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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.42 by root, Tue Dec 10 15:31:40 2013 UTC

…		…
19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)	19	# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20	#endif	20	#endif
21	#ifndef HvNAMEUTF8	21	#ifndef HvNAMEUTF8
22	# define HvNAMEUTF8(hv) 0	22	# define HvNAMEUTF8(hv) 0
23	#endif	23	#endif
		24	#ifndef SvREFCNT_dec_NN
		25	# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
		26	#endif
		27
		28	// known major and minor types
		29	enum cbor_type
		30	{
		31	MAJOR_SHIFT = 5,
		32	MINOR_MASK = 0x1f,
		33
		34	MAJOR_POS_INT = 0 << MAJOR_SHIFT,
		35	MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
		36	MAJOR_BYTES = 2 << MAJOR_SHIFT,
		37	MAJOR_TEXT = 3 << MAJOR_SHIFT,
		38	MAJOR_ARRAY = 4 << MAJOR_SHIFT,
		39	MAJOR_MAP = 5 << MAJOR_SHIFT,
		40	MAJOR_TAG = 6 << MAJOR_SHIFT,
		41	MAJOR_MISC = 7 << MAJOR_SHIFT,
		42
		43	// INT/STRING/ARRAY/MAP subtypes
		44	LENGTH_EXT1 = 24,
		45	LENGTH_EXT2 = 25,
		46	LENGTH_EXT4 = 26,
		47	LENGTH_EXT8 = 27,
		48
		49	// SIMPLE types (effectively MISC subtypes)
		50	SIMPLE_FALSE = 20,
		51	SIMPLE_TRUE = 21,
		52	SIMPLE_NULL = 22,
		53	SIMPLE_UNDEF = 23,
		54
		55	// MISC subtype (unused)
		56	MISC_EXT1 = 24,
		57	MISC_FLOAT16 = 25,
		58	MISC_FLOAT32 = 26,
		59	MISC_FLOAT64 = 27,
		60
		61	// BYTES/TEXT/ARRAY/MAP
		62	MINOR_INDEF = 31,
		63	};
24		64
25	// known tags	65	// known tags
26	enum cbor_tag	66	enum cbor_tag
27	{	67	{
28	// inofficial extensions (pending iana registration)	68	// extensions
		69	CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29	CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object	70	CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30	CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object	71	CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31	CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing	72	CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32	CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing	73	CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33	CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref	74	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	75	CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36		76
37	// rfc7049	77	// rfc7049
38	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8	78	CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any	79	CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40	CBOR_TAG_POS_BIGNUM = 2, // byte string	80	CBOR_TAG_POS_BIGNUM = 2, // byte string
41	CBOR_TAG_NEG_BIGNUM = 3, // byte string	81	CBOR_TAG_NEG_BIGNUM = 3, // byte string
42	CBOR_TAG_DECIMAL = 4, // decimal fraction, array	82	CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43	CBOR_TAG_BIGFLOAT = 5, // array	83	CBOR_TAG_BIGFLOAT = 5, // array
44		84
45	CBOR_TAG_CONV_B64U = 21, // base64url, any	85	CBOR_TAG_CONV_B64U = 21, // base64url, any
46	CBOR_TAG_CONV_B64 = 22, // base64, any	86	CBOR_TAG_CONV_B64 = 22, // base64, any
47	CBOR_TAG_CONV_HEX = 23, // base16, any	87	CBOR_TAG_CONV_HEX = 23, // base16, any
48	CBOR_TAG_CBOR = 24, // embedded cbor, byte string	88	CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49		89
50	CBOR_TAG_URI = 32, // URI rfc3986, utf-8	90	CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8	91	CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8	92	CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8	93	CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8	94	CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55		95
56	CBOR_TAG_MAGIC = 55799 // self-describe cbor	96	CBOR_TAG_MAGIC = 55799, // self-describe cbor
57	};	97	};
58		98
59	#define F_SHRINK 0x00000001UL	99	#define F_SHRINK 0x00000001UL
60	#define F_ALLOW_UNKNOWN 0x00000002UL	100	#define F_ALLOW_UNKNOWN 0x00000002UL
61	#define F_ALLOW_SHARING 0x00000004UL //TODO	101	#define F_ALLOW_SHARING 0x00000004UL
		102	#define F_ALLOW_CYCLES 0x00000008UL
62	#define F_DEDUP_STRINGS 0x00000008UL //TODO	103	#define F_PACK_STRINGS 0x00000010UL
63	#define F_DEDUP_KEYS 0x00000010UL //TODO	104	#define F_VALIDATE_UTF8 0x00000020UL
64		105
65	#define INIT_SIZE 32 // initial scalar size to be allocated	106	#define INIT_SIZE 32 // initial scalar size to be allocated
66		107
67	#define SB do {	108	#define SB do {
68	#define SE } while (0)	109	#define SE } while (0)
…		…
80	# define CBOR_SLOW 0	121	# define CBOR_SLOW 0
81	# define CBOR_STASH cbor_stash	122	# define CBOR_STASH cbor_stash
82	#endif	123	#endif
83		124
84	static HV cbor_stash, types_boolean_stash, types_error_stash, cbor_tagged_stash; // CBOR::XS::	125	static HV cbor_stash, types_boolean_stash, types_error_stash, cbor_tagged_stash; // CBOR::XS::
85	static SV types_true, types_false, types_error, sv_cbor;	126	static SV types_true, types_false, types_error, sv_cbor, *default_filter;
86		127
87	typedef struct {	128	typedef struct {
88	U32 flags;	129	U32 flags;
89	U32 max_depth;	130	U32 max_depth;
90	STRLEN max_size;	131	STRLEN max_size;
		132	SV *filter;
		133
		134	// for the incremental parser
		135	STRLEN incr_pos; // the current offset into the text
		136	STRLEN incr_need; // minimum bytes needed to decode
		137	AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91	} CBOR;	138	} CBOR;
92		139
93	ecb_inline void	140	ecb_inline void
94	cbor_init (CBOR *cbor)	141	cbor_init (CBOR *cbor)
95	{	142	{
96	Zero (cbor, 1, CBOR);	143	Zero (cbor, 1, CBOR);
97	cbor->max_depth = 512;	144	cbor->max_depth = 512;
		145	}
		146
		147	ecb_inline void
		148	cbor_free (CBOR *cbor)
		149	{
		150	SvREFCNT_dec (cbor->filter);
		151	SvREFCNT_dec (cbor->incr_count);
98	}	152	}
99		153
100	/////////////////////////////////////////////////////////////////////////////	154	/////////////////////////////////////////////////////////////////////////////
101	// utility functions	155	// utility functions
102		156
…		…
122	SvPV_shrink_to_cur (sv);	176	SvPV_shrink_to_cur (sv);
123	#elif defined (SvPV_renew)	177	#elif defined (SvPV_renew)
124	SvPV_renew (sv, SvCUR (sv) + 1);	178	SvPV_renew (sv, SvCUR (sv) + 1);
125	#endif	179	#endif
126	}	180	}
		181	}
		182
		183	// minimum length of a string to be registered for stringref
		184	ecb_inline int
		185	minimum_string_length (UV idx)
		186	{
		187	return idx > 23
		188	? idx > 0xffU
		189	? idx > 0xffffU
		190	? idx > 0xffffffffU
		191	? 11
		192	: 7
		193	: 5
		194	: 4
		195	: 3;
127	}	196	}
128		197
129	/////////////////////////////////////////////////////////////////////////////	198	/////////////////////////////////////////////////////////////////////////////
130	// encoder	199	// encoder
131		200
…		…
165	static void	234	static void
166	encode_uint (enc_t *enc, int major, UV len)	235	encode_uint (enc_t *enc, int major, UV len)
167	{	236	{
168	need (enc, 9);	237	need (enc, 9);
169		238
170	if (len < 24)	239	if (ecb_expect_true (len < LENGTH_EXT1))
171	*enc->cur++ = major \| len;	240	*enc->cur++ = major \| len;
172	else if (len <= 0xff)	241	else if (ecb_expect_true (len <= 0xffU))
173	{	242	{
174	*enc->cur++ = major \| 24;	243	*enc->cur++ = major \| LENGTH_EXT1;
175	*enc->cur++ = len;	244	*enc->cur++ = len;
176	}	245	}
177	else if (len <= 0xffff)	246	else if (len <= 0xffffU)
178	{	247	{
179	*enc->cur++ = major \| 25;	248	*enc->cur++ = major \| LENGTH_EXT2;
180	*enc->cur++ = len >> 8;	249	*enc->cur++ = len >> 8;
181	*enc->cur++ = len;	250	*enc->cur++ = len;
182	}	251	}
183	else if (len <= 0xffffffff)	252	else if (len <= 0xffffffffU)
184	{	253	{
185	*enc->cur++ = major \| 26;	254	*enc->cur++ = major \| LENGTH_EXT4;
186	*enc->cur++ = len >> 24;	255	*enc->cur++ = len >> 24;
187	*enc->cur++ = len >> 16;	256	*enc->cur++ = len >> 16;
188	*enc->cur++ = len >> 8;	257	*enc->cur++ = len >> 8;
189	*enc->cur++ = len;	258	*enc->cur++ = len;
190	}	259	}
191	else	260	else
192	{	261	{
193	*enc->cur++ = major \| 27;	262	*enc->cur++ = major \| LENGTH_EXT8;
194	*enc->cur++ = len >> 56;	263	*enc->cur++ = len >> 56;
195	*enc->cur++ = len >> 48;	264	*enc->cur++ = len >> 48;
196	*enc->cur++ = len >> 40;	265	*enc->cur++ = len >> 40;
197	*enc->cur++ = len >> 32;	266	*enc->cur++ = len >> 32;
198	*enc->cur++ = len >> 24;	267	*enc->cur++ = len >> 24;
…		…
200	*enc->cur++ = len >> 8;	269	*enc->cur++ = len >> 8;
201	*enc->cur++ = len;	270	*enc->cur++ = len;
202	}	271	}
203	}	272	}
204		273
205	static void	274	ecb_inline void
		275	encode_tag (enc_t *enc, UV tag)
		276	{
		277	encode_uint (enc, MAJOR_TAG, tag);
		278	}
		279
		280	ecb_inline void
206	encode_str (enc_t enc, int utf8, char str, STRLEN len)	281	encode_str (enc_t enc, int utf8, char str, STRLEN len)
207	{	282	{
208	encode_uint (enc, utf8 ? 0x60 : 0x40, len);	283	encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
209	need (enc, len);	284	need (enc, len);
210	memcpy (enc->cur, str, len);	285	memcpy (enc->cur, str, len);
211	enc->cur += len;	286	enc->cur += len;
212	}	287	}
213		288
214	ecb_inline void	289	static void
215	encode_tag (enc_t *enc, UV tag)	290	encode_strref (enc_t enc, int utf8, char str, STRLEN len)
216	{	291	{
217	encode_uint (enc, 0xc0, tag);	292	if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
		293	{
		294	SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
		295
		296	if (SvOK (*svp))
		297	{
		298	// already registered, use stringref
		299	encode_tag (enc, CBOR_TAG_STRINGREF);
		300	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		301	return;
		302	}
		303	else if (len >= minimum_string_length (enc->stringref_idx))
		304	{
		305	// register only
		306	sv_setuv (*svp, enc->stringref_idx);
		307	++enc->stringref_idx;
		308	}
		309	}
		310
		311	encode_str (enc, utf8, str, len);
218	}	312	}
219		313
220	static void encode_sv (enc_t enc, SV sv);	314	static void encode_sv (enc_t enc, SV sv);
221		315
222	static void	316	static void
…		…
227	if (enc->depth >= enc->cbor.max_depth)	321	if (enc->depth >= enc->cbor.max_depth)
228	croak (ERR_NESTING_EXCEEDED);	322	croak (ERR_NESTING_EXCEEDED);
229		323
230	++enc->depth;	324	++enc->depth;
231		325
232	encode_uint (enc, 0x80, len + 1);	326	encode_uint (enc, MAJOR_ARRAY, len + 1);
233		327
234	for (i = 0; i <= len; ++i)	328	for (i = 0; i <= len; ++i)
235	{	329	{
236	SV **svp = av_fetch (av, i, 0);	330	SV **svp = av_fetch (av, i, 0);
237	encode_sv (enc, svp ? *svp : &PL_sv_undef);	331	encode_sv (enc, svp ? *svp : &PL_sv_undef);
238	}	332	}
239		333
240	--enc->depth;	334	--enc->depth;
241	}	335	}
242		336
243	ecb_inline void
244	encode_he (enc_t enc, HE he)
245	{
246	if (HeKLEN (he) == HEf_SVKEY)
247	encode_sv (enc, HeSVKEY (he));
248	else
249	encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250	}
251
252	static void	337	static void
253	encode_hv (enc_t enc, HV hv)	338	encode_hv (enc_t enc, HV hv)
254	{	339	{
255	HE *he;	340	HE *he;
256		341
…		…
261		346
262	int pairs = hv_iterinit (hv);	347	int pairs = hv_iterinit (hv);
263	int mg = SvMAGICAL (hv);	348	int mg = SvMAGICAL (hv);
264		349
265	if (mg)	350	if (mg)
266	encode_ch (enc, 0xa0 \| 31);	351	encode_ch (enc, MAJOR_MAP \| MINOR_INDEF);
267	else	352	else
268	encode_uint (enc, 0xa0, pairs);	353	encode_uint (enc, MAJOR_MAP, pairs);
269		354
270	while ((he = hv_iternext (hv)))	355	while ((he = hv_iternext (hv)))
271	{	356	{
272	if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS \| F_DEDUP_KEYS)))
273	{
274	SV **svp;
275
276	if (HeKLEN (he) == HEf_SVKEY)	357	if (HeKLEN (he) == HEf_SVKEY)
277	svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/	358	encode_sv (enc, HeSVKEY (he));
278	else
279	svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281	if (SvOK (*svp))
282	{
283	encode_tag (enc, CBOR_TAG_STRINGREF);
284	encode_uint (enc, 0x00, SvUV (*svp));
285	}
286	else
287	{
288	sv_setuv (*svp, enc->stringref_idx);
289	++enc->stringref_idx;
290	encode_he (enc, he);
291	}
292	}
293	else	359	else
294	encode_he (enc, he);	360	encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295		361
296	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));	362	encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
297	}	363	}
298		364
299	if (mg)	365	if (mg)
300	encode_ch (enc, 0xe0 \| 31);	366	encode_ch (enc, MAJOR_MISC \| MINOR_INDEF);
301		367
302	--enc->depth;	368	--enc->depth;
303	}	369	}
304		370
305	// encode objects, arrays and special \0=false and \1=true values.	371	// encode objects, arrays and special \0=false and \1=true values.
306	static void	372	static void
307	encode_rv (enc_t enc, SV sv)	373	encode_rv (enc_t enc, SV sv)
308	{	374	{
309	SvGETMAGIC (sv);	375	SvGETMAGIC (sv);
310
311	if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312	&& ecb_expect_false (SvREFCNT (sv) > 1))
313	{
314	if (!enc->shareable)
315	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
316
317	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
318
319	if (SvOK (*svp))
320	{
321	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322	encode_uint (enc, 0x00, SvUV (*svp));
323	return;
324	}
325	else
326	{
327	sv_setuv (*svp, enc->shareable_idx);
328	++enc->shareable_idx;
329	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330	}
331	}
332		376
333	svtype svt = SvTYPE (sv);	377	svtype svt = SvTYPE (sv);
334		378
335	if (ecb_expect_false (SvOBJECT (sv)))	379	if (ecb_expect_false (SvOBJECT (sv)))
336	{	380	{
…		…
343	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	387	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
344	? cbor_tagged_stash	388	? cbor_tagged_stash
345	: gv_stashpv ("CBOR::XS::Tagged" , 1);	389	: gv_stashpv ("CBOR::XS::Tagged" , 1);
346		390
347	HV *stash = SvSTASH (sv);	391	HV *stash = SvSTASH (sv);
348	GV *method;
349		392
350	if (stash == boolean_stash)	393	if (stash == boolean_stash)
351	encode_ch (enc, SvIV (sv) ? 0xe0 \| 21 : 0xe0 \| 20);	394	{
		395	encode_ch (enc, SvIV (sv) ? MAJOR_MISC \| SIMPLE_TRUE : MAJOR_MISC \| SIMPLE_FALSE);
		396	return;
		397	}
352	else if (stash == error_stash)	398	else if (stash == error_stash)
353	encode_ch (enc, 0xe0 \| 23);	399	{
		400	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
		401	return;
		402	}
354	else if (stash == tagged_stash)	403	else if (stash == tagged_stash)
355	{	404	{
356	if (svt != SVt_PVAV)	405	if (svt != SVt_PVAV)
357	croak ("encountered CBOR::XS::Tagged object that isn't an array");	406	croak ("encountered CBOR::XS::Tagged object that isn't an array");
358		407
359	encode_uint (enc, 0xc0, SvUV (av_fetch ((AV )sv, 0, 1)));	408	encode_uint (enc, MAJOR_TAG, SvUV (av_fetch ((AV )sv, 0, 1)));
360	encode_sv (enc, av_fetch ((AV )sv, 1, 1));	409	encode_sv (enc, av_fetch ((AV )sv, 1, 1));
		410
		411	return;
		412	}
		413	}
		414
		415	if (ecb_expect_false (SvREFCNT (sv) > 1)
		416	&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
		417	{
		418	if (!enc->shareable)
		419	enc->shareable = (HV )sv_2mortal ((SV )newHV ());
		420
		421	SV *svp = hv_fetch (enc->shareable, (char )&sv, sizeof (sv), 1);
		422
		423	if (SvOK (*svp))
361	}	424	{
		425	encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
		426	encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
		427	return;
		428	}
		429	else
		430	{
		431	sv_setuv (*svp, enc->shareable_idx);
		432	++enc->shareable_idx;
		433	encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
		434	}
		435	}
		436
		437	if (ecb_expect_false (SvOBJECT (sv)))
		438	{
		439	HV *stash = SvSTASH (sv);
		440	GV *method;
		441
362	else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))	442	if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363	{	443	{
364	dSP;	444	dSP;
365		445
366	ENTER; SAVETMPS; PUSHMARK (SP);	446	ENTER; SAVETMPS; PUSHMARK (SP);
367	// we re-bless the reference to get overload and other niceties right	447	// we re-bless the reference to get overload and other niceties right
…		…
399	// catch this surprisingly common error	479	// catch this surprisingly common error
400	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)	480	if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));	481	croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402		482
403	encode_tag (enc, CBOR_TAG_PERL_OBJECT);	483	encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404	encode_uint (enc, 0x80, count + 1);	484	encode_uint (enc, MAJOR_ARRAY, count + 1);
405	encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));	485	encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406		486
407	while (count)	487	while (count)
408	encode_sv (enc, SP[1 - count--]);	488	encode_sv (enc, SP[1 - count--]);
409		489
410	PUTBACK;	490	PUTBACK;
…		…
431	{	511	{
432	double nv = SvNVX (sv);	512	double nv = SvNVX (sv);
433		513
434	need (enc, 9);	514	need (enc, 9);
435		515
436	if (ecb_expect_false (nv == (U32)nv))	516	if (ecb_expect_false (nv == (NV)(U32)nv))
437	encode_uint (enc, 0x00, (U32)nv);	517	encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438	//TODO: maybe I32?	518	//TODO: maybe I32?
439	else if (ecb_expect_false (nv == (float)nv))	519	else if (ecb_expect_false (nv == (float)nv))
440	{	520	{
441	uint32_t fp = ecb_float_to_binary32 (nv);	521	uint32_t fp = ecb_float_to_binary32 (nv);
442		522
443	*enc->cur++ = 0xe0 \| 26;	523	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT32;
444		524
445	if (!ecb_big_endian ())	525	if (!ecb_big_endian ())
446	fp = ecb_bswap32 (fp);	526	fp = ecb_bswap32 (fp);
447		527
448	memcpy (enc->cur, &fp, 4);	528	memcpy (enc->cur, &fp, 4);
…		…
450	}	530	}
451	else	531	else
452	{	532	{
453	uint64_t fp = ecb_double_to_binary64 (nv);	533	uint64_t fp = ecb_double_to_binary64 (nv);
454		534
455	*enc->cur++ = 0xe0 \| 27;	535	*enc->cur++ = MAJOR_MISC \| MISC_FLOAT64;
456		536
457	if (!ecb_big_endian ())	537	if (!ecb_big_endian ())
458	fp = ecb_bswap64 (fp);	538	fp = ecb_bswap64 (fp);
459		539
460	memcpy (enc->cur, &fp, 8);	540	memcpy (enc->cur, &fp, 8);
…		…
469		549
470	if (SvPOKp (sv))	550	if (SvPOKp (sv))
471	{	551	{
472	STRLEN len;	552	STRLEN len;
473	char *str = SvPV (sv, len);	553	char *str = SvPV (sv, len);
474	encode_str (enc, SvUTF8 (sv), str, len);	554	encode_strref (enc, SvUTF8 (sv), str, len);
475	}	555	}
476	else if (SvNOKp (sv))	556	else if (SvNOKp (sv))
477	encode_nv (enc, sv);	557	encode_nv (enc, sv);
478	else if (SvIOKp (sv))	558	else if (SvIOKp (sv))
479	{	559	{
480	if (SvIsUV (sv))	560	if (SvIsUV (sv))
481	encode_uint (enc, 0x00, SvUVX (sv));	561	encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
482	else if (SvIVX (sv) >= 0)	562	else if (SvIVX (sv) >= 0)
483	encode_uint (enc, 0x00, SvIVX (sv));	563	encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
484	else	564	else
485	encode_uint (enc, 0x20, -(SvIVX (sv) + 1));	565	encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
486	}	566	}
487	else if (SvROK (sv))	567	else if (SvROK (sv))
488	encode_rv (enc, SvRV (sv));	568	encode_rv (enc, SvRV (sv));
489	else if (!SvOK (sv))	569	else if (!SvOK (sv))
490	encode_ch (enc, 0xe0 \| 22);	570	encode_ch (enc, MAJOR_MISC \| SIMPLE_NULL);
491	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)	571	else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492	encode_ch (enc, 0xe0 \| 23);	572	encode_ch (enc, MAJOR_MISC \| SIMPLE_UNDEF);
493	else	573	else
494	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",	574	croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
495	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));	575	SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496	}	576	}
497		577
…		…
505	enc.cur = SvPVX (enc.sv);	585	enc.cur = SvPVX (enc.sv);
506	enc.end = SvEND (enc.sv);	586	enc.end = SvEND (enc.sv);
507		587
508	SvPOK_only (enc.sv);	588	SvPOK_only (enc.sv);
509		589
510	if (cbor->flags & (F_DEDUP_STRINGS \| F_DEDUP_KEYS))	590	if (cbor->flags & F_PACK_STRINGS)
511	{	591	{
512	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);	592	encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());	593	enc.stringref[0]= (HV )sv_2mortal ((SV )newHV ());
514	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());	594	enc.stringref[1]= (HV )sv_2mortal ((SV )newHV ());
515	}	595	}
…		…
537	CBOR cbor;	617	CBOR cbor;
538	U32 depth; // recursion depth	618	U32 depth; // recursion depth
539	U32 maxdepth; // recursion depth limit	619	U32 maxdepth; // recursion depth limit
540	AV *shareable;	620	AV *shareable;
541	AV *stringref;	621	AV *stringref;
		622	SV *decode_tagged;
542	} dec_t;	623	} dec_t;
543		624
544	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE	625	#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
545		626
546	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")	627	#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
…		…
549	#define DEC_DEC_DEPTH --dec->depth	630	#define DEC_DEC_DEPTH --dec->depth
550		631
551	static UV	632	static UV
552	decode_uint (dec_t *dec)	633	decode_uint (dec_t *dec)
553	{	634	{
554	switch (*dec->cur & 31)	635	U8 m = *dec->cur & MINOR_MASK;
555	{	636	++dec->cur;
556	case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557	case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558	case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559	return *dec->cur++ & 31;
560		637
561	case 24:	638	if (ecb_expect_true (m < LENGTH_EXT1))
		639	return m;
		640	else if (ecb_expect_true (m == LENGTH_EXT1))
		641	{
562	WANT (2);	642	WANT (1);
563	dec->cur += 2;	643	dec->cur += 1;
564	return dec->cur[-1];	644	return dec->cur[-1];
565		645	}
566	case 25:	646	else if (ecb_expect_true (m == LENGTH_EXT2))
		647	{
567	WANT (3);	648	WANT (2);
568	dec->cur += 3;	649	dec->cur += 2;
569	return (((UV)dec->cur[-2]) << 8)	650	return (((UV)dec->cur[-2]) << 8)
570	\| ((UV)dec->cur[-1]);	651	\| ((UV)dec->cur[-1]);
571		652	}
572	case 26:	653	else if (ecb_expect_true (m == LENGTH_EXT4))
		654	{
573	WANT (5);	655	WANT (4);
574	dec->cur += 5;	656	dec->cur += 4;
575	return (((UV)dec->cur[-4]) << 24)	657	return (((UV)dec->cur[-4]) << 24)
576	\| (((UV)dec->cur[-3]) << 16)	658	\| (((UV)dec->cur[-3]) << 16)
577	\| (((UV)dec->cur[-2]) << 8)	659	\| (((UV)dec->cur[-2]) << 8)
578	\| ((UV)dec->cur[-1]);	660	\| ((UV)dec->cur[-1]);
579		661	}
580	case 27:	662	else if (ecb_expect_true (m == LENGTH_EXT8))
		663	{
581	WANT (9);	664	WANT (8);
582	dec->cur += 9;	665	dec->cur += 8;
		666
		667	return
		668	#if UVSIZE < 8
		669	0
		670	#else
583	return (((UV)dec->cur[-8]) << 56)	671	(((UV)dec->cur[-8]) << 56)
584	\| (((UV)dec->cur[-7]) << 48)	672	\| (((UV)dec->cur[-7]) << 48)
585	\| (((UV)dec->cur[-6]) << 40)	673	\| (((UV)dec->cur[-6]) << 40)
586	\| (((UV)dec->cur[-5]) << 32)	674	\| (((UV)dec->cur[-5]) << 32)
		675	#endif
587	\| (((UV)dec->cur[-4]) << 24)	676	\| (((UV)dec->cur[-4]) << 24)
588	\| (((UV)dec->cur[-3]) << 16)	677	\| (((UV)dec->cur[-3]) << 16)
589	\| (((UV)dec->cur[-2]) << 8)	678	\| (((UV)dec->cur[-2]) << 8)
590	\| ((UV)dec->cur[-1]);	679	\| ((UV)dec->cur[-1]);
591		680	}
592	default:	681	else
593	ERR ("corrupted CBOR data (unsupported integer minor encoding)");	682	ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594	}
595		683
596	fail:	684	fail:
597	return 0;	685	return 0;
598	}	686	}
599		687
…		…
604	{	692	{
605	AV *av = newAV ();	693	AV *av = newAV ();
606		694
607	DEC_INC_DEPTH;	695	DEC_INC_DEPTH;
608		696
609	if ((*dec->cur & 31) == 31)	697	if (*dec->cur == (MAJOR_ARRAY \| MINOR_INDEF))
610	{	698	{
611	++dec->cur;	699	++dec->cur;
612		700
613	for (;;)	701	for (;;)
614	{	702	{
615	WANT (1);	703	WANT (1);
616		704
617	if (*dec->cur == (0xe0 \| 31))	705	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
618	{	706	{
619	++dec->cur;	707	++dec->cur;
620	break;	708	break;
621	}	709	}
622		710
…		…
625	}	713	}
626	else	714	else
627	{	715	{
628	int i, len = decode_uint (dec);	716	int i, len = decode_uint (dec);
629		717
		718	WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630	av_fill (av, len - 1);	719	av_fill (av, len - 1);
631		720
632	for (i = 0; i < len; ++i)	721	for (i = 0; i < len; ++i)
633	AvARRAY (av)[i] = decode_sv (dec);	722	AvARRAY (av)[i] = decode_sv (dec);
634	}	723	}
…		…
644		733
645	static void	734	static void
646	decode_he (dec_t dec, HV hv)	735	decode_he (dec_t dec, HV hv)
647	{	736	{
648	// for speed reasons, we specialcase single-string	737	// for speed reasons, we specialcase single-string
649	// byte or utf-8 strings as keys.	738	// byte or utf-8 strings as keys, but only when !stringref
650		739
651	if (dec->cur >= 0x40 && dec->cur <= 0x40 + 27)	740	if (ecb_expect_true (!dec->stringref))
		741	if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652	{	742	{
653	I32 len = decode_uint (dec);	743	I32 len = decode_uint (dec);
654	char key = (char )dec->cur;	744	char key = (char )dec->cur;
655		745
656	dec->cur += len;	746	dec->cur += len;
657		747
658	if (ecb_expect_false (dec->stringref))
659	av_push (dec->stringref, newSVpvn (key, len));
660
661	hv_store (hv, key, len, decode_sv (dec), 0);	748	hv_store (hv, key, len, decode_sv (dec), 0);
		749
		750	return;
662	}	751	}
663	else if (dec->cur >= 0x60 && dec->cur <= 0x60 + 27)	752	else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664	{	753	{
665	I32 len = decode_uint (dec);	754	I32 len = decode_uint (dec);
666	char key = (char )dec->cur;	755	char key = (char )dec->cur;
667		756
668	dec->cur += len;	757	dec->cur += len;
669		758
670	if (ecb_expect_false (dec->stringref))	759	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671	av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));	760	if (!is_utf8_string (key, len))
		761	ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672		762
673	hv_store (hv, key, -len, decode_sv (dec), 0);	763	hv_store (hv, key, -len, decode_sv (dec), 0);
		764
		765	return;
674	}	766	}
675	else	767
676	{
677	SV *k = decode_sv (dec);	768	SV *k = decode_sv (dec);
678	SV *v = decode_sv (dec);	769	SV *v = decode_sv (dec);
679		770
680	hv_store_ent (hv, k, v, 0);	771	hv_store_ent (hv, k, v, 0);
681	SvREFCNT_dec (k);	772	SvREFCNT_dec (k);
682	}	773
		774	fail:
		775	;
683	}	776	}
684		777
685	static SV *	778	static SV *
686	decode_hv (dec_t *dec)	779	decode_hv (dec_t *dec)
687	{	780	{
688	HV *hv = newHV ();	781	HV *hv = newHV ();
689		782
690	DEC_INC_DEPTH;	783	DEC_INC_DEPTH;
691		784
692	if ((*dec->cur & 31) == 31)	785	if (*dec->cur == (MAJOR_MAP \| MINOR_INDEF))
693	{	786	{
694	++dec->cur;	787	++dec->cur;
695		788
696	for (;;)	789	for (;;)
697	{	790	{
698	WANT (1);	791	WANT (1);
699		792
700	if (*dec->cur == (0xe0 \| 31))	793	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
701	{	794	{
702	++dec->cur;	795	++dec->cur;
703	break;	796	break;
704	}	797	}
705		798
…		…
726	static SV *	819	static SV *
727	decode_str (dec_t *dec, int utf8)	820	decode_str (dec_t *dec, int utf8)
728	{	821	{
729	SV *sv = 0;	822	SV *sv = 0;
730		823
731	if ((*dec->cur & 31) == 31)	824	if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732	{	825	{
		826	// indefinite length strings
733	++dec->cur;	827	++dec->cur;
734		828
		829	U8 major = *dec->cur & MAJOR_MISC;
		830
735	sv = newSVpvn ("", 0);	831	sv = newSVpvn ("", 0);
736		832
737	// not very fast, and certainly not robust against illegal input
738	for (;;)	833	for (;;)
739	{	834	{
740	WANT (1);	835	WANT (1);
741		836
742	if (*dec->cur == (0xe0 \| 31))	837	if ((*dec->cur - major) > LENGTH_EXT8)
		838	if (*dec->cur == (MAJOR_MISC \| MINOR_INDEF))
743	{	839	{
744	++dec->cur;	840	++dec->cur;
745	break;	841	break;
746	}	842	}
		843	else
		844	ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747		845
748	sv_catsv (sv, decode_sv (dec));	846	STRLEN len = decode_uint (dec);
		847
		848	WANT (len);
		849	sv_catpvn (sv, dec->cur, len);
		850	dec->cur += len;
749	}	851	}
750	}	852	}
751	else	853	else
752	{	854	{
753	STRLEN len = decode_uint (dec);	855	STRLEN len = decode_uint (dec);
754		856
755	WANT (len);	857	WANT (len);
756	sv = newSVpvn (dec->cur, len);	858	sv = newSVpvn (dec->cur, len);
757	dec->cur += len;	859	dec->cur += len;
		860
		861	if (ecb_expect_false (dec->stringref)
		862	&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
		863	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
758	}	864	}
759		865
760	if (utf8)	866	if (utf8)
		867	{
		868	if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
		869	if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
		870	ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
		871
761	SvUTF8_on (sv);	872	SvUTF8_on (sv);
762		873	}
763	if (ecb_expect_false (dec->stringref))
764	av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765		874
766	return sv;	875	return sv;
767		876
768	fail:	877	fail:
769	SvREFCNT_dec (sv);	878	SvREFCNT_dec (sv);
…		…
801	}	910	}
802	break;	911	break;
803		912
804	case CBOR_TAG_STRINGREF:	913	case CBOR_TAG_STRINGREF:
805	{	914	{
806	if ((*dec->cur >> 5) != 0)	915	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");	916	ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808		917
809	UV idx = decode_uint (dec);	918	UV idx = decode_uint (dec);
810		919
811	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))	920	if (!dec->stringref \|\| (int)idx > AvFILLp (dec->stringref))
…		…
818	case CBOR_TAG_VALUE_SHAREABLE:	927	case CBOR_TAG_VALUE_SHAREABLE:
819	{	928	{
820	if (ecb_expect_false (!dec->shareable))	929	if (ecb_expect_false (!dec->shareable))
821	dec->shareable = (AV )sv_2mortal ((SV )newAV ());	930	dec->shareable = (AV )sv_2mortal ((SV )newAV ());
822		931
		932	if (dec->cbor.flags & F_ALLOW_CYCLES)
		933	{
823	sv = newSV (0);	934	sv = newSV (0);
824	av_push (dec->shareable, SvREFCNT_inc_NN (sv));	935	av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825		936
826	SV *osv = decode_sv (dec);	937	SV *osv = decode_sv (dec);
827	sv_setsv (sv, osv);	938	sv_setsv (sv, osv);
828	SvREFCNT_dec_NN (osv);	939	SvREFCNT_dec_NN (osv);
		940	}
		941	else
		942	{
		943	av_push (dec->shareable, &PL_sv_undef);
		944	int idx = AvFILLp (dec->shareable);
		945	sv = decode_sv (dec);
		946	av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
		947	}
829	}	948	}
830	break;	949	break;
831		950
832	case CBOR_TAG_VALUE_SHAREDREF:	951	case CBOR_TAG_VALUE_SHAREDREF:
833	{	952	{
834	if ((*dec->cur >> 5) != 0)	953	if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");	954	ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836		955
837	UV idx = decode_uint (dec);	956	UV idx = decode_uint (dec);
838		957
839	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))	958	if (!dec->shareable \|\| (int)idx > AvFILLp (dec->shareable))
840	ERR ("corrupted CBOR data (sharedref index out of bounds)");	959	ERR ("corrupted CBOR data (sharedref index out of bounds)");
841		960
842	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);	961	sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
		962
		963	if (sv == &PL_sv_undef)
		964	ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
843	}	965	}
844	break;	966	break;
845		967
846	case CBOR_TAG_PERL_OBJECT:	968	case CBOR_TAG_PERL_OBJECT:
847	{	969	{
…		…
896		1018
897	default:	1019	default:
898	{	1020	{
899	sv = decode_sv (dec);	1021	sv = decode_sv (dec);
900		1022
		1023	dSP;
		1024	ENTER; SAVETMPS; PUSHMARK (SP);
		1025	EXTEND (SP, 2);
		1026	PUSHs (newSVuv (tag));
		1027	PUSHs (sv);
		1028
		1029	PUTBACK;
		1030	int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY \| G_EVAL);
		1031	SPAGAIN;
		1032
		1033	if (SvTRUE (ERRSV))
		1034	{
		1035	FREETMPS; LEAVE;
		1036	ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
		1037	}
		1038
		1039	if (count)
		1040	{
		1041	SvREFCNT_dec (sv);
		1042	sv = SvREFCNT_inc (POPs);
		1043	}
		1044	else
		1045	{
901	AV *av = newAV ();	1046	AV *av = newAV ();
902	av_push (av, newSVuv (tag));	1047	av_push (av, newSVuv (tag));
903	av_push (av, sv);	1048	av_push (av, sv);
904		1049
905	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash	1050	HV *tagged_stash = !CBOR_SLOW \|\| cbor_tagged_stash
906	? cbor_tagged_stash	1051	? cbor_tagged_stash
907	: gv_stashpv ("CBOR::XS::Tagged" , 1);	1052	: gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);	1053	sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
		1054	}
		1055
		1056	PUTBACK;
		1057
		1058	FREETMPS; LEAVE;
910	}	1059	}
911	break;	1060	break;
912	}	1061	}
913		1062
914	return sv;	1063	return sv;
…		…
921	static SV *	1070	static SV *
922	decode_sv (dec_t *dec)	1071	decode_sv (dec_t *dec)
923	{	1072	{
924	WANT (1);	1073	WANT (1);
925		1074
926	switch (*dec->cur >> 5)	1075	switch (*dec->cur >> MAJOR_SHIFT)
927	{	1076	{
928	case 0: // unsigned int	1077	case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929	return newSVuv (decode_uint (dec));	1078	case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930	case 1: // negative int	1079	case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931	return newSViv (-1 - (IV)decode_uint (dec));	1080	case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932	case 2: // octet string	1081	case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933	return decode_str (dec, 0);	1082	case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934	case 3: // utf-8 string	1083	case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935	return decode_str (dec, 1);	1084
936	case 4: // array	1085	case MAJOR_MISC >> MAJOR_SHIFT:
937	return decode_av (dec);
938	case 5: // map
939	return decode_hv (dec);
940	case 6: // tag
941	return decode_tagged (dec);
942	case 7: // misc
943	switch (*dec->cur++ & 31)	1086	switch (*dec->cur++ & MINOR_MASK)
944	{	1087	{
945	case 20:	1088	case SIMPLE_FALSE:
946	#if CBOR_SLOW	1089	#if CBOR_SLOW
947	types_false = get_bool ("Types::Serialiser::false");	1090	types_false = get_bool ("Types::Serialiser::false");
948	#endif	1091	#endif
949	return newSVsv (types_false);	1092	return newSVsv (types_false);
950	case 21:	1093	case SIMPLE_TRUE:
951	#if CBOR_SLOW	1094	#if CBOR_SLOW
952	types_true = get_bool ("Types::Serialiser::true");	1095	types_true = get_bool ("Types::Serialiser::true");
953	#endif	1096	#endif
954	return newSVsv (types_true);	1097	return newSVsv (types_true);
955	case 22:	1098	case SIMPLE_NULL:
956	return newSVsv (&PL_sv_undef);	1099	return newSVsv (&PL_sv_undef);
957	case 23:	1100	case SIMPLE_UNDEF:
958	#if CBOR_SLOW	1101	#if CBOR_SLOW
959	types_error = get_bool ("Types::Serialiser::error");	1102	types_error = get_bool ("Types::Serialiser::error");
960	#endif	1103	#endif
961	return newSVsv (types_error);	1104	return newSVsv (types_error);
962		1105
963	case 25:	1106	case MISC_FLOAT16:
964	{	1107	{
965	WANT (2);	1108	WANT (2);
966		1109
967	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];	1110	uint16_t fp = (dec->cur[0] << 8) \| dec->cur[1];
968	dec->cur += 2;	1111	dec->cur += 2;
969		1112
970	return newSVnv (ecb_binary16_to_float (fp));	1113	return newSVnv (ecb_binary16_to_float (fp));
971	}	1114	}
972		1115
973	case 26:	1116	case MISC_FLOAT32:
974	{	1117	{
975	uint32_t fp;	1118	uint32_t fp;
976	WANT (4);	1119	WANT (4);
977	memcpy (&fp, dec->cur, 4);	1120	memcpy (&fp, dec->cur, 4);
978	dec->cur += 4;	1121	dec->cur += 4;
…		…
981	fp = ecb_bswap32 (fp);	1124	fp = ecb_bswap32 (fp);
982		1125
983	return newSVnv (ecb_binary32_to_float (fp));	1126	return newSVnv (ecb_binary32_to_float (fp));
984	}	1127	}
985		1128
986	case 27:	1129	case MISC_FLOAT64:
987	{	1130	{
988	uint64_t fp;	1131	uint64_t fp;
989	WANT (8);	1132	WANT (8);
990	memcpy (&fp, dec->cur, 8);	1133	memcpy (&fp, dec->cur, 8);
991	dec->cur += 8;	1134	dec->cur += 8;
…		…
994	fp = ecb_bswap64 (fp);	1137	fp = ecb_bswap64 (fp);
995		1138
996	return newSVnv (ecb_binary64_to_double (fp));	1139	return newSVnv (ecb_binary64_to_double (fp));
997	}	1140	}
998		1141
999	// 0..19 unassigned	1142	// 0..19 unassigned simple
1000	// 24 reserved + unassigned (reserved values are not encodable)	1143	// 24 reserved + unassigned simple (reserved values are not encodable)
		1144	// 28-30 unassigned misc
		1145	// 31 break code
1001	default:	1146	default:
1002	ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");	1147	ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003	}	1148	}
1004		1149
1005	break;	1150	break;
1006	}	1151	}
1007		1152
…		…
1034	if (dec.cur != dec.end && !dec.err)	1179	if (dec.cur != dec.end && !dec.err)
1035	dec.err = "garbage after CBOR object";	1180	dec.err = "garbage after CBOR object";
1036		1181
1037	if (dec.err)	1182	if (dec.err)
1038	{	1183	{
		1184	if (dec.shareable)
		1185	{
		1186	// need to break cyclic links, which whould all be in shareable
		1187	int i;
		1188	SV **svp;
		1189
		1190	for (i = av_len (dec.shareable) + 1; i--; )
		1191	if ((svp = av_fetch (dec.shareable, i, 0)))
		1192	sv_setsv (*svp, &PL_sv_undef);
		1193	}
		1194
1039	SvREFCNT_dec (sv);	1195	SvREFCNT_dec (sv);
1040	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);	1196	croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 )data, (int)(uint8_t)dec.cur);
1041	}	1197	}
1042		1198
1043	sv = sv_2mortal (sv);	1199	sv = sv_2mortal (sv);
1044		1200
1045	return sv;	1201	return sv;
1046	}	1202	}
1047		1203
		1204	/////////////////////////////////////////////////////////////////////////////
		1205	// incremental parser
		1206
		1207	#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
		1208
		1209	// returns 0 for notyet, 1 for success or error
		1210	static int
		1211	incr_parse (CBOR self, SV cborstr)
		1212	{
		1213	STRLEN cur;
		1214	SvPV (cborstr, cur);
		1215
		1216	while (ecb_expect_true (self->incr_need <= cur))
		1217	{
		1218	// table of integer count bytes
		1219	static I8 incr_len[MINOR_MASK + 1] = {
		1220	0, 0, 0, 0, 0, 0, 0, 0,
		1221	0, 0, 0, 0, 0, 0, 0, 0,
		1222	0, 0, 0, 0, 0, 0, 0, 0,
		1223	1, 2, 4, 8,-1,-1,-1,-2
		1224	};
		1225
		1226	const U8 *p = SvPVX (cborstr) + self->incr_pos;
		1227	U8 m = *p & MINOR_MASK;
		1228	IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1229	I8 ilen = incr_len[m];
		1230
		1231	self->incr_need = self->incr_pos + 1;
		1232
		1233	if (ecb_expect_false (ilen < 0))
		1234	{
		1235	if (m != MINOR_INDEF)
		1236	return 1; // error
		1237
		1238	if (*p == (MAJOR_MISC \| MINOR_INDEF))
		1239	{
		1240	if (count >= 0)
		1241	return 1; // error
		1242
		1243	count = 1;
		1244	}
		1245	else
		1246	{
		1247	av_push (self->incr_count, newSViv (-1)); //TODO: nest
		1248	count = -1;
		1249	}
		1250	}
		1251	else
		1252	{
		1253	self->incr_need += ilen;
		1254	if (ecb_expect_false (self->incr_need > cur))
		1255	return 0;
		1256
		1257	int major = *p >> MAJOR_SHIFT;
		1258
		1259	switch (major)
		1260	{
		1261	case MAJOR_BYTES >> MAJOR_SHIFT:
		1262	case MAJOR_TEXT >> MAJOR_SHIFT:
		1263	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1264	case MAJOR_MAP >> MAJOR_SHIFT:
		1265	{
		1266	UV len;
		1267
		1268	if (ecb_expect_false (ilen))
		1269	{
		1270	len = 0;
		1271
		1272	do {
		1273	len = (len << 8) \| *++p;
		1274	} while (--ilen);
		1275	}
		1276	else
		1277	len = m;
		1278
		1279	switch (major)
		1280	{
		1281	case MAJOR_BYTES >> MAJOR_SHIFT:
		1282	case MAJOR_TEXT >> MAJOR_SHIFT:
		1283	self->incr_need += len;
		1284	if (ecb_expect_false (self->incr_need > cur))
		1285	return 0;
		1286
		1287	break;
		1288
		1289	case MAJOR_MAP >> MAJOR_SHIFT:
		1290	len <<= 1;
		1291	case MAJOR_ARRAY >> MAJOR_SHIFT:
		1292	if (len)
		1293	{
		1294	av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
		1295	count = len + 1;
		1296	}
		1297	break;
		1298	}
		1299	}
		1300	}
		1301	}
		1302
		1303	self->incr_pos = self->incr_need;
		1304
		1305	if (count > 0)
		1306	{
		1307	while (!--count)
		1308	{
		1309	if (!AvFILLp (self->incr_count))
		1310	return 1; // done
		1311
		1312	SvREFCNT_dec_NN (av_pop (self->incr_count));
		1313	count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
		1314	}
		1315
		1316	SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
		1317	}
		1318	}
		1319
		1320	return 0;
		1321	}
		1322
		1323
1048	/////////////////////////////////////////////////////////////////////////////	1324	/////////////////////////////////////////////////////////////////////////////
1049	// XS interface functions	1325	// XS interface functions
1050		1326
1051	MODULE = CBOR::XS PACKAGE = CBOR::XS	1327	MODULE = CBOR::XS PACKAGE = CBOR::XS
1052		1328
…		…
1059	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);	1335	types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1060		1336
1061	types_true = get_bool ("Types::Serialiser::true" );	1337	types_true = get_bool ("Types::Serialiser::true" );
1062	types_false = get_bool ("Types::Serialiser::false");	1338	types_false = get_bool ("Types::Serialiser::false");
1063	types_error = get_bool ("Types::Serialiser::error");	1339	types_error = get_bool ("Types::Serialiser::error");
		1340
		1341	default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1064		1342
1065	sv_cbor = newSVpv ("CBOR", 0);	1343	sv_cbor = newSVpv ("CBOR", 0);
1066	SvREADONLY_on (sv_cbor);	1344	SvREADONLY_on (sv_cbor);
1067	}	1345	}
1068		1346
…		…
1090	void shrink (CBOR *self, int enable = 1)	1368	void shrink (CBOR *self, int enable = 1)
1091	ALIAS:	1369	ALIAS:
1092	shrink = F_SHRINK	1370	shrink = F_SHRINK
1093	allow_unknown = F_ALLOW_UNKNOWN	1371	allow_unknown = F_ALLOW_UNKNOWN
1094	allow_sharing = F_ALLOW_SHARING	1372	allow_sharing = F_ALLOW_SHARING
1095	dedup_keys = F_DEDUP_KEYS	1373	allow_cycles = F_ALLOW_CYCLES
1096	dedup_strings = F_DEDUP_STRINGS	1374	pack_strings = F_PACK_STRINGS
		1375	validate_utf8 = F_VALIDATE_UTF8
1097	PPCODE:	1376	PPCODE:
1098	{	1377	{
1099	if (enable)	1378	if (enable)
1100	self->flags \|= ix;	1379	self->flags \|= ix;
1101	else	1380	else
…		…
1107	void get_shrink (CBOR *self)	1386	void get_shrink (CBOR *self)
1108	ALIAS:	1387	ALIAS:
1109	get_shrink = F_SHRINK	1388	get_shrink = F_SHRINK
1110	get_allow_unknown = F_ALLOW_UNKNOWN	1389	get_allow_unknown = F_ALLOW_UNKNOWN
1111	get_allow_sharing = F_ALLOW_SHARING	1390	get_allow_sharing = F_ALLOW_SHARING
1112	get_dedup_keys = F_DEDUP_KEYS	1391	get_allow_cycles = F_ALLOW_CYCLES
1113	get_dedup_strings = F_DEDUP_STRINGS	1392	get_pack_strings = F_PACK_STRINGS
		1393	get_validate_utf8 = F_VALIDATE_UTF8
1114	PPCODE:	1394	PPCODE:
1115	XPUSHs (boolSV (self->flags & ix));	1395	XPUSHs (boolSV (self->flags & ix));
1116		1396
1117	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)	1397	void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118	PPCODE:	1398	PPCODE:
…		…
1131	XPUSHs (ST (0));	1411	XPUSHs (ST (0));
1132		1412
1133	int get_max_size (CBOR *self)	1413	int get_max_size (CBOR *self)
1134	CODE:	1414	CODE:
1135	RETVAL = self->max_size;	1415	RETVAL = self->max_size;
		1416	OUTPUT:
		1417	RETVAL
		1418
		1419	void filter (CBOR self, SV filter = 0)
		1420	PPCODE:
		1421	SvREFCNT_dec (self->filter);
		1422	self->filter = filter ? newSVsv (filter) : filter;
		1423	XPUSHs (ST (0));
		1424
		1425	SV get_filter (CBOR self)
		1426	CODE:
		1427	RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136	OUTPUT:	1428	OUTPUT:
1137	RETVAL	1429	RETVAL
1138		1430
1139	void encode (CBOR self, SV scalar)	1431	void encode (CBOR self, SV scalar)
1140	PPCODE:	1432	PPCODE:
…		…
1155	EXTEND (SP, 2);	1447	EXTEND (SP, 2);
1156	PUSHs (sv);	1448	PUSHs (sv);
1157	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));	1449	PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158	}	1450	}
1159		1451
		1452	void incr_parse (CBOR self, SV cborstr)
		1453	ALIAS:
		1454	incr_parse_multiple = 1
		1455	PPCODE:
		1456	{
		1457	if (SvUTF8 (cborstr))
		1458	sv_utf8_downgrade (cborstr, 0);
		1459
		1460	if (!self->incr_count)
		1461	{
		1462	self->incr_count = newAV ();
		1463	self->incr_pos = 0;
		1464	self->incr_need = 1;
		1465
		1466	av_push (self->incr_count, newSViv (1));
		1467	}
		1468
		1469	do
		1470	{
		1471	if (!incr_parse (self, cborstr))
		1472	{
		1473	if (self->incr_need > self->max_size && self->max_size)
		1474	croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
		1475	(unsigned long)self->incr_need, (unsigned long)self->max_size);
		1476
		1477	break;
		1478	}
		1479
		1480	SV *sv;
		1481	char *offset;
		1482
		1483	PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
		1484	XPUSHs (sv);
		1485
		1486	sv_chop (cborstr, offset);
		1487
		1488	av_clear (self->incr_count);
		1489	av_push (self->incr_count, newSViv (1));
		1490
		1491	self->incr_pos = 0;
		1492	self->incr_need = self->incr_pos + 1;
		1493	}
		1494	while (ix);
		1495	}
		1496
		1497	void incr_reset (CBOR *self)
		1498	CODE:
		1499	{
		1500	SvREFCNT_dec (self->incr_count);
		1501	self->incr_count = 0;
		1502	}
		1503
		1504	void DESTROY (CBOR *self)
		1505	PPCODE:
		1506	cbor_free (self);
		1507
1160	PROTOTYPES: ENABLE	1508	PROTOTYPES: ENABLE
1161		1509
1162	void encode_cbor (SV *scalar)	1510	void encode_cbor (SV *scalar)
		1511	ALIAS:
		1512	encode_cbor = 0
		1513	encode_cbor_sharing = F_ALLOW_SHARING
1163	PPCODE:	1514	PPCODE:
1164	{	1515	{
1165	CBOR cbor;	1516	CBOR cbor;
1166	cbor_init (&cbor);	1517	cbor_init (&cbor);
		1518	cbor.flags \|= ix;
1167	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;	1519	PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168	XPUSHs (scalar);	1520	XPUSHs (scalar);
1169	}	1521	}
1170		1522
1171	void decode_cbor (SV *cborstr)	1523	void decode_cbor (SV *cborstr)

Diff Legend

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

Comparing CBOR-XS/XS.xs (file contents): Revision 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs. Revision 1.42 by root, Tue Dec 10 15:31:40 2013 UTC

Diff Legend

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.42 by root, Tue Dec 10 15:31:40 2013 UTC