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 1.22 by root, Wed Nov 20 15:05:46 2013 UTC vs.
Revision 1.75 by root, Thu Sep 7 23:52:24 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
57}; 105};
58 106
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
119};
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
124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 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_DEDUP_STRINGS)) 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}
275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279} 537}
280 538
281static void 539static void
282encode_hv (enc_t *enc, HV *hv) 540encode_hv (enc_t *enc, HV *hv)
283{ 541{
289 ++enc->depth; 547 ++enc->depth;
290 548
291 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
292 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
293 551
294 if (mg) 552 if (ecb_expect_false (mg))
295 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
296 else 554 else
297 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
298 556
299 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
300 { 558 {
301 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
302 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
303 else 561 else
304 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));
305 563
306 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));
307 } 565 }
308 566
309 if (mg) 567 if (ecb_expect_false (mg))
310 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
311 569
312 --enc->depth; 570 --enc->depth;
313} 571}
314 572
315// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
316static void 574static void
317encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
318{ 576{
319 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
320
321 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
322 && ecb_expect_false (SvREFCNT (sv) > 1))
323 {
324 if (!enc->shareable)
325 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
326
327 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
328
329 if (SvOK (*svp))
330 {
331 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
332 encode_uint (enc, 0x00, SvUV (*svp));
333 return;
334 }
335 else
336 {
337 sv_setuv (*svp, enc->shareable_idx);
338 ++enc->shareable_idx;
339 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
340 }
341 }
342 578
343 svtype svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
344 580
345 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
346 { 582 {
353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
354 ? cbor_tagged_stash 590 ? cbor_tagged_stash
355 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
356 592
357 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
358 GV *method;
359 594
360 if (stash == boolean_stash) 595 if (stash == boolean_stash)
361 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
362 else if (stash == error_stash) 600 else if (stash == error_stash)
363 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
364 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
365 { 606 {
366 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
367 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
368 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
370 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))
371 } 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))));
372 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
373 { 662 {
374 dSP; 663 dSP;
375 664
376 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
377 // 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
378 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
379 669
380 PUTBACK; 670 PUTBACK;
381 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
394 } 684 }
395 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
396 { 686 {
397 dSP; 687 dSP;
398 688
399 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
400 EXTEND (SP, 2); 691 EXTEND (SP, 2);
401 // 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
402 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
403 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
404 695
409 // catch this surprisingly common error 700 // catch this surprisingly common error
410 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
411 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));
412 703
413 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
414 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
415 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
416 707
417 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
418 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
419 716
420 PUTBACK; 717 PUTBACK;
421 718
422 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
423 } 720 }
441{ 738{
442 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
443 740
444 need (enc, 9); 741 need (enc, 9);
445 742
446 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
447 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
448 //TODO: maybe I32? 745 //TODO: maybe I32?
449 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
450 { 747 encode_float32 (enc, nv);
451 uint32_t fp = ecb_float_to_binary32 (nv);
452
453 *enc->cur++ = 0xe0 | 26;
454
455 if (!ecb_big_endian ())
456 fp = ecb_bswap32 (fp);
457
458 memcpy (enc->cur, &fp, 4);
459 enc->cur += 4;
460 }
461 else 748 else
462 { 749 encode_float64 (enc, nv);
463 uint64_t fp = ecb_double_to_binary64 (nv);
464
465 *enc->cur++ = 0xe0 | 27;
466
467 if (!ecb_big_endian ())
468 fp = ecb_bswap64 (fp);
469
470 memcpy (enc->cur, &fp, 8);
471 enc->cur += 8;
472 }
473} 750}
474 751
475static void 752static void
476encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
477{ 754{
479 756
480 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
481 { 758 {
482 STRLEN len; 759 STRLEN len;
483 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
484 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
485 } 762 }
486 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
487 encode_nv (enc, sv); 764 encode_nv (enc, sv);
488 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
489 { 766 encode_int (enc, sv);
490 if (SvIsUV (sv))
491 encode_uint (enc, 0x00, SvUVX (sv));
492 else if (SvIVX (sv) >= 0)
493 encode_uint (enc, 0x00, SvIVX (sv));
494 else
495 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
496 }
497 else if (SvROK (sv)) 767 else if (SvROK (sv))
498 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
499 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
500 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
501 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
502 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
503 else 773 else
504 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",
505 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
506} 776}
507 777
508static SV * 778static SV *
509encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
510{ 780{
511 enc_t enc = { }; 781 enc_t enc = { 0 };
512 782
513 enc.cbor = *cbor; 783 enc.cbor = *cbor;
514 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
515 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
516 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
517 787
518 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
519 789
520 if (cbor->flags & F_DEDUP_STRINGS) 790 if (cbor->flags & F_PACK_STRINGS)
521 { 791 {
522 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 792 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
523 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 793 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
524 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 794 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
525 } 795 }
547 CBOR cbor; 817 CBOR cbor;
548 U32 depth; // recursion depth 818 U32 depth; // recursion depth
549 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
550 AV *shareable; 820 AV *shareable;
551 AV *stringref; 821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
552} dec_t; 824} dec_t;
553 825
554#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);
555 833
556#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;
557 837
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
558#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)
559#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
560 864
561static UV 865static UV
562decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
563{ 867{
564 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
565 { 869 ++dec->cur;
566 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
567 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
568 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
569 return *dec->cur++ & 31;
570 870
571 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
572 WANT (2); 875 WANT (1);
573 dec->cur += 2; 876 dec->cur += 1;
574 return dec->cur[-1]; 877 return dec->cur[-1];
575 878 }
576 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
577 WANT (3); 881 WANT (2);
578 dec->cur += 3; 882 dec->cur += 2;
579 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
580 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
581 885 }
582 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
583 WANT (5); 888 WANT (4);
584 dec->cur += 5; 889 dec->cur += 4;
585 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
586 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
587 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
588 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
589 894 }
590 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
591 WANT (9); 897 WANT (8);
592 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
593 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
594 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
595 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
596 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
597 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
601 913 }
602 default: 914 else
603 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
604 }
605 916
606fail: 917fail:
607 return 0; 918 return 0;
608} 919}
609 920
614{ 925{
615 AV *av = newAV (); 926 AV *av = newAV ();
616 927
617 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
618 929
619 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
620 { 931 {
621 ++dec->cur; 932 ++dec->cur;
622 933
623 for (;;) 934 for (;;)
624 { 935 {
625 WANT (1); 936 WANT (1);
626 937
627 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
628 { 939 {
629 ++dec->cur; 940 ++dec->cur;
630 break; 941 break;
631 } 942 }
632 943
633 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
634 } 945 }
635 } 946 }
636 else 947 else
637 { 948 {
638 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
639 950
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
640 av_fill (av, len - 1); 952 av_fill (av, len - 1);
641 953
642 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
643 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
644 } 956 }
645 957
646 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
647 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
648 960
649fail: 961fail:
650 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
651 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
652 return &PL_sv_undef; 964 return &PL_sv_undef;
653} 965}
654 966
655static void 967static void
656decode_he (dec_t *dec, HV *hv) 968decode_he (dec_t *dec, HV *hv)
657{ 969{
658 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
659 // byte or utf-8 strings as keys, but only when !stringref 971 // byte or utf-8 strings as keys, but only when !stringref
660 972
661 if (expect_true (!dec->stringref)) 973 if (ecb_expect_true (!dec->stringref))
662 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
663 { 975 {
664 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
665 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
666 978
979 WANT (len);
667 dec->cur += len; 980 dec->cur += len;
668 981
669 if (ecb_expect_false (dec->stringref))
670 av_push (dec->stringref, newSVpvn (key, len));
671
672 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
673 983
674 return; 984 return;
675 } 985 }
676 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
677 { 987 {
678 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
679 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
680 990
991 WANT (len);
681 dec->cur += len; 992 dec->cur += len;
682 993
683 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
684 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)");
685 997
686 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
687 999
688 return; 1000 return;
689 } 1001 }
690 1002
691 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
692 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
693 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
694 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
695 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
696} 1039}
697 1040
698static SV * 1041static SV *
699decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
700{ 1043{
701 HV *hv = newHV (); 1044 HV *hv = newHV ();
702 1045
703 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
704 1047
705 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
706 { 1049 {
707 ++dec->cur; 1050 ++dec->cur;
708 1051
709 for (;;) 1052 for (;;)
710 { 1053 {
711 WANT (1); 1054 WANT (1);
712 1055
713 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
714 { 1057 {
715 ++dec->cur; 1058 ++dec->cur;
716 break; 1059 break;
717 } 1060 }
718 1061
719 decode_he (dec, hv); 1062 decode_he (dec, hv);
720 } 1063 }
721 } 1064 }
722 else 1065 else
723 { 1066 {
724 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
725 1070
726 while (pairs--) 1071 while (pairs--)
727 decode_he (dec, hv); 1072 decode_he (dec, hv);
728 } 1073 }
729 1074
730 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
731 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
732 1077
733fail: 1078fail:
734 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
735 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
736 return &PL_sv_undef; 1081 return &PL_sv_undef;
737} 1082}
738 1083
739static SV * 1084static SV *
740decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
741{ 1086{
742 SV *sv = 0; 1087 SV *sv = 0;
743 1088
744 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
745 { 1090 {
1091 // indefinite length strings
746 ++dec->cur; 1092 ++dec->cur;
747 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
748 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
749 1097
750 // not very fast, and certainly not robust against illegal input
751 for (;;) 1098 for (;;)
752 { 1099 {
753 WANT (1); 1100 WANT (1);
754 1101
755 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
756 { 1104 {
757 ++dec->cur; 1105 ++dec->cur;
758 break; 1106 break;
759 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
760 1110
761 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;
762 } 1116 }
763 } 1117 }
764 else 1118 else
765 { 1119 {
766 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
767 1121
768 WANT (len); 1122 WANT (len);
769 sv = newSVpvn (dec->cur, len); 1123 sv = newSVpvn (dec->cur, len);
770 dec->cur += len; 1124 dec->cur += len;
1125
1126 if (ecb_expect_false (dec->stringref)
1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
771 } 1129 }
772 1130
773 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
774 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
775 1138 }
776 if (ecb_expect_false (dec->stringref)
777 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
778 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
779 1139
780 return sv; 1140 return sv;
781 1141
782fail: 1142fail:
783 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
802 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
803 break; 1163 break;
804 1164
805 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
806 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
807 ENTER; SAVETMPS; 1169 ENTER;
808 1170
809 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
810 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
811 1173
812 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
813 1175
814 FREETMPS; LEAVE; 1176 LEAVE;
815 } 1177 }
816 break; 1178 break;
817 1179
818 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
819 { 1181 {
820 if ((*dec->cur >> 5) != 0) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
821 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
822 1184
823 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
824 1186
825 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
826 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)");
827 1189
828 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
829 } 1191 }
830 break; 1192 break;
832 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
833 { 1195 {
834 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
835 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
836 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
837 sv = newSV (0); 1201 sv = newSV (0);
838 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1202 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
839 1203
840 SV *osv = decode_sv (dec); 1204 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv); 1205 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv); 1206 SvREFCNT_dec_NN (osv);
1207 }
1208 else
1209 {
1210 av_push (dec->shareable, &PL_sv_undef);
1211 int idx = AvFILLp (dec->shareable);
1212 sv = decode_sv (dec);
1213 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1214 }
843 } 1215 }
844 break; 1216 break;
845 1217
846 case CBOR_TAG_VALUE_SHAREDREF: 1218 case CBOR_TAG_VALUE_SHAREDREF:
847 { 1219 {
848 if ((*dec->cur >> 5) != 0) 1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
850 1222
851 UV idx = decode_uint (dec); 1223 UV idx = decode_uint (dec);
852 1224
853 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
854 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
855 1227
856 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1228 sv = newRV_inc (AvARRAY (dec->shareable)[idx]);
1229
1230 if (sv == &PL_sv_undef)
1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
857 } 1232 }
858 break; 1233 break;
859 1234
860 case CBOR_TAG_PERL_OBJECT: 1235 case CBOR_TAG_PERL_OBJECT:
861 { 1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
862 sv = decode_sv (dec); 1240 sv = decode_sv (dec);
863 1241
864 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
865 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
866 1244
876 if (!method) 1254 if (!method)
877 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1255 ERR ("cannot decode perl-object (package does not have a THAW method)");
878 1256
879 dSP; 1257 dSP;
880 1258
881 ENTER; SAVETMPS; PUSHMARK (SP); 1259 ENTER; SAVETMPS;
1260 PUSHMARK (SP);
882 EXTEND (SP, len + 1); 1261 EXTEND (SP, len + 1);
883 // we re-bless the reference to get overload and other niceties right 1262 // we re-bless the reference to get overload and other niceties right
884 PUSHs (*av_fetch (av, 0, 1)); 1263 PUSHs (*av_fetch (av, 0, 1));
885 PUSHs (sv_cbor); 1264 PUSHs (sv_cbor);
886 1265
894 SPAGAIN; 1273 SPAGAIN;
895 1274
896 if (SvTRUE (ERRSV)) 1275 if (SvTRUE (ERRSV))
897 { 1276 {
898 FREETMPS; LEAVE; 1277 FREETMPS; LEAVE;
899 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1278 ERR_ERRSV;
900 } 1279 }
901 1280
902 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
903 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
904 1283
905 PUTBACK; 1284 PUTBACK;
906 1285
907 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
908 } 1287 }
909 break; 1288 break;
910 1289
911 default: 1290 default:
1291 filter:
912 { 1292 {
1293 SV *tag_sv = newSVuv (tag);
1294
913 sv = decode_sv (dec); 1295 sv = decode_sv (dec);
914 1296
1297 dSP;
1298 ENTER; SAVETMPS;
1299 PUSHMARK (SP);
1300 EXTEND (SP, 2);
1301 PUSHs (tag_sv);
1302 PUSHs (sv);
1303
1304 PUTBACK;
1305 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1306 SPAGAIN;
1307
1308 if (SvTRUE (ERRSV))
1309 {
1310 SvREFCNT_dec_NN (tag_sv);
1311 FREETMPS; LEAVE;
1312 ERR_ERRSV;
1313 }
1314
1315 if (count)
1316 {
1317 SvREFCNT_dec_NN (tag_sv);
1318 SvREFCNT_dec_NN (sv);
1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1321 }
1322 else
1323 {
915 AV *av = newAV (); 1324 AV *av = newAV ();
916 av_push (av, newSVuv (tag)); 1325 av_push (av, tag_sv);
917 av_push (av, sv); 1326 av_push (av, sv);
918 1327
919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1328 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
920 ? cbor_tagged_stash 1329 ? cbor_tagged_stash
921 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1330 : gv_stashpv ("CBOR::XS::Tagged" , 1);
922
923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1331 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1332 }
1333
1334 PUTBACK;
1335
1336 FREETMPS; LEAVE;
924 } 1337 }
925 break; 1338 break;
926 } 1339 }
927 1340
928 return sv; 1341 return sv;
935static SV * 1348static SV *
936decode_sv (dec_t *dec) 1349decode_sv (dec_t *dec)
937{ 1350{
938 WANT (1); 1351 WANT (1);
939 1352
940 switch (*dec->cur >> 5) 1353 switch (*dec->cur >> MAJOR_SHIFT)
941 { 1354 {
942 case 0: // unsigned int 1355 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
943 return newSVuv (decode_uint (dec)); 1356 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
944 case 1: // negative int 1357 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
945 return newSViv (-1 - (IV)decode_uint (dec)); 1358 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
946 case 2: // octet string 1359 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
947 return decode_str (dec, 0); 1360 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
948 case 3: // utf-8 string 1361 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
949 return decode_str (dec, 1); 1362
950 case 4: // array 1363 case MAJOR_MISC >> MAJOR_SHIFT:
951 return decode_av (dec);
952 case 5: // map
953 return decode_hv (dec);
954 case 6: // tag
955 return decode_tagged (dec);
956 case 7: // misc
957 switch (*dec->cur++ & 31) 1364 switch (*dec->cur++ & MINOR_MASK)
958 { 1365 {
959 case 20: 1366 case SIMPLE_FALSE:
960#if CBOR_SLOW 1367#if CBOR_SLOW
961 types_false = get_bool ("Types::Serialiser::false"); 1368 types_false = get_bool ("Types::Serialiser::false");
962#endif 1369#endif
963 return newSVsv (types_false); 1370 return newSVsv (types_false);
964 case 21: 1371 case SIMPLE_TRUE:
965#if CBOR_SLOW 1372#if CBOR_SLOW
966 types_true = get_bool ("Types::Serialiser::true"); 1373 types_true = get_bool ("Types::Serialiser::true");
967#endif 1374#endif
968 return newSVsv (types_true); 1375 return newSVsv (types_true);
969 case 22: 1376 case SIMPLE_NULL:
970 return newSVsv (&PL_sv_undef); 1377 return newSVsv (&PL_sv_undef);
971 case 23: 1378 case SIMPLE_UNDEF:
972#if CBOR_SLOW 1379#if CBOR_SLOW
973 types_error = get_bool ("Types::Serialiser::error"); 1380 types_error = get_bool ("Types::Serialiser::error");
974#endif 1381#endif
975 return newSVsv (types_error); 1382 return newSVsv (types_error);
976 1383
977 case 25: 1384 case MISC_FLOAT16:
978 { 1385 {
979 WANT (2); 1386 WANT (2);
980 1387
981 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1388 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
982 dec->cur += 2; 1389 dec->cur += 2;
983 1390
984 return newSVnv (ecb_binary16_to_float (fp)); 1391 return newSVnv (ecb_binary16_to_float (fp));
985 } 1392 }
986 1393
987 case 26: 1394 case MISC_FLOAT32:
988 { 1395 {
989 uint32_t fp; 1396 uint32_t fp;
990 WANT (4); 1397 WANT (4);
991 memcpy (&fp, dec->cur, 4); 1398 memcpy (&fp, dec->cur, 4);
992 dec->cur += 4; 1399 dec->cur += 4;
995 fp = ecb_bswap32 (fp); 1402 fp = ecb_bswap32 (fp);
996 1403
997 return newSVnv (ecb_binary32_to_float (fp)); 1404 return newSVnv (ecb_binary32_to_float (fp));
998 } 1405 }
999 1406
1000 case 27: 1407 case MISC_FLOAT64:
1001 { 1408 {
1002 uint64_t fp; 1409 uint64_t fp;
1003 WANT (8); 1410 WANT (8);
1004 memcpy (&fp, dec->cur, 8); 1411 memcpy (&fp, dec->cur, 8);
1005 dec->cur += 8; 1412 dec->cur += 8;
1008 fp = ecb_bswap64 (fp); 1415 fp = ecb_bswap64 (fp);
1009 1416
1010 return newSVnv (ecb_binary64_to_double (fp)); 1417 return newSVnv (ecb_binary64_to_double (fp));
1011 } 1418 }
1012 1419
1013 // 0..19 unassigned 1420 // 0..19 unassigned simple
1014 // 24 reserved + unassigned (reserved values are not encodable) 1421 // 24 reserved + unassigned simple (reserved values are not encodable)
1422 // 28-30 unassigned misc
1423 // 31 break code
1015 default: 1424 default:
1016 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1425 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1017 } 1426 }
1018 1427
1019 break; 1428 break;
1020 } 1429 }
1021 1430
1024} 1433}
1025 1434
1026static SV * 1435static SV *
1027decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1436decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1028{ 1437{
1029 dec_t dec = { }; 1438 dec_t dec = { 0 };
1030 SV *sv; 1439 SV *sv;
1031 STRLEN len; 1440 STRLEN len;
1032 char *data = SvPVbyte (string, len); 1441 char *data = SvPVbyte (string, len);
1033 1442
1034 if (len > cbor->max_size && cbor->max_size) 1443 if (len > cbor->max_size && cbor->max_size)
1048 if (dec.cur != dec.end && !dec.err) 1457 if (dec.cur != dec.end && !dec.err)
1049 dec.err = "garbage after CBOR object"; 1458 dec.err = "garbage after CBOR object";
1050 1459
1051 if (dec.err) 1460 if (dec.err)
1052 { 1461 {
1462 if (dec.shareable)
1463 {
1464 // need to break cyclic links, which would all be in shareable
1465 int i;
1466 SV **svp;
1467
1468 for (i = av_len (dec.shareable) + 1; i--; )
1469 if ((svp = av_fetch (dec.shareable, i, 0)))
1470 sv_setsv (*svp, &PL_sv_undef);
1471 }
1472
1053 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
1054 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1478 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1055 } 1479 }
1056 1480
1057 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
1058 1482
1059 return sv; 1483 return sv;
1060} 1484}
1061 1485
1486/////////////////////////////////////////////////////////////////////////////
1487// incremental parser
1488
1489#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1490
1491// returns 0 for notyet, 1 for success or error
1492static int
1493incr_parse (CBOR *self, SV *cborstr)
1494{
1495 STRLEN cur;
1496 SvPV (cborstr, cur);
1497
1498 while (ecb_expect_true (self->incr_need <= cur))
1499 {
1500 // table of integer count bytes
1501 static I8 incr_len[MINOR_MASK + 1] = {
1502 0, 0, 0, 0, 0, 0, 0, 0,
1503 0, 0, 0, 0, 0, 0, 0, 0,
1504 0, 0, 0, 0, 0, 0, 0, 0,
1505 1, 2, 4, 8,-1,-1,-1,-2
1506 };
1507
1508 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1509 U8 m = *p & MINOR_MASK;
1510 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1511 I8 ilen = incr_len[m];
1512
1513 self->incr_need = self->incr_pos + 1;
1514
1515 if (ecb_expect_false (ilen < 0))
1516 {
1517 if (m != MINOR_INDEF)
1518 return 1; // error
1519
1520 if (*p == (MAJOR_MISC | MINOR_INDEF))
1521 {
1522 if (count >= 0)
1523 return 1; // error
1524
1525 count = 1;
1526 }
1527 else
1528 {
1529 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1530 count = -1;
1531 }
1532 }
1533 else
1534 {
1535 self->incr_need += ilen;
1536 if (ecb_expect_false (self->incr_need > cur))
1537 return 0;
1538
1539 int major = *p >> MAJOR_SHIFT;
1540
1541 switch (major)
1542 {
1543 case MAJOR_TAG >> MAJOR_SHIFT:
1544 ++count; // tags merely prefix another value
1545 break;
1546
1547 case MAJOR_BYTES >> MAJOR_SHIFT:
1548 case MAJOR_TEXT >> MAJOR_SHIFT:
1549 case MAJOR_ARRAY >> MAJOR_SHIFT:
1550 case MAJOR_MAP >> MAJOR_SHIFT:
1551 {
1552 UV len;
1553
1554 if (ecb_expect_false (ilen))
1555 {
1556 len = 0;
1557
1558 do {
1559 len = (len << 8) | *++p;
1560 } while (--ilen);
1561 }
1562 else
1563 len = m;
1564
1565 switch (major)
1566 {
1567 case MAJOR_BYTES >> MAJOR_SHIFT:
1568 case MAJOR_TEXT >> MAJOR_SHIFT:
1569 self->incr_need += len;
1570 if (ecb_expect_false (self->incr_need > cur))
1571 return 0;
1572
1573 break;
1574
1575 case MAJOR_MAP >> MAJOR_SHIFT:
1576 len <<= 1;
1577 /* FALLTHROUGH */
1578 case MAJOR_ARRAY >> MAJOR_SHIFT:
1579 if (len)
1580 {
1581 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1582 count = len + 1;
1583 }
1584 break;
1585 }
1586 }
1587 }
1588 }
1589
1590 self->incr_pos = self->incr_need;
1591
1592 if (count > 0)
1593 {
1594 while (!--count)
1595 {
1596 if (!AvFILLp (self->incr_count))
1597 return 1; // done
1598
1599 SvREFCNT_dec_NN (av_pop (self->incr_count));
1600 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1601 }
1602
1603 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1604 }
1605 }
1606
1607 return 0;
1608}
1609
1610
1062///////////////////////////////////////////////////////////////////////////// 1611/////////////////////////////////////////////////////////////////////////////
1063// XS interface functions 1612// XS interface functions
1064 1613
1065MODULE = CBOR::XS PACKAGE = CBOR::XS 1614MODULE = CBOR::XS PACKAGE = CBOR::XS
1066 1615
1074 1623
1075 types_true = get_bool ("Types::Serialiser::true" ); 1624 types_true = get_bool ("Types::Serialiser::true" );
1076 types_false = get_bool ("Types::Serialiser::false"); 1625 types_false = get_bool ("Types::Serialiser::false");
1077 types_error = get_bool ("Types::Serialiser::error"); 1626 types_error = get_bool ("Types::Serialiser::error");
1078 1627
1628 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1629
1079 sv_cbor = newSVpv ("CBOR", 0); 1630 sv_cbor = newSVpv ("CBOR", 0);
1080 SvREADONLY_on (sv_cbor); 1631 SvREADONLY_on (sv_cbor);
1632
1633 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1081} 1634}
1082 1635
1083PROTOTYPES: DISABLE 1636PROTOTYPES: DISABLE
1084 1637
1085void CLONE (...) 1638void CLONE (...)
1104void shrink (CBOR *self, int enable = 1) 1657void shrink (CBOR *self, int enable = 1)
1105 ALIAS: 1658 ALIAS:
1106 shrink = F_SHRINK 1659 shrink = F_SHRINK
1107 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING 1661 allow_sharing = F_ALLOW_SHARING
1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1109 dedup_strings = F_DEDUP_STRINGS 1664 pack_strings = F_PACK_STRINGS
1665 text_keys = F_TEXT_KEYS
1666 text_strings = F_TEXT_STRINGS
1667 validate_utf8 = F_VALIDATE_UTF8
1110 PPCODE: 1668 PPCODE:
1111{ 1669{
1112 if (enable) 1670 if (enable)
1113 self->flags |= ix; 1671 self->flags |= ix;
1114 else 1672 else
1120void get_shrink (CBOR *self) 1678void get_shrink (CBOR *self)
1121 ALIAS: 1679 ALIAS:
1122 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
1123 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING 1682 get_allow_sharing = F_ALLOW_SHARING
1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1125 get_dedup_strings = F_DEDUP_STRINGS 1685 get_pack_strings = F_PACK_STRINGS
1686 get_text_keys = F_TEXT_KEYS
1687 get_text_strings = F_TEXT_STRINGS
1688 get_validate_utf8 = F_VALIDATE_UTF8
1126 PPCODE: 1689 PPCODE:
1127 XPUSHs (boolSV (self->flags & ix)); 1690 XPUSHs (boolSV (self->flags & ix));
1128 1691
1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1692void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1130 PPCODE: 1693 PPCODE:
1143 XPUSHs (ST (0)); 1706 XPUSHs (ST (0));
1144 1707
1145int get_max_size (CBOR *self) 1708int get_max_size (CBOR *self)
1146 CODE: 1709 CODE:
1147 RETVAL = self->max_size; 1710 RETVAL = self->max_size;
1711 OUTPUT:
1712 RETVAL
1713
1714void filter (CBOR *self, SV *filter = 0)
1715 PPCODE:
1716 SvREFCNT_dec (self->filter);
1717 self->filter = filter ? newSVsv (filter) : filter;
1718 XPUSHs (ST (0));
1719
1720SV *get_filter (CBOR *self)
1721 CODE:
1722 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1148 OUTPUT: 1723 OUTPUT:
1149 RETVAL 1724 RETVAL
1150 1725
1151void encode (CBOR *self, SV *scalar) 1726void encode (CBOR *self, SV *scalar)
1152 PPCODE: 1727 PPCODE:
1167 EXTEND (SP, 2); 1742 EXTEND (SP, 2);
1168 PUSHs (sv); 1743 PUSHs (sv);
1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1744 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1170} 1745}
1171 1746
1747void incr_parse (CBOR *self, SV *cborstr)
1748 ALIAS:
1749 incr_parse_multiple = 1
1750 PPCODE:
1751{
1752 if (SvUTF8 (cborstr))
1753 sv_utf8_downgrade (cborstr, 0);
1754
1755 if (!self->incr_count)
1756 {
1757 self->incr_count = newAV ();
1758 self->incr_pos = 0;
1759 self->incr_need = 1;
1760
1761 av_push (self->incr_count, newSViv (1));
1762 }
1763
1764 do
1765 {
1766 if (!incr_parse (self, cborstr))
1767 {
1768 if (self->incr_need > self->max_size && self->max_size)
1769 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1770 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1771
1772 break;
1773 }
1774
1775 SV *sv;
1776 char *offset;
1777
1778 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1779 XPUSHs (sv);
1780
1781 sv_chop (cborstr, offset);
1782
1783 av_clear (self->incr_count);
1784 av_push (self->incr_count, newSViv (1));
1785
1786 self->incr_pos = 0;
1787 self->incr_need = self->incr_pos + 1;
1788 }
1789 while (ix);
1790}
1791
1792void incr_reset (CBOR *self)
1793 CODE:
1794{
1795 SvREFCNT_dec (self->incr_count);
1796 self->incr_count = 0;
1797}
1798
1799void DESTROY (CBOR *self)
1800 PPCODE:
1801 cbor_free (self);
1802
1172PROTOTYPES: ENABLE 1803PROTOTYPES: ENABLE
1173 1804
1174void encode_cbor (SV *scalar) 1805void encode_cbor (SV *scalar)
1806 ALIAS:
1807 encode_cbor = 0
1808 encode_cbor_sharing = F_ALLOW_SHARING
1175 PPCODE: 1809 PPCODE:
1176{ 1810{
1177 CBOR cbor; 1811 CBOR cbor;
1178 cbor_init (&cbor); 1812 cbor_init (&cbor);
1813 cbor.flags |= ix;
1179 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1814 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1180 XPUSHs (scalar); 1815 XPUSHs (scalar);
1181} 1816}
1182 1817
1183void decode_cbor (SV *cborstr) 1818void decode_cbor (SV *cborstr)
1187 cbor_init (&cbor); 1822 cbor_init (&cbor);
1188 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1189 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1190} 1825}
1191 1826
1827#ifdef __AFL_COMPILER
1828
1829void
1830afl_init ()
1831 CODE:
1832 __AFL_INIT ();
1833
1834int
1835afl_loop (unsigned int count = 10000)
1836 CODE:
1837 RETVAL = __AFL_LOOP (count);
1838 OUTPUT:
1839 RETVAL
1840
1841#endif
1842

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines