ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.15 by root, Tue Oct 29 18:37:31 2013 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 2020 UTC

…		…
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
		11	#include <inttypes.h>
11		12
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
		29	#ifndef SvREFCNT_dec_NN
		30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		31	#endif
		32
		33	// known major and minor types
		34	enum cbor_type
		35	{
		36	MAJOR_SHIFT = 5,
		37	MINOR_MASK = 0x1f,
		38
		39	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		40	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		41	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		42	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		43	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		44	MAJOR_MAP = 5 << MAJOR_SHIFT,
		45	MAJOR_TAG = 6 << MAJOR_SHIFT,
		46	MAJOR_MISC = 7 << MAJOR_SHIFT,
		47
		48	// INT/STRING/ARRAY/MAP subtypes
		49	LENGTH_EXT1 = 24,
		50	LENGTH_EXT2 = 25,
		51	LENGTH_EXT4 = 26,
		52	LENGTH_EXT8 = 27,
		53
		54	// SIMPLE types (effectively MISC subtypes)
		55	SIMPLE_FALSE = 20,
		56	SIMPLE_TRUE = 21,
		57	SIMPLE_NULL = 22,
		58	SIMPLE_UNDEF = 23,
		59
		60	// MISC subtype (unused)
		61	MISC_EXT1 = 24,
		62	MISC_FLOAT16 = 25,
		63	MISC_FLOAT32 = 26,
		64	MISC_FLOAT64 = 27,
		65
		66	// BYTES/TEXT/ARRAY/MAP
		67	MINOR_INDEF = 31,
		68	};
24		69
25	// known tags	70	// known tags
26	enum cbor_tag	71	enum cbor_tag
27	{	72	{
28	// inofficial extensions (pending iana registration)	73	// extensions
29	CBOR_TAG_PERL_OBJECT = 256,	74	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30	CBOR_TAG_GENERIC_OBJECT = 257,	75	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
		76	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
		77	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
		78	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
		79	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
		80	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31		81
32	// rfc7049	82	// rfc7049
33	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	83	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	84	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35	CBOR_TAG_POS_BIGNUM = 2, // byte string	85	CBOR_TAG_POS_BIGNUM = 2, // byte string
36	CBOR_TAG_NEG_BIGNUM = 3, // byte string	86	CBOR_TAG_NEG_BIGNUM = 3, // byte string
37	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	87	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38	CBOR_TAG_BIGFLOAT = 5, // array	88	CBOR_TAG_BIGFLOAT = 5, // array
39		89
40	CBOR_TAG_CONV_B64U = 21, // base64url, any	90	CBOR_TAG_CONV_B64U = 21, // base64url, any
41	CBOR_TAG_CONV_B64 = 22, // base64, any	91	CBOR_TAG_CONV_B64 = 22, // base64, any
42	CBOR_TAG_CONV_HEX = 23, // base16, any	92	CBOR_TAG_CONV_HEX = 23, // base16, any
43	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	93	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44		94
45	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	95	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	96	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	97	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	98	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50		100
51	CBOR_TAG_MAGIC = 55799 // self-describe cbor	101	CBOR_TAG_MAGIC = 55799, // self-describe cbor
52	};	102	};
53		103
		104	// known forced types, also hardcoded in CBOR.pm
		105	enum
		106	{
		107	AS_CBOR = 0,
		108	AS_BYTES = 1,
		109	AS_TEXT = 2,
		110	AS_FLOAT16 = 3,
		111	AS_FLOAT32 = 4,
		112	AS_FLOAT64 = 5,
		113	// possibly future enhancements: float, integer
		114	};
		115
54	#define F_SHRINK 0x00000200UL	116	#define F_SHRINK 0x00000001UL
55	#define F_ALLOW_UNKNOWN 0x00002000UL	117	#define F_ALLOW_UNKNOWN 0x00000002UL
		118	#define F_ALLOW_SHARING 0x00000004UL
		119	#define F_ALLOW_CYCLES 0x00000008UL
		120	#define F_FORBID_OBJECTS 0x00000010UL
		121	#define F_PACK_STRINGS 0x00000020UL
		122	#define F_TEXT_KEYS 0x00000040UL
		123	#define F_TEXT_STRINGS 0x00000080UL
		124	#define F_VALIDATE_UTF8 0x00000100UL
56		125
57	#define INIT_SIZE 32 // initial scalar size to be allocated	126	#define INIT_SIZE 32 // initial scalar size to be allocated
58		127
59	#define SB do {	128	#define SB do {
60	#define SE } while (0)	129	#define SE } while (0)
…		…
72	# define CBOR_SLOW 0	141	# define CBOR_SLOW 0
73	# define CBOR_STASH cbor_stash	142	# define CBOR_STASH cbor_stash
74	#endif	143	#endif
75		144
76	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::	145	static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
77	static SV *types_true, *types_false, *types_error, *sv_cbor;	146	static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78		147
79	typedef struct {	148	typedef struct {
80	U32 flags;	149	U32 flags;
81	U32 max_depth;	150	U32 max_depth;
82	STRLEN max_size;	151	STRLEN max_size;
		152	SV *filter;
		153
		154	// for the incremental parser
		155	STRLEN incr_pos; // the current offset into the text
		156	STRLEN incr_need; // minimum bytes needed to decode
		157	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
83	} CBOR;	158	} CBOR;
84		159
85	ecb_inline void	160	ecb_inline void
86	cbor_init (CBOR *cbor)	161	cbor_init (CBOR *cbor)
87	{	162	{
88	Zero (cbor, 1, CBOR);	163	Zero (cbor, 1, CBOR);
89	cbor->max_depth = 512;	164	cbor->max_depth = 512;
		165	}
		166
		167	ecb_inline void
		168	cbor_free (CBOR *cbor)
		169	{
		170	SvREFCNT_dec (cbor->filter);
		171	SvREFCNT_dec (cbor->incr_count);
90	}	172	}
91		173
92	/////////////////////////////////////////////////////////////////////////////	174	/////////////////////////////////////////////////////////////////////////////
93	// utility functions	175	// utility functions
94		176
…		…
116	SvPV_renew (sv, SvCUR (sv) + 1);	198	SvPV_renew (sv, SvCUR (sv) + 1);
117	#endif	199	#endif
118	}	200	}
119	}	201	}
120		202
121	/////////////////////////////////////////////////////////////////////////////	203	// minimum length of a string to be registered for stringref
122	// fp hell	204	ecb_inline int
123		205	minimum_string_length (UV idx)
124	//TODO	206	{
		207	return idx <= 23 ? 3
		208	: idx <= 0xffU ? 4
		209	: idx <= 0xffffU ? 5
		210	: idx <= 0xffffffffU ? 7
		211	: 11;
		212	}
125		213
126	/////////////////////////////////////////////////////////////////////////////	214	/////////////////////////////////////////////////////////////////////////////
127	// encoder	215	// encoder
128		216
129	// structure used for encoding CBOR	217	// structure used for encoding CBOR
…		…
132	char *cur; // SvPVX (sv) + current output position	220	char *cur; // SvPVX (sv) + current output position
133	char *end; // SvEND (sv)	221	char *end; // SvEND (sv)
134	SV *sv; // result scalar	222	SV *sv; // result scalar
135	CBOR cbor;	223	CBOR cbor;
136	U32 depth; // recursion level	224	U32 depth; // recursion level
		225	HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
		226	UV stringref_idx;
		227	HV *shareable; // ptr => index, or 0
		228	UV shareable_idx;
137	} enc_t;	229	} enc_t;
138		230
139	ecb_inline void	231	ecb_inline void
140	need (enc_t *enc, STRLEN len)	232	need (enc_t *enc, STRLEN len)
141	{	233	{
142	if (ecb_expect_false (enc->cur + len >= enc->end))	234	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
143	{	235	{
144	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	236	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
145	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	237	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
146	enc->cur = SvPVX (enc->sv) + cur;	238	enc->cur = SvPVX (enc->sv) + cur;
147	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	239	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
153	{	245	{
154	need (enc, 1);	246	need (enc, 1);
155	*enc->cur++ = ch;	247	*enc->cur++ = ch;
156	}	248	}
157		249
		250	// used for tags, intregers, element counts and so on
158	static void	251	static void
159	encode_uint (enc_t *enc, int major, UV len)	252	encode_uint (enc_t *enc, int major, UV len)
160	{	253	{
161	need (enc, 9);	254	need (enc, 9);
162		255
163	if (len < 24)	256	if (ecb_expect_true (len < LENGTH_EXT1))
164	*enc->cur++ = major | len;	257	*enc->cur++ = major | len;
165	else if (len <= 0xff)	258	else if (ecb_expect_true (len <= 0xffU))
166	{	259	{
167	*enc->cur++ = major | 24;	260	*enc->cur++ = major | LENGTH_EXT1;
168	*enc->cur++ = len;	261	*enc->cur++ = len;
169	}	262	}
170	else if (len <= 0xffff)	263	else if (len <= 0xffffU)
171	{	264	{
172	*enc->cur++ = major | 25;	265	*enc->cur++ = major | LENGTH_EXT2;
173	*enc->cur++ = len >> 8;	266	*enc->cur++ = len >> 8;
174	*enc->cur++ = len;	267	*enc->cur++ = len;
175	}	268	}
176	else if (len <= 0xffffffff)	269	else if (len <= 0xffffffffU)
177	{	270	{
178	*enc->cur++ = major | 26;	271	*enc->cur++ = major | LENGTH_EXT4;
179	*enc->cur++ = len >> 24;	272	*enc->cur++ = len >> 24;
180	*enc->cur++ = len >> 16;	273	*enc->cur++ = len >> 16;
181	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
182	*enc->cur++ = len;	275	*enc->cur++ = len;
183	}	276	}
184	else	277	else
185	{	278	{
186	*enc->cur++ = major | 27;	279	*enc->cur++ = major | LENGTH_EXT8;
187	*enc->cur++ = len >> 56;	280	*enc->cur++ = len >> 56;
188	*enc->cur++ = len >> 48;	281	*enc->cur++ = len >> 48;
189	*enc->cur++ = len >> 40;	282	*enc->cur++ = len >> 40;
190	*enc->cur++ = len >> 32;	283	*enc->cur++ = len >> 32;
191	*enc->cur++ = len >> 24;	284	*enc->cur++ = len >> 24;
…		…
193	*enc->cur++ = len >> 8;	286	*enc->cur++ = len >> 8;
194	*enc->cur++ = len;	287	*enc->cur++ = len;
195	}	288	}
196	}	289	}
197		290
198	static void	291	// encodes a perl value into a CBOR integer
		292	ecb_inline
		293	encode_int (enc_t *enc, SV *sv)
		294	{
		295	if (SvIsUV (sv))
		296	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		297	else if (SvIVX (sv) >= 0)
		298	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		299	else
		300	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		301	}
		302
		303	ecb_inline void
		304	encode_tag (enc_t *enc, UV tag)
		305	{
		306	encode_uint (enc, MAJOR_TAG, tag);
		307	}
		308
		309	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		310	ecb_noinline static void
199	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	311	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200	{	312	{
201	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	313	STRLEN ulen = len;
		314	U8 *p, *pend = (U8 *)str + len;
		315
		316	for (p = (U8 *)str; p < pend; ++p)
		317	ulen += *p >> 7; // count set high bits
		318
		319	encode_uint (enc, MAJOR_TEXT, ulen);
		320
		321	need (enc, ulen);
		322	for (p = (U8 *)str; p < pend; ++p)
		323	if (*p < 0x80)
		324	*enc->cur++ = *p;
		325	else
		326	{
		327	*enc->cur++ = 0xc0 + (*p >> 6);
		328	*enc->cur++ = 0x80 + (*p & 63);
		329	}
		330	}
		331
		332	ecb_inline void
		333	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		334	{
		335	if (ecb_expect_false (upgrade_utf8))
		336	if (!utf8)
		337	{
		338	encode_str_utf8 (enc, utf8, str, len);
		339	return;
		340	}
		341
		342	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
202	need (enc, len);	343	need (enc, len);
203	memcpy (enc->cur, str, len);	344	memcpy (enc->cur, str, len);
204	enc->cur += len;	345	enc->cur += len;
205	}	346	}
206		347
		348	ecb_inline void
		349	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
		350	{
		351	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		352	{
		353	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		354
		355	if (SvOK (*svp))
		356	{
		357	// already registered, use stringref
		358	encode_tag (enc, CBOR_TAG_STRINGREF);
		359	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		360	return;
		361	}
		362	else if (len >= minimum_string_length (enc->stringref_idx))
		363	{
		364	// register only
		365	sv_setuv (*svp, enc->stringref_idx);
		366	++enc->stringref_idx;
		367	}
		368	}
		369
		370	encode_str (enc, upgrade_utf8, utf8, str, len);
		371	}
		372
		373	ecb_inline void
		374	encode_float16 (enc_t *enc, NV nv)
		375	{
		376	need (enc, 1+2);
		377
		378	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		379
		380	uint16_t fp = ecb_float_to_binary16 (nv);
		381
		382	if (!ecb_big_endian ())
		383	fp = ecb_bswap16 (fp);
		384
		385	memcpy (enc->cur, &fp, 2);
		386	enc->cur += 2;
		387	}
		388
		389	ecb_inline void
		390	encode_float32 (enc_t *enc, NV nv)
		391	{
		392	need (enc, 1+4);
		393
		394	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		395
		396	uint32_t fp = ecb_float_to_binary32 (nv);
		397
		398	if (!ecb_big_endian ())
		399	fp = ecb_bswap32 (fp);
		400
		401	memcpy (enc->cur, &fp, 4);
		402	enc->cur += 4;
		403	}
		404
		405	ecb_inline void
		406	encode_float64 (enc_t *enc, NV nv)
		407	{
		408	need (enc, 1+8);
		409
		410	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		411
		412	uint64_t fp = ecb_double_to_binary64 (nv);
		413
		414	if (!ecb_big_endian ())
		415	fp = ecb_bswap64 (fp);
		416
		417	memcpy (enc->cur, &fp, 8);
		418	enc->cur += 8;
		419	}
		420
		421	ecb_inline void
		422	encode_forced (enc_t *enc, UV type, SV *sv)
		423	{
		424	switch (type)
		425	{
		426	case AS_CBOR:
		427	{
		428	STRLEN len;
		429	char *str = SvPVbyte (sv, len);
		430
		431	need (enc, len);
		432	memcpy (enc->cur, str, len);
		433	enc->cur += len;
		434	}
		435	break;
		436
		437	case AS_BYTES:
		438	{
		439	STRLEN len;
		440	char *str = SvPVbyte (sv, len);
		441	encode_strref (enc, 0, 0, str, len);
		442	}
		443	break;
		444
		445	case AS_TEXT:
		446	{
		447	STRLEN len;
		448	char *str = SvPVutf8 (sv, len);
		449	encode_strref (enc, 1, 1, str, len);
		450	}
		451	break;
		452
		453	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		454	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		455	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		456
		457	default:
		458	croak ("encountered malformed CBOR::XS::Tagged object");
		459	}
		460	}
		461
207	static void encode_sv (enc_t *enc, SV *sv);	462	static void encode_sv (enc_t *enc, SV *sv);
208		463
209	static void	464	static void
210	encode_av (enc_t *enc, AV *av)	465	encode_av (enc_t *enc, AV *av)
211	{	466	{
…		…
214	if (enc->depth >= enc->cbor.max_depth)	469	if (enc->depth >= enc->cbor.max_depth)
215	croak (ERR_NESTING_EXCEEDED);	470	croak (ERR_NESTING_EXCEEDED);
216		471
217	++enc->depth;	472	++enc->depth;
218		473
219	encode_uint (enc, 0x80, len + 1);	474	encode_uint (enc, MAJOR_ARRAY, len + 1);
220		475
		476	if (ecb_expect_false (SvMAGICAL (av)))
221	for (i = 0; i <= len; ++i)	477	for (i = 0; i <= len; ++i)
222	{	478	{
223	SV **svp = av_fetch (av, i, 0);	479	SV **svp = av_fetch (av, i, 0);
224	encode_sv (enc, svp ? *svp : &PL_sv_undef);	480	encode_sv (enc, svp ? *svp : &PL_sv_undef);
225	}	481	}
		482	else
		483	for (i = 0; i <= len; ++i)
		484	{
		485	SV *sv = AvARRAY (av)[i];
		486	encode_sv (enc, sv ? sv : &PL_sv_undef);
		487	}
226		488
227	--enc->depth;	489	--enc->depth;
228	}	490	}
229		491
230	static void	492	static void
…		…
238	++enc->depth;	500	++enc->depth;
239		501
240	int pairs = hv_iterinit (hv);	502	int pairs = hv_iterinit (hv);
241	int mg = SvMAGICAL (hv);	503	int mg = SvMAGICAL (hv);
242		504
243	if (mg)	505	if (ecb_expect_false (mg))
244	encode_ch (enc, 0xa0 | 31);	506	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245	else	507	else
246	encode_uint (enc, 0xa0, pairs);	508	encode_uint (enc, MAJOR_MAP, pairs);
247		509
248	while ((he = hv_iternext (hv)))	510	while ((he = hv_iternext (hv)))
249	{	511	{
250	if (HeKLEN (he) == HEf_SVKEY)	512	if (HeKLEN (he) == HEf_SVKEY)
251	encode_sv (enc, HeSVKEY (he));	513	encode_sv (enc, HeSVKEY (he));
252	else	514	else
253	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	515	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254		516
255	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	517	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
256	}	518	}
257		519
258	if (mg)	520	if (ecb_expect_false (mg))
259	encode_ch (enc, 0xe0 | 31);	521	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260		522
261	--enc->depth;	523	--enc->depth;
262	}	524	}
263		525
264	// encode objects, arrays and special \0=false and \1=true values.	526	// encode objects, arrays and special \0=false and \1=true values.
265	static void	527	static void
266	encode_rv (enc_t *enc, SV *sv)	528	encode_rv (enc_t *enc, SV *sv)
267	{	529	{
268	svtype svt;
269
270	SvGETMAGIC (sv);	530	SvGETMAGIC (sv);
		531
271	svt = SvTYPE (sv);	532	svtype svt = SvTYPE (sv);
272		533
273	if (ecb_expect_false (SvOBJECT (sv)))	534	if (ecb_expect_false (SvOBJECT (sv)))
274	{	535	{
275	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash	536	HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276	? types_boolean_stash	537	? types_boolean_stash
…		…
281	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	542	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282	? cbor_tagged_stash	543	? cbor_tagged_stash
283	: gv_stashpv ("CBOR::XS::Tagged" , 1);	544	: gv_stashpv ("CBOR::XS::Tagged" , 1);
284		545
285	HV *stash = SvSTASH (sv);	546	HV *stash = SvSTASH (sv);
286	GV *method;
287		547
288	if (stash == boolean_stash)	548	if (stash == boolean_stash)
289	encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);	549	{
		550	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		551	return;
		552	}
290	else if (stash == error_stash)	553	else if (stash == error_stash)
291	encode_ch (enc, 0xe0 | 23);	554	{
		555	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
		556	return;
		557	}
292	else if (stash == tagged_stash)	558	else if (stash == tagged_stash)
293	{	559	{
294	if (svt != SVt_PVAV)	560	if (svt != SVt_PVAV)
295	croak ("encountered CBOR::XS::Tagged object that isn't an array");	561	croak ("encountered CBOR::XS::Tagged object that isn't an array");
296		562
		563	switch (av_len ((AV *)sv))
		564	{
		565	case 2-1:
		566	// actually a tagged value
297	encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));	567	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	568	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		569	break;
		570
		571	case 3-1:
		572	// a forced type [value, type, undef]
		573	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		574	break;
		575
		576	default:
		577	croak ("encountered malformed CBOR::XS::Tagged object");
		578	}
		579
		580	return;
		581	}
		582	}
		583
		584	if (ecb_expect_false (SvREFCNT (sv) > 1)
		585	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		586	{
		587	if (ecb_expect_false (!enc->shareable))
		588	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
		589
		590	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
		591
		592	if (SvOK (*svp))
299	}	593	{
		594	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		595	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		596	return;
		597	}
		598	else
		599	{
		600	sv_setuv (*svp, enc->shareable_idx);
		601	++enc->shareable_idx;
		602	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		603	}
		604	}
		605
		606	if (ecb_expect_false (SvOBJECT (sv)))
		607	{
		608	HV *stash = SvSTASH (sv);
		609	GV *method;
		610
		611	if (enc->cbor.flags & F_FORBID_OBJECTS)
		612	croak ("encountered object '%s', but forbid_objects is enabled",
		613	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
300	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	614	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301	{	615	{
302	dSP;	616	dSP;
303		617
304	ENTER; SAVETMPS; PUSHMARK (SP);	618	ENTER; SAVETMPS;
		619	PUSHMARK (SP);
305	// we re-bless the reference to get overload and other niceties right	620	// we re-bless the reference to get overload and other niceties right
306	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	621	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307		622
308	PUTBACK;	623	PUTBACK;
309	// G_SCALAR ensures that return value is 1	624	// G_SCALAR ensures that return value is 1
…		…
322	}	637	}
323	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	638	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324	{	639	{
325	dSP;	640	dSP;
326		641
327	ENTER; SAVETMPS; PUSHMARK (SP);	642	ENTER; SAVETMPS;
		643	PUSHMARK (SP);
328	EXTEND (SP, 2);	644	EXTEND (SP, 2);
329	// we re-bless the reference to get overload and other niceties right	645	// we re-bless the reference to get overload and other niceties right
330	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	646	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331	PUSHs (sv_cbor);	647	PUSHs (sv_cbor);
332		648
…		…
336		652
337	// catch this surprisingly common error	653	// catch this surprisingly common error
338	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	654	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	655	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340		656
341	encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT);	657	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342	encode_uint (enc, 0x80, count + 1);	658	encode_uint (enc, MAJOR_ARRAY, count + 1);
343	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	659	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344		660
345	while (count)	661	{
		662	int i;
		663
		664	for (i = 0; i < count; ++i)
346	encode_sv (enc, SP[1 - count--]);	665	encode_sv (enc, SP[i + 1 - count]);
		666
		667	SP -= count;
		668	}
347		669
348	PUTBACK;	670	PUTBACK;
349		671
350	FREETMPS; LEAVE;	672	FREETMPS; LEAVE;
351	}	673	}
…		…
355	}	677	}
356	else if (svt == SVt_PVHV)	678	else if (svt == SVt_PVHV)
357	encode_hv (enc, (HV *)sv);	679	encode_hv (enc, (HV *)sv);
358	else if (svt == SVt_PVAV)	680	else if (svt == SVt_PVAV)
359	encode_av (enc, (AV *)sv);	681	encode_av (enc, (AV *)sv);
360	else if (svt < SVt_PVAV)
361	{
362	STRLEN len = 0;
363	char *pv = svt ? SvPV (sv, len) : 0;
364
365	if (len == 1 && *pv == '1')
366	encode_ch (enc, 0xe0 | 21);
367	else if (len == 1 && *pv == '0')
368	encode_ch (enc, 0xe0 | 20);
369	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370	encode_ch (enc, 0xe0 | 23);
371	else
372	croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374	}
375	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376	encode_ch (enc, 0xe0 | 23);
377	else	682	else
378	croak ("encountered %s, but CBOR can only represent references to arrays or hashes",	683	{
379	SvPV_nolen (sv_2mortal (newRV_inc (sv))));	684	encode_tag (enc, CBOR_TAG_INDIRECTION);
		685	encode_sv (enc, sv);
		686	}
380	}	687	}
381		688
382	static void	689	static void
383	encode_nv (enc_t *enc, SV *sv)	690	encode_nv (enc_t *enc, SV *sv)
384	{	691	{
385	double nv = SvNVX (sv);	692	double nv = SvNVX (sv);
386		693
387	need (enc, 9);	694	need (enc, 9);
388		695
389	if (ecb_expect_false (nv == (U32)nv))	696	if (ecb_expect_false (nv == (NV)(U32)nv))
390	encode_uint (enc, 0x00, (U32)nv);	697	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391	//TODO: maybe I32?	698	//TODO: maybe I32?
392	else if (ecb_expect_false (nv == (float)nv))	699	else if (ecb_expect_false (nv == (float)nv))
393	{	700	encode_float32 (enc, nv);
394	uint32_t fp = ecb_float_to_binary32 (nv);
395
396	*enc->cur++ = 0xe0 | 26;
397
398	if (!ecb_big_endian ())
399	fp = ecb_bswap32 (fp);
400
401	memcpy (enc->cur, &fp, 4);
402	enc->cur += 4;
403	}
404	else	701	else
405	{	702	encode_float64 (enc, nv);
406	uint64_t fp = ecb_double_to_binary64 (nv);
407
408	*enc->cur++ = 0xe0 | 27;
409
410	if (!ecb_big_endian ())
411	fp = ecb_bswap64 (fp);
412
413	memcpy (enc->cur, &fp, 8);
414	enc->cur += 8;
415	}
416	}	703	}
417		704
418	static void	705	static void
419	encode_sv (enc_t *enc, SV *sv)	706	encode_sv (enc_t *enc, SV *sv)
420	{	707	{
…		…
422		709
423	if (SvPOKp (sv))	710	if (SvPOKp (sv))
424	{	711	{
425	STRLEN len;	712	STRLEN len;
426	char *str = SvPV (sv, len);	713	char *str = SvPV (sv, len);
427	encode_str (enc, SvUTF8 (sv), str, len);	714	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428	}	715	}
429	else if (SvNOKp (sv))	716	else if (SvNOKp (sv))
430	encode_nv (enc, sv);	717	encode_nv (enc, sv);
431	else if (SvIOKp (sv))	718	else if (SvIOKp (sv))
432	{	719	encode_int (e,v sv);
433	if (SvIsUV (sv))
434	encode_uint (enc, 0x00, SvUVX (sv));
435	else if (SvIVX (sv) >= 0)
436	encode_uint (enc, 0x00, SvIVX (sv));
437	else
438	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
439	}
440	else if (SvROK (sv))	720	else if (SvROK (sv))
441	encode_rv (enc, SvRV (sv));	721	encode_rv (enc, SvRV (sv));
442	else if (!SvOK (sv))	722	else if (!SvOK (sv))
443	encode_ch (enc, 0xe0 | 22);	723	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	724	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445	encode_ch (enc, 0xe0 | 23);	725	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446	else	726	else
447	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	727	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
448	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	728	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449	}	729	}
450		730
451	static SV *	731	static SV *
452	encode_cbor (SV *scalar, CBOR *cbor)	732	encode_cbor (SV *scalar, CBOR *cbor)
453	{	733	{
454	enc_t enc;	734	enc_t enc = { 0 };
455		735
456	enc.cbor = *cbor;	736	enc.cbor = *cbor;
457	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	737	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458	enc.cur = SvPVX (enc.sv);	738	enc.cur = SvPVX (enc.sv);
459	enc.end = SvEND (enc.sv);	739	enc.end = SvEND (enc.sv);
460	enc.depth = 0;
461		740
462	SvPOK_only (enc.sv);	741	SvPOK_only (enc.sv);
		742
		743	if (cbor->flags & F_PACK_STRINGS)
		744	{
		745	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
		746	enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
		747	enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
		748	}
		749
463	encode_sv (&enc, scalar);	750	encode_sv (&enc, scalar);
464		751
465	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));	752	SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings	753	*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
467		754
…		…
481	U8 *end; // end of input string	768	U8 *end; // end of input string
482	const char *err; // parse error, if != 0	769	const char *err; // parse error, if != 0
483	CBOR cbor;	770	CBOR cbor;
484	U32 depth; // recursion depth	771	U32 depth; // recursion depth
485	U32 maxdepth; // recursion depth limit	772	U32 maxdepth; // recursion depth limit
		773	AV *shareable;
		774	AV *stringref;
		775	SV *decode_tagged;
		776	SV *err_sv; // optional sv for error, needs to be freed
486	} dec_t;	777	} dec_t;
487		778
488	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	779	// set dec->err to ERRSV
		780	ecb_cold static void
		781	err_errsv (dec_t *dec)
		782	{
		783	if (!dec->err)
		784	{
		785	dec->err_sv = newSVsv (ERRSV);
489		786
490	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	787	// chop off the trailing \n
		788	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		789	*SvEND (dec->err_sv) = 0;
491		790
		791	dec->err = SvPVutf8_nolen (dec->err_sv);
		792	}
		793	}
		794
		795	// the following functions are used to reduce code size and help the compiler to optimise
		796	ecb_cold static void
		797	err_set (dec_t *dec, const char *reason)
		798	{
		799	if (!dec->err)
		800	dec->err = reason;
		801	}
		802
		803	ecb_cold static void
		804	err_unexpected_end (dec_t *dec)
		805	{
		806	err_set (dec, "unexpected end of CBOR data");
		807	}
		808
		809	#define ERR_DO(do) SB do; goto fail; SE
		810	#define ERR(reason) ERR_DO (err_set (dec, reason))
		811	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		812
		813	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		814
492	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	815	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
493	#define DEC_DEC_DEPTH --dec->depth	816	#define DEC_DEC_DEPTH --dec->depth
494		817
495	static UV	818	static UV
496	decode_uint (dec_t *dec)	819	decode_uint (dec_t *dec)
497	{	820	{
498	switch (*dec->cur & 31)	821	U8 m = *dec->cur & MINOR_MASK;
499	{	822	++dec->cur;
500	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503	return *dec->cur++ & 31;
504		823
505	case 24:	824	if (ecb_expect_true (m < LENGTH_EXT1))
		825	return m;
		826	else if (ecb_expect_true (m == LENGTH_EXT1))
		827	{
506	WANT (2);	828	WANT (1);
507	dec->cur += 2;	829	dec->cur += 1;
508	return dec->cur[-1];	830	return dec->cur[-1];
509		831	}
510	case 25:	832	else if (ecb_expect_true (m == LENGTH_EXT2))
		833	{
511	WANT (3);	834	WANT (2);
512	dec->cur += 3;	835	dec->cur += 2;
513	return (((UV)dec->cur[-2]) << 8)	836	return (((UV)dec->cur[-2]) << 8)
514	| ((UV)dec->cur[-1]);	837	| ((UV)dec->cur[-1]);
515		838	}
516	case 26:	839	else if (ecb_expect_true (m == LENGTH_EXT4))
		840	{
517	WANT (5);	841	WANT (4);
518	dec->cur += 5;	842	dec->cur += 4;
519	return (((UV)dec->cur[-4]) << 24)	843	return (((UV)dec->cur[-4]) << 24)
520	| (((UV)dec->cur[-3]) << 16)	844	| (((UV)dec->cur[-3]) << 16)
521	| (((UV)dec->cur[-2]) << 8)	845	| (((UV)dec->cur[-2]) << 8)
522	| ((UV)dec->cur[-1]);	846	| ((UV)dec->cur[-1]);
523		847	}
524	case 27:	848	else if (ecb_expect_true (m == LENGTH_EXT8))
		849	{
525	WANT (9);	850	WANT (8);
526	dec->cur += 9;	851	dec->cur += 8;
		852
		853	return
		854	#if UVSIZE < 8
		855	0
		856	#else
527	return (((UV)dec->cur[-8]) << 56)	857	(((UV)dec->cur[-8]) << 56)
528	| (((UV)dec->cur[-7]) << 48)	858	| (((UV)dec->cur[-7]) << 48)
529	| (((UV)dec->cur[-6]) << 40)	859	| (((UV)dec->cur[-6]) << 40)
530	| (((UV)dec->cur[-5]) << 32)	860	| (((UV)dec->cur[-5]) << 32)
		861	#endif
531	| (((UV)dec->cur[-4]) << 24)	862	| (((UV)dec->cur[-4]) << 24)
532	| (((UV)dec->cur[-3]) << 16)	863	| (((UV)dec->cur[-3]) << 16)
533	| (((UV)dec->cur[-2]) << 8)	864	| (((UV)dec->cur[-2]) << 8)
534	| ((UV)dec->cur[-1]);	865	| ((UV)dec->cur[-1]);
535		866	}
536	default:	867	else
537	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	868	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538	}
539		869
540	fail:	870	fail:
541	return 0;	871	return 0;
542	}	872	}
543		873
…		…
548	{	878	{
549	AV *av = newAV ();	879	AV *av = newAV ();
550		880
551	DEC_INC_DEPTH;	881	DEC_INC_DEPTH;
552		882
553	if ((*dec->cur & 31) == 31)	883	if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554	{	884	{
555	++dec->cur;	885	++dec->cur;
556		886
557	for (;;)	887	for (;;)
558	{	888	{
559	WANT (1);	889	WANT (1);
560		890
561	if (*dec->cur == (0xe0 | 31))	891	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
562	{	892	{
563	++dec->cur;	893	++dec->cur;
564	break;	894	break;
565	}	895	}
566		896
567	av_push (av, decode_sv (dec));	897	av_push (av, decode_sv (dec));
568	}	898	}
569	}	899	}
570	else	900	else
571	{	901	{
572	int i, len = decode_uint (dec);	902	UV i, len = decode_uint (dec);
573		903
		904	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574	av_fill (av, len - 1);	905	av_fill (av, len - 1);
575		906
576	for (i = 0; i < len; ++i)	907	for (i = 0; i < len; ++i)
577	AvARRAY (av)[i] = decode_sv (dec);	908	AvARRAY (av)[i] = decode_sv (dec);
578	}	909	}
579		910
580	DEC_DEC_DEPTH;	911	DEC_DEC_DEPTH;
581	return newRV_noinc ((SV *)av);	912	return newRV_noinc ((SV *)av);
582		913
583	fail:	914	fail:
584	SvREFCNT_dec (av);	915	SvREFCNT_dec_NN (av);
585	DEC_DEC_DEPTH;	916	DEC_DEC_DEPTH;
586	return &PL_sv_undef;	917	return &PL_sv_undef;
587	}	918	}
588		919
		920	static void
		921	decode_he (dec_t *dec, HV *hv)
		922	{
		923	// for speed reasons, we specialcase single-string
		924	// byte or utf-8 strings as keys, but only when !stringref
		925
		926	if (ecb_expect_true (!dec->stringref))
		927	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
		928	{
		929	STRLEN len = decode_uint (dec);
		930	char *key = (char *)dec->cur;
		931
		932	WANT (len);
		933	dec->cur += len;
		934
		935	hv_store (hv, key, len, decode_sv (dec), 0);
		936
		937	return;
		938	}
		939	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
		940	{
		941	STRLEN len = decode_uint (dec);
		942	char *key = (char *)dec->cur;
		943
		944	WANT (len);
		945	dec->cur += len;
		946
		947	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		948	if (!is_utf8_string (key, len))
		949	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
		950
		951	hv_store (hv, key, -len, decode_sv (dec), 0);
		952
		953	return;
		954	}
		955
		956	SV *k = decode_sv (dec);
		957	SV *v = decode_sv (dec);
		958
		959	// we leak memory if uncaught exceptions are thrown by random magical
		960	// methods, and this is hopefully the only place where it can happen,
		961	// so if there is a chance of an exception, take the very slow path.
		962	// since catching exceptions is "undocumented/internal/forbidden" by
		963	// the new p5p powers, we need to call out to a perl function :/
		964	if (ecb_expect_false (SvAMAGIC (k)))
		965	{
		966	dSP;
		967
		968	ENTER; SAVETMPS;
		969	PUSHMARK (SP);
		970	EXTEND (SP, 3);
		971	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		972	PUSHs (sv_2mortal (k));
		973	PUSHs (sv_2mortal (v));
		974
		975	PUTBACK;
		976	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		977	SPAGAIN;
		978
		979	FREETMPS; LEAVE;
		980
		981	if (SvTRUE (ERRSV))
		982	ERR_ERRSV;
		983
		984	return;
		985	}
		986
		987	hv_store_ent (hv, k, v, 0);
		988	SvREFCNT_dec_NN (k);
		989
		990	fail:
		991	;
		992	}
		993
589	static SV *	994	static SV *
590	decode_hv (dec_t *dec)	995	decode_hv (dec_t *dec)
591	{	996	{
592	HV *hv = newHV ();	997	HV *hv = newHV ();
593		998
594	DEC_INC_DEPTH;	999	DEC_INC_DEPTH;
595		1000
596	if ((*dec->cur & 31) == 31)	1001	if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
597	{	1002	{
598	++dec->cur;	1003	++dec->cur;
599		1004
600	for (;;)	1005	for (;;)
601	{	1006	{
602	WANT (1);	1007	WANT (1);
603		1008
604	if (*dec->cur == (0xe0 | 31))	1009	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
605	{	1010	{
606	++dec->cur;	1011	++dec->cur;
607	break;	1012	break;
608	}	1013	}
609		1014
610	SV *k = decode_sv (dec);	1015	decode_he (dec, hv);
611	SV *v = decode_sv (dec);
612
613	hv_store_ent (hv, k, v, 0);
614	SvREFCNT_dec (k);
615	}	1016	}
616	}	1017	}
617	else	1018	else
618	{	1019	{
619	int len = decode_uint (dec);	1020	UV pairs = decode_uint (dec);
620		1021
		1022	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
		1023
621	while (len--)	1024	while (pairs--)
622	{	1025	decode_he (dec, hv);
623	SV *k = decode_sv (dec);
624	SV *v = decode_sv (dec);
625
626	hv_store_ent (hv, k, v, 0);
627	SvREFCNT_dec (k);
628	}
629	}	1026	}
630		1027
631	DEC_DEC_DEPTH;	1028	DEC_DEC_DEPTH;
632	return newRV_noinc ((SV *)hv);	1029	return newRV_noinc ((SV *)hv);
633		1030
634	fail:	1031	fail:
635	SvREFCNT_dec (hv);	1032	SvREFCNT_dec_NN (hv);
636	DEC_DEC_DEPTH;	1033	DEC_DEC_DEPTH;
637	return &PL_sv_undef;	1034	return &PL_sv_undef;
638	}	1035	}
639		1036
640	static SV *	1037	static SV *
641	decode_str (dec_t *dec, int utf8)	1038	decode_str (dec_t *dec, int utf8)
642	{	1039	{
643	SV *sv = 0;	1040	SV *sv = 0;
644		1041
645	if ((*dec->cur & 31) == 31)	1042	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
646	{	1043	{
		1044	// indefinite length strings
647	++dec->cur;	1045	++dec->cur;
648		1046
		1047	U8 major = *dec->cur & MAJOR_MISC;
		1048
649	sv = newSVpvn ("", 0);	1049	sv = newSVpvn ("", 0);
650		1050
651	// not very fast, and certainly not robust against illegal input
652	for (;;)	1051	for (;;)
653	{	1052	{
654	WANT (1);	1053	WANT (1);
655		1054
656	if (*dec->cur == (0xe0 | 31))	1055	if ((*dec->cur - major) > LENGTH_EXT8)
		1056	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657	{	1057	{
658	++dec->cur;	1058	++dec->cur;
659	break;	1059	break;
660	}	1060	}
		1061	else
		1062	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
661		1063
662	sv_catsv (sv, decode_sv (dec));	1064	STRLEN len = decode_uint (dec);
		1065
		1066	WANT (len);
		1067	sv_catpvn (sv, dec->cur, len);
		1068	dec->cur += len;
663	}	1069	}
664	}	1070	}
665	else	1071	else
666	{	1072	{
667	STRLEN len = decode_uint (dec);	1073	STRLEN len = decode_uint (dec);
668		1074
669	WANT (len);	1075	WANT (len);
670	sv = newSVpvn (dec->cur, len);	1076	sv = newSVpvn (dec->cur, len);
671	dec->cur += len;	1077	dec->cur += len;
		1078
		1079	if (ecb_expect_false (dec->stringref)
		1080	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		1081	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
672	}	1082	}
673		1083
674	if (utf8)	1084	if (utf8)
		1085	{
		1086	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1087	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1088	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1089
675	SvUTF8_on (sv);	1090	SvUTF8_on (sv);
		1091	}
676		1092
677	return sv;	1093	return sv;
678		1094
679	fail:	1095	fail:
680	SvREFCNT_dec (sv);	1096	SvREFCNT_dec (sv);
…		…
682	}	1098	}
683		1099
684	static SV *	1100	static SV *
685	decode_tagged (dec_t *dec)	1101	decode_tagged (dec_t *dec)
686	{	1102	{
		1103	SV *sv = 0;
687	UV tag = decode_uint (dec);	1104	UV tag = decode_uint (dec);
		1105
		1106	WANT (1);
		1107
		1108	switch (tag)
		1109	{
		1110	case CBOR_TAG_MAGIC:
688	SV *sv = decode_sv (dec);	1111	sv = decode_sv (dec);
		1112	break;
689		1113
690	if (tag == CBOR_TAG_MAGIC)	1114	case CBOR_TAG_INDIRECTION:
691	return sv;	1115	sv = newRV_noinc (decode_sv (dec));
692	else if (tag == CBOR_TAG_PERL_OBJECT)	1116	break;
693	{	1117
		1118	case CBOR_TAG_STRINGREF_NAMESPACE:
		1119	{
		1120	// do not use SAVETMPS/FREETMPS, as these will
		1121	// erase mortalised caches, e.g. "shareable"
		1122	ENTER;
		1123
		1124	SAVESPTR (dec->stringref);
		1125	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
		1126
		1127	sv = decode_sv (dec);
		1128
		1129	LEAVE;
		1130	}
		1131	break;
		1132
		1133	case CBOR_TAG_STRINGREF:
		1134	{
		1135	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1136	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
		1137
		1138	UV idx = decode_uint (dec);
		1139
		1140	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
		1141	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
		1142
		1143	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
		1144	}
		1145	break;
		1146
		1147	case CBOR_TAG_VALUE_SHAREABLE:
		1148	{
		1149	if (ecb_expect_false (!dec->shareable))
		1150	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
		1151
		1152	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1153	{
		1154	sv = newSV (0);
		1155	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
		1156
		1157	SV *osv = decode_sv (dec);
		1158	sv_setsv (sv, osv);
		1159	SvREFCNT_dec_NN (osv);
		1160	}
		1161	else
		1162	{
		1163	av_push (dec->shareable, &PL_sv_undef);
		1164	int idx = AvFILLp (dec->shareable);
		1165	sv = decode_sv (dec);
		1166	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1167	}
		1168	}
		1169	break;
		1170
		1171	case CBOR_TAG_VALUE_SHAREDREF:
		1172	{
		1173	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
		1174	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
		1175
		1176	UV idx = decode_uint (dec);
		1177
		1178	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
		1179	ERR ("corrupted CBOR data (sharedref index out of bounds)");
		1180
		1181	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1182
		1183	if (sv == &PL_sv_undef)
		1184	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
		1185	}
		1186	break;
		1187
		1188	case CBOR_TAG_PERL_OBJECT:
		1189	{
		1190	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1191	goto filter;
		1192
		1193	sv = decode_sv (dec);
		1194
694	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1195	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
695	ERR ("corrupted CBOR data (non-array perl object)");	1196	ERR ("corrupted CBOR data (non-array perl object)");
696		1197
697	AV *av = (AV *)SvRV (sv);	1198	AV *av = (AV *)SvRV (sv);
698	int len = av_len (av) + 1;	1199	int len = av_len (av) + 1;
699	HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);	1200	HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
700		1201
701	if (!stash)	1202	if (!stash)
702	ERR ("cannot decode perl-object (package does not exist)");	1203	ERR ("cannot decode perl-object (package does not exist)");
703		1204
704	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);	1205	GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
705		1206
706	if (!method)	1207	if (!method)
707	ERR ("cannot decode perl-object (package does not have a THAW method)");	1208	ERR ("cannot decode perl-object (package does not have a THAW method)");
708		1209
709	dSP;	1210	dSP;
710		1211
711	ENTER; SAVETMPS; PUSHMARK (SP);	1212	ENTER; SAVETMPS;
		1213	PUSHMARK (SP);
712	EXTEND (SP, len + 1);	1214	EXTEND (SP, len + 1);
713	// we re-bless the reference to get overload and other niceties right	1215	// we re-bless the reference to get overload and other niceties right
714	PUSHs (*av_fetch (av, 0, 1));	1216	PUSHs (*av_fetch (av, 0, 1));
715	PUSHs (sv_cbor);	1217	PUSHs (sv_cbor);
716		1218
717	int i;	1219	int i;
718		1220
719	for (i = 1; i < len; ++i)	1221	for (i = 1; i < len; ++i)
720	PUSHs (*av_fetch (av, i, 1));	1222	PUSHs (*av_fetch (av, i, 1));
721		1223
722	PUTBACK;	1224	PUTBACK;
723	call_sv ((SV *)GvCV (method), G_SCALAR);	1225	call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
724	SPAGAIN;	1226	SPAGAIN;
725		1227
		1228	if (SvTRUE (ERRSV))
		1229	{
		1230	FREETMPS; LEAVE;
		1231	ERR_ERRSV;
		1232	}
		1233
726	SvREFCNT_dec (sv);	1234	SvREFCNT_dec_NN (sv);
727	sv = SvREFCNT_inc (POPs);	1235	sv = SvREFCNT_inc (POPs);
728		1236
729	PUTBACK;	1237	PUTBACK;
730		1238
731	FREETMPS; LEAVE;	1239	FREETMPS; LEAVE;
		1240	}
		1241	break;
732		1242
733	return sv;	1243	default:
734	}	1244	filter:
735	else	1245	{
736	{	1246	SV *tag_sv = newSVuv (tag);
		1247
		1248	sv = decode_sv (dec);
		1249
		1250	dSP;
		1251	ENTER; SAVETMPS;
		1252	PUSHMARK (SP);
		1253	EXTEND (SP, 2);
		1254	PUSHs (tag_sv);
		1255	PUSHs (sv);
		1256
		1257	PUTBACK;
		1258	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
		1259	SPAGAIN;
		1260
		1261	if (SvTRUE (ERRSV))
		1262	{
		1263	SvREFCNT_dec_NN (tag_sv);
		1264	FREETMPS; LEAVE;
		1265	ERR_ERRSV;
		1266	}
		1267
		1268	if (count)
		1269	{
		1270	SvREFCNT_dec_NN (tag_sv);
		1271	SvREFCNT_dec_NN (sv);
		1272	sv = SvREFCNT_inc_NN (TOPs);
		1273	SP -= count;
		1274	}
		1275	else
		1276	{
737	AV *av = newAV ();	1277	AV *av = newAV ();
738	av_push (av, newSVuv (tag));	1278	av_push (av, tag_sv);
739	av_push (av, sv);	1279	av_push (av, sv);
740		1280
741	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1281	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
742	? cbor_tagged_stash	1282	? cbor_tagged_stash
743	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1283	: gv_stashpv ("CBOR::XS::Tagged" , 1);
744
745	return sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1284	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1285	}
		1286
		1287	PUTBACK;
		1288
		1289	FREETMPS; LEAVE;
		1290	}
		1291	break;
746	}	1292	}
		1293
		1294	return sv;
747		1295
748	fail:	1296	fail:
749	SvREFCNT_dec (sv);	1297	SvREFCNT_dec (sv);
750	return &PL_sv_undef;	1298	return &PL_sv_undef;
751	}	1299	}
…		…
753	static SV *	1301	static SV *
754	decode_sv (dec_t *dec)	1302	decode_sv (dec_t *dec)
755	{	1303	{
756	WANT (1);	1304	WANT (1);
757		1305
758	switch (*dec->cur >> 5)	1306	switch (*dec->cur >> MAJOR_SHIFT)
759	{	1307	{
760	case 0: // unsigned int	1308	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
761	return newSVuv (decode_uint (dec));	1309	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
762	case 1: // negative int	1310	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
763	return newSViv (-1 - (IV)decode_uint (dec));	1311	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
764	case 2: // octet string	1312	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
765	return decode_str (dec, 0);	1313	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
766	case 3: // utf-8 string	1314	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
767	return decode_str (dec, 1);	1315
768	case 4: // array	1316	case MAJOR_MISC >> MAJOR_SHIFT:
769	return decode_av (dec);
770	case 5: // map
771	return decode_hv (dec);
772	case 6: // tag
773	return decode_tagged (dec);
774	case 7: // misc
775	switch (*dec->cur++ & 31)	1317	switch (*dec->cur++ & MINOR_MASK)
776	{	1318	{
777	case 20:	1319	case SIMPLE_FALSE:
778	#if CBOR_SLOW	1320	#if CBOR_SLOW
779	types_false = get_bool ("Types::Serialiser::false");	1321	types_false = get_bool ("Types::Serialiser::false");
780	#endif	1322	#endif
781	return newSVsv (types_false);	1323	return newSVsv (types_false);
782	case 21:	1324	case SIMPLE_TRUE:
783	#if CBOR_SLOW	1325	#if CBOR_SLOW
784	types_true = get_bool ("Types::Serialiser::true");	1326	types_true = get_bool ("Types::Serialiser::true");
785	#endif	1327	#endif
786	return newSVsv (types_true);	1328	return newSVsv (types_true);
787	case 22:	1329	case SIMPLE_NULL:
788	return newSVsv (&PL_sv_undef);	1330	return newSVsv (&PL_sv_undef);
789	case 23:	1331	case SIMPLE_UNDEF:
790	#if CBOR_SLOW	1332	#if CBOR_SLOW
791	types_error = get_bool ("Types::Serialiser::error");	1333	types_error = get_bool ("Types::Serialiser::error");
792	#endif	1334	#endif
793	return newSVsv (types_error);	1335	return newSVsv (types_error);
794		1336
795	case 25:	1337	case MISC_FLOAT16:
796	{	1338	{
797	WANT (2);	1339	WANT (2);
798		1340
799	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];	1341	uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
800	dec->cur += 2;	1342	dec->cur += 2;
801		1343
802	return newSVnv (ecb_binary16_to_float (fp));	1344	return newSVnv (ecb_binary16_to_float (fp));
803	}	1345	}
804		1346
805	case 26:	1347	case MISC_FLOAT32:
806	{	1348	{
807	uint32_t fp;	1349	uint32_t fp;
808	WANT (4);	1350	WANT (4);
809	memcpy (&fp, dec->cur, 4);	1351	memcpy (&fp, dec->cur, 4);
810	dec->cur += 4;	1352	dec->cur += 4;
…		…
813	fp = ecb_bswap32 (fp);	1355	fp = ecb_bswap32 (fp);
814		1356
815	return newSVnv (ecb_binary32_to_float (fp));	1357	return newSVnv (ecb_binary32_to_float (fp));
816	}	1358	}
817		1359
818	case 27:	1360	case MISC_FLOAT64:
819	{	1361	{
820	uint64_t fp;	1362	uint64_t fp;
821	WANT (8);	1363	WANT (8);
822	memcpy (&fp, dec->cur, 8);	1364	memcpy (&fp, dec->cur, 8);
823	dec->cur += 8;	1365	dec->cur += 8;
…		…
826	fp = ecb_bswap64 (fp);	1368	fp = ecb_bswap64 (fp);
827		1369
828	return newSVnv (ecb_binary64_to_double (fp));	1370	return newSVnv (ecb_binary64_to_double (fp));
829	}	1371	}
830		1372
831	// 0..19 unassigned	1373	// 0..19 unassigned simple
832	// 24 reserved + unassigned (reserved values are not encodable)	1374	// 24 reserved + unassigned simple (reserved values are not encodable)
		1375	// 28-30 unassigned misc
		1376	// 31 break code
833	default:	1377	default:
834	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1378	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
835	}	1379	}
836		1380
837	break;	1381	break;
838	}	1382	}
839		1383
…		…
842	}	1386	}
843		1387
844	static SV *	1388	static SV *
845	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1389	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
846	{	1390	{
847	dec_t dec;	1391	dec_t dec = { 0 };
848	SV *sv;	1392	SV *sv;
		1393	STRLEN len;
		1394	char *data = SvPVbyte (string, len);
849		1395
850	/* work around bugs in 5.10 where manipulating magic values
851	* makes perl ignore the magic in subsequent accesses.
852	* also make a copy of non-PV values, to get them into a clean
853	* state (SvPV should do that, but it's buggy, see below).
854	*/
855	/*SvGETMAGIC (string);*/
856	if (SvMAGICAL (string) || !SvPOK (string))
857	string = sv_2mortal (newSVsv (string));
858
859	SvUPGRADE (string, SVt_PV);
860
861	/* work around a bug in perl 5.10, which causes SvCUR to fail an
862	* assertion with -DDEBUGGING, although SvCUR is documented to
863	* return the xpv_cur field which certainly exists after upgrading.
864	* according to nicholas clark, calling SvPOK fixes this.
865	* But it doesn't fix it, so try another workaround, call SvPV_nolen
866	* and hope for the best.
867	* Damnit, SvPV_nolen still trips over yet another assertion. This
868	* assertion business is seriously broken, try yet another workaround
869	* for the broken -DDEBUGGING.
870	*/
871	{
872	#ifdef DEBUGGING
873	STRLEN offset = SvOK (string) ? sv_len (string) : 0;
874	#else
875	STRLEN offset = SvCUR (string);
876	#endif
877
878	if (offset > cbor->max_size && cbor->max_size)	1396	if (len > cbor->max_size && cbor->max_size)
879	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",	1397	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
880	(unsigned long)SvCUR (string), (unsigned long)cbor->max_size);	1398	(unsigned long)len, (unsigned long)cbor->max_size);
881	}
882
883	sv_utf8_downgrade (string, 0);
884		1399
885	dec.cbor = *cbor;	1400	dec.cbor = *cbor;
886	dec.cur = (U8 *)SvPVX (string);	1401	dec.cur = (U8 *)data;
887	dec.end = (U8 *)SvEND (string);	1402	dec.end = (U8 *)data + len;
888	dec.err = 0;
889	dec.depth = 0;
890		1403
891	sv = decode_sv (&dec);	1404	sv = decode_sv (&dec);
892		1405
893	if (offset_return)	1406	if (offset_return)
894	*offset_return = dec.cur;	1407	*offset_return = dec.cur;
…		…
897	if (dec.cur != dec.end && !dec.err)	1410	if (dec.cur != dec.end && !dec.err)
898	dec.err = "garbage after CBOR object";	1411	dec.err = "garbage after CBOR object";
899		1412
900	if (dec.err)	1413	if (dec.err)
901	{	1414	{
		1415	if (dec.shareable)
		1416	{
		1417	// need to break cyclic links, which would all be in shareable
		1418	int i;
		1419	SV **svp;
		1420
		1421	for (i = av_len (dec.shareable) + 1; i--; )
		1422	if ((svp = av_fetch (dec.shareable, i, 0)))
		1423	sv_setsv (*svp, &PL_sv_undef);
		1424	}
		1425
902	SvREFCNT_dec (sv);	1426	SvREFCNT_dec_NN (sv);
		1427
		1428	if (dec.err_sv)
		1429	sv_2mortal (dec.err_sv);
		1430
903	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur);	1431	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
904	}	1432	}
905		1433
906	sv = sv_2mortal (sv);	1434	sv = sv_2mortal (sv);
907		1435
908	return sv;	1436	return sv;
909	}	1437	}
910		1438
		1439	/////////////////////////////////////////////////////////////////////////////
		1440	// incremental parser
		1441
		1442	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1443
		1444	// returns 0 for notyet, 1 for success or error
		1445	static int
		1446	incr_parse (CBOR *self, SV *cborstr)
		1447	{
		1448	STRLEN cur;
		1449	SvPV (cborstr, cur);
		1450
		1451	while (ecb_expect_true (self->incr_need <= cur))
		1452	{
		1453	// table of integer count bytes
		1454	static I8 incr_len[MINOR_MASK + 1] = {
		1455	0, 0, 0, 0, 0, 0, 0, 0,
		1456	0, 0, 0, 0, 0, 0, 0, 0,
		1457	0, 0, 0, 0, 0, 0, 0, 0,
		1458	1, 2, 4, 8,-1,-1,-1,-2
		1459	};
		1460
		1461	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1462	U8 m = *p & MINOR_MASK;
		1463	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1464	I8 ilen = incr_len[m];
		1465
		1466	self->incr_need = self->incr_pos + 1;
		1467
		1468	if (ecb_expect_false (ilen < 0))
		1469	{
		1470	if (m != MINOR_INDEF)
		1471	return 1; // error
		1472
		1473	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1474	{
		1475	if (count >= 0)
		1476	return 1; // error
		1477
		1478	count = 1;
		1479	}
		1480	else
		1481	{
		1482	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1483	count = -1;
		1484	}
		1485	}
		1486	else
		1487	{
		1488	self->incr_need += ilen;
		1489	if (ecb_expect_false (self->incr_need > cur))
		1490	return 0;
		1491
		1492	int major = *p >> MAJOR_SHIFT;
		1493
		1494	switch (major)
		1495	{
		1496	case MAJOR_TAG >> MAJOR_SHIFT:
		1497	++count; // tags merely prefix another value
		1498	break;
		1499
		1500	case MAJOR_BYTES >> MAJOR_SHIFT:
		1501	case MAJOR_TEXT >> MAJOR_SHIFT:
		1502	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1503	case MAJOR_MAP >> MAJOR_SHIFT:
		1504	{
		1505	UV len;
		1506
		1507	if (ecb_expect_false (ilen))
		1508	{
		1509	len = 0;
		1510
		1511	do {
		1512	len = (len << 8) | *++p;
		1513	} while (--ilen);
		1514	}
		1515	else
		1516	len = m;
		1517
		1518	switch (major)
		1519	{
		1520	case MAJOR_BYTES >> MAJOR_SHIFT:
		1521	case MAJOR_TEXT >> MAJOR_SHIFT:
		1522	self->incr_need += len;
		1523	if (ecb_expect_false (self->incr_need > cur))
		1524	return 0;
		1525
		1526	break;
		1527
		1528	case MAJOR_MAP >> MAJOR_SHIFT:
		1529	len <<= 1;
		1530	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1531	if (len)
		1532	{
		1533	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1534	count = len + 1;
		1535	}
		1536	break;
		1537	}
		1538	}
		1539	}
		1540	}
		1541
		1542	self->incr_pos = self->incr_need;
		1543
		1544	if (count > 0)
		1545	{
		1546	while (!--count)
		1547	{
		1548	if (!AvFILLp (self->incr_count))
		1549	return 1; // done
		1550
		1551	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1552	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1553	}
		1554
		1555	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1556	}
		1557	}
		1558
		1559	return 0;
		1560	}
		1561
		1562
911	/////////////////////////////////////////////////////////////////////////////	1563	/////////////////////////////////////////////////////////////////////////////
912	// XS interface functions	1564	// XS interface functions
913		1565
914	MODULE = CBOR::XS PACKAGE = CBOR::XS	1566	MODULE = CBOR::XS PACKAGE = CBOR::XS
915		1567
…		…
923		1575
924	types_true = get_bool ("Types::Serialiser::true" );	1576	types_true = get_bool ("Types::Serialiser::true" );
925	types_false = get_bool ("Types::Serialiser::false");	1577	types_false = get_bool ("Types::Serialiser::false");
926	types_error = get_bool ("Types::Serialiser::error");	1578	types_error = get_bool ("Types::Serialiser::error");
927		1579
		1580	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
		1581
928	sv_cbor = newSVpv ("CBOR", 0);	1582	sv_cbor = newSVpv ("CBOR", 0);
929	SvREADONLY_on (sv_cbor);	1583	SvREADONLY_on (sv_cbor);
		1584
		1585	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
930	}	1586	}
931		1587
932	PROTOTYPES: DISABLE	1588	PROTOTYPES: DISABLE
933		1589
934	void CLONE (...)	1590	void CLONE (...)
…		…
952		1608
953	void shrink (CBOR *self, int enable = 1)	1609	void shrink (CBOR *self, int enable = 1)
954	ALIAS:	1610	ALIAS:
955	shrink = F_SHRINK	1611	shrink = F_SHRINK
956	allow_unknown = F_ALLOW_UNKNOWN	1612	allow_unknown = F_ALLOW_UNKNOWN
		1613	allow_sharing = F_ALLOW_SHARING
		1614	allow_cycles = F_ALLOW_CYCLES
		1615	forbid_objects = F_FORBID_OBJECTS
		1616	pack_strings = F_PACK_STRINGS
		1617	text_keys = F_TEXT_KEYS
		1618	text_strings = F_TEXT_STRINGS
		1619	validate_utf8 = F_VALIDATE_UTF8
957	PPCODE:	1620	PPCODE:
958	{	1621	{
959	if (enable)	1622	if (enable)
960	self->flags |= ix;	1623	self->flags |= ix;
961	else	1624	else
…		…
966		1629
967	void get_shrink (CBOR *self)	1630	void get_shrink (CBOR *self)
968	ALIAS:	1631	ALIAS:
969	get_shrink = F_SHRINK	1632	get_shrink = F_SHRINK
970	get_allow_unknown = F_ALLOW_UNKNOWN	1633	get_allow_unknown = F_ALLOW_UNKNOWN
		1634	get_allow_sharing = F_ALLOW_SHARING
		1635	get_allow_cycles = F_ALLOW_CYCLES
		1636	get_forbid_objects = F_FORBID_OBJECTS
		1637	get_pack_strings = F_PACK_STRINGS
		1638	get_text_keys = F_TEXT_KEYS
		1639	get_text_strings = F_TEXT_STRINGS
		1640	get_validate_utf8 = F_VALIDATE_UTF8
971	PPCODE:	1641	PPCODE:
972	XPUSHs (boolSV (self->flags & ix));	1642	XPUSHs (boolSV (self->flags & ix));
973		1643
974	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1644	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
975	PPCODE:	1645	PPCODE:
…		…
988	XPUSHs (ST (0));	1658	XPUSHs (ST (0));
989		1659
990	int get_max_size (CBOR *self)	1660	int get_max_size (CBOR *self)
991	CODE:	1661	CODE:
992	RETVAL = self->max_size;	1662	RETVAL = self->max_size;
		1663	OUTPUT:
		1664	RETVAL
		1665
		1666	void filter (CBOR *self, SV *filter = 0)
		1667	PPCODE:
		1668	SvREFCNT_dec (self->filter);
		1669	self->filter = filter ? newSVsv (filter) : filter;
		1670	XPUSHs (ST (0));
		1671
		1672	SV *get_filter (CBOR *self)
		1673	CODE:
		1674	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
993	OUTPUT:	1675	OUTPUT:
994	RETVAL	1676	RETVAL
995		1677
996	void encode (CBOR *self, SV *scalar)	1678	void encode (CBOR *self, SV *scalar)
997	PPCODE:	1679	PPCODE:
…		…
1012	EXTEND (SP, 2);	1694	EXTEND (SP, 2);
1013	PUSHs (sv);	1695	PUSHs (sv);
1014	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1696	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1015	}	1697	}
1016		1698
		1699	void incr_parse (CBOR *self, SV *cborstr)
		1700	ALIAS:
		1701	incr_parse_multiple = 1
		1702	PPCODE:
		1703	{
		1704	if (SvUTF8 (cborstr))
		1705	sv_utf8_downgrade (cborstr, 0);
		1706
		1707	if (!self->incr_count)
		1708	{
		1709	self->incr_count = newAV ();
		1710	self->incr_pos = 0;
		1711	self->incr_need = 1;
		1712
		1713	av_push (self->incr_count, newSViv (1));
		1714	}
		1715
		1716	do
		1717	{
		1718	if (!incr_parse (self, cborstr))
		1719	{
		1720	if (self->incr_need > self->max_size && self->max_size)
		1721	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1722	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1723
		1724	break;
		1725	}
		1726
		1727	SV *sv;
		1728	char *offset;
		1729
		1730	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1731	XPUSHs (sv);
		1732
		1733	sv_chop (cborstr, offset);
		1734
		1735	av_clear (self->incr_count);
		1736	av_push (self->incr_count, newSViv (1));
		1737
		1738	self->incr_pos = 0;
		1739	self->incr_need = self->incr_pos + 1;
		1740	}
		1741	while (ix);
		1742	}
		1743
		1744	void incr_reset (CBOR *self)
		1745	CODE:
		1746	{
		1747	SvREFCNT_dec (self->incr_count);
		1748	self->incr_count = 0;
		1749	}
		1750
		1751	void DESTROY (CBOR *self)
		1752	PPCODE:
		1753	cbor_free (self);
		1754
1017	PROTOTYPES: ENABLE	1755	PROTOTYPES: ENABLE
1018		1756
1019	void encode_cbor (SV *scalar)	1757	void encode_cbor (SV *scalar)
		1758	ALIAS:
		1759	encode_cbor = 0
		1760	encode_cbor_sharing = F_ALLOW_SHARING
1020	PPCODE:	1761	PPCODE:
1021	{	1762	{
1022	CBOR cbor;	1763	CBOR cbor;
1023	cbor_init (&cbor);	1764	cbor_init (&cbor);
		1765	cbor.flags |= ix;
1024	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1766	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1025	XPUSHs (scalar);	1767	XPUSHs (scalar);
1026	}	1768	}
1027		1769
1028	void decode_cbor (SV *cborstr)	1770	void decode_cbor (SV *cborstr)
…		…
1032	cbor_init (&cbor);	1774	cbor_init (&cbor);
1033	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1775	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1034	XPUSHs (cborstr);	1776	XPUSHs (cborstr);
1035	}	1777	}
1036		1778
		1779	#ifdef __AFL_COMPILER
		1780
		1781	void
		1782	afl_init ()
		1783	CODE:
		1784	__AFL_INIT ();
		1785
		1786	int
		1787	afl_loop (unsigned int count = 10000)
		1788	CODE:
		1789	RETVAL = __AFL_LOOP (count);
		1790	OUTPUT:
		1791	RETVAL
		1792
		1793	#endif
		1794

Diff Legend

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