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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC

…		…
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		11
		12	#define ECB_NO_THREADS 1
12	#include "ecb.h"	13	#include "ecb.h"
13		14
14	// compatibility with perl <5.18	15	// compatibility with perl <5.18
15	#ifndef HvNAMELEN_get	16	#ifndef HvNAMELEN_get
16	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))	17	# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	20	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	21	#endif
21	#ifndef HvNAMEUTF8	22	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	23	# define HvNAMEUTF8(hv) 0
23	#endif	24	#endif
		25	#ifndef SvREFCNT_dec_NN
		26	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		27	#endif
		28
		29	// known major and minor types
		30	enum cbor_type
		31	{
		32	MAJOR_SHIFT = 5,
		33	MINOR_MASK = 0x1f,
		34
		35	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		36	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		37	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		38	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		39	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		40	MAJOR_MAP = 5 << MAJOR_SHIFT,
		41	MAJOR_TAG = 6 << MAJOR_SHIFT,
		42	MAJOR_MISC = 7 << MAJOR_SHIFT,
		43
		44	// INT/STRING/ARRAY/MAP subtypes
		45	LENGTH_EXT1 = 24,
		46	LENGTH_EXT2 = 25,
		47	LENGTH_EXT4 = 26,
		48	LENGTH_EXT8 = 27,
		49
		50	// SIMPLE types (effectively MISC subtypes)
		51	SIMPLE_FALSE = 20,
		52	SIMPLE_TRUE = 21,
		53	SIMPLE_NULL = 22,
		54	SIMPLE_UNDEF = 23,
		55
		56	// MISC subtype (unused)
		57	MISC_EXT1 = 24,
		58	MISC_FLOAT16 = 25,
		59	MISC_FLOAT32 = 26,
		60	MISC_FLOAT64 = 27,
		61
		62	// BYTES/TEXT/ARRAY/MAP
		63	MINOR_INDEF = 31,
		64	};
24		65
25	// known tags	66	// known tags
26	enum cbor_tag	67	enum cbor_tag
27	{	68	{
28	// inofficial extensions (pending iana registration)	69	// extensions
		70	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	71	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	72	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	73	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	74	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	75	CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34	CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection	76	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36		77
37	// rfc7049	78	// rfc7049
38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	79	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	80	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40	CBOR_TAG_POS_BIGNUM = 2, // byte string	81	CBOR_TAG_POS_BIGNUM = 2, // byte string
41	CBOR_TAG_NEG_BIGNUM = 3, // byte string	82	CBOR_TAG_NEG_BIGNUM = 3, // byte string
42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	83	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43	CBOR_TAG_BIGFLOAT = 5, // array	84	CBOR_TAG_BIGFLOAT = 5, // array
44		85
45	CBOR_TAG_CONV_B64U = 21, // base64url, any	86	CBOR_TAG_CONV_B64U = 21, // base64url, any
46	CBOR_TAG_CONV_B64 = 22, // base64, any	87	CBOR_TAG_CONV_B64 = 22, // base64, any
47	CBOR_TAG_CONV_HEX = 23, // base16, any	88	CBOR_TAG_CONV_HEX = 23, // base16, any
48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	89	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49		90
50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	91	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	92	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	93	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	94	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	95	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55		96
56	CBOR_TAG_MAGIC = 55799 // self-describe cbor	97	CBOR_TAG_MAGIC = 55799, // self-describe cbor
57	};	98	};
58		99
59	#define F_SHRINK 0x00000001UL	100	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	101	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	102	#define F_ALLOW_SHARING 0x00000004UL
62	#define F_ALLOW_STRINGREF 0x00000008UL //TODO	103	#define F_ALLOW_CYCLES 0x00000008UL
		104	#define F_PACK_STRINGS 0x00000010UL
		105	#define F_TEXT_KEYS 0x00000020UL
		106	#define F_TEXT_STRINGS 0x00000040UL
		107	#define F_VALIDATE_UTF8 0x00000080UL
63		108
64	#define INIT_SIZE 32 // initial scalar size to be allocated	109	#define INIT_SIZE 32 // initial scalar size to be allocated
65		110
66	#define SB do {	111	#define SB do {
67	#define SE } while (0)	112	#define SE } while (0)
…		…
86	typedef struct {	131	typedef struct {
87	U32 flags;	132	U32 flags;
88	U32 max_depth;	133	U32 max_depth;
89	STRLEN max_size;	134	STRLEN max_size;
90	SV *filter;	135	SV *filter;
		136
		137	// for the incremental parser
		138	STRLEN incr_pos; // the current offset into the text
		139	STRLEN incr_need; // minimum bytes needed to decode
		140	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	141	} CBOR;
92		142
93	ecb_inline void	143	ecb_inline void
94	cbor_init (CBOR *cbor)	144	cbor_init (CBOR *cbor)
95	{	145	{
…		…
99		149
100	ecb_inline void	150	ecb_inline void
101	cbor_free (CBOR *cbor)	151	cbor_free (CBOR *cbor)
102	{	152	{
103	SvREFCNT_dec (cbor->filter);	153	SvREFCNT_dec (cbor->filter);
		154	SvREFCNT_dec (cbor->incr_count);
104	}	155	}
105		156
106	/////////////////////////////////////////////////////////////////////////////	157	/////////////////////////////////////////////////////////////////////////////
107	// utility functions	158	// utility functions
108		159
…		…
138	{	189	{
139	return idx > 23	190	return idx > 23
140	? idx > 0xffU	191	? idx > 0xffU
141	? idx > 0xffffU	192	? idx > 0xffffU
142	? idx > 0xffffffffU	193	? idx > 0xffffffffU
		194	? 11
143	? 7	195	: 7
144	: 6
145	: 5	196	: 5
146	: 4	197	: 4
147	: 3;	198	: 3;
148	}	199	}
149		200
…		…
186	static void	237	static void
187	encode_uint (enc_t *enc, int major, UV len)	238	encode_uint (enc_t *enc, int major, UV len)
188	{	239	{
189	need (enc, 9);	240	need (enc, 9);
190		241
191	if (len < 24)	242	if (ecb_expect_true (len < LENGTH_EXT1))
192	*enc->cur++ = major \| len;	243	*enc->cur++ = major \| len;
193	else if (len <= 0xff)	244	else if (ecb_expect_true (len <= 0xffU))
194	{	245	{
195	*enc->cur++ = major \| 24;	246	*enc->cur++ = major \| LENGTH_EXT1;
196	*enc->cur++ = len;	247	*enc->cur++ = len;
197	}	248	}
198	else if (len <= 0xffff)	249	else if (len <= 0xffffU)
199	{	250	{
200	*enc->cur++ = major \| 25;	251	*enc->cur++ = major \| LENGTH_EXT2;
201	*enc->cur++ = len >> 8;	252	*enc->cur++ = len >> 8;
202	*enc->cur++ = len;	253	*enc->cur++ = len;
203	}	254	}
204	else if (len <= 0xffffffff)	255	else if (len <= 0xffffffffU)
205	{	256	{
206	*enc->cur++ = major \| 26;	257	*enc->cur++ = major \| LENGTH_EXT4;
207	*enc->cur++ = len >> 24;	258	*enc->cur++ = len >> 24;
208	*enc->cur++ = len >> 16;	259	*enc->cur++ = len >> 16;
209	*enc->cur++ = len >> 8;	260	*enc->cur++ = len >> 8;
210	*enc->cur++ = len;	261	*enc->cur++ = len;
211	}	262	}
212	else	263	else
213	{	264	{
214	*enc->cur++ = major \| 27;	265	*enc->cur++ = major \| LENGTH_EXT8;
215	*enc->cur++ = len >> 56;	266	*enc->cur++ = len >> 56;
216	*enc->cur++ = len >> 48;	267	*enc->cur++ = len >> 48;
217	*enc->cur++ = len >> 40;	268	*enc->cur++ = len >> 40;
218	*enc->cur++ = len >> 32;	269	*enc->cur++ = len >> 32;
219	*enc->cur++ = len >> 24;	270	*enc->cur++ = len >> 24;
…		…
224	}	275	}
225		276
226	ecb_inline void	277	ecb_inline void
227	encode_tag (enc_t *enc, UV tag)	278	encode_tag (enc_t *enc, UV tag)
228	{	279	{
229	encode_uint (enc, 0xc0, tag);	280	encode_uint (enc, MAJOR_TAG, tag);
230	}	281	}
231		282
232	static void	283	// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
		284	ecb_noinline static void
233	encode_str (enc_t enc, int utf8, char str, STRLEN len)	285	encode_str_utf8 (enc_t enc, int utf8, char str, STRLEN len)
234	{	286	{
		287	STRLEN ulen = len;
		288	U8 p, pend = (U8 *)str + len;
		289
		290	for (p = (U8 *)str; p < pend; ++p)
		291	ulen += *p >> 7; // count set high bits
		292
		293	encode_uint (enc, MAJOR_TEXT, ulen);
		294
		295	need (enc, ulen);
		296	for (p = (U8 *)str; p < pend; ++p)
		297	if (*p < 0x80)
		298	enc->cur++ = p;
		299	else
		300	{
		301	enc->cur++ = 0xc0 + (p >> 6);
		302	enc->cur++ = 0x80 + (p & 63);
		303	}
		304	}
		305
		306	ecb_inline void
		307	encode_str (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
		308	{
		309	if (ecb_expect_false (upgrade_utf8))
		310	if (!utf8)
		311	{
		312	encode_str_utf8 (enc, utf8, str, len);
		313	return;
		314	}
		315
		316	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		317	need (enc, len);
		318	memcpy (enc->cur, str, len);
		319	enc->cur += len;
		320	}
		321
		322	ecb_inline void
		323	encode_strref (enc_t enc, int upgrade_utf8, int utf8, char str, STRLEN len)
		324	{
235	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	325	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236	{	326	{
237	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	327	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238		328
239	if (SvOK (*svp))	329	if (SvOK (*svp))
240	{	330	{
241	// already registered, use stringref	331	// already registered, use stringref
242	encode_tag (enc, CBOR_TAG_STRINGREF);	332	encode_tag (enc, CBOR_TAG_STRINGREF);
243	encode_uint (enc, 0x00, SvUV (*svp));	333	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244	return;	334	return;
245	}	335	}
246	else if (len >= minimum_string_length (enc->stringref_idx))	336	else if (len >= minimum_string_length (enc->stringref_idx))
247	{	337	{
248	// register only	338	// register only
249	sv_setuv (*svp, enc->stringref_idx);	339	sv_setuv (*svp, enc->stringref_idx);
250	++enc->stringref_idx;	340	++enc->stringref_idx;
251	}	341	}
252	}	342	}
253		343
254	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	344	encode_str (enc, upgrade_utf8, utf8, str, len);
255	need (enc, len);
256	memcpy (enc->cur, str, len);
257	enc->cur += len;
258	}	345	}
259		346
260	static void encode_sv (enc_t enc, SV sv);	347	static void encode_sv (enc_t enc, SV sv);
261		348
262	static void	349	static void
…		…
267	if (enc->depth >= enc->cbor.max_depth)	354	if (enc->depth >= enc->cbor.max_depth)
268	croak (ERR_NESTING_EXCEEDED);	355	croak (ERR_NESTING_EXCEEDED);
269		356
270	++enc->depth;	357	++enc->depth;
271		358
272	encode_uint (enc, 0x80, len + 1);	359	encode_uint (enc, MAJOR_ARRAY, len + 1);
273		360
		361	if (SvMAGICAL (av))
274	for (i = 0; i <= len; ++i)	362	for (i = 0; i <= len; ++i)
275	{	363	{
276	SV **svp = av_fetch (av, i, 0);	364	SV **svp = av_fetch (av, i, 0);
277	encode_sv (enc, svp ? *svp : &PL_sv_undef);	365	encode_sv (enc, svp ? *svp : &PL_sv_undef);
278	}	366	}
		367	else
		368	for (i = 0; i <= len; ++i)
		369	{
		370	SV *sv = AvARRAY (av)[i];
		371	encode_sv (enc, sv ? sv : &PL_sv_undef);
		372	}
279		373
280	--enc->depth;	374	--enc->depth;
281	}	375	}
282		376
283	static void	377	static void
…		…
292		386
293	int pairs = hv_iterinit (hv);	387	int pairs = hv_iterinit (hv);
294	int mg = SvMAGICAL (hv);	388	int mg = SvMAGICAL (hv);
295		389
296	if (mg)	390	if (mg)
297	encode_ch (enc, 0xa0 \| 31);	391	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
298	else	392	else
299	encode_uint (enc, 0xa0, pairs);	393	encode_uint (enc, MAJOR_MAP, pairs);
300		394
301	while ((he = hv_iternext (hv)))	395	while ((he = hv_iternext (hv)))
302	{	396	{
303	if (HeKLEN (he) == HEf_SVKEY)	397	if (HeKLEN (he) == HEf_SVKEY)
304	encode_sv (enc, HeSVKEY (he));	398	encode_sv (enc, HeSVKEY (he));
305	else	399	else
306	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	400	encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS \| F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307		401
308	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	402	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309	}	403	}
310		404
311	if (mg)	405	if (mg)
312	encode_ch (enc, 0xe0 \| 31);	406	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
313		407
314	--enc->depth;	408	--enc->depth;
315	}	409	}
316		410
317	// encode objects, arrays and special \0=false and \1=true values.	411	// encode objects, arrays and special \0=false and \1=true values.
318	static void	412	static void
319	encode_rv (enc_t enc, SV sv)	413	encode_rv (enc_t enc, SV sv)
320	{	414	{
321	SvGETMAGIC (sv);	415	SvGETMAGIC (sv);
322
323	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
324	&& ecb_expect_false (SvREFCNT (sv) > 1))
325	{
326	if (!enc->shareable)
327	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
328
329	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
330
331	if (SvOK (*svp))
332	{
333	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
334	encode_uint (enc, 0x00, SvUV (*svp));
335	return;
336	}
337	else
338	{
339	sv_setuv (*svp, enc->shareable_idx);
340	++enc->shareable_idx;
341	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
342	}
343	}
344		416
345	svtype svt = SvTYPE (sv);	417	svtype svt = SvTYPE (sv);
346		418
347	if (ecb_expect_false (SvOBJECT (sv)))	419	if (ecb_expect_false (SvOBJECT (sv)))
348	{	420	{
…		…
355	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	427	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
356	? cbor_tagged_stash	428	? cbor_tagged_stash
357	: gv_stashpv ("CBOR::XS::Tagged" , 1);	429	: gv_stashpv ("CBOR::XS::Tagged" , 1);
358		430
359	HV *stash = SvSTASH (sv);	431	HV *stash = SvSTASH (sv);
360	GV *method;
361		432
362	if (stash == boolean_stash)	433	if (stash == boolean_stash)
363	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	434	{
		435	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);
		436	return;
		437	}
364	else if (stash == error_stash)	438	else if (stash == error_stash)
365	encode_ch (enc, 0xe0 \| 23);	439	{
		440	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
		441	return;
		442	}
366	else if (stash == tagged_stash)	443	else if (stash == tagged_stash)
367	{	444	{
368	if (svt != SVt_PVAV)	445	if (svt != SVt_PVAV)
369	croak ("encountered CBOR::XS::Tagged object that isn't an array");	446	croak ("encountered CBOR::XS::Tagged object that isn't an array");
370		447
371	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	448	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
372	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	449	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		450
		451	return;
		452	}
		453	}
		454
		455	if (ecb_expect_false (SvREFCNT (sv) > 1)
		456	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		457	{
		458	if (!enc->shareable)
		459	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		460
		461	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		462
		463	if (SvOK (*svp))
373	}	464	{
		465	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		466	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		467	return;
		468	}
		469	else
		470	{
		471	sv_setuv (*svp, enc->shareable_idx);
		472	++enc->shareable_idx;
		473	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		474	}
		475	}
		476
		477	if (ecb_expect_false (SvOBJECT (sv)))
		478	{
		479	HV *stash = SvSTASH (sv);
		480	GV *method;
		481
374	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	482	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375	{	483	{
376	dSP;	484	dSP;
377		485
378	ENTER; SAVETMPS; PUSHMARK (SP);	486	ENTER; SAVETMPS;
		487	PUSHMARK (SP);
379	// we re-bless the reference to get overload and other niceties right	488	// we re-bless the reference to get overload and other niceties right
380	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	489	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381		490
382	PUTBACK;	491	PUTBACK;
383	// G_SCALAR ensures that return value is 1	492	// G_SCALAR ensures that return value is 1
…		…
396	}	505	}
397	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	506	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398	{	507	{
399	dSP;	508	dSP;
400		509
401	ENTER; SAVETMPS; PUSHMARK (SP);	510	ENTER; SAVETMPS;
		511	SAVESTACK_POS ();
		512	PUSHMARK (SP);
402	EXTEND (SP, 2);	513	EXTEND (SP, 2);
403	// we re-bless the reference to get overload and other niceties right	514	// we re-bless the reference to get overload and other niceties right
404	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	515	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405	PUSHs (sv_cbor);	516	PUSHs (sv_cbor);
406		517
…		…
411	// catch this surprisingly common error	522	// catch this surprisingly common error
412	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	523	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
413	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	524	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414		525
415	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	526	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416	encode_uint (enc, 0x80, count + 1);	527	encode_uint (enc, MAJOR_ARRAY, count + 1);
417	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	528	encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418		529
419	while (count)	530	while (count)
420	encode_sv (enc, SP[1 - count--]);	531	encode_sv (enc, SP[1 - count--]);
421		532
422	PUTBACK;	533	PUTBACK;
…		…
443	{	554	{
444	double nv = SvNVX (sv);	555	double nv = SvNVX (sv);
445		556
446	need (enc, 9);	557	need (enc, 9);
447		558
448	if (ecb_expect_false (nv == (U32)nv))	559	if (ecb_expect_false (nv == (NV)(U32)nv))
449	encode_uint (enc, 0x00, (U32)nv);	560	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450	//TODO: maybe I32?	561	//TODO: maybe I32?
451	else if (ecb_expect_false (nv == (float)nv))	562	else if (ecb_expect_false (nv == (float)nv))
452	{	563	{
453	uint32_t fp = ecb_float_to_binary32 (nv);	564	uint32_t fp = ecb_float_to_binary32 (nv);
454		565
455	*enc->cur++ = 0xe0 \| 26;	566	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
456		567
457	if (!ecb_big_endian ())	568	if (!ecb_big_endian ())
458	fp = ecb_bswap32 (fp);	569	fp = ecb_bswap32 (fp);
459		570
460	memcpy (enc->cur, &fp, 4);	571	memcpy (enc->cur, &fp, 4);
…		…
462	}	573	}
463	else	574	else
464	{	575	{
465	uint64_t fp = ecb_double_to_binary64 (nv);	576	uint64_t fp = ecb_double_to_binary64 (nv);
466		577
467	*enc->cur++ = 0xe0 \| 27;	578	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
468		579
469	if (!ecb_big_endian ())	580	if (!ecb_big_endian ())
470	fp = ecb_bswap64 (fp);	581	fp = ecb_bswap64 (fp);
471		582
472	memcpy (enc->cur, &fp, 8);	583	memcpy (enc->cur, &fp, 8);
…		…
481		592
482	if (SvPOKp (sv))	593	if (SvPOKp (sv))
483	{	594	{
484	STRLEN len;	595	STRLEN len;
485	char *str = SvPV (sv, len);	596	char *str = SvPV (sv, len);
486	encode_str (enc, SvUTF8 (sv), str, len);	597	encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
487	}	598	}
488	else if (SvNOKp (sv))	599	else if (SvNOKp (sv))
489	encode_nv (enc, sv);	600	encode_nv (enc, sv);
490	else if (SvIOKp (sv))	601	else if (SvIOKp (sv))
491	{	602	{
492	if (SvIsUV (sv))	603	if (SvIsUV (sv))
493	encode_uint (enc, 0x00, SvUVX (sv));	604	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494	else if (SvIVX (sv) >= 0)	605	else if (SvIVX (sv) >= 0)
495	encode_uint (enc, 0x00, SvIVX (sv));	606	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496	else	607	else
497	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	608	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498	}	609	}
499	else if (SvROK (sv))	610	else if (SvROK (sv))
500	encode_rv (enc, SvRV (sv));	611	encode_rv (enc, SvRV (sv));
501	else if (!SvOK (sv))	612	else if (!SvOK (sv))
502	encode_ch (enc, 0xe0 \| 22);	613	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
503	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	614	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504	encode_ch (enc, 0xe0 \| 23);	615	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
505	else	616	else
506	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	617	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	618	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508	}	619	}
509		620
510	static SV *	621	static SV *
511	encode_cbor (SV scalar, CBOR cbor)	622	encode_cbor (SV scalar, CBOR cbor)
512	{	623	{
513	enc_t enc = { };	624	enc_t enc = { 0 };
514		625
515	enc.cbor = *cbor;	626	enc.cbor = *cbor;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	627	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	628	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	629	enc.end = SvEND (enc.sv);
519		630
520	SvPOK_only (enc.sv);	631	SvPOK_only (enc.sv);
521		632
522	if (cbor->flags & F_ALLOW_STRINGREF)	633	if (cbor->flags & F_PACK_STRINGS)
523	{	634	{
524	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	635	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());	636	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
526	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());	637	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
527	}	638	}
…		…
550	U32 depth; // recursion depth	661	U32 depth; // recursion depth
551	U32 maxdepth; // recursion depth limit	662	U32 maxdepth; // recursion depth limit
552	AV *shareable;	663	AV *shareable;
553	AV *stringref;	664	AV *stringref;
554	SV *decode_tagged;	665	SV *decode_tagged;
		666	SV *err_sv; // optional sv for error, needs to be freed
555	} dec_t;	667	} dec_t;
556		668
557	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	669	// set dec->err to ERRSV
		670	ecb_cold static void
		671	err_errsv (dec_t *dec)
		672	{
		673	if (!dec->err)
		674	{
		675	dec->err_sv = newSVsv (ERRSV);
558		676
559	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	677	// chop off the trailing \n
		678	SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
		679	*SvEND (dec->err_sv) = 0;
560		680
		681	dec->err = SvPVutf8_nolen (dec->err_sv);
		682	}
		683	}
		684
		685	// the following functions are used to reduce code size and help the compiler to optimise
		686	ecb_cold static void
		687	err_set (dec_t dec, const char reason)
		688	{
		689	if (!dec->err)
		690	dec->err = reason;
		691	}
		692
		693	ecb_cold static void
		694	err_unexpected_end (dec_t *dec)
		695	{
		696	err_set (dec, "unexpected end of CBOR data");
		697	}
		698
		699	ecb_cold static void
		700	err_nesting_exceeded (dec_t *dec)
		701	{
		702	err_set (dec, ERR_NESTING_EXCEEDED);
		703	}
		704
		705	#define ERR_DO(do) SB do; goto fail; SE
		706	#define ERR(reason) ERR_DO (err_set (dec, reason))
		707	#define ERR_ERRSV ERR_DO (err_errsv (dec))
		708
		709	#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
		710
561	#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)	711	#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
562	#define DEC_DEC_DEPTH --dec->depth	712	#define DEC_DEC_DEPTH --dec->depth
563		713
564	static UV	714	static UV
565	decode_uint (dec_t *dec)	715	decode_uint (dec_t *dec)
566	{	716	{
567	switch (*dec->cur & 31)	717	U8 m = *dec->cur & MINOR_MASK;
568	{	718	++dec->cur;
569	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
570	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
571	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
572	return *dec->cur++ & 31;
573		719
574	case 24:	720	if (ecb_expect_true (m < LENGTH_EXT1))
		721	return m;
		722	else if (ecb_expect_true (m == LENGTH_EXT1))
		723	{
575	WANT (2);	724	WANT (1);
576	dec->cur += 2;	725	dec->cur += 1;
577	return dec->cur[-1];	726	return dec->cur[-1];
578		727	}
579	case 25:	728	else if (ecb_expect_true (m == LENGTH_EXT2))
		729	{
580	WANT (3);	730	WANT (2);
581	dec->cur += 3;	731	dec->cur += 2;
582	return (((UV)dec->cur[-2]) << 8)	732	return (((UV)dec->cur[-2]) << 8)
583	\| ((UV)dec->cur[-1]);	733	\| ((UV)dec->cur[-1]);
584		734	}
585	case 26:	735	else if (ecb_expect_true (m == LENGTH_EXT4))
		736	{
586	WANT (5);	737	WANT (4);
587	dec->cur += 5;	738	dec->cur += 4;
588	return (((UV)dec->cur[-4]) << 24)	739	return (((UV)dec->cur[-4]) << 24)
589	\| (((UV)dec->cur[-3]) << 16)	740	\| (((UV)dec->cur[-3]) << 16)
590	\| (((UV)dec->cur[-2]) << 8)	741	\| (((UV)dec->cur[-2]) << 8)
591	\| ((UV)dec->cur[-1]);	742	\| ((UV)dec->cur[-1]);
592		743	}
593	case 27:	744	else if (ecb_expect_true (m == LENGTH_EXT8))
		745	{
594	WANT (9);	746	WANT (8);
595	dec->cur += 9;	747	dec->cur += 8;
		748
		749	return
		750	#if UVSIZE < 8
		751	0
		752	#else
596	return (((UV)dec->cur[-8]) << 56)	753	(((UV)dec->cur[-8]) << 56)
597	\| (((UV)dec->cur[-7]) << 48)	754	\| (((UV)dec->cur[-7]) << 48)
598	\| (((UV)dec->cur[-6]) << 40)	755	\| (((UV)dec->cur[-6]) << 40)
599	\| (((UV)dec->cur[-5]) << 32)	756	\| (((UV)dec->cur[-5]) << 32)
		757	#endif
600	\| (((UV)dec->cur[-4]) << 24)	758	\| (((UV)dec->cur[-4]) << 24)
601	\| (((UV)dec->cur[-3]) << 16)	759	\| (((UV)dec->cur[-3]) << 16)
602	\| (((UV)dec->cur[-2]) << 8)	760	\| (((UV)dec->cur[-2]) << 8)
603	\| ((UV)dec->cur[-1]);	761	\| ((UV)dec->cur[-1]);
604		762	}
605	default:	763	else
606	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	764	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607	}
608		765
609	fail:	766	fail:
610	return 0;	767	return 0;
611	}	768	}
612		769
…		…
617	{	774	{
618	AV *av = newAV ();	775	AV *av = newAV ();
619		776
620	DEC_INC_DEPTH;	777	DEC_INC_DEPTH;
621		778
622	if ((*dec->cur & 31) == 31)	779	if (*dec->cur == (MAJOR_ARRAY \| MINOR_INDEF))
623	{	780	{
624	++dec->cur;	781	++dec->cur;
625		782
626	for (;;)	783	for (;;)
627	{	784	{
628	WANT (1);	785	WANT (1);
629		786
630	if (*dec->cur == (0xe0 \| 31))	787	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
631	{	788	{
632	++dec->cur;	789	++dec->cur;
633	break;	790	break;
634	}	791	}
635		792
636	av_push (av, decode_sv (dec));	793	av_push (av, decode_sv (dec));
637	}	794	}
638	}	795	}
639	else	796	else
640	{	797	{
641	int i, len = decode_uint (dec);	798	UV i, len = decode_uint (dec);
642		799
		800	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
643	av_fill (av, len - 1);	801	av_fill (av, len - 1);
644		802
645	for (i = 0; i < len; ++i)	803	for (i = 0; i < len; ++i)
646	AvARRAY (av)[i] = decode_sv (dec);	804	AvARRAY (av)[i] = decode_sv (dec);
647	}	805	}
…		…
660	{	818	{
661	// for speed reasons, we specialcase single-string	819	// for speed reasons, we specialcase single-string
662	// byte or utf-8 strings as keys, but only when !stringref	820	// byte or utf-8 strings as keys, but only when !stringref
663		821
664	if (ecb_expect_true (!dec->stringref))	822	if (ecb_expect_true (!dec->stringref))
665	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	823	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666	{	824	{
667	I32 len = decode_uint (dec);	825	STRLEN len = decode_uint (dec);
668	char key = (char )dec->cur;	826	char key = (char )dec->cur;
669		827
		828	WANT (len);
670	dec->cur += len;	829	dec->cur += len;
671		830
672	if (ecb_expect_false (dec->stringref))
673	av_push (dec->stringref, newSVpvn (key, len));
674
675	hv_store (hv, key, len, decode_sv (dec), 0);	831	hv_store (hv, key, len, decode_sv (dec), 0);
676		832
677	return;	833	return;
678	}	834	}
679	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	835	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680	{	836	{
681	I32 len = decode_uint (dec);	837	STRLEN len = decode_uint (dec);
682	char key = (char )dec->cur;	838	char key = (char )dec->cur;
683		839
		840	WANT (len);
684	dec->cur += len;	841	dec->cur += len;
685		842
686	if (ecb_expect_false (dec->stringref))	843	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	844	if (!is_utf8_string (key, len))
		845	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688		846
689	hv_store (hv, key, -len, decode_sv (dec), 0);	847	hv_store (hv, key, -len, decode_sv (dec), 0);
690		848
691	return;	849	return;
692	}	850	}
693		851
694	SV *k = decode_sv (dec);	852	SV *k = decode_sv (dec);
695	SV *v = decode_sv (dec);	853	SV *v = decode_sv (dec);
696		854
		855	// we leak memory if uncaught exceptions are thrown by random magical
		856	// methods, and this is hopefully the only place where it can happen,
		857	// so if there is a chance of an exception, take the very slow path.
		858	// since catching exceptions is "undocumented/internal/forbidden" by
		859	// the new p5p powers, we need to call out to a perl function :/
		860	if (ecb_expect_false (SvAMAGIC (k)))
		861	{
		862	dSP;
		863
		864	ENTER; SAVETMPS;
		865	PUSHMARK (SP);
		866	EXTEND (SP, 3);
		867	PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
		868	PUSHs (sv_2mortal (k));
		869	PUSHs (sv_2mortal (v));
		870
		871	PUTBACK;
		872	call_pv ("CBOR::XS::_hv_store", G_VOID \| G_DISCARD \| G_EVAL);
		873	SPAGAIN;
		874
		875	FREETMPS; LEAVE;
		876
		877	if (SvTRUE (ERRSV))
		878	ERR_ERRSV;
		879
		880	return;
		881	}
		882
697	hv_store_ent (hv, k, v, 0);	883	hv_store_ent (hv, k, v, 0);
698	SvREFCNT_dec (k);	884	SvREFCNT_dec (k);
		885
		886	fail:
		887	;
699	}	888	}
700		889
701	static SV *	890	static SV *
702	decode_hv (dec_t *dec)	891	decode_hv (dec_t *dec)
703	{	892	{
704	HV *hv = newHV ();	893	HV *hv = newHV ();
705		894
706	DEC_INC_DEPTH;	895	DEC_INC_DEPTH;
707		896
708	if ((*dec->cur & 31) == 31)	897	if (*dec->cur == (MAJOR_MAP \| MINOR_INDEF))
709	{	898	{
710	++dec->cur;	899	++dec->cur;
711		900
712	for (;;)	901	for (;;)
713	{	902	{
714	WANT (1);	903	WANT (1);
715		904
716	if (*dec->cur == (0xe0 \| 31))	905	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
717	{	906	{
718	++dec->cur;	907	++dec->cur;
719	break;	908	break;
720	}	909	}
721		910
722	decode_he (dec, hv);	911	decode_he (dec, hv);
723	}	912	}
724	}	913	}
725	else	914	else
726	{	915	{
727	int pairs = decode_uint (dec);	916	UV pairs = decode_uint (dec);
		917
		918	WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
728		919
729	while (pairs--)	920	while (pairs--)
730	decode_he (dec, hv);	921	decode_he (dec, hv);
731	}	922	}
732		923
…		…
742	static SV *	933	static SV *
743	decode_str (dec_t *dec, int utf8)	934	decode_str (dec_t *dec, int utf8)
744	{	935	{
745	SV *sv = 0;	936	SV *sv = 0;
746		937
747	if ((*dec->cur & 31) == 31)	938	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748	{	939	{
		940	// indefinite length strings
749	++dec->cur;	941	++dec->cur;
750		942
		943	U8 major = *dec->cur & MAJOR_MISC;
		944
751	sv = newSVpvn ("", 0);	945	sv = newSVpvn ("", 0);
752		946
753	// not very fast, and certainly not robust against illegal input
754	for (;;)	947	for (;;)
755	{	948	{
756	WANT (1);	949	WANT (1);
757		950
758	if (*dec->cur == (0xe0 \| 31))	951	if ((*dec->cur - major) > LENGTH_EXT8)
		952	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
759	{	953	{
760	++dec->cur;	954	++dec->cur;
761	break;	955	break;
762	}	956	}
		957	else
		958	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763		959
764	sv_catsv (sv, decode_sv (dec));	960	STRLEN len = decode_uint (dec);
		961
		962	WANT (len);
		963	sv_catpvn (sv, dec->cur, len);
		964	dec->cur += len;
765	}	965	}
766	}	966	}
767	else	967	else
768	{	968	{
769	STRLEN len = decode_uint (dec);	969	STRLEN len = decode_uint (dec);
…		…
776	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	976	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	977	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778	}	978	}
779		979
780	if (utf8)	980	if (utf8)
		981	{
		982	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		983	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		984	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		985
781	SvUTF8_on (sv);	986	SvUTF8_on (sv);
		987	}
782		988
783	return sv;	989	return sv;
784		990
785	fail:	991	fail:
786	SvREFCNT_dec (sv);	992	SvREFCNT_dec (sv);
…		…
805	sv = newRV_noinc (decode_sv (dec));	1011	sv = newRV_noinc (decode_sv (dec));
806	break;	1012	break;
807		1013
808	case CBOR_TAG_STRINGREF_NAMESPACE:	1014	case CBOR_TAG_STRINGREF_NAMESPACE:
809	{	1015	{
		1016	// do nmot use SAVETMPS/FREETMPS, as these will
		1017	// erase mortalised caches, e.g. "shareable"
810	ENTER; SAVETMPS;	1018	ENTER;
811		1019
812	SAVESPTR (dec->stringref);	1020	SAVESPTR (dec->stringref);
813	dec->stringref = (AV )sv_2mortal ((SV )newAV ());	1021	dec->stringref = (AV )sv_2mortal ((SV )newAV ());
814		1022
815	sv = decode_sv (dec);	1023	sv = decode_sv (dec);
816		1024
817	FREETMPS; LEAVE;	1025	LEAVE;
818	}	1026	}
819	break;	1027	break;
820		1028
821	case CBOR_TAG_STRINGREF:	1029	case CBOR_TAG_STRINGREF:
822	{	1030	{
823	if ((*dec->cur >> 5) != 0)	1031	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	1032	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825		1033
826	UV idx = decode_uint (dec);	1034	UV idx = decode_uint (dec);
827		1035
828	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	1036	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
…		…
835	case CBOR_TAG_VALUE_SHAREABLE:	1043	case CBOR_TAG_VALUE_SHAREABLE:
836	{	1044	{
837	if (ecb_expect_false (!dec->shareable))	1045	if (ecb_expect_false (!dec->shareable))
838	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	1046	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
839		1047
		1048	if (dec->cbor.flags & F_ALLOW_CYCLES)
		1049	{
840	sv = newSV (0);	1050	sv = newSV (0);
841	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	1051	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842		1052
843	SV *osv = decode_sv (dec);	1053	SV *osv = decode_sv (dec);
844	sv_setsv (sv, osv);	1054	sv_setsv (sv, osv);
845	SvREFCNT_dec_NN (osv);	1055	SvREFCNT_dec_NN (osv);
		1056	}
		1057	else
		1058	{
		1059	av_push (dec->shareable, &PL_sv_undef);
		1060	int idx = AvFILLp (dec->shareable);
		1061	sv = decode_sv (dec);
		1062	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		1063	}
846	}	1064	}
847	break;	1065	break;
848		1066
849	case CBOR_TAG_VALUE_SHAREDREF:	1067	case CBOR_TAG_VALUE_SHAREDREF:
850	{	1068	{
851	if ((*dec->cur >> 5) != 0)	1069	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	1070	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853		1071
854	UV idx = decode_uint (dec);	1072	UV idx = decode_uint (dec);
855		1073
856	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	1074	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
857	ERR ("corrupted CBOR data (sharedref index out of bounds)");	1075	ERR ("corrupted CBOR data (sharedref index out of bounds)");
858		1076
859	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	1077	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		1078
		1079	if (sv == &PL_sv_undef)
		1080	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860	}	1081	}
861	break;	1082	break;
862		1083
863	case CBOR_TAG_PERL_OBJECT:	1084	case CBOR_TAG_PERL_OBJECT:
864	{	1085	{
…		…
879	if (!method)	1100	if (!method)
880	ERR ("cannot decode perl-object (package does not have a THAW method)");	1101	ERR ("cannot decode perl-object (package does not have a THAW method)");
881		1102
882	dSP;	1103	dSP;
883		1104
884	ENTER; SAVETMPS; PUSHMARK (SP);	1105	ENTER; SAVETMPS;
		1106	PUSHMARK (SP);
885	EXTEND (SP, len + 1);	1107	EXTEND (SP, len + 1);
886	// we re-bless the reference to get overload and other niceties right	1108	// we re-bless the reference to get overload and other niceties right
887	PUSHs (*av_fetch (av, 0, 1));	1109	PUSHs (*av_fetch (av, 0, 1));
888	PUSHs (sv_cbor);	1110	PUSHs (sv_cbor);
889		1111
…		…
897	SPAGAIN;	1119	SPAGAIN;
898		1120
899	if (SvTRUE (ERRSV))	1121	if (SvTRUE (ERRSV))
900	{	1122	{
901	FREETMPS; LEAVE;	1123	FREETMPS; LEAVE;
902	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1124	ERR_ERRSV;
903	}	1125	}
904		1126
905	SvREFCNT_dec (sv);	1127	SvREFCNT_dec (sv);
906	sv = SvREFCNT_inc (POPs);	1128	sv = SvREFCNT_inc (POPs);
907		1129
…		…
911	}	1133	}
912	break;	1134	break;
913		1135
914	default:	1136	default:
915	{	1137	{
		1138	SV *tag_sv = newSVuv (tag);
		1139
916	sv = decode_sv (dec);	1140	sv = decode_sv (dec);
917		1141
918	dSP;	1142	dSP;
919	ENTER; SAVETMPS; PUSHMARK (SP);	1143	ENTER; SAVETMPS;
		1144	SAVESTACK_POS ();
		1145	PUSHMARK (SP);
920	EXTEND (SP, 2);	1146	EXTEND (SP, 2);
921	PUSHs (newSVuv (tag));	1147	PUSHs (tag_sv);
922	PUSHs (sv);	1148	PUSHs (sv);
923		1149
924	PUTBACK;	1150	PUTBACK;
925	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);	1151	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
926	SPAGAIN;	1152	SPAGAIN;
927		1153
928	if (SvTRUE (ERRSV))	1154	if (SvTRUE (ERRSV))
929	{	1155	{
		1156	SvREFCNT_dec (tag_sv);
930	FREETMPS; LEAVE;	1157	FREETMPS; LEAVE;
931	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));	1158	ERR_ERRSV;
932	}	1159	}
933		1160
934	if (count)	1161	if (count)
935	{	1162	{
		1163	SvREFCNT_dec (tag_sv);
936	SvREFCNT_dec (sv);	1164	SvREFCNT_dec (sv);
937	sv = SvREFCNT_inc (POPs);	1165	sv = SvREFCNT_inc (POPs);
938	}	1166	}
939	else	1167	else
940	{	1168	{
941	AV *av = newAV ();	1169	AV *av = newAV ();
942	av_push (av, newSVuv (tag));	1170	av_push (av, tag_sv);
943	av_push (av, sv);	1171	av_push (av, sv);
944		1172
945	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1173	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
946	? cbor_tagged_stash	1174	? cbor_tagged_stash
947	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1175	: gv_stashpv ("CBOR::XS::Tagged" , 1);
…		…
965	static SV *	1193	static SV *
966	decode_sv (dec_t *dec)	1194	decode_sv (dec_t *dec)
967	{	1195	{
968	WANT (1);	1196	WANT (1);
969		1197
970	switch (*dec->cur >> 5)	1198	switch (*dec->cur >> MAJOR_SHIFT)
971	{	1199	{
972	case 0: // unsigned int	1200	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973	return newSVuv (decode_uint (dec));	1201	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974	case 1: // negative int	1202	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975	return newSViv (-1 - (IV)decode_uint (dec));	1203	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976	case 2: // octet string	1204	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977	return decode_str (dec, 0);	1205	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978	case 3: // utf-8 string	1206	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979	return decode_str (dec, 1);	1207
980	case 4: // array	1208	case MAJOR_MISC >> MAJOR_SHIFT:
981	return decode_av (dec);
982	case 5: // map
983	return decode_hv (dec);
984	case 6: // tag
985	return decode_tagged (dec);
986	case 7: // misc
987	switch (*dec->cur++ & 31)	1209	switch (*dec->cur++ & MINOR_MASK)
988	{	1210	{
989	case 20:	1211	case SIMPLE_FALSE:
990	#if CBOR_SLOW	1212	#if CBOR_SLOW
991	types_false = get_bool ("Types::Serialiser::false");	1213	types_false = get_bool ("Types::Serialiser::false");
992	#endif	1214	#endif
993	return newSVsv (types_false);	1215	return newSVsv (types_false);
994	case 21:	1216	case SIMPLE_TRUE:
995	#if CBOR_SLOW	1217	#if CBOR_SLOW
996	types_true = get_bool ("Types::Serialiser::true");	1218	types_true = get_bool ("Types::Serialiser::true");
997	#endif	1219	#endif
998	return newSVsv (types_true);	1220	return newSVsv (types_true);
999	case 22:	1221	case SIMPLE_NULL:
1000	return newSVsv (&PL_sv_undef);	1222	return newSVsv (&PL_sv_undef);
1001	case 23:	1223	case SIMPLE_UNDEF:
1002	#if CBOR_SLOW	1224	#if CBOR_SLOW
1003	types_error = get_bool ("Types::Serialiser::error");	1225	types_error = get_bool ("Types::Serialiser::error");
1004	#endif	1226	#endif
1005	return newSVsv (types_error);	1227	return newSVsv (types_error);
1006		1228
1007	case 25:	1229	case MISC_FLOAT16:
1008	{	1230	{
1009	WANT (2);	1231	WANT (2);
1010		1232
1011	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];	1233	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];
1012	dec->cur += 2;	1234	dec->cur += 2;
1013		1235
1014	return newSVnv (ecb_binary16_to_float (fp));	1236	return newSVnv (ecb_binary16_to_float (fp));
1015	}	1237	}
1016		1238
1017	case 26:	1239	case MISC_FLOAT32:
1018	{	1240	{
1019	uint32_t fp;	1241	uint32_t fp;
1020	WANT (4);	1242	WANT (4);
1021	memcpy (&fp, dec->cur, 4);	1243	memcpy (&fp, dec->cur, 4);
1022	dec->cur += 4;	1244	dec->cur += 4;
…		…
1025	fp = ecb_bswap32 (fp);	1247	fp = ecb_bswap32 (fp);
1026		1248
1027	return newSVnv (ecb_binary32_to_float (fp));	1249	return newSVnv (ecb_binary32_to_float (fp));
1028	}	1250	}
1029		1251
1030	case 27:	1252	case MISC_FLOAT64:
1031	{	1253	{
1032	uint64_t fp;	1254	uint64_t fp;
1033	WANT (8);	1255	WANT (8);
1034	memcpy (&fp, dec->cur, 8);	1256	memcpy (&fp, dec->cur, 8);
1035	dec->cur += 8;	1257	dec->cur += 8;
…		…
1038	fp = ecb_bswap64 (fp);	1260	fp = ecb_bswap64 (fp);
1039		1261
1040	return newSVnv (ecb_binary64_to_double (fp));	1262	return newSVnv (ecb_binary64_to_double (fp));
1041	}	1263	}
1042		1264
1043	// 0..19 unassigned	1265	// 0..19 unassigned simple
1044	// 24 reserved + unassigned (reserved values are not encodable)	1266	// 24 reserved + unassigned simple (reserved values are not encodable)
		1267	// 28-30 unassigned misc
		1268	// 31 break code
1045	default:	1269	default:
1046	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1270	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047	}	1271	}
1048		1272
1049	break;	1273	break;
1050	}	1274	}
1051		1275
…		…
1054	}	1278	}
1055		1279
1056	static SV *	1280	static SV *
1057	decode_cbor (SV string, CBOR cbor, char **offset_return)	1281	decode_cbor (SV string, CBOR cbor, char **offset_return)
1058	{	1282	{
1059	dec_t dec = { };	1283	dec_t dec = { 0 };
1060	SV *sv;	1284	SV *sv;
1061	STRLEN len;	1285	STRLEN len;
1062	char *data = SvPVbyte (string, len);	1286	char *data = SvPVbyte (string, len);
1063		1287
1064	if (len > cbor->max_size && cbor->max_size)	1288	if (len > cbor->max_size && cbor->max_size)
…		…
1078	if (dec.cur != dec.end && !dec.err)	1302	if (dec.cur != dec.end && !dec.err)
1079	dec.err = "garbage after CBOR object";	1303	dec.err = "garbage after CBOR object";
1080		1304
1081	if (dec.err)	1305	if (dec.err)
1082	{	1306	{
		1307	if (dec.shareable)
		1308	{
		1309	// need to break cyclic links, which would all be in shareable
		1310	int i;
		1311	SV **svp;
		1312
		1313	for (i = av_len (dec.shareable) + 1; i--; )
		1314	if ((svp = av_fetch (dec.shareable, i, 0)))
		1315	sv_setsv (*svp, &PL_sv_undef);
		1316	}
		1317
1083	SvREFCNT_dec (sv);	1318	SvREFCNT_dec (sv);
		1319
		1320	if (dec.err_sv)
		1321	sv_2mortal (dec.err_sv);
		1322
1084	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1323	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1085	}	1324	}
1086		1325
1087	sv = sv_2mortal (sv);	1326	sv = sv_2mortal (sv);
1088		1327
1089	return sv;	1328	return sv;
1090	}	1329	}
1091		1330
		1331	/////////////////////////////////////////////////////////////////////////////
		1332	// incremental parser
		1333
		1334	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1335
		1336	// returns 0 for notyet, 1 for success or error
		1337	static int
		1338	incr_parse (CBOR self, SV cborstr)
		1339	{
		1340	STRLEN cur;
		1341	SvPV (cborstr, cur);
		1342
		1343	while (ecb_expect_true (self->incr_need <= cur))
		1344	{
		1345	// table of integer count bytes
		1346	static I8 incr_len[MINOR_MASK + 1] = {
		1347	0, 0, 0, 0, 0, 0, 0, 0,
		1348	0, 0, 0, 0, 0, 0, 0, 0,
		1349	0, 0, 0, 0, 0, 0, 0, 0,
		1350	1, 2, 4, 8,-1,-1,-1,-2
		1351	};
		1352
		1353	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1354	U8 m = *p & MINOR_MASK;
		1355	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1356	I8 ilen = incr_len[m];
		1357
		1358	self->incr_need = self->incr_pos + 1;
		1359
		1360	if (ecb_expect_false (ilen < 0))
		1361	{
		1362	if (m != MINOR_INDEF)
		1363	return 1; // error
		1364
		1365	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1366	{
		1367	if (count >= 0)
		1368	return 1; // error
		1369
		1370	count = 1;
		1371	}
		1372	else
		1373	{
		1374	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1375	count = -1;
		1376	}
		1377	}
		1378	else
		1379	{
		1380	self->incr_need += ilen;
		1381	if (ecb_expect_false (self->incr_need > cur))
		1382	return 0;
		1383
		1384	int major = *p >> MAJOR_SHIFT;
		1385
		1386	switch (major)
		1387	{
		1388	case MAJOR_TAG >> MAJOR_SHIFT:
		1389	++count; // tags merely prefix another value
		1390	break;
		1391
		1392	case MAJOR_BYTES >> MAJOR_SHIFT:
		1393	case MAJOR_TEXT >> MAJOR_SHIFT:
		1394	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1395	case MAJOR_MAP >> MAJOR_SHIFT:
		1396	{
		1397	UV len;
		1398
		1399	if (ecb_expect_false (ilen))
		1400	{
		1401	len = 0;
		1402
		1403	do {
		1404	len = (len << 8) \| *++p;
		1405	} while (--ilen);
		1406	}
		1407	else
		1408	len = m;
		1409
		1410	switch (major)
		1411	{
		1412	case MAJOR_BYTES >> MAJOR_SHIFT:
		1413	case MAJOR_TEXT >> MAJOR_SHIFT:
		1414	self->incr_need += len;
		1415	if (ecb_expect_false (self->incr_need > cur))
		1416	return 0;
		1417
		1418	break;
		1419
		1420	case MAJOR_MAP >> MAJOR_SHIFT:
		1421	len <<= 1;
		1422	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1423	if (len)
		1424	{
		1425	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1426	count = len + 1;
		1427	}
		1428	break;
		1429	}
		1430	}
		1431	}
		1432	}
		1433
		1434	self->incr_pos = self->incr_need;
		1435
		1436	if (count > 0)
		1437	{
		1438	while (!--count)
		1439	{
		1440	if (!AvFILLp (self->incr_count))
		1441	return 1; // done
		1442
		1443	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1444	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1445	}
		1446
		1447	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1448	}
		1449	}
		1450
		1451	return 0;
		1452	}
		1453
		1454
1092	/////////////////////////////////////////////////////////////////////////////	1455	/////////////////////////////////////////////////////////////////////////////
1093	// XS interface functions	1456	// XS interface functions
1094		1457
1095	MODULE = CBOR::XS PACKAGE = CBOR::XS	1458	MODULE = CBOR::XS PACKAGE = CBOR::XS
1096		1459
…		…
1108		1471
1109	default_filter = newSVpv ("CBOR::XS::default_filter", 0);	1472	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1110		1473
1111	sv_cbor = newSVpv ("CBOR", 0);	1474	sv_cbor = newSVpv ("CBOR", 0);
1112	SvREADONLY_on (sv_cbor);	1475	SvREADONLY_on (sv_cbor);
		1476
		1477	assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1113	}	1478	}
1114		1479
1115	PROTOTYPES: DISABLE	1480	PROTOTYPES: DISABLE
1116		1481
1117	void CLONE (...)	1482	void CLONE (...)
…		…
1136	void shrink (CBOR *self, int enable = 1)	1501	void shrink (CBOR *self, int enable = 1)
1137	ALIAS:	1502	ALIAS:
1138	shrink = F_SHRINK	1503	shrink = F_SHRINK
1139	allow_unknown = F_ALLOW_UNKNOWN	1504	allow_unknown = F_ALLOW_UNKNOWN
1140	allow_sharing = F_ALLOW_SHARING	1505	allow_sharing = F_ALLOW_SHARING
1141	allow_stringref = F_ALLOW_STRINGREF	1506	allow_cycles = F_ALLOW_CYCLES
		1507	pack_strings = F_PACK_STRINGS
		1508	text_keys = F_TEXT_KEYS
		1509	text_strings = F_TEXT_STRINGS
		1510	validate_utf8 = F_VALIDATE_UTF8
1142	PPCODE:	1511	PPCODE:
1143	{	1512	{
1144	if (enable)	1513	if (enable)
1145	self->flags \|= ix;	1514	self->flags \|= ix;
1146	else	1515	else
…		…
1152	void get_shrink (CBOR *self)	1521	void get_shrink (CBOR *self)
1153	ALIAS:	1522	ALIAS:
1154	get_shrink = F_SHRINK	1523	get_shrink = F_SHRINK
1155	get_allow_unknown = F_ALLOW_UNKNOWN	1524	get_allow_unknown = F_ALLOW_UNKNOWN
1156	get_allow_sharing = F_ALLOW_SHARING	1525	get_allow_sharing = F_ALLOW_SHARING
1157	get_allow_stringref = F_ALLOW_STRINGREF	1526	get_allow_cycles = F_ALLOW_CYCLES
		1527	get_pack_strings = F_PACK_STRINGS
		1528	get_text_keys = F_TEXT_KEYS
		1529	get_text_strings = F_TEXT_STRINGS
		1530	get_validate_utf8 = F_VALIDATE_UTF8
1158	PPCODE:	1531	PPCODE:
1159	XPUSHs (boolSV (self->flags & ix));	1532	XPUSHs (boolSV (self->flags & ix));
1160		1533
1161	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1534	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162	PPCODE:	1535	PPCODE:
…		…
1211	EXTEND (SP, 2);	1584	EXTEND (SP, 2);
1212	PUSHs (sv);	1585	PUSHs (sv);
1213	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1586	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214	}	1587	}
1215		1588
		1589	void incr_parse (CBOR self, SV cborstr)
		1590	ALIAS:
		1591	incr_parse_multiple = 1
		1592	PPCODE:
		1593	{
		1594	if (SvUTF8 (cborstr))
		1595	sv_utf8_downgrade (cborstr, 0);
		1596
		1597	if (!self->incr_count)
		1598	{
		1599	self->incr_count = newAV ();
		1600	self->incr_pos = 0;
		1601	self->incr_need = 1;
		1602
		1603	av_push (self->incr_count, newSViv (1));
		1604	}
		1605
		1606	do
		1607	{
		1608	if (!incr_parse (self, cborstr))
		1609	{
		1610	if (self->incr_need > self->max_size && self->max_size)
		1611	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1612	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1613
		1614	break;
		1615	}
		1616
		1617	SV *sv;
		1618	char *offset;
		1619
		1620	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1621	XPUSHs (sv);
		1622
		1623	sv_chop (cborstr, offset);
		1624
		1625	av_clear (self->incr_count);
		1626	av_push (self->incr_count, newSViv (1));
		1627
		1628	self->incr_pos = 0;
		1629	self->incr_need = self->incr_pos + 1;
		1630	}
		1631	while (ix);
		1632	}
		1633
		1634	void incr_reset (CBOR *self)
		1635	CODE:
		1636	{
		1637	SvREFCNT_dec (self->incr_count);
		1638	self->incr_count = 0;
		1639	}
		1640
1216	void DESTROY (CBOR *self)	1641	void DESTROY (CBOR *self)
1217	PPCODE:	1642	PPCODE:
1218	cbor_free (self);	1643	cbor_free (self);
1219		1644
1220	PROTOTYPES: ENABLE	1645	PROTOTYPES: ENABLE
1221		1646
1222	void encode_cbor (SV *scalar)	1647	void encode_cbor (SV *scalar)
		1648	ALIAS:
		1649	encode_cbor = 0
		1650	encode_cbor_sharing = F_ALLOW_SHARING
1223	PPCODE:	1651	PPCODE:
1224	{	1652	{
1225	CBOR cbor;	1653	CBOR cbor;
1226	cbor_init (&cbor);	1654	cbor_init (&cbor);
		1655	cbor.flags \|= ix;
1227	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1656	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228	XPUSHs (scalar);	1657	XPUSHs (scalar);
1229	}	1658	}
1230		1659
1231	void decode_cbor (SV *cborstr)	1660	void decode_cbor (SV *cborstr)

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs. Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.27 by root, Fri Nov 22 15:28:38 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 2016 UTC