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 1.26 by root, Fri Nov 22 09:40:13 2013 UTC vs.
Revision 1.78 by root, Fri Sep 8 06:01:12 2023 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
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 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// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
35
36// known major and minor types
37enum cbor_type
38{
39 MAJOR_SHIFT = 5,
40 MINOR_MASK = 0x1f,
41
42 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
43 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
44 MAJOR_BYTES = 2 << MAJOR_SHIFT,
45 MAJOR_TEXT = 3 << MAJOR_SHIFT,
46 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
47 MAJOR_MAP = 5 << MAJOR_SHIFT,
48 MAJOR_TAG = 6 << MAJOR_SHIFT,
49 MAJOR_MISC = 7 << MAJOR_SHIFT,
50
51 // INT/STRING/ARRAY/MAP subtypes
52 LENGTH_EXT1 = 24,
53 LENGTH_EXT2 = 25,
54 LENGTH_EXT4 = 26,
55 LENGTH_EXT8 = 27,
56
57 // SIMPLE types (effectively MISC subtypes)
58 SIMPLE_FALSE = 20,
59 SIMPLE_TRUE = 21,
60 SIMPLE_NULL = 22,
61 SIMPLE_UNDEF = 23,
62
63 // MISC subtype (unused)
64 MISC_EXT1 = 24,
65 MISC_FLOAT16 = 25,
66 MISC_FLOAT32 = 26,
67 MISC_FLOAT64 = 27,
68
69 // BYTES/TEXT/ARRAY/MAP
70 MINOR_INDEF = 31,
71};
24 72
25// known tags 73// known tags
26enum cbor_tag 74enum cbor_tag
27{ 75{
28 // inofficial extensions (pending iana registration) 76 // extensions
77 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 78 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 79 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 80 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 81 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 82 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 83 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 84
37 // rfc7049 85 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 86 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 87 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 88 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 89 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 90 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 91 CBOR_TAG_BIGFLOAT = 5, // array
44 92
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 93 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 94 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 95 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 96 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 97
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 98 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 99 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 100 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 101 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 103
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
105};
106
107// known forced types, also hardcoded in CBOR.pm
108enum
109{
110 AS_CBOR = 0,
111 AS_INT = 1,
112 AS_BYTES = 2,
113 AS_TEXT = 3,
114 AS_FLOAT16 = 4,
115 AS_FLOAT32 = 5,
116 AS_FLOAT64 = 6,
117 AS_MAP = 7,
118 // possibly future enhancements: (generic) float, (generic) string
57}; 119};
58 120
59#define F_SHRINK 0x00000001UL 121#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 123#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
126#define F_PACK_STRINGS 0x00000020UL
127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
63 130
64#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
65 132
66#define SB do { 133#define SB do {
67#define SE } while (0) 134#define SE } while (0)
79# define CBOR_SLOW 0 146# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 147# define CBOR_STASH cbor_stash
81#endif 148#endif
82 149
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 150static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 151static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 152
86typedef struct { 153typedef struct {
87 U32 flags; 154 U32 flags;
88 U32 max_depth; 155 U32 max_depth;
89 STRLEN max_size; 156 STRLEN max_size;
157 SV *filter;
158
159 // for the incremental parser
160 STRLEN incr_pos; // the current offset into the text
161 STRLEN incr_need; // minimum bytes needed to decode
162 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
90} CBOR; 163} CBOR;
91 164
92ecb_inline void 165ecb_inline void
93cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
94{ 167{
95 Zero (cbor, 1, CBOR); 168 Zero (cbor, 1, CBOR);
96 cbor->max_depth = 512; 169 cbor->max_depth = 512;
170}
171
172ecb_inline void
173cbor_free (CBOR *cbor)
174{
175 SvREFCNT_dec (cbor->filter);
176 SvREFCNT_dec (cbor->incr_count);
97} 177}
98 178
99///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
100// utility functions 180// utility functions
101 181
124#endif 204#endif
125 } 205 }
126} 206}
127 207
128// minimum length of a string to be registered for stringref 208// minimum length of a string to be registered for stringref
129ecb_inline int 209ecb_inline STRLEN
130minimum_string_length (UV idx) 210minimum_string_length (UV idx)
131{ 211{
132 return idx > 23 212 return idx <= 23 ? 3
133 ? idx > 0xffU 213 : idx <= 0xffU ? 4
134 ? idx > 0xffffU 214 : idx <= 0xffffU ? 5
135 ? idx > 0xffffffffU 215 : idx <= 0xffffffffU ? 7
136 ? 7 216 : 11;
137 : 6
138 : 5
139 : 4
140 : 3;
141} 217}
142 218
143///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
144// encoder 220// encoder
145 221
158} enc_t; 234} enc_t;
159 235
160ecb_inline void 236ecb_inline void
161need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
162{ 238{
163 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
164 { 240 {
165 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
166 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
167 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
168 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
169 } 245 }
170} 246}
171 247
248static void encode_sv (enc_t *enc, SV *sv);
249
172ecb_inline void 250ecb_inline void
173encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
174{ 252{
175 need (enc, 1); 253 need (enc, 1);
176 *enc->cur++ = ch; 254 *enc->cur++ = ch;
177} 255}
178 256
257// used for tags, intregers, element counts and so on
179static void 258static void
180encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
181{ 260{
182 need (enc, 9); 261 need (enc, 9);
183 262
184 if (len < 24) 263 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
186 else if (len <= 0xff) 265 else if (ecb_expect_true (len <= 0xffU))
187 { 266 {
188 *enc->cur++ = major | 24; 267 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 268 *enc->cur++ = len;
190 } 269 }
191 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
192 { 271 {
193 *enc->cur++ = major | 25; 272 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 274 *enc->cur++ = len;
196 } 275 }
197 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
198 { 277 {
199 *enc->cur++ = major | 26; 278 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 282 *enc->cur++ = len;
204 } 283 }
205 else 284 else
206 { 285 {
207 *enc->cur++ = major | 27; 286 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 287 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 288 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 289 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 290 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 291 *enc->cur++ = len >> 24;
214 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
215 *enc->cur++ = len; 294 *enc->cur++ = len;
216 } 295 }
217} 296}
218 297
298// encodes a perl value into a CBOR integer
299ecb_inline void
300encode_int (enc_t *enc, SV *sv)
301{
302 if (SvIsUV (sv))
303 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
304 else if (SvIVX (sv) >= 0)
305 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
306 else
307 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
308}
309
219ecb_inline void 310ecb_inline void
220encode_tag (enc_t *enc, UV tag) 311encode_tag (enc_t *enc, UV tag)
221{ 312{
222 encode_uint (enc, 0xc0, tag); 313 encode_uint (enc, MAJOR_TAG, tag);
223} 314}
224 315
225static void 316// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
317ecb_noinline static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 319{
320 STRLEN ulen = len;
321 U8 *p, *pend = (U8 *)str + len;
322
323 for (p = (U8 *)str; p < pend; ++p)
324 ulen += *p >> 7; // count set high bits
325
326 encode_uint (enc, MAJOR_TEXT, ulen);
327
328 need (enc, ulen);
329 for (p = (U8 *)str; p < pend; ++p)
330 if (*p < 0x80)
331 *enc->cur++ = *p;
332 else
333 {
334 *enc->cur++ = 0xc0 + (*p >> 6);
335 *enc->cur++ = 0x80 + (*p & 63);
336 }
337}
338
339ecb_inline void
340encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
341{
342 if (ecb_expect_false (upgrade_utf8))
343 if (!utf8)
344 {
345 encode_str_utf8 (enc, utf8, str, len);
346 return;
347 }
348
349 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
350 need (enc, len);
351 memcpy (enc->cur, str, len);
352 enc->cur += len;
353}
354
355ecb_inline void
356encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
357{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 358 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 359 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 360 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 361
232 if (SvOK (*svp)) 362 if (SvOK (*svp))
233 { 363 {
234 // already registered, use stringref 364 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 365 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 366 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 367 return;
238 } 368 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 369 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 370 {
241 // register only 371 // register only
242 sv_setuv (*svp, enc->stringref_idx); 372 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 373 ++enc->stringref_idx;
244 } 374 }
245 } 375 }
246 376
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 377 encode_str (enc, upgrade_utf8, utf8, str, len);
378}
379
380ecb_inline void
381encode_float16 (enc_t *enc, NV nv)
382{
248 need (enc, len); 383 need (enc, 1+2);
384
385 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
386
387 uint16_t fp = ecb_float_to_binary16 (nv);
388
389 if (!ecb_big_endian ())
390 fp = ecb_bswap16 (fp);
391
249 memcpy (enc->cur, str, len); 392 memcpy (enc->cur, &fp, 2);
250 enc->cur += len; 393 enc->cur += 2;
251} 394}
252 395
253static void encode_sv (enc_t *enc, SV *sv); 396ecb_inline void
397encode_float32 (enc_t *enc, NV nv)
398{
399 need (enc, 1+4);
400
401 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
402
403 uint32_t fp = ecb_float_to_binary32 (nv);
404
405 if (!ecb_big_endian ())
406 fp = ecb_bswap32 (fp);
407
408 memcpy (enc->cur, &fp, 4);
409 enc->cur += 4;
410}
411
412ecb_inline void
413encode_float64 (enc_t *enc, NV nv)
414{
415 need (enc, 1+8);
416
417 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
418
419 uint64_t fp = ecb_double_to_binary64 (nv);
420
421 if (!ecb_big_endian ())
422 fp = ecb_bswap64 (fp);
423
424 memcpy (enc->cur, &fp, 8);
425 enc->cur += 8;
426}
427
428ecb_inline void
429encode_bool (enc_t *enc, int istrue)
430{
431 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
432}
433
434// encodes an arrayref containing key-value pairs as CBOR map
435ecb_inline void
436encode_array_as_map (enc_t *enc, SV *sv)
437{
438 if (enc->depth >= enc->cbor.max_depth)
439 croak (ERR_NESTING_EXCEEDED);
440
441 ++enc->depth;
442
443 // as_map does error checking for us, but we re-check in case
444 // things have changed.
445
446 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
447 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
448
449 AV *av = (AV *)SvRV (sv);
450 int i, len = av_len (av);
451
452 if (!(len & 1))
453 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
454
455 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
456
457 for (i = 0; i <= len; ++i)
458 {
459 SV **svp = av_fetch (av, i, 0);
460 encode_sv (enc, svp ? *svp : &PL_sv_undef);
461 }
462
463 --enc->depth;
464}
465
466ecb_inline void
467encode_forced (enc_t *enc, UV type, SV *sv)
468{
469 switch (type)
470 {
471 case AS_CBOR:
472 {
473 STRLEN len;
474 char *str = SvPVbyte (sv, len);
475
476 need (enc, len);
477 memcpy (enc->cur, str, len);
478 enc->cur += len;
479 }
480 break;
481
482 case AS_BYTES:
483 {
484 STRLEN len;
485 char *str = SvPVbyte (sv, len);
486 encode_strref (enc, 0, 0, str, len);
487 }
488 break;
489
490 case AS_TEXT:
491 {
492 STRLEN len;
493 char *str = SvPVutf8 (sv, len);
494 encode_strref (enc, 1, 1, str, len);
495 }
496 break;
497
498 case AS_INT: encode_int (enc, sv); break;
499
500 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
501 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
502 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
503
504 case AS_MAP: encode_array_as_map (enc, sv); break;
505
506 default:
507 croak ("encountered malformed CBOR::XS::Tagged object");
508 }
509}
254 510
255static void 511static void
256encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
257{ 513{
258 int i, len = av_len (av); 514 int i, len = av_len (av);
260 if (enc->depth >= enc->cbor.max_depth) 516 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 517 croak (ERR_NESTING_EXCEEDED);
262 518
263 ++enc->depth; 519 ++enc->depth;
264 520
265 encode_uint (enc, 0x80, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 522
523 if (ecb_expect_false (SvMAGICAL (av)))
267 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
268 { 525 {
269 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 528 }
529 else
530 for (i = 0; i <= len; ++i)
531 {
532 SV *sv = AvARRAY (av)[i];
533 encode_sv (enc, sv ? sv : &PL_sv_undef);
534 }
272 535
273 --enc->depth; 536 --enc->depth;
274} 537}
275 538
276static void 539static void
284 ++enc->depth; 547 ++enc->depth;
285 548
286 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
288 551
289 if (mg) 552 if (ecb_expect_false (mg))
290 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 554 else
292 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
293 556
294 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
295 { 558 {
296 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
298 else 561 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 563
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 564 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 565 }
303 566
304 if (mg) 567 if (ecb_expect_false (mg))
305 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 569
307 --enc->depth; 570 --enc->depth;
308} 571}
309 572
310// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
311static void 574static void
312encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
313{ 576{
314 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
315
316 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317 && ecb_expect_false (SvREFCNT (sv) > 1))
318 {
319 if (!enc->shareable)
320 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324 if (SvOK (*svp))
325 {
326 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327 encode_uint (enc, 0x00, SvUV (*svp));
328 return;
329 }
330 else
331 {
332 sv_setuv (*svp, enc->shareable_idx);
333 ++enc->shareable_idx;
334 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335 }
336 }
337 578
338 svtype svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
339 580
340 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
341 { 582 {
348 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349 ? cbor_tagged_stash 590 ? cbor_tagged_stash
350 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
351 592
352 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
353 GV *method;
354 594
355 if (stash == boolean_stash) 595 if (stash == boolean_stash)
356 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
357 else if (stash == error_stash) 600 else if (stash == error_stash)
358 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
359 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
360 { 606 {
361 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
362 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
363 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
364 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 615 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
616 break;
617
618 case 3-1:
619 // a forced type [value, type, undef]
620 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
621 break;
622
623 default:
624 croak ("encountered malformed CBOR::XS::Tagged object");
625 }
626
627 return;
628 }
629 }
630
631 if (ecb_expect_false (SvREFCNT (sv) > 1)
632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
633 {
634 if (ecb_expect_false (!enc->shareable))
635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
636
637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
638
639 if (SvOK (*svp))
366 } 640 {
641 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
642 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
643 return;
644 }
645 else
646 {
647 sv_setuv (*svp, enc->shareable_idx);
648 ++enc->shareable_idx;
649 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
650 }
651 }
652
653 if (ecb_expect_false (SvOBJECT (sv)))
654 {
655 HV *stash = SvSTASH (sv);
656 GV *method;
657
658 if (enc->cbor.flags & F_FORBID_OBJECTS)
659 croak ("encountered object '%s', but forbid_objects is enabled",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
367 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368 { 662 {
369 dSP; 663 dSP;
370 664
371 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
372 // we re-bless the reference to get overload and other niceties right 667 // we re-bless the reference to get overload and other niceties right
373 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
374 669
375 PUTBACK; 670 PUTBACK;
376 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
389 } 684 }
390 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
391 { 686 {
392 dSP; 687 dSP;
393 688
394 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
395 EXTEND (SP, 2); 691 EXTEND (SP, 2);
396 // we re-bless the reference to get overload and other niceties right 692 // we re-bless the reference to get overload and other niceties right
397 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
398 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
399 695
404 // catch this surprisingly common error 700 // catch this surprisingly common error
405 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 702 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
407 703
408 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
410 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411 707
412 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
413 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
414 716
415 PUTBACK; 717 PUTBACK;
416 718
417 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
418 } 720 }
436{ 738{
437 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
438 740
439 need (enc, 9); 741 need (enc, 9);
440 742
441 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
442 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443 //TODO: maybe I32? 745 //TODO: maybe I32?
444 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
445 { 747 encode_float32 (enc, nv);
446 uint32_t fp = ecb_float_to_binary32 (nv);
447
448 *enc->cur++ = 0xe0 | 26;
449
450 if (!ecb_big_endian ())
451 fp = ecb_bswap32 (fp);
452
453 memcpy (enc->cur, &fp, 4);
454 enc->cur += 4;
455 }
456 else 748 else
457 { 749 encode_float64 (enc, nv);
458 uint64_t fp = ecb_double_to_binary64 (nv);
459
460 *enc->cur++ = 0xe0 | 27;
461
462 if (!ecb_big_endian ())
463 fp = ecb_bswap64 (fp);
464
465 memcpy (enc->cur, &fp, 8);
466 enc->cur += 8;
467 }
468} 750}
469 751
470static void 752static void
471encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
472{ 754{
474 756
475 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
476 { 758 {
477 STRLEN len; 759 STRLEN len;
478 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
479 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
480 } 762 }
481 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
482 encode_nv (enc, sv); 764 encode_nv (enc, sv);
483 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
484 { 766 encode_int (enc, sv);
485 if (SvIsUV (sv))
486 encode_uint (enc, 0x00, SvUVX (sv));
487 else if (SvIVX (sv) >= 0)
488 encode_uint (enc, 0x00, SvIVX (sv));
489 else
490 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
491 }
492 else if (SvROK (sv)) 767 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
495 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498 else 773 else
499 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 774 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
500 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501} 776}
502 777
503static SV * 778static SV *
504encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
505{ 780{
506 enc_t enc = { }; 781 enc_t enc = { 0 };
507 782
508 enc.cbor = *cbor; 783 enc.cbor = *cbor;
509 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
510 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
511 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
512 787
513 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
514 789
515 if (cbor->flags & F_ALLOW_STRINGREF) 790 if (cbor->flags & F_PACK_STRINGS)
516 { 791 {
517 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 792 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 793 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 794 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520 } 795 }
542 CBOR cbor; 817 CBOR cbor;
543 U32 depth; // recursion depth 818 U32 depth; // recursion depth
544 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
545 AV *shareable; 820 AV *shareable;
546 AV *stringref; 821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
547} dec_t; 824} dec_t;
548 825
549#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 826// set dec->err to ERRSV
827ecb_cold static void
828err_errsv (dec_t *dec)
829{
830 if (!dec->err)
831 {
832 dec->err_sv = newSVsv (ERRSV);
550 833
551#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 834 // chop off the trailing \n
835 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
836 *SvEND (dec->err_sv) = 0;
552 837
838 dec->err = SvPVutf8_nolen (dec->err_sv);
839 }
840}
841
842// the following functions are used to reduce code size and help the compiler to optimise
843ecb_cold static void
844err_set (dec_t *dec, const char *reason)
845{
846 if (!dec->err)
847 dec->err = reason;
848}
849
850ecb_cold static void
851err_unexpected_end (dec_t *dec)
852{
853 err_set (dec, "unexpected end of CBOR data");
854}
855
856#define ERR_DO(do) SB do; goto fail; SE
857#define ERR(reason) ERR_DO (err_set (dec, reason))
858#define ERR_ERRSV ERR_DO (err_errsv (dec))
859
860#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
861
553#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 862#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
554#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
555 864
556static UV 865static UV
557decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
558{ 867{
559 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
560 { 869 ++dec->cur;
561 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564 return *dec->cur++ & 31;
565 870
566 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
567 WANT (2); 875 WANT (1);
568 dec->cur += 2; 876 dec->cur += 1;
569 return dec->cur[-1]; 877 return dec->cur[-1];
570 878 }
571 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
572 WANT (3); 881 WANT (2);
573 dec->cur += 3; 882 dec->cur += 2;
574 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
575 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
576 885 }
577 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
578 WANT (5); 888 WANT (4);
579 dec->cur += 5; 889 dec->cur += 4;
580 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
581 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
582 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
584 894 }
585 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
586 WANT (9); 897 WANT (8);
587 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
588 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
589 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
590 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
591 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
592 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
593 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
594 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
595 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
596 913 }
597 default: 914 else
598 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599 }
600 916
601fail: 917fail:
602 return 0; 918 return 0;
603} 919}
604 920
609{ 925{
610 AV *av = newAV (); 926 AV *av = newAV ();
611 927
612 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
613 929
614 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615 { 931 {
616 ++dec->cur; 932 ++dec->cur;
617 933
618 for (;;) 934 for (;;)
619 { 935 {
620 WANT (1); 936 WANT (1);
621 937
622 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
623 { 939 {
624 ++dec->cur; 940 ++dec->cur;
625 break; 941 break;
626 } 942 }
627 943
628 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
629 } 945 }
630 } 946 }
631 else 947 else
632 { 948 {
633 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
634 950
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635 av_fill (av, len - 1); 952 av_fill (av, len - 1);
636 953
637 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
638 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
639 } 956 }
640 957
641 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
642 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
643 960
644fail: 961fail:
645 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
646 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
647 return &PL_sv_undef; 964 return &PL_sv_undef;
648} 965}
649 966
650static void 967static void
652{ 969{
653 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
654 // byte or utf-8 strings as keys, but only when !stringref 971 // byte or utf-8 strings as keys, but only when !stringref
655 972
656 if (ecb_expect_true (!dec->stringref)) 973 if (ecb_expect_true (!dec->stringref))
657 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658 { 975 {
659 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
660 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
661 978
979 WANT (len);
662 dec->cur += len; 980 dec->cur += len;
663 981
664 if (ecb_expect_false (dec->stringref))
665 av_push (dec->stringref, newSVpvn (key, len));
666
667 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
668 983
669 return; 984 return;
670 } 985 }
671 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672 { 987 {
673 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
674 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
675 990
991 WANT (len);
676 dec->cur += len; 992 dec->cur += len;
677 993
678 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 995 if (!cbor_is_utf8_string ((U8 *)key, len))
996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
680 997
681 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
682 999
683 return; 1000 return;
684 } 1001 }
685 1002
686 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
687 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
688 1005
1006 // we leak memory if uncaught exceptions are thrown by random magical
1007 // methods, and this is hopefully the only place where it can happen,
1008 // so if there is a chance of an exception, take the very slow path.
1009 // since catching exceptions is "undocumented/internal/forbidden" by
1010 // the new p5p powers, we need to call out to a perl function :/
1011 if (ecb_expect_false (SvAMAGIC (k)))
1012 {
1013 dSP;
1014
1015 ENTER; SAVETMPS;
1016 PUSHMARK (SP);
1017 EXTEND (SP, 3);
1018 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1019 PUSHs (sv_2mortal (k));
1020 PUSHs (sv_2mortal (v));
1021
1022 PUTBACK;
1023 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1024 SPAGAIN;
1025
1026 FREETMPS; LEAVE;
1027
1028 if (SvTRUE (ERRSV))
1029 ERR_ERRSV;
1030
1031 return;
1032 }
1033
689 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
690 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
691} 1039}
692 1040
693static SV * 1041static SV *
694decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
695{ 1043{
696 HV *hv = newHV (); 1044 HV *hv = newHV ();
697 1045
698 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
699 1047
700 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701 { 1049 {
702 ++dec->cur; 1050 ++dec->cur;
703 1051
704 for (;;) 1052 for (;;)
705 { 1053 {
706 WANT (1); 1054 WANT (1);
707 1055
708 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
709 { 1057 {
710 ++dec->cur; 1058 ++dec->cur;
711 break; 1059 break;
712 } 1060 }
713 1061
714 decode_he (dec, hv); 1062 decode_he (dec, hv);
715 } 1063 }
716 } 1064 }
717 else 1065 else
718 { 1066 {
719 int pairs = decode_uint (dec); 1067 UV pairs = decode_uint (dec);
1068
1069 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
720 1070
721 while (pairs--) 1071 while (pairs--)
722 decode_he (dec, hv); 1072 decode_he (dec, hv);
723 } 1073 }
724 1074
725 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
727 1077
728fail: 1078fail:
729 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
730 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
731 return &PL_sv_undef; 1081 return &PL_sv_undef;
732} 1082}
733 1083
734static SV * 1084static SV *
735decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
736{ 1086{
737 SV *sv = 0; 1087 SV *sv = 0;
738 1088
739 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
740 { 1090 {
1091 // indefinite length strings
741 ++dec->cur; 1092 ++dec->cur;
742 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
743 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
744 1097
745 // not very fast, and certainly not robust against illegal input
746 for (;;) 1098 for (;;)
747 { 1099 {
748 WANT (1); 1100 WANT (1);
749 1101
750 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751 { 1104 {
752 ++dec->cur; 1105 ++dec->cur;
753 break; 1106 break;
754 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755 1110
756 sv_catsv (sv, decode_sv (dec)); 1111 STRLEN len = decode_uint (dec);
1112
1113 WANT (len);
1114 sv_catpvn (sv, dec->cur, len);
1115 dec->cur += len;
757 } 1116 }
758 } 1117 }
759 else 1118 else
760 { 1119 {
761 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
768 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 1129 }
771 1130
772 if (utf8) 1131 if (utf8)
1132 {
1133 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1134 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1135 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1136
773 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
774 1139
775 return sv; 1140 return sv;
776 1141
777fail: 1142fail:
778 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
797 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
798 break; 1163 break;
799 1164
800 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
801 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
802 ENTER; SAVETMPS; 1169 ENTER;
803 1170
804 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
805 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
806 1173
807 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
808 1175
809 FREETMPS; LEAVE; 1176 LEAVE;
810 } 1177 }
811 break; 1178 break;
812 1179
813 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
814 { 1181 {
815 if ((*dec->cur >> 5) != 0) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817 1184
818 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
819 1186
820 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
821 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
822 1189
823 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
824 } 1191 }
825 break; 1192 break;
827 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
828 { 1195 {
829 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
830 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
1201 // if cycles are allowed, then we store an AV as value
1202 // while it is being decoded, and gather unresolved
1203 // references in it, to be re4solved after decoding.
1204 int idx, i;
832 sv = newSV (0); 1205 AV *av = newAV ();
833 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1206 av_push (dec->shareable, (SV *)av);
1207 idx = AvFILLp (dec->shareable);
834 1208
835 SV *osv = decode_sv (dec); 1209 sv = decode_sv (dec);
836 sv_setsv (sv, osv); 1210
1211 // the AV now contains \undef for all unresolved references,
1212 // so we fix them up here.
1213 for (i = 0; i <= AvFILLp (av); ++i)
1214 SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
1215
1216 // now replace the AV by a reference to the completed value
837 SvREFCNT_dec_NN (osv); 1217 SvREFCNT_dec_NN ((SV *)av);
1218 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1219 }
1220 else
1221 {
1222 av_push (dec->shareable, &PL_sv_undef);
1223 int idx = AvFILLp (dec->shareable);
1224 sv = decode_sv (dec);
1225 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1226 }
838 } 1227 }
839 break; 1228 break;
840 1229
841 case CBOR_TAG_VALUE_SHAREDREF: 1230 case CBOR_TAG_VALUE_SHAREDREF:
842 { 1231 {
843 if ((*dec->cur >> 5) != 0) 1232 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1233 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845 1234
846 UV idx = decode_uint (dec); 1235 UV idx = decode_uint (dec);
847 1236
848 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1237 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
849 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1238 ERR ("corrupted CBOR data (sharedref index out of bounds)");
850 1239
851 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1240 sv = AvARRAY (dec->shareable)[idx];
1241
1242 // reference to cycle, we create a new \undef and use that, and also
1243 // registerr it in the AV for later fixing
1244 if (SvTYPE (sv) == SVt_PVAV)
1245 {
1246 AV *av = (AV *)sv;
1247 sv = newRV_noinc (&PL_sv_undef);
1248 av_push (av, sv);
1249 }
1250 else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed
1251 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1252 else // we decoded the object earlier, no cycle
1253 sv = newSVsv (sv);
852 } 1254 }
853 break; 1255 break;
854 1256
855 case CBOR_TAG_PERL_OBJECT: 1257 case CBOR_TAG_PERL_OBJECT:
856 { 1258 {
1259 if (dec->cbor.flags & F_FORBID_OBJECTS)
1260 goto filter;
1261
857 sv = decode_sv (dec); 1262 sv = decode_sv (dec);
858 1263
859 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1264 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
860 ERR ("corrupted CBOR data (non-array perl object)"); 1265 ERR ("corrupted CBOR data (non-array perl object)");
861 1266
871 if (!method) 1276 if (!method)
872 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1277 ERR ("cannot decode perl-object (package does not have a THAW method)");
873 1278
874 dSP; 1279 dSP;
875 1280
876 ENTER; SAVETMPS; PUSHMARK (SP); 1281 ENTER; SAVETMPS;
1282 PUSHMARK (SP);
877 EXTEND (SP, len + 1); 1283 EXTEND (SP, len + 1);
878 // we re-bless the reference to get overload and other niceties right 1284 // we re-bless the reference to get overload and other niceties right
879 PUSHs (*av_fetch (av, 0, 1)); 1285 PUSHs (*av_fetch (av, 0, 1));
880 PUSHs (sv_cbor); 1286 PUSHs (sv_cbor);
881 1287
889 SPAGAIN; 1295 SPAGAIN;
890 1296
891 if (SvTRUE (ERRSV)) 1297 if (SvTRUE (ERRSV))
892 { 1298 {
893 FREETMPS; LEAVE; 1299 FREETMPS; LEAVE;
894 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1300 ERR_ERRSV;
895 } 1301 }
896 1302
897 SvREFCNT_dec (sv); 1303 SvREFCNT_dec_NN (sv);
898 sv = SvREFCNT_inc (POPs); 1304 sv = SvREFCNT_inc (POPs);
899 1305
900 PUTBACK; 1306 PUTBACK;
901 1307
902 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
903 } 1309 }
904 break; 1310 break;
905 1311
906 default: 1312 default:
1313 filter:
907 { 1314 {
1315 SV *tag_sv = newSVuv (tag);
1316
908 sv = decode_sv (dec); 1317 sv = decode_sv (dec);
909 1318
1319 dSP;
1320 ENTER; SAVETMPS;
1321 PUSHMARK (SP);
1322 EXTEND (SP, 2);
1323 PUSHs (tag_sv);
1324 PUSHs (sv);
1325
1326 PUTBACK;
1327 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1328 SPAGAIN;
1329
1330 if (SvTRUE (ERRSV))
1331 {
1332 SvREFCNT_dec_NN (tag_sv);
1333 FREETMPS; LEAVE;
1334 ERR_ERRSV;
1335 }
1336
1337 if (count)
1338 {
1339 SvREFCNT_dec_NN (tag_sv);
1340 SvREFCNT_dec_NN (sv);
1341 sv = SvREFCNT_inc_NN (TOPs);
1342 SP -= count;
1343 }
1344 else
1345 {
910 AV *av = newAV (); 1346 AV *av = newAV ();
911 av_push (av, newSVuv (tag)); 1347 av_push (av, tag_sv);
912 av_push (av, sv); 1348 av_push (av, sv);
913 1349
914 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1350 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915 ? cbor_tagged_stash 1351 ? cbor_tagged_stash
916 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 : gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1353 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1354 }
1355
1356 PUTBACK;
1357
1358 FREETMPS; LEAVE;
919 } 1359 }
920 break; 1360 break;
921 } 1361 }
922 1362
923 return sv; 1363 return sv;
930static SV * 1370static SV *
931decode_sv (dec_t *dec) 1371decode_sv (dec_t *dec)
932{ 1372{
933 WANT (1); 1373 WANT (1);
934 1374
935 switch (*dec->cur >> 5) 1375 switch (*dec->cur >> MAJOR_SHIFT)
936 { 1376 {
937 case 0: // unsigned int 1377 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938 return newSVuv (decode_uint (dec)); 1378 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939 case 1: // negative int 1379 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940 return newSViv (-1 - (IV)decode_uint (dec)); 1380 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941 case 2: // octet string 1381 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942 return decode_str (dec, 0); 1382 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943 case 3: // utf-8 string 1383 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944 return decode_str (dec, 1); 1384
945 case 4: // array 1385 case MAJOR_MISC >> MAJOR_SHIFT:
946 return decode_av (dec);
947 case 5: // map
948 return decode_hv (dec);
949 case 6: // tag
950 return decode_tagged (dec);
951 case 7: // misc
952 switch (*dec->cur++ & 31) 1386 switch (*dec->cur++ & MINOR_MASK)
953 { 1387 {
954 case 20: 1388 case SIMPLE_FALSE:
955#if CBOR_SLOW 1389#if CBOR_SLOW
956 types_false = get_bool ("Types::Serialiser::false"); 1390 types_false = get_bool ("Types::Serialiser::false");
957#endif 1391#endif
958 return newSVsv (types_false); 1392 return newSVsv (types_false);
959 case 21: 1393 case SIMPLE_TRUE:
960#if CBOR_SLOW 1394#if CBOR_SLOW
961 types_true = get_bool ("Types::Serialiser::true"); 1395 types_true = get_bool ("Types::Serialiser::true");
962#endif 1396#endif
963 return newSVsv (types_true); 1397 return newSVsv (types_true);
964 case 22: 1398 case SIMPLE_NULL:
965 return newSVsv (&PL_sv_undef); 1399 return newSVsv (&PL_sv_undef);
966 case 23: 1400 case SIMPLE_UNDEF:
967#if CBOR_SLOW 1401#if CBOR_SLOW
968 types_error = get_bool ("Types::Serialiser::error"); 1402 types_error = get_bool ("Types::Serialiser::error");
969#endif 1403#endif
970 return newSVsv (types_error); 1404 return newSVsv (types_error);
971 1405
972 case 25: 1406 case MISC_FLOAT16:
973 { 1407 {
974 WANT (2); 1408 WANT (2);
975 1409
976 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1410 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977 dec->cur += 2; 1411 dec->cur += 2;
978 1412
979 return newSVnv (ecb_binary16_to_float (fp)); 1413 return newSVnv (ecb_binary16_to_float (fp));
980 } 1414 }
981 1415
982 case 26: 1416 case MISC_FLOAT32:
983 { 1417 {
984 uint32_t fp; 1418 uint32_t fp;
985 WANT (4); 1419 WANT (4);
986 memcpy (&fp, dec->cur, 4); 1420 memcpy (&fp, dec->cur, 4);
987 dec->cur += 4; 1421 dec->cur += 4;
990 fp = ecb_bswap32 (fp); 1424 fp = ecb_bswap32 (fp);
991 1425
992 return newSVnv (ecb_binary32_to_float (fp)); 1426 return newSVnv (ecb_binary32_to_float (fp));
993 } 1427 }
994 1428
995 case 27: 1429 case MISC_FLOAT64:
996 { 1430 {
997 uint64_t fp; 1431 uint64_t fp;
998 WANT (8); 1432 WANT (8);
999 memcpy (&fp, dec->cur, 8); 1433 memcpy (&fp, dec->cur, 8);
1000 dec->cur += 8; 1434 dec->cur += 8;
1003 fp = ecb_bswap64 (fp); 1437 fp = ecb_bswap64 (fp);
1004 1438
1005 return newSVnv (ecb_binary64_to_double (fp)); 1439 return newSVnv (ecb_binary64_to_double (fp));
1006 } 1440 }
1007 1441
1008 // 0..19 unassigned 1442 // 0..19 unassigned simple
1009 // 24 reserved + unassigned (reserved values are not encodable) 1443 // 24 reserved + unassigned simple (reserved values are not encodable)
1444 // 28-30 unassigned misc
1445 // 31 break code
1010 default: 1446 default:
1011 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1447 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012 } 1448 }
1013 1449
1014 break; 1450 break;
1015 } 1451 }
1016 1452
1019} 1455}
1020 1456
1021static SV * 1457static SV *
1022decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1458decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1023{ 1459{
1024 dec_t dec = { }; 1460 dec_t dec = { 0 };
1025 SV *sv; 1461 SV *sv;
1026 STRLEN len; 1462 STRLEN len;
1027 char *data = SvPVbyte (string, len); 1463 char *data = SvPVbyte (string, len);
1028 1464
1029 if (len > cbor->max_size && cbor->max_size) 1465 if (len > cbor->max_size && cbor->max_size)
1043 if (dec.cur != dec.end && !dec.err) 1479 if (dec.cur != dec.end && !dec.err)
1044 dec.err = "garbage after CBOR object"; 1480 dec.err = "garbage after CBOR object";
1045 1481
1046 if (dec.err) 1482 if (dec.err)
1047 { 1483 {
1484 if (dec.shareable)
1485 {
1486 // need to break cyclic links, which would all be in shareable
1487 int i;
1488 SV **svp;
1489
1490 for (i = av_len (dec.shareable) + 1; i--; )
1491 if ((svp = av_fetch (dec.shareable, i, 0)))
1492 sv_setsv (*svp, &PL_sv_undef);
1493 }
1494
1048 SvREFCNT_dec (sv); 1495 SvREFCNT_dec_NN (sv);
1496
1497 if (dec.err_sv)
1498 sv_2mortal (dec.err_sv);
1499
1049 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1500 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1050 } 1501 }
1051 1502
1052 sv = sv_2mortal (sv); 1503 sv = sv_2mortal (sv);
1053 1504
1054 return sv; 1505 return sv;
1055} 1506}
1056 1507
1508/////////////////////////////////////////////////////////////////////////////
1509// incremental parser
1510
1511#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1512
1513// returns 0 for notyet, 1 for success or error
1514static int
1515incr_parse (CBOR *self, SV *cborstr)
1516{
1517 STRLEN cur;
1518 SvPV (cborstr, cur);
1519
1520 while (ecb_expect_true (self->incr_need <= cur))
1521 {
1522 // table of integer count bytes
1523 static I8 incr_len[MINOR_MASK + 1] = {
1524 0, 0, 0, 0, 0, 0, 0, 0,
1525 0, 0, 0, 0, 0, 0, 0, 0,
1526 0, 0, 0, 0, 0, 0, 0, 0,
1527 1, 2, 4, 8,-1,-1,-1,-2
1528 };
1529
1530 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1531 U8 m = *p & MINOR_MASK;
1532 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1533 I8 ilen = incr_len[m];
1534
1535 self->incr_need = self->incr_pos + 1;
1536
1537 if (ecb_expect_false (ilen < 0))
1538 {
1539 if (m != MINOR_INDEF)
1540 return 1; // error
1541
1542 if (*p == (MAJOR_MISC | MINOR_INDEF))
1543 {
1544 if (count >= 0)
1545 return 1; // error
1546
1547 count = 1;
1548 }
1549 else
1550 {
1551 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1552 count = -1;
1553 }
1554 }
1555 else
1556 {
1557 self->incr_need += ilen;
1558 if (ecb_expect_false (self->incr_need > cur))
1559 return 0;
1560
1561 int major = *p >> MAJOR_SHIFT;
1562
1563 switch (major)
1564 {
1565 case MAJOR_TAG >> MAJOR_SHIFT:
1566 ++count; // tags merely prefix another value
1567 break;
1568
1569 case MAJOR_BYTES >> MAJOR_SHIFT:
1570 case MAJOR_TEXT >> MAJOR_SHIFT:
1571 case MAJOR_ARRAY >> MAJOR_SHIFT:
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 {
1574 UV len;
1575
1576 if (ecb_expect_false (ilen))
1577 {
1578 len = 0;
1579
1580 do {
1581 len = (len << 8) | *++p;
1582 } while (--ilen);
1583 }
1584 else
1585 len = m;
1586
1587 switch (major)
1588 {
1589 case MAJOR_BYTES >> MAJOR_SHIFT:
1590 case MAJOR_TEXT >> MAJOR_SHIFT:
1591 self->incr_need += len;
1592 if (ecb_expect_false (self->incr_need > cur))
1593 return 0;
1594
1595 break;
1596
1597 case MAJOR_MAP >> MAJOR_SHIFT:
1598 len <<= 1;
1599 /* FALLTHROUGH */
1600 case MAJOR_ARRAY >> MAJOR_SHIFT:
1601 if (len)
1602 {
1603 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1604 count = len + 1;
1605 }
1606 break;
1607 }
1608 }
1609 }
1610 }
1611
1612 self->incr_pos = self->incr_need;
1613
1614 if (count > 0)
1615 {
1616 while (!--count)
1617 {
1618 if (!AvFILLp (self->incr_count))
1619 return 1; // done
1620
1621 SvREFCNT_dec_NN (av_pop (self->incr_count));
1622 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1623 }
1624
1625 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1626 }
1627 }
1628
1629 return 0;
1630}
1631
1632
1057///////////////////////////////////////////////////////////////////////////// 1633/////////////////////////////////////////////////////////////////////////////
1058// XS interface functions 1634// XS interface functions
1059 1635
1060MODULE = CBOR::XS PACKAGE = CBOR::XS 1636MODULE = CBOR::XS PACKAGE = CBOR::XS
1061 1637
1069 1645
1070 types_true = get_bool ("Types::Serialiser::true" ); 1646 types_true = get_bool ("Types::Serialiser::true" );
1071 types_false = get_bool ("Types::Serialiser::false"); 1647 types_false = get_bool ("Types::Serialiser::false");
1072 types_error = get_bool ("Types::Serialiser::error"); 1648 types_error = get_bool ("Types::Serialiser::error");
1073 1649
1650 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1651
1074 sv_cbor = newSVpv ("CBOR", 0); 1652 sv_cbor = newSVpv ("CBOR", 0);
1075 SvREADONLY_on (sv_cbor); 1653 SvREADONLY_on (sv_cbor);
1654
1655 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1076} 1656}
1077 1657
1078PROTOTYPES: DISABLE 1658PROTOTYPES: DISABLE
1079 1659
1080void CLONE (...) 1660void CLONE (...)
1099void shrink (CBOR *self, int enable = 1) 1679void shrink (CBOR *self, int enable = 1)
1100 ALIAS: 1680 ALIAS:
1101 shrink = F_SHRINK 1681 shrink = F_SHRINK
1102 allow_unknown = F_ALLOW_UNKNOWN 1682 allow_unknown = F_ALLOW_UNKNOWN
1103 allow_sharing = F_ALLOW_SHARING 1683 allow_sharing = F_ALLOW_SHARING
1104 allow_stringref = F_ALLOW_STRINGREF 1684 allow_cycles = F_ALLOW_CYCLES
1685 forbid_objects = F_FORBID_OBJECTS
1686 pack_strings = F_PACK_STRINGS
1687 text_keys = F_TEXT_KEYS
1688 text_strings = F_TEXT_STRINGS
1689 validate_utf8 = F_VALIDATE_UTF8
1105 PPCODE: 1690 PPCODE:
1106{ 1691{
1107 if (enable) 1692 if (enable)
1108 self->flags |= ix; 1693 self->flags |= ix;
1109 else 1694 else
1115void get_shrink (CBOR *self) 1700void get_shrink (CBOR *self)
1116 ALIAS: 1701 ALIAS:
1117 get_shrink = F_SHRINK 1702 get_shrink = F_SHRINK
1118 get_allow_unknown = F_ALLOW_UNKNOWN 1703 get_allow_unknown = F_ALLOW_UNKNOWN
1119 get_allow_sharing = F_ALLOW_SHARING 1704 get_allow_sharing = F_ALLOW_SHARING
1120 get_allow_stringref = F_ALLOW_STRINGREF 1705 get_allow_cycles = F_ALLOW_CYCLES
1706 get_forbid_objects = F_FORBID_OBJECTS
1707 get_pack_strings = F_PACK_STRINGS
1708 get_text_keys = F_TEXT_KEYS
1709 get_text_strings = F_TEXT_STRINGS
1710 get_validate_utf8 = F_VALIDATE_UTF8
1121 PPCODE: 1711 PPCODE:
1122 XPUSHs (boolSV (self->flags & ix)); 1712 XPUSHs (boolSV (self->flags & ix));
1123 1713
1124void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1714void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125 PPCODE: 1715 PPCODE:
1138 XPUSHs (ST (0)); 1728 XPUSHs (ST (0));
1139 1729
1140int get_max_size (CBOR *self) 1730int get_max_size (CBOR *self)
1141 CODE: 1731 CODE:
1142 RETVAL = self->max_size; 1732 RETVAL = self->max_size;
1733 OUTPUT:
1734 RETVAL
1735
1736void filter (CBOR *self, SV *filter = 0)
1737 PPCODE:
1738 SvREFCNT_dec (self->filter);
1739 self->filter = filter ? newSVsv (filter) : filter;
1740 XPUSHs (ST (0));
1741
1742SV *get_filter (CBOR *self)
1743 CODE:
1744 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1143 OUTPUT: 1745 OUTPUT:
1144 RETVAL 1746 RETVAL
1145 1747
1146void encode (CBOR *self, SV *scalar) 1748void encode (CBOR *self, SV *scalar)
1147 PPCODE: 1749 PPCODE:
1162 EXTEND (SP, 2); 1764 EXTEND (SP, 2);
1163 PUSHs (sv); 1765 PUSHs (sv);
1164 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1766 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165} 1767}
1166 1768
1769void incr_parse (CBOR *self, SV *cborstr)
1770 ALIAS:
1771 incr_parse_multiple = 1
1772 PPCODE:
1773{
1774 if (SvUTF8 (cborstr))
1775 sv_utf8_downgrade (cborstr, 0);
1776
1777 if (!self->incr_count)
1778 {
1779 self->incr_count = newAV ();
1780 self->incr_pos = 0;
1781 self->incr_need = 1;
1782
1783 av_push (self->incr_count, newSViv (1));
1784 }
1785
1786 do
1787 {
1788 if (!incr_parse (self, cborstr))
1789 {
1790 if (self->incr_need > self->max_size && self->max_size)
1791 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1792 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1793
1794 break;
1795 }
1796
1797 SV *sv;
1798 char *offset;
1799
1800 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1801 XPUSHs (sv);
1802
1803 sv_chop (cborstr, offset);
1804
1805 av_clear (self->incr_count);
1806 av_push (self->incr_count, newSViv (1));
1807
1808 self->incr_pos = 0;
1809 self->incr_need = self->incr_pos + 1;
1810 }
1811 while (ix);
1812}
1813
1814void incr_reset (CBOR *self)
1815 CODE:
1816{
1817 SvREFCNT_dec (self->incr_count);
1818 self->incr_count = 0;
1819}
1820
1821void DESTROY (CBOR *self)
1822 PPCODE:
1823 cbor_free (self);
1824
1167PROTOTYPES: ENABLE 1825PROTOTYPES: ENABLE
1168 1826
1169void encode_cbor (SV *scalar) 1827void encode_cbor (SV *scalar)
1828 ALIAS:
1829 encode_cbor = 0
1830 encode_cbor_sharing = F_ALLOW_SHARING
1170 PPCODE: 1831 PPCODE:
1171{ 1832{
1172 CBOR cbor; 1833 CBOR cbor;
1173 cbor_init (&cbor); 1834 cbor_init (&cbor);
1835 cbor.flags |= ix;
1174 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1836 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175 XPUSHs (scalar); 1837 XPUSHs (scalar);
1176} 1838}
1177 1839
1178void decode_cbor (SV *cborstr) 1840void decode_cbor (SV *cborstr)
1182 cbor_init (&cbor); 1844 cbor_init (&cbor);
1183 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1845 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1184 XPUSHs (cborstr); 1846 XPUSHs (cborstr);
1185} 1847}
1186 1848
1849#ifdef __AFL_COMPILER
1850
1851void
1852afl_init ()
1853 CODE:
1854 __AFL_INIT ();
1855
1856int
1857afl_loop (unsigned int count = 10000)
1858 CODE:
1859 RETVAL = __AFL_LOOP (count);
1860 OUTPUT:
1861 RETVAL
1862
1863#endif
1864

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines