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 1.72 by root, Thu Oct 21 01:14:58 2021 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// 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;
990 printf ("len %d\n", len);//D
626 991
992 WANT (len);
627 dec->cur += len; 993 dec->cur += len;
628 994
995 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
996 if (!cbor_is_utf8_string ((U8 *)key, len))
997 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
998
629 hv_store (hv, key, -len, decode_sv (dec), 0); 999 hv_store (hv, key, -len, decode_sv (dec), 0);
1000
1001 return;
630 } 1002 }
631 else 1003
632 {
633 SV *k = decode_sv (dec); 1004 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 1005 SV *v = decode_sv (dec);
635 1006
1007 // we leak memory if uncaught exceptions are thrown by random magical
1008 // methods, and this is hopefully the only place where it can happen,
1009 // so if there is a chance of an exception, take the very slow path.
1010 // since catching exceptions is "undocumented/internal/forbidden" by
1011 // the new p5p powers, we need to call out to a perl function :/
1012 if (ecb_expect_false (SvAMAGIC (k)))
1013 {
1014 dSP;
1015
1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
1018 EXTEND (SP, 3);
1019 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1020 PUSHs (sv_2mortal (k));
1021 PUSHs (sv_2mortal (v));
1022
1023 PUTBACK;
1024 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1025 SPAGAIN;
1026
1027 FREETMPS; LEAVE;
1028
1029 if (SvTRUE (ERRSV))
1030 ERR_ERRSV;
1031
1032 return;
1033 }
1034
636 hv_store_ent (hv, k, v, 0); 1035 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 1036 SvREFCNT_dec_NN (k);
638 } 1037
1038fail:
1039 ;
639} 1040}
640 1041
641static SV * 1042static SV *
642decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
643{ 1044{
644 HV *hv = newHV (); 1045 HV *hv = newHV ();
645 1046
646 DEC_INC_DEPTH; 1047 DEC_INC_DEPTH;
647 1048
648 if ((*dec->cur & 31) == 31) 1049 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 1050 {
650 ++dec->cur; 1051 ++dec->cur;
651 1052
652 for (;;) 1053 for (;;)
653 { 1054 {
654 WANT (1); 1055 WANT (1);
655 1056
656 if (*dec->cur == (0xe0 | 31)) 1057 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
657 { 1058 {
658 ++dec->cur; 1059 ++dec->cur;
659 break; 1060 break;
660 } 1061 }
661 1062
662 decode_he (dec, hv); 1063 decode_he (dec, hv);
663 } 1064 }
664 } 1065 }
665 else 1066 else
666 { 1067 {
667 int pairs = decode_uint (dec); 1068 UV pairs = decode_uint (dec);
1069
1070 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
668 1071
669 while (pairs--) 1072 while (pairs--)
670 decode_he (dec, hv); 1073 decode_he (dec, hv);
671 } 1074 }
672 1075
673 DEC_DEC_DEPTH; 1076 DEC_DEC_DEPTH;
674 return newRV_noinc ((SV *)hv); 1077 return newRV_noinc ((SV *)hv);
675 1078
676fail: 1079fail:
677 SvREFCNT_dec (hv); 1080 SvREFCNT_dec_NN (hv);
678 DEC_DEC_DEPTH; 1081 DEC_DEC_DEPTH;
679 return &PL_sv_undef; 1082 return &PL_sv_undef;
680} 1083}
681 1084
682static SV * 1085static SV *
683decode_str (dec_t *dec, int utf8) 1086decode_str (dec_t *dec, int utf8)
684{ 1087{
685 SV *sv = 0; 1088 SV *sv = 0;
686 1089
687 if ((*dec->cur & 31) == 31) 1090 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
688 { 1091 {
1092 // indefinite length strings
689 ++dec->cur; 1093 ++dec->cur;
690 1094
1095 U8 major = *dec->cur & MAJOR_MISC;
1096
691 sv = newSVpvn ("", 0); 1097 sv = newSVpvn ("", 0);
692 1098
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 1099 for (;;)
695 { 1100 {
696 WANT (1); 1101 WANT (1);
697 1102
698 if (*dec->cur == (0xe0 | 31)) 1103 if ((*dec->cur - major) > LENGTH_EXT8)
1104 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 1105 {
700 ++dec->cur; 1106 ++dec->cur;
701 break; 1107 break;
702 } 1108 }
1109 else
1110 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 1111
704 sv_catsv (sv, decode_sv (dec)); 1112 STRLEN len = decode_uint (dec);
1113
1114 WANT (len);
1115 sv_catpvn (sv, dec->cur, len);
1116 dec->cur += len;
705 } 1117 }
706 } 1118 }
707 else 1119 else
708 { 1120 {
709 STRLEN len = decode_uint (dec); 1121 STRLEN len = decode_uint (dec);
710 1122
711 WANT (len); 1123 WANT (len);
712 sv = newSVpvn (dec->cur, len); 1124 sv = newSVpvn (dec->cur, len);
713 dec->cur += len; 1125 dec->cur += len;
1126
1127 if (ecb_expect_false (dec->stringref)
1128 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1129 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
714 } 1130 }
715 1131
716 if (utf8) 1132 if (utf8)
1133 {
1134 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1135 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1136 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1137
717 SvUTF8_on (sv); 1138 SvUTF8_on (sv);
1139 }
718 1140
719 return sv; 1141 return sv;
720 1142
721fail: 1143fail:
722 SvREFCNT_dec (sv); 1144 SvREFCNT_dec (sv);
732 WANT (1); 1154 WANT (1);
733 1155
734 switch (tag) 1156 switch (tag)
735 { 1157 {
736 case CBOR_TAG_MAGIC: 1158 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 1159 sv = decode_sv (dec);
1160 break;
738 1161
739 case CBOR_TAG_INDIRECTION: 1162 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 1163 sv = newRV_noinc (decode_sv (dec));
1164 break;
1165
1166 case CBOR_TAG_STRINGREF_NAMESPACE:
1167 {
1168 // do not use SAVETMPS/FREETMPS, as these will
1169 // erase mortalised caches, e.g. "shareable"
1170 ENTER;
1171
1172 SAVESPTR (dec->stringref);
1173 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1174
1175 sv = decode_sv (dec);
1176
1177 LEAVE;
1178 }
1179 break;
1180
1181 case CBOR_TAG_STRINGREF:
1182 {
1183 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1184 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1185
1186 UV idx = decode_uint (dec);
1187
1188 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1189 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1190
1191 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1192 }
1193 break;
741 1194
742 case CBOR_TAG_VALUE_SHAREABLE: 1195 case CBOR_TAG_VALUE_SHAREABLE:
743 { 1196 {
744 if (ecb_expect_false (!dec->shareable)) 1197 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1198 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 1199
1200 if (dec->cbor.flags & F_ALLOW_CYCLES)
1201 {
747 sv = newSV (0); 1202 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1203 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749 1204
750 SV *osv = decode_sv (dec); 1205 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv); 1206 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv); 1207 SvREFCNT_dec_NN (osv);
753 } 1208 }
754 1209 else
755 return sv; 1210 {
1211 av_push (dec->shareable, &PL_sv_undef);
1212 int idx = AvFILLp (dec->shareable);
1213 sv = decode_sv (dec);
1214 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1215 }
1216 }
1217 break;
756 1218
757 case CBOR_TAG_VALUE_SHAREDREF: 1219 case CBOR_TAG_VALUE_SHAREDREF:
758 { 1220 {
759 if ((*dec->cur >> 5) != 0) 1221 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1222 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 1223
762 UV idx = decode_uint (dec); 1224 UV idx = decode_uint (dec);
763 1225
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 1226 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1227 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 1228
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1229 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1230
1231 if (sv == &PL_sv_undef)
1232 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
768 } 1233 }
1234 break;
769 1235
770 case CBOR_TAG_PERL_OBJECT: 1236 case CBOR_TAG_PERL_OBJECT:
771 { 1237 {
1238 if (dec->cbor.flags & F_FORBID_OBJECTS)
1239 goto filter;
1240
772 sv = decode_sv (dec); 1241 sv = decode_sv (dec);
773 1242
774 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1243 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
775 ERR ("corrupted CBOR data (non-array perl object)"); 1244 ERR ("corrupted CBOR data (non-array perl object)");
776 1245
786 if (!method) 1255 if (!method)
787 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1256 ERR ("cannot decode perl-object (package does not have a THAW method)");
788 1257
789 dSP; 1258 dSP;
790 1259
791 ENTER; SAVETMPS; PUSHMARK (SP); 1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
792 EXTEND (SP, len + 1); 1262 EXTEND (SP, len + 1);
793 // we re-bless the reference to get overload and other niceties right 1263 // we re-bless the reference to get overload and other niceties right
794 PUSHs (*av_fetch (av, 0, 1)); 1264 PUSHs (*av_fetch (av, 0, 1));
795 PUSHs (sv_cbor); 1265 PUSHs (sv_cbor);
796 1266
804 SPAGAIN; 1274 SPAGAIN;
805 1275
806 if (SvTRUE (ERRSV)) 1276 if (SvTRUE (ERRSV))
807 { 1277 {
808 FREETMPS; LEAVE; 1278 FREETMPS; LEAVE;
809 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1279 ERR_ERRSV;
810 } 1280 }
811 1281
812 SvREFCNT_dec (sv); 1282 SvREFCNT_dec_NN (sv);
813 sv = SvREFCNT_inc (POPs); 1283 sv = SvREFCNT_inc (POPs);
814 1284
815 PUTBACK; 1285 PUTBACK;
816 1286
817 FREETMPS; LEAVE; 1287 FREETMPS; LEAVE;
818
819 return sv;
820 } 1288 }
1289 break;
821 1290
822 default: 1291 default:
1292 filter:
823 { 1293 {
1294 SV *tag_sv = newSVuv (tag);
1295
824 sv = decode_sv (dec); 1296 sv = decode_sv (dec);
825 1297
1298 dSP;
1299 ENTER; SAVETMPS;
1300 PUSHMARK (SP);
1301 EXTEND (SP, 2);
1302 PUSHs (tag_sv);
1303 PUSHs (sv);
1304
1305 PUTBACK;
1306 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1307 SPAGAIN;
1308
1309 if (SvTRUE (ERRSV))
1310 {
1311 SvREFCNT_dec_NN (tag_sv);
1312 FREETMPS; LEAVE;
1313 ERR_ERRSV;
1314 }
1315
1316 if (count)
1317 {
1318 SvREFCNT_dec_NN (tag_sv);
1319 SvREFCNT_dec_NN (sv);
1320 sv = SvREFCNT_inc_NN (TOPs);
1321 SP -= count;
1322 }
1323 else
1324 {
826 AV *av = newAV (); 1325 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1326 av_push (av, tag_sv);
828 av_push (av, sv); 1327 av_push (av, sv);
829 1328
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1329 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1330 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1331 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1332 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1333 }
1334
1335 PUTBACK;
1336
1337 FREETMPS; LEAVE;
1338 }
1339 break;
836 } 1340 }
1341
1342 return sv;
837 1343
838fail: 1344fail:
839 SvREFCNT_dec (sv); 1345 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1346 return &PL_sv_undef;
841} 1347}
843static SV * 1349static SV *
844decode_sv (dec_t *dec) 1350decode_sv (dec_t *dec)
845{ 1351{
846 WANT (1); 1352 WANT (1);
847 1353
848 switch (*dec->cur >> 5) 1354 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1355 {
850 case 0: // unsigned int 1356 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1357 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1358 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1359 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1360 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1361 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1362 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1363
858 case 4: // array 1364 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) 1365 switch (*dec->cur++ & MINOR_MASK)
866 { 1366 {
867 case 20: 1367 case SIMPLE_FALSE:
868#if CBOR_SLOW 1368#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1369 types_false = get_bool ("Types::Serialiser::false");
870#endif 1370#endif
871 return newSVsv (types_false); 1371 return newSVsv (types_false);
872 case 21: 1372 case SIMPLE_TRUE:
873#if CBOR_SLOW 1373#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1374 types_true = get_bool ("Types::Serialiser::true");
875#endif 1375#endif
876 return newSVsv (types_true); 1376 return newSVsv (types_true);
877 case 22: 1377 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1378 return newSVsv (&PL_sv_undef);
879 case 23: 1379 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1380#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1381 types_error = get_bool ("Types::Serialiser::error");
882#endif 1382#endif
883 return newSVsv (types_error); 1383 return newSVsv (types_error);
884 1384
885 case 25: 1385 case MISC_FLOAT16:
886 { 1386 {
887 WANT (2); 1387 WANT (2);
888 1388
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1389 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1390 dec->cur += 2;
891 1391
892 return newSVnv (ecb_binary16_to_float (fp)); 1392 return newSVnv (ecb_binary16_to_float (fp));
893 } 1393 }
894 1394
895 case 26: 1395 case MISC_FLOAT32:
896 { 1396 {
897 uint32_t fp; 1397 uint32_t fp;
898 WANT (4); 1398 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1399 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1400 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1403 fp = ecb_bswap32 (fp);
904 1404
905 return newSVnv (ecb_binary32_to_float (fp)); 1405 return newSVnv (ecb_binary32_to_float (fp));
906 } 1406 }
907 1407
908 case 27: 1408 case MISC_FLOAT64:
909 { 1409 {
910 uint64_t fp; 1410 uint64_t fp;
911 WANT (8); 1411 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1412 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1413 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1416 fp = ecb_bswap64 (fp);
917 1417
918 return newSVnv (ecb_binary64_to_double (fp)); 1418 return newSVnv (ecb_binary64_to_double (fp));
919 } 1419 }
920 1420
921 // 0..19 unassigned 1421 // 0..19 unassigned simple
922 // 24 reserved + unassigned (reserved values are not encodable) 1422 // 24 reserved + unassigned simple (reserved values are not encodable)
1423 // 28-30 unassigned misc
1424 // 31 break code
923 default: 1425 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1426 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1427 }
926 1428
927 break; 1429 break;
928 } 1430 }
929 1431
932} 1434}
933 1435
934static SV * 1436static SV *
935decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1437decode_cbor (SV *string, CBOR *cbor, char **offset_return)
936{ 1438{
937 dec_t dec = { }; 1439 dec_t dec = { 0 };
938 SV *sv; 1440 SV *sv;
939 STRLEN len; 1441 STRLEN len;
940 char *data = SvPVbyte (string, len); 1442 char *data = SvPVbyte (string, len);
941 1443
942 if (len > cbor->max_size && cbor->max_size) 1444 if (len > cbor->max_size && cbor->max_size)
956 if (dec.cur != dec.end && !dec.err) 1458 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1459 dec.err = "garbage after CBOR object";
958 1460
959 if (dec.err) 1461 if (dec.err)
960 { 1462 {
1463 if (dec.shareable)
1464 {
1465 // need to break cyclic links, which would all be in shareable
1466 int i;
1467 SV **svp;
1468
1469 for (i = av_len (dec.shareable) + 1; i--; )
1470 if ((svp = av_fetch (dec.shareable, i, 0)))
1471 sv_setsv (*svp, &PL_sv_undef);
1472 }
1473
961 SvREFCNT_dec (sv); 1474 SvREFCNT_dec_NN (sv);
1475
1476 if (dec.err_sv)
1477 sv_2mortal (dec.err_sv);
1478
962 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1479 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
963 } 1480 }
964 1481
965 sv = sv_2mortal (sv); 1482 sv = sv_2mortal (sv);
966 1483
967 return sv; 1484 return sv;
968} 1485}
969 1486
1487/////////////////////////////////////////////////////////////////////////////
1488// incremental parser
1489
1490#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1491
1492// returns 0 for notyet, 1 for success or error
1493static int
1494incr_parse (CBOR *self, SV *cborstr)
1495{
1496 STRLEN cur;
1497 SvPV (cborstr, cur);
1498
1499 while (ecb_expect_true (self->incr_need <= cur))
1500 {
1501 // table of integer count bytes
1502 static I8 incr_len[MINOR_MASK + 1] = {
1503 0, 0, 0, 0, 0, 0, 0, 0,
1504 0, 0, 0, 0, 0, 0, 0, 0,
1505 0, 0, 0, 0, 0, 0, 0, 0,
1506 1, 2, 4, 8,-1,-1,-1,-2
1507 };
1508
1509 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1510 U8 m = *p & MINOR_MASK;
1511 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1512 I8 ilen = incr_len[m];
1513
1514 self->incr_need = self->incr_pos + 1;
1515
1516 if (ecb_expect_false (ilen < 0))
1517 {
1518 if (m != MINOR_INDEF)
1519 return 1; // error
1520
1521 if (*p == (MAJOR_MISC | MINOR_INDEF))
1522 {
1523 if (count >= 0)
1524 return 1; // error
1525
1526 count = 1;
1527 }
1528 else
1529 {
1530 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1531 count = -1;
1532 }
1533 }
1534 else
1535 {
1536 self->incr_need += ilen;
1537 if (ecb_expect_false (self->incr_need > cur))
1538 return 0;
1539
1540 int major = *p >> MAJOR_SHIFT;
1541
1542 switch (major)
1543 {
1544 case MAJOR_TAG >> MAJOR_SHIFT:
1545 ++count; // tags merely prefix another value
1546 break;
1547
1548 case MAJOR_BYTES >> MAJOR_SHIFT:
1549 case MAJOR_TEXT >> MAJOR_SHIFT:
1550 case MAJOR_ARRAY >> MAJOR_SHIFT:
1551 case MAJOR_MAP >> MAJOR_SHIFT:
1552 {
1553 UV len;
1554
1555 if (ecb_expect_false (ilen))
1556 {
1557 len = 0;
1558
1559 do {
1560 len = (len << 8) | *++p;
1561 } while (--ilen);
1562 }
1563 else
1564 len = m;
1565
1566 switch (major)
1567 {
1568 case MAJOR_BYTES >> MAJOR_SHIFT:
1569 case MAJOR_TEXT >> MAJOR_SHIFT:
1570 self->incr_need += len;
1571 if (ecb_expect_false (self->incr_need > cur))
1572 return 0;
1573
1574 break;
1575
1576 case MAJOR_MAP >> MAJOR_SHIFT:
1577 len <<= 1;
1578 /* FALLTHROUGH */
1579 case MAJOR_ARRAY >> MAJOR_SHIFT:
1580 if (len)
1581 {
1582 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1583 count = len + 1;
1584 }
1585 break;
1586 }
1587 }
1588 }
1589 }
1590
1591 self->incr_pos = self->incr_need;
1592
1593 if (count > 0)
1594 {
1595 while (!--count)
1596 {
1597 if (!AvFILLp (self->incr_count))
1598 return 1; // done
1599
1600 SvREFCNT_dec_NN (av_pop (self->incr_count));
1601 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1602 }
1603
1604 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1605 }
1606 }
1607
1608 return 0;
1609}
1610
1611
970///////////////////////////////////////////////////////////////////////////// 1612/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1613// XS interface functions
972 1614
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1615MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1616
982 1624
983 types_true = get_bool ("Types::Serialiser::true" ); 1625 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1626 types_false = get_bool ("Types::Serialiser::false");
985 types_error = get_bool ("Types::Serialiser::error"); 1627 types_error = get_bool ("Types::Serialiser::error");
986 1628
1629 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1630
987 sv_cbor = newSVpv ("CBOR", 0); 1631 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1632 SvREADONLY_on (sv_cbor);
1633
1634 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
989} 1635}
990 1636
991PROTOTYPES: DISABLE 1637PROTOTYPES: DISABLE
992 1638
993void CLONE (...) 1639void CLONE (...)
1012void shrink (CBOR *self, int enable = 1) 1658void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1659 ALIAS:
1014 shrink = F_SHRINK 1660 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1661 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1662 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1663 allow_cycles = F_ALLOW_CYCLES
1664 forbid_objects = F_FORBID_OBJECTS
1018 dedup_strings = F_DEDUP_STRINGS 1665 pack_strings = F_PACK_STRINGS
1666 text_keys = F_TEXT_KEYS
1667 text_strings = F_TEXT_STRINGS
1668 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1669 PPCODE:
1020{ 1670{
1021 if (enable) 1671 if (enable)
1022 self->flags |= ix; 1672 self->flags |= ix;
1023 else 1673 else
1029void get_shrink (CBOR *self) 1679void get_shrink (CBOR *self)
1030 ALIAS: 1680 ALIAS:
1031 get_shrink = F_SHRINK 1681 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1682 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1683 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1684 get_allow_cycles = F_ALLOW_CYCLES
1685 get_forbid_objects = F_FORBID_OBJECTS
1035 get_dedup_strings = F_DEDUP_STRINGS 1686 get_pack_strings = F_PACK_STRINGS
1687 get_text_keys = F_TEXT_KEYS
1688 get_text_strings = F_TEXT_STRINGS
1689 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1690 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1691 XPUSHs (boolSV (self->flags & ix));
1038 1692
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1693void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1694 PPCODE:
1053 XPUSHs (ST (0)); 1707 XPUSHs (ST (0));
1054 1708
1055int get_max_size (CBOR *self) 1709int get_max_size (CBOR *self)
1056 CODE: 1710 CODE:
1057 RETVAL = self->max_size; 1711 RETVAL = self->max_size;
1712 OUTPUT:
1713 RETVAL
1714
1715void filter (CBOR *self, SV *filter = 0)
1716 PPCODE:
1717 SvREFCNT_dec (self->filter);
1718 self->filter = filter ? newSVsv (filter) : filter;
1719 XPUSHs (ST (0));
1720
1721SV *get_filter (CBOR *self)
1722 CODE:
1723 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1058 OUTPUT: 1724 OUTPUT:
1059 RETVAL 1725 RETVAL
1060 1726
1061void encode (CBOR *self, SV *scalar) 1727void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1728 PPCODE:
1077 EXTEND (SP, 2); 1743 EXTEND (SP, 2);
1078 PUSHs (sv); 1744 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1745 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1746}
1081 1747
1748void incr_parse (CBOR *self, SV *cborstr)
1749 ALIAS:
1750 incr_parse_multiple = 1
1751 PPCODE:
1752{
1753 if (SvUTF8 (cborstr))
1754 sv_utf8_downgrade (cborstr, 0);
1755
1756 if (!self->incr_count)
1757 {
1758 self->incr_count = newAV ();
1759 self->incr_pos = 0;
1760 self->incr_need = 1;
1761
1762 av_push (self->incr_count, newSViv (1));
1763 }
1764
1765 do
1766 {
1767 if (!incr_parse (self, cborstr))
1768 {
1769 if (self->incr_need > self->max_size && self->max_size)
1770 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1771 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1772
1773 break;
1774 }
1775
1776 SV *sv;
1777 char *offset;
1778
1779 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1780 XPUSHs (sv);
1781
1782 sv_chop (cborstr, offset);
1783
1784 av_clear (self->incr_count);
1785 av_push (self->incr_count, newSViv (1));
1786
1787 self->incr_pos = 0;
1788 self->incr_need = self->incr_pos + 1;
1789 }
1790 while (ix);
1791}
1792
1793void incr_reset (CBOR *self)
1794 CODE:
1795{
1796 SvREFCNT_dec (self->incr_count);
1797 self->incr_count = 0;
1798}
1799
1800void DESTROY (CBOR *self)
1801 PPCODE:
1802 cbor_free (self);
1803
1082PROTOTYPES: ENABLE 1804PROTOTYPES: ENABLE
1083 1805
1084void encode_cbor (SV *scalar) 1806void encode_cbor (SV *scalar)
1807 ALIAS:
1808 encode_cbor = 0
1809 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1810 PPCODE:
1086{ 1811{
1087 CBOR cbor; 1812 CBOR cbor;
1088 cbor_init (&cbor); 1813 cbor_init (&cbor);
1814 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1815 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1816 XPUSHs (scalar);
1091} 1817}
1092 1818
1093void decode_cbor (SV *cborstr) 1819void decode_cbor (SV *cborstr)
1097 cbor_init (&cbor); 1823 cbor_init (&cbor);
1098 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1824 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1099 XPUSHs (cborstr); 1825 XPUSHs (cborstr);
1100} 1826}
1101 1827
1828#ifdef __AFL_COMPILER
1829
1830void
1831afl_init ()
1832 CODE:
1833 __AFL_INIT ();
1834
1835int
1836afl_loop (unsigned int count = 10000)
1837 CODE:
1838 RETVAL = __AFL_LOOP (count);
1839 OUTPUT:
1840 RETVAL
1841
1842#endif
1843

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines