ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC

…		…
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
		11	#include <inttypes.h>
11		12
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
24	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	31	#endif
		32
		33	// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
		34	#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
27		35
28	// known major and minor types	36	// known major and minor types
29	enum cbor_type	37	enum cbor_type
30	{	38	{
31	MAJOR_SHIFT = 5,	39	MAJOR_SHIFT = 5,
…		…
94	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	102	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95		103
96	CBOR_TAG_MAGIC = 55799, // self-describe cbor	104	CBOR_TAG_MAGIC = 55799, // self-describe cbor
97	};	105	};
98		106
		107	// known forced types, also hardcoded in CBOR.pm
		108	enum
		109	{
		110	AS_CBOR = 0,
		111	AS_INT = 1,
		112	AS_BYTES = 2,
		113	AS_TEXT = 3,
		114	AS_FLOAT16 = 4,
		115	AS_FLOAT32 = 5,
		116	AS_FLOAT64 = 6,
		117	AS_MAP = 7,
		118	// possibly future enhancements: (generic) float, (generic) string
		119	};
		120
99	#define F_SHRINK 0x00000001UL	121	#define F_SHRINK 0x00000001UL
100	#define F_ALLOW_UNKNOWN 0x00000002UL	122	#define F_ALLOW_UNKNOWN 0x00000002UL
101	#define F_ALLOW_SHARING 0x00000004UL	123	#define F_ALLOW_SHARING 0x00000004UL
		124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_ALLOW_WEAK_CYCLES 0x00000010UL
		126	#define F_FORBID_OBJECTS 0x00000020UL
102	#define F_PACK_STRINGS 0x00000008UL	127	#define F_PACK_STRINGS 0x00000040UL
		128	#define F_TEXT_KEYS 0x00000080UL
		129	#define F_TEXT_STRINGS 0x00000100UL
		130	#define F_VALIDATE_UTF8 0x00000200UL
103		131
104	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
105		133
106	#define SB do {	134	#define SB do {
107	#define SE } while (0)	135	#define SE } while (0)
…		…
126	typedef struct {	154	typedef struct {
127	U32 flags;	155	U32 flags;
128	U32 max_depth;	156	U32 max_depth;
129	STRLEN max_size;	157	STRLEN max_size;
130	SV *filter;	158	SV *filter;
		159
		160	// for the incremental parser
		161	STRLEN incr_pos; // the current offset into the text
		162	STRLEN incr_need; // minimum bytes needed to decode
		163	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131	} CBOR;	164	} CBOR;
132		165
133	ecb_inline void	166	ecb_inline void
134	cbor_init (CBOR *cbor)	167	cbor_init (CBOR *cbor)
135	{	168	{
…		…
139		172
140	ecb_inline void	173	ecb_inline void
141	cbor_free (CBOR *cbor)	174	cbor_free (CBOR *cbor)
142	{	175	{
143	SvREFCNT_dec (cbor->filter);	176	SvREFCNT_dec (cbor->filter);
		177	SvREFCNT_dec (cbor->incr_count);
144	}	178	}
145		179
146	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
147	// utility functions	181	// utility functions
148		182
…		…
171	#endif	205	#endif
172	}	206	}
173	}	207	}
174		208
175	// minimum length of a string to be registered for stringref	209	// minimum length of a string to be registered for stringref
176	ecb_inline int	210	ecb_inline STRLEN
177	minimum_string_length (UV idx)	211	minimum_string_length (UV idx)
178	{	212	{
179	return idx > 23	213	return idx <= 23 ? 3
180	? idx > 0xffU	214	: idx <= 0xffU ? 4
181	? idx > 0xffffU	215	: idx <= 0xffffU ? 5
182	? idx > 0xffffffffU	216	: idx <= 0xffffffffU ? 7
183	? 11	217	: 11;
184	: 7
185	: 5
186	: 4
187	: 3;
188	}	218	}
189		219
190	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
191	// encoder	221	// encoder
192		222
…		…
205	} enc_t;	235	} enc_t;
206		236
207	ecb_inline void	237	ecb_inline void
208	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
209	{	239	{
210	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211	{	241	{
212	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
215	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
216	}	246	}
217	}	247	}
218		248
		249	static void encode_sv (enc_t *enc, SV *sv);
		250
219	ecb_inline void	251	ecb_inline void
220	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
221	{	253	{
222	need (enc, 1);	254	need (enc, 1);
223	*enc->cur++ = ch;	255	*enc->cur++ = ch;
224	}	256	}
225		257
		258	// used for tags, intregers, element counts and so on
226	static void	259	static void
227	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
228	{	261	{
229	need (enc, 9);	262	need (enc, 9);
230		263
231	if (ecb_expect_true (len < LENGTH_EXT1))	264	if (ecb_expect_true (len < LENGTH_EXT1))
232	*enc->cur++ = major | len;	265	*enc->cur++ = major | len;
233	else if (ecb_expect_true (len <= 0xff))	266	else if (ecb_expect_true (len <= 0xffU))
234	{	267	{
235	*enc->cur++ = major | LENGTH_EXT1;	268	*enc->cur++ = major | LENGTH_EXT1;
236	*enc->cur++ = len;	269	*enc->cur++ = len;
237	}	270	}
238	else if (len <= 0xffff)	271	else if (len <= 0xffffU)
239	{	272	{
240	*enc->cur++ = major | LENGTH_EXT2;	273	*enc->cur++ = major | LENGTH_EXT2;
241	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
242	*enc->cur++ = len;	275	*enc->cur++ = len;
243	}	276	}
244	else if (len <= 0xffffffff)	277	else if (len <= 0xffffffffU)
245	{	278	{
246	*enc->cur++ = major | LENGTH_EXT4;	279	*enc->cur++ = major | LENGTH_EXT4;
247	*enc->cur++ = len >> 24;	280	*enc->cur++ = len >> 24;
248	*enc->cur++ = len >> 16;	281	*enc->cur++ = len >> 16;
249	*enc->cur++ = len >> 8;	282	*enc->cur++ = len >> 8;
…		…
261	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
262	*enc->cur++ = len;	295	*enc->cur++ = len;
263	}	296	}
264	}	297	}
265		298
		299	// encodes a perl value into a CBOR integer
		300	ecb_inline void
		301	encode_int (enc_t *enc, SV *sv)
		302	{
		303	if (SvIsUV (sv))
		304	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		305	else if (SvIVX (sv) >= 0)
		306	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		307	else
		308	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		309	}
		310
266	ecb_inline void	311	ecb_inline void
267	encode_tag (enc_t *enc, UV tag)	312	encode_tag (enc_t *enc, UV tag)
268	{	313	{
269	encode_uint (enc, MAJOR_TAG, tag);	314	encode_uint (enc, MAJOR_TAG, tag);
270	}	315	}
271		316
		317	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		318	ecb_noinline static void
		319	encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
		320	{
		321	STRLEN ulen = len;
		322	U8 *p, *pend = (U8 *)str + len;
		323
		324	for (p = (U8 *)str; p < pend; ++p)
		325	ulen += *p >> 7; // count set high bits
		326
		327	encode_uint (enc, MAJOR_TEXT, ulen);
		328
		329	need (enc, ulen);
		330	for (p = (U8 *)str; p < pend; ++p)
		331	if (*p < 0x80)
		332	*enc->cur++ = *p;
		333	else
		334	{
		335	*enc->cur++ = 0xc0 + (*p >> 6);
		336	*enc->cur++ = 0x80 + (*p & 63);
		337	}
		338	}
		339
272	ecb_inline void	340	ecb_inline void
273	encode_str (enc_t *enc, int utf8, char *str, STRLEN len)	341	encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
274	{	342	{
		343	if (ecb_expect_false (upgrade_utf8))
		344	if (!utf8)
		345	{
		346	encode_str_utf8 (enc, utf8, str, len);
		347	return;
		348	}
		349
275	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);	350	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276	need (enc, len);	351	need (enc, len);
277	memcpy (enc->cur, str, len);	352	memcpy (enc->cur, str, len);
278	enc->cur += len;	353	enc->cur += len;
279	}	354	}
280		355
281	static void	356	ecb_inline void
282	encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)	357	encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283	{	358	{
284	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))	359	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285	{	360	{
286	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	361	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287		362
…		…
298	sv_setuv (*svp, enc->stringref_idx);	373	sv_setuv (*svp, enc->stringref_idx);
299	++enc->stringref_idx;	374	++enc->stringref_idx;
300	}	375	}
301	}	376	}
302		377
303	encode_str (enc, utf8, str, len);	378	encode_str (enc, upgrade_utf8, utf8, str, len);
304	}	379	}
305		380
306	static void encode_sv (enc_t *enc, SV *sv);	381	ecb_inline void
		382	encode_float16 (enc_t *enc, NV nv)
		383	{
		384	need (enc, 1+2);
		385
		386	*enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
		387
		388	uint16_t fp = ecb_float_to_binary16 (nv);
		389
		390	if (!ecb_big_endian ())
		391	fp = ecb_bswap16 (fp);
		392
		393	memcpy (enc->cur, &fp, 2);
		394	enc->cur += 2;
		395	}
		396
		397	ecb_inline void
		398	encode_float32 (enc_t *enc, NV nv)
		399	{
		400	need (enc, 1+4);
		401
		402	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
		403
		404	uint32_t fp = ecb_float_to_binary32 (nv);
		405
		406	if (!ecb_big_endian ())
		407	fp = ecb_bswap32 (fp);
		408
		409	memcpy (enc->cur, &fp, 4);
		410	enc->cur += 4;
		411	}
		412
		413	ecb_inline void
		414	encode_float64 (enc_t *enc, NV nv)
		415	{
		416	need (enc, 1+8);
		417
		418	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
		419
		420	uint64_t fp = ecb_double_to_binary64 (nv);
		421
		422	if (!ecb_big_endian ())
		423	fp = ecb_bswap64 (fp);
		424
		425	memcpy (enc->cur, &fp, 8);
		426	enc->cur += 8;
		427	}
		428
		429	ecb_inline void
		430	encode_bool (enc_t *enc, int istrue)
		431	{
		432	encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
		433	}
		434
		435	// encodes an arrayref containing key-value pairs as CBOR map
		436	ecb_inline void
		437	encode_array_as_map (enc_t *enc, SV *sv)
		438	{
		439	if (enc->depth >= enc->cbor.max_depth)
		440	croak (ERR_NESTING_EXCEEDED);
		441
		442	++enc->depth;
		443
		444	// as_map does error checking for us, but we re-check in case
		445	// things have changed.
		446
		447	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
		448	croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
		449
		450	AV *av = (AV *)SvRV (sv);
		451	int i, len = av_len (av);
		452
		453	if (!(len & 1))
		454	croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
		455
		456	encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
		457
		458	for (i = 0; i <= len; ++i)
		459	{
		460	SV **svp = av_fetch (av, i, 0);
		461	encode_sv (enc, svp ? *svp : &PL_sv_undef);
		462	}
		463
		464	--enc->depth;
		465	}
		466
		467	ecb_inline void
		468	encode_forced (enc_t *enc, UV type, SV *sv)
		469	{
		470	switch (type)
		471	{
		472	case AS_CBOR:
		473	{
		474	STRLEN len;
		475	char *str = SvPVbyte (sv, len);
		476
		477	need (enc, len);
		478	memcpy (enc->cur, str, len);
		479	enc->cur += len;
		480	}
		481	break;
		482
		483	case AS_BYTES:
		484	{
		485	STRLEN len;
		486	char *str = SvPVbyte (sv, len);
		487	encode_strref (enc, 0, 0, str, len);
		488	}
		489	break;
		490
		491	case AS_TEXT:
		492	{
		493	STRLEN len;
		494	char *str = SvPVutf8 (sv, len);
		495	encode_strref (enc, 1, 1, str, len);
		496	}
		497	break;
		498
		499	case AS_INT: encode_int (enc, sv); break;
		500
		501	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		502	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		503	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		504
		505	case AS_MAP: encode_array_as_map (enc, sv); break;
		506
		507	default:
		508	croak ("encountered malformed CBOR::XS::Tagged object");
		509	}
		510	}
307		511
308	static void	512	static void
309	encode_av (enc_t *enc, AV *av)	513	encode_av (enc_t *enc, AV *av)
310	{	514	{
311	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
315		519
316	++enc->depth;	520	++enc->depth;
317		521
318	encode_uint (enc, MAJOR_ARRAY, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
319		523
		524	if (ecb_expect_false (SvMAGICAL (av)))
320	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
321	{	526	{
322	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
323	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
324	}	529	}
		530	else
		531	for (i = 0; i <= len; ++i)
		532	{
		533	SV *sv = AvARRAY (av)[i];
		534	encode_sv (enc, sv ? sv : &PL_sv_undef);
		535	}
325		536
326	--enc->depth;	537	--enc->depth;
327	}	538	}
328		539
329	static void	540	static void
…		…
337	++enc->depth;	548	++enc->depth;
338		549
339	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
340	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
341		552
342	if (mg)	553	if (ecb_expect_false (mg))
343	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);	554	encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
344	else	555	else
345	encode_uint (enc, MAJOR_MAP, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
346		557
347	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
348	{	559	{
349	if (HeKLEN (he) == HEf_SVKEY)	560	if (HeKLEN (he) == HEf_SVKEY)
350	encode_sv (enc, HeSVKEY (he));	561	encode_sv (enc, HeSVKEY (he));
351	else	562	else
352	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	563	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353		564
354	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	565	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355	}	566	}
356		567
357	if (mg)	568	if (ecb_expect_false (mg))
358	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);	569	encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
359		570
360	--enc->depth;	571	--enc->depth;
361	}	572	}
362		573
…		…
382		593
383	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
384		595
385	if (stash == boolean_stash)	596	if (stash == boolean_stash)
386	{	597	{
387	encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);	598	encode_bool (enc, SvIV (sv));
388	return;	599	return;
389	}	600	}
390	else if (stash == error_stash)	601	else if (stash == error_stash)
391	{	602	{
392	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);	603	encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
…		…
395	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
396	{	607	{
397	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
398	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
399		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
400	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));	615	encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
401	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));	616	encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
		617	break;
		618
		619	case 3-1:
		620	// a forced type [value, type, undef]
		621	encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
		622	break;
		623
		624	default:
		625	croak ("encountered malformed CBOR::XS::Tagged object");
		626	}
402		627
403	return;	628	return;
404	}	629	}
405	}	630	}
406		631
407	if (ecb_expect_false (SvREFCNT (sv) > 1)	632	if (ecb_expect_false (SvREFCNT (sv) > 1)
408	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))	633	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409	{	634	{
410	if (!enc->shareable)	635	if (ecb_expect_false (!enc->shareable))
411	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());	636	enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
412		637
413	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);	638	SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
414		639
415	if (SvOK (*svp))	640	if (SvOK (*svp))
…		…
429	if (ecb_expect_false (SvOBJECT (sv)))	654	if (ecb_expect_false (SvOBJECT (sv)))
430	{	655	{
431	HV *stash = SvSTASH (sv);	656	HV *stash = SvSTASH (sv);
432	GV *method;	657	GV *method;
433		658
		659	if (enc->cbor.flags & F_FORBID_OBJECTS)
		660	croak ("encountered object '%s', but forbid_objects is enabled",
		661	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
434	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435	{	663	{
436	dSP;	664	dSP;
437		665
438	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
439	// we re-bless the reference to get overload and other niceties right	668	// we re-bless the reference to get overload and other niceties right
440	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441		670
442	PUTBACK;	671	PUTBACK;
443	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
456	}	685	}
457	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458	{	687	{
459	dSP;	688	dSP;
460		689
461	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
462	EXTEND (SP, 2);	692	EXTEND (SP, 2);
463	// we re-bless the reference to get overload and other niceties right	693	// we re-bless the reference to get overload and other niceties right
464	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
466		696
…		…
472	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
473	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	703	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474		704
475	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476	encode_uint (enc, MAJOR_ARRAY, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
477	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478		708
479	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
480	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
481		717
482	PUTBACK;	718	PUTBACK;
483		719
484	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
485	}	721	}
…		…
507		743
508	if (ecb_expect_false (nv == (NV)(U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
509	encode_uint (enc, MAJOR_POS_INT, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510	//TODO: maybe I32?	746	//TODO: maybe I32?
511	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
512	{	748	encode_float32 (enc, nv);
513	uint32_t fp = ecb_float_to_binary32 (nv);
514
515	*enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
516
517	if (!ecb_big_endian ())
518	fp = ecb_bswap32 (fp);
519
520	memcpy (enc->cur, &fp, 4);
521	enc->cur += 4;
522	}
523	else	749	else
524	{	750	encode_float64 (enc, nv);
525	uint64_t fp = ecb_double_to_binary64 (nv);
526
527	*enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
528
529	if (!ecb_big_endian ())
530	fp = ecb_bswap64 (fp);
531
532	memcpy (enc->cur, &fp, 8);
533	enc->cur += 8;
534	}
535	}	751	}
536		752
537	static void	753	static void
538	encode_sv (enc_t *enc, SV *sv)	754	encode_sv (enc_t *enc, SV *sv)
539	{	755	{
…		…
541		757
542	if (SvPOKp (sv))	758	if (SvPOKp (sv))
543	{	759	{
544	STRLEN len;	760	STRLEN len;
545	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
546	encode_strref (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547	}	763	}
548	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
549	encode_nv (enc, sv);	765	encode_nv (enc, sv);
550	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
551	{	767	encode_int (enc, sv);
552	if (SvIsUV (sv))
553	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
554	else if (SvIVX (sv) >= 0)
555	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
556	else
557	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
558	}
559	else if (SvROK (sv))	768	else if (SvROK (sv))
560	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
561	else if (!SvOK (sv))	770	else if (!SvOK (sv))
562	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);	771	encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
563	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
…		…
568	}	777	}
569		778
570	static SV *	779	static SV *
571	encode_cbor (SV *scalar, CBOR *cbor)	780	encode_cbor (SV *scalar, CBOR *cbor)
572	{	781	{
573	enc_t enc = { };	782	enc_t enc = { 0 };
574		783
575	enc.cbor = *cbor;	784	enc.cbor = *cbor;
576	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
578	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
579		788
580	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
581		790
582	if (cbor->flags & F_PACK_STRINGS)	791	if (cbor->flags & F_PACK_STRINGS)
583	{	792	{
…		…
610	U32 depth; // recursion depth	819	U32 depth; // recursion depth
611	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
612	AV *shareable;	821	AV *shareable;
613	AV *stringref;	822	AV *stringref;
614	SV *decode_tagged;	823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
615	} dec_t;	825	} dec_t;
616		826
617	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	827	// set dec->err to ERRSV
		828	ecb_cold static void
		829	err_errsv (dec_t *dec)
		830	{
		831	if (!dec->err)
		832	{
		833	dec->err_sv = newSVsv (ERRSV);
618		834
619	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	835	// chop off the trailing \n
		836	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		837	*SvEND (dec->err_sv) = 0;
620		838
		839	dec->err = SvPVutf8_nolen (dec->err_sv);
		840	}
		841	}
		842
		843	// the following functions are used to reduce code size and help the compiler to optimise
		844	ecb_cold static void
		845	err_set (dec_t *dec, const char *reason)
		846	{
		847	if (!dec->err)
		848	dec->err = reason;
		849	}
		850
		851	ecb_cold static void
		852	err_unexpected_end (dec_t *dec)
		853	{
		854	err_set (dec, "unexpected end of CBOR data");
		855	}
		856
		857	#define ERR_DO(do) SB do; goto fail; SE
		858	#define ERR(reason) ERR_DO (err_set (dec, reason))
		859	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		860
		861	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		862
621	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	863	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
623		865
624	static UV	866	static UV
625	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
626	{	868	{
627	U8 m = *dec->cur & MINOR_MASK;	869	U8 m = *dec->cur & MINOR_MASK;
628	++dec->cur;	870	++dec->cur;
629		871
630	if (ecb_expect_true (m < LENGTH_EXT1))	872	if (ecb_expect_true (m < LENGTH_EXT1))
631	return m;	873	return m;
632		874	else if (ecb_expect_true (m == LENGTH_EXT1))
633	switch (m)
634	{	875	{
635	case LENGTH_EXT1:
636	WANT (1);	876	WANT (1);
637	dec->cur += 1;	877	dec->cur += 1;
638	return dec->cur[-1];	878	return dec->cur[-1];
639		879	}
640	case LENGTH_EXT2:	880	else if (ecb_expect_true (m == LENGTH_EXT2))
		881	{
641	WANT (2);	882	WANT (2);
642	dec->cur += 2;	883	dec->cur += 2;
643	return (((UV)dec->cur[-2]) << 8)	884	return (((UV)dec->cur[-2]) << 8)
644	| ((UV)dec->cur[-1]);	885	| ((UV)dec->cur[-1]);
645		886	}
646	case LENGTH_EXT4:	887	else if (ecb_expect_true (m == LENGTH_EXT4))
		888	{
647	WANT (4);	889	WANT (4);
648	dec->cur += 4;	890	dec->cur += 4;
649	return (((UV)dec->cur[-4]) << 24)	891	return (((UV)dec->cur[-4]) << 24)
650	| (((UV)dec->cur[-3]) << 16)	892	| (((UV)dec->cur[-3]) << 16)
651	| (((UV)dec->cur[-2]) << 8)	893	| (((UV)dec->cur[-2]) << 8)
652	| ((UV)dec->cur[-1]);	894	| ((UV)dec->cur[-1]);
653		895	}
654	case LENGTH_EXT8:	896	else if (ecb_expect_true (m == LENGTH_EXT8))
		897	{
655	WANT (8);	898	WANT (8);
656	dec->cur += 8;	899	dec->cur += 8;
657		900
658	return	901	return
659	#if UVSIZE < 8	902	#if UVSIZE < 8
660	0	903	0
661	#else	904	#else
662	(((UV)dec->cur[-8]) << 56)	905	(((UV)dec->cur[-8]) << 56)
663	| (((UV)dec->cur[-7]) << 48)	906	| (((UV)dec->cur[-7]) << 48)
664	| (((UV)dec->cur[-6]) << 40)	907	| (((UV)dec->cur[-6]) << 40)
665	| (((UV)dec->cur[-5]) << 32)	908	| (((UV)dec->cur[-5]) << 32)
666	#endif	909	#endif
667	| (((UV)dec->cur[-4]) << 24)	910	| (((UV)dec->cur[-4]) << 24)
668	| (((UV)dec->cur[-3]) << 16)	911	| (((UV)dec->cur[-3]) << 16)
669	| (((UV)dec->cur[-2]) << 8)	912	| (((UV)dec->cur[-2]) << 8)
670	| ((UV)dec->cur[-1]);	913	| ((UV)dec->cur[-1]);
671		914	}
672	default:	915	else
673	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	916	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674	}
675		917
676	fail:	918	fail:
677	return 0;	919	return 0;
678	}	920	}
679		921
…		…
692		934
693	for (;;)	935	for (;;)
694	{	936	{
695	WANT (1);	937	WANT (1);
696		938
697	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))	939	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
698	{	940	{
699	++dec->cur;	941	++dec->cur;
700	break;	942	break;
701	}	943	}
702		944
703	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
704	}	946	}
705	}	947	}
706	else	948	else
707	{	949	{
708	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
709		951
		952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
710	av_fill (av, len - 1);	953	av_fill (av, len - 1);
711		954
712	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
713	AvARRAY (av)[i] = decode_sv (dec);	956	AvARRAY (av)[i] = decode_sv (dec);
714	}	957	}
715		958
716	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
717	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
718		961
719	fail:	962	fail:
720	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
721	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
722	return &PL_sv_undef;	965	return &PL_sv_undef;
723	}	966	}
724		967
725	static void	968	static void
…		…
727	{	970	{
728	// for speed reasons, we specialcase single-string	971	// for speed reasons, we specialcase single-string
729	// byte or utf-8 strings as keys, but only when !stringref	972	// byte or utf-8 strings as keys, but only when !stringref
730		973
731	if (ecb_expect_true (!dec->stringref))	974	if (ecb_expect_true (!dec->stringref))
732	if ((*dec->cur - MAJOR_BYTES) <= 27)	975	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733	{	976	{
734	I32 len = decode_uint (dec);	977	STRLEN len = decode_uint (dec);
735	char *key = (char *)dec->cur;	978	char *key = (char *)dec->cur;
736		979
		980	WANT (len);
737	dec->cur += len;	981	dec->cur += len;
738		982
739	if (ecb_expect_false (dec->stringref))
740	av_push (dec->stringref, newSVpvn (key, len));
741
742	hv_store (hv, key, len, decode_sv (dec), 0);	983	hv_store (hv, key, len, decode_sv (dec), 0);
743		984
744	return;	985	return;
745	}	986	}
746	else if ((*dec->cur - MAJOR_TEXT) <= 27)	987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747	{	988	{
748	I32 len = decode_uint (dec);	989	STRLEN len = decode_uint (dec);
749	char *key = (char *)dec->cur;	990	char *key = (char *)dec->cur;
750		991
		992	WANT (len);
751	dec->cur += len;	993	dec->cur += len;
752		994
753	if (ecb_expect_false (dec->stringref))	995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	996	if (!cbor_is_utf8_string ((U8 *)key, len))
		997	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755		998
756	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
757		1000
758	return;	1001	return;
759	}	1002	}
760		1003
761	SV *k = decode_sv (dec);	1004	SV *k = decode_sv (dec);
762	SV *v = decode_sv (dec);	1005	SV *v = decode_sv (dec);
763		1006
		1007	// we leak memory if uncaught exceptions are thrown by random magical
		1008	// methods, and this is hopefully the only place where it can happen,
		1009	// so if there is a chance of an exception, take the very slow path.
		1010	// since catching exceptions is "undocumented/internal/forbidden" by
		1011	// the new p5p powers, we need to call out to a perl function :/
		1012	if (ecb_expect_false (SvAMAGIC (k)))
		1013	{
		1014	dSP;
		1015
		1016	ENTER; SAVETMPS;
		1017	PUSHMARK (SP);
		1018	EXTEND (SP, 3);
		1019	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		1020	PUSHs (sv_2mortal (k));
		1021	PUSHs (sv_2mortal (v));
		1022
		1023	PUTBACK;
		1024	call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
		1025	SPAGAIN;
		1026
		1027	FREETMPS; LEAVE;
		1028
		1029	if (SvTRUE (ERRSV))
		1030	ERR_ERRSV;
		1031
		1032	return;
		1033	}
		1034
764	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
765	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
		1037
		1038	fail:
		1039	;
766	}	1040	}
767		1041
768	static SV *	1042	static SV *
769	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
770	{	1044	{
…		…
778		1052
779	for (;;)	1053	for (;;)
780	{	1054	{
781	WANT (1);	1055	WANT (1);
782		1056
783	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))	1057	if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
784	{	1058	{
785	++dec->cur;	1059	++dec->cur;
786	break;	1060	break;
787	}	1061	}
788		1062
789	decode_he (dec, hv);	1063	decode_he (dec, hv);
790	}	1064	}
791	}	1065	}
792	else	1066	else
793	{	1067	{
794	int pairs = decode_uint (dec);	1068	UV pairs = decode_uint (dec);
		1069
		1070	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795		1071
796	while (pairs--)	1072	while (pairs--)
797	decode_he (dec, hv);	1073	decode_he (dec, hv);
798	}	1074	}
799		1075
800	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
801	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
802		1078
803	fail:	1079	fail:
804	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
805	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
806	return &PL_sv_undef;	1082	return &PL_sv_undef;
807	}	1083	}
808		1084
809	static SV *	1085	static SV *
810	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
811	{	1087	{
812	SV *sv = 0;	1088	SV *sv = 0;
813		1089
814	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815	{	1091	{
816	// indefinite length strings	1092	// indefinite length strings
817	++dec->cur;	1093	++dec->cur;
818		1094
819	U8 major = *dec->cur & MAJOR_MISC;	1095	U8 major = *dec->cur & MAJOR_MISC;
…		…
852	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1128	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1129	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854	}	1130	}
855		1131
856	if (utf8)	1132	if (utf8)
		1133	{
		1134	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1135	if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1136	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1137
857	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
		1139	}
858		1140
859	return sv;	1141	return sv;
860		1142
861	fail:	1143	fail:
862	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
881	sv = newRV_noinc (decode_sv (dec));	1163	sv = newRV_noinc (decode_sv (dec));
882	break;	1164	break;
883		1165
884	case CBOR_TAG_STRINGREF_NAMESPACE:	1166	case CBOR_TAG_STRINGREF_NAMESPACE:
885	{	1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
886	ENTER; SAVETMPS;	1170	ENTER;
887		1171
888	SAVESPTR (dec->stringref);	1172	SAVESPTR (dec->stringref);
889	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());	1173	dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
890		1174
891	sv = decode_sv (dec);	1175	sv = decode_sv (dec);
892		1176
893	FREETMPS; LEAVE;	1177	LEAVE;
894	}	1178	}
895	break;	1179	break;
896		1180
897	case CBOR_TAG_STRINGREF:	1181	case CBOR_TAG_STRINGREF:
898	{	1182	{
899	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901		1185
902	UV idx = decode_uint (dec);	1186	UV idx = decode_uint (dec);
903		1187
904	if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))	1188	if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
905	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1189	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906		1190
907	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908	}	1192	}
909	break;	1193	break;
…		…
911	case CBOR_TAG_VALUE_SHAREABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
912	{	1196	{
913	if (ecb_expect_false (!dec->shareable))	1197	if (ecb_expect_false (!dec->shareable))
914	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());	1198	dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
915		1199
		1200	if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES | F_ALLOW_WEAK_CYCLES)))
		1201	{
		1202	// if cycles are allowed, then we store an AV as value
		1203	// while it is being decoded, and gather unresolved
		1204	// references in it, to be re4solved after decoding.
		1205	int idx, i;
916	sv = newSV (0);	1206	AV *av = newAV ();
917	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
918		1209
919	SV *osv = decode_sv (dec);	1210	sv = decode_sv (dec);
920	sv_setsv (sv, osv);	1211
		1212	// the AV now contains \undef for all unresolved references,
		1213	// so we fix them up here.
		1214	for (i = 0; i <= AvFILLp (av); ++i)
		1215	SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
		1216
		1217	// weaken all recursive references
		1218	if (dec->cbor.flags & F_ALLOW_WEAK_CYCLES)
		1219	for (i = 0; i <= AvFILLp (av); ++i)
		1220	sv_rvweaken (AvARRAY (av)[i]);
		1221
		1222	// now replace the AV by a reference to the completed value
921	SvREFCNT_dec_NN (osv);	1223	SvREFCNT_dec_NN ((SV *)av);
		1224	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1225	}
		1226	else
		1227	{
		1228	av_push (dec->shareable, &PL_sv_undef);
		1229	int idx = AvFILLp (dec->shareable);
		1230	sv = decode_sv (dec);
		1231	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
		1232	}
922	}	1233	}
923	break;	1234	break;
924		1235
925	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
926	{	1237	{
927	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929		1240
930	UV idx = decode_uint (dec);	1241	UV idx = decode_uint (dec);
931		1242
932	if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))	1243	if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
933	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
934		1245
935	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1246	sv = AvARRAY (dec->shareable)[idx];
		1247
		1248	// reference to cycle, we create a new \undef and use that, and also
		1249	// registerr it in the AV for later fixing
		1250	if (ecb_expect_false (SvTYPE (sv) == SVt_PVAV))
		1251	{
		1252	AV *av = (AV *)sv;
		1253	sv = newRV_noinc (&PL_sv_undef);
		1254	av_push (av, SvREFCNT_inc_NN (sv));
		1255	}
		1256	else if (ecb_expect_false (sv == &PL_sv_undef)) // not yet decoded, but cycles not allowed
		1257	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
		1258	else // we decoded the object earlier, no cycle
		1259	sv = newSVsv (sv);
936	}	1260	}
937	break;	1261	break;
938		1262
939	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
940	{	1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
941	sv = decode_sv (dec);	1268	sv = decode_sv (dec);
942		1269
943	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
944	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
945		1272
…		…
955	if (!method)	1282	if (!method)
956	ERR ("cannot decode perl-object (package does not have a THAW method)");	1283	ERR ("cannot decode perl-object (package does not have a THAW method)");
957		1284
958	dSP;	1285	dSP;
959		1286
960	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
961	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
962	// we re-bless the reference to get overload and other niceties right	1290	// we re-bless the reference to get overload and other niceties right
963	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
964	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
965		1293
…		…
973	SPAGAIN;	1301	SPAGAIN;
974		1302
975	if (SvTRUE (ERRSV))	1303	if (SvTRUE (ERRSV))
976	{	1304	{
977	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
978	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1306	ERR_ERRSV;
979	}	1307	}
980		1308
981	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
982	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
983		1311
984	PUTBACK;	1312	PUTBACK;
985		1313
986	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
987	}	1315	}
988	break;	1316	break;
989		1317
990	default:	1318	default:
		1319	filter:
991	{	1320	{
		1321	SV *tag_sv = newSVuv (tag);
		1322
992	sv = decode_sv (dec);	1323	sv = decode_sv (dec);
993		1324
994	dSP;	1325	dSP;
995	ENTER; SAVETMPS; PUSHMARK (SP);	1326	ENTER; SAVETMPS;
		1327	PUSHMARK (SP);
996	EXTEND (SP, 2);	1328	EXTEND (SP, 2);
997	PUSHs (newSVuv (tag));	1329	PUSHs (tag_sv);
998	PUSHs (sv);	1330	PUSHs (sv);
999		1331
1000	PUTBACK;	1332	PUTBACK;
1001	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);	1333	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1002	SPAGAIN;	1334	SPAGAIN;
1003		1335
1004	if (SvTRUE (ERRSV))	1336	if (SvTRUE (ERRSV))
1005	{	1337	{
		1338	SvREFCNT_dec_NN (tag_sv);
1006	FREETMPS; LEAVE;	1339	FREETMPS; LEAVE;
1007	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1340	ERR_ERRSV;
1008	}	1341	}
1009		1342
1010	if (count)	1343	if (count)
1011	{	1344	{
		1345	SvREFCNT_dec_NN (tag_sv);
1012	SvREFCNT_dec (sv);	1346	SvREFCNT_dec_NN (sv);
1013	sv = SvREFCNT_inc (POPs);	1347	sv = SvREFCNT_inc_NN (TOPs);
		1348	SP -= count;
1014	}	1349	}
1015	else	1350	else
1016	{	1351	{
1017	AV *av = newAV ();	1352	AV *av = newAV ();
1018	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
1019	av_push (av, sv);	1354	av_push (av, sv);
1020		1355
1021	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1022	? cbor_tagged_stash	1357	? cbor_tagged_stash
1023	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
1109		1444
1110	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
1111	}	1446	}
1112		1447
1113	// 0..19 unassigned simple	1448	// 0..19 unassigned simple
1114	// 24 reserved + unassigned (reserved values are not encodable)	1449	// 24 reserved + unassigned simple (reserved values are not encodable)
		1450	// 28-30 unassigned misc
		1451	// 31 break code
1115	default:	1452	default:
1116	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117	}	1454	}
1118		1455
1119	break;	1456	break;
1120	}	1457	}
1121		1458
…		…
1124	}	1461	}
1125		1462
1126	static SV *	1463	static SV *
1127	decode_cbor (SV *string, CBOR *cbor, char **offset_return)	1464	decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1128	{	1465	{
1129	dec_t dec = { };	1466	dec_t dec = { 0 };
1130	SV *sv;	1467	SV *sv;
1131	STRLEN len;	1468	STRLEN len;
1132	char *data = SvPVbyte (string, len);	1469	char *data = SvPVbyte (string, len);
1133		1470
1134	if (len > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
…		…
1148	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
1149	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
1150		1487
1151	if (dec.err)	1488	if (dec.err)
1152	{	1489	{
		1490	if (dec.shareable)
		1491	{
		1492	// need to break cyclic links, which would all be in shareable
		1493	int i;
		1494	SV **svp;
		1495
		1496	for (i = av_len (dec.shareable) + 1; i--; )
		1497	if ((svp = av_fetch (dec.shareable, i, 0)))
		1498	sv_setsv (*svp, &PL_sv_undef);
		1499	}
		1500
1153	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
1154	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);	1506	croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1155	}	1507	}
1156		1508
1157	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
1158		1510
1159	return sv;	1511	return sv;
1160	}	1512	}
1161		1513
		1514	/////////////////////////////////////////////////////////////////////////////
		1515	// incremental parser
		1516
		1517	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1518
		1519	// returns 0 for notyet, 1 for success or error
		1520	static int
		1521	incr_parse (CBOR *self, SV *cborstr)
		1522	{
		1523	STRLEN cur;
		1524	SvPV (cborstr, cur);
		1525
		1526	while (ecb_expect_true (self->incr_need <= cur))
		1527	{
		1528	// table of integer count bytes
		1529	static I8 incr_len[MINOR_MASK + 1] = {
		1530	0, 0, 0, 0, 0, 0, 0, 0,
		1531	0, 0, 0, 0, 0, 0, 0, 0,
		1532	0, 0, 0, 0, 0, 0, 0, 0,
		1533	1, 2, 4, 8,-1,-1,-1,-2
		1534	};
		1535
		1536	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1537	U8 m = *p & MINOR_MASK;
		1538	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1539	I8 ilen = incr_len[m];
		1540
		1541	self->incr_need = self->incr_pos + 1;
		1542
		1543	if (ecb_expect_false (ilen < 0))
		1544	{
		1545	if (m != MINOR_INDEF)
		1546	return 1; // error
		1547
		1548	if (*p == (MAJOR_MISC | MINOR_INDEF))
		1549	{
		1550	if (count >= 0)
		1551	return 1; // error
		1552
		1553	count = 1;
		1554	}
		1555	else
		1556	{
		1557	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1558	count = -1;
		1559	}
		1560	}
		1561	else
		1562	{
		1563	self->incr_need += ilen;
		1564	if (ecb_expect_false (self->incr_need > cur))
		1565	return 0;
		1566
		1567	int major = *p >> MAJOR_SHIFT;
		1568
		1569	switch (major)
		1570	{
		1571	case MAJOR_TAG >> MAJOR_SHIFT:
		1572	++count; // tags merely prefix another value
		1573	break;
		1574
		1575	case MAJOR_BYTES >> MAJOR_SHIFT:
		1576	case MAJOR_TEXT >> MAJOR_SHIFT:
		1577	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1578	case MAJOR_MAP >> MAJOR_SHIFT:
		1579	{
		1580	UV len;
		1581
		1582	if (ecb_expect_false (ilen))
		1583	{
		1584	len = 0;
		1585
		1586	do {
		1587	len = (len << 8) | *++p;
		1588	} while (--ilen);
		1589	}
		1590	else
		1591	len = m;
		1592
		1593	switch (major)
		1594	{
		1595	case MAJOR_BYTES >> MAJOR_SHIFT:
		1596	case MAJOR_TEXT >> MAJOR_SHIFT:
		1597	self->incr_need += len;
		1598	if (ecb_expect_false (self->incr_need > cur))
		1599	return 0;
		1600
		1601	break;
		1602
		1603	case MAJOR_MAP >> MAJOR_SHIFT:
		1604	len <<= 1;
		1605	/* FALLTHROUGH */
		1606	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1607	if (len)
		1608	{
		1609	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1610	count = len + 1;
		1611	}
		1612	break;
		1613	}
		1614	}
		1615	}
		1616	}
		1617
		1618	self->incr_pos = self->incr_need;
		1619
		1620	if (count > 0)
		1621	{
		1622	while (!--count)
		1623	{
		1624	if (!AvFILLp (self->incr_count))
		1625	return 1; // done
		1626
		1627	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1628	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1629	}
		1630
		1631	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1632	}
		1633	}
		1634
		1635	return 0;
		1636	}
		1637
		1638
1162	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
1163	// XS interface functions	1640	// XS interface functions
1164		1641
1165	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
1166		1643
…		…
1178		1655
1179	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180		1657
1181	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
1182	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183	}	1662	}
1184		1663
1185	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
1186		1665
1187	void CLONE (...)	1666	void CLONE (...)
…		…
1203	)));	1682	)));
1204	}	1683	}
1205		1684
1206	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1207	ALIAS:	1686	ALIAS:
1208	shrink = F_SHRINK	1687	shrink = F_SHRINK
1209	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1210	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
		1690	allow_cycles = F_ALLOW_CYCLES
		1691	allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1692	forbid_objects = F_FORBID_OBJECTS
1211	pack_strings = F_PACK_STRINGS	1693	pack_strings = F_PACK_STRINGS
		1694	text_keys = F_TEXT_KEYS
		1695	text_strings = F_TEXT_STRINGS
		1696	validate_utf8 = F_VALIDATE_UTF8
1212	PPCODE:	1697	PPCODE:
1213	{	1698	{
1214	if (enable)	1699	if (enable)
1215	self->flags |= ix;	1700	self->flags |= ix;
1216	else	1701	else
…		…
1219	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1220	}	1705	}
1221		1706
1222	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1223	ALIAS:	1708	ALIAS:
1224	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1225	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1226	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
		1712	get_allow_cycles = F_ALLOW_CYCLES
		1713	get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1714	get_forbid_objects = F_FORBID_OBJECTS
1227	get_pack_strings = F_PACK_STRINGS	1715	get_pack_strings = F_PACK_STRINGS
		1716	get_text_keys = F_TEXT_KEYS
		1717	get_text_strings = F_TEXT_STRINGS
		1718	get_validate_utf8 = F_VALIDATE_UTF8
1228	PPCODE:	1719	PPCODE:
1229	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1230		1721
1231	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232	PPCODE:	1723	PPCODE:
…		…
1281	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1282	PUSHs (sv);	1773	PUSHs (sv);
1283	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284	}	1775	}
1285		1776
		1777	void incr_parse (CBOR *self, SV *cborstr)
		1778	ALIAS:
		1779	incr_parse_multiple = 1
		1780	PPCODE:
		1781	{
		1782	if (SvUTF8 (cborstr))
		1783	sv_utf8_downgrade (cborstr, 0);
		1784
		1785	if (!self->incr_count)
		1786	{
		1787	self->incr_count = newAV ();
		1788	self->incr_pos = 0;
		1789	self->incr_need = 1;
		1790
		1791	av_push (self->incr_count, newSViv (1));
		1792	}
		1793
		1794	do
		1795	{
		1796	if (!incr_parse (self, cborstr))
		1797	{
		1798	if (self->incr_need > self->max_size && self->max_size)
		1799	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1800	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1801
		1802	break;
		1803	}
		1804
		1805	SV *sv;
		1806	char *offset;
		1807
		1808	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1809	XPUSHs (sv);
		1810
		1811	sv_chop (cborstr, offset);
		1812
		1813	av_clear (self->incr_count);
		1814	av_push (self->incr_count, newSViv (1));
		1815
		1816	self->incr_pos = 0;
		1817	self->incr_need = self->incr_pos + 1;
		1818	}
		1819	while (ix);
		1820	}
		1821
		1822	void incr_reset (CBOR *self)
		1823	CODE:
		1824	{
		1825	SvREFCNT_dec (self->incr_count);
		1826	self->incr_count = 0;
		1827	}
		1828
1286	void DESTROY (CBOR *self)	1829	void DESTROY (CBOR *self)
1287	PPCODE:	1830	PPCODE:
1288	cbor_free (self);	1831	cbor_free (self);
1289		1832
1290	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
1291		1834
1292	void encode_cbor (SV *scalar)	1835	void encode_cbor (SV *scalar)
		1836	ALIAS:
		1837	encode_cbor = 0
		1838	encode_cbor_sharing = F_ALLOW_SHARING
1293	PPCODE:	1839	PPCODE:
1294	{	1840	{
1295	CBOR cbor;	1841	CBOR cbor;
1296	cbor_init (&cbor);	1842	cbor_init (&cbor);
		1843	cbor.flags |= ix;
1297	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1844	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298	XPUSHs (scalar);	1845	XPUSHs (scalar);
1299	}	1846	}
1300		1847
1301	void decode_cbor (SV *cborstr)	1848	void decode_cbor (SV *cborstr)
…		…
1305	cbor_init (&cbor);	1852	cbor_init (&cbor);
1306	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1308	}	1855	}
1309		1856
		1857	#ifdef __AFL_COMPILER
		1858
		1859	void
		1860	afl_init ()
		1861	CODE:
		1862	__AFL_INIT ();
		1863
		1864	int
		1865	afl_loop (unsigned int count = 10000)
		1866	CODE:
		1867	RETVAL = __AFL_LOOP (count);
		1868	OUTPUT:
		1869	RETVAL
		1870
		1871	#endif
		1872

Diff Legend

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