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 1.19 by root, Wed Nov 20 02:03:09 2013 UTC vs.
Revision 1.78 by root, Fri Sep 8 06:01:12 2023 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
35
36// known major and minor types
37enum cbor_type
38{
39 MAJOR_SHIFT = 5,
40 MINOR_MASK = 0x1f,
41
42 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
43 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
44 MAJOR_BYTES = 2 << MAJOR_SHIFT,
45 MAJOR_TEXT = 3 << MAJOR_SHIFT,
46 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
47 MAJOR_MAP = 5 << MAJOR_SHIFT,
48 MAJOR_TAG = 6 << MAJOR_SHIFT,
49 MAJOR_MISC = 7 << MAJOR_SHIFT,
50
51 // INT/STRING/ARRAY/MAP subtypes
52 LENGTH_EXT1 = 24,
53 LENGTH_EXT2 = 25,
54 LENGTH_EXT4 = 26,
55 LENGTH_EXT8 = 27,
56
57 // SIMPLE types (effectively MISC subtypes)
58 SIMPLE_FALSE = 20,
59 SIMPLE_TRUE = 21,
60 SIMPLE_NULL = 22,
61 SIMPLE_UNDEF = 23,
62
63 // MISC subtype (unused)
64 MISC_EXT1 = 24,
65 MISC_FLOAT16 = 25,
66 MISC_FLOAT32 = 26,
67 MISC_FLOAT64 = 27,
68
69 // BYTES/TEXT/ARRAY/MAP
70 MINOR_INDEF = 31,
71};
24 72
25// known tags 73// known tags
26enum cbor_tag 74enum cbor_tag
27{ 75{
28 // inofficial extensions (pending iana registration) 76 // extensions
77 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 78 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 79 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 80 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 81 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 82 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 83 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 84
37 // rfc7049 85 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 86 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 87 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 88 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 89 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 90 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 91 CBOR_TAG_BIGFLOAT = 5, // array
44 92
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 93 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 94 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 95 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 96 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 97
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 98 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 99 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 100 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 101 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 103
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
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
124 SvPV_renew (sv, SvCUR (sv) + 1); 203 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 204#endif
126 } 205 }
127} 206}
128 207
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;
217}
218
129///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
130// encoder 220// encoder
131 221
132// structure used for encoding CBOR 222// structure used for encoding CBOR
133typedef struct 223typedef struct
135 char *cur; // SvPVX (sv) + current output position 225 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 226 char *end; // SvEND (sv)
137 SV *sv; // result scalar 227 SV *sv; // result scalar
138 CBOR cbor; 228 CBOR cbor;
139 U32 depth; // recursion level 229 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 230 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
231 UV stringref_idx;
141 HV *shareable; // ptr => index, or 0 232 HV *shareable; // ptr => index, or 0
142 UV shareable_idx; 233 UV shareable_idx;
143} enc_t; 234} enc_t;
144 235
145ecb_inline void 236ecb_inline void
146need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
147{ 238{
148 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
149 { 240 {
150 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
151 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
152 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
153 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
154 } 245 }
155} 246}
156 247
248static void encode_sv (enc_t *enc, SV *sv);
249
157ecb_inline void 250ecb_inline void
158encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
159{ 252{
160 need (enc, 1); 253 need (enc, 1);
161 *enc->cur++ = ch; 254 *enc->cur++ = ch;
162} 255}
163 256
257// used for tags, intregers, element counts and so on
164static void 258static void
165encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
166{ 260{
167 need (enc, 9); 261 need (enc, 9);
168 262
169 if (len < 24) 263 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
171 else if (len <= 0xff) 265 else if (ecb_expect_true (len <= 0xffU))
172 { 266 {
173 *enc->cur++ = major | 24; 267 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 268 *enc->cur++ = len;
175 } 269 }
176 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
177 { 271 {
178 *enc->cur++ = major | 25; 272 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 274 *enc->cur++ = len;
181 } 275 }
182 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
183 { 277 {
184 *enc->cur++ = major | 26; 278 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 282 *enc->cur++ = len;
189 } 283 }
190 else 284 else
191 { 285 {
192 *enc->cur++ = major | 27; 286 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 287 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 288 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 289 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 290 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 291 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 294 *enc->cur++ = len;
201 } 295 }
202} 296}
203 297
204static 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
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 319{
207 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);
208 need (enc, len); 350 need (enc, len);
209 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
210 enc->cur += len; 352 enc->cur += len;
211} 353}
212 354
213ecb_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
214encode_tag (enc_t *enc, UV tag) 381encode_float16 (enc_t *enc, NV nv)
215{ 382{
216 encode_uint (enc, 0xc0, tag); 383 need (enc, 1+2);
217}
218 384
219static 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}
220 510
221static void 511static void
222encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
223{ 513{
224 int i, len = av_len (av); 514 int i, len = av_len (av);
226 if (enc->depth >= enc->cbor.max_depth) 516 if (enc->depth >= enc->cbor.max_depth)
227 croak (ERR_NESTING_EXCEEDED); 517 croak (ERR_NESTING_EXCEEDED);
228 518
229 ++enc->depth; 519 ++enc->depth;
230 520
231 encode_uint (enc, 0x80, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
232 522
523 if (ecb_expect_false (SvMAGICAL (av)))
233 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
234 { 525 {
235 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
236 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
237 } 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 }
238 535
239 --enc->depth; 536 --enc->depth;
240} 537}
241 538
242static void 539static void
250 ++enc->depth; 547 ++enc->depth;
251 548
252 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
253 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
254 551
255 if (mg) 552 if (ecb_expect_false (mg))
256 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
257 else 554 else
258 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
259 556
260 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
261 { 558 {
262 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
263 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
264 else 561 else
265 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
266 563
267 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));
268 } 565 }
269 566
270 if (mg) 567 if (ecb_expect_false (mg))
271 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
272 569
273 --enc->depth; 570 --enc->depth;
274} 571}
275 572
276// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
277static void 574static void
278encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
279{ 576{
280 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303 578
304 svtype svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
305 580
306 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
307 { 582 {
314 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
315 ? cbor_tagged_stash 590 ? cbor_tagged_stash
316 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
317 592
318 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
319 GV *method;
320 594
321 if (stash == boolean_stash) 595 if (stash == boolean_stash)
322 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
323 else if (stash == error_stash) 600 else if (stash == error_stash)
324 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
325 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
326 { 606 {
327 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
328 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
329 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
330 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
331 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))
332 } 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))));
333 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
334 { 662 {
335 dSP; 663 dSP;
336 664
337 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
338 // 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
339 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
340 669
341 PUTBACK; 670 PUTBACK;
342 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
355 } 684 }
356 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
357 { 686 {
358 dSP; 687 dSP;
359 688
360 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
361 EXTEND (SP, 2); 691 EXTEND (SP, 2);
362 // 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
363 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
364 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
365 695
370 // catch this surprisingly common error 700 // catch this surprisingly common error
371 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
372 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));
373 703
374 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
375 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
377 707
378 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
379 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
380 716
381 PUTBACK; 717 PUTBACK;
382 718
383 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
384 } 720 }
402{ 738{
403 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
404 740
405 need (enc, 9); 741 need (enc, 9);
406 742
407 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
408 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
409 //TODO: maybe I32? 745 //TODO: maybe I32?
410 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
411 { 747 encode_float32 (enc, nv);
412 uint32_t fp = ecb_float_to_binary32 (nv);
413
414 *enc->cur++ = 0xe0 | 26;
415
416 if (!ecb_big_endian ())
417 fp = ecb_bswap32 (fp);
418
419 memcpy (enc->cur, &fp, 4);
420 enc->cur += 4;
421 }
422 else 748 else
423 { 749 encode_float64 (enc, nv);
424 uint64_t fp = ecb_double_to_binary64 (nv);
425
426 *enc->cur++ = 0xe0 | 27;
427
428 if (!ecb_big_endian ())
429 fp = ecb_bswap64 (fp);
430
431 memcpy (enc->cur, &fp, 8);
432 enc->cur += 8;
433 }
434} 750}
435 751
436static void 752static void
437encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
438{ 754{
440 756
441 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
442 { 758 {
443 STRLEN len; 759 STRLEN len;
444 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
445 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
446 } 762 }
447 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
448 encode_nv (enc, sv); 764 encode_nv (enc, sv);
449 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
450 { 766 encode_int (enc, sv);
451 if (SvIsUV (sv))
452 encode_uint (enc, 0x00, SvUVX (sv));
453 else if (SvIVX (sv) >= 0)
454 encode_uint (enc, 0x00, SvIVX (sv));
455 else
456 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
457 }
458 else if (SvROK (sv)) 767 else if (SvROK (sv))
459 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
460 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
461 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
462 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
463 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
464 else 773 else
465 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",
466 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
467} 776}
468 777
469static SV * 778static SV *
470encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
471{ 780{
472 enc_t enc = { }; 781 enc_t enc = { 0 };
473 782
474 enc.cbor = *cbor; 783 enc.cbor = *cbor;
475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
476 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
477 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
478 787
479 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
789
790 if (cbor->flags & F_PACK_STRINGS)
791 {
792 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
793 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
794 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
795 }
796
480 encode_sv (&enc, scalar); 797 encode_sv (&enc, scalar);
481 798
482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 799 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
483 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 800 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
484 801
499 const char *err; // parse error, if != 0 816 const char *err; // parse error, if != 0
500 CBOR cbor; 817 CBOR cbor;
501 U32 depth; // recursion depth 818 U32 depth; // recursion depth
502 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
503 AV *shareable; 820 AV *shareable;
821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
504} dec_t; 824} dec_t;
505 825
506#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);
507 833
508#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;
509 837
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
510#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)
511#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
512 864
513static UV 865static UV
514decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
515{ 867{
516 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
517 { 869 ++dec->cur;
518 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
519 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
520 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
521 return *dec->cur++ & 31;
522 870
523 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
524 WANT (2); 875 WANT (1);
525 dec->cur += 2; 876 dec->cur += 1;
526 return dec->cur[-1]; 877 return dec->cur[-1];
527 878 }
528 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
529 WANT (3); 881 WANT (2);
530 dec->cur += 3; 882 dec->cur += 2;
531 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
532 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
533 885 }
534 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
535 WANT (5); 888 WANT (4);
536 dec->cur += 5; 889 dec->cur += 4;
537 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
538 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
539 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
540 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
541 894 }
542 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
543 WANT (9); 897 WANT (8);
544 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
545 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
546 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
547 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
548 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
549 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
550 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
551 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
552 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
553 913 }
554 default: 914 else
555 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
556 }
557 916
558fail: 917fail:
559 return 0; 918 return 0;
560} 919}
561 920
566{ 925{
567 AV *av = newAV (); 926 AV *av = newAV ();
568 927
569 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
570 929
571 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
572 { 931 {
573 ++dec->cur; 932 ++dec->cur;
574 933
575 for (;;) 934 for (;;)
576 { 935 {
577 WANT (1); 936 WANT (1);
578 937
579 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
580 { 939 {
581 ++dec->cur; 940 ++dec->cur;
582 break; 941 break;
583 } 942 }
584 943
585 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
586 } 945 }
587 } 946 }
588 else 947 else
589 { 948 {
590 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
591 950
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
592 av_fill (av, len - 1); 952 av_fill (av, len - 1);
593 953
594 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
595 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
596 } 956 }
597 957
598 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
599 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
600 960
601fail: 961fail:
602 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
603 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
604 return &PL_sv_undef; 964 return &PL_sv_undef;
605} 965}
606 966
607static void 967static void
608decode_he (dec_t *dec, HV *hv) 968decode_he (dec_t *dec, HV *hv)
609{ 969{
610 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
611 // byte or utf-8 strings as keys. 971 // byte or utf-8 strings as keys, but only when !stringref
612 972
613 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))
614 { 975 {
615 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
616 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
617 978
979 WANT (len);
618 dec->cur += len; 980 dec->cur += len;
619 981
620 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
983
984 return;
621 } 985 }
622 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
623 { 987 {
624 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
625 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
626 990
991 WANT (len);
627 dec->cur += len; 992 dec->cur += len;
628 993
994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
995 if (!cbor_is_utf8_string ((U8 *)key, len))
996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
997
629 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
999
1000 return;
630 } 1001 }
631 else 1002
632 {
633 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
635 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
636 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
638 } 1036
1037fail:
1038 ;
639} 1039}
640 1040
641static SV * 1041static SV *
642decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
643{ 1043{
644 HV *hv = newHV (); 1044 HV *hv = newHV ();
645 1045
646 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
647 1047
648 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 1049 {
650 ++dec->cur; 1050 ++dec->cur;
651 1051
652 for (;;) 1052 for (;;)
653 { 1053 {
654 WANT (1); 1054 WANT (1);
655 1055
656 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
657 { 1057 {
658 ++dec->cur; 1058 ++dec->cur;
659 break; 1059 break;
660 } 1060 }
661 1061
662 decode_he (dec, hv); 1062 decode_he (dec, hv);
663 } 1063 }
664 } 1064 }
665 else 1065 else
666 { 1066 {
667 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
668 1070
669 while (pairs--) 1071 while (pairs--)
670 decode_he (dec, hv); 1072 decode_he (dec, hv);
671 } 1073 }
672 1074
673 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
674 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
675 1077
676fail: 1078fail:
677 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
678 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
679 return &PL_sv_undef; 1081 return &PL_sv_undef;
680} 1082}
681 1083
682static SV * 1084static SV *
683decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
684{ 1086{
685 SV *sv = 0; 1087 SV *sv = 0;
686 1088
687 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
688 { 1090 {
1091 // indefinite length strings
689 ++dec->cur; 1092 ++dec->cur;
690 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
691 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
692 1097
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 1098 for (;;)
695 { 1099 {
696 WANT (1); 1100 WANT (1);
697 1101
698 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 1104 {
700 ++dec->cur; 1105 ++dec->cur;
701 break; 1106 break;
702 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 1110
704 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;
705 } 1116 }
706 } 1117 }
707 else 1118 else
708 { 1119 {
709 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
710 1121
711 WANT (len); 1122 WANT (len);
712 sv = newSVpvn (dec->cur, len); 1123 sv = newSVpvn (dec->cur, len);
713 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));
714 } 1129 }
715 1130
716 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
717 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
718 1139
719 return sv; 1140 return sv;
720 1141
721fail: 1142fail:
722 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
732 WANT (1); 1153 WANT (1);
733 1154
734 switch (tag) 1155 switch (tag)
735 { 1156 {
736 case CBOR_TAG_MAGIC: 1157 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 1158 sv = decode_sv (dec);
1159 break;
738 1160
739 case CBOR_TAG_INDIRECTION: 1161 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 1162 sv = newRV_noinc (decode_sv (dec));
1163 break;
1164
1165 case CBOR_TAG_STRINGREF_NAMESPACE:
1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
1169 ENTER;
1170
1171 SAVESPTR (dec->stringref);
1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1173
1174 sv = decode_sv (dec);
1175
1176 LEAVE;
1177 }
1178 break;
1179
1180 case CBOR_TAG_STRINGREF:
1181 {
1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1184
1185 UV idx = decode_uint (dec);
1186
1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1189
1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1191 }
1192 break;
741 1193
742 case CBOR_TAG_VALUE_SHAREABLE: 1194 case CBOR_TAG_VALUE_SHAREABLE:
743 { 1195 {
744 if (ecb_expect_false (!dec->shareable)) 1196 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
1201 // if cycles are allowed, then we store an AV as value
1202 // while it is being decoded, and gather unresolved
1203 // references in it, to be re4solved after decoding.
1204 int idx, i;
747 sv = newSV (0); 1205 AV *av = newAV ();
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1206 av_push (dec->shareable, (SV *)av);
1207 idx = AvFILLp (dec->shareable);
749 1208
750 SV *osv = decode_sv (dec); 1209 sv = decode_sv (dec);
751 sv_setsv (sv, osv); 1210
1211 // the AV now contains \undef for all unresolved references,
1212 // so we fix them up here.
1213 for (i = 0; i <= AvFILLp (av); ++i)
1214 SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
1215
1216 // now replace the AV by a reference to the completed value
752 SvREFCNT_dec_NN (osv); 1217 SvREFCNT_dec_NN ((SV *)av);
1218 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
753 } 1219 }
754 1220 else
755 return sv; 1221 {
1222 av_push (dec->shareable, &PL_sv_undef);
1223 int idx = AvFILLp (dec->shareable);
1224 sv = decode_sv (dec);
1225 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1226 }
1227 }
1228 break;
756 1229
757 case CBOR_TAG_VALUE_SHAREDREF: 1230 case CBOR_TAG_VALUE_SHAREDREF:
758 { 1231 {
759 if ((*dec->cur >> 5) != 0) 1232 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1233 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 1234
762 UV idx = decode_uint (dec); 1235 UV idx = decode_uint (dec);
763 1236
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 1237 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1238 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 1239
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1240 sv = AvARRAY (dec->shareable)[idx];
1241
1242 // reference to cycle, we create a new \undef and use that, and also
1243 // registerr it in the AV for later fixing
1244 if (SvTYPE (sv) == SVt_PVAV)
1245 {
1246 AV *av = (AV *)sv;
1247 sv = newRV_noinc (&PL_sv_undef);
1248 av_push (av, sv);
768 } 1249 }
1250 else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed
1251 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1252 else // we decoded the object earlier, no cycle
1253 sv = newSVsv (sv);
1254 }
1255 break;
769 1256
770 case CBOR_TAG_PERL_OBJECT: 1257 case CBOR_TAG_PERL_OBJECT:
771 { 1258 {
1259 if (dec->cbor.flags & F_FORBID_OBJECTS)
1260 goto filter;
1261
772 sv = decode_sv (dec); 1262 sv = decode_sv (dec);
773 1263
774 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1264 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
775 ERR ("corrupted CBOR data (non-array perl object)"); 1265 ERR ("corrupted CBOR data (non-array perl object)");
776 1266
786 if (!method) 1276 if (!method)
787 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1277 ERR ("cannot decode perl-object (package does not have a THAW method)");
788 1278
789 dSP; 1279 dSP;
790 1280
791 ENTER; SAVETMPS; PUSHMARK (SP); 1281 ENTER; SAVETMPS;
1282 PUSHMARK (SP);
792 EXTEND (SP, len + 1); 1283 EXTEND (SP, len + 1);
793 // we re-bless the reference to get overload and other niceties right 1284 // we re-bless the reference to get overload and other niceties right
794 PUSHs (*av_fetch (av, 0, 1)); 1285 PUSHs (*av_fetch (av, 0, 1));
795 PUSHs (sv_cbor); 1286 PUSHs (sv_cbor);
796 1287
804 SPAGAIN; 1295 SPAGAIN;
805 1296
806 if (SvTRUE (ERRSV)) 1297 if (SvTRUE (ERRSV))
807 { 1298 {
808 FREETMPS; LEAVE; 1299 FREETMPS; LEAVE;
809 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1300 ERR_ERRSV;
810 } 1301 }
811 1302
812 SvREFCNT_dec (sv); 1303 SvREFCNT_dec_NN (sv);
813 sv = SvREFCNT_inc (POPs); 1304 sv = SvREFCNT_inc (POPs);
814 1305
815 PUTBACK; 1306 PUTBACK;
816 1307
817 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
818
819 return sv;
820 } 1309 }
1310 break;
821 1311
822 default: 1312 default:
1313 filter:
823 { 1314 {
1315 SV *tag_sv = newSVuv (tag);
1316
824 sv = decode_sv (dec); 1317 sv = decode_sv (dec);
825 1318
1319 dSP;
1320 ENTER; SAVETMPS;
1321 PUSHMARK (SP);
1322 EXTEND (SP, 2);
1323 PUSHs (tag_sv);
1324 PUSHs (sv);
1325
1326 PUTBACK;
1327 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1328 SPAGAIN;
1329
1330 if (SvTRUE (ERRSV))
1331 {
1332 SvREFCNT_dec_NN (tag_sv);
1333 FREETMPS; LEAVE;
1334 ERR_ERRSV;
1335 }
1336
1337 if (count)
1338 {
1339 SvREFCNT_dec_NN (tag_sv);
1340 SvREFCNT_dec_NN (sv);
1341 sv = SvREFCNT_inc_NN (TOPs);
1342 SP -= count;
1343 }
1344 else
1345 {
826 AV *av = newAV (); 1346 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1347 av_push (av, tag_sv);
828 av_push (av, sv); 1348 av_push (av, sv);
829 1349
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1350 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1351 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1353 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1354 }
1355
1356 PUTBACK;
1357
1358 FREETMPS; LEAVE;
1359 }
1360 break;
836 } 1361 }
1362
1363 return sv;
837 1364
838fail: 1365fail:
839 SvREFCNT_dec (sv); 1366 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1367 return &PL_sv_undef;
841} 1368}
843static SV * 1370static SV *
844decode_sv (dec_t *dec) 1371decode_sv (dec_t *dec)
845{ 1372{
846 WANT (1); 1373 WANT (1);
847 1374
848 switch (*dec->cur >> 5) 1375 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1376 {
850 case 0: // unsigned int 1377 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1378 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1379 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1380 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1381 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1382 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1383 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1384
858 case 4: // array 1385 case MAJOR_MISC >> MAJOR_SHIFT:
859 return decode_av (dec);
860 case 5: // map
861 return decode_hv (dec);
862 case 6: // tag
863 return decode_tagged (dec);
864 case 7: // misc
865 switch (*dec->cur++ & 31) 1386 switch (*dec->cur++ & MINOR_MASK)
866 { 1387 {
867 case 20: 1388 case SIMPLE_FALSE:
868#if CBOR_SLOW 1389#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1390 types_false = get_bool ("Types::Serialiser::false");
870#endif 1391#endif
871 return newSVsv (types_false); 1392 return newSVsv (types_false);
872 case 21: 1393 case SIMPLE_TRUE:
873#if CBOR_SLOW 1394#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1395 types_true = get_bool ("Types::Serialiser::true");
875#endif 1396#endif
876 return newSVsv (types_true); 1397 return newSVsv (types_true);
877 case 22: 1398 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1399 return newSVsv (&PL_sv_undef);
879 case 23: 1400 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1401#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1402 types_error = get_bool ("Types::Serialiser::error");
882#endif 1403#endif
883 return newSVsv (types_error); 1404 return newSVsv (types_error);
884 1405
885 case 25: 1406 case MISC_FLOAT16:
886 { 1407 {
887 WANT (2); 1408 WANT (2);
888 1409
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1410 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1411 dec->cur += 2;
891 1412
892 return newSVnv (ecb_binary16_to_float (fp)); 1413 return newSVnv (ecb_binary16_to_float (fp));
893 } 1414 }
894 1415
895 case 26: 1416 case MISC_FLOAT32:
896 { 1417 {
897 uint32_t fp; 1418 uint32_t fp;
898 WANT (4); 1419 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1420 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1421 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1424 fp = ecb_bswap32 (fp);
904 1425
905 return newSVnv (ecb_binary32_to_float (fp)); 1426 return newSVnv (ecb_binary32_to_float (fp));
906 } 1427 }
907 1428
908 case 27: 1429 case MISC_FLOAT64:
909 { 1430 {
910 uint64_t fp; 1431 uint64_t fp;
911 WANT (8); 1432 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1433 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1434 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1437 fp = ecb_bswap64 (fp);
917 1438
918 return newSVnv (ecb_binary64_to_double (fp)); 1439 return newSVnv (ecb_binary64_to_double (fp));
919 } 1440 }
920 1441
921 // 0..19 unassigned 1442 // 0..19 unassigned simple
922 // 24 reserved + unassigned (reserved values are not encodable) 1443 // 24 reserved + unassigned simple (reserved values are not encodable)
1444 // 28-30 unassigned misc
1445 // 31 break code
923 default: 1446 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1447 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1448 }
926 1449
927 break; 1450 break;
928 } 1451 }
929 1452
932} 1455}
933 1456
934static SV * 1457static SV *
935decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1458decode_cbor (SV *string, CBOR *cbor, char **offset_return)
936{ 1459{
937 dec_t dec = { }; 1460 dec_t dec = { 0 };
938 SV *sv; 1461 SV *sv;
939 STRLEN len; 1462 STRLEN len;
940 char *data = SvPVbyte (string, len); 1463 char *data = SvPVbyte (string, len);
941 1464
942 if (len > cbor->max_size && cbor->max_size) 1465 if (len > cbor->max_size && cbor->max_size)
956 if (dec.cur != dec.end && !dec.err) 1479 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1480 dec.err = "garbage after CBOR object";
958 1481
959 if (dec.err) 1482 if (dec.err)
960 { 1483 {
1484 if (dec.shareable)
1485 {
1486 // need to break cyclic links, which would all be in shareable
1487 int i;
1488 SV **svp;
1489
1490 for (i = av_len (dec.shareable) + 1; i--; )
1491 if ((svp = av_fetch (dec.shareable, i, 0)))
1492 sv_setsv (*svp, &PL_sv_undef);
1493 }
1494
961 SvREFCNT_dec (sv); 1495 SvREFCNT_dec_NN (sv);
1496
1497 if (dec.err_sv)
1498 sv_2mortal (dec.err_sv);
1499
962 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1500 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
963 } 1501 }
964 1502
965 sv = sv_2mortal (sv); 1503 sv = sv_2mortal (sv);
966 1504
967 return sv; 1505 return sv;
968} 1506}
969 1507
1508/////////////////////////////////////////////////////////////////////////////
1509// incremental parser
1510
1511#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1512
1513// returns 0 for notyet, 1 for success or error
1514static int
1515incr_parse (CBOR *self, SV *cborstr)
1516{
1517 STRLEN cur;
1518 SvPV (cborstr, cur);
1519
1520 while (ecb_expect_true (self->incr_need <= cur))
1521 {
1522 // table of integer count bytes
1523 static I8 incr_len[MINOR_MASK + 1] = {
1524 0, 0, 0, 0, 0, 0, 0, 0,
1525 0, 0, 0, 0, 0, 0, 0, 0,
1526 0, 0, 0, 0, 0, 0, 0, 0,
1527 1, 2, 4, 8,-1,-1,-1,-2
1528 };
1529
1530 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1531 U8 m = *p & MINOR_MASK;
1532 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1533 I8 ilen = incr_len[m];
1534
1535 self->incr_need = self->incr_pos + 1;
1536
1537 if (ecb_expect_false (ilen < 0))
1538 {
1539 if (m != MINOR_INDEF)
1540 return 1; // error
1541
1542 if (*p == (MAJOR_MISC | MINOR_INDEF))
1543 {
1544 if (count >= 0)
1545 return 1; // error
1546
1547 count = 1;
1548 }
1549 else
1550 {
1551 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1552 count = -1;
1553 }
1554 }
1555 else
1556 {
1557 self->incr_need += ilen;
1558 if (ecb_expect_false (self->incr_need > cur))
1559 return 0;
1560
1561 int major = *p >> MAJOR_SHIFT;
1562
1563 switch (major)
1564 {
1565 case MAJOR_TAG >> MAJOR_SHIFT:
1566 ++count; // tags merely prefix another value
1567 break;
1568
1569 case MAJOR_BYTES >> MAJOR_SHIFT:
1570 case MAJOR_TEXT >> MAJOR_SHIFT:
1571 case MAJOR_ARRAY >> MAJOR_SHIFT:
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 {
1574 UV len;
1575
1576 if (ecb_expect_false (ilen))
1577 {
1578 len = 0;
1579
1580 do {
1581 len = (len << 8) | *++p;
1582 } while (--ilen);
1583 }
1584 else
1585 len = m;
1586
1587 switch (major)
1588 {
1589 case MAJOR_BYTES >> MAJOR_SHIFT:
1590 case MAJOR_TEXT >> MAJOR_SHIFT:
1591 self->incr_need += len;
1592 if (ecb_expect_false (self->incr_need > cur))
1593 return 0;
1594
1595 break;
1596
1597 case MAJOR_MAP >> MAJOR_SHIFT:
1598 len <<= 1;
1599 /* FALLTHROUGH */
1600 case MAJOR_ARRAY >> MAJOR_SHIFT:
1601 if (len)
1602 {
1603 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1604 count = len + 1;
1605 }
1606 break;
1607 }
1608 }
1609 }
1610 }
1611
1612 self->incr_pos = self->incr_need;
1613
1614 if (count > 0)
1615 {
1616 while (!--count)
1617 {
1618 if (!AvFILLp (self->incr_count))
1619 return 1; // done
1620
1621 SvREFCNT_dec_NN (av_pop (self->incr_count));
1622 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1623 }
1624
1625 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1626 }
1627 }
1628
1629 return 0;
1630}
1631
1632
970///////////////////////////////////////////////////////////////////////////// 1633/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1634// XS interface functions
972 1635
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1636MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1637
982 1645
983 types_true = get_bool ("Types::Serialiser::true" ); 1646 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1647 types_false = get_bool ("Types::Serialiser::false");
985 types_error = get_bool ("Types::Serialiser::error"); 1648 types_error = get_bool ("Types::Serialiser::error");
986 1649
1650 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1651
987 sv_cbor = newSVpv ("CBOR", 0); 1652 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1653 SvREADONLY_on (sv_cbor);
1654
1655 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
989} 1656}
990 1657
991PROTOTYPES: DISABLE 1658PROTOTYPES: DISABLE
992 1659
993void CLONE (...) 1660void CLONE (...)
1012void shrink (CBOR *self, int enable = 1) 1679void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1680 ALIAS:
1014 shrink = F_SHRINK 1681 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1682 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1683 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1684 allow_cycles = F_ALLOW_CYCLES
1685 forbid_objects = F_FORBID_OBJECTS
1018 dedup_strings = F_DEDUP_STRINGS 1686 pack_strings = F_PACK_STRINGS
1687 text_keys = F_TEXT_KEYS
1688 text_strings = F_TEXT_STRINGS
1689 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1690 PPCODE:
1020{ 1691{
1021 if (enable) 1692 if (enable)
1022 self->flags |= ix; 1693 self->flags |= ix;
1023 else 1694 else
1029void get_shrink (CBOR *self) 1700void get_shrink (CBOR *self)
1030 ALIAS: 1701 ALIAS:
1031 get_shrink = F_SHRINK 1702 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1703 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1704 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1705 get_allow_cycles = F_ALLOW_CYCLES
1706 get_forbid_objects = F_FORBID_OBJECTS
1035 get_dedup_strings = F_DEDUP_STRINGS 1707 get_pack_strings = F_PACK_STRINGS
1708 get_text_keys = F_TEXT_KEYS
1709 get_text_strings = F_TEXT_STRINGS
1710 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1711 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1712 XPUSHs (boolSV (self->flags & ix));
1038 1713
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1714void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1715 PPCODE:
1053 XPUSHs (ST (0)); 1728 XPUSHs (ST (0));
1054 1729
1055int get_max_size (CBOR *self) 1730int get_max_size (CBOR *self)
1056 CODE: 1731 CODE:
1057 RETVAL = self->max_size; 1732 RETVAL = self->max_size;
1733 OUTPUT:
1734 RETVAL
1735
1736void filter (CBOR *self, SV *filter = 0)
1737 PPCODE:
1738 SvREFCNT_dec (self->filter);
1739 self->filter = filter ? newSVsv (filter) : filter;
1740 XPUSHs (ST (0));
1741
1742SV *get_filter (CBOR *self)
1743 CODE:
1744 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1058 OUTPUT: 1745 OUTPUT:
1059 RETVAL 1746 RETVAL
1060 1747
1061void encode (CBOR *self, SV *scalar) 1748void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1749 PPCODE:
1077 EXTEND (SP, 2); 1764 EXTEND (SP, 2);
1078 PUSHs (sv); 1765 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1766 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1767}
1081 1768
1769void incr_parse (CBOR *self, SV *cborstr)
1770 ALIAS:
1771 incr_parse_multiple = 1
1772 PPCODE:
1773{
1774 if (SvUTF8 (cborstr))
1775 sv_utf8_downgrade (cborstr, 0);
1776
1777 if (!self->incr_count)
1778 {
1779 self->incr_count = newAV ();
1780 self->incr_pos = 0;
1781 self->incr_need = 1;
1782
1783 av_push (self->incr_count, newSViv (1));
1784 }
1785
1786 do
1787 {
1788 if (!incr_parse (self, cborstr))
1789 {
1790 if (self->incr_need > self->max_size && self->max_size)
1791 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1792 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1793
1794 break;
1795 }
1796
1797 SV *sv;
1798 char *offset;
1799
1800 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1801 XPUSHs (sv);
1802
1803 sv_chop (cborstr, offset);
1804
1805 av_clear (self->incr_count);
1806 av_push (self->incr_count, newSViv (1));
1807
1808 self->incr_pos = 0;
1809 self->incr_need = self->incr_pos + 1;
1810 }
1811 while (ix);
1812}
1813
1814void incr_reset (CBOR *self)
1815 CODE:
1816{
1817 SvREFCNT_dec (self->incr_count);
1818 self->incr_count = 0;
1819}
1820
1821void DESTROY (CBOR *self)
1822 PPCODE:
1823 cbor_free (self);
1824
1082PROTOTYPES: ENABLE 1825PROTOTYPES: ENABLE
1083 1826
1084void encode_cbor (SV *scalar) 1827void encode_cbor (SV *scalar)
1828 ALIAS:
1829 encode_cbor = 0
1830 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1831 PPCODE:
1086{ 1832{
1087 CBOR cbor; 1833 CBOR cbor;
1088 cbor_init (&cbor); 1834 cbor_init (&cbor);
1835 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1836 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1837 XPUSHs (scalar);
1091} 1838}
1092 1839
1093void decode_cbor (SV *cborstr) 1840void decode_cbor (SV *cborstr)
1097 cbor_init (&cbor); 1844 cbor_init (&cbor);
1098 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1845 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1099 XPUSHs (cborstr); 1846 XPUSHs (cborstr);
1100} 1847}
1101 1848
1849#ifdef __AFL_COMPILER
1850
1851void
1852afl_init ()
1853 CODE:
1854 __AFL_INIT ();
1855
1856int
1857afl_loop (unsigned int count = 10000)
1858 CODE:
1859 RETVAL = __AFL_LOOP (count);
1860 OUTPUT:
1861 RETVAL
1862
1863#endif
1864

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines