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 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.75 by root, Thu Sep 7 23:52:24 2023 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
35
36// known major and minor types
37enum cbor_type
38{
39 MAJOR_SHIFT = 5,
40 MINOR_MASK = 0x1f,
41
42 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
43 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
44 MAJOR_BYTES = 2 << MAJOR_SHIFT,
45 MAJOR_TEXT = 3 << MAJOR_SHIFT,
46 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
47 MAJOR_MAP = 5 << MAJOR_SHIFT,
48 MAJOR_TAG = 6 << MAJOR_SHIFT,
49 MAJOR_MISC = 7 << MAJOR_SHIFT,
50
51 // INT/STRING/ARRAY/MAP subtypes
52 LENGTH_EXT1 = 24,
53 LENGTH_EXT2 = 25,
54 LENGTH_EXT4 = 26,
55 LENGTH_EXT8 = 27,
56
57 // SIMPLE types (effectively MISC subtypes)
58 SIMPLE_FALSE = 20,
59 SIMPLE_TRUE = 21,
60 SIMPLE_NULL = 22,
61 SIMPLE_UNDEF = 23,
62
63 // MISC subtype (unused)
64 MISC_EXT1 = 24,
65 MISC_FLOAT16 = 25,
66 MISC_FLOAT32 = 26,
67 MISC_FLOAT64 = 27,
68
69 // BYTES/TEXT/ARRAY/MAP
70 MINOR_INDEF = 31,
71};
24 72
25// known tags 73// known tags
26enum cbor_tag 74enum cbor_tag
27{ 75{
28 // inofficial extensions (pending iana registration) 76 // extensions
77 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 78 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 79 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 80 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 81 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 82 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 83 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 84
37 // rfc7049 85 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 86 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 87 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 88 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 89 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 90 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 91 CBOR_TAG_BIGFLOAT = 5, // array
44 92
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 93 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 94 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 95 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 96 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 97
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 98 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 99 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 100 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 101 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 103
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 105};
58 106
107// known forced types, also hardcoded in CBOR.pm
108enum
109{
110 AS_CBOR = 0,
111 AS_INT = 1,
112 AS_BYTES = 2,
113 AS_TEXT = 3,
114 AS_FLOAT16 = 4,
115 AS_FLOAT32 = 5,
116 AS_FLOAT64 = 6,
117 AS_MAP = 7,
118 // possibly future enhancements: (generic) float, (generic) string
119};
120
59#define F_SHRINK 0x00000001UL 121#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 123#define F_ALLOW_SHARING 0x00000004UL
124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 126#define F_PACK_STRINGS 0x00000020UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
64 130
65#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
66 132
67#define SB do { 133#define SB do {
68#define SE } while (0) 134#define SE } while (0)
80# define CBOR_SLOW 0 146# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 147# define CBOR_STASH cbor_stash
82#endif 148#endif
83 149
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 150static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 151static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 152
87typedef struct { 153typedef struct {
88 U32 flags; 154 U32 flags;
89 U32 max_depth; 155 U32 max_depth;
90 STRLEN max_size; 156 STRLEN max_size;
157 SV *filter;
158
159 // for the incremental parser
160 STRLEN incr_pos; // the current offset into the text
161 STRLEN incr_need; // minimum bytes needed to decode
162 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 163} CBOR;
92 164
93ecb_inline void 165ecb_inline void
94cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
95{ 167{
96 Zero (cbor, 1, CBOR); 168 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 169 cbor->max_depth = 512;
170}
171
172ecb_inline void
173cbor_free (CBOR *cbor)
174{
175 SvREFCNT_dec (cbor->filter);
176 SvREFCNT_dec (cbor->incr_count);
98} 177}
99 178
100///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
101// utility functions 180// utility functions
102 181
122 SvPV_shrink_to_cur (sv); 201 SvPV_shrink_to_cur (sv);
123#elif defined (SvPV_renew) 202#elif defined (SvPV_renew)
124 SvPV_renew (sv, SvCUR (sv) + 1); 203 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 204#endif
126 } 205 }
206}
207
208// minimum length of a string to be registered for stringref
209ecb_inline STRLEN
210minimum_string_length (UV idx)
211{
212 return idx <= 23 ? 3
213 : idx <= 0xffU ? 4
214 : idx <= 0xffffU ? 5
215 : idx <= 0xffffffffU ? 7
216 : 11;
127} 217}
128 218
129///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
130// encoder 220// encoder
131 221
144} enc_t; 234} enc_t;
145 235
146ecb_inline void 236ecb_inline void
147need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
148{ 238{
149 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150 { 240 {
151 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
154 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
155 } 245 }
156} 246}
157 247
248static void encode_sv (enc_t *enc, SV *sv);
249
158ecb_inline void 250ecb_inline void
159encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
160{ 252{
161 need (enc, 1); 253 need (enc, 1);
162 *enc->cur++ = ch; 254 *enc->cur++ = ch;
163} 255}
164 256
257// used for tags, intregers, element counts and so on
165static void 258static void
166encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
167{ 260{
168 need (enc, 9); 261 need (enc, 9);
169 262
170 if (len < 24) 263 if (ecb_expect_true (len < LENGTH_EXT1))
171 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
172 else if (len <= 0xff) 265 else if (ecb_expect_true (len <= 0xffU))
173 { 266 {
174 *enc->cur++ = major | 24; 267 *enc->cur++ = major | LENGTH_EXT1;
175 *enc->cur++ = len; 268 *enc->cur++ = len;
176 } 269 }
177 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
178 { 271 {
179 *enc->cur++ = major | 25; 272 *enc->cur++ = major | LENGTH_EXT2;
180 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
181 *enc->cur++ = len; 274 *enc->cur++ = len;
182 } 275 }
183 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
184 { 277 {
185 *enc->cur++ = major | 26; 278 *enc->cur++ = major | LENGTH_EXT4;
186 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
187 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
188 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
189 *enc->cur++ = len; 282 *enc->cur++ = len;
190 } 283 }
191 else 284 else
192 { 285 {
193 *enc->cur++ = major | 27; 286 *enc->cur++ = major | LENGTH_EXT8;
194 *enc->cur++ = len >> 56; 287 *enc->cur++ = len >> 56;
195 *enc->cur++ = len >> 48; 288 *enc->cur++ = len >> 48;
196 *enc->cur++ = len >> 40; 289 *enc->cur++ = len >> 40;
197 *enc->cur++ = len >> 32; 290 *enc->cur++ = len >> 32;
198 *enc->cur++ = len >> 24; 291 *enc->cur++ = len >> 24;
200 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
201 *enc->cur++ = len; 294 *enc->cur++ = len;
202 } 295 }
203} 296}
204 297
205static void 298// encodes a perl value into a CBOR integer
299ecb_inline void
300encode_int (enc_t *enc, SV *sv)
301{
302 if (SvIsUV (sv))
303 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
304 else if (SvIVX (sv) >= 0)
305 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
306 else
307 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
308}
309
310ecb_inline void
311encode_tag (enc_t *enc, UV tag)
312{
313 encode_uint (enc, MAJOR_TAG, tag);
314}
315
316// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
317ecb_noinline static void
206encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207{ 319{
208 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 320 STRLEN ulen = len;
321 U8 *p, *pend = (U8 *)str + len;
322
323 for (p = (U8 *)str; p < pend; ++p)
324 ulen += *p >> 7; // count set high bits
325
326 encode_uint (enc, MAJOR_TEXT, ulen);
327
328 need (enc, ulen);
329 for (p = (U8 *)str; p < pend; ++p)
330 if (*p < 0x80)
331 *enc->cur++ = *p;
332 else
333 {
334 *enc->cur++ = 0xc0 + (*p >> 6);
335 *enc->cur++ = 0x80 + (*p & 63);
336 }
337}
338
339ecb_inline void
340encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
341{
342 if (ecb_expect_false (upgrade_utf8))
343 if (!utf8)
344 {
345 encode_str_utf8 (enc, utf8, str, len);
346 return;
347 }
348
349 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
209 need (enc, len); 350 need (enc, len);
210 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
211 enc->cur += len; 352 enc->cur += len;
212} 353}
213 354
214ecb_inline void 355ecb_inline void
356encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
357{
358 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
359 {
360 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
361
362 if (SvOK (*svp))
363 {
364 // already registered, use stringref
365 encode_tag (enc, CBOR_TAG_STRINGREF);
366 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
367 return;
368 }
369 else if (len >= minimum_string_length (enc->stringref_idx))
370 {
371 // register only
372 sv_setuv (*svp, enc->stringref_idx);
373 ++enc->stringref_idx;
374 }
375 }
376
377 encode_str (enc, upgrade_utf8, utf8, str, len);
378}
379
380ecb_inline void
215encode_tag (enc_t *enc, UV tag) 381encode_float16 (enc_t *enc, NV nv)
216{ 382{
217 encode_uint (enc, 0xc0, tag); 383 need (enc, 1+2);
218}
219 384
220static void encode_sv (enc_t *enc, SV *sv); 385 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
386
387 uint16_t fp = ecb_float_to_binary16 (nv);
388
389 if (!ecb_big_endian ())
390 fp = ecb_bswap16 (fp);
391
392 memcpy (enc->cur, &fp, 2);
393 enc->cur += 2;
394}
395
396ecb_inline void
397encode_float32 (enc_t *enc, NV nv)
398{
399 need (enc, 1+4);
400
401 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
402
403 uint32_t fp = ecb_float_to_binary32 (nv);
404
405 if (!ecb_big_endian ())
406 fp = ecb_bswap32 (fp);
407
408 memcpy (enc->cur, &fp, 4);
409 enc->cur += 4;
410}
411
412ecb_inline void
413encode_float64 (enc_t *enc, NV nv)
414{
415 need (enc, 1+8);
416
417 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
418
419 uint64_t fp = ecb_double_to_binary64 (nv);
420
421 if (!ecb_big_endian ())
422 fp = ecb_bswap64 (fp);
423
424 memcpy (enc->cur, &fp, 8);
425 enc->cur += 8;
426}
427
428ecb_inline void
429encode_bool (enc_t *enc, int istrue)
430{
431 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
432}
433
434// encodes an arrayref containing key-value pairs as CBOR map
435ecb_inline void
436encode_array_as_map (enc_t *enc, SV *sv)
437{
438 if (enc->depth >= enc->cbor.max_depth)
439 croak (ERR_NESTING_EXCEEDED);
440
441 ++enc->depth;
442
443 // as_map does error checking for us, but we re-check in case
444 // things have changed.
445
446 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
447 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
448
449 AV *av = (AV *)SvRV (sv);
450 int i, len = av_len (av);
451
452 if (!(len & 1))
453 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
454
455 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
456
457 for (i = 0; i <= len; ++i)
458 {
459 SV **svp = av_fetch (av, i, 0);
460 encode_sv (enc, svp ? *svp : &PL_sv_undef);
461 }
462
463 --enc->depth;
464}
465
466ecb_inline void
467encode_forced (enc_t *enc, UV type, SV *sv)
468{
469 switch (type)
470 {
471 case AS_CBOR:
472 {
473 STRLEN len;
474 char *str = SvPVbyte (sv, len);
475
476 need (enc, len);
477 memcpy (enc->cur, str, len);
478 enc->cur += len;
479 }
480 break;
481
482 case AS_BYTES:
483 {
484 STRLEN len;
485 char *str = SvPVbyte (sv, len);
486 encode_strref (enc, 0, 0, str, len);
487 }
488 break;
489
490 case AS_TEXT:
491 {
492 STRLEN len;
493 char *str = SvPVutf8 (sv, len);
494 encode_strref (enc, 1, 1, str, len);
495 }
496 break;
497
498 case AS_INT: encode_int (enc, sv); break;
499
500 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
501 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
502 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
503
504 case AS_MAP: encode_array_as_map (enc, sv); break;
505
506 default:
507 croak ("encountered malformed CBOR::XS::Tagged object");
508 }
509}
221 510
222static void 511static void
223encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
224{ 513{
225 int i, len = av_len (av); 514 int i, len = av_len (av);
227 if (enc->depth >= enc->cbor.max_depth) 516 if (enc->depth >= enc->cbor.max_depth)
228 croak (ERR_NESTING_EXCEEDED); 517 croak (ERR_NESTING_EXCEEDED);
229 518
230 ++enc->depth; 519 ++enc->depth;
231 520
232 encode_uint (enc, 0x80, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
233 522
523 if (ecb_expect_false (SvMAGICAL (av)))
234 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
235 { 525 {
236 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
237 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
238 } 528 }
529 else
530 for (i = 0; i <= len; ++i)
531 {
532 SV *sv = AvARRAY (av)[i];
533 encode_sv (enc, sv ? sv : &PL_sv_undef);
534 }
239 535
240 --enc->depth; 536 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250} 537}
251 538
252static void 539static void
253encode_hv (enc_t *enc, HV *hv) 540encode_hv (enc_t *enc, HV *hv)
254{ 541{
260 ++enc->depth; 547 ++enc->depth;
261 548
262 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
263 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
264 551
265 if (mg) 552 if (ecb_expect_false (mg))
266 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267 else 554 else
268 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
269 556
270 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
271 { 558 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
276 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/ 560 encode_sv (enc, HeSVKEY (he));
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
293 else 561 else
294 encode_he (enc, he); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295 563
296 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 564 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
297 } 565 }
298 566
299 if (mg) 567 if (ecb_expect_false (mg))
300 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301 569
302 --enc->depth; 570 --enc->depth;
303} 571}
304 572
305// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
306static void 574static void
307encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
308{ 576{
309 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332 578
333 svtype svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
334 580
335 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
336 { 582 {
343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344 ? cbor_tagged_stash 590 ? cbor_tagged_stash
345 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
346 592
347 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
348 GV *method;
349 594
350 if (stash == boolean_stash) 595 if (stash == boolean_stash)
351 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
352 else if (stash == error_stash) 600 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
354 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
355 { 606 {
356 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
357 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
358 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 615 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
616 break;
617
618 case 3-1:
619 // a forced type [value, type, undef]
620 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
621 break;
622
623 default:
624 croak ("encountered malformed CBOR::XS::Tagged object");
625 }
626
627 return;
628 }
629 }
630
631 if (ecb_expect_false (SvREFCNT (sv) > 1)
632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
633 {
634 if (ecb_expect_false (!enc->shareable))
635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
636
637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
638
639 if (SvOK (*svp))
361 } 640 {
641 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
642 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
643 return;
644 }
645 else
646 {
647 sv_setuv (*svp, enc->shareable_idx);
648 ++enc->shareable_idx;
649 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
650 }
651 }
652
653 if (ecb_expect_false (SvOBJECT (sv)))
654 {
655 HV *stash = SvSTASH (sv);
656 GV *method;
657
658 if (enc->cbor.flags & F_FORBID_OBJECTS)
659 croak ("encountered object '%s', but forbid_objects is enabled",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 { 662 {
364 dSP; 663 dSP;
365 664
366 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
367 // we re-bless the reference to get overload and other niceties right 667 // we re-bless the reference to get overload and other niceties right
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369 669
370 PUTBACK; 670 PUTBACK;
371 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
384 } 684 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 { 686 {
387 dSP; 687 dSP;
388 688
389 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
390 EXTEND (SP, 2); 691 EXTEND (SP, 2);
391 // we re-bless the reference to get overload and other niceties right 692 // we re-bless the reference to get overload and other niceties right
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
394 695
399 // catch this surprisingly common error 700 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 702 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402 703
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406 707
407 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
408 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
409 716
410 PUTBACK; 717 PUTBACK;
411 718
412 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
413 } 720 }
431{ 738{
432 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
433 740
434 need (enc, 9); 741 need (enc, 9);
435 742
436 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
437 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438 //TODO: maybe I32? 745 //TODO: maybe I32?
439 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
440 { 747 encode_float32 (enc, nv);
441 uint32_t fp = ecb_float_to_binary32 (nv);
442
443 *enc->cur++ = 0xe0 | 26;
444
445 if (!ecb_big_endian ())
446 fp = ecb_bswap32 (fp);
447
448 memcpy (enc->cur, &fp, 4);
449 enc->cur += 4;
450 }
451 else 748 else
452 { 749 encode_float64 (enc, nv);
453 uint64_t fp = ecb_double_to_binary64 (nv);
454
455 *enc->cur++ = 0xe0 | 27;
456
457 if (!ecb_big_endian ())
458 fp = ecb_bswap64 (fp);
459
460 memcpy (enc->cur, &fp, 8);
461 enc->cur += 8;
462 }
463} 750}
464 751
465static void 752static void
466encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
467{ 754{
469 756
470 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
471 { 758 {
472 STRLEN len; 759 STRLEN len;
473 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
474 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475 } 762 }
476 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
477 encode_nv (enc, sv); 764 encode_nv (enc, sv);
478 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
479 { 766 encode_int (enc, sv);
480 if (SvIsUV (sv))
481 encode_uint (enc, 0x00, SvUVX (sv));
482 else if (SvIVX (sv) >= 0)
483 encode_uint (enc, 0x00, SvIVX (sv));
484 else
485 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
486 }
487 else if (SvROK (sv)) 767 else if (SvROK (sv))
488 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
489 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
490 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493 else 773 else
494 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 774 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
495 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496} 776}
497 777
498static SV * 778static SV *
499encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
500{ 780{
501 enc_t enc = { }; 781 enc_t enc = { 0 };
502 782
503 enc.cbor = *cbor; 783 enc.cbor = *cbor;
504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
506 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
507 787
508 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
509 789
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)) 790 if (cbor->flags & F_PACK_STRINGS)
511 { 791 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 792 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 793 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 794 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 } 795 }
537 CBOR cbor; 817 CBOR cbor;
538 U32 depth; // recursion depth 818 U32 depth; // recursion depth
539 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
540 AV *shareable; 820 AV *shareable;
541 AV *stringref; 821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
542} dec_t; 824} dec_t;
543 825
544#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 826// set dec->err to ERRSV
827ecb_cold static void
828err_errsv (dec_t *dec)
829{
830 if (!dec->err)
831 {
832 dec->err_sv = newSVsv (ERRSV);
545 833
546#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 834 // chop off the trailing \n
835 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
836 *SvEND (dec->err_sv) = 0;
547 837
838 dec->err = SvPVutf8_nolen (dec->err_sv);
839 }
840}
841
842// the following functions are used to reduce code size and help the compiler to optimise
843ecb_cold static void
844err_set (dec_t *dec, const char *reason)
845{
846 if (!dec->err)
847 dec->err = reason;
848}
849
850ecb_cold static void
851err_unexpected_end (dec_t *dec)
852{
853 err_set (dec, "unexpected end of CBOR data");
854}
855
856#define ERR_DO(do) SB do; goto fail; SE
857#define ERR(reason) ERR_DO (err_set (dec, reason))
858#define ERR_ERRSV ERR_DO (err_errsv (dec))
859
860#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
861
548#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 862#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
549#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
550 864
551static UV 865static UV
552decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
553{ 867{
554 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
555 { 869 ++dec->cur;
556 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559 return *dec->cur++ & 31;
560 870
561 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
562 WANT (2); 875 WANT (1);
563 dec->cur += 2; 876 dec->cur += 1;
564 return dec->cur[-1]; 877 return dec->cur[-1];
565 878 }
566 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
567 WANT (3); 881 WANT (2);
568 dec->cur += 3; 882 dec->cur += 2;
569 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
570 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
571 885 }
572 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
573 WANT (5); 888 WANT (4);
574 dec->cur += 5; 889 dec->cur += 4;
575 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
576 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
577 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
578 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
579 894 }
580 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
581 WANT (9); 897 WANT (8);
582 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
583 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
584 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
585 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
586 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
587 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
588 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
589 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
590 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
591 913 }
592 default: 914 else
593 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594 }
595 916
596fail: 917fail:
597 return 0; 918 return 0;
598} 919}
599 920
604{ 925{
605 AV *av = newAV (); 926 AV *av = newAV ();
606 927
607 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
608 929
609 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610 { 931 {
611 ++dec->cur; 932 ++dec->cur;
612 933
613 for (;;) 934 for (;;)
614 { 935 {
615 WANT (1); 936 WANT (1);
616 937
617 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
618 { 939 {
619 ++dec->cur; 940 ++dec->cur;
620 break; 941 break;
621 } 942 }
622 943
623 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
624 } 945 }
625 } 946 }
626 else 947 else
627 { 948 {
628 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
629 950
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630 av_fill (av, len - 1); 952 av_fill (av, len - 1);
631 953
632 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
633 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
634 } 956 }
635 957
636 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
637 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
638 960
639fail: 961fail:
640 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
641 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
642 return &PL_sv_undef; 964 return &PL_sv_undef;
643} 965}
644 966
645static void 967static void
646decode_he (dec_t *dec, HV *hv) 968decode_he (dec_t *dec, HV *hv)
647{ 969{
648 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys. 971 // byte or utf-8 strings as keys, but only when !stringref
650 972
651 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 973 if (ecb_expect_true (!dec->stringref))
974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652 { 975 {
653 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
654 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
655 978
979 WANT (len);
656 dec->cur += len; 980 dec->cur += len;
657 981
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
983
984 return;
662 } 985 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664 { 987 {
665 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
666 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
667 990
991 WANT (len);
668 dec->cur += len; 992 dec->cur += len;
669 993
670 if (ecb_expect_false (dec->stringref)) 994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 995 if (!cbor_is_utf8_string ((U8 *)key, len))
996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672 997
673 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
999
1000 return;
674 } 1001 }
675 else 1002
676 {
677 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
679 1005
1006 // we leak memory if uncaught exceptions are thrown by random magical
1007 // methods, and this is hopefully the only place where it can happen,
1008 // so if there is a chance of an exception, take the very slow path.
1009 // since catching exceptions is "undocumented/internal/forbidden" by
1010 // the new p5p powers, we need to call out to a perl function :/
1011 if (ecb_expect_false (SvAMAGIC (k)))
1012 {
1013 dSP;
1014
1015 ENTER; SAVETMPS;
1016 PUSHMARK (SP);
1017 EXTEND (SP, 3);
1018 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1019 PUSHs (sv_2mortal (k));
1020 PUSHs (sv_2mortal (v));
1021
1022 PUTBACK;
1023 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1024 SPAGAIN;
1025
1026 FREETMPS; LEAVE;
1027
1028 if (SvTRUE (ERRSV))
1029 ERR_ERRSV;
1030
1031 return;
1032 }
1033
680 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
682 } 1036
1037fail:
1038 ;
683} 1039}
684 1040
685static SV * 1041static SV *
686decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
687{ 1043{
688 HV *hv = newHV (); 1044 HV *hv = newHV ();
689 1045
690 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
691 1047
692 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693 { 1049 {
694 ++dec->cur; 1050 ++dec->cur;
695 1051
696 for (;;) 1052 for (;;)
697 { 1053 {
698 WANT (1); 1054 WANT (1);
699 1055
700 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
701 { 1057 {
702 ++dec->cur; 1058 ++dec->cur;
703 break; 1059 break;
704 } 1060 }
705 1061
706 decode_he (dec, hv); 1062 decode_he (dec, hv);
707 } 1063 }
708 } 1064 }
709 else 1065 else
710 { 1066 {
711 int pairs = decode_uint (dec); 1067 UV pairs = decode_uint (dec);
1068
1069 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
712 1070
713 while (pairs--) 1071 while (pairs--)
714 decode_he (dec, hv); 1072 decode_he (dec, hv);
715 } 1073 }
716 1074
717 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
718 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
719 1077
720fail: 1078fail:
721 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
722 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 1081 return &PL_sv_undef;
724} 1082}
725 1083
726static SV * 1084static SV *
727decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
728{ 1086{
729 SV *sv = 0; 1087 SV *sv = 0;
730 1088
731 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
732 { 1090 {
1091 // indefinite length strings
733 ++dec->cur; 1092 ++dec->cur;
734 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
735 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
736 1097
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 1098 for (;;)
739 { 1099 {
740 WANT (1); 1100 WANT (1);
741 1101
742 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 1104 {
744 ++dec->cur; 1105 ++dec->cur;
745 break; 1106 break;
746 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 1110
748 sv_catsv (sv, decode_sv (dec)); 1111 STRLEN len = decode_uint (dec);
1112
1113 WANT (len);
1114 sv_catpvn (sv, dec->cur, len);
1115 dec->cur += len;
749 } 1116 }
750 } 1117 }
751 else 1118 else
752 { 1119 {
753 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
754 1121
755 WANT (len); 1122 WANT (len);
756 sv = newSVpvn (dec->cur, len); 1123 sv = newSVpvn (dec->cur, len);
757 dec->cur += len; 1124 dec->cur += len;
1125
1126 if (ecb_expect_false (dec->stringref)
1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
758 } 1129 }
759 1130
760 if (utf8) 1131 if (utf8)
1132 {
1133 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1134 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1135 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1136
761 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
762 1138 }
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765 1139
766 return sv; 1140 return sv;
767 1141
768fail: 1142fail:
769 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
788 sv = newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
789 break; 1163 break;
790 1164
791 case CBOR_TAG_STRINGREF_NAMESPACE: 1165 case CBOR_TAG_STRINGREF_NAMESPACE:
792 { 1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
793 ENTER; SAVETMPS; 1169 ENTER;
794 1170
795 SAVESPTR (dec->stringref); 1171 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797 1173
798 sv = decode_sv (dec); 1174 sv = decode_sv (dec);
799 1175
800 FREETMPS; LEAVE; 1176 LEAVE;
801 } 1177 }
802 break; 1178 break;
803 1179
804 case CBOR_TAG_STRINGREF: 1180 case CBOR_TAG_STRINGREF:
805 { 1181 {
806 if ((*dec->cur >> 5) != 0) 1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808 1184
809 UV idx = decode_uint (dec); 1185 UV idx = decode_uint (dec);
810 1186
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
812 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813 1189
814 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815 } 1191 }
816 break; 1192 break;
818 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
819 { 1195 {
820 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
823 sv = newSV (0); 1201 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1202 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825 1203
826 SV *osv = decode_sv (dec); 1204 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv); 1205 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (osv); 1206 SvREFCNT_dec_NN (osv);
1207 }
1208 else
1209 {
1210 av_push (dec->shareable, &PL_sv_undef);
1211 int idx = AvFILLp (dec->shareable);
1212 sv = decode_sv (dec);
1213 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1214 }
829 } 1215 }
830 break; 1216 break;
831 1217
832 case CBOR_TAG_VALUE_SHAREDREF: 1218 case CBOR_TAG_VALUE_SHAREDREF:
833 { 1219 {
834 if ((*dec->cur >> 5) != 0) 1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836 1222
837 UV idx = decode_uint (dec); 1223 UV idx = decode_uint (dec);
838 1224
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841 1227
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1228 sv = newRV_inc (AvARRAY (dec->shareable)[idx]);
1229
1230 if (sv == &PL_sv_undef)
1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
843 } 1232 }
844 break; 1233 break;
845 1234
846 case CBOR_TAG_PERL_OBJECT: 1235 case CBOR_TAG_PERL_OBJECT:
847 { 1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
848 sv = decode_sv (dec); 1240 sv = decode_sv (dec);
849 1241
850 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
852 1244
862 if (!method) 1254 if (!method)
863 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1255 ERR ("cannot decode perl-object (package does not have a THAW method)");
864 1256
865 dSP; 1257 dSP;
866 1258
867 ENTER; SAVETMPS; PUSHMARK (SP); 1259 ENTER; SAVETMPS;
1260 PUSHMARK (SP);
868 EXTEND (SP, len + 1); 1261 EXTEND (SP, len + 1);
869 // we re-bless the reference to get overload and other niceties right 1262 // we re-bless the reference to get overload and other niceties right
870 PUSHs (*av_fetch (av, 0, 1)); 1263 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor); 1264 PUSHs (sv_cbor);
872 1265
880 SPAGAIN; 1273 SPAGAIN;
881 1274
882 if (SvTRUE (ERRSV)) 1275 if (SvTRUE (ERRSV))
883 { 1276 {
884 FREETMPS; LEAVE; 1277 FREETMPS; LEAVE;
885 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1278 ERR_ERRSV;
886 } 1279 }
887 1280
888 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
889 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
890 1283
891 PUTBACK; 1284 PUTBACK;
892 1285
893 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
894 } 1287 }
895 break; 1288 break;
896 1289
897 default: 1290 default:
1291 filter:
898 { 1292 {
1293 SV *tag_sv = newSVuv (tag);
1294
899 sv = decode_sv (dec); 1295 sv = decode_sv (dec);
900 1296
1297 dSP;
1298 ENTER; SAVETMPS;
1299 PUSHMARK (SP);
1300 EXTEND (SP, 2);
1301 PUSHs (tag_sv);
1302 PUSHs (sv);
1303
1304 PUTBACK;
1305 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1306 SPAGAIN;
1307
1308 if (SvTRUE (ERRSV))
1309 {
1310 SvREFCNT_dec_NN (tag_sv);
1311 FREETMPS; LEAVE;
1312 ERR_ERRSV;
1313 }
1314
1315 if (count)
1316 {
1317 SvREFCNT_dec_NN (tag_sv);
1318 SvREFCNT_dec_NN (sv);
1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1321 }
1322 else
1323 {
901 AV *av = newAV (); 1324 AV *av = newAV ();
902 av_push (av, newSVuv (tag)); 1325 av_push (av, tag_sv);
903 av_push (av, sv); 1326 av_push (av, sv);
904 1327
905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1328 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906 ? cbor_tagged_stash 1329 ? cbor_tagged_stash
907 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1330 : gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1331 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1332 }
1333
1334 PUTBACK;
1335
1336 FREETMPS; LEAVE;
910 } 1337 }
911 break; 1338 break;
912 } 1339 }
913 1340
914 return sv; 1341 return sv;
921static SV * 1348static SV *
922decode_sv (dec_t *dec) 1349decode_sv (dec_t *dec)
923{ 1350{
924 WANT (1); 1351 WANT (1);
925 1352
926 switch (*dec->cur >> 5) 1353 switch (*dec->cur >> MAJOR_SHIFT)
927 { 1354 {
928 case 0: // unsigned int 1355 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929 return newSVuv (decode_uint (dec)); 1356 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930 case 1: // negative int 1357 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931 return newSViv (-1 - (IV)decode_uint (dec)); 1358 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932 case 2: // octet string 1359 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933 return decode_str (dec, 0); 1360 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934 case 3: // utf-8 string 1361 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935 return decode_str (dec, 1); 1362
936 case 4: // array 1363 case MAJOR_MISC >> MAJOR_SHIFT:
937 return decode_av (dec);
938 case 5: // map
939 return decode_hv (dec);
940 case 6: // tag
941 return decode_tagged (dec);
942 case 7: // misc
943 switch (*dec->cur++ & 31) 1364 switch (*dec->cur++ & MINOR_MASK)
944 { 1365 {
945 case 20: 1366 case SIMPLE_FALSE:
946#if CBOR_SLOW 1367#if CBOR_SLOW
947 types_false = get_bool ("Types::Serialiser::false"); 1368 types_false = get_bool ("Types::Serialiser::false");
948#endif 1369#endif
949 return newSVsv (types_false); 1370 return newSVsv (types_false);
950 case 21: 1371 case SIMPLE_TRUE:
951#if CBOR_SLOW 1372#if CBOR_SLOW
952 types_true = get_bool ("Types::Serialiser::true"); 1373 types_true = get_bool ("Types::Serialiser::true");
953#endif 1374#endif
954 return newSVsv (types_true); 1375 return newSVsv (types_true);
955 case 22: 1376 case SIMPLE_NULL:
956 return newSVsv (&PL_sv_undef); 1377 return newSVsv (&PL_sv_undef);
957 case 23: 1378 case SIMPLE_UNDEF:
958#if CBOR_SLOW 1379#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error"); 1380 types_error = get_bool ("Types::Serialiser::error");
960#endif 1381#endif
961 return newSVsv (types_error); 1382 return newSVsv (types_error);
962 1383
963 case 25: 1384 case MISC_FLOAT16:
964 { 1385 {
965 WANT (2); 1386 WANT (2);
966 1387
967 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1388 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968 dec->cur += 2; 1389 dec->cur += 2;
969 1390
970 return newSVnv (ecb_binary16_to_float (fp)); 1391 return newSVnv (ecb_binary16_to_float (fp));
971 } 1392 }
972 1393
973 case 26: 1394 case MISC_FLOAT32:
974 { 1395 {
975 uint32_t fp; 1396 uint32_t fp;
976 WANT (4); 1397 WANT (4);
977 memcpy (&fp, dec->cur, 4); 1398 memcpy (&fp, dec->cur, 4);
978 dec->cur += 4; 1399 dec->cur += 4;
981 fp = ecb_bswap32 (fp); 1402 fp = ecb_bswap32 (fp);
982 1403
983 return newSVnv (ecb_binary32_to_float (fp)); 1404 return newSVnv (ecb_binary32_to_float (fp));
984 } 1405 }
985 1406
986 case 27: 1407 case MISC_FLOAT64:
987 { 1408 {
988 uint64_t fp; 1409 uint64_t fp;
989 WANT (8); 1410 WANT (8);
990 memcpy (&fp, dec->cur, 8); 1411 memcpy (&fp, dec->cur, 8);
991 dec->cur += 8; 1412 dec->cur += 8;
994 fp = ecb_bswap64 (fp); 1415 fp = ecb_bswap64 (fp);
995 1416
996 return newSVnv (ecb_binary64_to_double (fp)); 1417 return newSVnv (ecb_binary64_to_double (fp));
997 } 1418 }
998 1419
999 // 0..19 unassigned 1420 // 0..19 unassigned simple
1000 // 24 reserved + unassigned (reserved values are not encodable) 1421 // 24 reserved + unassigned simple (reserved values are not encodable)
1422 // 28-30 unassigned misc
1423 // 31 break code
1001 default: 1424 default:
1002 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1425 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003 } 1426 }
1004 1427
1005 break; 1428 break;
1006 } 1429 }
1007 1430
1010} 1433}
1011 1434
1012static SV * 1435static SV *
1013decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1436decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014{ 1437{
1015 dec_t dec = { }; 1438 dec_t dec = { 0 };
1016 SV *sv; 1439 SV *sv;
1017 STRLEN len; 1440 STRLEN len;
1018 char *data = SvPVbyte (string, len); 1441 char *data = SvPVbyte (string, len);
1019 1442
1020 if (len > cbor->max_size && cbor->max_size) 1443 if (len > cbor->max_size && cbor->max_size)
1034 if (dec.cur != dec.end && !dec.err) 1457 if (dec.cur != dec.end && !dec.err)
1035 dec.err = "garbage after CBOR object"; 1458 dec.err = "garbage after CBOR object";
1036 1459
1037 if (dec.err) 1460 if (dec.err)
1038 { 1461 {
1462 if (dec.shareable)
1463 {
1464 // need to break cyclic links, which would all be in shareable
1465 int i;
1466 SV **svp;
1467
1468 for (i = av_len (dec.shareable) + 1; i--; )
1469 if ((svp = av_fetch (dec.shareable, i, 0)))
1470 sv_setsv (*svp, &PL_sv_undef);
1471 }
1472
1039 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1478 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
1041 } 1479 }
1042 1480
1043 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
1044 1482
1045 return sv; 1483 return sv;
1046} 1484}
1047 1485
1486/////////////////////////////////////////////////////////////////////////////
1487// incremental parser
1488
1489#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1490
1491// returns 0 for notyet, 1 for success or error
1492static int
1493incr_parse (CBOR *self, SV *cborstr)
1494{
1495 STRLEN cur;
1496 SvPV (cborstr, cur);
1497
1498 while (ecb_expect_true (self->incr_need <= cur))
1499 {
1500 // table of integer count bytes
1501 static I8 incr_len[MINOR_MASK + 1] = {
1502 0, 0, 0, 0, 0, 0, 0, 0,
1503 0, 0, 0, 0, 0, 0, 0, 0,
1504 0, 0, 0, 0, 0, 0, 0, 0,
1505 1, 2, 4, 8,-1,-1,-1,-2
1506 };
1507
1508 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1509 U8 m = *p & MINOR_MASK;
1510 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1511 I8 ilen = incr_len[m];
1512
1513 self->incr_need = self->incr_pos + 1;
1514
1515 if (ecb_expect_false (ilen < 0))
1516 {
1517 if (m != MINOR_INDEF)
1518 return 1; // error
1519
1520 if (*p == (MAJOR_MISC | MINOR_INDEF))
1521 {
1522 if (count >= 0)
1523 return 1; // error
1524
1525 count = 1;
1526 }
1527 else
1528 {
1529 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1530 count = -1;
1531 }
1532 }
1533 else
1534 {
1535 self->incr_need += ilen;
1536 if (ecb_expect_false (self->incr_need > cur))
1537 return 0;
1538
1539 int major = *p >> MAJOR_SHIFT;
1540
1541 switch (major)
1542 {
1543 case MAJOR_TAG >> MAJOR_SHIFT:
1544 ++count; // tags merely prefix another value
1545 break;
1546
1547 case MAJOR_BYTES >> MAJOR_SHIFT:
1548 case MAJOR_TEXT >> MAJOR_SHIFT:
1549 case MAJOR_ARRAY >> MAJOR_SHIFT:
1550 case MAJOR_MAP >> MAJOR_SHIFT:
1551 {
1552 UV len;
1553
1554 if (ecb_expect_false (ilen))
1555 {
1556 len = 0;
1557
1558 do {
1559 len = (len << 8) | *++p;
1560 } while (--ilen);
1561 }
1562 else
1563 len = m;
1564
1565 switch (major)
1566 {
1567 case MAJOR_BYTES >> MAJOR_SHIFT:
1568 case MAJOR_TEXT >> MAJOR_SHIFT:
1569 self->incr_need += len;
1570 if (ecb_expect_false (self->incr_need > cur))
1571 return 0;
1572
1573 break;
1574
1575 case MAJOR_MAP >> MAJOR_SHIFT:
1576 len <<= 1;
1577 /* FALLTHROUGH */
1578 case MAJOR_ARRAY >> MAJOR_SHIFT:
1579 if (len)
1580 {
1581 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1582 count = len + 1;
1583 }
1584 break;
1585 }
1586 }
1587 }
1588 }
1589
1590 self->incr_pos = self->incr_need;
1591
1592 if (count > 0)
1593 {
1594 while (!--count)
1595 {
1596 if (!AvFILLp (self->incr_count))
1597 return 1; // done
1598
1599 SvREFCNT_dec_NN (av_pop (self->incr_count));
1600 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1601 }
1602
1603 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1604 }
1605 }
1606
1607 return 0;
1608}
1609
1610
1048///////////////////////////////////////////////////////////////////////////// 1611/////////////////////////////////////////////////////////////////////////////
1049// XS interface functions 1612// XS interface functions
1050 1613
1051MODULE = CBOR::XS PACKAGE = CBOR::XS 1614MODULE = CBOR::XS PACKAGE = CBOR::XS
1052 1615
1060 1623
1061 types_true = get_bool ("Types::Serialiser::true" ); 1624 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false"); 1625 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error"); 1626 types_error = get_bool ("Types::Serialiser::error");
1064 1627
1628 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1629
1065 sv_cbor = newSVpv ("CBOR", 0); 1630 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor); 1631 SvREADONLY_on (sv_cbor);
1632
1633 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067} 1634}
1068 1635
1069PROTOTYPES: DISABLE 1636PROTOTYPES: DISABLE
1070 1637
1071void CLONE (...) 1638void CLONE (...)
1090void shrink (CBOR *self, int enable = 1) 1657void shrink (CBOR *self, int enable = 1)
1091 ALIAS: 1658 ALIAS:
1092 shrink = F_SHRINK 1659 shrink = F_SHRINK
1093 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING 1661 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS 1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1096 dedup_strings = F_DEDUP_STRINGS 1664 pack_strings = F_PACK_STRINGS
1665 text_keys = F_TEXT_KEYS
1666 text_strings = F_TEXT_STRINGS
1667 validate_utf8 = F_VALIDATE_UTF8
1097 PPCODE: 1668 PPCODE:
1098{ 1669{
1099 if (enable) 1670 if (enable)
1100 self->flags |= ix; 1671 self->flags |= ix;
1101 else 1672 else
1107void get_shrink (CBOR *self) 1678void get_shrink (CBOR *self)
1108 ALIAS: 1679 ALIAS:
1109 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
1110 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING 1682 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS 1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1113 get_dedup_strings = F_DEDUP_STRINGS 1685 get_pack_strings = F_PACK_STRINGS
1686 get_text_keys = F_TEXT_KEYS
1687 get_text_strings = F_TEXT_STRINGS
1688 get_validate_utf8 = F_VALIDATE_UTF8
1114 PPCODE: 1689 PPCODE:
1115 XPUSHs (boolSV (self->flags & ix)); 1690 XPUSHs (boolSV (self->flags & ix));
1116 1691
1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1692void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118 PPCODE: 1693 PPCODE:
1131 XPUSHs (ST (0)); 1706 XPUSHs (ST (0));
1132 1707
1133int get_max_size (CBOR *self) 1708int get_max_size (CBOR *self)
1134 CODE: 1709 CODE:
1135 RETVAL = self->max_size; 1710 RETVAL = self->max_size;
1711 OUTPUT:
1712 RETVAL
1713
1714void filter (CBOR *self, SV *filter = 0)
1715 PPCODE:
1716 SvREFCNT_dec (self->filter);
1717 self->filter = filter ? newSVsv (filter) : filter;
1718 XPUSHs (ST (0));
1719
1720SV *get_filter (CBOR *self)
1721 CODE:
1722 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136 OUTPUT: 1723 OUTPUT:
1137 RETVAL 1724 RETVAL
1138 1725
1139void encode (CBOR *self, SV *scalar) 1726void encode (CBOR *self, SV *scalar)
1140 PPCODE: 1727 PPCODE:
1155 EXTEND (SP, 2); 1742 EXTEND (SP, 2);
1156 PUSHs (sv); 1743 PUSHs (sv);
1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1744 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158} 1745}
1159 1746
1747void incr_parse (CBOR *self, SV *cborstr)
1748 ALIAS:
1749 incr_parse_multiple = 1
1750 PPCODE:
1751{
1752 if (SvUTF8 (cborstr))
1753 sv_utf8_downgrade (cborstr, 0);
1754
1755 if (!self->incr_count)
1756 {
1757 self->incr_count = newAV ();
1758 self->incr_pos = 0;
1759 self->incr_need = 1;
1760
1761 av_push (self->incr_count, newSViv (1));
1762 }
1763
1764 do
1765 {
1766 if (!incr_parse (self, cborstr))
1767 {
1768 if (self->incr_need > self->max_size && self->max_size)
1769 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1770 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1771
1772 break;
1773 }
1774
1775 SV *sv;
1776 char *offset;
1777
1778 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1779 XPUSHs (sv);
1780
1781 sv_chop (cborstr, offset);
1782
1783 av_clear (self->incr_count);
1784 av_push (self->incr_count, newSViv (1));
1785
1786 self->incr_pos = 0;
1787 self->incr_need = self->incr_pos + 1;
1788 }
1789 while (ix);
1790}
1791
1792void incr_reset (CBOR *self)
1793 CODE:
1794{
1795 SvREFCNT_dec (self->incr_count);
1796 self->incr_count = 0;
1797}
1798
1799void DESTROY (CBOR *self)
1800 PPCODE:
1801 cbor_free (self);
1802
1160PROTOTYPES: ENABLE 1803PROTOTYPES: ENABLE
1161 1804
1162void encode_cbor (SV *scalar) 1805void encode_cbor (SV *scalar)
1806 ALIAS:
1807 encode_cbor = 0
1808 encode_cbor_sharing = F_ALLOW_SHARING
1163 PPCODE: 1809 PPCODE:
1164{ 1810{
1165 CBOR cbor; 1811 CBOR cbor;
1166 cbor_init (&cbor); 1812 cbor_init (&cbor);
1813 cbor.flags |= ix;
1167 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1814 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168 XPUSHs (scalar); 1815 XPUSHs (scalar);
1169} 1816}
1170 1817
1171void decode_cbor (SV *cborstr) 1818void decode_cbor (SV *cborstr)
1175 cbor_init (&cbor); 1822 cbor_init (&cbor);
1176 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1177 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1178} 1825}
1179 1826
1827#ifdef __AFL_COMPILER
1828
1829void
1830afl_init ()
1831 CODE:
1832 __AFL_INIT ();
1833
1834int
1835afl_loop (unsigned int count = 10000)
1836 CODE:
1837 RETVAL = __AFL_LOOP (count);
1838 OUTPUT:
1839 RETVAL
1840
1841#endif
1842

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines