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 1.25 by root, Fri Nov 22 05:54:07 2013 UTC vs.
Revision 1.71 by root, Fri Mar 19 17:30:27 2021 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// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
24 69
25// known tags 70// known tags
26enum cbor_tag 71enum cbor_tag
27{ 72{
28 // inofficial extensions (pending iana registration) 73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 79 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 80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 81
37 // rfc7049 82 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
44 89
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 94
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 100
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
57}; 116};
58 117
59#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 120#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
123#define F_PACK_STRINGS 0x00000020UL
124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
63 127
64#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
65 129
66#define SB do { 130#define SB do {
67#define SE } while (0) 131#define SE } while (0)
79# define CBOR_SLOW 0 143# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
81#endif 145#endif
82 146
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 147static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 149
86typedef struct { 150typedef struct {
87 U32 flags; 151 U32 flags;
88 U32 max_depth; 152 U32 max_depth;
89 STRLEN max_size; 153 STRLEN max_size;
154 SV *filter;
155
156 // for the incremental parser
157 STRLEN incr_pos; // the current offset into the text
158 STRLEN incr_need; // minimum bytes needed to decode
159 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
90} CBOR; 160} CBOR;
91 161
92ecb_inline void 162ecb_inline void
93cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
94{ 164{
95 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
96 cbor->max_depth = 512; 166 cbor->max_depth = 512;
167}
168
169ecb_inline void
170cbor_free (CBOR *cbor)
171{
172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
97} 174}
98 175
99///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
100// utility functions 177// utility functions
101 178
124#endif 201#endif
125 } 202 }
126} 203}
127 204
128// minimum length of a string to be registered for stringref 205// minimum length of a string to be registered for stringref
129ecb_inline int 206ecb_inline STRLEN
130minimum_string_length (UV idx) 207minimum_string_length (UV idx)
131{ 208{
132 return idx > 23 209 return idx <= 23 ? 3
133 ? idx > 0xffU 210 : idx <= 0xffU ? 4
134 ? idx > 0xffffU 211 : idx <= 0xffffU ? 5
135 ? idx > 0xffffffffU 212 : idx <= 0xffffffffU ? 7
136 ? 7 213 : 11;
137 : 6
138 : 5
139 : 4
140 : 3;
141} 214}
142 215
143///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
144// encoder 217// encoder
145 218
158} enc_t; 231} enc_t;
159 232
160ecb_inline void 233ecb_inline void
161need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
162{ 235{
163 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
164 { 237 {
165 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
166 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
167 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
168 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
169 } 242 }
170} 243}
171 244
245static void encode_sv (enc_t *enc, SV *sv);
246
172ecb_inline void 247ecb_inline void
173encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
174{ 249{
175 need (enc, 1); 250 need (enc, 1);
176 *enc->cur++ = ch; 251 *enc->cur++ = ch;
177} 252}
178 253
254// used for tags, intregers, element counts and so on
179static void 255static void
180encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
181{ 257{
182 need (enc, 9); 258 need (enc, 9);
183 259
184 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
186 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
187 { 263 {
188 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 265 *enc->cur++ = len;
190 } 266 }
191 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
192 { 268 {
193 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 271 *enc->cur++ = len;
196 } 272 }
197 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
198 { 274 {
199 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 279 *enc->cur++ = len;
204 } 280 }
205 else 281 else
206 { 282 {
207 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
214 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
215 *enc->cur++ = len; 291 *enc->cur++ = len;
216 } 292 }
217} 293}
218 294
295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
219ecb_inline void 307ecb_inline void
220encode_tag (enc_t *enc, UV tag) 308encode_tag (enc_t *enc, UV tag)
221{ 309{
222 encode_uint (enc, 0xc0, tag); 310 encode_uint (enc, MAJOR_TAG, tag);
223} 311}
224 312
225static void 313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 316{
317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
336ecb_inline void
337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
347 need (enc, len);
348 memcpy (enc->cur, str, len);
349 enc->cur += len;
350}
351
352ecb_inline void
353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
354{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 356 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 358
232 if (SvOK (*svp)) 359 if (SvOK (*svp))
233 { 360 {
234 // already registered, use stringref 361 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 362 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 363 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 364 return;
238 } 365 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 366 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 367 {
241 // register only 368 // register only
242 sv_setuv (*svp, enc->stringref_idx); 369 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 370 ++enc->stringref_idx;
244 } 371 }
245 } 372 }
246 373
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 374 encode_str (enc, upgrade_utf8, utf8, str, len);
375}
376
377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
248 need (enc, len); 380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
249 memcpy (enc->cur, str, len); 389 memcpy (enc->cur, &fp, 2);
250 enc->cur += len; 390 enc->cur += 2;
251} 391}
252 392
253static void encode_sv (enc_t *enc, SV *sv); 393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
254 507
255static void 508static void
256encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
257{ 510{
258 int i, len = av_len (av); 511 int i, len = av_len (av);
260 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
262 515
263 ++enc->depth; 516 ++enc->depth;
264 517
265 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 519
520 if (ecb_expect_false (SvMAGICAL (av)))
267 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
268 { 522 {
269 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 525 }
526 else
527 for (i = 0; i <= len; ++i)
528 {
529 SV *sv = AvARRAY (av)[i];
530 encode_sv (enc, sv ? sv : &PL_sv_undef);
531 }
272 532
273 --enc->depth; 533 --enc->depth;
274}
275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279} 534}
280 535
281static void 536static void
282encode_hv (enc_t *enc, HV *hv) 537encode_hv (enc_t *enc, HV *hv)
283{ 538{
289 ++enc->depth; 544 ++enc->depth;
290 545
291 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
292 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
293 548
294 if (mg) 549 if (ecb_expect_false (mg))
295 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
296 else 551 else
297 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
298 553
299 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
300 { 555 {
301 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
302 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
303 else 558 else
304 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
305 560
306 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
307 } 562 }
308 563
309 if (mg) 564 if (ecb_expect_false (mg))
310 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
311 566
312 --enc->depth; 567 --enc->depth;
313} 568}
314 569
315// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
316static void 571static void
317encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
318{ 573{
319 SvGETMAGIC (sv); 574 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 }
342 575
343 svtype svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
344 577
345 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
346 { 579 {
353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
354 ? cbor_tagged_stash 587 ? cbor_tagged_stash
355 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
356 589
357 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
358 GV *method;
359 591
360 if (stash == boolean_stash) 592 if (stash == boolean_stash)
361 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
362 else if (stash == error_stash) 597 else if (stash == error_stash)
363 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
364 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
365 { 603 {
366 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
367 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
368 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
370 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
623
624 return;
625 }
626 }
627
628 if (ecb_expect_false (SvREFCNT (sv) > 1)
629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
630 {
631 if (ecb_expect_false (!enc->shareable))
632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
633
634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
635
636 if (SvOK (*svp))
371 } 637 {
638 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
639 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
640 return;
641 }
642 else
643 {
644 sv_setuv (*svp, enc->shareable_idx);
645 ++enc->shareable_idx;
646 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
647 }
648 }
649
650 if (ecb_expect_false (SvOBJECT (sv)))
651 {
652 HV *stash = SvSTASH (sv);
653 GV *method;
654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
372 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
373 { 659 {
374 dSP; 660 dSP;
375 661
376 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
377 // we re-bless the reference to get overload and other niceties right 664 // we re-bless the reference to get overload and other niceties right
378 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
379 666
380 PUTBACK; 667 PUTBACK;
381 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
394 } 681 }
395 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
396 { 683 {
397 dSP; 684 dSP;
398 685
399 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
400 EXTEND (SP, 2); 688 EXTEND (SP, 2);
401 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
402 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
403 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
404 692
409 // catch this surprisingly common error 697 // catch this surprisingly common error
410 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 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)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
412 700
413 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
414 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
415 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
416 704
417 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
418 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
419 713
420 PUTBACK; 714 PUTBACK;
421 715
422 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
423 } 717 }
441{ 735{
442 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
443 737
444 need (enc, 9); 738 need (enc, 9);
445 739
446 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
447 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
448 //TODO: maybe I32? 742 //TODO: maybe I32?
449 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
450 { 744 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 745 else
462 { 746 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} 747}
474 748
475static void 749static void
476encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
477{ 751{
479 753
480 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
481 { 755 {
482 STRLEN len; 756 STRLEN len;
483 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
484 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
485 } 759 }
486 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
487 encode_nv (enc, sv); 761 encode_nv (enc, sv);
488 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
489 { 763 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)) 764 else if (SvROK (sv))
498 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
499 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
500 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
501 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
502 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
503 else 770 else
504 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 771 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
505 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
506} 773}
507 774
508static SV * 775static SV *
509encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
510{ 777{
511 enc_t enc = { }; 778 enc_t enc = { 0 };
512 779
513 enc.cbor = *cbor; 780 enc.cbor = *cbor;
514 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
515 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
516 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
517 784
518 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
519 786
520 if (cbor->flags & F_ALLOW_STRINGREF) 787 if (cbor->flags & F_PACK_STRINGS)
521 { 788 {
522 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
523 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
524 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
525 } 792 }
547 CBOR cbor; 814 CBOR cbor;
548 U32 depth; // recursion depth 815 U32 depth; // recursion depth
549 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
550 AV *shareable; 817 AV *shareable;
551 AV *stringref; 818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
552} dec_t; 821} dec_t;
553 822
554#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
555 830
556#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
557 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
558#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#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 860#define DEC_DEC_DEPTH --dec->depth
560 861
561static UV 862static UV
562decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
563{ 864{
564 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
565 { 866 ++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;
570 867
571 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
572 WANT (2); 872 WANT (1);
573 dec->cur += 2; 873 dec->cur += 1;
574 return dec->cur[-1]; 874 return dec->cur[-1];
575 875 }
576 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
577 WANT (3); 878 WANT (2);
578 dec->cur += 3; 879 dec->cur += 2;
579 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
580 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
581 882 }
582 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
583 WANT (5); 885 WANT (4);
584 dec->cur += 5; 886 dec->cur += 4;
585 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
586 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
587 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
588 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
589 891 }
590 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
591 WANT (9); 894 WANT (8);
592 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
593 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
594 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
595 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
596 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
597 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
601 910 }
602 default: 911 else
603 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
604 }
605 913
606fail: 914fail:
607 return 0; 915 return 0;
608} 916}
609 917
614{ 922{
615 AV *av = newAV (); 923 AV *av = newAV ();
616 924
617 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
618 926
619 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
620 { 928 {
621 ++dec->cur; 929 ++dec->cur;
622 930
623 for (;;) 931 for (;;)
624 { 932 {
625 WANT (1); 933 WANT (1);
626 934
627 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
628 { 936 {
629 ++dec->cur; 937 ++dec->cur;
630 break; 938 break;
631 } 939 }
632 940
633 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
634 } 942 }
635 } 943 }
636 else 944 else
637 { 945 {
638 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
639 947
948 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); 949 av_fill (av, len - 1);
641 950
642 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
643 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
644 } 953 }
645 954
646 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
647 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
648 957
649fail: 958fail:
650 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
651 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
652 return &PL_sv_undef; 961 return &PL_sv_undef;
653} 962}
654 963
655static void 964static void
657{ 966{
658 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
659 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
660 969
661 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
662 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
663 { 972 {
664 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
665 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
666 975
976 WANT (len);
667 dec->cur += len; 977 dec->cur += len;
668 978
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); 979 hv_store (hv, key, len, decode_sv (dec), 0);
673 980
674 return; 981 return;
675 } 982 }
676 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
677 { 984 {
678 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
679 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
680 987
988 WANT (len);
681 dec->cur += len; 989 dec->cur += len;
682 990
683 if (ecb_expect_false (dec->stringref)) 991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
684 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 992 if (!is_utf8_string ((U8 *)key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
685 994
686 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
687 996
688 return; 997 return;
689 } 998 }
690 999
691 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
692 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
693 1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
694 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
695 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
1033
1034fail:
1035 ;
696} 1036}
697 1037
698static SV * 1038static SV *
699decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
700{ 1040{
701 HV *hv = newHV (); 1041 HV *hv = newHV ();
702 1042
703 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
704 1044
705 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
706 { 1046 {
707 ++dec->cur; 1047 ++dec->cur;
708 1048
709 for (;;) 1049 for (;;)
710 { 1050 {
711 WANT (1); 1051 WANT (1);
712 1052
713 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
714 { 1054 {
715 ++dec->cur; 1055 ++dec->cur;
716 break; 1056 break;
717 } 1057 }
718 1058
719 decode_he (dec, hv); 1059 decode_he (dec, hv);
720 } 1060 }
721 } 1061 }
722 else 1062 else
723 { 1063 {
724 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
725 1067
726 while (pairs--) 1068 while (pairs--)
727 decode_he (dec, hv); 1069 decode_he (dec, hv);
728 } 1070 }
729 1071
730 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
731 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
732 1074
733fail: 1075fail:
734 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
735 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
736 return &PL_sv_undef; 1078 return &PL_sv_undef;
737} 1079}
738 1080
739static SV * 1081static SV *
740decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
741{ 1083{
742 SV *sv = 0; 1084 SV *sv = 0;
743 1085
744 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
745 { 1087 {
1088 // indefinite length strings
746 ++dec->cur; 1089 ++dec->cur;
747 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
748 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
749 1094
750 // not very fast, and certainly not robust against illegal input
751 for (;;) 1095 for (;;)
752 { 1096 {
753 WANT (1); 1097 WANT (1);
754 1098
755 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
756 { 1101 {
757 ++dec->cur; 1102 ++dec->cur;
758 break; 1103 break;
759 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
760 1107
761 sv_catsv (sv, decode_sv (dec)); 1108 STRLEN len = decode_uint (dec);
1109
1110 WANT (len);
1111 sv_catpvn (sv, dec->cur, len);
1112 dec->cur += len;
762 } 1113 }
763 } 1114 }
764 else 1115 else
765 { 1116 {
766 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
773 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
774 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
775 } 1126 }
776 1127
777 if (utf8) 1128 if (utf8)
1129 {
1130 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1131 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1132 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1133
778 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
779 1136
780 return sv; 1137 return sv;
781 1138
782fail: 1139fail:
783 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
802 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
803 break; 1160 break;
804 1161
805 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
806 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
807 ENTER; SAVETMPS; 1166 ENTER;
808 1167
809 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
810 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
811 1170
812 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
813 1172
814 FREETMPS; LEAVE; 1173 LEAVE;
815 } 1174 }
816 break; 1175 break;
817 1176
818 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
819 { 1178 {
820 if ((*dec->cur >> 5) != 0) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
821 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
822 1181
823 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
824 1183
825 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
826 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
827 1186
828 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
829 } 1188 }
830 break; 1189 break;
832 case CBOR_TAG_VALUE_SHAREABLE: 1191 case CBOR_TAG_VALUE_SHAREABLE:
833 { 1192 {
834 if (ecb_expect_false (!dec->shareable)) 1193 if (ecb_expect_false (!dec->shareable))
835 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
836 1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
837 sv = newSV (0); 1198 sv = newSV (0);
838 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
839 1200
840 SV *osv = decode_sv (dec); 1201 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv); 1202 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv); 1203 SvREFCNT_dec_NN (osv);
1204 }
1205 else
1206 {
1207 av_push (dec->shareable, &PL_sv_undef);
1208 int idx = AvFILLp (dec->shareable);
1209 sv = decode_sv (dec);
1210 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1211 }
843 } 1212 }
844 break; 1213 break;
845 1214
846 case CBOR_TAG_VALUE_SHAREDREF: 1215 case CBOR_TAG_VALUE_SHAREDREF:
847 { 1216 {
848 if ((*dec->cur >> 5) != 0) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
850 1219
851 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
852 1221
853 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
854 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
855 1224
856 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1226
1227 if (sv == &PL_sv_undef)
1228 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
857 } 1229 }
858 break; 1230 break;
859 1231
860 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
861 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
862 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
863 1238
864 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
865 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
866 1241
876 if (!method) 1251 if (!method)
877 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1252 ERR ("cannot decode perl-object (package does not have a THAW method)");
878 1253
879 dSP; 1254 dSP;
880 1255
881 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
882 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
883 // we re-bless the reference to get overload and other niceties right 1259 // we re-bless the reference to get overload and other niceties right
884 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
885 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
886 1262
894 SPAGAIN; 1270 SPAGAIN;
895 1271
896 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
897 { 1273 {
898 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
899 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
900 } 1276 }
901 1277
902 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
903 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
904 1280
905 PUTBACK; 1281 PUTBACK;
906 1282
907 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
908 } 1284 }
909 break; 1285 break;
910 1286
911 default: 1287 default:
1288 filter:
912 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
913 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
914 1293
1294 dSP;
1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1297 EXTEND (SP, 2);
1298 PUSHs (tag_sv);
1299 PUSHs (sv);
1300
1301 PUTBACK;
1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1303 SPAGAIN;
1304
1305 if (SvTRUE (ERRSV))
1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1308 FREETMPS; LEAVE;
1309 ERR_ERRSV;
1310 }
1311
1312 if (count)
1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1315 SvREFCNT_dec_NN (sv);
1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1318 }
1319 else
1320 {
915 AV *av = newAV (); 1321 AV *av = newAV ();
916 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
917 av_push (av, sv); 1323 av_push (av, sv);
918 1324
919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
920 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
921 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
922
923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1328 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1329 }
1330
1331 PUTBACK;
1332
1333 FREETMPS; LEAVE;
924 } 1334 }
925 break; 1335 break;
926 } 1336 }
927 1337
928 return sv; 1338 return sv;
935static SV * 1345static SV *
936decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
937{ 1347{
938 WANT (1); 1348 WANT (1);
939 1349
940 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
941 { 1351 {
942 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
943 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
944 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
945 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
946 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
947 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
948 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
949 return decode_str (dec, 1); 1359
950 case 4: // array 1360 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) 1361 switch (*dec->cur++ & MINOR_MASK)
958 { 1362 {
959 case 20: 1363 case SIMPLE_FALSE:
960#if CBOR_SLOW 1364#if CBOR_SLOW
961 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
962#endif 1366#endif
963 return newSVsv (types_false); 1367 return newSVsv (types_false);
964 case 21: 1368 case SIMPLE_TRUE:
965#if CBOR_SLOW 1369#if CBOR_SLOW
966 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
967#endif 1371#endif
968 return newSVsv (types_true); 1372 return newSVsv (types_true);
969 case 22: 1373 case SIMPLE_NULL:
970 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
971 case 23: 1375 case SIMPLE_UNDEF:
972#if CBOR_SLOW 1376#if CBOR_SLOW
973 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
974#endif 1378#endif
975 return newSVsv (types_error); 1379 return newSVsv (types_error);
976 1380
977 case 25: 1381 case MISC_FLOAT16:
978 { 1382 {
979 WANT (2); 1383 WANT (2);
980 1384
981 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
982 dec->cur += 2; 1386 dec->cur += 2;
983 1387
984 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
985 } 1389 }
986 1390
987 case 26: 1391 case MISC_FLOAT32:
988 { 1392 {
989 uint32_t fp; 1393 uint32_t fp;
990 WANT (4); 1394 WANT (4);
991 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
992 dec->cur += 4; 1396 dec->cur += 4;
995 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
996 1400
997 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
998 } 1402 }
999 1403
1000 case 27: 1404 case MISC_FLOAT64:
1001 { 1405 {
1002 uint64_t fp; 1406 uint64_t fp;
1003 WANT (8); 1407 WANT (8);
1004 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
1005 dec->cur += 8; 1409 dec->cur += 8;
1008 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
1009 1413
1010 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
1011 } 1415 }
1012 1416
1013 // 0..19 unassigned 1417 // 0..19 unassigned simple
1014 // 24 reserved + unassigned (reserved values are not encodable) 1418 // 24 reserved + unassigned simple (reserved values are not encodable)
1419 // 28-30 unassigned misc
1420 // 31 break code
1015 default: 1421 default:
1016 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1017 } 1423 }
1018 1424
1019 break; 1425 break;
1020 } 1426 }
1021 1427
1024} 1430}
1025 1431
1026static SV * 1432static SV *
1027decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1028{ 1434{
1029 dec_t dec = { }; 1435 dec_t dec = { 0 };
1030 SV *sv; 1436 SV *sv;
1031 STRLEN len; 1437 STRLEN len;
1032 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
1033 1439
1034 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
1048 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
1049 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
1050 1456
1051 if (dec.err) 1457 if (dec.err)
1052 { 1458 {
1459 if (dec.shareable)
1460 {
1461 // need to break cyclic links, which would all be in shareable
1462 int i;
1463 SV **svp;
1464
1465 for (i = av_len (dec.shareable) + 1; i--; )
1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1467 sv_setsv (*svp, &PL_sv_undef);
1468 }
1469
1053 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1054 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1055 } 1476 }
1056 1477
1057 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1058 1479
1059 return sv; 1480 return sv;
1060} 1481}
1061 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 /* FALLTHROUGH */
1575 case MAJOR_ARRAY >> MAJOR_SHIFT:
1576 if (len)
1577 {
1578 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1579 count = len + 1;
1580 }
1581 break;
1582 }
1583 }
1584 }
1585 }
1586
1587 self->incr_pos = self->incr_need;
1588
1589 if (count > 0)
1590 {
1591 while (!--count)
1592 {
1593 if (!AvFILLp (self->incr_count))
1594 return 1; // done
1595
1596 SvREFCNT_dec_NN (av_pop (self->incr_count));
1597 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1598 }
1599
1600 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1601 }
1602 }
1603
1604 return 0;
1605}
1606
1607
1062///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
1063// XS interface functions 1609// XS interface functions
1064 1610
1065MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
1066 1612
1074 1620
1075 types_true = get_bool ("Types::Serialiser::true" ); 1621 types_true = get_bool ("Types::Serialiser::true" );
1076 types_false = get_bool ("Types::Serialiser::false"); 1622 types_false = get_bool ("Types::Serialiser::false");
1077 types_error = get_bool ("Types::Serialiser::error"); 1623 types_error = get_bool ("Types::Serialiser::error");
1078 1624
1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1626
1079 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
1080 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1081} 1631}
1082 1632
1083PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
1084 1634
1085void CLONE (...) 1635void CLONE (...)
1104void shrink (CBOR *self, int enable = 1) 1654void shrink (CBOR *self, int enable = 1)
1105 ALIAS: 1655 ALIAS:
1106 shrink = F_SHRINK 1656 shrink = F_SHRINK
1107 allow_unknown = F_ALLOW_UNKNOWN 1657 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING 1658 allow_sharing = F_ALLOW_SHARING
1109 allow_stringref = F_ALLOW_STRINGREF 1659 allow_cycles = F_ALLOW_CYCLES
1660 forbid_objects = F_FORBID_OBJECTS
1661 pack_strings = F_PACK_STRINGS
1662 text_keys = F_TEXT_KEYS
1663 text_strings = F_TEXT_STRINGS
1664 validate_utf8 = F_VALIDATE_UTF8
1110 PPCODE: 1665 PPCODE:
1111{ 1666{
1112 if (enable) 1667 if (enable)
1113 self->flags |= ix; 1668 self->flags |= ix;
1114 else 1669 else
1120void get_shrink (CBOR *self) 1675void get_shrink (CBOR *self)
1121 ALIAS: 1676 ALIAS:
1122 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
1123 get_allow_unknown = F_ALLOW_UNKNOWN 1678 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING 1679 get_allow_sharing = F_ALLOW_SHARING
1125 get_allow_stringref = F_ALLOW_STRINGREF 1680 get_allow_cycles = F_ALLOW_CYCLES
1681 get_forbid_objects = F_FORBID_OBJECTS
1682 get_pack_strings = F_PACK_STRINGS
1683 get_text_keys = F_TEXT_KEYS
1684 get_text_strings = F_TEXT_STRINGS
1685 get_validate_utf8 = F_VALIDATE_UTF8
1126 PPCODE: 1686 PPCODE:
1127 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
1128 1688
1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1130 PPCODE: 1690 PPCODE:
1143 XPUSHs (ST (0)); 1703 XPUSHs (ST (0));
1144 1704
1145int get_max_size (CBOR *self) 1705int get_max_size (CBOR *self)
1146 CODE: 1706 CODE:
1147 RETVAL = self->max_size; 1707 RETVAL = self->max_size;
1708 OUTPUT:
1709 RETVAL
1710
1711void filter (CBOR *self, SV *filter = 0)
1712 PPCODE:
1713 SvREFCNT_dec (self->filter);
1714 self->filter = filter ? newSVsv (filter) : filter;
1715 XPUSHs (ST (0));
1716
1717SV *get_filter (CBOR *self)
1718 CODE:
1719 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1148 OUTPUT: 1720 OUTPUT:
1149 RETVAL 1721 RETVAL
1150 1722
1151void encode (CBOR *self, SV *scalar) 1723void encode (CBOR *self, SV *scalar)
1152 PPCODE: 1724 PPCODE:
1167 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1168 PUSHs (sv); 1740 PUSHs (sv);
1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1170} 1742}
1171 1743
1744void incr_parse (CBOR *self, SV *cborstr)
1745 ALIAS:
1746 incr_parse_multiple = 1
1747 PPCODE:
1748{
1749 if (SvUTF8 (cborstr))
1750 sv_utf8_downgrade (cborstr, 0);
1751
1752 if (!self->incr_count)
1753 {
1754 self->incr_count = newAV ();
1755 self->incr_pos = 0;
1756 self->incr_need = 1;
1757
1758 av_push (self->incr_count, newSViv (1));
1759 }
1760
1761 do
1762 {
1763 if (!incr_parse (self, cborstr))
1764 {
1765 if (self->incr_need > self->max_size && self->max_size)
1766 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1767 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1768
1769 break;
1770 }
1771
1772 SV *sv;
1773 char *offset;
1774
1775 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1776 XPUSHs (sv);
1777
1778 sv_chop (cborstr, offset);
1779
1780 av_clear (self->incr_count);
1781 av_push (self->incr_count, newSViv (1));
1782
1783 self->incr_pos = 0;
1784 self->incr_need = self->incr_pos + 1;
1785 }
1786 while (ix);
1787}
1788
1789void incr_reset (CBOR *self)
1790 CODE:
1791{
1792 SvREFCNT_dec (self->incr_count);
1793 self->incr_count = 0;
1794}
1795
1796void DESTROY (CBOR *self)
1797 PPCODE:
1798 cbor_free (self);
1799
1172PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
1173 1801
1174void encode_cbor (SV *scalar) 1802void encode_cbor (SV *scalar)
1803 ALIAS:
1804 encode_cbor = 0
1805 encode_cbor_sharing = F_ALLOW_SHARING
1175 PPCODE: 1806 PPCODE:
1176{ 1807{
1177 CBOR cbor; 1808 CBOR cbor;
1178 cbor_init (&cbor); 1809 cbor_init (&cbor);
1810 cbor.flags |= ix;
1179 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1811 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1180 XPUSHs (scalar); 1812 XPUSHs (scalar);
1181} 1813}
1182 1814
1183void decode_cbor (SV *cborstr) 1815void decode_cbor (SV *cborstr)
1187 cbor_init (&cbor); 1819 cbor_init (&cbor);
1188 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1189 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1190} 1822}
1191 1823
1824#ifdef __AFL_COMPILER
1825
1826void
1827afl_init ()
1828 CODE:
1829 __AFL_INIT ();
1830
1831int
1832afl_loop (unsigned int count = 10000)
1833 CODE:
1834 RETVAL = __AFL_LOOP (count);
1835 OUTPUT:
1836 RETVAL
1837
1838#endif
1839

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines