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 1.7 by root, Sat Oct 26 23:02:55 2013 UTC vs.
Revision 1.68 by root, Mon Nov 30 18:30:29 2020 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
14#define F_SHRINK 0x00000200UL 117#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
119#define F_ALLOW_SHARING 0x00000004UL
120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
125#define F_VALIDATE_UTF8 0x00000100UL
16 126
17#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
18 128
19#define SB do { 129#define SB do {
20#define SE } while (0) 130#define SE } while (0)
31#else 141#else
32# define CBOR_SLOW 0 142# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 143# define CBOR_STASH cbor_stash
34#endif 144#endif
35 145
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 146static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 147static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 148
39typedef struct { 149typedef struct {
40 U32 flags; 150 U32 flags;
41 U32 max_depth; 151 U32 max_depth;
42 STRLEN max_size; 152 STRLEN max_size;
153 SV *filter;
154
155 // for the incremental parser
156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 159} CBOR;
44 160
45ecb_inline void 161ecb_inline void
46cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
47{ 163{
48 Zero (cbor, 1, CBOR); 164 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 165 cbor->max_depth = 512;
166}
167
168ecb_inline void
169cbor_free (CBOR *cbor)
170{
171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
50} 173}
51 174
52///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
53// utility functions 176// utility functions
54 177
76 SvPV_renew (sv, SvCUR (sv) + 1); 199 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 200#endif
78 } 201 }
79} 202}
80 203
81///////////////////////////////////////////////////////////////////////////// 204// minimum length of a string to be registered for stringref
82// fp hell 205ecb_inline int
83 206minimum_string_length (UV idx)
84//TODO 207{
208 return idx <= 23 ? 3
209 : idx <= 0xffU ? 4
210 : idx <= 0xffffU ? 5
211 : idx <= 0xffffffffU ? 7
212 : 11;
213}
85 214
86///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
87// encoder 216// encoder
88 217
89// structure used for encoding CBOR 218// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 221 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 222 char *end; // SvEND (sv)
94 SV *sv; // result scalar 223 SV *sv; // result scalar
95 CBOR cbor; 224 CBOR cbor;
96 U32 depth; // recursion level 225 U32 depth; // recursion level
226 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
227 UV stringref_idx;
228 HV *shareable; // ptr => index, or 0
229 UV shareable_idx;
97} enc_t; 230} enc_t;
98 231
99ecb_inline void 232ecb_inline void
100need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
101{ 234{
102 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
103 { 236 {
104 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
105 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
106 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
107 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
113{ 246{
114 need (enc, 1); 247 need (enc, 1);
115 *enc->cur++ = ch; 248 *enc->cur++ = ch;
116} 249}
117 250
251// used for tags, intregers, element counts and so on
118static void 252static void
119encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
120{ 254{
121 need (enc, 9); 255 need (enc, 9);
122 256
123 if (len < 24) 257 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
125 else if (len <= 0xff) 259 else if (ecb_expect_true (len <= 0xffU))
126 { 260 {
127 *enc->cur++ = major | 24; 261 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 262 *enc->cur++ = len;
129 } 263 }
130 else if (len <= 0xffff) 264 else if (len <= 0xffffU)
131 { 265 {
132 *enc->cur++ = major | 25; 266 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 268 *enc->cur++ = len;
135 } 269 }
136 else if (len <= 0xffffffff) 270 else if (len <= 0xffffffffU)
137 { 271 {
138 *enc->cur++ = major | 26; 272 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 276 *enc->cur++ = len;
143 } 277 }
144 else 278 else
145 { 279 {
146 *enc->cur++ = major | 27; 280 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 281 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 282 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 283 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 284 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 285 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 288 *enc->cur++ = len;
155 } 289 }
156} 290}
157 291
158static void 292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
304ecb_inline void
305encode_tag (enc_t *enc, UV tag)
306{
307 encode_uint (enc, MAJOR_TAG, tag);
308}
309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 313{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
333ecb_inline void
334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 344 need (enc, len);
163 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
164 enc->cur += len; 346 enc->cur += len;
165} 347}
166 348
349ecb_inline void
350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
351{
352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
353 {
354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
355
356 if (SvOK (*svp))
357 {
358 // already registered, use stringref
359 encode_tag (enc, CBOR_TAG_STRINGREF);
360 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
361 return;
362 }
363 else if (len >= minimum_string_length (enc->stringref_idx))
364 {
365 // register only
366 sv_setuv (*svp, enc->stringref_idx);
367 ++enc->stringref_idx;
368 }
369 }
370
371 encode_str (enc, upgrade_utf8, utf8, str, len);
372}
373
374ecb_inline void
375encode_float16 (enc_t *enc, NV nv)
376{
377 need (enc, 1+2);
378
379 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
380
381 uint16_t fp = ecb_float_to_binary16 (nv);
382
383 if (!ecb_big_endian ())
384 fp = ecb_bswap16 (fp);
385
386 memcpy (enc->cur, &fp, 2);
387 enc->cur += 2;
388}
389
390ecb_inline void
391encode_float32 (enc_t *enc, NV nv)
392{
393 need (enc, 1+4);
394
395 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
396
397 uint32_t fp = ecb_float_to_binary32 (nv);
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap32 (fp);
401
402 memcpy (enc->cur, &fp, 4);
403 enc->cur += 4;
404}
405
406ecb_inline void
407encode_float64 (enc_t *enc, NV nv)
408{
409 need (enc, 1+8);
410
411 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
412
413 uint64_t fp = ecb_double_to_binary64 (nv);
414
415 if (!ecb_big_endian ())
416 fp = ecb_bswap64 (fp);
417
418 memcpy (enc->cur, &fp, 8);
419 enc->cur += 8;
420}
421
422ecb_inline void
423encode_forced (enc_t *enc, UV type, SV *sv)
424{
425 switch (type)
426 {
427 case AS_CBOR:
428 {
429 STRLEN len;
430 char *str = SvPVbyte (sv, len);
431
432 need (enc, len);
433 memcpy (enc->cur, str, len);
434 enc->cur += len;
435 }
436 break;
437
438 case AS_BYTES:
439 {
440 STRLEN len;
441 char *str = SvPVbyte (sv, len);
442 encode_strref (enc, 0, 0, str, len);
443 }
444 break;
445
446 case AS_TEXT:
447 {
448 STRLEN len;
449 char *str = SvPVutf8 (sv, len);
450 encode_strref (enc, 1, 1, str, len);
451 }
452 break;
453
454 case AS_INT: encode_int (enc, sv); break;
455 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
456 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
457 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
458
459 default:
460 croak ("encountered malformed CBOR::XS::Tagged object");
461 }
462}
463
167static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
168 465
169static void 466static void
170encode_av (enc_t *enc, AV *av) 467encode_av (enc_t *enc, AV *av)
171{ 468{
174 if (enc->depth >= enc->cbor.max_depth) 471 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 472 croak (ERR_NESTING_EXCEEDED);
176 473
177 ++enc->depth; 474 ++enc->depth;
178 475
179 encode_uint (enc, 0x80, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 477
478 if (ecb_expect_false (SvMAGICAL (av)))
181 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
182 { 480 {
183 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 483 }
484 else
485 for (i = 0; i <= len; ++i)
486 {
487 SV *sv = AvARRAY (av)[i];
488 encode_sv (enc, sv ? sv : &PL_sv_undef);
489 }
186 490
187 --enc->depth; 491 --enc->depth;
188} 492}
189 493
190static void 494static void
198 ++enc->depth; 502 ++enc->depth;
199 503
200 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
202 506
203 if (mg) 507 if (ecb_expect_false (mg))
204 encode_ch (enc, 0xa0 | 31); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 509 else
206 encode_uint (enc, 0xa0, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
207 511
208 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
209 { 513 {
210 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
212 else 516 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 517 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 518
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 519 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 520 }
217 521
218 if (mg) 522 if (ecb_expect_false (mg))
219 encode_ch (enc, 0xe0 | 31); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 524
221 --enc->depth; 525 --enc->depth;
222} 526}
223 527
224// encode objects, arrays and special \0=false and \1=true values. 528// encode objects, arrays and special \0=false and \1=true values.
225static void 529static void
226encode_rv (enc_t *enc, SV *sv) 530encode_rv (enc_t *enc, SV *sv)
227{ 531{
228 svtype svt;
229
230 SvGETMAGIC (sv); 532 SvGETMAGIC (sv);
533
231 svt = SvTYPE (sv); 534 svtype svt = SvTYPE (sv);
232 535
233 if (ecb_expect_false (SvOBJECT (sv))) 536 if (ecb_expect_false (SvOBJECT (sv)))
234 { 537 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 538 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 539 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 540 : gv_stashpv ("Types::Serialiser::Boolean", 1);
541 HV *error_stash = !CBOR_SLOW || types_error_stash
542 ? types_error_stash
543 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 544 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 545 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 546 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 547
548 HV *stash = SvSTASH (sv);
549
242 if (SvSTASH (sv) == boolean_stash) 550 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 551 {
552 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
553 return;
554 }
555 else if (stash == error_stash)
556 {
557 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
558 return;
559 }
244 else if (SvSTASH (sv) == tagged_stash) 560 else if (stash == tagged_stash)
245 { 561 {
246 if (svt != SVt_PVAV) 562 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 569 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 570 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
571 break;
572
573 case 3-1:
574 // a forced type [value, type, undef]
575 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
576 break;
577
578 default:
579 croak ("encountered malformed CBOR::XS::Tagged object");
580 }
581
582 return;
583 }
584 }
585
586 if (ecb_expect_false (SvREFCNT (sv) > 1)
587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
588 {
589 if (ecb_expect_false (!enc->shareable))
590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
591
592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
593
594 if (SvOK (*svp))
595 {
596 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
597 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
598 return;
251 } 599 }
252 else 600 else
253 { 601 {
602 sv_setuv (*svp, enc->shareable_idx);
603 ++enc->shareable_idx;
604 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
605 }
606 }
607
608 if (ecb_expect_false (SvOBJECT (sv)))
609 {
610 HV *stash = SvSTASH (sv);
611 GV *method;
612
613 if (enc->cbor.flags & F_FORBID_OBJECTS)
614 croak ("encountered object '%s', but forbid_objects is enabled",
615 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
617 {
618 dSP;
619
620 ENTER; SAVETMPS;
621 PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 622 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 623 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 624
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 625 PUTBACK;
626 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 627 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 628 SPAGAIN;
268 629
269 // catch this surprisingly common error 630 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 631 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 632 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 633
273 sv = POPs; 634 encode_sv (enc, POPs);
635
274 PUTBACK; 636 PUTBACK;
275 637
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 638 FREETMPS; LEAVE;
639 }
640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
641 {
642 dSP;
643
644 ENTER; SAVETMPS;
645 PUSHMARK (SP);
646 EXTEND (SP, 2);
647 // we re-bless the reference to get overload and other niceties right
648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
649 PUSHs (sv_cbor);
650
651 PUTBACK;
652 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
653 SPAGAIN;
654
655 // catch this surprisingly common error
656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
657 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
658
659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
660 encode_uint (enc, MAJOR_ARRAY, count + 1);
661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
662
663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
279 } 670 }
671
672 PUTBACK;
673
674 FREETMPS; LEAVE;
675 }
280 else 676 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 677 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 678 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 679 }
285 else if (svt == SVt_PVHV) 680 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 681 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 682 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 683 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 684 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 685 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 686 encode_tag (enc, CBOR_TAG_INDIRECTION);
687 encode_sv (enc, sv);
688 }
309} 689}
310 690
311static void 691static void
312encode_nv (enc_t *enc, SV *sv) 692encode_nv (enc_t *enc, SV *sv)
313{ 693{
314 double nv = SvNVX (sv); 694 double nv = SvNVX (sv);
315 695
316 need (enc, 9); 696 need (enc, 9);
317 697
318 if (ecb_expect_false (nv == (U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 700 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
322 { 702 encode_float32 (enc, nv);
323 uint32_t fp = ecb_float_to_binary32 (nv);
324
325 *enc->cur++ = 0xe0 | 26;
326
327 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp);
329
330 memcpy (enc->cur, &fp, 4);
331 enc->cur += 4;
332 }
333 else 703 else
334 { 704 encode_float64 (enc, nv);
335 uint64_t fp = ecb_double_to_binary64 (nv);
336
337 *enc->cur++ = 0xe0 | 27;
338
339 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp);
341
342 memcpy (enc->cur, &fp, 8);
343 enc->cur += 8;
344 }
345} 705}
346 706
347static void 707static void
348encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
349{ 709{
351 711
352 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
353 { 713 {
354 STRLEN len; 714 STRLEN len;
355 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
357 } 717 }
358 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 719 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
361 { 721 encode_int (enc, sv);
362 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv));
364 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv));
366 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
368 }
369 else if (SvROK (sv)) 722 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 727 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 728 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 729 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 730 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 731}
379 732
380static SV * 733static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
382{ 735{
383 enc_t enc; 736 enc_t enc = { 0 };
384 737
385 enc.cbor = *cbor; 738 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 742
391 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
744
745 if (cbor->flags & F_PACK_STRINGS)
746 {
747 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
748 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
749 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
750 }
751
392 encode_sv (&enc, scalar); 752 encode_sv (&enc, scalar);
393 753
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 754 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 755 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 756
410 U8 *end; // end of input string 770 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 771 const char *err; // parse error, if != 0
412 CBOR cbor; 772 CBOR cbor;
413 U32 depth; // recursion depth 773 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
775 AV *shareable;
776 AV *stringref;
777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
415} dec_t; 779} dec_t;
416 780
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 781// set dec->err to ERRSV
782ecb_cold static void
783err_errsv (dec_t *dec)
784{
785 if (!dec->err)
786 {
787 dec->err_sv = newSVsv (ERRSV);
418 788
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 789 // chop off the trailing \n
790 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
791 *SvEND (dec->err_sv) = 0;
420 792
793 dec->err = SvPVutf8_nolen (dec->err_sv);
794 }
795}
796
797// the following functions are used to reduce code size and help the compiler to optimise
798ecb_cold static void
799err_set (dec_t *dec, const char *reason)
800{
801 if (!dec->err)
802 dec->err = reason;
803}
804
805ecb_cold static void
806err_unexpected_end (dec_t *dec)
807{
808 err_set (dec, "unexpected end of CBOR data");
809}
810
811#define ERR_DO(do) SB do; goto fail; SE
812#define ERR(reason) ERR_DO (err_set (dec, reason))
813#define ERR_ERRSV ERR_DO (err_errsv (dec))
814
815#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
816
421#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 817#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
422#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
423 819
424static UV 820static UV
425decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
426{ 822{
427 switch (*dec->cur & 31) 823 U8 m = *dec->cur & MINOR_MASK;
428 { 824 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 825
434 case 24: 826 if (ecb_expect_true (m < LENGTH_EXT1))
827 return m;
828 else if (ecb_expect_true (m == LENGTH_EXT1))
829 {
435 WANT (2); 830 WANT (1);
436 dec->cur += 2; 831 dec->cur += 1;
437 return dec->cur[-1]; 832 return dec->cur[-1];
438 833 }
439 case 25: 834 else if (ecb_expect_true (m == LENGTH_EXT2))
835 {
440 WANT (3); 836 WANT (2);
441 dec->cur += 3; 837 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 838 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 839 | ((UV)dec->cur[-1]);
444 840 }
445 case 26: 841 else if (ecb_expect_true (m == LENGTH_EXT4))
842 {
446 WANT (5); 843 WANT (4);
447 dec->cur += 5; 844 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 845 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 846 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 847 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 848 | ((UV)dec->cur[-1]);
452 849 }
453 case 27: 850 else if (ecb_expect_true (m == LENGTH_EXT8))
851 {
454 WANT (9); 852 WANT (8);
455 dec->cur += 9; 853 dec->cur += 8;
854
855 return
856#if UVSIZE < 8
857 0
858#else
456 return (((UV)dec->cur[-8]) << 56) 859 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 860 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 861 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 862 | (((UV)dec->cur[-5]) << 32)
863#endif
460 | (((UV)dec->cur[-4]) << 24) 864 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 865 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 866 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 867 | ((UV)dec->cur[-1]);
464 868 }
465 default: 869 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 870 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 871
469fail: 872fail:
470 return 0; 873 return 0;
471} 874}
472 875
477{ 880{
478 AV *av = newAV (); 881 AV *av = newAV ();
479 882
480 DEC_INC_DEPTH; 883 DEC_INC_DEPTH;
481 884
482 if ((*dec->cur & 31) == 31) 885 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 886 {
484 ++dec->cur; 887 ++dec->cur;
485 888
486 for (;;) 889 for (;;)
487 { 890 {
488 WANT (1); 891 WANT (1);
489 892
490 if (*dec->cur == (0xe0 | 31)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
491 { 894 {
492 ++dec->cur; 895 ++dec->cur;
493 break; 896 break;
494 } 897 }
495 898
496 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
497 } 900 }
498 } 901 }
499 else 902 else
500 { 903 {
501 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
502 905
906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 907 av_fill (av, len - 1);
504 908
505 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 910 AvARRAY (av)[i] = decode_sv (dec);
507 } 911 }
508 912
509 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
510 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
511 915
512fail: 916fail:
513 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
514 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 919 return &PL_sv_undef;
516} 920}
517 921
922static void
923decode_he (dec_t *dec, HV *hv)
924{
925 // for speed reasons, we specialcase single-string
926 // byte or utf-8 strings as keys, but only when !stringref
927
928 if (ecb_expect_true (!dec->stringref))
929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
930 {
931 STRLEN len = decode_uint (dec);
932 char *key = (char *)dec->cur;
933
934 WANT (len);
935 dec->cur += len;
936
937 hv_store (hv, key, len, decode_sv (dec), 0);
938
939 return;
940 }
941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
942 {
943 STRLEN len = decode_uint (dec);
944 char *key = (char *)dec->cur;
945
946 WANT (len);
947 dec->cur += len;
948
949 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
950 if (!is_utf8_string (key, len))
951 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
952
953 hv_store (hv, key, -len, decode_sv (dec), 0);
954
955 return;
956 }
957
958 SV *k = decode_sv (dec);
959 SV *v = decode_sv (dec);
960
961 // we leak memory if uncaught exceptions are thrown by random magical
962 // methods, and this is hopefully the only place where it can happen,
963 // so if there is a chance of an exception, take the very slow path.
964 // since catching exceptions is "undocumented/internal/forbidden" by
965 // the new p5p powers, we need to call out to a perl function :/
966 if (ecb_expect_false (SvAMAGIC (k)))
967 {
968 dSP;
969
970 ENTER; SAVETMPS;
971 PUSHMARK (SP);
972 EXTEND (SP, 3);
973 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
974 PUSHs (sv_2mortal (k));
975 PUSHs (sv_2mortal (v));
976
977 PUTBACK;
978 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
979 SPAGAIN;
980
981 FREETMPS; LEAVE;
982
983 if (SvTRUE (ERRSV))
984 ERR_ERRSV;
985
986 return;
987 }
988
989 hv_store_ent (hv, k, v, 0);
990 SvREFCNT_dec_NN (k);
991
992fail:
993 ;
994}
995
518static SV * 996static SV *
519decode_hv (dec_t *dec) 997decode_hv (dec_t *dec)
520{ 998{
521 HV *hv = newHV (); 999 HV *hv = newHV ();
522 1000
523 DEC_INC_DEPTH; 1001 DEC_INC_DEPTH;
524 1002
525 if ((*dec->cur & 31) == 31) 1003 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 1004 {
527 ++dec->cur; 1005 ++dec->cur;
528 1006
529 for (;;) 1007 for (;;)
530 { 1008 {
531 WANT (1); 1009 WANT (1);
532 1010
533 if (*dec->cur == (0xe0 | 31)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
534 { 1012 {
535 ++dec->cur; 1013 ++dec->cur;
536 break; 1014 break;
537 } 1015 }
538 1016
539 SV *k = decode_sv (dec); 1017 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 1018 }
544 } 1019 }
545 else 1020 else
546 { 1021 {
547 int len = decode_uint (dec); 1022 UV pairs = decode_uint (dec);
548 1023
1024 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1025
549 while (len--) 1026 while (pairs--)
550 { 1027 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 1028 }
557 1029
558 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
560 1032
561fail: 1033fail:
562 SvREFCNT_dec (hv); 1034 SvREFCNT_dec_NN (hv);
563 DEC_DEC_DEPTH; 1035 DEC_DEC_DEPTH;
564 return &PL_sv_undef; 1036 return &PL_sv_undef;
565} 1037}
566 1038
567static SV * 1039static SV *
568decode_str (dec_t *dec, int utf8) 1040decode_str (dec_t *dec, int utf8)
569{ 1041{
570 SV *sv = 0; 1042 SV *sv = 0;
571 1043
572 if ((*dec->cur & 31) == 31) 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
573 { 1045 {
1046 // indefinite length strings
574 ++dec->cur; 1047 ++dec->cur;
575 1048
1049 U8 major = *dec->cur & MAJOR_MISC;
1050
576 sv = newSVpvn ("", 0); 1051 sv = newSVpvn ("", 0);
577 1052
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 1053 for (;;)
580 { 1054 {
581 WANT (1); 1055 WANT (1);
582 1056
583 if (*dec->cur == (0xe0 | 31)) 1057 if ((*dec->cur - major) > LENGTH_EXT8)
1058 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 1059 {
585 ++dec->cur; 1060 ++dec->cur;
586 break; 1061 break;
587 } 1062 }
1063 else
1064 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 1065
589 sv_catsv (sv, decode_sv (dec)); 1066 STRLEN len = decode_uint (dec);
1067
1068 WANT (len);
1069 sv_catpvn (sv, dec->cur, len);
1070 dec->cur += len;
590 } 1071 }
591 } 1072 }
592 else 1073 else
593 { 1074 {
594 STRLEN len = decode_uint (dec); 1075 STRLEN len = decode_uint (dec);
595 1076
596 WANT (len); 1077 WANT (len);
597 sv = newSVpvn (dec->cur, len); 1078 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 1079 dec->cur += len;
1080
1081 if (ecb_expect_false (dec->stringref)
1082 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1083 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 1084 }
600 1085
601 if (utf8) 1086 if (utf8)
1087 {
1088 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1089 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1090 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1091
602 SvUTF8_on (sv); 1092 SvUTF8_on (sv);
1093 }
603 1094
604 return sv; 1095 return sv;
605 1096
606fail: 1097fail:
607 SvREFCNT_dec (sv); 1098 SvREFCNT_dec (sv);
609} 1100}
610 1101
611static SV * 1102static SV *
612decode_tagged (dec_t *dec) 1103decode_tagged (dec_t *dec)
613{ 1104{
1105 SV *sv = 0;
614 UV tag = decode_uint (dec); 1106 UV tag = decode_uint (dec);
1107
1108 WANT (1);
1109
1110 switch (tag)
1111 {
1112 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 1113 sv = decode_sv (dec);
1114 break;
616 1115
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1116 case CBOR_TAG_INDIRECTION:
1117 sv = newRV_noinc (decode_sv (dec));
1118 break;
1119
1120 case CBOR_TAG_STRINGREF_NAMESPACE:
1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
1124 ENTER;
1125
1126 SAVESPTR (dec->stringref);
1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1128
1129 sv = decode_sv (dec);
1130
1131 LEAVE;
1132 }
1133 break;
1134
1135 case CBOR_TAG_STRINGREF:
1136 {
1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1139
1140 UV idx = decode_uint (dec);
1141
1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1143 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1144
1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1146 }
1147 break;
1148
1149 case CBOR_TAG_VALUE_SHAREABLE:
1150 {
1151 if (ecb_expect_false (!dec->shareable))
1152 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1153
1154 if (dec->cbor.flags & F_ALLOW_CYCLES)
1155 {
1156 sv = newSV (0);
1157 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1158
1159 SV *osv = decode_sv (dec);
1160 sv_setsv (sv, osv);
1161 SvREFCNT_dec_NN (osv);
1162 }
1163 else
1164 {
1165 av_push (dec->shareable, &PL_sv_undef);
1166 int idx = AvFILLp (dec->shareable);
1167 sv = decode_sv (dec);
1168 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1169 }
1170 }
1171 break;
1172
1173 case CBOR_TAG_VALUE_SHAREDREF:
1174 {
1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1177
1178 UV idx = decode_uint (dec);
1179
1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1182
1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1184
1185 if (sv == &PL_sv_undef)
1186 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1187 }
1188 break;
1189
1190 case CBOR_TAG_PERL_OBJECT:
1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
1195 sv = decode_sv (dec);
1196
1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1198 ERR ("corrupted CBOR data (non-array perl object)");
1199
1200 AV *av = (AV *)SvRV (sv);
1201 int len = av_len (av) + 1;
1202 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1203
1204 if (!stash)
1205 ERR ("cannot decode perl-object (package does not exist)");
1206
1207 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1208
1209 if (!method)
1210 ERR ("cannot decode perl-object (package does not have a THAW method)");
1211
1212 dSP;
1213
1214 ENTER; SAVETMPS;
1215 PUSHMARK (SP);
1216 EXTEND (SP, len + 1);
1217 // we re-bless the reference to get overload and other niceties right
1218 PUSHs (*av_fetch (av, 0, 1));
1219 PUSHs (sv_cbor);
1220
1221 int i;
1222
1223 for (i = 1; i < len; ++i)
1224 PUSHs (*av_fetch (av, i, 1));
1225
1226 PUTBACK;
1227 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1228 SPAGAIN;
1229
1230 if (SvTRUE (ERRSV))
1231 {
1232 FREETMPS; LEAVE;
1233 ERR_ERRSV;
1234 }
1235
1236 SvREFCNT_dec_NN (sv);
1237 sv = SvREFCNT_inc (POPs);
1238
1239 PUTBACK;
1240
1241 FREETMPS; LEAVE;
1242 }
1243 break;
1244
1245 default:
1246 filter:
1247 {
1248 SV *tag_sv = newSVuv (tag);
1249
1250 sv = decode_sv (dec);
1251
1252 dSP;
1253 ENTER; SAVETMPS;
1254 PUSHMARK (SP);
1255 EXTEND (SP, 2);
1256 PUSHs (tag_sv);
1257 PUSHs (sv);
1258
1259 PUTBACK;
1260 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1261 SPAGAIN;
1262
1263 if (SvTRUE (ERRSV))
1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1266 FREETMPS; LEAVE;
1267 ERR_ERRSV;
1268 }
1269
1270 if (count)
1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1273 SvREFCNT_dec_NN (sv);
1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1276 }
1277 else
1278 {
1279 AV *av = newAV ();
1280 av_push (av, tag_sv);
1281 av_push (av, sv);
1282
1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1284 ? cbor_tagged_stash
1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1286 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1287 }
1288
1289 PUTBACK;
1290
1291 FREETMPS; LEAVE;
1292 }
1293 break;
1294 }
1295
618 return sv; 1296 return sv;
619 1297
620 AV *av = newAV (); 1298fail:
621 av_push (av, newSVuv (tag)); 1299 SvREFCNT_dec (sv);
622 av_push (av, sv); 1300 return &PL_sv_undef;
623
624 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
625 ? cbor_tagged_stash
626 : gv_stashpv ("CBOR::XS::Tagged" , 1);
627
628 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
629} 1301}
630 1302
631static SV * 1303static SV *
632decode_sv (dec_t *dec) 1304decode_sv (dec_t *dec)
633{ 1305{
634 WANT (1); 1306 WANT (1);
635 1307
636 switch (*dec->cur >> 5) 1308 switch (*dec->cur >> MAJOR_SHIFT)
637 { 1309 {
638 case 0: // unsigned int 1310 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
639 return newSVuv (decode_uint (dec)); 1311 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
640 case 1: // negative int 1312 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
641 return newSViv (-1 - (IV)decode_uint (dec)); 1313 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
642 case 2: // octet string 1314 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
643 return decode_str (dec, 0); 1315 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
644 case 3: // utf-8 string 1316 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
645 return decode_str (dec, 1); 1317
646 case 4: // array 1318 case MAJOR_MISC >> MAJOR_SHIFT:
647 return decode_av (dec);
648 case 5: // map
649 return decode_hv (dec);
650 case 6: // tag
651 return decode_tagged (dec);
652 case 7: // misc
653 switch (*dec->cur++ & 31) 1319 switch (*dec->cur++ & MINOR_MASK)
654 { 1320 {
655 case 20: 1321 case SIMPLE_FALSE:
656#if CBOR_SLOW 1322#if CBOR_SLOW
657 cbor_false = get_bool ("CBOR::XS::false"); 1323 types_false = get_bool ("Types::Serialiser::false");
658#endif 1324#endif
659 return newSVsv (cbor_false); 1325 return newSVsv (types_false);
660 case 21: 1326 case SIMPLE_TRUE:
661#if CBOR_SLOW 1327#if CBOR_SLOW
662 cbor_true = get_bool ("CBOR::XS::true"); 1328 types_true = get_bool ("Types::Serialiser::true");
663#endif 1329#endif
664 return newSVsv (cbor_true); 1330 return newSVsv (types_true);
665 case 22: 1331 case SIMPLE_NULL:
666 return newSVsv (&PL_sv_undef); 1332 return newSVsv (&PL_sv_undef);
1333 case SIMPLE_UNDEF:
1334#if CBOR_SLOW
1335 types_error = get_bool ("Types::Serialiser::error");
1336#endif
1337 return newSVsv (types_error);
667 1338
668 case 25: 1339 case MISC_FLOAT16:
669 { 1340 {
670 WANT (2); 1341 WANT (2);
671 1342
672 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1343 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
673 dec->cur += 2; 1344 dec->cur += 2;
674 1345
675 return newSVnv (ecb_binary16_to_float (fp)); 1346 return newSVnv (ecb_binary16_to_float (fp));
676 } 1347 }
677 1348
678 case 26: 1349 case MISC_FLOAT32:
679 { 1350 {
680 uint32_t fp; 1351 uint32_t fp;
681 WANT (4); 1352 WANT (4);
682 memcpy (&fp, dec->cur, 4); 1353 memcpy (&fp, dec->cur, 4);
683 dec->cur += 4; 1354 dec->cur += 4;
686 fp = ecb_bswap32 (fp); 1357 fp = ecb_bswap32 (fp);
687 1358
688 return newSVnv (ecb_binary32_to_float (fp)); 1359 return newSVnv (ecb_binary32_to_float (fp));
689 } 1360 }
690 1361
691 case 27: 1362 case MISC_FLOAT64:
692 { 1363 {
693 uint64_t fp; 1364 uint64_t fp;
694 WANT (8); 1365 WANT (8);
695 memcpy (&fp, dec->cur, 8); 1366 memcpy (&fp, dec->cur, 8);
696 dec->cur += 8; 1367 dec->cur += 8;
699 fp = ecb_bswap64 (fp); 1370 fp = ecb_bswap64 (fp);
700 1371
701 return newSVnv (ecb_binary64_to_double (fp)); 1372 return newSVnv (ecb_binary64_to_double (fp));
702 } 1373 }
703 1374
704 // 0..19 unassigned 1375 // 0..19 unassigned simple
705 // 24 reserved + unassigned (reserved values are not encodable) 1376 // 24 reserved + unassigned simple (reserved values are not encodable)
1377 // 28-30 unassigned misc
1378 // 31 break code
706 default: 1379 default:
707 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1380 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
708 } 1381 }
709 1382
710 break; 1383 break;
711 } 1384 }
712 1385
715} 1388}
716 1389
717static SV * 1390static SV *
718decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1391decode_cbor (SV *string, CBOR *cbor, char **offset_return)
719{ 1392{
720 dec_t dec; 1393 dec_t dec = { 0 };
721 SV *sv; 1394 SV *sv;
1395 STRLEN len;
1396 char *data = SvPVbyte (string, len);
722 1397
723 /* work around bugs in 5.10 where manipulating magic values
724 * makes perl ignore the magic in subsequent accesses.
725 * also make a copy of non-PV values, to get them into a clean
726 * state (SvPV should do that, but it's buggy, see below).
727 */
728 /*SvGETMAGIC (string);*/
729 if (SvMAGICAL (string) || !SvPOK (string))
730 string = sv_2mortal (newSVsv (string));
731
732 SvUPGRADE (string, SVt_PV);
733
734 /* work around a bug in perl 5.10, which causes SvCUR to fail an
735 * assertion with -DDEBUGGING, although SvCUR is documented to
736 * return the xpv_cur field which certainly exists after upgrading.
737 * according to nicholas clark, calling SvPOK fixes this.
738 * But it doesn't fix it, so try another workaround, call SvPV_nolen
739 * and hope for the best.
740 * Damnit, SvPV_nolen still trips over yet another assertion. This
741 * assertion business is seriously broken, try yet another workaround
742 * for the broken -DDEBUGGING.
743 */
744 {
745#ifdef DEBUGGING
746 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
747#else
748 STRLEN offset = SvCUR (string);
749#endif
750
751 if (offset > cbor->max_size && cbor->max_size) 1398 if (len > cbor->max_size && cbor->max_size)
752 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1399 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
753 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1400 (unsigned long)len, (unsigned long)cbor->max_size);
754 }
755
756 sv_utf8_downgrade (string, 0);
757 1401
758 dec.cbor = *cbor; 1402 dec.cbor = *cbor;
759 dec.cur = (U8 *)SvPVX (string); 1403 dec.cur = (U8 *)data;
760 dec.end = (U8 *)SvEND (string); 1404 dec.end = (U8 *)data + len;
761 dec.err = 0;
762 dec.depth = 0;
763 1405
764 sv = decode_sv (&dec); 1406 sv = decode_sv (&dec);
765 1407
766 if (offset_return) 1408 if (offset_return)
767 *offset_return = dec.cur; 1409 *offset_return = dec.cur;
770 if (dec.cur != dec.end && !dec.err) 1412 if (dec.cur != dec.end && !dec.err)
771 dec.err = "garbage after CBOR object"; 1413 dec.err = "garbage after CBOR object";
772 1414
773 if (dec.err) 1415 if (dec.err)
774 { 1416 {
1417 if (dec.shareable)
1418 {
1419 // need to break cyclic links, which would all be in shareable
1420 int i;
1421 SV **svp;
1422
1423 for (i = av_len (dec.shareable) + 1; i--; )
1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1425 sv_setsv (*svp, &PL_sv_undef);
1426 }
1427
775 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
776 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1433 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
777 } 1434 }
778 1435
779 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
780 1437
781 return sv; 1438 return sv;
782} 1439}
783 1440
1441/////////////////////////////////////////////////////////////////////////////
1442// incremental parser
1443
1444#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1445
1446// returns 0 for notyet, 1 for success or error
1447static int
1448incr_parse (CBOR *self, SV *cborstr)
1449{
1450 STRLEN cur;
1451 SvPV (cborstr, cur);
1452
1453 while (ecb_expect_true (self->incr_need <= cur))
1454 {
1455 // table of integer count bytes
1456 static I8 incr_len[MINOR_MASK + 1] = {
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 0, 0, 0, 0, 0, 0, 0, 0,
1459 0, 0, 0, 0, 0, 0, 0, 0,
1460 1, 2, 4, 8,-1,-1,-1,-2
1461 };
1462
1463 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1464 U8 m = *p & MINOR_MASK;
1465 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1466 I8 ilen = incr_len[m];
1467
1468 self->incr_need = self->incr_pos + 1;
1469
1470 if (ecb_expect_false (ilen < 0))
1471 {
1472 if (m != MINOR_INDEF)
1473 return 1; // error
1474
1475 if (*p == (MAJOR_MISC | MINOR_INDEF))
1476 {
1477 if (count >= 0)
1478 return 1; // error
1479
1480 count = 1;
1481 }
1482 else
1483 {
1484 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1485 count = -1;
1486 }
1487 }
1488 else
1489 {
1490 self->incr_need += ilen;
1491 if (ecb_expect_false (self->incr_need > cur))
1492 return 0;
1493
1494 int major = *p >> MAJOR_SHIFT;
1495
1496 switch (major)
1497 {
1498 case MAJOR_TAG >> MAJOR_SHIFT:
1499 ++count; // tags merely prefix another value
1500 break;
1501
1502 case MAJOR_BYTES >> MAJOR_SHIFT:
1503 case MAJOR_TEXT >> MAJOR_SHIFT:
1504 case MAJOR_ARRAY >> MAJOR_SHIFT:
1505 case MAJOR_MAP >> MAJOR_SHIFT:
1506 {
1507 UV len;
1508
1509 if (ecb_expect_false (ilen))
1510 {
1511 len = 0;
1512
1513 do {
1514 len = (len << 8) | *++p;
1515 } while (--ilen);
1516 }
1517 else
1518 len = m;
1519
1520 switch (major)
1521 {
1522 case MAJOR_BYTES >> MAJOR_SHIFT:
1523 case MAJOR_TEXT >> MAJOR_SHIFT:
1524 self->incr_need += len;
1525 if (ecb_expect_false (self->incr_need > cur))
1526 return 0;
1527
1528 break;
1529
1530 case MAJOR_MAP >> MAJOR_SHIFT:
1531 len <<= 1;
1532 case MAJOR_ARRAY >> MAJOR_SHIFT:
1533 if (len)
1534 {
1535 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1536 count = len + 1;
1537 }
1538 break;
1539 }
1540 }
1541 }
1542 }
1543
1544 self->incr_pos = self->incr_need;
1545
1546 if (count > 0)
1547 {
1548 while (!--count)
1549 {
1550 if (!AvFILLp (self->incr_count))
1551 return 1; // done
1552
1553 SvREFCNT_dec_NN (av_pop (self->incr_count));
1554 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1555 }
1556
1557 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1558 }
1559 }
1560
1561 return 0;
1562}
1563
1564
784///////////////////////////////////////////////////////////////////////////// 1565/////////////////////////////////////////////////////////////////////////////
785// XS interface functions 1566// XS interface functions
786 1567
787MODULE = CBOR::XS PACKAGE = CBOR::XS 1568MODULE = CBOR::XS PACKAGE = CBOR::XS
788 1569
789BOOT: 1570BOOT:
790{ 1571{
791 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1572 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
792 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
793 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1573 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
794 1574
795 cbor_true = get_bool ("CBOR::XS::true"); 1575 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
796 cbor_false = get_bool ("CBOR::XS::false"); 1576 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1577
1578 types_true = get_bool ("Types::Serialiser::true" );
1579 types_false = get_bool ("Types::Serialiser::false");
1580 types_error = get_bool ("Types::Serialiser::error");
1581
1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1583
1584 sv_cbor = newSVpv ("CBOR", 0);
1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
797} 1588}
798 1589
799PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
800 1591
801void CLONE (...) 1592void CLONE (...)
802 CODE: 1593 CODE:
803 cbor_stash = 0; 1594 cbor_stash = 0;
804 cbor_boolean_stash = 0;
805 cbor_tagged_stash = 0; 1595 cbor_tagged_stash = 0;
1596 types_error_stash = 0;
1597 types_boolean_stash = 0;
806 1598
807void new (char *klass) 1599void new (char *klass)
808 PPCODE: 1600 PPCODE:
809{ 1601{
810 SV *pv = NEWSV (0, sizeof (CBOR)); 1602 SV *pv = NEWSV (0, sizeof (CBOR));
818 1610
819void shrink (CBOR *self, int enable = 1) 1611void shrink (CBOR *self, int enable = 1)
820 ALIAS: 1612 ALIAS:
821 shrink = F_SHRINK 1613 shrink = F_SHRINK
822 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1615 allow_sharing = F_ALLOW_SHARING
1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1620 text_strings = F_TEXT_STRINGS
1621 validate_utf8 = F_VALIDATE_UTF8
823 PPCODE: 1622 PPCODE:
824{ 1623{
825 if (enable) 1624 if (enable)
826 self->flags |= ix; 1625 self->flags |= ix;
827 else 1626 else
832 1631
833void get_shrink (CBOR *self) 1632void get_shrink (CBOR *self)
834 ALIAS: 1633 ALIAS:
835 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
836 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1636 get_allow_sharing = F_ALLOW_SHARING
1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1639 get_pack_strings = F_PACK_STRINGS
1640 get_text_keys = F_TEXT_KEYS
1641 get_text_strings = F_TEXT_STRINGS
1642 get_validate_utf8 = F_VALIDATE_UTF8
837 PPCODE: 1643 PPCODE:
838 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
839 1645
840void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
841 PPCODE: 1647 PPCODE:
857 CODE: 1663 CODE:
858 RETVAL = self->max_size; 1664 RETVAL = self->max_size;
859 OUTPUT: 1665 OUTPUT:
860 RETVAL 1666 RETVAL
861 1667
862#if 0 //TODO 1668void filter (CBOR *self, SV *filter = 0)
863
864void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
865 PPCODE: 1669 PPCODE:
866{
867 SvREFCNT_dec (self->cb_object); 1670 SvREFCNT_dec (self->filter);
868 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1671 self->filter = filter ? newSVsv (filter) : filter;
869
870 XPUSHs (ST (0)); 1672 XPUSHs (ST (0));
871}
872 1673
873void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1674SV *get_filter (CBOR *self)
874 PPCODE: 1675 CODE:
875{ 1676 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
876 if (!self->cb_sk_object) 1677 OUTPUT:
877 self->cb_sk_object = newHV (); 1678 RETVAL
878
879 if (SvOK (cb))
880 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
881 else
882 {
883 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
884
885 if (!HvKEYS (self->cb_sk_object))
886 {
887 SvREFCNT_dec (self->cb_sk_object);
888 self->cb_sk_object = 0;
889 }
890 }
891
892 XPUSHs (ST (0));
893}
894
895#endif
896 1679
897void encode (CBOR *self, SV *scalar) 1680void encode (CBOR *self, SV *scalar)
898 PPCODE: 1681 PPCODE:
899 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1682 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
900 XPUSHs (scalar); 1683 XPUSHs (scalar);
913 EXTEND (SP, 2); 1696 EXTEND (SP, 2);
914 PUSHs (sv); 1697 PUSHs (sv);
915 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1698 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
916} 1699}
917 1700
918#if 0 1701void incr_parse (CBOR *self, SV *cborstr)
1702 ALIAS:
1703 incr_parse_multiple = 1
1704 PPCODE:
1705{
1706 if (SvUTF8 (cborstr))
1707 sv_utf8_downgrade (cborstr, 0);
1708
1709 if (!self->incr_count)
1710 {
1711 self->incr_count = newAV ();
1712 self->incr_pos = 0;
1713 self->incr_need = 1;
1714
1715 av_push (self->incr_count, newSViv (1));
1716 }
1717
1718 do
1719 {
1720 if (!incr_parse (self, cborstr))
1721 {
1722 if (self->incr_need > self->max_size && self->max_size)
1723 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1724 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1725
1726 break;
1727 }
1728
1729 SV *sv;
1730 char *offset;
1731
1732 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1733 XPUSHs (sv);
1734
1735 sv_chop (cborstr, offset);
1736
1737 av_clear (self->incr_count);
1738 av_push (self->incr_count, newSViv (1));
1739
1740 self->incr_pos = 0;
1741 self->incr_need = self->incr_pos + 1;
1742 }
1743 while (ix);
1744}
1745
1746void incr_reset (CBOR *self)
1747 CODE:
1748{
1749 SvREFCNT_dec (self->incr_count);
1750 self->incr_count = 0;
1751}
919 1752
920void DESTROY (CBOR *self) 1753void DESTROY (CBOR *self)
921 CODE: 1754 PPCODE:
922 SvREFCNT_dec (self->cb_sk_object); 1755 cbor_free (self);
923 SvREFCNT_dec (self->cb_object);
924
925#endif
926 1756
927PROTOTYPES: ENABLE 1757PROTOTYPES: ENABLE
928 1758
929void encode_cbor (SV *scalar) 1759void encode_cbor (SV *scalar)
1760 ALIAS:
1761 encode_cbor = 0
1762 encode_cbor_sharing = F_ALLOW_SHARING
930 PPCODE: 1763 PPCODE:
931{ 1764{
932 CBOR cbor; 1765 CBOR cbor;
933 cbor_init (&cbor); 1766 cbor_init (&cbor);
1767 cbor.flags |= ix;
934 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1768 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
935 XPUSHs (scalar); 1769 XPUSHs (scalar);
936} 1770}
937 1771
938void decode_cbor (SV *cborstr) 1772void decode_cbor (SV *cborstr)
942 cbor_init (&cbor); 1776 cbor_init (&cbor);
943 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
944 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
945} 1779}
946 1780
1781#ifdef __AFL_COMPILER
1782
1783void
1784afl_init ()
1785 CODE:
1786 __AFL_INIT ();
1787
1788int
1789afl_loop (unsigned int count = 10000)
1790 CODE:
1791 RETVAL = __AFL_LOOP (count);
1792 OUTPUT:
1793 RETVAL
1794
1795#endif
1796

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines