ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.4 by root, Sat Oct 26 21:06:37 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
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
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
30enum 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};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
16 107
17#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
18 109
19#define SB do { 110#define SB do {
20#define SE } while (0) 111#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 112
33#define IN_RANGE_INC(type,val,beg,end) \ 113#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 114 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 115 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 116
42#else 122#else
43# define CBOR_SLOW 0 123# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
45#endif 125#endif
46 126
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 129
50typedef struct { 130typedef struct {
51 U32 flags; 131 U32 flags;
52 U32 max_depth; 132 U32 max_depth;
53 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
54 135
55 SV *cb_object; 136 // for the incremental parser
56 HV *cb_sk_object; 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.
57} CBOR; 140} CBOR;
58 141
59INLINE void 142ecb_inline void
60cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
61{ 144{
62 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 146 cbor->max_depth = 512;
64} 147}
65 148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
154}
155
66///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
67// utility functions 157// utility functions
68 158
69INLINE SV * 159ecb_inline SV *
70get_bool (const char *name) 160get_bool (const char *name)
71{ 161{
72 SV *sv = get_sv (name, 1); 162 SV *sv = get_sv (name, 1);
73 163
74 SvREADONLY_on (sv); 164 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 165 SvREADONLY_on (SvRV (sv));
76 166
77 return sv; 167 return sv;
78} 168}
79 169
80INLINE void 170ecb_inline void
81shrink (SV *sv) 171shrink (SV *sv)
82{ 172{
83 sv_utf8_downgrade (sv, 1); 173 sv_utf8_downgrade (sv, 1);
84 174
85 if (SvLEN (sv) > SvCUR (sv) + 1) 175 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 181#endif
92 } 182 }
93} 183}
94 184
95///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
96// fp hell 186ecb_inline int
97 187minimum_string_length (UV idx)
98//TODO 188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
99 199
100///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
101// encoder 201// encoder
102 202
103// structure used for encoding CBOR 203// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
108 SV *sv; // result scalar 208 SV *sv; // result scalar
109 CBOR cbor; 209 CBOR cbor;
110 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
111} enc_t; 215} enc_t;
112 216
113INLINE void 217ecb_inline void
114need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
115{ 219{
116 if (expect_false (enc->cur + len >= enc->end)) 220 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 221 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 224 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 226 }
123} 227}
124 228
125INLINE void 229ecb_inline void
126encode_ch (enc_t *enc, char ch) 230encode_ch (enc_t *enc, char ch)
127{ 231{
128 need (enc, 1); 232 need (enc, 1);
129 *enc->cur++ = ch; 233 *enc->cur++ = ch;
130} 234}
132static void 236static void
133encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
134{ 238{
135 need (enc, 9); 239 need (enc, 9);
136 240
137 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
139 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
140 { 244 {
141 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 246 *enc->cur++ = len;
143 } 247 }
144 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
145 { 249 {
146 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 252 *enc->cur++ = len;
149 } 253 }
150 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
151 { 255 {
152 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 260 *enc->cur++ = len;
157 } 261 }
158 else 262 else
159 { 263 {
160 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 272 *enc->cur++ = len;
169 } 273 }
170} 274}
171 275
172static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 284{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 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);
176 need (enc, len); 299 need (enc, len);
177 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
178 enc->cur += len; 301 enc->cur += len;
179} 302}
180 303
304static void
305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
306{
307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310
311 if (SvOK (*svp))
312 {
313 // already registered, use stringref
314 encode_tag (enc, CBOR_TAG_STRINGREF);
315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
316 return;
317 }
318 else if (len >= minimum_string_length (enc->stringref_idx))
319 {
320 // register only
321 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx;
323 }
324 }
325
326 encode_str (enc, utf8, str, len);
327}
328
181static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
182 330
183static void 331static void
184encode_av (enc_t *enc, AV *av) 332encode_av (enc_t *enc, AV *av)
185{ 333{
188 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
190 338
191 ++enc->depth; 339 ++enc->depth;
192 340
193 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 342
343 if (SvMAGICAL (av))
195 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
196 { 345 {
197 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 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 }
200 355
201 --enc->depth; 356 --enc->depth;
202} 357}
203 358
204static void 359static void
213 368
214 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
216 371
217 if (mg) 372 if (mg)
218 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 374 else
220 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
221 376
222 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
223 { 378 {
224 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
226 else 381 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 383
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 385 }
231 386
232 if (mg) 387 if (mg)
233 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 389
235 --enc->depth; 390 --enc->depth;
236} 391}
237 392
238// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
239static void 394static void
240encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
241{ 396{
242 svtype svt;
243
244 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
398
245 svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
246 400
247 if (expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
248 { 402 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 403 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 404 ? types_boolean_stash
405 : gv_stashpv ("Types::Serialiser::Boolean", 1);
406 HV *error_stash = !CBOR_SLOW || types_error_stash
407 ? types_error_stash
408 : gv_stashpv ("Types::Serialiser::Error", 1);
409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
410 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 412
253 if (SvSTASH (sv) == stash) 413 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 414
415 if (stash == boolean_stash)
416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
420 else if (stash == error_stash)
421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
425 else if (stash == tagged_stash)
426 {
427 if (svt != SVt_PVAV)
428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
429
430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 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))
446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
450 }
255 else 451 else
256 { 452 {
257#if 0 //TODO 453 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 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
464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 465 {
466 dSP;
467
468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
260 // 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
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 472
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 473 PUTBACK;
474 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 475 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 476 SPAGAIN;
274 477
275 // catch this surprisingly common error 478 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 479 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 480 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 481
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 482 encode_sv (enc, POPs);
283 483
484 PUTBACK;
485
284 FREETMPS; LEAVE; 486 FREETMPS; LEAVE;
285 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED)
287 encode_str (enc, "null", 4, 0);
288 else
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
290 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291 } 487 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 489 {
490 dSP;
491
492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
495 EXTEND (SP, 2);
496 // we re-bless the reference to get overload and other niceties right
497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
498 PUSHs (sv_cbor);
499
500 PUTBACK;
501 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
502 SPAGAIN;
503
504 // catch this surprisingly common error
505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
507
508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
509 encode_uint (enc, MAJOR_ARRAY, count + 1);
510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
511
512 while (count)
513 encode_sv (enc, SP[1 - count--]);
514
515 PUTBACK;
516
517 FREETMPS; LEAVE;
518 }
294 else 519 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 520 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 522 }
300 else if (svt == SVt_PVHV) 523 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
304 else if (svt < SVt_PVAV)
305 {
306 STRLEN len = 0;
307 char *pv = svt ? SvPV (sv, len) : 0;
308
309 if (len == 1 && *pv == '1')
310 encode_ch (enc, 0xe0 | 21);
311 else if (len == 1 && *pv == '0')
312 encode_ch (enc, 0xe0 | 20);
313 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314 encode_ch (enc, 0xe0 | 23);
315 else
316 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318 }
319 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320 encode_ch (enc, 0xe0 | 23);
321 else 527 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 528 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 529 encode_tag (enc, CBOR_TAG_INDIRECTION);
530 encode_sv (enc, sv);
531 }
324} 532}
325 533
326static void 534static void
327encode_nv (enc_t *enc, SV *sv) 535encode_nv (enc_t *enc, SV *sv)
328{ 536{
329 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
330 538
331 need (enc, 9); 539 need (enc, 9);
332 540
333 if (expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 543 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
337 { 545 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
339 547
340 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341 549
342 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
344 552
345 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
347 } 555 }
348 else 556 else
349 { 557 {
350 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
351 559
352 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353 561
354 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
356 564
357 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
366 574
367 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
368 { 576 {
369 STRLEN len; 577 STRLEN len;
370 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
372 } 580 }
373 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 582 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
376 { 584 {
377 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 589 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 591 }
384 else if (SvROK (sv)) 592 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 598 else
391 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",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 601}
394 602
395static SV * 603static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
397{ 605{
398 enc_t enc; 606 enc_t enc = { 0 };
399 607
400 enc.cbor = *cbor; 608 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 612
406 SvPOK_only (enc.sv); 613 SvPOK_only (enc.sv);
614
615 if (cbor->flags & F_PACK_STRINGS)
616 {
617 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
618 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
619 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
620 }
621
407 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
408 623
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 625 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 626
425 U8 *end; // end of input string 640 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
427 CBOR cbor; 642 CBOR cbor;
428 U32 depth; // recursion depth 643 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
430} dec_t; 648} dec_t;
431 649
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
433 651
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
435 653
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 654#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
438 656
439static UV 657static UV
440decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
441{ 659{
442 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
443 { 661 ++dec->cur;
444 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447 return *dec->cur++ & 31;
448 662
449 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
450 WANT (2); 667 WANT (1);
451 dec->cur += 2; 668 dec->cur += 1;
452 return dec->cur[-1]; 669 return dec->cur[-1];
453 670 }
454 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
455 WANT (3); 673 WANT (2);
456 dec->cur += 3; 674 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
459 677 }
460 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
461 WANT (5); 680 WANT (4);
462 dec->cur += 5; 681 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
467 686 }
468 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
469 WANT (9); 689 WANT (8);
470 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
471 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
475 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
479 705 }
480 default: 706 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 708
484fail: 709fail:
485 return 0; 710 return 0;
486} 711}
487 712
492{ 717{
493 AV *av = newAV (); 718 AV *av = newAV ();
494 719
495 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
496 721
497 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 723 {
499 ++dec->cur; 724 ++dec->cur;
500 725
501 for (;;) 726 for (;;)
502 { 727 {
503 WANT (1); 728 WANT (1);
504 729
505 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506 { 731 {
507 ++dec->cur; 732 ++dec->cur;
508 break; 733 break;
509 } 734 }
510 735
513 } 738 }
514 else 739 else
515 { 740 {
516 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
517 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518 av_fill (av, len - 1); 744 av_fill (av, len - 1);
519 745
520 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
522 } 748 }
528 SvREFCNT_dec (av); 754 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 755 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 756 return &PL_sv_undef;
531} 757}
532 758
759static void
760decode_he (dec_t *dec, HV *hv)
761{
762 // for speed reasons, we specialcase single-string
763 // byte or utf-8 strings as keys, but only when !stringref
764
765 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
767 {
768 I32 len = decode_uint (dec);
769 char *key = (char *)dec->cur;
770
771 WANT (len);
772 dec->cur += len;
773
774 hv_store (hv, key, len, decode_sv (dec), 0);
775
776 return;
777 }
778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
779 {
780 I32 len = decode_uint (dec);
781 char *key = (char *)dec->cur;
782
783 WANT (len);
784 dec->cur += len;
785
786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
789
790 hv_store (hv, key, -len, decode_sv (dec), 0);
791
792 return;
793 }
794
795 SV *k = decode_sv (dec);
796 SV *v = decode_sv (dec);
797
798 hv_store_ent (hv, k, v, 0);
799 SvREFCNT_dec (k);
800
801fail:
802 ;
803}
804
533static SV * 805static SV *
534decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
535{ 807{
536 HV *hv = newHV (); 808 HV *hv = newHV ();
537 809
538 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
539 811
540 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 813 {
542 ++dec->cur; 814 ++dec->cur;
543 815
544 for (;;) 816 for (;;)
545 { 817 {
546 WANT (1); 818 WANT (1);
547 819
548 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549 { 821 {
550 ++dec->cur; 822 ++dec->cur;
551 break; 823 break;
552 } 824 }
553 825
554 SV *k = decode_sv (dec); 826 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 827 }
559 } 828 }
560 else 829 else
561 { 830 {
562 int len = decode_uint (dec); 831 int pairs = decode_uint (dec);
563 832
564 while (len--) 833 while (pairs--)
565 { 834 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 835 }
572 836
573 DEC_DEC_DEPTH; 837 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 838 return newRV_noinc ((SV *)hv);
575
576#if 0
577 SV *sv;
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH;
581 decode_ws (dec);
582
583 for (;;)
584 {
585 // heuristic: assume that
586 // a) decode_str + hv_store_ent are abysmally slow.
587 // b) most hash keys are short, simple ascii text.
588 // => try to "fast-match" such strings to avoid
589 // the overhead of decode_str + hv_store_ent.
590 {
591 SV *value;
592 char *p = dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594
595 for (;;)
596 {
597 // the >= 0x80 is false on most architectures
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599 {
600 // slow path, back up and use decode_str
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break;
619 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 {
685 dSP;
686 int count;
687
688 ENTER; SAVETMPS; PUSHMARK (SP);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730 839
731fail: 840fail:
732 SvREFCNT_dec (hv); 841 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 842 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 843 return &PL_sv_undef;
735} 844}
736 845
737static SV * 846static SV *
738decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
739{ 848{
740 SV *sv; 849 SV *sv = 0;
741 850
742 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743 { 852 {
853 // indefinite length strings
744 ++dec->cur; 854 ++dec->cur;
745 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
746 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
747 859
748 // not very fast, and certainly not robust against illegal input
749 for (;;) 860 for (;;)
750 { 861 {
751 WANT (1); 862 WANT (1);
752 863
753 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754 { 866 {
755 ++dec->cur; 867 ++dec->cur;
756 break; 868 break;
757 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758 872
759 SV *sv2 = decode_sv (dec); 873 STRLEN len = decode_uint (dec);
760 sv_catsv (sv, sv2); 874
875 WANT (len);
876 sv_catpvn (sv, dec->cur, len);
877 dec->cur += len;
761 } 878 }
762 } 879 }
763 else 880 else
764 { 881 {
765 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
766 883
767 WANT (len); 884 WANT (len);
768 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 886 dec->cur += len;
887
888 if (ecb_expect_false (dec->stringref)
889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 891 }
771 892
772 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
773 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
774 901
775 return sv; 902 return sv;
776 903
777fail: 904fail:
905 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 906 return &PL_sv_undef;
779} 907}
780 908
781static SV * 909static SV *
782decode_tagged (dec_t *dec) 910decode_tagged (dec_t *dec)
783{ 911{
912 SV *sv = 0;
784 UV tag = decode_uint (dec); 913 UV tag = decode_uint (dec);
914
915 WANT (1);
916
917 switch (tag)
918 {
919 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 920 sv = decode_sv (dec);
921 break;
786 922
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 923 case CBOR_TAG_INDIRECTION:
924 sv = newRV_noinc (decode_sv (dec));
925 break;
926
927 case CBOR_TAG_STRINGREF_NAMESPACE:
928 {
929 ENTER; SAVETMPS;
930
931 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933
934 sv = decode_sv (dec);
935
936 FREETMPS; LEAVE;
937 }
938 break;
939
940 case CBOR_TAG_STRINGREF:
941 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944
945 UV idx = decode_uint (dec);
946
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 }
952 break;
953
954 case CBOR_TAG_VALUE_SHAREABLE:
955 {
956 if (ecb_expect_false (!dec->shareable))
957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
961 sv = newSV (0);
962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
963
964 SV *osv = decode_sv (dec);
965 sv_setsv (sv, 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 }
975 }
976 break;
977
978 case CBOR_TAG_VALUE_SHAREDREF:
979 {
980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
982
983 UV idx = decode_uint (dec);
984
985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
987
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");
992 }
993 break;
994
995 case CBOR_TAG_PERL_OBJECT:
996 {
997 sv = decode_sv (dec);
998
999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1000 ERR ("corrupted CBOR data (non-array perl object)");
1001
1002 AV *av = (AV *)SvRV (sv);
1003 int len = av_len (av) + 1;
1004 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1005
1006 if (!stash)
1007 ERR ("cannot decode perl-object (package does not exist)");
1008
1009 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1010
1011 if (!method)
1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
1013
1014 dSP;
1015
1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
1018 EXTEND (SP, len + 1);
1019 // we re-bless the reference to get overload and other niceties right
1020 PUSHs (*av_fetch (av, 0, 1));
1021 PUSHs (sv_cbor);
1022
1023 int i;
1024
1025 for (i = 1; i < len; ++i)
1026 PUSHs (*av_fetch (av, i, 1));
1027
1028 PUTBACK;
1029 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1030 SPAGAIN;
1031
1032 if (SvTRUE (ERRSV))
1033 {
1034 FREETMPS; LEAVE;
1035 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1036 }
1037
1038 SvREFCNT_dec (sv);
1039 sv = SvREFCNT_inc (POPs);
1040
1041 PUTBACK;
1042
1043 FREETMPS; LEAVE;
1044 }
1045 break;
1046
1047 default:
1048 {
1049 sv = decode_sv (dec);
1050
1051 dSP;
1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
1055 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag));
1057 PUSHs (sv);
1058
1059 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN;
1062
1063 if (SvTRUE (ERRSV))
1064 {
1065 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1067 }
1068
1069 if (count)
1070 {
1071 SvREFCNT_dec (sv);
1072 sv = SvREFCNT_inc (POPs);
1073 }
1074 else
1075 {
1076 AV *av = newAV ();
1077 av_push (av, newSVuv (tag));
1078 av_push (av, sv);
1079
1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1081 ? cbor_tagged_stash
1082 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1083 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1084 }
1085
1086 PUTBACK;
1087
1088 FREETMPS; LEAVE;
1089 }
1090 break;
1091 }
1092
788 return sv; 1093 return sv;
789 1094
790 AV *av = newAV (); 1095fail:
791 av_push (av, newSVuv (tag)); 1096 SvREFCNT_dec (sv);
792 av_push (av, sv); 1097 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1098}
795 1099
796static SV * 1100static SV *
797decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
798{ 1102{
799 WANT (1); 1103 WANT (1);
800 1104
801 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1106 {
803 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1114
811 return decode_str (dec, 1); 1115 case MAJOR_MISC >> MAJOR_SHIFT:
812 case 4: // array
813 return decode_av (dec);
814 case 5: // map
815 return decode_hv (dec);
816 case 6: // tag
817 return decode_tagged (dec);
818 case 7: // misc
819 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
820 { 1117 {
821 case 20: 1118 case SIMPLE_FALSE:
822#if CBOR_SLOW 1119#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
824#endif 1121#endif
825 return newSVsv (cbor_false); 1122 return newSVsv (types_false);
826 case 21: 1123 case SIMPLE_TRUE:
827#if CBOR_SLOW 1124#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
829#endif 1126#endif
830 return newSVsv (cbor_true); 1127 return newSVsv (types_true);
831 case 22: 1128 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
1130 case SIMPLE_UNDEF:
1131#if CBOR_SLOW
1132 types_error = get_bool ("Types::Serialiser::error");
1133#endif
1134 return newSVsv (types_error);
833 1135
834 case 25: 1136 case MISC_FLOAT16:
835 { 1137 {
836 WANT (2); 1138 WANT (2);
837 1139
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1141 dec->cur += 2;
840 1142
841 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
842 } 1144 }
843 1145
844 case 26: 1146 case MISC_FLOAT32:
845 { 1147 {
846 uint32_t fp; 1148 uint32_t fp;
847 WANT (4); 1149 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1151 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
853 1155
854 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
855 } 1157 }
856 1158
857 case 27: 1159 case MISC_FLOAT64:
858 { 1160 {
859 uint64_t fp; 1161 uint64_t fp;
860 WANT (8); 1162 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1164 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
866 1168
867 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
868 } 1170 }
869 1171
870 // 0..19 unassigned 1172 // 0..19 unassigned simple
871 // 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
872 default: 1176 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1178 }
875 1179
876 break; 1180 break;
877 } 1181 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 1182
935fail: 1183fail:
936 return &PL_sv_undef; 1184 return &PL_sv_undef;
937} 1185}
938 1186
939static SV * 1187static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1189{
942 dec_t dec; 1190 dec_t dec = { 0 };
943 SV *sv; 1191 SV *sv;
1192 STRLEN len;
1193 char *data = SvPVbyte (string, len);
944 1194
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1196 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1197 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1198
980 dec.cbor = *cbor; 1199 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1200 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1201 dec.end = (U8 *)data + len;
983 dec.err = 0;
984 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 1202
989 sv = decode_sv (&dec); 1203 sv = decode_sv (&dec);
990 1204
991 if (offset_return) 1205 if (offset_return)
992 *offset_return = dec.cur; 1206 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
997 1211
998 if (dec.err) 1212 if (dec.err)
999 { 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
1000 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (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);
1002 } 1227 }
1003 1228
1004 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
1005 1230
1006 return sv; 1231 return sv;
1007} 1232}
1008 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
1240static int
1241incr_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
1009///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1359// XS interface functions
1011 1360
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1362
1014BOOT: 1363BOOT:
1015{ 1364{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1365 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1366 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1367
1019 cbor_true = get_bool ("CBOR::XS::true"); 1368 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1369 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1370
1371 types_true = get_bool ("Types::Serialiser::true" );
1372 types_false = get_bool ("Types::Serialiser::false");
1373 types_error = get_bool ("Types::Serialiser::error");
1374
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1376
1377 sv_cbor = newSVpv ("CBOR", 0);
1378 SvREADONLY_on (sv_cbor);
1021} 1379}
1022 1380
1023PROTOTYPES: DISABLE 1381PROTOTYPES: DISABLE
1024 1382
1025void CLONE (...) 1383void CLONE (...)
1026 CODE: 1384 CODE:
1027 cbor_stash = 0; 1385 cbor_stash = 0;
1386 cbor_tagged_stash = 0;
1387 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1388 types_boolean_stash = 0;
1029 1389
1030void new (char *klass) 1390void new (char *klass)
1031 PPCODE: 1391 PPCODE:
1032{ 1392{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1393 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1401
1042void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1403 ALIAS:
1044 shrink = F_SHRINK 1404 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1406 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES
1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1411 PPCODE:
1047{ 1412{
1048 if (enable) 1413 if (enable)
1049 self->flags |= ix; 1414 self->flags |= ix;
1050 else 1415 else
1055 1420
1056void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
1057 ALIAS: 1422 ALIAS:
1058 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1425 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES
1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1429 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
1062 1431
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1433 PPCODE:
1080 CODE: 1449 CODE:
1081 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
1082 OUTPUT: 1451 OUTPUT:
1083 RETVAL 1452 RETVAL
1084 1453
1085#if 0 //TODO 1454void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1455 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1456 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1457 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
1094}
1095 1459
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1460SV *get_filter (CBOR *self)
1097 PPCODE: 1461 CODE:
1098{ 1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1463 OUTPUT:
1100 self->cb_sk_object = newHV (); 1464 RETVAL
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119 1465
1120void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1467 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1468 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1469 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
1137 PUSHs (sv); 1483 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1485}
1140 1486
1487void 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
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
1538
1141void DESTROY (CBOR *self) 1539void DESTROY (CBOR *self)
1142 CODE: 1540 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1541 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1542
1146PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
1147 1544
1148void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1549 PPCODE:
1150{ 1550{
1151 CBOR cbor; 1551 CBOR cbor;
1152 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1555 XPUSHs (scalar);
1155} 1556}
1156 1557
1157void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines