[ViewVC] Diff of: 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 1.66 by root, Sun Nov 29 21:32:01 2020 UTC

…		…
6	#include <string.h>	6	#include <string.h>
7	#include <stdlib.h>	7	#include <stdlib.h>
8	#include <stdio.h>	8	#include <stdio.h>
9	#include <limits.h>	9	#include <limits.h>
10	#include <float.h>	10	#include <float.h>
		11	#include <inttypes.h>
11		12
		13	#define ECB_NO_THREADS 1
12	#include "ecb.h"	14	#include "ecb.h"
13		15
14	// compatibility with perl <5.18	16	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	17	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	18	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	21	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	22	#endif
21	#ifndef HvNAMEUTF8	23	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	24	# define HvNAMEUTF8(hv) 0
23	#endif	25	#endif
		26	#ifndef SvREFCNT_inc_NN
		27	# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
		28	#endif
24	#ifndef SvREFCNT_dec_NN	29	#ifndef SvREFCNT_dec_NN
25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)	30	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26	#endif	31	#endif
27		32
28	// known major and minor types	33	// known major and minor types
…		…
94	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	99	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95		100
96	CBOR_TAG_MAGIC = 55799, // self-describe cbor	101	CBOR_TAG_MAGIC = 55799, // self-describe cbor
97	};	102	};
98		103
		104	// known forced types, also hardcoded in CBOR.pm
		105	enum
		106	{
		107	AS_CBOR = 0,
		108	AS_BYTES = 1,
		109	AS_TEXT = 2,
		110	AS_FLOAT16 = 3,
		111	AS_FLOAT32 = 4,
		112	AS_FLOAT64 = 5,
		113	// possibly future enhancements: float, integer
		114	};
		115
99	#define F_SHRINK 0x00000001UL	116	#define F_SHRINK 0x00000001UL
100	#define F_ALLOW_UNKNOWN 0x00000002UL	117	#define F_ALLOW_UNKNOWN 0x00000002UL
101	#define F_ALLOW_SHARING 0x00000004UL	118	#define F_ALLOW_SHARING 0x00000004UL
		119	#define F_ALLOW_CYCLES 0x00000008UL
		120	#define F_FORBID_OBJECTS 0x00000010UL
102	#define F_PACK_STRINGS 0x00000008UL	121	#define F_PACK_STRINGS 0x00000020UL
		122	#define F_TEXT_KEYS 0x00000040UL
		123	#define F_TEXT_STRINGS 0x00000080UL
		124	#define F_VALIDATE_UTF8 0x00000100UL
103		125
104	#define INIT_SIZE 32 // initial scalar size to be allocated	126	#define INIT_SIZE 32 // initial scalar size to be allocated
105		127
106	#define SB do {	128	#define SB do {
107	#define SE } while (0)	129	#define SE } while (0)
…		…
126	typedef struct {	148	typedef struct {
127	U32 flags;	149	U32 flags;
128	U32 max_depth;	150	U32 max_depth;
129	STRLEN max_size;	151	STRLEN max_size;
130	SV *filter;	152	SV *filter;
		153
		154	// for the incremental parser
		155	STRLEN incr_pos; // the current offset into the text
		156	STRLEN incr_need; // minimum bytes needed to decode
		157	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
131	} CBOR;	158	} CBOR;
132		159
133	ecb_inline void	160	ecb_inline void
134	cbor_init (CBOR *cbor)	161	cbor_init (CBOR *cbor)
135	{	162	{
…		…
139		166
140	ecb_inline void	167	ecb_inline void
141	cbor_free (CBOR *cbor)	168	cbor_free (CBOR *cbor)
142	{	169	{
143	SvREFCNT_dec (cbor->filter);	170	SvREFCNT_dec (cbor->filter);
		171	SvREFCNT_dec (cbor->incr_count);
144	}	172	}
145		173
146	/////////////////////////////////////////////////////////////////////////////	174	/////////////////////////////////////////////////////////////////////////////
147	// utility functions	175	// utility functions
148		176
…		…
174		202
175	// minimum length of a string to be registered for stringref	203	// minimum length of a string to be registered for stringref
176	ecb_inline int	204	ecb_inline int
177	minimum_string_length (UV idx)	205	minimum_string_length (UV idx)
178	{	206	{
179	return idx > 23	207	return idx <= 23 ? 3
180	? idx > 0xffU	208	: idx <= 0xffU ? 4
181	? idx > 0xffffU	209	: idx <= 0xffffU ? 5
182	? idx > 0xffffffffU	210	: idx <= 0xffffffffU ? 7
183	? 11	211	: 11;
184	: 7
185	: 5
186	: 4
187	: 3;
188	}	212	}
189		213
190	/////////////////////////////////////////////////////////////////////////////	214	/////////////////////////////////////////////////////////////////////////////
191	// encoder	215	// encoder
192		216
…		…
205	} enc_t;	229	} enc_t;
206		230
207	ecb_inline void	231	ecb_inline void
208	need (enc_t *enc, STRLEN len)	232	need (enc_t *enc, STRLEN len)
209	{	233	{
210	if (ecb_expect_false (enc->cur + len >= enc->end))	234	if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
211	{	235	{
212	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	236	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
213	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	237	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
214	enc->cur = SvPVX (enc->sv) + cur;	238	enc->cur = SvPVX (enc->sv) + cur;
215	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	239	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
…		…
221	{	245	{
222	need (enc, 1);	246	need (enc, 1);
223	*enc->cur++ = ch;	247	*enc->cur++ = ch;
224	}	248	}
225		249
		250	// used for tags, intregers, element counts and so on
226	static void	251	static void
227	encode_uint (enc_t *enc, int major, UV len)	252	encode_uint (enc_t *enc, int major, UV len)
228	{	253	{
229	need (enc, 9);	254	need (enc, 9);
230		255
231	if (ecb_expect_true (len < LENGTH_EXT1))	256	if (ecb_expect_true (len < LENGTH_EXT1))
232	*enc->cur++ = major \| len;	257	*enc->cur++ = major \| len;
233	else if (ecb_expect_true (len <= 0xff))	258	else if (ecb_expect_true (len <= 0xffU))
234	{	259	{
235	*enc->cur++ = major \| LENGTH_EXT1;	260	*enc->cur++ = major \| LENGTH_EXT1;
236	*enc->cur++ = len;	261	*enc->cur++ = len;
237	}	262	}
238	else if (len <= 0xffff)	263	else if (len <= 0xffffU)
239	{	264	{
240	*enc->cur++ = major \| LENGTH_EXT2;	265	*enc->cur++ = major \| LENGTH_EXT2;
241	*enc->cur++ = len >> 8;	266	*enc->cur++ = len >> 8;
242	*enc->cur++ = len;	267	*enc->cur++ = len;
243	}	268	}
244	else if (len <= 0xffffffff)	269	else if (len <= 0xffffffffU)
245	{	270	{
246	*enc->cur++ = major \| LENGTH_EXT4;	271	*enc->cur++ = major \| LENGTH_EXT4;
247	*enc->cur++ = len >> 24;	272	*enc->cur++ = len >> 24;
248	*enc->cur++ = len >> 16;	273	*enc->cur++ = len >> 16;
249	*enc->cur++ = len >> 8;	274	*enc->cur++ = len >> 8;
…		…
261	*enc->cur++ = len >> 8;	286	*enc->cur++ = len >> 8;
262	*enc->cur++ = len;	287	*enc->cur++ = len;
263	}	288	}
264	}	289	}
265		290
		291	// encodes a perl value into a CBOR integer
		292	ecb_inline
		293	encode_int (enc_t enc, SV sv)
		294	{
		295	if (SvIsUV (sv))
		296	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
		297	else if (SvIVX (sv) >= 0)
		298	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
		299	else
		300	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
		301	}
		302
266	ecb_inline void	303	ecb_inline void
267	encode_tag (enc_t *enc, UV tag)	304	encode_tag (enc_t *enc, UV tag)
268	{	305	{
269	encode_uint (enc, MAJOR_TAG, tag);	306	encode_uint (enc, MAJOR_TAG, tag);
270	}	307	}
271		308
		309	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		310	ecb_noinline static void
		311	encode_str_utf8 (enc_t enc, int utf8, char str, STRLEN len)
		312	{
		313	STRLEN ulen = len;
		314	U8 p, pend = (U8 *)str + len;
		315
		316	for (p = (U8 *)str; p < pend; ++p)
		317	ulen += *p >> 7; // count set high bits
		318
		319	encode_uint (enc, MAJOR_TEXT, ulen);
		320
		321	need (enc, ulen);
		322	for (p = (U8 *)str; p < pend; ++p)
		323	if (*p < 0x80)
		324	enc->cur++ = p;
		325	else
		326	{
		327	enc->cur++ = 0xc0 + (p >> 6);
		328	enc->cur++ = 0x80 + (p & 63);
		329	}
		330	}
		331
272	ecb_inline void	332	ecb_inline void
273	encode_str (enc_t enc, int utf8, char str, STRLEN len)	333	encode_str (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
274	{	334	{
		335	if (ecb_expect_false (upgrade_utf8))
		336	if (!utf8)
		337	{
		338	encode_str_utf8 (enc, utf8, str, len);
		339	return;
		340	}
		341
275	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);	342	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
276	need (enc, len);	343	need (enc, len);
277	memcpy (enc->cur, str, len);	344	memcpy (enc->cur, str, len);
278	enc->cur += len;	345	enc->cur += len;
279	}	346	}
280		347
281	static void	348	ecb_inline void
282	encode_strref (enc_t enc, int utf8, char str, STRLEN len)	349	encode_strref (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
283	{	350	{
284	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))	351	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
285	{	352	{
286	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	353	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
287		354
…		…
298	sv_setuv (*svp, enc->stringref_idx);	365	sv_setuv (*svp, enc->stringref_idx);
299	++enc->stringref_idx;	366	++enc->stringref_idx;
300	}	367	}
301	}	368	}
302		369
303	encode_str (enc, utf8, str, len);	370	encode_str (enc, upgrade_utf8, utf8, str, len);
		371	}
		372
		373	ecb_inline void
		374	encode_float16 (enc_t *enc, NV nv)
		375	{
		376	need (enc, 1+2);
		377
		378	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT16;
		379
		380	uint16_t fp = ecb_float_to_binary16 (nv);
		381
		382	if (!ecb_big_endian ())
		383	fp = ecb_bswap16 (fp);
		384
		385	memcpy (enc->cur, &fp, 2);
		386	enc->cur += 2;
		387	}
		388
		389	ecb_inline void
		390	encode_float32 (enc_t *enc, NV nv)
		391	{
		392	need (enc, 1+4);
		393
		394	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
		395
		396	uint32_t fp = ecb_float_to_binary32 (nv);
		397
		398	if (!ecb_big_endian ())
		399	fp = ecb_bswap32 (fp);
		400
		401	memcpy (enc->cur, &fp, 4);
		402	enc->cur += 4;
		403	}
		404
		405	ecb_inline void
		406	encode_float64 (enc_t *enc, NV nv)
		407	{
		408	need (enc, 1+8);
		409
		410	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
		411
		412	uint64_t fp = ecb_double_to_binary64 (nv);
		413
		414	if (!ecb_big_endian ())
		415	fp = ecb_bswap64 (fp);
		416
		417	memcpy (enc->cur, &fp, 8);
		418	enc->cur += 8;
		419	}
		420
		421	ecb_inline void
		422	encode_forced (enc_t enc, UV type, SV sv)
		423	{
		424	switch (type)
		425	{
		426	case AS_CBOR:
		427	{
		428	STRLEN len;
		429	char *str = SvPVbyte (sv, len);
		430
		431	need (enc, len);
		432	memcpy (enc->cur, str, len);
		433	enc->cur += len;
		434	}
		435	break;
		436
		437	case AS_BYTES:
		438	{
		439	STRLEN len;
		440	char *str = SvPVbyte (sv, len);
		441	encode_strref (enc, 0, 0, str, len);
		442	}
		443	break;
		444
		445	case AS_TEXT:
		446	{
		447	STRLEN len;
		448	char *str = SvPVutf8 (sv, len);
		449	encode_strref (enc, 1, 1, str, len);
		450	}
		451	break;
		452
		453	case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
		454	case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
		455	case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
		456
		457	default:
		458	croak ("encountered malformed CBOR::XS::Tagged object");
		459	}
304	}	460	}
305		461
306	static void encode_sv (enc_t enc, SV sv);	462	static void encode_sv (enc_t enc, SV sv);
307		463
308	static void	464	static void
…		…
315		471
316	++enc->depth;	472	++enc->depth;
317		473
318	encode_uint (enc, MAJOR_ARRAY, len + 1);	474	encode_uint (enc, MAJOR_ARRAY, len + 1);
319		475
		476	if (ecb_expect_false (SvMAGICAL (av)))
320	for (i = 0; i <= len; ++i)	477	for (i = 0; i <= len; ++i)
321	{	478	{
322	SV **svp = av_fetch (av, i, 0);	479	SV **svp = av_fetch (av, i, 0);
323	encode_sv (enc, svp ? *svp : &PL_sv_undef);	480	encode_sv (enc, svp ? *svp : &PL_sv_undef);
324	}	481	}
		482	else
		483	for (i = 0; i <= len; ++i)
		484	{
		485	SV *sv = AvARRAY (av)[i];
		486	encode_sv (enc, sv ? sv : &PL_sv_undef);
		487	}
325		488
326	--enc->depth;	489	--enc->depth;
327	}	490	}
328		491
329	static void	492	static void
…		…
337	++enc->depth;	500	++enc->depth;
338		501
339	int pairs = hv_iterinit (hv);	502	int pairs = hv_iterinit (hv);
340	int mg = SvMAGICAL (hv);	503	int mg = SvMAGICAL (hv);
341		504
342	if (mg)	505	if (ecb_expect_false (mg))
343	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);	506	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
344	else	507	else
345	encode_uint (enc, MAJOR_MAP, pairs);	508	encode_uint (enc, MAJOR_MAP, pairs);
346		509
347	while ((he = hv_iternext (hv)))	510	while ((he = hv_iternext (hv)))
348	{	511	{
349	if (HeKLEN (he) == HEf_SVKEY)	512	if (HeKLEN (he) == HEf_SVKEY)
350	encode_sv (enc, HeSVKEY (he));	513	encode_sv (enc, HeSVKEY (he));
351	else	514	else
352	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	515	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS \| F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
353		516
354	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	517	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
355	}	518	}
356		519
357	if (mg)	520	if (ecb_expect_false (mg))
358	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);	521	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
359		522
360	--enc->depth;	523	--enc->depth;
361	}	524	}
362		525
…		…
395	else if (stash == tagged_stash)	558	else if (stash == tagged_stash)
396	{	559	{
397	if (svt != SVt_PVAV)	560	if (svt != SVt_PVAV)
398	croak ("encountered CBOR::XS::Tagged object that isn't an array");	561	croak ("encountered CBOR::XS::Tagged object that isn't an array");
399		562
		563	switch (av_len ((AV *)sv))
		564	{
		565	case 2-1:
		566	// actually a tagged value
400	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));	567	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
401	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	568	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		569	break;
		570
		571	case 3-1:
		572	// a forced type [value, type, undef]
		573	encode_forced (enc, SvUV (av_fetch ((AV )sv, 1, 1)), av_fetch ((AV )sv, 0, 1));
		574	break;
		575
		576	default:
		577	croak ("encountered malformed CBOR::XS::Tagged object");
		578	}
402		579
403	return;	580	return;
404	}	581	}
405	}	582	}
406		583
407	if (ecb_expect_false (SvREFCNT (sv) > 1)	584	if (ecb_expect_false (SvREFCNT (sv) > 1)
408	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))	585	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
409	{	586	{
410	if (!enc->shareable)	587	if (ecb_expect_false (!enc->shareable))
411	enc->shareable = (HV )sv_2mortal ((SV )newHV ());	588	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
412		589
413	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);	590	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
414		591
415	if (SvOK (*svp))	592	if (SvOK (*svp))
…		…
429	if (ecb_expect_false (SvOBJECT (sv)))	606	if (ecb_expect_false (SvOBJECT (sv)))
430	{	607	{
431	HV *stash = SvSTASH (sv);	608	HV *stash = SvSTASH (sv);
432	GV *method;	609	GV *method;
433		610
		611	if (enc->cbor.flags & F_FORBID_OBJECTS)
		612	croak ("encountered object '%s', but forbid_objects is enabled",
		613	SvPV_nolen (sv_2mortal (newRV_inc (sv))));
434	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	614	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
435	{	615	{
436	dSP;	616	dSP;
437		617
438	ENTER; SAVETMPS; PUSHMARK (SP);	618	ENTER; SAVETMPS;
		619	PUSHMARK (SP);
439	// we re-bless the reference to get overload and other niceties right	620	// we re-bless the reference to get overload and other niceties right
440	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	621	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
441		622
442	PUTBACK;	623	PUTBACK;
443	// G_SCALAR ensures that return value is 1	624	// G_SCALAR ensures that return value is 1
…		…
456	}	637	}
457	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	638	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
458	{	639	{
459	dSP;	640	dSP;
460		641
461	ENTER; SAVETMPS; PUSHMARK (SP);	642	ENTER; SAVETMPS;
		643	PUSHMARK (SP);
462	EXTEND (SP, 2);	644	EXTEND (SP, 2);
463	// we re-bless the reference to get overload and other niceties right	645	// we re-bless the reference to get overload and other niceties right
464	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	646	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
465	PUSHs (sv_cbor);	647	PUSHs (sv_cbor);
466		648
…		…
472	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	654	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));	655	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
474		656
475	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	657	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
476	encode_uint (enc, MAJOR_ARRAY, count + 1);	658	encode_uint (enc, MAJOR_ARRAY, count + 1);
477	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	659	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
478		660
479	while (count)	661	{
		662	int i;
		663
		664	for (i = 0; i < count; ++i)
480	encode_sv (enc, SP[1 - count--]);	665	encode_sv (enc, SP[i + 1 - count]);
		666
		667	SP -= count;
		668	}
481		669
482	PUTBACK;	670	PUTBACK;
483		671
484	FREETMPS; LEAVE;	672	FREETMPS; LEAVE;
485	}	673	}
…		…
507		695
508	if (ecb_expect_false (nv == (NV)(U32)nv))	696	if (ecb_expect_false (nv == (NV)(U32)nv))
509	encode_uint (enc, MAJOR_POS_INT, (U32)nv);	697	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
510	//TODO: maybe I32?	698	//TODO: maybe I32?
511	else if (ecb_expect_false (nv == (float)nv))	699	else if (ecb_expect_false (nv == (float)nv))
512	{	700	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	701	else
524	{	702	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	}	703	}
536		704
537	static void	705	static void
538	encode_sv (enc_t enc, SV sv)	706	encode_sv (enc_t enc, SV sv)
539	{	707	{
…		…
541		709
542	if (SvPOKp (sv))	710	if (SvPOKp (sv))
543	{	711	{
544	STRLEN len;	712	STRLEN len;
545	char *str = SvPV (sv, len);	713	char *str = SvPV (sv, len);
546	encode_strref (enc, SvUTF8 (sv), str, len);	714	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
547	}	715	}
548	else if (SvNOKp (sv))	716	else if (SvNOKp (sv))
549	encode_nv (enc, sv);	717	encode_nv (enc, sv);
550	else if (SvIOKp (sv))	718	else if (SvIOKp (sv))
551	{	719	encode_int (e,v 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))	720	else if (SvROK (sv))
560	encode_rv (enc, SvRV (sv));	721	encode_rv (enc, SvRV (sv));
561	else if (!SvOK (sv))	722	else if (!SvOK (sv))
562	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);	723	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
563	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	724	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
…		…
568	}	729	}
569		730
570	static SV *	731	static SV *
571	encode_cbor (SV scalar, CBOR cbor)	732	encode_cbor (SV scalar, CBOR cbor)
572	{	733	{
573	enc_t enc = { };	734	enc_t enc = { 0 };
574		735
575	enc.cbor = *cbor;	736	enc.cbor = *cbor;
576	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	737	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
577	enc.cur = SvPVX (enc.sv);	738	enc.cur = SvPVX (enc.sv);
578	enc.end = SvEND (enc.sv);	739	enc.end = SvEND (enc.sv);
579		740
580	SvPOK_only (enc.sv);	741	SvPOK_only (enc.sv);
581		742
582	if (cbor->flags & F_PACK_STRINGS)	743	if (cbor->flags & F_PACK_STRINGS)
583	{	744	{
…		…
610	U32 depth; // recursion depth	771	U32 depth; // recursion depth
611	U32 maxdepth; // recursion depth limit	772	U32 maxdepth; // recursion depth limit
612	AV *shareable;	773	AV *shareable;
613	AV *stringref;	774	AV *stringref;
614	SV *decode_tagged;	775	SV *decode_tagged;
		776	SV *err_sv; // optional sv for error, needs to be freed
615	} dec_t;	777	} dec_t;
616		778
617	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	779	// set dec->err to ERRSV
		780	ecb_cold static void
		781	err_errsv (dec_t *dec)
		782	{
		783	if (!dec->err)
		784	{
		785	dec->err_sv = newSVsv (ERRSV);
618		786
619	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	787	// chop off the trailing \n
		788	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		789	*SvEND (dec->err_sv) = 0;
620		790
		791	dec->err = SvPVutf8_nolen (dec->err_sv);
		792	}
		793	}
		794
		795	// the following functions are used to reduce code size and help the compiler to optimise
		796	ecb_cold static void
		797	err_set (dec_t dec, const char reason)
		798	{
		799	if (!dec->err)
		800	dec->err = reason;
		801	}
		802
		803	ecb_cold static void
		804	err_unexpected_end (dec_t *dec)
		805	{
		806	err_set (dec, "unexpected end of CBOR data");
		807	}
		808
		809	#define ERR_DO(do) SB do; goto fail; SE
		810	#define ERR(reason) ERR_DO (err_set (dec, reason))
		811	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		812
		813	#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
		814
621	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	815	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
622	#define DEC_DEC_DEPTH --dec->depth	816	#define DEC_DEC_DEPTH --dec->depth
623		817
624	static UV	818	static UV
625	decode_uint (dec_t *dec)	819	decode_uint (dec_t *dec)
626	{	820	{
627	U8 m = *dec->cur & MINOR_MASK;	821	U8 m = *dec->cur & MINOR_MASK;
628	++dec->cur;	822	++dec->cur;
629		823
630	if (ecb_expect_true (m < LENGTH_EXT1))	824	if (ecb_expect_true (m < LENGTH_EXT1))
631	return m;	825	return m;
632		826	else if (ecb_expect_true (m == LENGTH_EXT1))
633	switch (m)
634	{	827	{
635	case LENGTH_EXT1:
636	WANT (1);	828	WANT (1);
637	dec->cur += 1;	829	dec->cur += 1;
638	return dec->cur[-1];	830	return dec->cur[-1];
639		831	}
640	case LENGTH_EXT2:	832	else if (ecb_expect_true (m == LENGTH_EXT2))
		833	{
641	WANT (2);	834	WANT (2);
642	dec->cur += 2;	835	dec->cur += 2;
643	return (((UV)dec->cur[-2]) << 8)	836	return (((UV)dec->cur[-2]) << 8)
644	\| ((UV)dec->cur[-1]);	837	\| ((UV)dec->cur[-1]);
645		838	}
646	case LENGTH_EXT4:	839	else if (ecb_expect_true (m == LENGTH_EXT4))
		840	{
647	WANT (4);	841	WANT (4);
648	dec->cur += 4;	842	dec->cur += 4;
649	return (((UV)dec->cur[-4]) << 24)	843	return (((UV)dec->cur[-4]) << 24)
650	\| (((UV)dec->cur[-3]) << 16)	844	\| (((UV)dec->cur[-3]) << 16)
651	\| (((UV)dec->cur[-2]) << 8)	845	\| (((UV)dec->cur[-2]) << 8)
652	\| ((UV)dec->cur[-1]);	846	\| ((UV)dec->cur[-1]);
653		847	}
654	case LENGTH_EXT8:	848	else if (ecb_expect_true (m == LENGTH_EXT8))
		849	{
655	WANT (8);	850	WANT (8);
656	dec->cur += 8;	851	dec->cur += 8;
657		852
658	return	853	return
659	#if UVSIZE < 8	854	#if UVSIZE < 8
660	0	855	0
661	#else	856	#else
662	(((UV)dec->cur[-8]) << 56)	857	(((UV)dec->cur[-8]) << 56)
663	\| (((UV)dec->cur[-7]) << 48)	858	\| (((UV)dec->cur[-7]) << 48)
664	\| (((UV)dec->cur[-6]) << 40)	859	\| (((UV)dec->cur[-6]) << 40)
665	\| (((UV)dec->cur[-5]) << 32)	860	\| (((UV)dec->cur[-5]) << 32)
666	#endif	861	#endif
667	\| (((UV)dec->cur[-4]) << 24)	862	\| (((UV)dec->cur[-4]) << 24)
668	\| (((UV)dec->cur[-3]) << 16)	863	\| (((UV)dec->cur[-3]) << 16)
669	\| (((UV)dec->cur[-2]) << 8)	864	\| (((UV)dec->cur[-2]) << 8)
670	\| ((UV)dec->cur[-1]);	865	\| ((UV)dec->cur[-1]);
671		866	}
672	default:	867	else
673	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	868	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
674	}
675		869
676	fail:	870	fail:
677	return 0;	871	return 0;
678	}	872	}
679		873
…		…
692		886
693	for (;;)	887	for (;;)
694	{	888	{
695	WANT (1);	889	WANT (1);
696		890
697	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	891	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
698	{	892	{
699	++dec->cur;	893	++dec->cur;
700	break;	894	break;
701	}	895	}
702		896
703	av_push (av, decode_sv (dec));	897	av_push (av, decode_sv (dec));
704	}	898	}
705	}	899	}
706	else	900	else
707	{	901	{
708	int i, len = decode_uint (dec);	902	UV i, len = decode_uint (dec);
709		903
		904	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);	905	av_fill (av, len - 1);
711		906
712	for (i = 0; i < len; ++i)	907	for (i = 0; i < len; ++i)
713	AvARRAY (av)[i] = decode_sv (dec);	908	AvARRAY (av)[i] = decode_sv (dec);
714	}	909	}
715		910
716	DEC_DEC_DEPTH;	911	DEC_DEC_DEPTH;
717	return newRV_noinc ((SV *)av);	912	return newRV_noinc ((SV *)av);
718		913
719	fail:	914	fail:
720	SvREFCNT_dec (av);	915	SvREFCNT_dec_NN (av);
721	DEC_DEC_DEPTH;	916	DEC_DEC_DEPTH;
722	return &PL_sv_undef;	917	return &PL_sv_undef;
723	}	918	}
724		919
725	static void	920	static void
…		…
727	{	922	{
728	// for speed reasons, we specialcase single-string	923	// for speed reasons, we specialcase single-string
729	// byte or utf-8 strings as keys, but only when !stringref	924	// byte or utf-8 strings as keys, but only when !stringref
730		925
731	if (ecb_expect_true (!dec->stringref))	926	if (ecb_expect_true (!dec->stringref))
732	if ((*dec->cur - MAJOR_BYTES) <= 27)	927	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
733	{	928	{
734	I32 len = decode_uint (dec);	929	STRLEN len = decode_uint (dec);
735	char key = (char )dec->cur;	930	char key = (char )dec->cur;
736		931
		932	WANT (len);
737	dec->cur += len;	933	dec->cur += len;
738		934
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);	935	hv_store (hv, key, len, decode_sv (dec), 0);
743		936
744	return;	937	return;
745	}	938	}
746	else if ((*dec->cur - MAJOR_TEXT) <= 27)	939	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747	{	940	{
748	I32 len = decode_uint (dec);	941	STRLEN len = decode_uint (dec);
749	char key = (char )dec->cur;	942	char key = (char )dec->cur;
750		943
		944	WANT (len);
751	dec->cur += len;	945	dec->cur += len;
752		946
753	if (ecb_expect_false (dec->stringref))	947	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	948	if (!is_utf8_string (key, len))
		949	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
755		950
756	hv_store (hv, key, -len, decode_sv (dec), 0);	951	hv_store (hv, key, -len, decode_sv (dec), 0);
757		952
758	return;	953	return;
759	}	954	}
760		955
761	SV *k = decode_sv (dec);	956	SV *k = decode_sv (dec);
762	SV *v = decode_sv (dec);	957	SV *v = decode_sv (dec);
763		958
		959	// we leak memory if uncaught exceptions are thrown by random magical
		960	// methods, and this is hopefully the only place where it can happen,
		961	// so if there is a chance of an exception, take the very slow path.
		962	// since catching exceptions is "undocumented/internal/forbidden" by
		963	// the new p5p powers, we need to call out to a perl function :/
		964	if (ecb_expect_false (SvAMAGIC (k)))
		965	{
		966	dSP;
		967
		968	ENTER; SAVETMPS;
		969	PUSHMARK (SP);
		970	EXTEND (SP, 3);
		971	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		972	PUSHs (sv_2mortal (k));
		973	PUSHs (sv_2mortal (v));
		974
		975	PUTBACK;
		976	call_pv ("CBOR::XS::_hv_store", G_VOID \| G_DISCARD \| G_EVAL);
		977	SPAGAIN;
		978
		979	FREETMPS; LEAVE;
		980
		981	if (SvTRUE (ERRSV))
		982	ERR_ERRSV;
		983
		984	return;
		985	}
		986
764	hv_store_ent (hv, k, v, 0);	987	hv_store_ent (hv, k, v, 0);
765	SvREFCNT_dec (k);	988	SvREFCNT_dec_NN (k);
		989
		990	fail:
		991	;
766	}	992	}
767		993
768	static SV *	994	static SV *
769	decode_hv (dec_t *dec)	995	decode_hv (dec_t *dec)
770	{	996	{
…		…
778		1004
779	for (;;)	1005	for (;;)
780	{	1006	{
781	WANT (1);	1007	WANT (1);
782		1008
783	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))	1009	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF) \|\| dec->err)
784	{	1010	{
785	++dec->cur;	1011	++dec->cur;
786	break;	1012	break;
787	}	1013	}
788		1014
789	decode_he (dec, hv);	1015	decode_he (dec, hv);
790	}	1016	}
791	}	1017	}
792	else	1018	else
793	{	1019	{
794	int pairs = decode_uint (dec);	1020	UV pairs = decode_uint (dec);
		1021
		1022	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
795		1023
796	while (pairs--)	1024	while (pairs--)
797	decode_he (dec, hv);	1025	decode_he (dec, hv);
798	}	1026	}
799		1027
800	DEC_DEC_DEPTH;	1028	DEC_DEC_DEPTH;
801	return newRV_noinc ((SV *)hv);	1029	return newRV_noinc ((SV *)hv);
802		1030
803	fail:	1031	fail:
804	SvREFCNT_dec (hv);	1032	SvREFCNT_dec_NN (hv);
805	DEC_DEC_DEPTH;	1033	DEC_DEC_DEPTH;
806	return &PL_sv_undef;	1034	return &PL_sv_undef;
807	}	1035	}
808		1036
809	static SV *	1037	static SV *
810	decode_str (dec_t *dec, int utf8)	1038	decode_str (dec_t *dec, int utf8)
811	{	1039	{
812	SV *sv = 0;	1040	SV *sv = 0;
813		1041
814	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)	1042	if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
815	{	1043	{
816	// indefinite length strings	1044	// indefinite length strings
817	++dec->cur;	1045	++dec->cur;
818		1046
819	U8 major = *dec->cur & MAJOR_MISC;	1047	U8 major = *dec->cur & MAJOR_MISC;
…		…
852	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	1080	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
853	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	1081	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
854	}	1082	}
855		1083
856	if (utf8)	1084	if (utf8)
		1085	{
		1086	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		1087	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		1088	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		1089
857	SvUTF8_on (sv);	1090	SvUTF8_on (sv);
		1091	}
858		1092
859	return sv;	1093	return sv;
860		1094
861	fail:	1095	fail:
862	SvREFCNT_dec (sv);	1096	SvREFCNT_dec (sv);
…		…
881	sv = newRV_noinc (decode_sv (dec));	1115	sv = newRV_noinc (decode_sv (dec));
882	break;	1116	break;
883		1117
884	case CBOR_TAG_STRINGREF_NAMESPACE:	1118	case CBOR_TAG_STRINGREF_NAMESPACE:
885	{	1119	{
		1120	// do not use SAVETMPS/FREETMPS, as these will
		1121	// erase mortalised caches, e.g. "shareable"
886	ENTER; SAVETMPS;	1122	ENTER;
887		1123
888	SAVESPTR (dec->stringref);	1124	SAVESPTR (dec->stringref);
889	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1125	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
890		1126
891	sv = decode_sv (dec);	1127	sv = decode_sv (dec);
892		1128
893	FREETMPS; LEAVE;	1129	LEAVE;
894	}	1130	}
895	break;	1131	break;
896		1132
897	case CBOR_TAG_STRINGREF:	1133	case CBOR_TAG_STRINGREF:
898	{	1134	{
899	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1135	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
900	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1136	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
901		1137
902	UV idx = decode_uint (dec);	1138	UV idx = decode_uint (dec);
903		1139
904	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	1140	if (!dec->stringref \|\| idx >= (UV)(1 + AvFILLp (dec->stringref)))
905	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");	1141	ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
906		1142
907	sv = newSVsv (AvARRAY (dec->stringref)[idx]);	1143	sv = newSVsv (AvARRAY (dec->stringref)[idx]);
908	}	1144	}
909	break;	1145	break;
…		…
911	case CBOR_TAG_VALUE_SHAREABLE:	1147	case CBOR_TAG_VALUE_SHAREABLE:
912	{	1148	{
913	if (ecb_expect_false (!dec->shareable))	1149	if (ecb_expect_false (!dec->shareable))
914	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1150	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
915		1151
		1152	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1153	{
916	sv = newSV (0);	1154	sv = newSV (0);
917	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1155	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
918		1156
919	SV *osv = decode_sv (dec);	1157	SV *osv = decode_sv (dec);
920	sv_setsv (sv, osv);	1158	sv_setsv (sv, osv);
921	SvREFCNT_dec_NN (osv);	1159	SvREFCNT_dec_NN (osv);
		1160	}
		1161	else
		1162	{
		1163	av_push (dec->shareable, &PL_sv_undef);
		1164	int idx = AvFILLp (dec->shareable);
		1165	sv = decode_sv (dec);
		1166	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1167	}
922	}	1168	}
923	break;	1169	break;
924		1170
925	case CBOR_TAG_VALUE_SHAREDREF:	1171	case CBOR_TAG_VALUE_SHAREDREF:
926	{	1172	{
927	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))	1173	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
928	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1174	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
929		1175
930	UV idx = decode_uint (dec);	1176	UV idx = decode_uint (dec);
931		1177
932	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1178	if (!dec->shareable \|\| idx >= (UV)(1 + AvFILLp (dec->shareable)))
933	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1179	ERR ("corrupted CBOR data (sharedref index out of bounds)");
934		1180
935	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1181	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1182
		1183	if (sv == &PL_sv_undef)
		1184	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
936	}	1185	}
937	break;	1186	break;
938		1187
939	case CBOR_TAG_PERL_OBJECT:	1188	case CBOR_TAG_PERL_OBJECT:
940	{	1189	{
		1190	if (dec->cbor.flags & F_FORBID_OBJECTS)
		1191	goto filter;
		1192
941	sv = decode_sv (dec);	1193	sv = decode_sv (dec);
942		1194
943	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)	1195	if (!SvROK (sv) \|\| SvTYPE (SvRV (sv)) != SVt_PVAV)
944	ERR ("corrupted CBOR data (non-array perl object)");	1196	ERR ("corrupted CBOR data (non-array perl object)");
945		1197
…		…
955	if (!method)	1207	if (!method)
956	ERR ("cannot decode perl-object (package does not have a THAW method)");	1208	ERR ("cannot decode perl-object (package does not have a THAW method)");
957		1209
958	dSP;	1210	dSP;
959		1211
960	ENTER; SAVETMPS; PUSHMARK (SP);	1212	ENTER; SAVETMPS;
		1213	PUSHMARK (SP);
961	EXTEND (SP, len + 1);	1214	EXTEND (SP, len + 1);
962	// we re-bless the reference to get overload and other niceties right	1215	// we re-bless the reference to get overload and other niceties right
963	PUSHs (*av_fetch (av, 0, 1));	1216	PUSHs (*av_fetch (av, 0, 1));
964	PUSHs (sv_cbor);	1217	PUSHs (sv_cbor);
965		1218
…		…
973	SPAGAIN;	1226	SPAGAIN;
974		1227
975	if (SvTRUE (ERRSV))	1228	if (SvTRUE (ERRSV))
976	{	1229	{
977	FREETMPS; LEAVE;	1230	FREETMPS; LEAVE;
978	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1231	ERR_ERRSV;
979	}	1232	}
980		1233
981	SvREFCNT_dec (sv);	1234	SvREFCNT_dec_NN (sv);
982	sv = SvREFCNT_inc (POPs);	1235	sv = SvREFCNT_inc (POPs);
983		1236
984	PUTBACK;	1237	PUTBACK;
985		1238
986	FREETMPS; LEAVE;	1239	FREETMPS; LEAVE;
987	}	1240	}
988	break;	1241	break;
989		1242
990	default:	1243	default:
		1244	filter:
991	{	1245	{
		1246	SV *tag_sv = newSVuv (tag);
		1247
992	sv = decode_sv (dec);	1248	sv = decode_sv (dec);
993		1249
994	dSP;	1250	dSP;
995	ENTER; SAVETMPS; PUSHMARK (SP);	1251	ENTER; SAVETMPS;
		1252	PUSHMARK (SP);
996	EXTEND (SP, 2);	1253	EXTEND (SP, 2);
997	PUSHs (newSVuv (tag));	1254	PUSHs (tag_sv);
998	PUSHs (sv);	1255	PUSHs (sv);
999		1256
1000	PUTBACK;	1257	PUTBACK;
1001	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);	1258	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
1002	SPAGAIN;	1259	SPAGAIN;
1003		1260
1004	if (SvTRUE (ERRSV))	1261	if (SvTRUE (ERRSV))
1005	{	1262	{
		1263	SvREFCNT_dec_NN (tag_sv);
1006	FREETMPS; LEAVE;	1264	FREETMPS; LEAVE;
1007	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1265	ERR_ERRSV;
1008	}	1266	}
1009		1267
1010	if (count)	1268	if (count)
1011	{	1269	{
		1270	SvREFCNT_dec_NN (tag_sv);
1012	SvREFCNT_dec (sv);	1271	SvREFCNT_dec_NN (sv);
1013	sv = SvREFCNT_inc (POPs);	1272	sv = SvREFCNT_inc_NN (TOPs);
		1273	SP -= count;
1014	}	1274	}
1015	else	1275	else
1016	{	1276	{
1017	AV *av = newAV ();	1277	AV *av = newAV ();
1018	av_push (av, newSVuv (tag));	1278	av_push (av, tag_sv);
1019	av_push (av, sv);	1279	av_push (av, sv);
1020		1280
1021	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1281	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
1022	? cbor_tagged_stash	1282	? cbor_tagged_stash
1023	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1283	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
1109		1369
1110	return newSVnv (ecb_binary64_to_double (fp));	1370	return newSVnv (ecb_binary64_to_double (fp));
1111	}	1371	}
1112		1372
1113	// 0..19 unassigned simple	1373	// 0..19 unassigned simple
1114	// 24 reserved + unassigned (reserved values are not encodable)	1374	// 24 reserved + unassigned simple (reserved values are not encodable)
		1375	// 28-30 unassigned misc
		1376	// 31 break code
1115	default:	1377	default:
1116	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1378	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1117	}	1379	}
1118		1380
1119	break;	1381	break;
1120	}	1382	}
1121		1383
…		…
1124	}	1386	}
1125		1387
1126	static SV *	1388	static SV *
1127	decode_cbor (SV string, CBOR cbor, char **offset_return)	1389	decode_cbor (SV string, CBOR cbor, char **offset_return)
1128	{	1390	{
1129	dec_t dec = { };	1391	dec_t dec = { 0 };
1130	SV *sv;	1392	SV *sv;
1131	STRLEN len;	1393	STRLEN len;
1132	char *data = SvPVbyte (string, len);	1394	char *data = SvPVbyte (string, len);
1133		1395
1134	if (len > cbor->max_size && cbor->max_size)	1396	if (len > cbor->max_size && cbor->max_size)
…		…
1148	if (dec.cur != dec.end && !dec.err)	1410	if (dec.cur != dec.end && !dec.err)
1149	dec.err = "garbage after CBOR object";	1411	dec.err = "garbage after CBOR object";
1150		1412
1151	if (dec.err)	1413	if (dec.err)
1152	{	1414	{
		1415	if (dec.shareable)
		1416	{
		1417	// need to break cyclic links, which would all be in shareable
		1418	int i;
		1419	SV **svp;
		1420
		1421	for (i = av_len (dec.shareable) + 1; i--; )
		1422	if ((svp = av_fetch (dec.shareable, i, 0)))
		1423	sv_setsv (*svp, &PL_sv_undef);
		1424	}
		1425
1153	SvREFCNT_dec (sv);	1426	SvREFCNT_dec_NN (sv);
		1427
		1428	if (dec.err_sv)
		1429	sv_2mortal (dec.err_sv);
		1430
1154	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1431	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1155	}	1432	}
1156		1433
1157	sv = sv_2mortal (sv);	1434	sv = sv_2mortal (sv);
1158		1435
1159	return sv;	1436	return sv;
1160	}	1437	}
1161		1438
		1439	/////////////////////////////////////////////////////////////////////////////
		1440	// incremental parser
		1441
		1442	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1443
		1444	// returns 0 for notyet, 1 for success or error
		1445	static int
		1446	incr_parse (CBOR self, SV cborstr)
		1447	{
		1448	STRLEN cur;
		1449	SvPV (cborstr, cur);
		1450
		1451	while (ecb_expect_true (self->incr_need <= cur))
		1452	{
		1453	// table of integer count bytes
		1454	static I8 incr_len[MINOR_MASK + 1] = {
		1455	0, 0, 0, 0, 0, 0, 0, 0,
		1456	0, 0, 0, 0, 0, 0, 0, 0,
		1457	0, 0, 0, 0, 0, 0, 0, 0,
		1458	1, 2, 4, 8,-1,-1,-1,-2
		1459	};
		1460
		1461	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1462	U8 m = *p & MINOR_MASK;
		1463	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1464	I8 ilen = incr_len[m];
		1465
		1466	self->incr_need = self->incr_pos + 1;
		1467
		1468	if (ecb_expect_false (ilen < 0))
		1469	{
		1470	if (m != MINOR_INDEF)
		1471	return 1; // error
		1472
		1473	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1474	{
		1475	if (count >= 0)
		1476	return 1; // error
		1477
		1478	count = 1;
		1479	}
		1480	else
		1481	{
		1482	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1483	count = -1;
		1484	}
		1485	}
		1486	else
		1487	{
		1488	self->incr_need += ilen;
		1489	if (ecb_expect_false (self->incr_need > cur))
		1490	return 0;
		1491
		1492	int major = *p >> MAJOR_SHIFT;
		1493
		1494	switch (major)
		1495	{
		1496	case MAJOR_TAG >> MAJOR_SHIFT:
		1497	++count; // tags merely prefix another value
		1498	break;
		1499
		1500	case MAJOR_BYTES >> MAJOR_SHIFT:
		1501	case MAJOR_TEXT >> MAJOR_SHIFT:
		1502	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1503	case MAJOR_MAP >> MAJOR_SHIFT:
		1504	{
		1505	UV len;
		1506
		1507	if (ecb_expect_false (ilen))
		1508	{
		1509	len = 0;
		1510
		1511	do {
		1512	len = (len << 8) \| *++p;
		1513	} while (--ilen);
		1514	}
		1515	else
		1516	len = m;
		1517
		1518	switch (major)
		1519	{
		1520	case MAJOR_BYTES >> MAJOR_SHIFT:
		1521	case MAJOR_TEXT >> MAJOR_SHIFT:
		1522	self->incr_need += len;
		1523	if (ecb_expect_false (self->incr_need > cur))
		1524	return 0;
		1525
		1526	break;
		1527
		1528	case MAJOR_MAP >> MAJOR_SHIFT:
		1529	len <<= 1;
		1530	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1531	if (len)
		1532	{
		1533	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1534	count = len + 1;
		1535	}
		1536	break;
		1537	}
		1538	}
		1539	}
		1540	}
		1541
		1542	self->incr_pos = self->incr_need;
		1543
		1544	if (count > 0)
		1545	{
		1546	while (!--count)
		1547	{
		1548	if (!AvFILLp (self->incr_count))
		1549	return 1; // done
		1550
		1551	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1552	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1553	}
		1554
		1555	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1556	}
		1557	}
		1558
		1559	return 0;
		1560	}
		1561
		1562
1162	/////////////////////////////////////////////////////////////////////////////	1563	/////////////////////////////////////////////////////////////////////////////
1163	// XS interface functions	1564	// XS interface functions
1164		1565
1165	MODULE = CBOR::XS PACKAGE = CBOR::XS	1566	MODULE = CBOR::XS PACKAGE = CBOR::XS
1166		1567
…		…
1178		1579
1179	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1580	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1180		1581
1181	sv_cbor = newSVpv ("CBOR", 0);	1582	sv_cbor = newSVpv ("CBOR", 0);
1182	SvREADONLY_on (sv_cbor);	1583	SvREADONLY_on (sv_cbor);
		1584
		1585	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1183	}	1586	}
1184		1587
1185	PROTOTYPES: DISABLE	1588	PROTOTYPES: DISABLE
1186		1589
1187	void CLONE (...)	1590	void CLONE (...)
…		…
1206	void shrink (CBOR *self, int enable = 1)	1609	void shrink (CBOR *self, int enable = 1)
1207	ALIAS:	1610	ALIAS:
1208	shrink = F_SHRINK	1611	shrink = F_SHRINK
1209	allow_unknown = F_ALLOW_UNKNOWN	1612	allow_unknown = F_ALLOW_UNKNOWN
1210	allow_sharing = F_ALLOW_SHARING	1613	allow_sharing = F_ALLOW_SHARING
		1614	allow_cycles = F_ALLOW_CYCLES
		1615	forbid_objects = F_FORBID_OBJECTS
1211	pack_strings = F_PACK_STRINGS	1616	pack_strings = F_PACK_STRINGS
		1617	text_keys = F_TEXT_KEYS
		1618	text_strings = F_TEXT_STRINGS
		1619	validate_utf8 = F_VALIDATE_UTF8
1212	PPCODE:	1620	PPCODE:
1213	{	1621	{
1214	if (enable)	1622	if (enable)
1215	self->flags \|= ix;	1623	self->flags \|= ix;
1216	else	1624	else
…		…
1222	void get_shrink (CBOR *self)	1630	void get_shrink (CBOR *self)
1223	ALIAS:	1631	ALIAS:
1224	get_shrink = F_SHRINK	1632	get_shrink = F_SHRINK
1225	get_allow_unknown = F_ALLOW_UNKNOWN	1633	get_allow_unknown = F_ALLOW_UNKNOWN
1226	get_allow_sharing = F_ALLOW_SHARING	1634	get_allow_sharing = F_ALLOW_SHARING
		1635	get_allow_cycles = F_ALLOW_CYCLES
		1636	get_forbid_objects = F_FORBID_OBJECTS
1227	get_pack_strings = F_PACK_STRINGS	1637	get_pack_strings = F_PACK_STRINGS
		1638	get_text_keys = F_TEXT_KEYS
		1639	get_text_strings = F_TEXT_STRINGS
		1640	get_validate_utf8 = F_VALIDATE_UTF8
1228	PPCODE:	1641	PPCODE:
1229	XPUSHs (boolSV (self->flags & ix));	1642	XPUSHs (boolSV (self->flags & ix));
1230		1643
1231	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1644	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1232	PPCODE:	1645	PPCODE:
…		…
1281	EXTEND (SP, 2);	1694	EXTEND (SP, 2);
1282	PUSHs (sv);	1695	PUSHs (sv);
1283	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1696	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1284	}	1697	}
1285		1698
		1699	void incr_parse (CBOR self, SV cborstr)
		1700	ALIAS:
		1701	incr_parse_multiple = 1
		1702	PPCODE:
		1703	{
		1704	if (SvUTF8 (cborstr))
		1705	sv_utf8_downgrade (cborstr, 0);
		1706
		1707	if (!self->incr_count)
		1708	{
		1709	self->incr_count = newAV ();
		1710	self->incr_pos = 0;
		1711	self->incr_need = 1;
		1712
		1713	av_push (self->incr_count, newSViv (1));
		1714	}
		1715
		1716	do
		1717	{
		1718	if (!incr_parse (self, cborstr))
		1719	{
		1720	if (self->incr_need > self->max_size && self->max_size)
		1721	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1722	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1723
		1724	break;
		1725	}
		1726
		1727	SV *sv;
		1728	char *offset;
		1729
		1730	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1731	XPUSHs (sv);
		1732
		1733	sv_chop (cborstr, offset);
		1734
		1735	av_clear (self->incr_count);
		1736	av_push (self->incr_count, newSViv (1));
		1737
		1738	self->incr_pos = 0;
		1739	self->incr_need = self->incr_pos + 1;
		1740	}
		1741	while (ix);
		1742	}
		1743
		1744	void incr_reset (CBOR *self)
		1745	CODE:
		1746	{
		1747	SvREFCNT_dec (self->incr_count);
		1748	self->incr_count = 0;
		1749	}
		1750
1286	void DESTROY (CBOR *self)	1751	void DESTROY (CBOR *self)
1287	PPCODE:	1752	PPCODE:
1288	cbor_free (self);	1753	cbor_free (self);
1289		1754
1290	PROTOTYPES: ENABLE	1755	PROTOTYPES: ENABLE
1291		1756
1292	void encode_cbor (SV *scalar)	1757	void encode_cbor (SV *scalar)
		1758	ALIAS:
		1759	encode_cbor = 0
		1760	encode_cbor_sharing = F_ALLOW_SHARING
1293	PPCODE:	1761	PPCODE:
1294	{	1762	{
1295	CBOR cbor;	1763	CBOR cbor;
1296	cbor_init (&cbor);	1764	cbor_init (&cbor);
		1765	cbor.flags \|= ix;
1297	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1766	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1298	XPUSHs (scalar);	1767	XPUSHs (scalar);
1299	}	1768	}
1300		1769
1301	void decode_cbor (SV *cborstr)	1770	void decode_cbor (SV *cborstr)
…		…
1305	cbor_init (&cbor);	1774	cbor_init (&cbor);
1306	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;	1775	PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1307	XPUSHs (cborstr);	1776	XPUSHs (cborstr);
1308	}	1777	}
1309		1778
		1779	#ifdef __AFL_COMPILER
		1780
		1781	void
		1782	afl_init ()
		1783	CODE:
		1784	__AFL_INIT ();
		1785
		1786	int
		1787	afl_loop (unsigned int count = 10000)
		1788	CODE:
		1789	RETVAL = __AFL_LOOP (count);
		1790	OUTPUT:
		1791	RETVAL
		1792
		1793	#endif
		1794

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs. Revision 1.66 by root, Sun Nov 29 21:32:01 2020 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.35 by root, Sat Nov 30 17:19:34 2013 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 2020 UTC