ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.71 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
		29	#ifndef SvREFCNT_dec_NN
		30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		31	#endif
		32
		33	// known major and minor types
		34	enum cbor_type
		35	{
		36	MAJOR_SHIFT = 5,
		37	MINOR_MASK = 0x1f,
		38
		39	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		40	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		41	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		42	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		43	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		44	MAJOR_MAP = 5 << MAJOR_SHIFT,
		45	MAJOR_TAG = 6 << MAJOR_SHIFT,
		46	MAJOR_MISC = 7 << MAJOR_SHIFT,
		47
		48	// INT/STRING/ARRAY/MAP subtypes
		49	LENGTH_EXT1 = 24,
		50	LENGTH_EXT2 = 25,
		51	LENGTH_EXT4 = 26,
		52	LENGTH_EXT8 = 27,
		53
		54	// SIMPLE types (effectively MISC subtypes)
		55	SIMPLE_FALSE = 20,
		56	SIMPLE_TRUE = 21,
		57	SIMPLE_NULL = 22,
		58	SIMPLE_UNDEF = 23,
		59
		60	// MISC subtype (unused)
		61	MISC_EXT1 = 24,
		62	MISC_FLOAT16 = 25,
		63	MISC_FLOAT32 = 26,
		64	MISC_FLOAT64 = 27,
		65
		66	// BYTES/TEXT/ARRAY/MAP
		67	MINOR_INDEF = 31,
		68	};
24		69
25	// known tags	70	// known tags
26	enum cbor_tag	71	enum cbor_tag
27	{	72	{
28	// inofficial extensions (pending iana registration)	73	// extensions
		74	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	75	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	76	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	77	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	78	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	79	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	80	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36		81
37	// rfc7049	82	// rfc7049
38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	83	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	84	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40	CBOR_TAG_POS_BIGNUM = 2, // byte string	85	CBOR_TAG_POS_BIGNUM = 2, // byte string
41	CBOR_TAG_NEG_BIGNUM = 3, // byte string	86	CBOR_TAG_NEG_BIGNUM = 3, // byte string
42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	87	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43	CBOR_TAG_BIGFLOAT = 5, // array	88	CBOR_TAG_BIGFLOAT = 5, // array
44		89
45	CBOR_TAG_CONV_B64U = 21, // base64url, any	90	CBOR_TAG_CONV_B64U = 21, // base64url, any
46	CBOR_TAG_CONV_B64 = 22, // base64, any	91	CBOR_TAG_CONV_B64 = 22, // base64, any
47	CBOR_TAG_CONV_HEX = 23, // base16, any	92	CBOR_TAG_CONV_HEX = 23, // base16, any
48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	93	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49		94
50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	95	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	96	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	97	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	98	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55		100
56	CBOR_TAG_MAGIC = 55799 // self-describe cbor	101	CBOR_TAG_MAGIC = 55799, // self-describe cbor
		102	};
		103
		104	// known forced types, also hardcoded in CBOR.pm
		105	enum
		106	{
		107	AS_CBOR = 0,
		108	AS_INT = 1,
		109	AS_BYTES = 2,
		110	AS_TEXT = 3,
		111	AS_FLOAT16 = 4,
		112	AS_FLOAT32 = 5,
		113	AS_FLOAT64 = 6,
		114	AS_MAP = 7,
		115	// possibly future enhancements: (generic) float, (generic) string
57	};	116	};
58		117
59	#define F_SHRINK 0x00000001UL	118	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	119	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	120	#define F_ALLOW_SHARING 0x00000004UL
62	#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
63		127
64	#define INIT_SIZE 32 // initial scalar size to be allocated	128	#define INIT_SIZE 32 // initial scalar size to be allocated
65		129
66	#define SB do {	130	#define SB do {
67	#define SE } while (0)	131	#define SE } while (0)
…		…
86	typedef struct {	150	typedef struct {
87	U32 flags;	151	U32 flags;
88	U32 max_depth;	152	U32 max_depth;
89	STRLEN max_size;	153	STRLEN max_size;
90	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.
91	} CBOR;	160	} CBOR;
92		161
93	ecb_inline void	162	ecb_inline void
94	cbor_init (CBOR *cbor)	163	cbor_init (CBOR *cbor)
95	{	164	{
…		…
99		168
100	ecb_inline void	169	ecb_inline void
101	cbor_free (CBOR *cbor)	170	cbor_free (CBOR *cbor)
102	{	171	{
103	SvREFCNT_dec (cbor->filter);	172	SvREFCNT_dec (cbor->filter);
		173	SvREFCNT_dec (cbor->incr_count);
104	}	174	}
105		175
106	/////////////////////////////////////////////////////////////////////////////	176	/////////////////////////////////////////////////////////////////////////////
107	// utility functions	177	// utility functions
108		178
…		…
131	#endif	201	#endif
132	}	202	}
133	}	203	}
134		204
135	// minimum length of a string to be registered for stringref	205	// minimum length of a string to be registered for stringref
136	ecb_inline int	206	ecb_inline STRLEN
137	minimum_string_length (UV idx)	207	minimum_string_length (UV idx)
138	{	208	{
139	return idx > 23	209	return idx <= 23 ? 3
140	? idx > 0xffU	210	: idx <= 0xffU ? 4
141	? idx > 0xffffU	211	: idx <= 0xffffU ? 5
142	? idx > 0xffffffffU	212	: idx <= 0xffffffffU ? 7
143	? 7	213	: 11;
144	: 6
145	: 5
146	: 4
147	: 3;
148	}	214	}
149		215
150	/////////////////////////////////////////////////////////////////////////////	216	/////////////////////////////////////////////////////////////////////////////
151	// encoder	217	// encoder
152		218
…		…
165	} enc_t;	231	} enc_t;
166		232
167	ecb_inline void	233	ecb_inline void
168	need (enc_t *enc, STRLEN len)	234	need (enc_t *enc, STRLEN len)
169	{	235	{
170	if (ecb_expect_false (enc->cur + len >= enc->end))	236	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
171	{	237	{
172	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	238	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
173	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	239	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
174	enc->cur = SvPVX (enc->sv) + cur;	240	enc->cur = SvPVX (enc->sv) + cur;
175	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	241	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
176	}	242	}
177	}	243	}
178		244
		245	static void encode_sv (enc_t *enc, SV *sv);
		246
179	ecb_inline void	247	ecb_inline void
180	encode_ch (enc_t *enc, char ch)	248	encode_ch (enc_t *enc, char ch)
181	{	249	{
182	need (enc, 1);	250	need (enc, 1);
183	*enc->cur++ = ch;	251	*enc->cur++ = ch;
184	}	252	}
185		253
		254	// used for tags, intregers, element counts and so on
186	static void	255	static void
187	encode_uint (enc_t *enc, int major, UV len)	256	encode_uint (enc_t *enc, int major, UV len)
188	{	257	{
189	need (enc, 9);	258	need (enc, 9);
190		259
191	if (len < 24)	260	if (ecb_expect_true (len < LENGTH_EXT1))
192	*enc->cur++ = major | len;	261	*enc->cur++ = major | len;
193	else if (len <= 0xff)	262	else if (ecb_expect_true (len <= 0xffU))
194	{	263	{
195	*enc->cur++ = major | 24;	264	*enc->cur++ = major | LENGTH_EXT1;
196	*enc->cur++ = len;	265	*enc->cur++ = len;
197	}	266	}
198	else if (len <= 0xffff)	267	else if (len <= 0xffffU)
199	{	268	{
200	*enc->cur++ = major | 25;	269	*enc->cur++ = major | LENGTH_EXT2;
201	*enc->cur++ = len >> 8;	270	*enc->cur++ = len >> 8;
202	*enc->cur++ = len;	271	*enc->cur++ = len;
203	}	272	}
204	else if (len <= 0xffffffff)	273	else if (len <= 0xffffffffU)
205	{	274	{
206	*enc->cur++ = major | 26;	275	*enc->cur++ = major | LENGTH_EXT4;
207	*enc->cur++ = len >> 24;	276	*enc->cur++ = len >> 24;
208	*enc->cur++ = len >> 16;	277	*enc->cur++ = len >> 16;
209	*enc->cur++ = len >> 8;	278	*enc->cur++ = len >> 8;
210	*enc->cur++ = len;	279	*enc->cur++ = len;
211	}	280	}
212	else	281	else
213	{	282	{
214	*enc->cur++ = major | 27;	283	*enc->cur++ = major | LENGTH_EXT8;
215	*enc->cur++ = len >> 56;	284	*enc->cur++ = len >> 56;
216	*enc->cur++ = len >> 48;	285	*enc->cur++ = len >> 48;
217	*enc->cur++ = len >> 40;	286	*enc->cur++ = len >> 40;
218	*enc->cur++ = len >> 32;	287	*enc->cur++ = len >> 32;
219	*enc->cur++ = len >> 24;	288	*enc->cur++ = len >> 24;
…		…
221	*enc->cur++ = len >> 8;	290	*enc->cur++ = len >> 8;
222	*enc->cur++ = len;	291	*enc->cur++ = len;
223	}	292	}
224	}	293	}
225		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
226	ecb_inline void	307	ecb_inline void
227	encode_tag (enc_t *enc, UV tag)	308	encode_tag (enc_t *enc, UV tag)
228	{	309	{
229	encode_uint (enc, 0xc0, tag);	310	encode_uint (enc, MAJOR_TAG, tag);
230	}	311	}
231		312
232	static void	313	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		314	ecb_noinline static void
233	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	315	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
234	{	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	}
		334	}
		335
		336	ecb_inline void
		337	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		338	{
		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);
		347	need (enc, len);
		348	memcpy (enc->cur, str, len);
		349	enc->cur += len;
		350	}
		351
		352	ecb_inline void
		353	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		354	{
235	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	355	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236	{	356	{
237	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	357	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238		358
239	if (SvOK (*svp))	359	if (SvOK (*svp))
240	{	360	{
241	// already registered, use stringref	361	// already registered, use stringref
242	encode_tag (enc, CBOR_TAG_STRINGREF);	362	encode_tag (enc, CBOR_TAG_STRINGREF);
243	encode_uint (enc, 0x00, SvUV (*svp));	363	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244	return;	364	return;
245	}	365	}
246	else if (len >= minimum_string_length (enc->stringref_idx))	366	else if (len >= minimum_string_length (enc->stringref_idx))
247	{	367	{
248	// register only	368	// register only
249	sv_setuv (*svp, enc->stringref_idx);	369	sv_setuv (*svp, enc->stringref_idx);
250	++enc->stringref_idx;	370	++enc->stringref_idx;
251	}	371	}
252	}	372	}
253		373
254	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	374	encode_str (enc, upgrade_utf8, utf8, str, len);
		375	}
		376
		377	ecb_inline void
		378	encode_float16 (enc_t *enc, NV nv)
		379	{
255	need (enc, len);	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
256	memcpy (enc->cur, str, len);	389	memcpy (enc->cur, &fp, 2);
257	enc->cur += len;	390	enc->cur += 2;
258	}	391	}
259		392
260	static void encode_sv (enc_t *enc, SV *sv);	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	}
261		507
262	static void	508	static void
263	encode_av (enc_t *enc, AV *av)	509	encode_av (enc_t *enc, AV *av)
264	{	510	{
265	int i, len = av_len (av);	511	int i, len = av_len (av);
…		…
267	if (enc->depth >= enc->cbor.max_depth)	513	if (enc->depth >= enc->cbor.max_depth)
268	croak (ERR_NESTING_EXCEEDED);	514	croak (ERR_NESTING_EXCEEDED);
269		515
270	++enc->depth;	516	++enc->depth;
271		517
272	encode_uint (enc, 0x80, len + 1);	518	encode_uint (enc, MAJOR_ARRAY, len + 1);
273		519
		520	if (ecb_expect_false (SvMAGICAL (av)))
274	for (i = 0; i <= len; ++i)	521	for (i = 0; i <= len; ++i)
275	{	522	{
276	SV **svp = av_fetch (av, i, 0);	523	SV **svp = av_fetch (av, i, 0);
277	encode_sv (enc, svp ? *svp : &PL_sv_undef);	524	encode_sv (enc, svp ? *svp : &PL_sv_undef);
278	}	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	}
279		532
280	--enc->depth;	533	--enc->depth;
281	}	534	}
282		535
283	static void	536	static void
…		…
291	++enc->depth;	544	++enc->depth;
292		545
293	int pairs = hv_iterinit (hv);	546	int pairs = hv_iterinit (hv);
294	int mg = SvMAGICAL (hv);	547	int mg = SvMAGICAL (hv);
295		548
296	if (mg)	549	if (ecb_expect_false (mg))
297	encode_ch (enc, 0xa0 | 31);	550	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
298	else	551	else
299	encode_uint (enc, 0xa0, pairs);	552	encode_uint (enc, MAJOR_MAP, pairs);
300		553
301	while ((he = hv_iternext (hv)))	554	while ((he = hv_iternext (hv)))
302	{	555	{
303	if (HeKLEN (he) == HEf_SVKEY)	556	if (HeKLEN (he) == HEf_SVKEY)
304	encode_sv (enc, HeSVKEY (he));	557	encode_sv (enc, HeSVKEY (he));
305	else	558	else
306	encode_str (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));
307		560
308	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));
309	}	562	}
310		563
311	if (mg)	564	if (ecb_expect_false (mg))
312	encode_ch (enc, 0xe0 | 31);	565	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
313		566
314	--enc->depth;	567	--enc->depth;
315	}	568	}
316		569
317	// encode objects, arrays and special \0=false and \1=true values.	570	// encode objects, arrays and special \0=false and \1=true values.
318	static void	571	static void
319	encode_rv (enc_t *enc, SV *sv)	572	encode_rv (enc_t *enc, SV *sv)
320	{	573	{
321	SvGETMAGIC (sv);	574	SvGETMAGIC (sv);
322
323	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324	&& ecb_expect_false (SvREFCNT (sv) > 1))
325	{
326	if (!enc->shareable)
327	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
328
329	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
330
331	if (SvOK (*svp))
332	{
333	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334	encode_uint (enc, 0x00, SvUV (*svp));
335	return;
336	}
337	else
338	{
339	sv_setuv (*svp, enc->shareable_idx);
340	++enc->shareable_idx;
341	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342	}
343	}
344		575
345	svtype svt = SvTYPE (sv);	576	svtype svt = SvTYPE (sv);
346		577
347	if (ecb_expect_false (SvOBJECT (sv)))	578	if (ecb_expect_false (SvOBJECT (sv)))
348	{	579	{
…		…
355	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	586	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
356	? cbor_tagged_stash	587	? cbor_tagged_stash
357	: gv_stashpv ("CBOR::XS::Tagged" , 1);	588	: gv_stashpv ("CBOR::XS::Tagged" , 1);
358		589
359	HV *stash = SvSTASH (sv);	590	HV *stash = SvSTASH (sv);
360	GV *method;
361		591
362	if (stash == boolean_stash)	592	if (stash == boolean_stash)
363	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	593	{
		594	encode_bool (enc, SvIV (sv));
		595	return;
		596	}
364	else if (stash == error_stash)	597	else if (stash == error_stash)
365	encode_ch (enc, 0xe0 | 23);	598	{
		599	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		600	return;
		601	}
366	else if (stash == tagged_stash)	602	else if (stash == tagged_stash)
367	{	603	{
368	if (svt != SVt_PVAV)	604	if (svt != SVt_PVAV)
369	croak ("encountered CBOR::XS::Tagged object that isn't an array");	605	croak ("encountered CBOR::XS::Tagged object that isn't an array");
370		606
		607	switch (av_len ((AV *)sv))
		608	{
		609	case 2-1:
		610	// actually a tagged value
371	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	611	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
372	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))
373	}	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))));
374	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	658	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375	{	659	{
376	dSP;	660	dSP;
377		661
378	ENTER; SAVETMPS; PUSHMARK (SP);	662	ENTER; SAVETMPS;
		663	PUSHMARK (SP);
379	// 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
380	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	665	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381		666
382	PUTBACK;	667	PUTBACK;
383	// G_SCALAR ensures that return value is 1	668	// G_SCALAR ensures that return value is 1
…		…
396	}	681	}
397	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	682	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398	{	683	{
399	dSP;	684	dSP;
400		685
401	ENTER; SAVETMPS; PUSHMARK (SP);	686	ENTER; SAVETMPS;
		687	PUSHMARK (SP);
402	EXTEND (SP, 2);	688	EXTEND (SP, 2);
403	// 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
404	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	690	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405	PUSHs (sv_cbor);	691	PUSHs (sv_cbor);
406		692
…		…
411	// catch this surprisingly common error	697	// catch this surprisingly common error
412	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	698	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	699	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414		700
415	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	701	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416	encode_uint (enc, 0x80, count + 1);	702	encode_uint (enc, MAJOR_ARRAY, count + 1);
417	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	703	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418		704
419	while (count)	705	{
		706	int i;
		707
		708	for (i = 0; i < count; ++i)
420	encode_sv (enc, SP[1 - count--]);	709	encode_sv (enc, SP[i + 1 - count]);
		710
		711	SP -= count;
		712	}
421		713
422	PUTBACK;	714	PUTBACK;
423		715
424	FREETMPS; LEAVE;	716	FREETMPS; LEAVE;
425	}	717	}
…		…
443	{	735	{
444	double nv = SvNVX (sv);	736	double nv = SvNVX (sv);
445		737
446	need (enc, 9);	738	need (enc, 9);
447		739
448	if (ecb_expect_false (nv == (U32)nv))	740	if (ecb_expect_false (nv == (NV)(U32)nv))
449	encode_uint (enc, 0x00, (U32)nv);	741	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450	//TODO: maybe I32?	742	//TODO: maybe I32?
451	else if (ecb_expect_false (nv == (float)nv))	743	else if (ecb_expect_false (nv == (float)nv))
452	{	744	encode_float32 (enc, nv);
453	uint32_t fp = ecb_float_to_binary32 (nv);
454
455	*enc->cur++ = 0xe0 | 26;
456
457	if (!ecb_big_endian ())
458	fp = ecb_bswap32 (fp);
459
460	memcpy (enc->cur, &fp, 4);
461	enc->cur += 4;
462	}
463	else	745	else
464	{	746	encode_float64 (enc, nv);
465	uint64_t fp = ecb_double_to_binary64 (nv);
466
467	*enc->cur++ = 0xe0 | 27;
468
469	if (!ecb_big_endian ())
470	fp = ecb_bswap64 (fp);
471
472	memcpy (enc->cur, &fp, 8);
473	enc->cur += 8;
474	}
475	}	747	}
476		748
477	static void	749	static void
478	encode_sv (enc_t *enc, SV *sv)	750	encode_sv (enc_t *enc, SV *sv)
479	{	751	{
…		…
481		753
482	if (SvPOKp (sv))	754	if (SvPOKp (sv))
483	{	755	{
484	STRLEN len;	756	STRLEN len;
485	char *str = SvPV (sv, len);	757	char *str = SvPV (sv, len);
486	encode_str (enc, SvUTF8 (sv), str, len);	758	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
487	}	759	}
488	else if (SvNOKp (sv))	760	else if (SvNOKp (sv))
489	encode_nv (enc, sv);	761	encode_nv (enc, sv);
490	else if (SvIOKp (sv))	762	else if (SvIOKp (sv))
491	{	763	encode_int (enc, sv);
492	if (SvIsUV (sv))
493	encode_uint (enc, 0x00, SvUVX (sv));
494	else if (SvIVX (sv) >= 0)
495	encode_uint (enc, 0x00, SvIVX (sv));
496	else
497	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
498	}
499	else if (SvROK (sv))	764	else if (SvROK (sv))
500	encode_rv (enc, SvRV (sv));	765	encode_rv (enc, SvRV (sv));
501	else if (!SvOK (sv))	766	else if (!SvOK (sv))
502	encode_ch (enc, 0xe0 | 22);	767	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
503	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	768	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504	encode_ch (enc, 0xe0 | 23);	769	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
505	else	770	else
506	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",
507	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	772	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508	}	773	}
509		774
510	static SV *	775	static SV *
511	encode_cbor (SV *scalar, CBOR *cbor)	776	encode_cbor (SV *scalar, CBOR *cbor)
512	{	777	{
513	enc_t enc = { };	778	enc_t enc = { 0 };
514		779
515	enc.cbor = *cbor;	780	enc.cbor = *cbor;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	781	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	782	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	783	enc.end = SvEND (enc.sv);
519		784
520	SvPOK_only (enc.sv);	785	SvPOK_only (enc.sv);
521		786
522	if (cbor->flags & F_ALLOW_STRINGREF)	787	if (cbor->flags & F_PACK_STRINGS)
523	{	788	{
524	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	789	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());	790	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
526	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());	791	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
527	}	792	}
…		…
550	U32 depth; // recursion depth	815	U32 depth; // recursion depth
551	U32 maxdepth; // recursion depth limit	816	U32 maxdepth; // recursion depth limit
552	AV *shareable;	817	AV *shareable;
553	AV *stringref;	818	AV *stringref;
554	SV *decode_tagged;	819	SV *decode_tagged;
		820	SV *err_sv; // optional sv for error, needs to be freed
555	} dec_t;	821	} dec_t;
556		822
557	#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);
558		830
559	#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;
560		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
561	#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)
562	#define DEC_DEC_DEPTH --dec->depth	860	#define DEC_DEC_DEPTH --dec->depth
563		861
564	static UV	862	static UV
565	decode_uint (dec_t *dec)	863	decode_uint (dec_t *dec)
566	{	864	{
567	switch (*dec->cur & 31)	865	U8 m = *dec->cur & MINOR_MASK;
568	{	866	++dec->cur;
569	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572	return *dec->cur++ & 31;
573		867
574	case 24:	868	if (ecb_expect_true (m < LENGTH_EXT1))
		869	return m;
		870	else if (ecb_expect_true (m == LENGTH_EXT1))
		871	{
575	WANT (2);	872	WANT (1);
576	dec->cur += 2;	873	dec->cur += 1;
577	return dec->cur[-1];	874	return dec->cur[-1];
578		875	}
579	case 25:	876	else if (ecb_expect_true (m == LENGTH_EXT2))
		877	{
580	WANT (3);	878	WANT (2);
581	dec->cur += 3;	879	dec->cur += 2;
582	return (((UV)dec->cur[-2]) << 8)	880	return (((UV)dec->cur[-2]) << 8)
583	| ((UV)dec->cur[-1]);	881	| ((UV)dec->cur[-1]);
584		882	}
585	case 26:	883	else if (ecb_expect_true (m == LENGTH_EXT4))
		884	{
586	WANT (5);	885	WANT (4);
587	dec->cur += 5;	886	dec->cur += 4;
588	return (((UV)dec->cur[-4]) << 24)	887	return (((UV)dec->cur[-4]) << 24)
589	| (((UV)dec->cur[-3]) << 16)	888	| (((UV)dec->cur[-3]) << 16)
590	| (((UV)dec->cur[-2]) << 8)	889	| (((UV)dec->cur[-2]) << 8)
591	| ((UV)dec->cur[-1]);	890	| ((UV)dec->cur[-1]);
592		891	}
593	case 27:	892	else if (ecb_expect_true (m == LENGTH_EXT8))
		893	{
594	WANT (9);	894	WANT (8);
595	dec->cur += 9;	895	dec->cur += 8;
		896
		897	return
		898	#if UVSIZE < 8
		899	0
		900	#else
596	return (((UV)dec->cur[-8]) << 56)	901	(((UV)dec->cur[-8]) << 56)
597	| (((UV)dec->cur[-7]) << 48)	902	| (((UV)dec->cur[-7]) << 48)
598	| (((UV)dec->cur[-6]) << 40)	903	| (((UV)dec->cur[-6]) << 40)
599	| (((UV)dec->cur[-5]) << 32)	904	| (((UV)dec->cur[-5]) << 32)
		905	#endif
600	| (((UV)dec->cur[-4]) << 24)	906	| (((UV)dec->cur[-4]) << 24)
601	| (((UV)dec->cur[-3]) << 16)	907	| (((UV)dec->cur[-3]) << 16)
602	| (((UV)dec->cur[-2]) << 8)	908	| (((UV)dec->cur[-2]) << 8)
603	| ((UV)dec->cur[-1]);	909	| ((UV)dec->cur[-1]);
604		910	}
605	default:	911	else
606	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	912	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607	}
608		913
609	fail:	914	fail:
610	return 0;	915	return 0;
611	}	916	}
612		917
…		…
617	{	922	{
618	AV *av = newAV ();	923	AV *av = newAV ();
619		924
620	DEC_INC_DEPTH;	925	DEC_INC_DEPTH;
621		926
622	if ((*dec->cur & 31) == 31)	927	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
623	{	928	{
624	++dec->cur;	929	++dec->cur;
625		930
626	for (;;)	931	for (;;)
627	{	932	{
628	WANT (1);	933	WANT (1);
629		934
630	if (*dec->cur == (0xe0 | 31))	935	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
631	{	936	{
632	++dec->cur;	937	++dec->cur;
633	break;	938	break;
634	}	939	}
635		940
636	av_push (av, decode_sv (dec));	941	av_push (av, decode_sv (dec));
637	}	942	}
638	}	943	}
639	else	944	else
640	{	945	{
641	int i, len = decode_uint (dec);	946	UV i, len = decode_uint (dec);
642		947
		948	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643	av_fill (av, len - 1);	949	av_fill (av, len - 1);
644		950
645	for (i = 0; i < len; ++i)	951	for (i = 0; i < len; ++i)
646	AvARRAY (av)[i] = decode_sv (dec);	952	AvARRAY (av)[i] = decode_sv (dec);
647	}	953	}
648		954
649	DEC_DEC_DEPTH;	955	DEC_DEC_DEPTH;
650	return newRV_noinc ((SV *)av);	956	return newRV_noinc ((SV *)av);
651		957
652	fail:	958	fail:
653	SvREFCNT_dec (av);	959	SvREFCNT_dec_NN (av);
654	DEC_DEC_DEPTH;	960	DEC_DEC_DEPTH;
655	return &PL_sv_undef;	961	return &PL_sv_undef;
656	}	962	}
657		963
658	static void	964	static void
…		…
660	{	966	{
661	// for speed reasons, we specialcase single-string	967	// for speed reasons, we specialcase single-string
662	// byte or utf-8 strings as keys, but only when !stringref	968	// byte or utf-8 strings as keys, but only when !stringref
663		969
664	if (ecb_expect_true (!dec->stringref))	970	if (ecb_expect_true (!dec->stringref))
665	if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)	971	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666	{	972	{
667	I32 len = decode_uint (dec);	973	STRLEN len = decode_uint (dec);
668	char *key = (char *)dec->cur;	974	char *key = (char *)dec->cur;
669		975
		976	WANT (len);
670	dec->cur += len;	977	dec->cur += len;
671		978
672	if (ecb_expect_false (dec->stringref))
673	av_push (dec->stringref, newSVpvn (key, len));
674
675	hv_store (hv, key, len, decode_sv (dec), 0);	979	hv_store (hv, key, len, decode_sv (dec), 0);
676		980
677	return;	981	return;
678	}	982	}
679	else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)	983	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680	{	984	{
681	I32 len = decode_uint (dec);	985	STRLEN len = decode_uint (dec);
682	char *key = (char *)dec->cur;	986	char *key = (char *)dec->cur;
683		987
		988	WANT (len);
684	dec->cur += len;	989	dec->cur += len;
685		990
686	if (ecb_expect_false (dec->stringref))	991	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687	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)");
688		994
689	hv_store (hv, key, -len, decode_sv (dec), 0);	995	hv_store (hv, key, -len, decode_sv (dec), 0);
690		996
691	return;	997	return;
692	}	998	}
693		999
694	SV *k = decode_sv (dec);	1000	SV *k = decode_sv (dec);
695	SV *v = decode_sv (dec);	1001	SV *v = decode_sv (dec);
696		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
697	hv_store_ent (hv, k, v, 0);	1031	hv_store_ent (hv, k, v, 0);
698	SvREFCNT_dec (k);	1032	SvREFCNT_dec_NN (k);
		1033
		1034	fail:
		1035	;
699	}	1036	}
700		1037
701	static SV *	1038	static SV *
702	decode_hv (dec_t *dec)	1039	decode_hv (dec_t *dec)
703	{	1040	{
704	HV *hv = newHV ();	1041	HV *hv = newHV ();
705		1042
706	DEC_INC_DEPTH;	1043	DEC_INC_DEPTH;
707		1044
708	if ((*dec->cur & 31) == 31)	1045	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
709	{	1046	{
710	++dec->cur;	1047	++dec->cur;
711		1048
712	for (;;)	1049	for (;;)
713	{	1050	{
714	WANT (1);	1051	WANT (1);
715		1052
716	if (*dec->cur == (0xe0 | 31))	1053	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
717	{	1054	{
718	++dec->cur;	1055	++dec->cur;
719	break;	1056	break;
720	}	1057	}
721		1058
722	decode_he (dec, hv);	1059	decode_he (dec, hv);
723	}	1060	}
724	}	1061	}
725	else	1062	else
726	{	1063	{
727	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
728		1067
729	while (pairs--)	1068	while (pairs--)
730	decode_he (dec, hv);	1069	decode_he (dec, hv);
731	}	1070	}
732		1071
733	DEC_DEC_DEPTH;	1072	DEC_DEC_DEPTH;
734	return newRV_noinc ((SV *)hv);	1073	return newRV_noinc ((SV *)hv);
735		1074
736	fail:	1075	fail:
737	SvREFCNT_dec (hv);	1076	SvREFCNT_dec_NN (hv);
738	DEC_DEC_DEPTH;	1077	DEC_DEC_DEPTH;
739	return &PL_sv_undef;	1078	return &PL_sv_undef;
740	}	1079	}
741		1080
742	static SV *	1081	static SV *
743	decode_str (dec_t *dec, int utf8)	1082	decode_str (dec_t *dec, int utf8)
744	{	1083	{
745	SV *sv = 0;	1084	SV *sv = 0;
746		1085
747	if ((*dec->cur & 31) == 31)	1086	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
748	{	1087	{
		1088	// indefinite length strings
749	++dec->cur;	1089	++dec->cur;
750		1090
		1091	U8 major = *dec->cur & MAJOR_MISC;
		1092
751	sv = newSVpvn ("", 0);	1093	sv = newSVpvn ("", 0);
752		1094
753	// not very fast, and certainly not robust against illegal input
754	for (;;)	1095	for (;;)
755	{	1096	{
756	WANT (1);	1097	WANT (1);
757		1098
758	if (*dec->cur == (0xe0 | 31))	1099	if ((*dec->cur - major) > LENGTH_EXT8)
		1100	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
759	{	1101	{
760	++dec->cur;	1102	++dec->cur;
761	break;	1103	break;
762	}	1104	}
		1105	else
		1106	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763		1107
764	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;
765	}	1113	}
766	}	1114	}
767	else	1115	else
768	{	1116	{
769	STRLEN len = decode_uint (dec);	1117	STRLEN len = decode_uint (dec);
…		…
776	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1124	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1125	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778	}	1126	}
779		1127
780	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
781	SvUTF8_on (sv);	1134	SvUTF8_on (sv);
		1135	}
782		1136
783	return sv;	1137	return sv;
784		1138
785	fail:	1139	fail:
786	SvREFCNT_dec (sv);	1140	SvREFCNT_dec (sv);
…		…
805	sv = newRV_noinc (decode_sv (dec));	1159	sv = newRV_noinc (decode_sv (dec));
806	break;	1160	break;
807		1161
808	case CBOR_TAG_STRINGREF_NAMESPACE:	1162	case CBOR_TAG_STRINGREF_NAMESPACE:
809	{	1163	{
		1164	// do not use SAVETMPS/FREETMPS, as these will
		1165	// erase mortalised caches, e.g. "shareable"
810	ENTER; SAVETMPS;	1166	ENTER;
811		1167
812	SAVESPTR (dec->stringref);	1168	SAVESPTR (dec->stringref);
813	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1169	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
814		1170
815	sv = decode_sv (dec);	1171	sv = decode_sv (dec);
816		1172
817	FREETMPS; LEAVE;	1173	LEAVE;
818	}	1174	}
819	break;	1175	break;
820		1176
821	case CBOR_TAG_STRINGREF:	1177	case CBOR_TAG_STRINGREF:
822	{	1178	{
823	if ((*dec->cur >> 5) != 0)	1179	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1180	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825		1181
826	UV idx = decode_uint (dec);	1182	UV idx = decode_uint (dec);
827		1183
828	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1184	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
829	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)");
830		1186
831	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1187	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
832	}	1188	}
833	break;	1189	break;
…		…
835	case CBOR_TAG_VALUE_SHAREABLE:	1191	case CBOR_TAG_VALUE_SHAREABLE:
836	{	1192	{
837	if (ecb_expect_false (!dec->shareable))	1193	if (ecb_expect_false (!dec->shareable))
838	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1194	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
839		1195
		1196	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1197	{
840	sv = newSV (0);	1198	sv = newSV (0);
841	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1199	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842		1200
843	SV *osv = decode_sv (dec);	1201	SV *osv = decode_sv (dec);
844	sv_setsv (sv, osv);	1202	sv_setsv (sv, osv);
845	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	}
846	}	1212	}
847	break;	1213	break;
848		1214
849	case CBOR_TAG_VALUE_SHAREDREF:	1215	case CBOR_TAG_VALUE_SHAREDREF:
850	{	1216	{
851	if ((*dec->cur >> 5) != 0)	1217	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1218	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853		1219
854	UV idx = decode_uint (dec);	1220	UV idx = decode_uint (dec);
855		1221
856	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1222	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
857	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1223	ERR ("corrupted CBOR data (sharedref index out of bounds)");
858		1224
859	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");
860	}	1229	}
861	break;	1230	break;
862		1231
863	case CBOR_TAG_PERL_OBJECT:	1232	case CBOR_TAG_PERL_OBJECT:
864	{	1233	{
		1234	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1235	goto filter;
		1236
865	sv = decode_sv (dec);	1237	sv = decode_sv (dec);
866		1238
867	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1239	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
868	ERR ("corrupted CBOR data (non-array perl object)");	1240	ERR ("corrupted CBOR data (non-array perl object)");
869		1241
…		…
879	if (!method)	1251	if (!method)
880	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)");
881		1253
882	dSP;	1254	dSP;
883		1255
884	ENTER; SAVETMPS; PUSHMARK (SP);	1256	ENTER; SAVETMPS;
		1257	PUSHMARK (SP);
885	EXTEND (SP, len + 1);	1258	EXTEND (SP, len + 1);
886	// 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
887	PUSHs (*av_fetch (av, 0, 1));	1260	PUSHs (*av_fetch (av, 0, 1));
888	PUSHs (sv_cbor);	1261	PUSHs (sv_cbor);
889		1262
…		…
897	SPAGAIN;	1270	SPAGAIN;
898		1271
899	if (SvTRUE (ERRSV))	1272	if (SvTRUE (ERRSV))
900	{	1273	{
901	FREETMPS; LEAVE;	1274	FREETMPS; LEAVE;
902	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1275	ERR_ERRSV;
903	}	1276	}
904		1277
905	SvREFCNT_dec (sv);	1278	SvREFCNT_dec_NN (sv);
906	sv = SvREFCNT_inc (POPs);	1279	sv = SvREFCNT_inc (POPs);
907		1280
908	PUTBACK;	1281	PUTBACK;
909		1282
910	FREETMPS; LEAVE;	1283	FREETMPS; LEAVE;
911	}	1284	}
912	break;	1285	break;
913		1286
914	default:	1287	default:
		1288	filter:
915	{	1289	{
		1290	SV *tag_sv = newSVuv (tag);
		1291
916	sv = decode_sv (dec);	1292	sv = decode_sv (dec);
917		1293
918	dSP;	1294	dSP;
919	ENTER; SAVETMPS; PUSHMARK (SP);	1295	ENTER; SAVETMPS;
		1296	PUSHMARK (SP);
920	EXTEND (SP, 2);	1297	EXTEND (SP, 2);
921	PUSHs (newSVuv (tag));	1298	PUSHs (tag_sv);
922	PUSHs (sv);	1299	PUSHs (sv);
923		1300
924	PUTBACK;	1301	PUTBACK;
925	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);
926	SPAGAIN;	1303	SPAGAIN;
927		1304
928	if (SvTRUE (ERRSV))	1305	if (SvTRUE (ERRSV))
929	{	1306	{
		1307	SvREFCNT_dec_NN (tag_sv);
930	FREETMPS; LEAVE;	1308	FREETMPS; LEAVE;
931	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1309	ERR_ERRSV;
932	}	1310	}
933		1311
934	if (count)	1312	if (count)
935	{	1313	{
		1314	SvREFCNT_dec_NN (tag_sv);
936	SvREFCNT_dec (sv);	1315	SvREFCNT_dec_NN (sv);
937	sv = SvREFCNT_inc (POPs);	1316	sv = SvREFCNT_inc_NN (TOPs);
		1317	SP -= count;
938	}	1318	}
939	else	1319	else
940	{	1320	{
941	AV *av = newAV ();	1321	AV *av = newAV ();
942	av_push (av, newSVuv (tag));	1322	av_push (av, tag_sv);
943	av_push (av, sv);	1323	av_push (av, sv);
944		1324
945	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1325	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
946	? cbor_tagged_stash	1326	? cbor_tagged_stash
947	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1327	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
965	static SV *	1345	static SV *
966	decode_sv (dec_t *dec)	1346	decode_sv (dec_t *dec)
967	{	1347	{
968	WANT (1);	1348	WANT (1);
969		1349
970	switch (*dec->cur >> 5)	1350	switch (*dec->cur >> MAJOR_SHIFT)
971	{	1351	{
972	case 0: // unsigned int	1352	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973	return newSVuv (decode_uint (dec));	1353	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974	case 1: // negative int	1354	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975	return newSViv (-1 - (IV)decode_uint (dec));	1355	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976	case 2: // octet string	1356	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977	return decode_str (dec, 0);	1357	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978	case 3: // utf-8 string	1358	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979	return decode_str (dec, 1);	1359
980	case 4: // array	1360	case MAJOR_MISC >> MAJOR_SHIFT:
981	return decode_av (dec);
982	case 5: // map
983	return decode_hv (dec);
984	case 6: // tag
985	return decode_tagged (dec);
986	case 7: // misc
987	switch (*dec->cur++ & 31)	1361	switch (*dec->cur++ & MINOR_MASK)
988	{	1362	{
989	case 20:	1363	case SIMPLE_FALSE:
990	#if CBOR_SLOW	1364	#if CBOR_SLOW
991	types_false = get_bool ("Types::Serialiser::false");	1365	types_false = get_bool ("Types::Serialiser::false");
992	#endif	1366	#endif
993	return newSVsv (types_false);	1367	return newSVsv (types_false);
994	case 21:	1368	case SIMPLE_TRUE:
995	#if CBOR_SLOW	1369	#if CBOR_SLOW
996	types_true = get_bool ("Types::Serialiser::true");	1370	types_true = get_bool ("Types::Serialiser::true");
997	#endif	1371	#endif
998	return newSVsv (types_true);	1372	return newSVsv (types_true);
999	case 22:	1373	case SIMPLE_NULL:
1000	return newSVsv (&PL_sv_undef);	1374	return newSVsv (&PL_sv_undef);
1001	case 23:	1375	case SIMPLE_UNDEF:
1002	#if CBOR_SLOW	1376	#if CBOR_SLOW
1003	types_error = get_bool ("Types::Serialiser::error");	1377	types_error = get_bool ("Types::Serialiser::error");
1004	#endif	1378	#endif
1005	return newSVsv (types_error);	1379	return newSVsv (types_error);
1006		1380
1007	case 25:	1381	case MISC_FLOAT16:
1008	{	1382	{
1009	WANT (2);	1383	WANT (2);
1010		1384
1011	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1385	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1012	dec->cur += 2;	1386	dec->cur += 2;
1013		1387
1014	return newSVnv (ecb_binary16_to_float (fp));	1388	return newSVnv (ecb_binary16_to_float (fp));
1015	}	1389	}
1016		1390
1017	case 26:	1391	case MISC_FLOAT32:
1018	{	1392	{
1019	uint32_t fp;	1393	uint32_t fp;
1020	WANT (4);	1394	WANT (4);
1021	memcpy (&fp, dec->cur, 4);	1395	memcpy (&fp, dec->cur, 4);
1022	dec->cur += 4;	1396	dec->cur += 4;
…		…
1025	fp = ecb_bswap32 (fp);	1399	fp = ecb_bswap32 (fp);
1026		1400
1027	return newSVnv (ecb_binary32_to_float (fp));	1401	return newSVnv (ecb_binary32_to_float (fp));
1028	}	1402	}
1029		1403
1030	case 27:	1404	case MISC_FLOAT64:
1031	{	1405	{
1032	uint64_t fp;	1406	uint64_t fp;
1033	WANT (8);	1407	WANT (8);
1034	memcpy (&fp, dec->cur, 8);	1408	memcpy (&fp, dec->cur, 8);
1035	dec->cur += 8;	1409	dec->cur += 8;
…		…
1038	fp = ecb_bswap64 (fp);	1412	fp = ecb_bswap64 (fp);
1039		1413
1040	return newSVnv (ecb_binary64_to_double (fp));	1414	return newSVnv (ecb_binary64_to_double (fp));
1041	}	1415	}
1042		1416
1043	// 0..19 unassigned	1417	// 0..19 unassigned simple
1044	// 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
1045	default:	1421	default:
1046	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1422	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047	}	1423	}
1048		1424
1049	break;	1425	break;
1050	}	1426	}
1051		1427
…		…
1054	}	1430	}
1055		1431
1056	static SV *	1432	static SV *
1057	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1433	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1058	{	1434	{
1059	dec_t dec = { };	1435	dec_t dec = { 0 };
1060	SV *sv;	1436	SV *sv;
1061	STRLEN len;	1437	STRLEN len;
1062	char *data = SvPVbyte (string, len);	1438	char *data = SvPVbyte (string, len);
1063		1439
1064	if (len > cbor->max_size && cbor->max_size)	1440	if (len > cbor->max_size && cbor->max_size)
…		…
1078	if (dec.cur != dec.end && !dec.err)	1454	if (dec.cur != dec.end && !dec.err)
1079	dec.err = "garbage after CBOR object";	1455	dec.err = "garbage after CBOR object";
1080		1456
1081	if (dec.err)	1457	if (dec.err)
1082	{	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
1083	SvREFCNT_dec (sv);	1470	SvREFCNT_dec_NN (sv);
		1471
		1472	if (dec.err_sv)
		1473	sv_2mortal (dec.err_sv);
		1474
1084	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);
1085	}	1476	}
1086		1477
1087	sv = sv_2mortal (sv);	1478	sv = sv_2mortal (sv);
1088		1479
1089	return sv;	1480	return sv;
1090	}	1481	}
1091		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
1092	/////////////////////////////////////////////////////////////////////////////	1608	/////////////////////////////////////////////////////////////////////////////
1093	// XS interface functions	1609	// XS interface functions
1094		1610
1095	MODULE = CBOR::XS PACKAGE = CBOR::XS	1611	MODULE = CBOR::XS PACKAGE = CBOR::XS
1096		1612
…		…
1108		1624
1109	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1625	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1110		1626
1111	sv_cbor = newSVpv ("CBOR", 0);	1627	sv_cbor = newSVpv ("CBOR", 0);
1112	SvREADONLY_on (sv_cbor);	1628	SvREADONLY_on (sv_cbor);
		1629
		1630	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1113	}	1631	}
1114		1632
1115	PROTOTYPES: DISABLE	1633	PROTOTYPES: DISABLE
1116		1634
1117	void CLONE (...)	1635	void CLONE (...)
…		…
1136	void shrink (CBOR *self, int enable = 1)	1654	void shrink (CBOR *self, int enable = 1)
1137	ALIAS:	1655	ALIAS:
1138	shrink = F_SHRINK	1656	shrink = F_SHRINK
1139	allow_unknown = F_ALLOW_UNKNOWN	1657	allow_unknown = F_ALLOW_UNKNOWN
1140	allow_sharing = F_ALLOW_SHARING	1658	allow_sharing = F_ALLOW_SHARING
1141	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
1142	PPCODE:	1665	PPCODE:
1143	{	1666	{
1144	if (enable)	1667	if (enable)
1145	self->flags |= ix;	1668	self->flags |= ix;
1146	else	1669	else
…		…
1152	void get_shrink (CBOR *self)	1675	void get_shrink (CBOR *self)
1153	ALIAS:	1676	ALIAS:
1154	get_shrink = F_SHRINK	1677	get_shrink = F_SHRINK
1155	get_allow_unknown = F_ALLOW_UNKNOWN	1678	get_allow_unknown = F_ALLOW_UNKNOWN
1156	get_allow_sharing = F_ALLOW_SHARING	1679	get_allow_sharing = F_ALLOW_SHARING
1157	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
1158	PPCODE:	1686	PPCODE:
1159	XPUSHs (boolSV (self->flags & ix));	1687	XPUSHs (boolSV (self->flags & ix));
1160		1688
1161	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1689	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162	PPCODE:	1690	PPCODE:
…		…
1211	EXTEND (SP, 2);	1739	EXTEND (SP, 2);
1212	PUSHs (sv);	1740	PUSHs (sv);
1213	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1741	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214	}	1742	}
1215		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
1216	void DESTROY (CBOR *self)	1796	void DESTROY (CBOR *self)
1217	PPCODE:	1797	PPCODE:
1218	cbor_free (self);	1798	cbor_free (self);
1219		1799
1220	PROTOTYPES: ENABLE	1800	PROTOTYPES: ENABLE
1221		1801
1222	void encode_cbor (SV *scalar)	1802	void encode_cbor (SV *scalar)
		1803	ALIAS:
		1804	encode_cbor = 0
		1805	encode_cbor_sharing = F_ALLOW_SHARING
1223	PPCODE:	1806	PPCODE:
1224	{	1807	{
1225	CBOR cbor;	1808	CBOR cbor;
1226	cbor_init (&cbor);	1809	cbor_init (&cbor);
		1810	cbor.flags |= ix;
1227	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1811	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228	XPUSHs (scalar);	1812	XPUSHs (scalar);
1229	}	1813	}
1230		1814
1231	void decode_cbor (SV *cborstr)	1815	void decode_cbor (SV *cborstr)
…		…
1235	cbor_init (&cbor);	1819	cbor_init (&cbor);
1236	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1820	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1237	XPUSHs (cborstr);	1821	XPUSHs (cborstr);
1238	}	1822	}
1239		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