[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 1.51 by root, Sun Apr 24 13:15:19 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_UTF8_STRINGS 0x00000020UL
		106	#define F_VALIDATE_UTF8 0x00000040UL
63		107
64	#define INIT_SIZE 32 // initial scalar size to be allocated	108	#define INIT_SIZE 32 // initial scalar size to be allocated
65		109
66	#define SB do {	110	#define SB do {
67	#define SE } while (0)	111	#define SE } while (0)
…		…
86	typedef struct {	130	typedef struct {
87	U32 flags;	131	U32 flags;
88	U32 max_depth;	132	U32 max_depth;
89	STRLEN max_size;	133	STRLEN max_size;
90	SV *filter;	134	SV *filter;
		135
		136	// for the incremental parser
		137	STRLEN incr_pos; // the current offset into the text
		138	STRLEN incr_need; // minimum bytes needed to decode
		139	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	140	} CBOR;
92		141
93	ecb_inline void	142	ecb_inline void
94	cbor_init (CBOR *cbor)	143	cbor_init (CBOR *cbor)
95	{	144	{
…		…
99		148
100	ecb_inline void	149	ecb_inline void
101	cbor_free (CBOR *cbor)	150	cbor_free (CBOR *cbor)
102	{	151	{
103	SvREFCNT_dec (cbor->filter);	152	SvREFCNT_dec (cbor->filter);
		153	SvREFCNT_dec (cbor->incr_count);
104	}	154	}
105		155
106	/////////////////////////////////////////////////////////////////////////////	156	/////////////////////////////////////////////////////////////////////////////
107	// utility functions	157	// utility functions
108		158
…		…
138	{	188	{
139	return idx > 23	189	return idx > 23
140	? idx > 0xffU	190	? idx > 0xffU
141	? idx > 0xffffU	191	? idx > 0xffffU
142	? idx > 0xffffffffU	192	? idx > 0xffffffffU
		193	? 11
143	? 7	194	: 7
144	: 6
145	: 5	195	: 5
146	: 4	196	: 4
147	: 3;	197	: 3;
148	}	198	}
149		199
…		…
186	static void	236	static void
187	encode_uint (enc_t *enc, int major, UV len)	237	encode_uint (enc_t *enc, int major, UV len)
188	{	238	{
189	need (enc, 9);	239	need (enc, 9);
190		240
191	if (len < 24)	241	if (ecb_expect_true (len < LENGTH_EXT1))
192	*enc->cur++ = major \| len;	242	*enc->cur++ = major \| len;
193	else if (len <= 0xff)	243	else if (ecb_expect_true (len <= 0xffU))
194	{	244	{
195	*enc->cur++ = major \| 24;	245	*enc->cur++ = major \| LENGTH_EXT1;
196	*enc->cur++ = len;	246	*enc->cur++ = len;
197	}	247	}
198	else if (len <= 0xffff)	248	else if (len <= 0xffffU)
199	{	249	{
200	*enc->cur++ = major \| 25;	250	*enc->cur++ = major \| LENGTH_EXT2;
201	*enc->cur++ = len >> 8;	251	*enc->cur++ = len >> 8;
202	*enc->cur++ = len;	252	*enc->cur++ = len;
203	}	253	}
204	else if (len <= 0xffffffff)	254	else if (len <= 0xffffffffU)
205	{	255	{
206	*enc->cur++ = major \| 26;	256	*enc->cur++ = major \| LENGTH_EXT4;
207	*enc->cur++ = len >> 24;	257	*enc->cur++ = len >> 24;
208	*enc->cur++ = len >> 16;	258	*enc->cur++ = len >> 16;
209	*enc->cur++ = len >> 8;	259	*enc->cur++ = len >> 8;
210	*enc->cur++ = len;	260	*enc->cur++ = len;
211	}	261	}
212	else	262	else
213	{	263	{
214	*enc->cur++ = major \| 27;	264	*enc->cur++ = major \| LENGTH_EXT8;
215	*enc->cur++ = len >> 56;	265	*enc->cur++ = len >> 56;
216	*enc->cur++ = len >> 48;	266	*enc->cur++ = len >> 48;
217	*enc->cur++ = len >> 40;	267	*enc->cur++ = len >> 40;
218	*enc->cur++ = len >> 32;	268	*enc->cur++ = len >> 32;
219	*enc->cur++ = len >> 24;	269	*enc->cur++ = len >> 24;
…		…
224	}	274	}
225		275
226	ecb_inline void	276	ecb_inline void
227	encode_tag (enc_t *enc, UV tag)	277	encode_tag (enc_t *enc, UV tag)
228	{	278	{
229	encode_uint (enc, 0xc0, tag);	279	encode_uint (enc, MAJOR_TAG, tag);
		280	}
		281
		282	ecb_inline void
		283	encode_str (enc_t enc, int utf8, char str, STRLEN len)
		284	{
		285	if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
		286	if (!utf8)
		287	{
		288	SV *sv = sv_newmortal ();
		289	char *s; STRLEN l;
		290
		291	sv_setpvn (sv, str, len);
		292
		293	s = SvPVutf8 (sv, l);
		294	encode_str (enc, 1, s, l);
		295	return;
		296	}
		297
		298	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
		299	need (enc, len);
		300	memcpy (enc->cur, str, len);
		301	enc->cur += len;
230	}	302	}
231		303
232	static void	304	static void
233	encode_str (enc_t enc, int utf8, char str, STRLEN len)	305	encode_strref (enc_t enc, int utf8, char str, STRLEN len)
234	{	306	{
235	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF))	307	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
236	{	308	{
237	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);	309	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
238		310
239	if (SvOK (*svp))	311	if (SvOK (*svp))
240	{	312	{
241	// already registered, use stringref	313	// already registered, use stringref
242	encode_tag (enc, CBOR_TAG_STRINGREF);	314	encode_tag (enc, CBOR_TAG_STRINGREF);
243	encode_uint (enc, 0x00, SvUV (*svp));	315	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
244	return;	316	return;
245	}	317	}
246	else if (len >= minimum_string_length (enc->stringref_idx))	318	else if (len >= minimum_string_length (enc->stringref_idx))
247	{	319	{
248	// register only	320	// register only
249	sv_setuv (*svp, enc->stringref_idx);	321	sv_setuv (*svp, enc->stringref_idx);
250	++enc->stringref_idx;	322	++enc->stringref_idx;
251	}	323	}
252	}	324	}
253		325
254	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	326	encode_str (enc, utf8, str, len);
255	need (enc, len);
256	memcpy (enc->cur, str, len);
257	enc->cur += len;
258	}	327	}
259		328
260	static void encode_sv (enc_t enc, SV sv);	329	static void encode_sv (enc_t enc, SV sv);
261		330
262	static void	331	static void
…		…
267	if (enc->depth >= enc->cbor.max_depth)	336	if (enc->depth >= enc->cbor.max_depth)
268	croak (ERR_NESTING_EXCEEDED);	337	croak (ERR_NESTING_EXCEEDED);
269		338
270	++enc->depth;	339	++enc->depth;
271		340
272	encode_uint (enc, 0x80, len + 1);	341	encode_uint (enc, MAJOR_ARRAY, len + 1);
273		342
		343	if (SvMAGICAL (av))
274	for (i = 0; i <= len; ++i)	344	for (i = 0; i <= len; ++i)
275	{	345	{
276	SV **svp = av_fetch (av, i, 0);	346	SV **svp = av_fetch (av, i, 0);
277	encode_sv (enc, svp ? *svp : &PL_sv_undef);	347	encode_sv (enc, svp ? *svp : &PL_sv_undef);
278	}	348	}
		349	else
		350	for (i = 0; i <= len; ++i)
		351	{
		352	SV *sv = AvARRAY (av)[i];
		353	encode_sv (enc, sv ? sv : &PL_sv_undef);
		354	}
279		355
280	--enc->depth;	356	--enc->depth;
281	}	357	}
282		358
283	static void	359	static void
…		…
292		368
293	int pairs = hv_iterinit (hv);	369	int pairs = hv_iterinit (hv);
294	int mg = SvMAGICAL (hv);	370	int mg = SvMAGICAL (hv);
295		371
296	if (mg)	372	if (mg)
297	encode_ch (enc, 0xa0 \| 31);	373	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
298	else	374	else
299	encode_uint (enc, 0xa0, pairs);	375	encode_uint (enc, MAJOR_MAP, pairs);
300		376
301	while ((he = hv_iternext (hv)))	377	while ((he = hv_iternext (hv)))
302	{	378	{
303	if (HeKLEN (he) == HEf_SVKEY)	379	if (HeKLEN (he) == HEf_SVKEY)
304	encode_sv (enc, HeSVKEY (he));	380	encode_sv (enc, HeSVKEY (he));
305	else	381	else
306	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));	382	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
307		383
308	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	384	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
309	}	385	}
310		386
311	if (mg)	387	if (mg)
312	encode_ch (enc, 0xe0 \| 31);	388	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
313		389
314	--enc->depth;	390	--enc->depth;
315	}	391	}
316		392
317	// encode objects, arrays and special \0=false and \1=true values.	393	// encode objects, arrays and special \0=false and \1=true values.
318	static void	394	static void
319	encode_rv (enc_t enc, SV sv)	395	encode_rv (enc_t enc, SV sv)
320	{	396	{
321	SvGETMAGIC (sv);	397	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		398
345	svtype svt = SvTYPE (sv);	399	svtype svt = SvTYPE (sv);
346		400
347	if (ecb_expect_false (SvOBJECT (sv)))	401	if (ecb_expect_false (SvOBJECT (sv)))
348	{	402	{
…		…
355	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	409	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
356	? cbor_tagged_stash	410	? cbor_tagged_stash
357	: gv_stashpv ("CBOR::XS::Tagged" , 1);	411	: gv_stashpv ("CBOR::XS::Tagged" , 1);
358		412
359	HV *stash = SvSTASH (sv);	413	HV *stash = SvSTASH (sv);
360	GV *method;
361		414
362	if (stash == boolean_stash)	415	if (stash == boolean_stash)
363	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	416	{
		417	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);
		418	return;
		419	}
364	else if (stash == error_stash)	420	else if (stash == error_stash)
365	encode_ch (enc, 0xe0 \| 23);	421	{
		422	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
		423	return;
		424	}
366	else if (stash == tagged_stash)	425	else if (stash == tagged_stash)
367	{	426	{
368	if (svt != SVt_PVAV)	427	if (svt != SVt_PVAV)
369	croak ("encountered CBOR::XS::Tagged object that isn't an array");	428	croak ("encountered CBOR::XS::Tagged object that isn't an array");
370		429
371	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	430	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
372	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	431	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		432
		433	return;
		434	}
		435	}
		436
		437	if (ecb_expect_false (SvREFCNT (sv) > 1)
		438	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		439	{
		440	if (!enc->shareable)
		441	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		442
		443	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		444
		445	if (SvOK (*svp))
373	}	446	{
		447	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		448	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		449	return;
		450	}
		451	else
		452	{
		453	sv_setuv (*svp, enc->shareable_idx);
		454	++enc->shareable_idx;
		455	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		456	}
		457	}
		458
		459	if (ecb_expect_false (SvOBJECT (sv)))
		460	{
		461	HV *stash = SvSTASH (sv);
		462	GV *method;
		463
374	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	464	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
375	{	465	{
376	dSP;	466	dSP;
377		467
378	ENTER; SAVETMPS; PUSHMARK (SP);	468	ENTER; SAVETMPS;
		469	PUSHMARK (SP);
379	// we re-bless the reference to get overload and other niceties right	470	// we re-bless the reference to get overload and other niceties right
380	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	471	XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
381		472
382	PUTBACK;	473	PUTBACK;
383	// G_SCALAR ensures that return value is 1	474	// G_SCALAR ensures that return value is 1
…		…
396	}	487	}
397	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)	488	else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
398	{	489	{
399	dSP;	490	dSP;
400		491
401	ENTER; SAVETMPS; PUSHMARK (SP);	492	ENTER; SAVETMPS;
		493	SAVESTACK_POS ();
		494	PUSHMARK (SP);
402	EXTEND (SP, 2);	495	EXTEND (SP, 2);
403	// we re-bless the reference to get overload and other niceties right	496	// we re-bless the reference to get overload and other niceties right
404	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	497	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
405	PUSHs (sv_cbor);	498	PUSHs (sv_cbor);
406		499
…		…
411	// catch this surprisingly common error	504	// catch this surprisingly common error
412	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	505	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));	506	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
414		507
415	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	508	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
416	encode_uint (enc, 0x80, count + 1);	509	encode_uint (enc, MAJOR_ARRAY, count + 1);
417	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	510	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
418		511
419	while (count)	512	while (count)
420	encode_sv (enc, SP[1 - count--]);	513	encode_sv (enc, SP[1 - count--]);
421		514
422	PUTBACK;	515	PUTBACK;
…		…
443	{	536	{
444	double nv = SvNVX (sv);	537	double nv = SvNVX (sv);
445		538
446	need (enc, 9);	539	need (enc, 9);
447		540
448	if (ecb_expect_false (nv == (U32)nv))	541	if (ecb_expect_false (nv == (NV)(U32)nv))
449	encode_uint (enc, 0x00, (U32)nv);	542	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
450	//TODO: maybe I32?	543	//TODO: maybe I32?
451	else if (ecb_expect_false (nv == (float)nv))	544	else if (ecb_expect_false (nv == (float)nv))
452	{	545	{
453	uint32_t fp = ecb_float_to_binary32 (nv);	546	uint32_t fp = ecb_float_to_binary32 (nv);
454		547
455	*enc->cur++ = 0xe0 \| 26;	548	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
456		549
457	if (!ecb_big_endian ())	550	if (!ecb_big_endian ())
458	fp = ecb_bswap32 (fp);	551	fp = ecb_bswap32 (fp);
459		552
460	memcpy (enc->cur, &fp, 4);	553	memcpy (enc->cur, &fp, 4);
…		…
462	}	555	}
463	else	556	else
464	{	557	{
465	uint64_t fp = ecb_double_to_binary64 (nv);	558	uint64_t fp = ecb_double_to_binary64 (nv);
466		559
467	*enc->cur++ = 0xe0 \| 27;	560	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
468		561
469	if (!ecb_big_endian ())	562	if (!ecb_big_endian ())
470	fp = ecb_bswap64 (fp);	563	fp = ecb_bswap64 (fp);
471		564
472	memcpy (enc->cur, &fp, 8);	565	memcpy (enc->cur, &fp, 8);
…		…
481		574
482	if (SvPOKp (sv))	575	if (SvPOKp (sv))
483	{	576	{
484	STRLEN len;	577	STRLEN len;
485	char *str = SvPV (sv, len);	578	char *str = SvPV (sv, len);
486	encode_str (enc, SvUTF8 (sv), str, len);	579	encode_strref (enc, SvUTF8 (sv), str, len);
487	}	580	}
488	else if (SvNOKp (sv))	581	else if (SvNOKp (sv))
489	encode_nv (enc, sv);	582	encode_nv (enc, sv);
490	else if (SvIOKp (sv))	583	else if (SvIOKp (sv))
491	{	584	{
492	if (SvIsUV (sv))	585	if (SvIsUV (sv))
493	encode_uint (enc, 0x00, SvUVX (sv));	586	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
494	else if (SvIVX (sv) >= 0)	587	else if (SvIVX (sv) >= 0)
495	encode_uint (enc, 0x00, SvIVX (sv));	588	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
496	else	589	else
497	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	590	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
498	}	591	}
499	else if (SvROK (sv))	592	else if (SvROK (sv))
500	encode_rv (enc, SvRV (sv));	593	encode_rv (enc, SvRV (sv));
501	else if (!SvOK (sv))	594	else if (!SvOK (sv))
502	encode_ch (enc, 0xe0 \| 22);	595	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
503	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	596	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
504	encode_ch (enc, 0xe0 \| 23);	597	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
505	else	598	else
506	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	599	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
507	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	600	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
508	}	601	}
509		602
510	static SV *	603	static SV *
511	encode_cbor (SV scalar, CBOR cbor)	604	encode_cbor (SV scalar, CBOR cbor)
512	{	605	{
513	enc_t enc = { };	606	enc_t enc = { 0 };
514		607
515	enc.cbor = *cbor;	608	enc.cbor = *cbor;
516	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));	609	enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
517	enc.cur = SvPVX (enc.sv);	610	enc.cur = SvPVX (enc.sv);
518	enc.end = SvEND (enc.sv);	611	enc.end = SvEND (enc.sv);
519		612
520	SvPOK_only (enc.sv);	613	SvPOK_only (enc.sv);
521		614
522	if (cbor->flags & F_ALLOW_STRINGREF)	615	if (cbor->flags & F_PACK_STRINGS)
523	{	616	{
524	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	617	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
525	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());	618	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
526	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());	619	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
527	}	620	}
…		…
562	#define DEC_DEC_DEPTH --dec->depth	655	#define DEC_DEC_DEPTH --dec->depth
563		656
564	static UV	657	static UV
565	decode_uint (dec_t *dec)	658	decode_uint (dec_t *dec)
566	{	659	{
567	switch (*dec->cur & 31)	660	U8 m = *dec->cur & MINOR_MASK;
568	{	661	++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		662
574	case 24:	663	if (ecb_expect_true (m < LENGTH_EXT1))
		664	return m;
		665	else if (ecb_expect_true (m == LENGTH_EXT1))
		666	{
575	WANT (2);	667	WANT (1);
576	dec->cur += 2;	668	dec->cur += 1;
577	return dec->cur[-1];	669	return dec->cur[-1];
578		670	}
579	case 25:	671	else if (ecb_expect_true (m == LENGTH_EXT2))
		672	{
580	WANT (3);	673	WANT (2);
581	dec->cur += 3;	674	dec->cur += 2;
582	return (((UV)dec->cur[-2]) << 8)	675	return (((UV)dec->cur[-2]) << 8)
583	\| ((UV)dec->cur[-1]);	676	\| ((UV)dec->cur[-1]);
584		677	}
585	case 26:	678	else if (ecb_expect_true (m == LENGTH_EXT4))
		679	{
586	WANT (5);	680	WANT (4);
587	dec->cur += 5;	681	dec->cur += 4;
588	return (((UV)dec->cur[-4]) << 24)	682	return (((UV)dec->cur[-4]) << 24)
589	\| (((UV)dec->cur[-3]) << 16)	683	\| (((UV)dec->cur[-3]) << 16)
590	\| (((UV)dec->cur[-2]) << 8)	684	\| (((UV)dec->cur[-2]) << 8)
591	\| ((UV)dec->cur[-1]);	685	\| ((UV)dec->cur[-1]);
592		686	}
593	case 27:	687	else if (ecb_expect_true (m == LENGTH_EXT8))
		688	{
594	WANT (9);	689	WANT (8);
595	dec->cur += 9;	690	dec->cur += 8;
		691
		692	return
		693	#if UVSIZE < 8
		694	0
		695	#else
596	return (((UV)dec->cur[-8]) << 56)	696	(((UV)dec->cur[-8]) << 56)
597	\| (((UV)dec->cur[-7]) << 48)	697	\| (((UV)dec->cur[-7]) << 48)
598	\| (((UV)dec->cur[-6]) << 40)	698	\| (((UV)dec->cur[-6]) << 40)
599	\| (((UV)dec->cur[-5]) << 32)	699	\| (((UV)dec->cur[-5]) << 32)
		700	#endif
600	\| (((UV)dec->cur[-4]) << 24)	701	\| (((UV)dec->cur[-4]) << 24)
601	\| (((UV)dec->cur[-3]) << 16)	702	\| (((UV)dec->cur[-3]) << 16)
602	\| (((UV)dec->cur[-2]) << 8)	703	\| (((UV)dec->cur[-2]) << 8)
603	\| ((UV)dec->cur[-1]);	704	\| ((UV)dec->cur[-1]);
604		705	}
605	default:	706	else
606	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	707	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
607	}
608		708
609	fail:	709	fail:
610	return 0;	710	return 0;
611	}	711	}
612		712
…		…
617	{	717	{
618	AV *av = newAV ();	718	AV *av = newAV ();
619		719
620	DEC_INC_DEPTH;	720	DEC_INC_DEPTH;
621		721
622	if ((*dec->cur & 31) == 31)	722	if (*dec->cur == (MAJOR_ARRAY \| MINOR_INDEF))
623	{	723	{
624	++dec->cur;	724	++dec->cur;
625		725
626	for (;;)	726	for (;;)
627	{	727	{
628	WANT (1);	728	WANT (1);
629		729
630	if (*dec->cur == (0xe0 \| 31))	730	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
631	{	731	{
632	++dec->cur;	732	++dec->cur;
633	break;	733	break;
634	}	734	}
635		735
…		…
638	}	738	}
639	else	739	else
640	{	740	{
641	int i, len = decode_uint (dec);	741	int i, len = decode_uint (dec);
642		742
		743	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);	744	av_fill (av, len - 1);
644		745
645	for (i = 0; i < len; ++i)	746	for (i = 0; i < len; ++i)
646	AvARRAY (av)[i] = decode_sv (dec);	747	AvARRAY (av)[i] = decode_sv (dec);
647	}	748	}
…		…
660	{	761	{
661	// for speed reasons, we specialcase single-string	762	// for speed reasons, we specialcase single-string
662	// byte or utf-8 strings as keys, but only when !stringref	763	// byte or utf-8 strings as keys, but only when !stringref
663		764
664	if (ecb_expect_true (!dec->stringref))	765	if (ecb_expect_true (!dec->stringref))
665	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	766	if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
666	{	767	{
667	I32 len = decode_uint (dec);	768	I32 len = decode_uint (dec);
668	char key = (char )dec->cur;	769	char key = (char )dec->cur;
669		770
		771	WANT (len);
670	dec->cur += len;	772	dec->cur += len;
671		773
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);	774	hv_store (hv, key, len, decode_sv (dec), 0);
676		775
677	return;	776	return;
678	}	777	}
679	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	778	else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
680	{	779	{
681	I32 len = decode_uint (dec);	780	I32 len = decode_uint (dec);
682	char key = (char )dec->cur;	781	char key = (char )dec->cur;
683		782
		783	WANT (len);
684	dec->cur += len;	784	dec->cur += len;
685		785
686	if (ecb_expect_false (dec->stringref))	786	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
687	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	787	if (!is_utf8_string (key, len))
		788	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
688		789
689	hv_store (hv, key, -len, decode_sv (dec), 0);	790	hv_store (hv, key, -len, decode_sv (dec), 0);
690		791
691	return;	792	return;
692	}	793	}
…		…
694	SV *k = decode_sv (dec);	795	SV *k = decode_sv (dec);
695	SV *v = decode_sv (dec);	796	SV *v = decode_sv (dec);
696		797
697	hv_store_ent (hv, k, v, 0);	798	hv_store_ent (hv, k, v, 0);
698	SvREFCNT_dec (k);	799	SvREFCNT_dec (k);
		800
		801	fail:
		802	;
699	}	803	}
700		804
701	static SV *	805	static SV *
702	decode_hv (dec_t *dec)	806	decode_hv (dec_t *dec)
703	{	807	{
704	HV *hv = newHV ();	808	HV *hv = newHV ();
705		809
706	DEC_INC_DEPTH;	810	DEC_INC_DEPTH;
707		811
708	if ((*dec->cur & 31) == 31)	812	if (*dec->cur == (MAJOR_MAP \| MINOR_INDEF))
709	{	813	{
710	++dec->cur;	814	++dec->cur;
711		815
712	for (;;)	816	for (;;)
713	{	817	{
714	WANT (1);	818	WANT (1);
715		819
716	if (*dec->cur == (0xe0 \| 31))	820	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
717	{	821	{
718	++dec->cur;	822	++dec->cur;
719	break;	823	break;
720	}	824	}
721		825
…		…
742	static SV *	846	static SV *
743	decode_str (dec_t *dec, int utf8)	847	decode_str (dec_t *dec, int utf8)
744	{	848	{
745	SV *sv = 0;	849	SV *sv = 0;
746		850
747	if ((*dec->cur & 31) == 31)	851	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
748	{	852	{
		853	// indefinite length strings
749	++dec->cur;	854	++dec->cur;
750		855
		856	U8 major = *dec->cur & MAJOR_MISC;
		857
751	sv = newSVpvn ("", 0);	858	sv = newSVpvn ("", 0);
752		859
753	// not very fast, and certainly not robust against illegal input
754	for (;;)	860	for (;;)
755	{	861	{
756	WANT (1);	862	WANT (1);
757		863
758	if (*dec->cur == (0xe0 \| 31))	864	if ((*dec->cur - major) > LENGTH_EXT8)
		865	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
759	{	866	{
760	++dec->cur;	867	++dec->cur;
761	break;	868	break;
762	}	869	}
		870	else
		871	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
763		872
764	sv_catsv (sv, decode_sv (dec));	873	STRLEN len = decode_uint (dec);
		874
		875	WANT (len);
		876	sv_catpvn (sv, dec->cur, len);
		877	dec->cur += len;
765	}	878	}
766	}	879	}
767	else	880	else
768	{	881	{
769	STRLEN len = decode_uint (dec);	882	STRLEN len = decode_uint (dec);
…		…
776	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))	889	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
777	av_push (dec->stringref, SvREFCNT_inc_NN (sv));	890	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
778	}	891	}
779		892
780	if (utf8)	893	if (utf8)
		894	{
		895	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		896	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		897	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		898
781	SvUTF8_on (sv);	899	SvUTF8_on (sv);
		900	}
782		901
783	return sv;	902	return sv;
784		903
785	fail:	904	fail:
786	SvREFCNT_dec (sv);	905	SvREFCNT_dec (sv);
…		…
818	}	937	}
819	break;	938	break;
820		939
821	case CBOR_TAG_STRINGREF:	940	case CBOR_TAG_STRINGREF:
822	{	941	{
823	if ((*dec->cur >> 5) != 0)	942	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
824	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	943	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
825		944
826	UV idx = decode_uint (dec);	945	UV idx = decode_uint (dec);
827		946
828	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	947	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
…		…
835	case CBOR_TAG_VALUE_SHAREABLE:	954	case CBOR_TAG_VALUE_SHAREABLE:
836	{	955	{
837	if (ecb_expect_false (!dec->shareable))	956	if (ecb_expect_false (!dec->shareable))
838	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	957	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
839		958
		959	if (dec->cbor.flags & F_ALLOW_CYCLES)
		960	{
840	sv = newSV (0);	961	sv = newSV (0);
841	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	962	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
842		963
843	SV *osv = decode_sv (dec);	964	SV *osv = decode_sv (dec);
844	sv_setsv (sv, osv);	965	sv_setsv (sv, osv);
845	SvREFCNT_dec_NN (osv);	966	SvREFCNT_dec_NN (osv);
		967	}
		968	else
		969	{
		970	av_push (dec->shareable, &PL_sv_undef);
		971	int idx = AvFILLp (dec->shareable);
		972	sv = decode_sv (dec);
		973	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		974	}
846	}	975	}
847	break;	976	break;
848		977
849	case CBOR_TAG_VALUE_SHAREDREF:	978	case CBOR_TAG_VALUE_SHAREDREF:
850	{	979	{
851	if ((*dec->cur >> 5) != 0)	980	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
852	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	981	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
853		982
854	UV idx = decode_uint (dec);	983	UV idx = decode_uint (dec);
855		984
856	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	985	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
857	ERR ("corrupted CBOR data (sharedref index out of bounds)");	986	ERR ("corrupted CBOR data (sharedref index out of bounds)");
858		987
859	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	988	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		989
		990	if (sv == &PL_sv_undef)
		991	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
860	}	992	}
861	break;	993	break;
862		994
863	case CBOR_TAG_PERL_OBJECT:	995	case CBOR_TAG_PERL_OBJECT:
864	{	996	{
…		…
879	if (!method)	1011	if (!method)
880	ERR ("cannot decode perl-object (package does not have a THAW method)");	1012	ERR ("cannot decode perl-object (package does not have a THAW method)");
881		1013
882	dSP;	1014	dSP;
883		1015
884	ENTER; SAVETMPS; PUSHMARK (SP);	1016	ENTER; SAVETMPS;
		1017	PUSHMARK (SP);
885	EXTEND (SP, len + 1);	1018	EXTEND (SP, len + 1);
886	// we re-bless the reference to get overload and other niceties right	1019	// we re-bless the reference to get overload and other niceties right
887	PUSHs (*av_fetch (av, 0, 1));	1020	PUSHs (*av_fetch (av, 0, 1));
888	PUSHs (sv_cbor);	1021	PUSHs (sv_cbor);
889		1022
…		…
914	default:	1047	default:
915	{	1048	{
916	sv = decode_sv (dec);	1049	sv = decode_sv (dec);
917		1050
918	dSP;	1051	dSP;
919	ENTER; SAVETMPS; PUSHMARK (SP);	1052	ENTER; SAVETMPS;
		1053	SAVESTACK_POS ();
		1054	PUSHMARK (SP);
920	EXTEND (SP, 2);	1055	EXTEND (SP, 2);
921	PUSHs (newSVuv (tag));	1056	PUSHs (newSVuv (tag));
922	PUSHs (sv);	1057	PUSHs (sv);
923		1058
924	PUTBACK;	1059	PUTBACK;
…		…
965	static SV *	1100	static SV *
966	decode_sv (dec_t *dec)	1101	decode_sv (dec_t *dec)
967	{	1102	{
968	WANT (1);	1103	WANT (1);
969		1104
970	switch (*dec->cur >> 5)	1105	switch (*dec->cur >> MAJOR_SHIFT)
971	{	1106	{
972	case 0: // unsigned int	1107	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
973	return newSVuv (decode_uint (dec));	1108	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
974	case 1: // negative int	1109	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
975	return newSViv (-1 - (IV)decode_uint (dec));	1110	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
976	case 2: // octet string	1111	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
977	return decode_str (dec, 0);	1112	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
978	case 3: // utf-8 string	1113	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
979	return decode_str (dec, 1);	1114
980	case 4: // array	1115	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)	1116	switch (*dec->cur++ & MINOR_MASK)
988	{	1117	{
989	case 20:	1118	case SIMPLE_FALSE:
990	#if CBOR_SLOW	1119	#if CBOR_SLOW
991	types_false = get_bool ("Types::Serialiser::false");	1120	types_false = get_bool ("Types::Serialiser::false");
992	#endif	1121	#endif
993	return newSVsv (types_false);	1122	return newSVsv (types_false);
994	case 21:	1123	case SIMPLE_TRUE:
995	#if CBOR_SLOW	1124	#if CBOR_SLOW
996	types_true = get_bool ("Types::Serialiser::true");	1125	types_true = get_bool ("Types::Serialiser::true");
997	#endif	1126	#endif
998	return newSVsv (types_true);	1127	return newSVsv (types_true);
999	case 22:	1128	case SIMPLE_NULL:
1000	return newSVsv (&PL_sv_undef);	1129	return newSVsv (&PL_sv_undef);
1001	case 23:	1130	case SIMPLE_UNDEF:
1002	#if CBOR_SLOW	1131	#if CBOR_SLOW
1003	types_error = get_bool ("Types::Serialiser::error");	1132	types_error = get_bool ("Types::Serialiser::error");
1004	#endif	1133	#endif
1005	return newSVsv (types_error);	1134	return newSVsv (types_error);
1006		1135
1007	case 25:	1136	case MISC_FLOAT16:
1008	{	1137	{
1009	WANT (2);	1138	WANT (2);
1010		1139
1011	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];	1140	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];
1012	dec->cur += 2;	1141	dec->cur += 2;
1013		1142
1014	return newSVnv (ecb_binary16_to_float (fp));	1143	return newSVnv (ecb_binary16_to_float (fp));
1015	}	1144	}
1016		1145
1017	case 26:	1146	case MISC_FLOAT32:
1018	{	1147	{
1019	uint32_t fp;	1148	uint32_t fp;
1020	WANT (4);	1149	WANT (4);
1021	memcpy (&fp, dec->cur, 4);	1150	memcpy (&fp, dec->cur, 4);
1022	dec->cur += 4;	1151	dec->cur += 4;
…		…
1025	fp = ecb_bswap32 (fp);	1154	fp = ecb_bswap32 (fp);
1026		1155
1027	return newSVnv (ecb_binary32_to_float (fp));	1156	return newSVnv (ecb_binary32_to_float (fp));
1028	}	1157	}
1029		1158
1030	case 27:	1159	case MISC_FLOAT64:
1031	{	1160	{
1032	uint64_t fp;	1161	uint64_t fp;
1033	WANT (8);	1162	WANT (8);
1034	memcpy (&fp, dec->cur, 8);	1163	memcpy (&fp, dec->cur, 8);
1035	dec->cur += 8;	1164	dec->cur += 8;
…		…
1038	fp = ecb_bswap64 (fp);	1167	fp = ecb_bswap64 (fp);
1039		1168
1040	return newSVnv (ecb_binary64_to_double (fp));	1169	return newSVnv (ecb_binary64_to_double (fp));
1041	}	1170	}
1042		1171
1043	// 0..19 unassigned	1172	// 0..19 unassigned simple
1044	// 24 reserved + unassigned (reserved values are not encodable)	1173	// 24 reserved + unassigned simple (reserved values are not encodable)
		1174	// 28-30 unassigned misc
		1175	// 31 break code
1045	default:	1176	default:
1046	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1177	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1047	}	1178	}
1048		1179
1049	break;	1180	break;
1050	}	1181	}
1051		1182
…		…
1054	}	1185	}
1055		1186
1056	static SV *	1187	static SV *
1057	decode_cbor (SV string, CBOR cbor, char **offset_return)	1188	decode_cbor (SV string, CBOR cbor, char **offset_return)
1058	{	1189	{
1059	dec_t dec = { };	1190	dec_t dec = { 0 };
1060	SV *sv;	1191	SV *sv;
1061	STRLEN len;	1192	STRLEN len;
1062	char *data = SvPVbyte (string, len);	1193	char *data = SvPVbyte (string, len);
1063		1194
1064	if (len > cbor->max_size && cbor->max_size)	1195	if (len > cbor->max_size && cbor->max_size)
…		…
1078	if (dec.cur != dec.end && !dec.err)	1209	if (dec.cur != dec.end && !dec.err)
1079	dec.err = "garbage after CBOR object";	1210	dec.err = "garbage after CBOR object";
1080		1211
1081	if (dec.err)	1212	if (dec.err)
1082	{	1213	{
		1214	if (dec.shareable)
		1215	{
		1216	// need to break cyclic links, which whould all be in shareable
		1217	int i;
		1218	SV **svp;
		1219
		1220	for (i = av_len (dec.shareable) + 1; i--; )
		1221	if ((svp = av_fetch (dec.shareable, i, 0)))
		1222	sv_setsv (*svp, &PL_sv_undef);
		1223	}
		1224
1083	SvREFCNT_dec (sv);	1225	SvREFCNT_dec (sv);
1084	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1226	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1085	}	1227	}
1086		1228
1087	sv = sv_2mortal (sv);	1229	sv = sv_2mortal (sv);
1088		1230
1089	return sv;	1231	return sv;
1090	}	1232	}
1091		1233
		1234	/////////////////////////////////////////////////////////////////////////////
		1235	// incremental parser
		1236
		1237	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1238
		1239	// returns 0 for notyet, 1 for success or error
		1240	static int
		1241	incr_parse (CBOR self, SV cborstr)
		1242	{
		1243	STRLEN cur;
		1244	SvPV (cborstr, cur);
		1245
		1246	while (ecb_expect_true (self->incr_need <= cur))
		1247	{
		1248	// table of integer count bytes
		1249	static I8 incr_len[MINOR_MASK + 1] = {
		1250	0, 0, 0, 0, 0, 0, 0, 0,
		1251	0, 0, 0, 0, 0, 0, 0, 0,
		1252	0, 0, 0, 0, 0, 0, 0, 0,
		1253	1, 2, 4, 8,-1,-1,-1,-2
		1254	};
		1255
		1256	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1257	U8 m = *p & MINOR_MASK;
		1258	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1259	I8 ilen = incr_len[m];
		1260
		1261	self->incr_need = self->incr_pos + 1;
		1262
		1263	if (ecb_expect_false (ilen < 0))
		1264	{
		1265	if (m != MINOR_INDEF)
		1266	return 1; // error
		1267
		1268	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1269	{
		1270	if (count >= 0)
		1271	return 1; // error
		1272
		1273	count = 1;
		1274	}
		1275	else
		1276	{
		1277	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1278	count = -1;
		1279	}
		1280	}
		1281	else
		1282	{
		1283	self->incr_need += ilen;
		1284	if (ecb_expect_false (self->incr_need > cur))
		1285	return 0;
		1286
		1287	int major = *p >> MAJOR_SHIFT;
		1288
		1289	switch (major)
		1290	{
		1291	case MAJOR_TAG >> MAJOR_SHIFT:
		1292	++count; // tags merely prefix another value
		1293	break;
		1294
		1295	case MAJOR_BYTES >> MAJOR_SHIFT:
		1296	case MAJOR_TEXT >> MAJOR_SHIFT:
		1297	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1298	case MAJOR_MAP >> MAJOR_SHIFT:
		1299	{
		1300	UV len;
		1301
		1302	if (ecb_expect_false (ilen))
		1303	{
		1304	len = 0;
		1305
		1306	do {
		1307	len = (len << 8) \| *++p;
		1308	} while (--ilen);
		1309	}
		1310	else
		1311	len = m;
		1312
		1313	switch (major)
		1314	{
		1315	case MAJOR_BYTES >> MAJOR_SHIFT:
		1316	case MAJOR_TEXT >> MAJOR_SHIFT:
		1317	self->incr_need += len;
		1318	if (ecb_expect_false (self->incr_need > cur))
		1319	return 0;
		1320
		1321	break;
		1322
		1323	case MAJOR_MAP >> MAJOR_SHIFT:
		1324	len <<= 1;
		1325	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1326	if (len)
		1327	{
		1328	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1329	count = len + 1;
		1330	}
		1331	break;
		1332	}
		1333	}
		1334	}
		1335	}
		1336
		1337	self->incr_pos = self->incr_need;
		1338
		1339	if (count > 0)
		1340	{
		1341	while (!--count)
		1342	{
		1343	if (!AvFILLp (self->incr_count))
		1344	return 1; // done
		1345
		1346	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1347	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1348	}
		1349
		1350	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1351	}
		1352	}
		1353
		1354	return 0;
		1355	}
		1356
		1357
1092	/////////////////////////////////////////////////////////////////////////////	1358	/////////////////////////////////////////////////////////////////////////////
1093	// XS interface functions	1359	// XS interface functions
1094		1360
1095	MODULE = CBOR::XS PACKAGE = CBOR::XS	1361	MODULE = CBOR::XS PACKAGE = CBOR::XS
1096		1362
…		…
1136	void shrink (CBOR *self, int enable = 1)	1402	void shrink (CBOR *self, int enable = 1)
1137	ALIAS:	1403	ALIAS:
1138	shrink = F_SHRINK	1404	shrink = F_SHRINK
1139	allow_unknown = F_ALLOW_UNKNOWN	1405	allow_unknown = F_ALLOW_UNKNOWN
1140	allow_sharing = F_ALLOW_SHARING	1406	allow_sharing = F_ALLOW_SHARING
1141	allow_stringref = F_ALLOW_STRINGREF	1407	allow_cycles = F_ALLOW_CYCLES
		1408	pack_strings = F_PACK_STRINGS
		1409	utf8_strings = F_UTF8_STRINGS
		1410	validate_utf8 = F_VALIDATE_UTF8
1142	PPCODE:	1411	PPCODE:
1143	{	1412	{
1144	if (enable)	1413	if (enable)
1145	self->flags \|= ix;	1414	self->flags \|= ix;
1146	else	1415	else
…		…
1152	void get_shrink (CBOR *self)	1421	void get_shrink (CBOR *self)
1153	ALIAS:	1422	ALIAS:
1154	get_shrink = F_SHRINK	1423	get_shrink = F_SHRINK
1155	get_allow_unknown = F_ALLOW_UNKNOWN	1424	get_allow_unknown = F_ALLOW_UNKNOWN
1156	get_allow_sharing = F_ALLOW_SHARING	1425	get_allow_sharing = F_ALLOW_SHARING
1157	get_allow_stringref = F_ALLOW_STRINGREF	1426	get_allow_cycles = F_ALLOW_CYCLES
		1427	get_pack_strings = F_PACK_STRINGS
		1428	get_validate_utf8 = F_VALIDATE_UTF8
1158	PPCODE:	1429	PPCODE:
1159	XPUSHs (boolSV (self->flags & ix));	1430	XPUSHs (boolSV (self->flags & ix));
1160		1431
1161	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1432	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1162	PPCODE:	1433	PPCODE:
…		…
1211	EXTEND (SP, 2);	1482	EXTEND (SP, 2);
1212	PUSHs (sv);	1483	PUSHs (sv);
1213	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1484	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1214	}	1485	}
1215		1486
		1487	void incr_parse (CBOR self, SV cborstr)
		1488	ALIAS:
		1489	incr_parse_multiple = 1
		1490	PPCODE:
		1491	{
		1492	if (SvUTF8 (cborstr))
		1493	sv_utf8_downgrade (cborstr, 0);
		1494
		1495	if (!self->incr_count)
		1496	{
		1497	self->incr_count = newAV ();
		1498	self->incr_pos = 0;
		1499	self->incr_need = 1;
		1500
		1501	av_push (self->incr_count, newSViv (1));
		1502	}
		1503
		1504	do
		1505	{
		1506	if (!incr_parse (self, cborstr))
		1507	{
		1508	if (self->incr_need > self->max_size && self->max_size)
		1509	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1510	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1511
		1512	break;
		1513	}
		1514
		1515	SV *sv;
		1516	char *offset;
		1517
		1518	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1519	XPUSHs (sv);
		1520
		1521	sv_chop (cborstr, offset);
		1522
		1523	av_clear (self->incr_count);
		1524	av_push (self->incr_count, newSViv (1));
		1525
		1526	self->incr_pos = 0;
		1527	self->incr_need = self->incr_pos + 1;
		1528	}
		1529	while (ix);
		1530	}
		1531
		1532	void incr_reset (CBOR *self)
		1533	CODE:
		1534	{
		1535	SvREFCNT_dec (self->incr_count);
		1536	self->incr_count = 0;
		1537	}
		1538
1216	void DESTROY (CBOR *self)	1539	void DESTROY (CBOR *self)
1217	PPCODE:	1540	PPCODE:
1218	cbor_free (self);	1541	cbor_free (self);
1219		1542
1220	PROTOTYPES: ENABLE	1543	PROTOTYPES: ENABLE
1221		1544
1222	void encode_cbor (SV *scalar)	1545	void encode_cbor (SV *scalar)
		1546	ALIAS:
		1547	encode_cbor = 0
		1548	encode_cbor_sharing = F_ALLOW_SHARING
1223	PPCODE:	1549	PPCODE:
1224	{	1550	{
1225	CBOR cbor;	1551	CBOR cbor;
1226	cbor_init (&cbor);	1552	cbor_init (&cbor);
		1553	cbor.flags \|= ix;
1227	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1554	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1228	XPUSHs (scalar);	1555	XPUSHs (scalar);
1229	}	1556	}
1230		1557
1231	void decode_cbor (SV *cborstr)	1558	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 1.51 by root, Sun Apr 24 13:15:19 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 1.51 by root, Sun Apr 24 13:15:19 2016 UTC