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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.26 by root, Fri Nov 22 09:40:13 2013 UTC vs.
Revision 1.71 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} 534}
275 535
276static void 536static void
284 ++enc->depth; 544 ++enc->depth;
285 545
286 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
288 548
289 if (mg) 549 if (ecb_expect_false (mg))
290 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 551 else
292 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
293 553
294 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
295 { 555 {
296 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
298 else 558 else
299 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));
300 560
301 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));
302 } 562 }
303 563
304 if (mg) 564 if (ecb_expect_false (mg))
305 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 566
307 --enc->depth; 567 --enc->depth;
308} 568}
309 569
310// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
311static void 571static void
312encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
313{ 573{
314 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
315
316 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317 && ecb_expect_false (SvREFCNT (sv) > 1))
318 {
319 if (!enc->shareable)
320 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324 if (SvOK (*svp))
325 {
326 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327 encode_uint (enc, 0x00, SvUV (*svp));
328 return;
329 }
330 else
331 {
332 sv_setuv (*svp, enc->shareable_idx);
333 ++enc->shareable_idx;
334 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335 }
336 }
337 575
338 svtype svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
339 577
340 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
341 { 579 {
348 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349 ? cbor_tagged_stash 587 ? cbor_tagged_stash
350 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
351 589
352 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
353 GV *method;
354 591
355 if (stash == boolean_stash) 592 if (stash == boolean_stash)
356 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
357 else if (stash == error_stash) 597 else if (stash == error_stash)
358 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
359 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
360 { 603 {
361 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
362 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
363 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
364 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365 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))
366 } 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))));
367 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368 { 659 {
369 dSP; 660 dSP;
370 661
371 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
372 // 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
373 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
374 666
375 PUTBACK; 667 PUTBACK;
376 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
389 } 681 }
390 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
391 { 683 {
392 dSP; 684 dSP;
393 685
394 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
395 EXTEND (SP, 2); 688 EXTEND (SP, 2);
396 // 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
397 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
398 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
399 692
404 // catch this surprisingly common error 697 // catch this surprisingly common error
405 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
407 700
408 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
410 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411 704
412 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
413 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
414 713
415 PUTBACK; 714 PUTBACK;
416 715
417 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
418 } 717 }
436{ 735{
437 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
438 737
439 need (enc, 9); 738 need (enc, 9);
440 739
441 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
442 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443 //TODO: maybe I32? 742 //TODO: maybe I32?
444 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
445 { 744 encode_float32 (enc, nv);
446 uint32_t fp = ecb_float_to_binary32 (nv);
447
448 *enc->cur++ = 0xe0 | 26;
449
450 if (!ecb_big_endian ())
451 fp = ecb_bswap32 (fp);
452
453 memcpy (enc->cur, &fp, 4);
454 enc->cur += 4;
455 }
456 else 745 else
457 { 746 encode_float64 (enc, nv);
458 uint64_t fp = ecb_double_to_binary64 (nv);
459
460 *enc->cur++ = 0xe0 | 27;
461
462 if (!ecb_big_endian ())
463 fp = ecb_bswap64 (fp);
464
465 memcpy (enc->cur, &fp, 8);
466 enc->cur += 8;
467 }
468} 747}
469 748
470static void 749static void
471encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
472{ 751{
474 753
475 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
476 { 755 {
477 STRLEN len; 756 STRLEN len;
478 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
479 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
480 } 759 }
481 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
482 encode_nv (enc, sv); 761 encode_nv (enc, sv);
483 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
484 { 763 encode_int (enc, sv);
485 if (SvIsUV (sv))
486 encode_uint (enc, 0x00, SvUVX (sv));
487 else if (SvIVX (sv) >= 0)
488 encode_uint (enc, 0x00, SvIVX (sv));
489 else
490 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
491 }
492 else if (SvROK (sv)) 764 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
495 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498 else 770 else
499 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",
500 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501} 773}
502 774
503static SV * 775static SV *
504encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
505{ 777{
506 enc_t enc = { }; 778 enc_t enc = { 0 };
507 779
508 enc.cbor = *cbor; 780 enc.cbor = *cbor;
509 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
510 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
511 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
512 784
513 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
514 786
515 if (cbor->flags & F_ALLOW_STRINGREF) 787 if (cbor->flags & F_PACK_STRINGS)
516 { 788 {
517 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520 } 792 }
542 CBOR cbor; 814 CBOR cbor;
543 U32 depth; // recursion depth 815 U32 depth; // recursion depth
544 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
545 AV *shareable; 817 AV *shareable;
546 AV *stringref; 818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
547} dec_t; 821} dec_t;
548 822
549#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);
550 830
551#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;
552 834
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
553#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)
554#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
555 861
556static UV 862static UV
557decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
558{ 864{
559 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
560 { 866 ++dec->cur;
561 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564 return *dec->cur++ & 31;
565 867
566 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
567 WANT (2); 872 WANT (1);
568 dec->cur += 2; 873 dec->cur += 1;
569 return dec->cur[-1]; 874 return dec->cur[-1];
570 875 }
571 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
572 WANT (3); 878 WANT (2);
573 dec->cur += 3; 879 dec->cur += 2;
574 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
575 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
576 882 }
577 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
578 WANT (5); 885 WANT (4);
579 dec->cur += 5; 886 dec->cur += 4;
580 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
581 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
582 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
584 891 }
585 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
586 WANT (9); 894 WANT (8);
587 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
588 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
589 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
590 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
591 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
592 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
593 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
594 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
595 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
596 910 }
597 default: 911 else
598 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599 }
600 913
601fail: 914fail:
602 return 0; 915 return 0;
603} 916}
604 917
609{ 922{
610 AV *av = newAV (); 923 AV *av = newAV ();
611 924
612 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
613 926
614 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615 { 928 {
616 ++dec->cur; 929 ++dec->cur;
617 930
618 for (;;) 931 for (;;)
619 { 932 {
620 WANT (1); 933 WANT (1);
621 934
622 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
623 { 936 {
624 ++dec->cur; 937 ++dec->cur;
625 break; 938 break;
626 } 939 }
627 940
628 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
629 } 942 }
630 } 943 }
631 else 944 else
632 { 945 {
633 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
634 947
948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635 av_fill (av, len - 1); 949 av_fill (av, len - 1);
636 950
637 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
638 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
639 } 953 }
640 954
641 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
642 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
643 957
644fail: 958fail:
645 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
646 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
647 return &PL_sv_undef; 961 return &PL_sv_undef;
648} 962}
649 963
650static void 964static void
652{ 966{
653 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
654 // byte or utf-8 strings as keys, but only when !stringref 968 // byte or utf-8 strings as keys, but only when !stringref
655 969
656 if (ecb_expect_true (!dec->stringref)) 970 if (ecb_expect_true (!dec->stringref))
657 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658 { 972 {
659 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
660 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
661 975
976 WANT (len);
662 dec->cur += len; 977 dec->cur += len;
663 978
664 if (ecb_expect_false (dec->stringref))
665 av_push (dec->stringref, newSVpvn (key, len));
666
667 hv_store (hv, key, len, decode_sv (dec), 0); 979 hv_store (hv, key, len, decode_sv (dec), 0);
668 980
669 return; 981 return;
670 } 982 }
671 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672 { 984 {
673 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
674 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
675 987
988 WANT (len);
676 dec->cur += len; 989 dec->cur += len;
677 990
678 if (ecb_expect_false (dec->stringref)) 991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679 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)");
680 994
681 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
682 996
683 return; 997 return;
684 } 998 }
685 999
686 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
687 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
688 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
689 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
690 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
1033
1034fail:
1035 ;
691} 1036}
692 1037
693static SV * 1038static SV *
694decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
695{ 1040{
696 HV *hv = newHV (); 1041 HV *hv = newHV ();
697 1042
698 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
699 1044
700 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701 { 1046 {
702 ++dec->cur; 1047 ++dec->cur;
703 1048
704 for (;;) 1049 for (;;)
705 { 1050 {
706 WANT (1); 1051 WANT (1);
707 1052
708 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
709 { 1054 {
710 ++dec->cur; 1055 ++dec->cur;
711 break; 1056 break;
712 } 1057 }
713 1058
714 decode_he (dec, hv); 1059 decode_he (dec, hv);
715 } 1060 }
716 } 1061 }
717 else 1062 else
718 { 1063 {
719 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
720 1067
721 while (pairs--) 1068 while (pairs--)
722 decode_he (dec, hv); 1069 decode_he (dec, hv);
723 } 1070 }
724 1071
725 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
727 1074
728fail: 1075fail:
729 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
730 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
731 return &PL_sv_undef; 1078 return &PL_sv_undef;
732} 1079}
733 1080
734static SV * 1081static SV *
735decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
736{ 1083{
737 SV *sv = 0; 1084 SV *sv = 0;
738 1085
739 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
740 { 1087 {
1088 // indefinite length strings
741 ++dec->cur; 1089 ++dec->cur;
742 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
743 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
744 1094
745 // not very fast, and certainly not robust against illegal input
746 for (;;) 1095 for (;;)
747 { 1096 {
748 WANT (1); 1097 WANT (1);
749 1098
750 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751 { 1101 {
752 ++dec->cur; 1102 ++dec->cur;
753 break; 1103 break;
754 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755 1107
756 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;
757 } 1113 }
758 } 1114 }
759 else 1115 else
760 { 1116 {
761 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
768 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 1126 }
771 1127
772 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
773 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
774 1136
775 return sv; 1137 return sv;
776 1138
777fail: 1139fail:
778 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
797 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
798 break; 1160 break;
799 1161
800 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
801 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
802 ENTER; SAVETMPS; 1166 ENTER;
803 1167
804 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
805 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
806 1170
807 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
808 1172
809 FREETMPS; LEAVE; 1173 LEAVE;
810 } 1174 }
811 break; 1175 break;
812 1176
813 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
814 { 1178 {
815 if ((*dec->cur >> 5) != 0) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817 1181
818 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
819 1183
820 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
821 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)");
822 1186
823 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
824 } 1188 }
825 break; 1189 break;
827 case CBOR_TAG_VALUE_SHAREABLE: 1191 case CBOR_TAG_VALUE_SHAREABLE:
828 { 1192 {
829 if (ecb_expect_false (!dec->shareable)) 1193 if (ecb_expect_false (!dec->shareable))
830 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831 1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
832 sv = newSV (0); 1198 sv = newSV (0);
833 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
834 1200
835 SV *osv = decode_sv (dec); 1201 SV *osv = decode_sv (dec);
836 sv_setsv (sv, osv); 1202 sv_setsv (sv, osv);
837 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 }
838 } 1212 }
839 break; 1213 break;
840 1214
841 case CBOR_TAG_VALUE_SHAREDREF: 1215 case CBOR_TAG_VALUE_SHAREDREF:
842 { 1216 {
843 if ((*dec->cur >> 5) != 0) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845 1219
846 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
847 1221
848 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
849 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
850 1224
851 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");
852 } 1229 }
853 break; 1230 break;
854 1231
855 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
856 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
857 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
858 1238
859 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
860 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
861 1241
871 if (!method) 1251 if (!method)
872 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)");
873 1253
874 dSP; 1254 dSP;
875 1255
876 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
877 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
878 // 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
879 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
880 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
881 1262
889 SPAGAIN; 1270 SPAGAIN;
890 1271
891 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
892 { 1273 {
893 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
894 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
895 } 1276 }
896 1277
897 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
898 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
899 1280
900 PUTBACK; 1281 PUTBACK;
901 1282
902 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
903 } 1284 }
904 break; 1285 break;
905 1286
906 default: 1287 default:
1288 filter:
907 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
908 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
909 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 {
910 AV *av = newAV (); 1321 AV *av = newAV ();
911 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
912 av_push (av, sv); 1323 av_push (av, sv);
913 1324
914 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
916 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918 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;
919 } 1334 }
920 break; 1335 break;
921 } 1336 }
922 1337
923 return sv; 1338 return sv;
930static SV * 1345static SV *
931decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
932{ 1347{
933 WANT (1); 1348 WANT (1);
934 1349
935 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
936 { 1351 {
937 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944 return decode_str (dec, 1); 1359
945 case 4: // array 1360 case MAJOR_MISC >> MAJOR_SHIFT:
946 return decode_av (dec);
947 case 5: // map
948 return decode_hv (dec);
949 case 6: // tag
950 return decode_tagged (dec);
951 case 7: // misc
952 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
953 { 1362 {
954 case 20: 1363 case SIMPLE_FALSE:
955#if CBOR_SLOW 1364#if CBOR_SLOW
956 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
957#endif 1366#endif
958 return newSVsv (types_false); 1367 return newSVsv (types_false);
959 case 21: 1368 case SIMPLE_TRUE:
960#if CBOR_SLOW 1369#if CBOR_SLOW
961 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
962#endif 1371#endif
963 return newSVsv (types_true); 1372 return newSVsv (types_true);
964 case 22: 1373 case SIMPLE_NULL:
965 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
966 case 23: 1375 case SIMPLE_UNDEF:
967#if CBOR_SLOW 1376#if CBOR_SLOW
968 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
969#endif 1378#endif
970 return newSVsv (types_error); 1379 return newSVsv (types_error);
971 1380
972 case 25: 1381 case MISC_FLOAT16:
973 { 1382 {
974 WANT (2); 1383 WANT (2);
975 1384
976 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977 dec->cur += 2; 1386 dec->cur += 2;
978 1387
979 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
980 } 1389 }
981 1390
982 case 26: 1391 case MISC_FLOAT32:
983 { 1392 {
984 uint32_t fp; 1393 uint32_t fp;
985 WANT (4); 1394 WANT (4);
986 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
987 dec->cur += 4; 1396 dec->cur += 4;
990 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
991 1400
992 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
993 } 1402 }
994 1403
995 case 27: 1404 case MISC_FLOAT64:
996 { 1405 {
997 uint64_t fp; 1406 uint64_t fp;
998 WANT (8); 1407 WANT (8);
999 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
1000 dec->cur += 8; 1409 dec->cur += 8;
1003 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
1004 1413
1005 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
1006 } 1415 }
1007 1416
1008 // 0..19 unassigned 1417 // 0..19 unassigned simple
1009 // 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
1010 default: 1421 default:
1011 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012 } 1423 }
1013 1424
1014 break; 1425 break;
1015 } 1426 }
1016 1427
1019} 1430}
1020 1431
1021static SV * 1432static SV *
1022decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1023{ 1434{
1024 dec_t dec = { }; 1435 dec_t dec = { 0 };
1025 SV *sv; 1436 SV *sv;
1026 STRLEN len; 1437 STRLEN len;
1027 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
1028 1439
1029 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
1043 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
1044 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
1045 1456
1046 if (dec.err) 1457 if (dec.err)
1047 { 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
1048 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1049 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);
1050 } 1476 }
1051 1477
1052 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1053 1479
1054 return sv; 1480 return sv;
1055} 1481}
1056 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
1057///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
1058// XS interface functions 1609// XS interface functions
1059 1610
1060MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
1061 1612
1069 1620
1070 types_true = get_bool ("Types::Serialiser::true" ); 1621 types_true = get_bool ("Types::Serialiser::true" );
1071 types_false = get_bool ("Types::Serialiser::false"); 1622 types_false = get_bool ("Types::Serialiser::false");
1072 types_error = get_bool ("Types::Serialiser::error"); 1623 types_error = get_bool ("Types::Serialiser::error");
1073 1624
1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1626
1074 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
1075 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1076} 1631}
1077 1632
1078PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
1079 1634
1080void CLONE (...) 1635void CLONE (...)
1099void shrink (CBOR *self, int enable = 1) 1654void shrink (CBOR *self, int enable = 1)
1100 ALIAS: 1655 ALIAS:
1101 shrink = F_SHRINK 1656 shrink = F_SHRINK
1102 allow_unknown = F_ALLOW_UNKNOWN 1657 allow_unknown = F_ALLOW_UNKNOWN
1103 allow_sharing = F_ALLOW_SHARING 1658 allow_sharing = F_ALLOW_SHARING
1104 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
1105 PPCODE: 1665 PPCODE:
1106{ 1666{
1107 if (enable) 1667 if (enable)
1108 self->flags |= ix; 1668 self->flags |= ix;
1109 else 1669 else
1115void get_shrink (CBOR *self) 1675void get_shrink (CBOR *self)
1116 ALIAS: 1676 ALIAS:
1117 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
1118 get_allow_unknown = F_ALLOW_UNKNOWN 1678 get_allow_unknown = F_ALLOW_UNKNOWN
1119 get_allow_sharing = F_ALLOW_SHARING 1679 get_allow_sharing = F_ALLOW_SHARING
1120 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
1121 PPCODE: 1686 PPCODE:
1122 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
1123 1688
1124void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125 PPCODE: 1690 PPCODE:
1138 XPUSHs (ST (0)); 1703 XPUSHs (ST (0));
1139 1704
1140int get_max_size (CBOR *self) 1705int get_max_size (CBOR *self)
1141 CODE: 1706 CODE:
1142 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);
1143 OUTPUT: 1720 OUTPUT:
1144 RETVAL 1721 RETVAL
1145 1722
1146void encode (CBOR *self, SV *scalar) 1723void encode (CBOR *self, SV *scalar)
1147 PPCODE: 1724 PPCODE:
1162 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1163 PUSHs (sv); 1740 PUSHs (sv);
1164 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165} 1742}
1166 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
1167PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
1168 1801
1169void encode_cbor (SV *scalar) 1802void encode_cbor (SV *scalar)
1803 ALIAS:
1804 encode_cbor = 0
1805 encode_cbor_sharing = F_ALLOW_SHARING
1170 PPCODE: 1806 PPCODE:
1171{ 1807{
1172 CBOR cbor; 1808 CBOR cbor;
1173 cbor_init (&cbor); 1809 cbor_init (&cbor);
1810 cbor.flags |= ix;
1174 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1811 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175 XPUSHs (scalar); 1812 XPUSHs (scalar);
1176} 1813}
1177 1814
1178void decode_cbor (SV *cborstr) 1815void decode_cbor (SV *cborstr)
1182 cbor_init (&cbor); 1819 cbor_init (&cbor);
1183 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1184 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1185} 1822}
1186 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