[ViewVC] Diff of: cvs/CBOR-XS/XS.xs

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.37 by root, Sat Nov 30 18:13:53 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
102	#define F_ALLOW_CYCLES 0x00000008UL	124	#define F_ALLOW_CYCLES 0x00000008UL
		125	#define F_ALLOW_WEAK_CYCLES 0x00000010UL
		126	#define F_FORBID_OBJECTS 0x00000020UL
103	#define F_PACK_STRINGS 0x00000010UL	127	#define F_PACK_STRINGS 0x00000040UL
		128	#define F_TEXT_KEYS 0x00000080UL
		129	#define F_TEXT_STRINGS 0x00000100UL
		130	#define F_VALIDATE_UTF8 0x00000200UL
104		131
105	#define INIT_SIZE 32 // initial scalar size to be allocated	132	#define INIT_SIZE 32 // initial scalar size to be allocated
106		133
107	#define SB do {	134	#define SB do {
108	#define SE } while (0)	135	#define SE } while (0)
…		…
127	typedef struct {	154	typedef struct {
128	U32 flags;	155	U32 flags;
129	U32 max_depth;	156	U32 max_depth;
130	STRLEN max_size;	157	STRLEN max_size;
131	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.
132	} CBOR;	164	} CBOR;
133		165
134	ecb_inline void	166	ecb_inline void
135	cbor_init (CBOR *cbor)	167	cbor_init (CBOR *cbor)
136	{	168	{
…		…
140		172
141	ecb_inline void	173	ecb_inline void
142	cbor_free (CBOR *cbor)	174	cbor_free (CBOR *cbor)
143	{	175	{
144	SvREFCNT_dec (cbor->filter);	176	SvREFCNT_dec (cbor->filter);
		177	SvREFCNT_dec (cbor->incr_count);
145	}	178	}
146		179
147	/////////////////////////////////////////////////////////////////////////////	180	/////////////////////////////////////////////////////////////////////////////
148	// utility functions	181	// utility functions
149		182
…		…
172	#endif	205	#endif
173	}	206	}
174	}	207	}
175		208
176	// minimum length of a string to be registered for stringref	209	// minimum length of a string to be registered for stringref
177	ecb_inline int	210	ecb_inline STRLEN
178	minimum_string_length (UV idx)	211	minimum_string_length (UV idx)
179	{	212	{
180	return idx > 23	213	return idx <= 23 ? 3
181	? idx > 0xffU	214	: idx <= 0xffU ? 4
182	? idx > 0xffffU	215	: idx <= 0xffffU ? 5
183	? idx > 0xffffffffU	216	: idx <= 0xffffffffU ? 7
184	? 11	217	: 11;
185	: 7
186	: 5
187	: 4
188	: 3;
189	}	218	}
190		219
191	/////////////////////////////////////////////////////////////////////////////	220	/////////////////////////////////////////////////////////////////////////////
192	// encoder	221	// encoder
193		222
…		…
206	} enc_t;	235	} enc_t;
207		236
208	ecb_inline void	237	ecb_inline void
209	need (enc_t *enc, STRLEN len)	238	need (enc_t *enc, STRLEN len)
210	{	239	{
211	if (ecb_expect_false (enc->cur + len >= enc->end))	240	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
212	{	241	{
213	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	242	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
214	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	243	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
215	enc->cur = SvPVX (enc->sv) + cur;	244	enc->cur = SvPVX (enc->sv) + cur;
216	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	245	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
217	}	246	}
218	}	247	}
219		248
		249	static void encode_sv (enc_t enc, SV sv);
		250
220	ecb_inline void	251	ecb_inline void
221	encode_ch (enc_t *enc, char ch)	252	encode_ch (enc_t *enc, char ch)
222	{	253	{
223	need (enc, 1);	254	need (enc, 1);
224	*enc->cur++ = ch;	255	*enc->cur++ = ch;
225	}	256	}
226		257
		258	// used for tags, intregers, element counts and so on
227	static void	259	static void
228	encode_uint (enc_t *enc, int major, UV len)	260	encode_uint (enc_t *enc, int major, UV len)
229	{	261	{
230	need (enc, 9);	262	need (enc, 9);
231		263
…		…
262	*enc->cur++ = len >> 8;	294	*enc->cur++ = len >> 8;
263	*enc->cur++ = len;	295	*enc->cur++ = len;
264	}	296	}
265	}	297	}
266		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
267	ecb_inline void	311	ecb_inline void
268	encode_tag (enc_t *enc, UV tag)	312	encode_tag (enc_t *enc, UV tag)
269	{	313	{
270	encode_uint (enc, MAJOR_TAG, tag);	314	encode_uint (enc, MAJOR_TAG, tag);
271	}	315	}
272		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
273	ecb_inline void	340	ecb_inline void
274	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)
275	{	342	{
		343	if (ecb_expect_false (upgrade_utf8))
		344	if (!utf8)
		345	{
		346	encode_str_utf8 (enc, utf8, str, len);
		347	return;
		348	}
		349
276	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);	350	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
277	need (enc, len);	351	need (enc, len);
278	memcpy (enc->cur, str, len);	352	memcpy (enc->cur, str, len);
279	enc->cur += len;	353	enc->cur += len;
280	}	354	}
281		355
282	static void	356	ecb_inline void
283	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)
284	{	358	{
285	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))	359	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286	{	360	{
287	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	361	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288		362
…		…
299	sv_setuv (*svp, enc->stringref_idx);	373	sv_setuv (*svp, enc->stringref_idx);
300	++enc->stringref_idx;	374	++enc->stringref_idx;
301	}	375	}
302	}	376	}
303		377
304	encode_str (enc, utf8, str, len);	378	encode_str (enc, upgrade_utf8, utf8, str, len);
305	}	379	}
306		380
307	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	}
308		511
309	static void	512	static void
310	encode_av (enc_t enc, AV av)	513	encode_av (enc_t enc, AV av)
311	{	514	{
312	int i, len = av_len (av);	515	int i, len = av_len (av);
…		…
316		519
317	++enc->depth;	520	++enc->depth;
318		521
319	encode_uint (enc, MAJOR_ARRAY, len + 1);	522	encode_uint (enc, MAJOR_ARRAY, len + 1);
320		523
		524	if (ecb_expect_false (SvMAGICAL (av)))
321	for (i = 0; i <= len; ++i)	525	for (i = 0; i <= len; ++i)
322	{	526	{
323	SV **svp = av_fetch (av, i, 0);	527	SV **svp = av_fetch (av, i, 0);
324	encode_sv (enc, svp ? *svp : &PL_sv_undef);	528	encode_sv (enc, svp ? *svp : &PL_sv_undef);
325	}	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	}
326		536
327	--enc->depth;	537	--enc->depth;
328	}	538	}
329		539
330	static void	540	static void
…		…
338	++enc->depth;	548	++enc->depth;
339		549
340	int pairs = hv_iterinit (hv);	550	int pairs = hv_iterinit (hv);
341	int mg = SvMAGICAL (hv);	551	int mg = SvMAGICAL (hv);
342		552
343	if (mg)	553	if (ecb_expect_false (mg))
344	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);	554	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
345	else	555	else
346	encode_uint (enc, MAJOR_MAP, pairs);	556	encode_uint (enc, MAJOR_MAP, pairs);
347		557
348	while ((he = hv_iternext (hv)))	558	while ((he = hv_iternext (hv)))
349	{	559	{
350	if (HeKLEN (he) == HEf_SVKEY)	560	if (HeKLEN (he) == HEf_SVKEY)
351	encode_sv (enc, HeSVKEY (he));	561	encode_sv (enc, HeSVKEY (he));
352	else	562	else
353	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));
354		564
355	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));
356	}	566	}
357		567
358	if (mg)	568	if (ecb_expect_false (mg))
359	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);	569	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
360		570
361	--enc->depth;	571	--enc->depth;
362	}	572	}
363		573
…		…
383		593
384	HV *stash = SvSTASH (sv);	594	HV *stash = SvSTASH (sv);
385		595
386	if (stash == boolean_stash)	596	if (stash == boolean_stash)
387	{	597	{
388	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);	598	encode_bool (enc, SvIV (sv));
389	return;	599	return;
390	}	600	}
391	else if (stash == error_stash)	601	else if (stash == error_stash)
392	{	602	{
393	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);	603	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
…		…
396	else if (stash == tagged_stash)	606	else if (stash == tagged_stash)
397	{	607	{
398	if (svt != SVt_PVAV)	608	if (svt != SVt_PVAV)
399	croak ("encountered CBOR::XS::Tagged object that isn't an array");	609	croak ("encountered CBOR::XS::Tagged object that isn't an array");
400		610
		611	switch (av_len ((AV *)sv))
		612	{
		613	case 2-1:
		614	// actually a tagged value
401	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)));
402	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	}
403		627
404	return;	628	return;
405	}	629	}
406	}	630	}
407		631
408	if (ecb_expect_false (SvREFCNT (sv) > 1)	632	if (ecb_expect_false (SvREFCNT (sv) > 1)
409	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))	633	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410	{	634	{
411	if (!enc->shareable)	635	if (ecb_expect_false (!enc->shareable))
412	enc->shareable = (HV )sv_2mortal ((SV )newHV ());	636	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
413		637
414	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);	638	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
415		639
416	if (SvOK (*svp))	640	if (SvOK (*svp))
…		…
430	if (ecb_expect_false (SvOBJECT (sv)))	654	if (ecb_expect_false (SvOBJECT (sv)))
431	{	655	{
432	HV *stash = SvSTASH (sv);	656	HV *stash = SvSTASH (sv);
433	GV *method;	657	GV *method;
434		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))));
435	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	662	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
436	{	663	{
437	dSP;	664	dSP;
438		665
439	ENTER; SAVETMPS; PUSHMARK (SP);	666	ENTER; SAVETMPS;
		667	PUSHMARK (SP);
440	// 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
441	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	669	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
442		670
443	PUTBACK;	671	PUTBACK;
444	// G_SCALAR ensures that return value is 1	672	// G_SCALAR ensures that return value is 1
…		…
457	}	685	}
458	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	686	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
459	{	687	{
460	dSP;	688	dSP;
461		689
462	ENTER; SAVETMPS; PUSHMARK (SP);	690	ENTER; SAVETMPS;
		691	PUSHMARK (SP);
463	EXTEND (SP, 2);	692	EXTEND (SP, 2);
464	// 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
465	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	694	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466	PUSHs (sv_cbor);	695	PUSHs (sv_cbor);
467		696
…		…
473	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	702	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474	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));
475		704
476	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	705	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477	encode_uint (enc, MAJOR_ARRAY, count + 1);	706	encode_uint (enc, MAJOR_ARRAY, count + 1);
478	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	707	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479		708
480	while (count)	709	{
		710	int i;
		711
		712	for (i = 0; i < count; ++i)
481	encode_sv (enc, SP[1 - count--]);	713	encode_sv (enc, SP[i + 1 - count]);
		714
		715	SP -= count;
		716	}
482		717
483	PUTBACK;	718	PUTBACK;
484		719
485	FREETMPS; LEAVE;	720	FREETMPS; LEAVE;
486	}	721	}
…		…
508		743
509	if (ecb_expect_false (nv == (NV)(U32)nv))	744	if (ecb_expect_false (nv == (NV)(U32)nv))
510	encode_uint (enc, MAJOR_POS_INT, (U32)nv);	745	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
511	//TODO: maybe I32?	746	//TODO: maybe I32?
512	else if (ecb_expect_false (nv == (float)nv))	747	else if (ecb_expect_false (nv == (float)nv))
513	{	748	encode_float32 (enc, nv);
514	uint32_t fp = ecb_float_to_binary32 (nv);
515
516	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
517
518	if (!ecb_big_endian ())
519	fp = ecb_bswap32 (fp);
520
521	memcpy (enc->cur, &fp, 4);
522	enc->cur += 4;
523	}
524	else	749	else
525	{	750	encode_float64 (enc, nv);
526	uint64_t fp = ecb_double_to_binary64 (nv);
527
528	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
529
530	if (!ecb_big_endian ())
531	fp = ecb_bswap64 (fp);
532
533	memcpy (enc->cur, &fp, 8);
534	enc->cur += 8;
535	}
536	}	751	}
537		752
538	static void	753	static void
539	encode_sv (enc_t enc, SV sv)	754	encode_sv (enc_t enc, SV sv)
540	{	755	{
…		…
542		757
543	if (SvPOKp (sv))	758	if (SvPOKp (sv))
544	{	759	{
545	STRLEN len;	760	STRLEN len;
546	char *str = SvPV (sv, len);	761	char *str = SvPV (sv, len);
547	encode_strref (enc, SvUTF8 (sv), str, len);	762	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
548	}	763	}
549	else if (SvNOKp (sv))	764	else if (SvNOKp (sv))
550	encode_nv (enc, sv);	765	encode_nv (enc, sv);
551	else if (SvIOKp (sv))	766	else if (SvIOKp (sv))
552	{	767	encode_int (enc, sv);
553	if (SvIsUV (sv))
554	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
555	else if (SvIVX (sv) >= 0)
556	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
557	else
558	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
559	}
560	else if (SvROK (sv))	768	else if (SvROK (sv))
561	encode_rv (enc, SvRV (sv));	769	encode_rv (enc, SvRV (sv));
562	else if (!SvOK (sv))	770	else if (!SvOK (sv))
563	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);	771	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
564	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	772	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
…		…
569	}	777	}
570		778
571	static SV *	779	static SV *
572	encode_cbor (SV scalar, CBOR cbor)	780	encode_cbor (SV scalar, CBOR cbor)
573	{	781	{
574	enc_t enc = { };	782	enc_t enc = { 0 };
575		783
576	enc.cbor = *cbor;	784	enc.cbor = *cbor;
577	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	785	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
578	enc.cur = SvPVX (enc.sv);	786	enc.cur = SvPVX (enc.sv);
579	enc.end = SvEND (enc.sv);	787	enc.end = SvEND (enc.sv);
580		788
581	SvPOK_only (enc.sv);	789	SvPOK_only (enc.sv);
582		790
583	if (cbor->flags & F_PACK_STRINGS)	791	if (cbor->flags & F_PACK_STRINGS)
584	{	792	{
…		…
611	U32 depth; // recursion depth	819	U32 depth; // recursion depth
612	U32 maxdepth; // recursion depth limit	820	U32 maxdepth; // recursion depth limit
613	AV *shareable;	821	AV *shareable;
614	AV *stringref;	822	AV *stringref;
615	SV *decode_tagged;	823	SV *decode_tagged;
		824	SV *err_sv; // optional sv for error, needs to be freed
616	} dec_t;	825	} dec_t;
617		826
618	#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);
619		834
620	#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;
621		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
622	#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)
623	#define DEC_DEC_DEPTH --dec->depth	864	#define DEC_DEC_DEPTH --dec->depth
624		865
625	static UV	866	static UV
626	decode_uint (dec_t *dec)	867	decode_uint (dec_t *dec)
627	{	868	{
…		…
693		934
694	for (;;)	935	for (;;)
695	{	936	{
696	WANT (1);	937	WANT (1);
697		938
698	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	939	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
699	{	940	{
700	++dec->cur;	941	++dec->cur;
701	break;	942	break;
702	}	943	}
703		944
704	av_push (av, decode_sv (dec));	945	av_push (av, decode_sv (dec));
705	}	946	}
706	}	947	}
707	else	948	else
708	{	949	{
709	int i, len = decode_uint (dec);	950	UV i, len = decode_uint (dec);
710		951
711	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays	952	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
712	av_fill (av, len - 1);	953	av_fill (av, len - 1);
713		954
714	for (i = 0; i < len; ++i)	955	for (i = 0; i < len; ++i)
…		…
717		958
718	DEC_DEC_DEPTH;	959	DEC_DEC_DEPTH;
719	return newRV_noinc ((SV *)av);	960	return newRV_noinc ((SV *)av);
720		961
721	fail:	962	fail:
722	SvREFCNT_dec (av);	963	SvREFCNT_dec_NN (av);
723	DEC_DEC_DEPTH;	964	DEC_DEC_DEPTH;
724	return &PL_sv_undef;	965	return &PL_sv_undef;
725	}	966	}
726		967
727	static void	968	static void
…		…
729	{	970	{
730	// for speed reasons, we specialcase single-string	971	// for speed reasons, we specialcase single-string
731	// byte or utf-8 strings as keys, but only when !stringref	972	// byte or utf-8 strings as keys, but only when !stringref
732		973
733	if (ecb_expect_true (!dec->stringref))	974	if (ecb_expect_true (!dec->stringref))
734	if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))	975	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735	{	976	{
736	I32 len = decode_uint (dec);	977	STRLEN len = decode_uint (dec);
737	char key = (char )dec->cur;	978	char key = (char )dec->cur;
738		979
		980	WANT (len);
739	dec->cur += len;	981	dec->cur += len;
740		982
741	if (ecb_expect_false (dec->stringref))
742	av_push (dec->stringref, newSVpvn (key, len));
743
744	hv_store (hv, key, len, decode_sv (dec), 0);	983	hv_store (hv, key, len, decode_sv (dec), 0);
745		984
746	return;	985	return;
747	}	986	}
748	else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))	987	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749	{	988	{
750	I32 len = decode_uint (dec);	989	STRLEN len = decode_uint (dec);
751	char key = (char )dec->cur;	990	char key = (char )dec->cur;
752		991
		992	WANT (len);
753	dec->cur += len;	993	dec->cur += len;
754		994
755	if (ecb_expect_false (dec->stringref))	995	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
756	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)");
757		998
758	hv_store (hv, key, -len, decode_sv (dec), 0);	999	hv_store (hv, key, -len, decode_sv (dec), 0);
759		1000
760	return;	1001	return;
761	}	1002	}
762		1003
763	SV *k = decode_sv (dec);	1004	SV *k = decode_sv (dec);
764	SV *v = decode_sv (dec);	1005	SV *v = decode_sv (dec);
765		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
766	hv_store_ent (hv, k, v, 0);	1035	hv_store_ent (hv, k, v, 0);
767	SvREFCNT_dec (k);	1036	SvREFCNT_dec_NN (k);
		1037
		1038	fail:
		1039	;
768	}	1040	}
769		1041
770	static SV *	1042	static SV *
771	decode_hv (dec_t *dec)	1043	decode_hv (dec_t *dec)
772	{	1044	{
…		…
780		1052
781	for (;;)	1053	for (;;)
782	{	1054	{
783	WANT (1);	1055	WANT (1);
784		1056
785	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	1057	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
786	{	1058	{
787	++dec->cur;	1059	++dec->cur;
788	break;	1060	break;
789	}	1061	}
790		1062
791	decode_he (dec, hv);	1063	decode_he (dec, hv);
792	}	1064	}
793	}	1065	}
794	else	1066	else
795	{	1067	{
796	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
797		1071
798	while (pairs--)	1072	while (pairs--)
799	decode_he (dec, hv);	1073	decode_he (dec, hv);
800	}	1074	}
801		1075
802	DEC_DEC_DEPTH;	1076	DEC_DEC_DEPTH;
803	return newRV_noinc ((SV *)hv);	1077	return newRV_noinc ((SV *)hv);
804		1078
805	fail:	1079	fail:
806	SvREFCNT_dec (hv);	1080	SvREFCNT_dec_NN (hv);
807	DEC_DEC_DEPTH;	1081	DEC_DEC_DEPTH;
808	return &PL_sv_undef;	1082	return &PL_sv_undef;
809	}	1083	}
810		1084
811	static SV *	1085	static SV *
812	decode_str (dec_t *dec, int utf8)	1086	decode_str (dec_t *dec, int utf8)
813	{	1087	{
814	SV *sv = 0;	1088	SV *sv = 0;
815		1089
816	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)	1090	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
817	{	1091	{
818	// indefinite length strings	1092	// indefinite length strings
819	++dec->cur;	1093	++dec->cur;
820		1094
821	U8 major = *dec->cur & MAJOR_MISC;	1095	U8 major = *dec->cur & MAJOR_MISC;
…		…
854	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1128	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1129	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
856	}	1130	}
857		1131
858	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
859	SvUTF8_on (sv);	1138	SvUTF8_on (sv);
		1139	}
860		1140
861	return sv;	1141	return sv;
862		1142
863	fail:	1143	fail:
864	SvREFCNT_dec (sv);	1144	SvREFCNT_dec (sv);
…		…
883	sv = newRV_noinc (decode_sv (dec));	1163	sv = newRV_noinc (decode_sv (dec));
884	break;	1164	break;
885		1165
886	case CBOR_TAG_STRINGREF_NAMESPACE:	1166	case CBOR_TAG_STRINGREF_NAMESPACE:
887	{	1167	{
		1168	// do not use SAVETMPS/FREETMPS, as these will
		1169	// erase mortalised caches, e.g. "shareable"
888	ENTER; SAVETMPS;	1170	ENTER;
889		1171
890	SAVESPTR (dec->stringref);	1172	SAVESPTR (dec->stringref);
891	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1173	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
892		1174
893	sv = decode_sv (dec);	1175	sv = decode_sv (dec);
894		1176
895	FREETMPS; LEAVE;	1177	LEAVE;
896	}	1178	}
897	break;	1179	break;
898		1180
899	case CBOR_TAG_STRINGREF:	1181	case CBOR_TAG_STRINGREF:
900	{	1182	{
901	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1183	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1184	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903		1185
904	UV idx = decode_uint (dec);	1186	UV idx = decode_uint (dec);
905		1187
906	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	1188	if (!dec->stringref \|\| idx >= (UV)(1 + AvFILLp (dec->stringref)))
907	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)");
908		1190
909	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1191	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910	}	1192	}
911	break;	1193	break;
…		…
913	case CBOR_TAG_VALUE_SHAREABLE:	1195	case CBOR_TAG_VALUE_SHAREABLE:
914	{	1196	{
915	if (ecb_expect_false (!dec->shareable))	1197	if (ecb_expect_false (!dec->shareable))
916	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1198	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
917		1199
918	if (dec->cbor.flags & F_ALLOW_CYCLES)	1200	if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES \| F_ALLOW_WEAK_CYCLES)))
919	{	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;
920	sv = newSV (0);	1206	AV *av = newAV ();
921	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1207	av_push (dec->shareable, (SV *)av);
		1208	idx = AvFILLp (dec->shareable);
922		1209
923	SV *osv = decode_sv (dec);	1210	sv = decode_sv (dec);
924	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
925	SvREFCNT_dec_NN (osv);	1223	SvREFCNT_dec_NN ((SV *)av);
		1224	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
926	}	1225	}
927	else	1226	else
928	{	1227	{
929	av_push (dec->shareable, &PL_sv_undef);	1228	av_push (dec->shareable, &PL_sv_undef);
930	int idx = AvFILLp (dec->shareable);	1229	int idx = AvFILLp (dec->shareable);
931	sv = decode_sv (dec);	1230	sv = decode_sv (dec);
932	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));	1231	AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
933	}	1232	}
934	}	1233	}
935	break;	1234	break;
936		1235
937	case CBOR_TAG_VALUE_SHAREDREF:	1236	case CBOR_TAG_VALUE_SHAREDREF:
…		…
939	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1238	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1239	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941		1240
942	UV idx = decode_uint (dec);	1241	UV idx = decode_uint (dec);
943		1242
944	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1243	if (!dec->shareable \|\| idx >= (UV)(1 + AvFILLp (dec->shareable)))
945	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1244	ERR ("corrupted CBOR data (sharedref index out of bounds)");
946		1245
947	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1246	sv = AvARRAY (dec->shareable)[idx];
948		1247
949	if (sv == &PL_sv_undef)	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
950	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");	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);
951	}	1260	}
952	break;	1261	break;
953		1262
954	case CBOR_TAG_PERL_OBJECT:	1263	case CBOR_TAG_PERL_OBJECT:
955	{	1264	{
		1265	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1266	goto filter;
		1267
956	sv = decode_sv (dec);	1268	sv = decode_sv (dec);
957		1269
958	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)	1270	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)
959	ERR ("corrupted CBOR data (non-array perl object)");	1271	ERR ("corrupted CBOR data (non-array perl object)");
960		1272
…		…
970	if (!method)	1282	if (!method)
971	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)");
972		1284
973	dSP;	1285	dSP;
974		1286
975	ENTER; SAVETMPS; PUSHMARK (SP);	1287	ENTER; SAVETMPS;
		1288	PUSHMARK (SP);
976	EXTEND (SP, len + 1);	1289	EXTEND (SP, len + 1);
977	// 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
978	PUSHs (*av_fetch (av, 0, 1));	1291	PUSHs (*av_fetch (av, 0, 1));
979	PUSHs (sv_cbor);	1292	PUSHs (sv_cbor);
980		1293
…		…
988	SPAGAIN;	1301	SPAGAIN;
989		1302
990	if (SvTRUE (ERRSV))	1303	if (SvTRUE (ERRSV))
991	{	1304	{
992	FREETMPS; LEAVE;	1305	FREETMPS; LEAVE;
993	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1306	ERR_ERRSV;
994	}	1307	}
995		1308
996	SvREFCNT_dec (sv);	1309	SvREFCNT_dec_NN (sv);
997	sv = SvREFCNT_inc (POPs);	1310	sv = SvREFCNT_inc (POPs);
998		1311
999	PUTBACK;	1312	PUTBACK;
1000		1313
1001	FREETMPS; LEAVE;	1314	FREETMPS; LEAVE;
1002	}	1315	}
1003	break;	1316	break;
1004		1317
1005	default:	1318	default:
		1319	filter:
1006	{	1320	{
		1321	SV *tag_sv = newSVuv (tag);
		1322
1007	sv = decode_sv (dec);	1323	sv = decode_sv (dec);
1008		1324
1009	dSP;	1325	dSP;
1010	ENTER; SAVETMPS; PUSHMARK (SP);	1326	ENTER; SAVETMPS;
		1327	PUSHMARK (SP);
1011	EXTEND (SP, 2);	1328	EXTEND (SP, 2);
1012	PUSHs (newSVuv (tag));	1329	PUSHs (tag_sv);
1013	PUSHs (sv);	1330	PUSHs (sv);
1014		1331
1015	PUTBACK;	1332	PUTBACK;
1016	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);
1017	SPAGAIN;	1334	SPAGAIN;
1018		1335
1019	if (SvTRUE (ERRSV))	1336	if (SvTRUE (ERRSV))
1020	{	1337	{
		1338	SvREFCNT_dec_NN (tag_sv);
1021	FREETMPS; LEAVE;	1339	FREETMPS; LEAVE;
1022	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1340	ERR_ERRSV;
1023	}	1341	}
1024		1342
1025	if (count)	1343	if (count)
1026	{	1344	{
		1345	SvREFCNT_dec_NN (tag_sv);
1027	SvREFCNT_dec (sv);	1346	SvREFCNT_dec_NN (sv);
1028	sv = SvREFCNT_inc (POPs);	1347	sv = SvREFCNT_inc_NN (TOPs);
		1348	SP -= count;
1029	}	1349	}
1030	else	1350	else
1031	{	1351	{
1032	AV *av = newAV ();	1352	AV *av = newAV ();
1033	av_push (av, newSVuv (tag));	1353	av_push (av, tag_sv);
1034	av_push (av, sv);	1354	av_push (av, sv);
1035		1355
1036	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1356	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
1037	? cbor_tagged_stash	1357	? cbor_tagged_stash
1038	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1358	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
1124		1444
1125	return newSVnv (ecb_binary64_to_double (fp));	1445	return newSVnv (ecb_binary64_to_double (fp));
1126	}	1446	}
1127		1447
1128	// 0..19 unassigned simple	1448	// 0..19 unassigned simple
1129	// 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
1130	default:	1452	default:
1131	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1453	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1132	}	1454	}
1133		1455
1134	break;	1456	break;
1135	}	1457	}
1136		1458
…		…
1139	}	1461	}
1140		1462
1141	static SV *	1463	static SV *
1142	decode_cbor (SV string, CBOR cbor, char **offset_return)	1464	decode_cbor (SV string, CBOR cbor, char **offset_return)
1143	{	1465	{
1144	dec_t dec = { };	1466	dec_t dec = { 0 };
1145	SV *sv;	1467	SV *sv;
1146	STRLEN len;	1468	STRLEN len;
1147	char *data = SvPVbyte (string, len);	1469	char *data = SvPVbyte (string, len);
1148		1470
1149	if (len > cbor->max_size && cbor->max_size)	1471	if (len > cbor->max_size && cbor->max_size)
…		…
1163	if (dec.cur != dec.end && !dec.err)	1485	if (dec.cur != dec.end && !dec.err)
1164	dec.err = "garbage after CBOR object";	1486	dec.err = "garbage after CBOR object";
1165		1487
1166	if (dec.err)	1488	if (dec.err)
1167	{	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
1168	SvREFCNT_dec (sv);	1501	SvREFCNT_dec_NN (sv);
		1502
		1503	if (dec.err_sv)
		1504	sv_2mortal (dec.err_sv);
		1505
1169	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);
1170	}	1507	}
1171		1508
1172	sv = sv_2mortal (sv);	1509	sv = sv_2mortal (sv);
1173		1510
1174	return sv;	1511	return sv;
1175	}	1512	}
1176		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
1177	/////////////////////////////////////////////////////////////////////////////	1639	/////////////////////////////////////////////////////////////////////////////
1178	// XS interface functions	1640	// XS interface functions
1179		1641
1180	MODULE = CBOR::XS PACKAGE = CBOR::XS	1642	MODULE = CBOR::XS PACKAGE = CBOR::XS
1181		1643
…		…
1193		1655
1194	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1656	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195		1657
1196	sv_cbor = newSVpv ("CBOR", 0);	1658	sv_cbor = newSVpv ("CBOR", 0);
1197	SvREADONLY_on (sv_cbor);	1659	SvREADONLY_on (sv_cbor);
		1660
		1661	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1198	}	1662	}
1199		1663
1200	PROTOTYPES: DISABLE	1664	PROTOTYPES: DISABLE
1201		1665
1202	void CLONE (...)	1666	void CLONE (...)
…		…
1218	)));	1682	)));
1219	}	1683	}
1220		1684
1221	void shrink (CBOR *self, int enable = 1)	1685	void shrink (CBOR *self, int enable = 1)
1222	ALIAS:	1686	ALIAS:
1223	shrink = F_SHRINK	1687	shrink = F_SHRINK
1224	allow_unknown = F_ALLOW_UNKNOWN	1688	allow_unknown = F_ALLOW_UNKNOWN
1225	allow_sharing = F_ALLOW_SHARING	1689	allow_sharing = F_ALLOW_SHARING
1226	allow_cycles = F_ALLOW_CYCLES	1690	allow_cycles = F_ALLOW_CYCLES
		1691	allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1692	forbid_objects = F_FORBID_OBJECTS
1227	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
1228	PPCODE:	1697	PPCODE:
1229	{	1698	{
1230	if (enable)	1699	if (enable)
1231	self->flags \|= ix;	1700	self->flags \|= ix;
1232	else	1701	else
…		…
1235	XPUSHs (ST (0));	1704	XPUSHs (ST (0));
1236	}	1705	}
1237		1706
1238	void get_shrink (CBOR *self)	1707	void get_shrink (CBOR *self)
1239	ALIAS:	1708	ALIAS:
1240	get_shrink = F_SHRINK	1709	get_shrink = F_SHRINK
1241	get_allow_unknown = F_ALLOW_UNKNOWN	1710	get_allow_unknown = F_ALLOW_UNKNOWN
1242	get_allow_sharing = F_ALLOW_SHARING	1711	get_allow_sharing = F_ALLOW_SHARING
1243	get_allow_cycles = F_ALLOW_CYCLES	1712	get_allow_cycles = F_ALLOW_CYCLES
		1713	get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
		1714	get_forbid_objects = F_FORBID_OBJECTS
1244	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
1245	PPCODE:	1719	PPCODE:
1246	XPUSHs (boolSV (self->flags & ix));	1720	XPUSHs (boolSV (self->flags & ix));
1247		1721
1248	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1722	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1249	PPCODE:	1723	PPCODE:
…		…
1298	EXTEND (SP, 2);	1772	EXTEND (SP, 2);
1299	PUSHs (sv);	1773	PUSHs (sv);
1300	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1774	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1301	}	1775	}
1302		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
1303	void DESTROY (CBOR *self)	1829	void DESTROY (CBOR *self)
1304	PPCODE:	1830	PPCODE:
1305	cbor_free (self);	1831	cbor_free (self);
1306		1832
1307	PROTOTYPES: ENABLE	1833	PROTOTYPES: ENABLE
…		…
1326	cbor_init (&cbor);	1852	cbor_init (&cbor);
1327	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1853	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1328	XPUSHs (cborstr);	1854	XPUSHs (cborstr);
1329	}	1855	}
1330		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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.37 by root, Sat Nov 30 18:13:53 2013 UTC vs. Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.37 by root, Sat Nov 30 18:13:53 2013 UTC vs.
Revision 1.80 by root, Fri Sep 8 20:03:06 2023 UTC