;	287	*enc->cur++ = len >> 32;
222	*enc->cur++ = len >> 24;	288	*enc->cur++ = len >> 24;
…		…
224	*enc->cur++ = len >> 8;	290	*enc->cur++ = len >> 8;
225	*enc->cur++ = len;	291	*enc->cur++ = len;
226	}	292	}
227	}	293	}
228		294
		295	// encodes a perl value into a CBOR integer
		296	ecb_inline void
		297	encode_int (enc_t *enc, SV *sv)
		298	{
		299	if (SvIsUV (sv))
		300	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		301	else if (SvIVX (sv) >= 0)
		302	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		303	else
		304	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		305	}
		306
229	ecb_inline void	307	ecb_inline void
230	encode_tag (enc_t *enc, UV tag)	308	encode_tag (enc_t *enc, UV tag)
231	{	309	{
232	encode_uint (enc, 0xc0, tag);	310	encode_uint (enc, MAJOR_TAG, tag);
		311	}
		312
		313	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		314	ecb_noinline static void
		315	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
		316	{
		317	STRLEN ulen = len;
		318	U8 *p, *pend = (U8 *)str + len;
		319
		320	for (p = (U8 *)str; p < pend; ++p)
		321	ulen += *p >> 7; // count set high bits
		322
		323	encode_uint (enc, MAJOR_TEXT, ulen);
		324
		325	need (enc, ulen);
		326	for (p = (U8 *)str; p < pend; ++p)
		327	if (*p < 0x80)
		328	*enc->cur++ = *p;
		329	else
		330	{
		331	*enc->cur++ = 0xc0 + (*p >> 6);
		332	*enc->cur++ = 0x80 + (*p & 63);
		333	}
233	}	334	}
234		335
235	ecb_inline void	336	ecb_inline void
236	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	337	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
237	{	338	{
238	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	339	if (ecb_expect_false (upgrade_utf8))
		340	if (!utf8)
		341	{
		342	encode_str_utf8 (enc, utf8, str, len);
		343	return;
		344	}
		345
		346	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
239	need (enc, len);	347	need (enc, len);
240	memcpy (enc->cur, str, len);	348	memcpy (enc->cur, str, len);
241	enc->cur += len;	349	enc->cur += len;
242	}	350	}
243		351
244	static void	352	ecb_inline void
245	encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)	353	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
246	{	354	{
247	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	355	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248	{	356	{
249	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	357	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250		358
251	if (SvOK (*svp))	359	if (SvOK (*svp))
252	{	360	{
253	// already registered, use stringref	361	// already registered, use stringref
254	encode_tag (enc, CBOR_TAG_STRINGREF);	362	encode_tag (enc, CBOR_TAG_STRINGREF);
255	encode_uint (enc, 0x00, SvUV (*svp));	363	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
256	return;	364	return;
257	}	365	}
258	else if (len >= minimum_string_length (enc->stringref_idx))	366	else if (len >= minimum_string_length (enc->stringref_idx))
259	{	367	{
260	// register only	368	// register only
261	sv_setuv (*svp, enc->stringref_idx);	369	sv_setuv (*svp, enc->stringref_idx);
262	++enc->stringref_idx;	370	++enc->stringref_idx;
263	}	371	}
264	}	372	}
265		373
266	encode_str (enc, utf8, str, len);	374	encode_str (enc, upgrade_utf8, utf8, str, len);
267	}	375	}
268		376
269	static void encode_sv (enc_t *enc, SV *sv);	377	ecb_inline void
		378	encode_float16 (enc_t *enc, NV nv)
		379	{
		380	need (enc, 1+2);
		381
		382	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		383
		384	uint16_t fp = ecb_float_to_binary16 (nv);
		385
		386	if (!ecb_big_endian ())
		387	fp = ecb_bswap16 (fp);
		388
		389	memcpy (enc->cur, &fp, 2);
		390	enc->cur += 2;
		391	}
		392
		393	ecb_inline void
		394	encode_float32 (enc_t *enc, NV nv)
		395	{
		396	need (enc, 1+4);
		397
		398	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		399
		400	uint32_t fp = ecb_float_to_binary32 (nv);
		401
		402	if (!ecb_big_endian ())
		403	fp = ecb_bswap32 (fp);
		404
		405	memcpy (enc->cur, &fp, 4);
		406	enc->cur += 4;
		407	}
		408
		409	ecb_inline void
		410	encode_float64 (enc_t *enc, NV nv)
		411	{
		412	need (enc, 1+8);
		413
		414	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		415
		416	uint64_t fp = ecb_double_to_binary64 (nv);
		417
		418	if (!ecb_big_endian ())
		419	fp = ecb_bswap64 (fp);
		420
		421	memcpy (enc->cur, &fp, 8);
		422	enc->cur += 8;
		423	}
		424
		425	ecb_inline void
		426	encode_bool (enc_t *enc, int istrue)
		427	{
		428	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		429	}
		430
		431	// encodes an arrayref containing key-value pairs as CBOR map
		432	ecb_inline void
		433	encode_array_as_map (enc_t *enc, SV *sv)
		434	{
		435	if (enc->depth >= enc->cbor.max_depth)
		436	croak (ERR_NESTING_EXCEEDED);
		437
		438	++enc->depth;
		439
		440	// as_map does error checking for us, but we re-check in case
		441	// things have changed.
		442
		443	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
		444	croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
		445
		446	AV *av = (AV *)SvRV (sv);
		447	int i, len = av_len (av);
		448
		449	if (!(len & 1))
		450	croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
		451
		452	encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
		453
		454	for (i = 0; i <= len; ++i)
		455	{
		456	SV **svp = av_fetch (av, i, 0);
		457	encode_sv (enc, svp ? *svp : &PL_sv_undef);
		458	}
		459
		460	--enc->depth;
		461	}
		462
		463	ecb_inline void
		464	encode_forced (enc_t *enc, UV type, SV *sv)
		465	{
		466	switch (type)
		467	{
		468	case AS_CBOR:
		469	{
		470	STRLEN len;
		471	char *str = SvPVbyte (sv, len);
		472
		473	need (enc, len);
		474	memcpy (enc->cur, str, len);
		475	enc->cur += len;
		476	}
		477	break;
		478
		479	case AS_BYTES:
		480	{
		481	STRLEN len;
		482	char *str = SvPVbyte (sv, len);
		483	encode_strref (enc, 0, 0, str, len);
		484	}
		485	break;
		486
		487	case AS_TEXT:
		488	{
		489	STRLEN len;
		490	char *str = SvPVutf8 (sv, len);
		491	encode_strref (enc, 1, 1, str, len);
		492	}
		493	break;
		494
		495	case AS_INT: encode_int (enc, sv); break;
		496
		497	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		498	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		499	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		500
		501	case AS_MAP: encode_array_as_map (enc, sv); break;
		502
		503	default:
		504	croak ("encountered malformed CBOR::XS::Tagged object");
		505	}
		506	}
270		507
271	static void	508	static void
272	encode_av (enc_t *enc, AV *av)	509	encode_av (enc_t *enc, AV *av)
273	{	510	{
274	int i, len = av_len (av);	511	int i, len = av_len (av);
…		…
276	if (enc->depth >= enc->cbor.max_depth)	513	if (enc->depth >= enc->cbor.max_depth)
277	croak (ERR_NESTING_EXCEEDED);	514	croak (ERR_NESTING_EXCEEDED);
278		515
279	++enc->depth;	516	++enc->depth;
280		517
281	encode_uint (enc, 0x80, len + 1);	518	encode_uint (enc, MAJOR_ARRAY, len + 1);
282		519
		520	if (ecb_expect_false (SvMAGICAL (av)))
283	for (i = 0; i <= len; ++i)	521	for (i = 0; i <= len; ++i)
284	{	522	{
285	SV **svp = av_fetch (av, i, 0);	523	SV **svp = av_fetch (av, i, 0);
286	encode_sv (enc, svp ? *svp : &PL_sv_undef);	524	encode_sv (enc, svp ? *svp : &PL_sv_undef);
287	}	525	}
		526	else
		527	for (i = 0; i <= len; ++i)
		528	{
		529	SV *sv = AvARRAY (av)[i];
		530	encode_sv (enc, sv ? sv : &PL_sv_undef);
		531	}
288		532
289	--enc->depth;	533	--enc->depth;
290	}	534	}
291		535
292	static void	536	static void
…		…
300	++enc->depth;	544	++enc->depth;
301		545
302	int pairs = hv_iterinit (hv);	546	int pairs = hv_iterinit (hv);
303	int mg = SvMAGICAL (hv);	547	int mg = SvMAGICAL (hv);
304		548
305	if (mg)	549	if (ecb_expect_false (mg))
306	encode_ch (enc, 0xa0 | 31);	550	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
307	else	551	else
308	encode_uint (enc, 0xa0, pairs);	552	encode_uint (enc, MAJOR_MAP, pairs);
309		553
310	while ((he = hv_iternext (hv)))	554	while ((he = hv_iternext (hv)))
311	{	555	{
312	if (HeKLEN (he) == HEf_SVKEY)	556	if (HeKLEN (he) == HEf_SVKEY)
313	encode_sv (enc, HeSVKEY (he));	557	encode_sv (enc, HeSVKEY (he));
314	else	558	else
315	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	559	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
316		560
317	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	561	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318	}	562	}
319		563
320	if (mg)	564	if (ecb_expect_false (mg))
321	encode_ch (enc, 0xe0 | 31);	565	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
322		566
323	--enc->depth;	567	--enc->depth;
324	}	568	}
325		569
326	// encode objects, arrays and special \0=false and \1=true values.	570	// encode objects, arrays and special \0=false and \1=true values.
327	static void	571	static void
328	encode_rv (enc_t *enc, SV *sv)	572	encode_rv (enc_t *enc, SV *sv)
329	{	573	{
330	SvGETMAGIC (sv);	574	SvGETMAGIC (sv);
331
332	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
333	&& ecb_expect_false (SvREFCNT (sv) > 1))
334	{
335	if (!enc->shareable)
336	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
337
338	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
339
340	if (SvOK (*svp))
341	{
342	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
343	encode_uint (enc, 0x00, SvUV (*svp));
344	return;
345	}
346	else
347	{
348	sv_setuv (*svp, enc->shareable_idx);
349	++enc->shareable_idx;
350	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
351	}
352	}
353		575
354	svtype svt = SvTYPE (sv);	576	svtype svt = SvTYPE (sv);
355		577
356	if (ecb_expect_false (SvOBJECT (sv)))	578	if (ecb_expect_false (SvOBJECT (sv)))
357	{	579	{
…		…
364	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	586	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
365	? cbor_tagged_stash	587	? cbor_tagged_stash
366	: gv_stashpv ("CBOR::XS::Tagged" , 1);	588	: gv_stashpv ("CBOR::XS::Tagged" , 1);
367		589
368	HV *stash = SvSTASH (sv);	590	HV *stash = SvSTASH (sv);
369	GV *method;
370		591
371	if (stash == boolean_stash)	592	if (stash == boolean_stash)
372	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	593	{
		594	encode_bool (enc, SvIV (sv));
		595	return;
		596	}
373	else if (stash == error_stash)	597	else if (stash == error_stash)
374	encode_ch (enc, 0xe0 | 23);	598	{
		599	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		600	return;
		601	}
375	else if (stash == tagged_stash)	602	else if (stash == tagged_stash)
376	{	603	{
377	if (svt != SVt_PVAV)	604	if (svt != SVt_PVAV)
378	croak ("encountered CBOR::XS::Tagged object that isn't an array");	605	croak ("encountered CBOR::XS::Tagged object that isn't an array");
379		606
		607	switch (av_len ((AV *)sv))
		608	{
		609	case 2-1:
		610	// actually a tagged value
380	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	611	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
381	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	612	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		613	break;
		614
		615	case 3-1:
		616	// a forced type [value, type, undef]
		617	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		618	break;
		619
		620	default:
		621	croak ("encountered malformed CBOR::XS::Tagged object");
		622	}
		623
		624	return;
		625	}
		626	}
		627
		628	if (ecb_expect_false (SvREFCNT (sv) > 1)
		629	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		630	{
		631	if (ecb_expect_false (!enc->shareable))
		632	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		633
		634	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		635
		636	if (SvOK (*svp))
382	}	637	{
		638	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		639	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		640	return;
		641	}
		642	else
		643	{
		644	sv_setuv (*svp, enc->shareable_idx);
		645	++enc->shareable_idx;
		646	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		647	}
		648	}
		649
		650	if (ecb_expect_false (SvOBJECT (sv)))
		651	{
		652	HV *stash = SvSTASH (sv);
		653	GV *method;
		654
		655	if (enc->cbor.flags & F_FORBID_OBJECTS)
		656	croak ("encountered object '%s', but forbid_objects is enabled",
		657	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
383	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	658	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
384	{	659	{
385	dSP;	660	dSP;
386		661
387	ENTER; SAVETMPS; PUSHMARK (SP);	662	ENTER; SAVETMPS;
		663	PUSHMARK (SP);
388	// we re-bless the reference to get overload and other niceties right	664	// we re-bless the reference to get overload and other niceties right
389	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	665	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
390		666
391	PUTBACK;	667	PUTBACK;
392	// G_SCALAR ensures that return value is 1	668	// G_SCALAR ensures that return value is 1
…		…
405	}	681	}
406	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	682	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
407	{	683	{
408	dSP;	684	dSP;
409		685
410	ENTER; SAVETMPS; PUSHMARK (SP);	686	ENTER; SAVETMPS;
		687	PUSHMARK (SP);
411	EXTEND (SP, 2);	688	EXTEND (SP, 2);
412	// we re-bless the reference to get overload and other niceties right	689	// we re-bless the reference to get overload and other niceties right
413	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	690	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
414	PUSHs (sv_cbor);	691	PUSHs (sv_cbor);
415		692
…		…
420	// catch this surprisingly common error	697	// catch this surprisingly common error
421	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	698	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
422	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	699	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
423		700
424	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	701	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
425	encode_uint (enc, 0x80, count + 1);	702	encode_uint (enc, MAJOR_ARRAY, count + 1);
426	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	703	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
427		704
428	while (count)	705	{
		706	int i;
		707
		708	for (i = 0; i < count; ++i)
429	encode_sv (enc, SP[1 - count--]);	709	encode_sv (enc, SP[i + 1 - count]);
		710
		711	SP -= count;
		712	}
430		713
431	PUTBACK;	714	PUTBACK;
432		715
433	FREETMPS; LEAVE;	716	FREETMPS; LEAVE;
434	}	717	}
…		…
452	{	735	{
453	double nv = SvNVX (sv);	736	double nv = SvNVX (sv);
454		737
455	need (enc, 9);	738	need (enc, 9);
456		739
457	if (ecb_expect_false (nv == (U32)nv))	740	if (ecb_expect_false (nv == (NV)(U32)nv))
458	encode_uint (enc, 0x00, (U32)nv);	741	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
459	//TODO: maybe I32?	742	//TODO: maybe I32?
460	else if (ecb_expect_false (nv == (float)nv))	743	else if (ecb_expect_false (nv == (float)nv))
461	{	744	encode_float32 (enc, nv);
462	uint32_t fp = ecb_float_to_binary32 (nv);
463
464	*enc->cur++ = 0xe0 | 26;
465
466	if (!ecb_big_endian ())
467	fp = ecb_bswap32 (fp);
468
469	memcpy (enc->cur, &fp, 4);
470	enc->cur += 4;
471	}
472	else	745	else
473	{	746	encode_float64 (enc, nv);
474	uint64_t fp = ecb_double_to_binary64 (nv);
475
476	*enc->cur++ = 0xe0 | 27;
477
478	if (!ecb_big_endian ())
479	fp = ecb_bswap64 (fp);
480
481	memcpy (enc->cur, &fp, 8);
482	enc->cur += 8;
483	}
484	}	747	}
485		748
486	static void	749	static void
487	encode_sv (enc_t *enc, SV *sv)	750	encode_sv (enc_t *enc, SV *sv)
488	{	751	{
…		…
490		753
491	if (SvPOKp (sv))	754	if (SvPOKp (sv))
492	{	755	{
493	STRLEN len;	756	STRLEN len;
494	char *str = SvPV (sv, len);	757	char *str = SvPV (sv, len);
495	encode_strref (enc, SvUTF8 (sv), str, len);	758	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
496	}	759	}
497	else if (SvNOKp (sv))	760	else if (SvNOKp (sv))
498	encode_nv (enc, sv);	761	encode_nv (enc, sv);
499	else if (SvIOKp (sv))	762	else if (SvIOKp (sv))
500	{	763	encode_int (enc, sv);
501	if (SvIsUV (sv))
502	encode_uint (enc, 0x00, SvUVX (sv));
503	else if (SvIVX (sv) >= 0)
504	encode_uint (enc, 0x00, SvIVX (sv));
505	else
506	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
507	}
508	else if (SvROK (sv))	764	else if (SvROK (sv))
509	encode_rv (enc, SvRV (sv));	765	encode_rv (enc, SvRV (sv));
510	else if (!SvOK (sv))	766	else if (!SvOK (sv))
511	encode_ch (enc, 0xe0 | 22);	767	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
512	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	768	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
513	encode_ch (enc, 0xe0 | 23);	769	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
514	else	770	else
515	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	771	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
516	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	772	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
517	}	773	}
518		774
519	static SV *	775	static SV *
520	encode_cbor (SV *scalar, CBOR *cbor)	776	encode_cbor (SV *scalar, CBOR *cbor)
521	{	777	{
522	enc_t enc = { };	778	enc_t enc = { 0 };
523		779
524	enc.cbor = *cbor;	780	enc.cbor = *cbor;
525	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	781	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
526	enc.cur = SvPVX (enc.sv);	782	enc.cur = SvPVX (enc.sv);
527	enc.end = SvEND (enc.sv);	783	enc.end = SvEND (enc.sv);
528		784
529	SvPOK_only (enc.sv);	785	SvPOK_only (enc.sv);
530		786
531	if (cbor->flags & F_ALLOW_STRINGREF)	787	if (cbor->flags & F_PACK_STRINGS)
532	{	788	{
533	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	789	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
534	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	790	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
535	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	791	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
536	}	792	}
…		…
559	U32 depth; // recursion depth	815	U32 depth; // recursion depth
560	U32 maxdepth; // recursion depth limit	816	U32 maxdepth; // recursion depth limit
561	AV *shareable;	817	AV *shareable;
562	AV *stringref;	818	AV *stringref;
563	SV *decode_tagged;	819	SV *decode_tagged;
		820	SV *err_sv; // optional sv for error, needs to be freed
564	} dec_t;	821	} dec_t;
565		822
566	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	823	// set dec->err to ERRSV
		824	ecb_cold static void
		825	err_errsv (dec_t *dec)
		826	{
		827	if (!dec->err)
		828	{
		829	dec->err_sv = newSVsv (ERRSV);
567		830
568	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	831	// chop off the trailing \n
		832	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		833	*SvEND (dec->err_sv) = 0;
569		834
		835	dec->err = SvPVutf8_nolen (dec->err_sv);
		836	}
		837	}
		838
		839	// the following functions are used to reduce code size and help the compiler to optimise
		840	ecb_cold static void
		841	err_set (dec_t *dec, const char *reason)
		842	{
		843	if (!dec->err)
		844	dec->err = reason;
		845	}
		846
		847	ecb_cold static void
		848	err_unexpected_end (dec_t *dec)
		849	{
		850	err_set (dec, "unexpected end of CBOR data");
		851	}
		852
		853	#define ERR_DO(do) SB do; goto fail; SE
		854	#define ERR(reason) ERR_DO (err_set (dec, reason))
		855	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		856
		857	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		858
570	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	859	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
571	#define DEC_DEC_DEPTH --dec->depth	860	#define DEC_DEC_DEPTH --dec->depth
572		861
573	static UV	862	static UV
574	decode_uint (dec_t *dec)	863	decode_uint (dec_t *dec)
575	{	864	{
576	switch (*dec->cur & 31)	865	U8 m = *dec->cur & MINOR_MASK;
577	{	866	++dec->cur;
578	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
579	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
580	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
581	return *dec->cur++ & 31;
582		867
583	case 24:	868	if (ecb_expect_true (m < LENGTH_EXT1))
		869	return m;
		870	else if (ecb_expect_true (m == LENGTH_EXT1))
		871	{
584	WANT (2);	872	WANT (1);
585	dec->cur += 2;	873	dec->cur += 1;
586	return dec->cur[-1];	874	return dec->cur[-1];
587		875	}
588	case 25:	876	else if (ecb_expect_true (m == LENGTH_EXT2))
		877	{
589	WANT (3);	878	WANT (2);
590	dec->cur += 3;	879	dec->cur += 2;
591	return (((UV)dec->cur[-2]) << 8)	880	return (((UV)dec->cur[-2]) << 8)
592	| ((UV)dec->cur[-1]);	881	| ((UV)dec->cur[-1]);
593		882	}
594	case 26:	883	else if (ecb_expect_true (m == LENGTH_EXT4))
		884	{
595	WANT (5);	885	WANT (4);
596	dec->cur += 5;	886	dec->cur += 4;
597	return (((UV)dec->cur[-4]) << 24)	887	return (((UV)dec->cur[-4]) << 24)
598	| (((UV)dec->cur[-3]) << 16)	888	| (((UV)dec->cur[-3]) << 16)
599	| (((UV)dec->cur[-2]) << 8)	889	| (((UV)dec->cur[-2]) << 8)
600	| ((UV)dec->cur[-1]);	890	| ((UV)dec->cur[-1]);
601		891	}
602	case 27:	892	else if (ecb_expect_true (m == LENGTH_EXT8))
		893	{
603	WANT (9);	894	WANT (8);
604	dec->cur += 9;	895	dec->cur += 8;
		896
		897	return
		898	#if UVSIZE < 8
		899	0
		900	#else
605	return (((UV)dec->cur[-8]) << 56)	901	(((UV)dec->cur[-8]) << 56)
606	| (((UV)dec->cur[-7]) << 48)	902	| (((UV)dec->cur[-7]) << 48)
607	| (((UV)dec->cur[-6]) << 40)	903	| (((UV)dec->cur[-6]) << 40)
608	| (((UV)dec->cur[-5]) << 32)	904	| (((UV)dec->cur[-5]) << 32)
		905	#endif
609	| (((UV)dec->cur[-4]) << 24)	906	| (((UV)dec->cur[-4]) << 24)
610	| (((UV)dec->cur[-3]) << 16)	907	| (((UV)dec->cur[-3]) << 16)
611	| (((UV)dec->cur[-2]) << 8)	908	| (((UV)dec->cur[-2]) << 8)
612	| ((UV)dec->cur[-1]);	909	| ((UV)dec->cur[-1]);
613		910	}
614	default:	911	else
615	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	912	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
616	}
617		913
618	fail:	914	fail:
619	return 0;	915	return 0;
620	}	916	}
621		917
…		…
626	{	922	{
627	AV *av = newAV ();	923	AV *av = newAV ();
628		924
629	DEC_INC_DEPTH;	925	DEC_INC_DEPTH;
630		926
631	if ((*dec->cur & 31) == 31)	927	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
632	{	928	{
633	++dec->cur;	929	++dec->cur;
634		930
635	for (;;)	931	for (;;)
636	{	932	{
637	WANT (1);	933	WANT (1);
638		934
639	if (*dec->cur == (0xe0 | 31))	935	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
640	{	936	{
641	++dec->cur;	937	++dec->cur;
642	break;	938	break;
643	}	939	}
644		940
645	av_push (av, decode_sv (dec));	941	av_push (av, decode_sv (dec));
646	}	942	}
647	}	943	}
648	else	944	else
649	{	945	{
650	int i, len = decode_uint (dec);	946	UV i, len = decode_uint (dec);
651		947
		948	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
652	av_fill (av, len - 1);	949	av_fill (av, len - 1);
653		950
654	for (i = 0; i < len; ++i)	951	for (i = 0; i < len; ++i)
655	AvARRAY (av)[i] = decode_sv (dec);	952	AvARRAY (av)[i] = decode_sv (dec);
656	}	953	}
657		954
658	DEC_DEC_DEPTH;	955	DEC_DEC_DEPTH;
659	return newRV_noinc ((SV *)av);	956	return newRV_noinc ((SV *)av);
660		957
661	fail:	958	fail:
662	SvREFCNT_dec (av);	959	SvREFCNT_dec_NN (av);
663	DEC_DEC_DEPTH;	960	DEC_DEC_DEPTH;
664	return &PL_sv_undef;	961	return &PL_sv_undef;
665	}	962	}
666		963
667	static void	964	static void
…		…
669	{	966	{
670	// for speed reasons, we specialcase single-string	967	// for speed reasons, we specialcase single-string
671	// byte or utf-8 strings as keys, but only when !stringref	968	// byte or utf-8 strings as keys, but only when !stringref
672		969
673	if (ecb_expect_true (!dec->stringref))	970	if (ecb_expect_true (!dec->stringref))
674	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	971	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
675	{	972	{
676	I32 len = decode_uint (dec);	973	STRLEN len = decode_uint (dec);
677	char *key = (char *)dec->cur;	974	char *key = (char *)dec->cur;
678		975
		976	WANT (len);
679	dec->cur += len;	977	dec->cur += len;
680		978
681	if (ecb_expect_false (dec->stringref))
682	av_push (dec->stringref, newSVpvn (key, len));
683
684	hv_store (hv, key, len, decode_sv (dec), 0);	979	hv_store (hv, key, len, decode_sv (dec), 0);
685		980
686	return;	981	return;
687	}	982	}
688	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	983	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
689	{	984	{
690	I32 len = decode_uint (dec);	985	STRLEN len = decode_uint (dec);
691	char *key = (char *)dec->cur;	986	char *key = (char *)dec->cur;
692		987
		988	WANT (len);
693	dec->cur += len;	989	dec->cur += len;
694		990
695	if (ecb_expect_false (dec->stringref))	991	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
696	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	992	if (!is_utf8_string ((U8 *)key, len))
		993	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
697		994
698	hv_store (hv, key, -len, decode_sv (dec), 0);	995	hv_store (hv, key, -len, decode_sv (dec), 0);
699		996
700	return;	997	return;
701	}	998	}
702		999
703	SV *k = decode_sv (dec);	1000	SV *k = decode_sv (dec);
704	SV *v = decode_sv (dec);	1001	SV *v = decode_sv (dec);
705		1002
		1003	// we leak memory if uncaught exceptions are thrown by random magical
		1004	// methods, and this is hopefully the only place where it can happen,
		1005	// so if there is a chance of an exception, take the very slow path.
		1006	// since catching exceptions is "undocumented/internal/forbidden" by
		1007	// the new p5p powers, we need to call out to a perl function :/
		1008	if (ecb_expect_false (SvAMAGIC (k)))
		1009	{
		1010	dSP;
		1011
		1012	ENTER; SAVETMPS;
		1013	PUSHMARK (SP);
		1014	EXTEND (SP, 3);
		1015	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		1016	PUSHs (sv_2mortal (k));
		1017	PUSHs (sv_2mortal (v));
		1018
		1019	PUTBACK;
		1020	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		1021	SPAGAIN;
		1022
		1023	FREETMPS; LEAVE;
		1024
		1025	if (SvTRUE (ERRSV))
		1026	ERR_ERRSV;
		1027
		1028	return;
		1029	}
		1030
706	hv_store_ent (hv, k, v, 0);	1031	hv_store_ent (hv, k, v, 0);
707	SvREFCNT_dec (k);	1032	SvREFCNT_dec_NN (k);
		1033
		1034	fail:
		1035	;
708	}	1036	}
709		1037
710	static SV *	1038	static SV *
711	decode_hv (dec_t *dec)	1039	decode_hv (dec_t *dec)
712	{	1040	{
713	HV *hv = newHV ();	1041	HV *hv = newHV ();
714		1042
715	DEC_INC_DEPTH;	1043	DEC_INC_DEPTH;
716		1044
717	if ((*dec->cur & 31) == 31)	1045	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
718	{	1046	{
719	++dec->cur;	1047	++dec->cur;
720		1048
721	for (;;)	1049	for (;;)
722	{	1050	{
723	WANT (1);	1051	WANT (1);
724		1052
725	if (*dec->cur == (0xe0 | 31))	1053	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
726	{	1054	{
727	++dec->cur;	1055	++dec->cur;
728	break;	1056	break;
729	}	1057	}
730		1058
731	decode_he (dec, hv);	1059	decode_he (dec, hv);
732	}	1060	}
733	}	1061	}
734	else	1062	else
735	{	1063	{
736	int pairs = decode_uint (dec);	1064	UV pairs = decode_uint (dec);
		1065
		1066	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
737		1067
738	while (pairs--)	1068	while (pairs--)
739	decode_he (dec, hv);	1069	decode_he (dec, hv);
740	}	1070	}
741		1071
742	DEC_DEC_DEPTH;	1072	DEC_DEC_DEPTH;
743	return newRV_noinc ((SV *)hv);	1073	return newRV_noinc ((SV *)hv);
744		1074
745	fail:	1075	fail:
746	SvREFCNT_dec (hv);	1076	SvREFCNT_dec_NN (hv);
747	DEC_DEC_DEPTH;	1077	DEC_DEC_DEPTH;
748	return &PL_sv_undef;	1078	return &PL_sv_undef;
749	}	1079	}
750		1080
751	static SV *	1081	static SV *
752	decode_str (dec_t *dec, int utf8)	1082	decode_str (dec_t *dec, int utf8)
753	{	1083	{
754	SV *sv = 0;	1084	SV *sv = 0;
755		1085
756	if ((*dec->cur & 31) == 31)	1086	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
757	{	1087	{
		1088	// indefinite length strings
758	++dec->cur;	1089	++dec->cur;
759		1090
		1091	U8 major = *dec->cur & MAJOR_MISC;
		1092
760	sv = newSVpvn ("", 0);	1093	sv = newSVpvn ("", 0);
761		1094
762	// not very fast, and certainly not robust against illegal input
763	for (;;)	1095	for (;;)
764	{	1096	{
765	WANT (1);	1097	WANT (1);
766		1098
767	if (*dec->cur == (0xe0 | 31))	1099	if ((*dec->cur - major) > LENGTH_EXT8)
		1100	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
768	{	1101	{
769	++dec->cur;	1102	++dec->cur;
770	break;	1103	break;
771	}	1104	}
		1105	else
		1106	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
772		1107
773	sv_catsv (sv, decode_sv (dec));	1108	STRLEN len = decode_uint (dec);
		1109
		1110	WANT (len);
		1111	sv_catpvn (sv, dec->cur, len);
		1112	dec->cur += len;
774	}	1113	}
775	}	1114	}
776	else	1115	else
777	{	1116	{
778	STRLEN len = decode_uint (dec);	1117	STRLEN len = decode_uint (dec);
…		…
785	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1124	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
786	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1125	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
787	}	1126	}
788		1127
789	if (utf8)	1128	if (utf8)
		1129	{
		1130	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1131	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1132	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1133
790	SvUTF8_on (sv);	1134	SvUTF8_on (sv);
		1135	}
791		1136
792	return sv;	1137	return sv;
793		1138
794	fail:	1139	fail:
795	SvREFCNT_dec (sv);	1140	SvREFCNT_dec (sv);
…		…
814	sv = newRV_noinc (decode_sv (dec));	1159	sv = newRV_noinc (decode_sv (dec));
815	break;	1160	break;
816		1161
817	case CBOR_TAG_STRINGREF_NAMESPACE:	1162	case CBOR_TAG_STRINGREF_NAMESPACE:
818	{	1163	{
		1164	// do not use SAVETMPS/FREETMPS, as these will
		1165	// erase mortalised caches, e.g. "shareable"
819	ENTER; SAVETMPS;	1166	ENTER;
820		1167
821	SAVESPTR (dec->stringref);	1168	SAVESPTR (dec->stringref);
822	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1169	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
823		1170
824	sv = decode_sv (dec);	1171	sv = decode_sv (dec);
825		1172
826	FREETMPS; LEAVE;	1173	LEAVE;
827	}	1174	}
828	break;	1175	break;
829		1176
830	case CBOR_TAG_STRINGREF:	1177	case CBOR_TAG_STRINGREF:
831	{	1178	{
832	if ((*dec->cur >> 5) != 0)	1179	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
833	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1180	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
834		1181
835	UV idx = decode_uint (dec);	1182	UV idx = decode_uint (dec);
836		1183
837	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1184	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
838	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1185	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
839		1186
840	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1187	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
841	}	1188	}
842	break;	1189	break;
…		…
844	case CBOR_TAG_VALUE_SHAREABLE:	1191	case CBOR_TAG_VALUE_SHAREABLE:
845	{	1192	{
846	if (ecb_expect_false (!dec->shareable))	1193	if (ecb_expect_false (!dec->shareable))
847	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1194	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
848		1195
		1196	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1197	{
849	sv = newSV (0);	1198	sv = newSV (0);
850	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1199	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
851		1200
852	SV *osv = decode_sv (dec);	1201	SV *osv = decode_sv (dec);
853	sv_setsv (sv, osv);	1202	sv_setsv (sv, osv);
854	SvREFCNT_dec_NN (osv);	1203	SvREFCNT_dec_NN (osv);
		1204	}
		1205	else
		1206	{
		1207	av_push (dec->shareable, &PL_sv_undef);
		1208	int idx = AvFILLp (dec->shareable);
		1209	sv = decode_sv (dec);
		1210	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1211	}
855	}	1212	}
856	break;	1213	break;
857		1214
858	case CBOR_TAG_VALUE_SHAREDREF:	1215	case CBOR_TAG_VALUE_SHAREDREF:
859	{	1216	{
860	if ((*dec->cur >> 5) != 0)	1217	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
861	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1218	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
862		1219
863	UV idx = decode_uint (dec);	1220	UV idx = decode_uint (dec);
864		1221
865	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1222	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
866	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1223	ERR ("corrupted CBOR data (sharedref index out of bounds)");
867		1224
868	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1225	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1226
		1227	if (sv == &PL_sv_undef)
		1228	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
869	}	1229	}
870	break;	1230	break;
871		1231
872	case CBOR_TAG_PERL_OBJECT:	1232	case CBOR_TAG_PERL_OBJECT:
873	{	1233	{
		1234	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1235	goto filter;
		1236
874	sv = decode_sv (dec);	1237	sv = decode_sv (dec);
875		1238
876	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1239	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
877	ERR ("corrupted CBOR data (non-array perl object)");	1240	ERR ("corrupted CBOR data (non-array perl object)");
878		1241
…		…
888	if (!method)	1251	if (!method)
889	ERR ("cannot decode perl-object (package does not have a THAW method)");	1252	ERR ("cannot decode perl-object (package does not have a THAW method)");
890		1253
891	dSP;	1254	dSP;
892		1255
893	ENTER; SAVETMPS; PUSHMARK (SP);	1256	ENTER; SAVETMPS;
		1257	PUSHMARK (SP);
894	EXTEND (SP, len + 1);	1258	EXTEND (SP, len + 1);
895	// we re-bless the reference to get overload and other niceties right	1259	// we re-bless the reference to get overload and other niceties right
896	PUSHs (*av_fetch (av, 0, 1));	1260	PUSHs (*av_fetch (av, 0, 1));
897	PUSHs (sv_cbor);	1261	PUSHs (sv_cbor);
898		1262
…		…
906	SPAGAIN;	1270	SPAGAIN;
907		1271
908	if (SvTRUE (ERRSV))	1272	if (SvTRUE (ERRSV))
909	{	1273	{
910	FREETMPS; LEAVE;	1274	FREETMPS; LEAVE;
911	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1275	ERR_ERRSV;
912	}	1276	}
913		1277
914	SvREFCNT_dec (sv);	1278	SvREFCNT_dec_NN (sv);
915	sv = SvREFCNT_inc (POPs);	1279	sv = SvREFCNT_inc (POPs);
916		1280
917	PUTBACK;	1281	PUTBACK;
918		1282
919	FREETMPS; LEAVE;	1283	FREETMPS; LEAVE;
920	}	1284	}
921	break;	1285	break;
922		1286
923	default:	1287	default:
		1288	filter:
924	{	1289	{
		1290	SV *tag_sv = newSVuv (tag);
		1291
925	sv = decode_sv (dec);	1292	sv = decode_sv (dec);
926		1293
927	dSP;	1294	dSP;
928	ENTER; SAVETMPS; PUSHMARK (SP);	1295	ENTER; SAVETMPS;
		1296	PUSHMARK (SP);
929	EXTEND (SP, 2);	1297	EXTEND (SP, 2);
930	PUSHs (newSVuv (tag));	1298	PUSHs (tag_sv);
931	PUSHs (sv);	1299	PUSHs (sv);
932		1300
933	PUTBACK;	1301	PUTBACK;
934	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1302	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
935	SPAGAIN;	1303	SPAGAIN;
936		1304
937	if (SvTRUE (ERRSV))	1305	if (SvTRUE (ERRSV))
938	{	1306	{
		1307	SvREFCNT_dec_NN (tag_sv);
939	FREETMPS; LEAVE;	1308	FREETMPS; LEAVE;
940	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1309	ERR_ERRSV;
941	}	1310	}
942		1311
943	if (count)	1312	if (count)
944	{	1313	{
		1314	SvREFCNT_dec_NN (tag_sv);
945	SvREFCNT_dec (sv);	1315	SvREFCNT_dec_NN (sv);
946	sv = SvREFCNT_inc (POPs);	1316	sv = SvREFCNT_inc_NN (TOPs);
		1317	SP -= count;
947	}	1318	}
948	else	1319	else
949	{	1320	{
950	AV *av = newAV ();	1321	AV *av = newAV ();
951	av_push (av, newSVuv (tag));	1322	av_push (av, tag_sv);
952	av_push (av, sv);	1323	av_push (av, sv);
953		1324
954	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1325	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
955	? cbor_tagged_stash	1326	? cbor_tagged_stash
956	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1327	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
974	static SV *	1345	static SV *
975	decode_sv (dec_t *dec)	1346	decode_sv (dec_t *dec)
976	{	1347	{
977	WANT (1);	1348	WANT (1);
978		1349
979	switch (*dec->cur >> 5)	1350	switch (*dec->cur >> MAJOR_SHIFT)
980	{	1351	{
981	case 0: // unsigned int	1352	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
982	return newSVuv (decode_uint (dec));	1353	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
983	case 1: // negative int	1354	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
984	return newSViv (-1 - (IV)decode_uint (dec));	1355	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
985	case 2: // octet string	1356	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
986	return decode_str (dec, 0);	1357	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
987	case 3: // utf-8 string	1358	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
988	return decode_str (dec, 1);	1359
989	case 4: // array	1360	case MAJOR_MISC >> MAJOR_SHIFT:
990	return decode_av (dec);
991	case 5: // map
992	return decode_hv (dec);
993	case 6: // tag
994	return decode_tagged (dec);
995	case 7: // misc
996	switch (*dec->cur++ & 31)	1361	switch (*dec->cur++ & MINOR_MASK)
997	{	1362	{
998	case 20:	1363	case SIMPLE_FALSE:
999	#if CBOR_SLOW	1364	#if CBOR_SLOW
1000	types_false = get_bool ("Types::Serialiser::false");	1365	types_false = get_bool ("Types::Serialiser::false");
1001	#endif	1366	#endif
1002	return newSVsv (types_false);	1367	return newSVsv (types_false);
1003	case 21:	1368	case SIMPLE_TRUE:
1004	#if CBOR_SLOW	1369	#if CBOR_SLOW
1005	types_true = get_bool ("Types::Serialiser::true");	1370	types_true = get_bool ("Types::Serialiser::true");
1006	#endif	1371	#endif
1007	return newSVsv (types_true);	1372	return newSVsv (types_true);
1008	case 22:	1373	case SIMPLE_NULL:
1009	return newSVsv (&PL_sv_undef);	1374	return newSVsv (&PL_sv_undef);
1010	case 23:	1375	case SIMPLE_UNDEF:
1011	#if CBOR_SLOW	1376	#if CBOR_SLOW
1012	types_error = get_bool ("Types::Serialiser::error");	1377	types_error = get_bool ("Types::Serialiser::error");
1013	#endif	1378	#endif
1014	return newSVsv (types_error);	1379	return newSVsv (types_error);
1015		1380
1016	case 25:	1381	case MISC_FLOAT16:
1017	{	1382	{
1018	WANT (2);	1383	WANT (2);
1019		1384
1020	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1385	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1021	dec->cur += 2;	1386	dec->cur += 2;
1022		1387
1023	return newSVnv (ecb_binary16_to_float (fp));	1388	return newSVnv (ecb_binary16_to_float (fp));
1024	}	1389	}
1025		1390
1026	case 26:	1391	case MISC_FLOAT32:
1027	{	1392	{
1028	uint32_t fp;	1393	uint32_t fp;
1029	WANT (4);	1394	WANT (4);
1030	memcpy (&fp, dec->cur, 4);	1395	memcpy (&fp, dec->cur, 4);
1031	dec->cur += 4;	1396	dec->cur += 4;
…		…
1034	fp = ecb_bswap32 (fp);	1399	fp = ecb_bswap32 (fp);
1035		1400
1036	return newSVnv (ecb_binary32_to_float (fp));	1401	return newSVnv (ecb_binary32_to_float (fp));
1037	}	1402	}
1038		1403
1039	case 27:	1404	case MISC_FLOAT64:
1040	{	1405	{
1041	uint64_t fp;	1406	uint64_t fp;
1042	WANT (8);	1407	WANT (8);
1043	memcpy (&fp, dec->cur, 8);	1408	memcpy (&fp, dec->cur, 8);
1044	dec->cur += 8;	1409	dec->cur += 8;
…		…
1047	fp = ecb_bswap64 (fp);	1412	fp = ecb_bswap64 (fp);
1048		1413
1049	return newSVnv (ecb_binary64_to_double (fp));	1414	return newSVnv (ecb_binary64_to_double (fp));
1050	}	1415	}
1051		1416
1052	// 0..19 unassigned	1417	// 0..19 unassigned simple
1053	// 24 reserved + unassigned (reserved values are not encodable)	1418	// 24 reserved + unassigned simple (reserved values are not encodable)
		1419	// 28-30 unassigned misc
		1420	// 31 break code
1054	default:	1421	default:
1055	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1422	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1056	}	1423	}
1057		1424
1058	break;	1425	break;
1059	}	1426	}
1060		1427
…		…
1063	}	1430	}
1064		1431
1065	static SV *	1432	static SV *
1066	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1433	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1067	{	1434	{
1068	dec_t dec = { };	1435	dec_t dec = { 0 };
1069	SV *sv;	1436	SV *sv;
1070	STRLEN len;	1437	STRLEN len;
1071	char *data = SvPVbyte (string, len);	1438	char *data = SvPVbyte (string, len);
1072		1439
1073	if (len > cbor->max_size && cbor->max_size)	1440	if (len > cbor->max_size && cbor->max_size)
…		…
1087	if (dec.cur != dec.end && !dec.err)	1454	if (dec.cur != dec.end && !dec.err)
1088	dec.err = "garbage after CBOR object";	1455	dec.err = "garbage after CBOR object";
1089		1456
1090	if (dec.err)	1457	if (dec.err)
1091	{	1458	{
		1459	if (dec.shareable)
		1460	{
		1461	// need to break cyclic links, which would all be in shareable
		1462	int i;
		1463	SV **svp;
		1464
		1465	for (i = av_len (dec.shareable) + 1; i--; )
		1466	if ((svp = av_fetch (dec.shareable, i, 0)))
		1467	sv_setsv (*svp, &PL_sv_undef);
		1468	}
		1469
1092	SvREFCNT_dec (sv);	1470	SvREFCNT_dec_NN (sv);
		1471
		1472	if (dec.err_sv)
		1473	sv_2mortal (dec.err_sv);
		1474
1093	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1475	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1094	}	1476	}
1095		1477
1096	sv = sv_2mortal (sv);	1478	sv = sv_2mortal (sv);
1097		1479
1098	return sv;	1480	return sv;
1099	}	1481	}
1100		1482
		1483	/////////////////////////////////////////////////////////////////////////////
		1484	// incremental parser
		1485
		1486	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1487
		1488	// returns 0 for notyet, 1 for success or error
		1489	static int
		1490	incr_parse (CBOR *self, SV *cborstr)
		1491	{
		1492	STRLEN cur;
		1493	SvPV (cborstr, cur);
		1494
		1495	while (ecb_expect_true (self->incr_need <= cur))
		1496	{
		1497	// table of integer count bytes
		1498	static I8 incr_len[MINOR_MASK + 1] = {
		1499	0, 0, 0, 0, 0, 0, 0, 0,
		1500	0, 0, 0, 0, 0, 0, 0, 0,
		1501	0, 0, 0, 0, 0, 0, 0, 0,
		1502	1, 2, 4, 8,-1,-1,-1,-2
		1503	};
		1504
		1505	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1506	U8 m = *p & MINOR_MASK;
		1507	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1508	I8 ilen = incr_len[m];
		1509
		1510	self->incr_need = self->incr_pos + 1;
		1511
		1512	if (ecb_expect_false (ilen < 0))
		1513	{
		1514	if (m != MINOR_INDEF)
		1515	return 1; // error
		1516
		1517	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1518	{
		1519	if (count >= 0)
		1520	return 1; // error
		1521
		1522	count = 1;
		1523	}
		1524	else
		1525	{
		1526	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1527	count = -1;
		1528	}
		1529	}
		1530	else
		1531	{
		1532	self->incr_need += ilen;
		1533	if (ecb_expect_false (self->incr_need > cur))
		1534	return 0;
		1535
		1536	int major = *p >> MAJOR_SHIFT;
		1537
		1538	switch (major)
		1539	{
		1540	case MAJOR_TAG >> MAJOR_SHIFT:
		1541	++count; // tags merely prefix another value
		1542	break;
		1543
		1544	case MAJOR_BYTES >> MAJOR_SHIFT:
		1545	case MAJOR_TEXT >> MAJOR_SHIFT:
		1546	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1547	case MAJOR_MAP >> MAJOR_SHIFT:
		1548	{
		1549	UV len;
		1550
		1551	if (ecb_expect_false (ilen))
		1552	{
		1553	len = 0;
		1554
		1555	do {
		1556	len = (len << 8) | *++p;
		1557	} while (--ilen);
		1558	}
		1559	else
		1560	len = m;
		1561
		1562	switch (major)
		1563	{
		1564	case MAJOR_BYTES >> MAJOR_SHIFT:
		1565	case MAJOR_TEXT >> MAJOR_SHIFT:
		1566	self->incr_need += len;
		1567	if (ecb_expect_false (self->incr_need > cur))
		1568	return 0;
		1569
		1570	break;
		1571
		1572	case MAJOR_MAP >> MAJOR_SHIFT:
		1573	len <<= 1;
		1574	/* FALLTHROUGH */
		1575	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1576	if (len)
		1577	{
		1578	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1579	count = len + 1;
		1580	}
		1581	break;
		1582	}
		1583	}
		1584	}
		1585	}
		1586
		1587	self->incr_pos = self->incr_need;
		1588
		1589	if (count > 0)
		1590	{
		1591	while (!--count)
		1592	{
		1593	if (!AvFILLp (self->incr_count))
		1594	return 1; // done
		1595
		1596	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1597	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1598	}
		1599
		1600	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1601	}
		1602	}
		1603
		1604	return 0;
		1605	}
		1606
		1607
1101	/////////////////////////////////////////////////////////////////////////////	1608	/////////////////////////////////////////////////////////////////////////////
1102	// XS interface functions	1609	// XS interface functions
1103		1610
1104	MODULE = CBOR::XS PACKAGE = CBOR::XS	1611	MODULE = CBOR::XS PACKAGE = CBOR::XS
1105		1612
…		…
1117		1624
1118	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1625	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1119		1626
1120	sv_cbor = newSVpv ("CBOR", 0);	1627	sv_cbor = newSVpv ("CBOR", 0);
1121	SvREADONLY_on (sv_cbor);	1628	SvREADONLY_on (sv_cbor);
		1629
		1630	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1122	}	1631	}
1123		1632
1124	PROTOTYPES: DISABLE	1633	PROTOTYPES: DISABLE
1125		1634
1126	void CLONE (...)	1635	void CLONE (...)
…		…
1145	void shrink (CBOR *self, int enable = 1)	1654	void shrink (CBOR *self, int enable = 1)
1146	ALIAS:	1655	ALIAS:
1147	shrink = F_SHRINK	1656	shrink = F_SHRINK
1148	allow_unknown = F_ALLOW_UNKNOWN	1657	allow_unknown = F_ALLOW_UNKNOWN
1149	allow_sharing = F_ALLOW_SHARING	1658	allow_sharing = F_ALLOW_SHARING
1150	allow_stringref = F_ALLOW_STRINGREF	1659	allow_cycles = F_ALLOW_CYCLES
		1660	forbid_objects = F_FORBID_OBJECTS
		1661	pack_strings = F_PACK_STRINGS
		1662	text_keys = F_TEXT_KEYS
		1663	text_strings = F_TEXT_STRINGS
		1664	validate_utf8 = F_VALIDATE_UTF8
1151	PPCODE:	1665	PPCODE:
1152	{	1666	{
1153	if (enable)	1667	if (enable)
1154	self->flags |= ix;	1668	self->flags |= ix;
1155	else	1669	else
…		…
1161	void get_shrink (CBOR *self)	1675	void get_shrink (CBOR *self)
1162	ALIAS:	1676	ALIAS:
1163	get_shrink = F_SHRINK	1677	get_shrink = F_SHRINK
1164	get_allow_unknown = F_ALLOW_UNKNOWN	1678	get_allow_unknown = F_ALLOW_UNKNOWN
1165	get_allow_sharing = F_ALLOW_SHARING	1679	get_allow_sharing = F_ALLOW_SHARING
1166	get_allow_stringref = F_ALLOW_STRINGREF	1680	get_allow_cycles = F_ALLOW_CYCLES
		1681	get_forbid_objects = F_FORBID_OBJECTS
		1682	get_pack_strings = F_PACK_STRINGS
		1683	get_text_keys = F_TEXT_KEYS
		1684	get_text_strings = F_TEXT_STRINGS
		1685	get_validate_utf8 = F_VALIDATE_UTF8
1167	PPCODE:	1686	PPCODE:
1168	XPUSHs (boolSV (self->flags & ix));	1687	XPUSHs (boolSV (self->flags & ix));
1169		1688
1170	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1689	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1171	PPCODE:	1690	PPCODE:
…		…
1220	EXTEND (SP, 2);	1739	EXTEND (SP, 2);
1221	PUSHs (sv);	1740	PUSHs (sv);
1222	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1741	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1223	}	1742	}
1224		1743
		1744	void incr_parse (CBOR *self, SV *cborstr)
		1745	ALIAS:
		1746	incr_parse_multiple = 1
		1747	PPCODE:
		1748	{
		1749	if (SvUTF8 (cborstr))
		1750	sv_utf8_downgrade (cborstr, 0);
		1751
		1752	if (!self->incr_count)
		1753	{
		1754	self->incr_count = newAV ();
		1755	self->incr_pos = 0;
		1756	self->incr_need = 1;
		1757
		1758	av_push (self->incr_count, newSViv (1));
		1759	}
		1760
		1761	do
		1762	{
		1763	if (!incr_parse (self, cborstr))
		1764	{
		1765	if (self->incr_need > self->max_size && self->max_size)
		1766	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1767	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1768
		1769	break;
		1770	}
		1771
		1772	SV *sv;
		1773	char *offset;
		1774
		1775	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1776	XPUSHs (sv);
		1777
		1778	sv_chop (cborstr, offset);
		1779
		1780	av_clear (self->incr_count);
		1781	av_push (self->incr_count, newSViv (1));
		1782
		1783	self->incr_pos = 0;
		1784	self->incr_need = self->incr_pos + 1;
		1785	}
		1786	while (ix);
		1787	}
		1788
		1789	void incr_reset (CBOR *self)
		1790	CODE:
		1791	{
		1792	SvREFCNT_dec (self->incr_count);
		1793	self->incr_count = 0;
		1794	}
		1795
1225	void DESTROY (CBOR *self)	1796	void DESTROY (CBOR *self)
1226	PPCODE:	1797	PPCODE:
1227	cbor_free (self);	1798	cbor_free (self);
1228		1799
1229	PROTOTYPES: ENABLE	1800	PROTOTYPES: ENABLE
1230		1801
1231	void encode_cbor (SV *scalar)	1802	void encode_cbor (SV *scalar)
		1803	ALIAS:
		1804	encode_cbor = 0
		1805	encode_cbor_sharing = F_ALLOW_SHARING
1232	PPCODE:	1806	PPCODE:
1233	{	1807	{
1234	CBOR cbor;	1808	CBOR cbor;
1235	cbor_init (&cbor);	1809	cbor_init (&cbor);
		1810	cbor.flags |= ix;
1236	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1811	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1237	XPUSHs (scalar);	1812	XPUSHs (scalar);
1238	}	1813	}
1239		1814
1240	void decode_cbor (SV *cborstr)	1815	void decode_cbor (SV *cborstr)
…		…
1244	cbor_init (&cbor);	1819	cbor_init (&cbor);
1245	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1820	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1246	XPUSHs (cborstr);	1821	XPUSHs (cborstr);
1247	}	1822	}
1248		1823
		1824	#ifdef __AFL_COMPILER
		1825
		1826	void
		1827	afl_init ()
		1828	CODE:
		1829	__AFL_INIT ();
		1830
		1831	int
		1832	afl_loop (unsigned int count = 10000)
		1833	CODE:
		1834	RETVAL = __AFL_LOOP (count);
		1835	OUTPUT:
		1836	RETVAL
		1837
		1838	#endif
		1839

Diff Legend

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