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 1.16 by root, Tue Oct 29 20:59:16 2013 UTC vs.
Revision 1.73 by root, Sat Oct 23 03:00:31 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
29 CBOR_TAG_PERL_OBJECT = 256, 77 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 78 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
79 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
80 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
81 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
82 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
83 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 84
32 // rfc7049 85 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 86 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 87 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 88 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 89 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 90 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 91 CBOR_TAG_BIGFLOAT = 5, // array
39 92
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 93 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 94 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 95 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 96 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 97
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 98 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 99 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 100 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 101 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 103
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 105};
53 106
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
54#define F_SHRINK 0x00000200UL 121#define F_SHRINK 0x00000001UL
55#define F_ALLOW_UNKNOWN 0x00002000UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
123#define F_ALLOW_SHARING 0x00000004UL
124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
126#define F_PACK_STRINGS 0x00000020UL
127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
56 130
57#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
58 132
59#define SB do { 133#define SB do {
60#define SE } while (0) 134#define SE } while (0)
72# define CBOR_SLOW 0 146# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 147# define CBOR_STASH cbor_stash
74#endif 148#endif
75 149
76static 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::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 151static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 152
79typedef struct { 153typedef struct {
80 U32 flags; 154 U32 flags;
81 U32 max_depth; 155 U32 max_depth;
82 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.
83} CBOR; 163} CBOR;
84 164
85ecb_inline void 165ecb_inline void
86cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
87{ 167{
88 Zero (cbor, 1, CBOR); 168 Zero (cbor, 1, CBOR);
89 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);
90} 177}
91 178
92///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
93// utility functions 180// utility functions
94 181
116 SvPV_renew (sv, SvCUR (sv) + 1); 203 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 204#endif
118 } 205 }
119} 206}
120 207
121///////////////////////////////////////////////////////////////////////////// 208// minimum length of a string to be registered for stringref
122// fp hell 209ecb_inline STRLEN
123 210minimum_string_length (UV idx)
124//TODO 211{
212 return idx <= 23 ? 3
213 : idx <= 0xffU ? 4
214 : idx <= 0xffffU ? 5
215 : idx <= 0xffffffffU ? 7
216 : 11;
217}
125 218
126///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
127// encoder 220// encoder
128 221
129// structure used for encoding CBOR 222// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 225 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 226 char *end; // SvEND (sv)
134 SV *sv; // result scalar 227 SV *sv; // result scalar
135 CBOR cbor; 228 CBOR cbor;
136 U32 depth; // recursion level 229 U32 depth; // recursion level
230 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
231 UV stringref_idx;
232 HV *shareable; // ptr => index, or 0
233 UV shareable_idx;
137} enc_t; 234} enc_t;
138 235
139ecb_inline void 236ecb_inline void
140need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
141{ 238{
142 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
143 { 240 {
144 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
145 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
146 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
147 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
148 } 245 }
149} 246}
150 247
248static void encode_sv (enc_t *enc, SV *sv);
249
151ecb_inline void 250ecb_inline void
152encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
153{ 252{
154 need (enc, 1); 253 need (enc, 1);
155 *enc->cur++ = ch; 254 *enc->cur++ = ch;
156} 255}
157 256
257// used for tags, intregers, element counts and so on
158static void 258static void
159encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
160{ 260{
161 need (enc, 9); 261 need (enc, 9);
162 262
163 if (len < 24) 263 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
165 else if (len <= 0xff) 265 else if (ecb_expect_true (len <= 0xffU))
166 { 266 {
167 *enc->cur++ = major | 24; 267 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 268 *enc->cur++ = len;
169 } 269 }
170 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
171 { 271 {
172 *enc->cur++ = major | 25; 272 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 274 *enc->cur++ = len;
175 } 275 }
176 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
177 { 277 {
178 *enc->cur++ = major | 26; 278 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 282 *enc->cur++ = len;
183 } 283 }
184 else 284 else
185 { 285 {
186 *enc->cur++ = major | 27; 286 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 287 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 288 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 289 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 290 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 291 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 294 *enc->cur++ = len;
195 } 295 }
196} 296}
197 297
198static 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
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 319{
201 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);
202 need (enc, len); 350 need (enc, len);
203 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
204 enc->cur += len; 352 enc->cur += len;
205} 353}
206 354
207static void encode_sv (enc_t *enc, SV *sv); 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
381encode_float16 (enc_t *enc, NV nv)
382{
383 need (enc, 1+2);
384
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}
208 510
209static void 511static void
210encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
211{ 513{
212 int i, len = av_len (av); 514 int i, len = av_len (av);
214 if (enc->depth >= enc->cbor.max_depth) 516 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 517 croak (ERR_NESTING_EXCEEDED);
216 518
217 ++enc->depth; 519 ++enc->depth;
218 520
219 encode_uint (enc, 0x80, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 522
523 if (ecb_expect_false (SvMAGICAL (av)))
221 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
222 { 525 {
223 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 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 }
226 535
227 --enc->depth; 536 --enc->depth;
228} 537}
229 538
230static void 539static void
238 ++enc->depth; 547 ++enc->depth;
239 548
240 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
242 551
243 if (mg) 552 if (ecb_expect_false (mg))
244 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 554 else
246 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
247 556
248 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
249 { 558 {
250 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
252 else 561 else
253 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));
254 563
255 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));
256 } 565 }
257 566
258 if (mg) 567 if (ecb_expect_false (mg))
259 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 569
261 --enc->depth; 570 --enc->depth;
262} 571}
263 572
264// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
265static void 574static void
266encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
267{ 576{
268 svtype svt;
269
270 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
578
271 svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
272 580
273 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
274 { 582 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 583 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 584 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 590 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 592
285 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
286 GV *method;
287 594
288 if (stash == boolean_stash) 595 if (stash == boolean_stash)
289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
290 else if (stash == error_stash) 600 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
292 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
293 { 606 {
294 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 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))
299 } 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))));
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 662 {
302 dSP; 663 dSP;
303 664
304 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
305 // 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
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307 669
308 PUTBACK; 670 PUTBACK;
309 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
322 } 684 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 { 686 {
325 dSP; 687 dSP;
326 688
327 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
328 EXTEND (SP, 2); 691 EXTEND (SP, 2);
329 // 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
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
332 695
336 699
337 // catch this surprisingly common error 700 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 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));
340 703
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 707
345 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
346 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
347 716
348 PUTBACK; 717 PUTBACK;
349 718
350 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
351 } 720 }
355 } 724 }
356 else if (svt == SVt_PVHV) 725 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 726 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 727 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 728 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 729 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 730 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 731 encode_tag (enc, CBOR_TAG_INDIRECTION);
732 encode_sv (enc, sv);
733 }
380} 734}
381 735
382static void 736static void
383encode_nv (enc_t *enc, SV *sv) 737encode_nv (enc_t *enc, SV *sv)
384{ 738{
385 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
386 740
387 need (enc, 9); 741 need (enc, 9);
388 742
389 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 745 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
393 { 747 encode_float32 (enc, nv);
394 uint32_t fp = ecb_float_to_binary32 (nv);
395
396 *enc->cur++ = 0xe0 | 26;
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403 }
404 else 748 else
405 { 749 encode_float64 (enc, nv);
406 uint64_t fp = ecb_double_to_binary64 (nv);
407
408 *enc->cur++ = 0xe0 | 27;
409
410 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp);
412
413 memcpy (enc->cur, &fp, 8);
414 enc->cur += 8;
415 }
416} 750}
417 751
418static void 752static void
419encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
420{ 754{
422 756
423 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
424 { 758 {
425 STRLEN len; 759 STRLEN len;
426 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428 } 762 }
429 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 764 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
432 { 766 encode_int (enc, sv);
433 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv));
435 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv));
437 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
439 }
440 else if (SvROK (sv)) 767 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 773 else
447 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",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 776}
450 777
451static SV * 778static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
453{ 780{
454 enc_t enc; 781 enc_t enc = { 0 };
455 782
456 enc.cbor = *cbor; 783 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 787
462 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
463 encode_sv (&enc, scalar); 797 encode_sv (&enc, scalar);
464 798
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 799 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *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
467 801
481 U8 *end; // end of input string 815 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 816 const char *err; // parse error, if != 0
483 CBOR cbor; 817 CBOR cbor;
484 U32 depth; // recursion depth 818 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
820 AV *shareable;
821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
486} dec_t; 824} dec_t;
487 825
488#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);
489 833
490#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;
491 837
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
492#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)
493#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
494 864
495static UV 865static UV
496decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
497{ 867{
498 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
499 { 869 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 870
505 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
506 WANT (2); 875 WANT (1);
507 dec->cur += 2; 876 dec->cur += 1;
508 return dec->cur[-1]; 877 return dec->cur[-1];
509 878 }
510 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
511 WANT (3); 881 WANT (2);
512 dec->cur += 3; 882 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
515 885 }
516 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
517 WANT (5); 888 WANT (4);
518 dec->cur += 5; 889 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
523 894 }
524 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
525 WANT (9); 897 WANT (8);
526 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
527 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
531 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
535 913 }
536 default: 914 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 916
540fail: 917fail:
541 return 0; 918 return 0;
542} 919}
543 920
548{ 925{
549 AV *av = newAV (); 926 AV *av = newAV ();
550 927
551 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
552 929
553 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 931 {
555 ++dec->cur; 932 ++dec->cur;
556 933
557 for (;;) 934 for (;;)
558 { 935 {
559 WANT (1); 936 WANT (1);
560 937
561 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
562 { 939 {
563 ++dec->cur; 940 ++dec->cur;
564 break; 941 break;
565 } 942 }
566 943
567 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
568 } 945 }
569 } 946 }
570 else 947 else
571 { 948 {
572 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
573 950
951 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 952 av_fill (av, len - 1);
575 953
576 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
578 } 956 }
579 957
580 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
581 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
582 960
583fail: 961fail:
584 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
585 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
586 return &PL_sv_undef; 964 return &PL_sv_undef;
587} 965}
588 966
589static void 967static void
590decode_he (dec_t *dec, HV *hv) 968decode_he (dec_t *dec, HV *hv)
591{ 969{
592 // for speed reasons, we specialcase single-string 970 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 971 // byte or utf-8 strings as keys, but only when !stringref
594 972
595 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))
596 { 975 {
597 I32 len = decode_uint (dec); 976 STRLEN len = decode_uint (dec);
598 char *key = (char *)dec->cur; 977 char *key = (char *)dec->cur;
599 978
979 WANT (len);
600 dec->cur += len; 980 dec->cur += len;
601 981
602 hv_store (hv, key, len, decode_sv (dec), 0); 982 hv_store (hv, key, len, decode_sv (dec), 0);
983
984 return;
603 } 985 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 987 {
606 I32 len = decode_uint (dec); 988 STRLEN len = decode_uint (dec);
607 char *key = (char *)dec->cur; 989 char *key = (char *)dec->cur;
608 990
991 WANT (len);
609 dec->cur += len; 992 dec->cur += len;
610 993
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 998 hv_store (hv, key, -len, decode_sv (dec), 0);
999
1000 return;
612 } 1001 }
613 else 1002
614 {
615 SV *k = decode_sv (dec); 1003 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 1004 SV *v = decode_sv (dec);
617 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
618 hv_store_ent (hv, k, v, 0); 1034 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 1035 SvREFCNT_dec_NN (k);
620 } 1036
1037fail:
1038 ;
621} 1039}
622 1040
623static SV * 1041static SV *
624decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
625{ 1043{
626 HV *hv = newHV (); 1044 HV *hv = newHV ();
627 1045
628 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
629 1047
630 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 1049 {
632 ++dec->cur; 1050 ++dec->cur;
633 1051
634 for (;;) 1052 for (;;)
635 { 1053 {
636 WANT (1); 1054 WANT (1);
637 1055
638 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
639 { 1057 {
640 ++dec->cur; 1058 ++dec->cur;
641 break; 1059 break;
642 } 1060 }
643 1061
644 decode_he (dec, hv); 1062 decode_he (dec, hv);
645 } 1063 }
646 } 1064 }
647 else 1065 else
648 { 1066 {
649 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
650 1070
651 while (pairs--) 1071 while (pairs--)
652 decode_he (dec, hv); 1072 decode_he (dec, hv);
653 } 1073 }
654 1074
655 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
656 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
657 1077
658fail: 1078fail:
659 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
660 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
661 return &PL_sv_undef; 1081 return &PL_sv_undef;
662} 1082}
663 1083
664static SV * 1084static SV *
665decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
666{ 1086{
667 SV *sv = 0; 1087 SV *sv = 0;
668 1088
669 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
670 { 1090 {
1091 // indefinite length strings
671 ++dec->cur; 1092 ++dec->cur;
672 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
673 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
674 1097
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 1098 for (;;)
677 { 1099 {
678 WANT (1); 1100 WANT (1);
679 1101
680 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 1104 {
682 ++dec->cur; 1105 ++dec->cur;
683 break; 1106 break;
684 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 1110
686 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;
687 } 1116 }
688 } 1117 }
689 else 1118 else
690 { 1119 {
691 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
692 1121
693 WANT (len); 1122 WANT (len);
694 sv = newSVpvn (dec->cur, len); 1123 sv = newSVpvn (dec->cur, len);
695 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));
696 } 1129 }
697 1130
698 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
699 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
700 1139
701 return sv; 1140 return sv;
702 1141
703fail: 1142fail:
704 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
706} 1145}
707 1146
708static SV * 1147static SV *
709decode_tagged (dec_t *dec) 1148decode_tagged (dec_t *dec)
710{ 1149{
1150 SV *sv = 0;
711 UV tag = decode_uint (dec); 1151 UV tag = decode_uint (dec);
1152
1153 WANT (1);
1154
1155 switch (tag)
1156 {
1157 case CBOR_TAG_MAGIC:
712 SV *sv = decode_sv (dec); 1158 sv = decode_sv (dec);
1159 break;
713 1160
714 if (tag == CBOR_TAG_MAGIC) 1161 case CBOR_TAG_INDIRECTION:
715 return sv; 1162 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 1163 break;
717 { 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;
1193
1194 case CBOR_TAG_VALUE_SHAREABLE:
1195 {
1196 if (ecb_expect_false (!dec->shareable))
1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
1201 sv = newSV (0);
1202 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1203
1204 SV *osv = decode_sv (dec);
1205 sv_setsv (sv, osv);
1206 SvREFCNT_dec_NN (osv);
1207 }
1208 else
1209 {
1210 av_push (dec->shareable, &PL_sv_undef);
1211 int idx = AvFILLp (dec->shareable);
1212 sv = decode_sv (dec);
1213 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1214 }
1215 }
1216 break;
1217
1218 case CBOR_TAG_VALUE_SHAREDREF:
1219 {
1220 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1221 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1222
1223 UV idx = decode_uint (dec);
1224
1225 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1226 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1227
1228 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1229
1230 if (sv == &PL_sv_undef)
1231 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1232 }
1233 break;
1234
1235 case CBOR_TAG_PERL_OBJECT:
1236 {
1237 if (dec->cbor.flags & F_FORBID_OBJECTS)
1238 goto filter;
1239
1240 sv = decode_sv (dec);
1241
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1242 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)"); 1243 ERR ("corrupted CBOR data (non-array perl object)");
720 1244
721 AV *av = (AV *)SvRV (sv); 1245 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1; 1246 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1247 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724 1248
725 if (!stash) 1249 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)"); 1250 ERR ("cannot decode perl-object (package does not exist)");
727 1251
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1252 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 1253
730 if (!method) 1254 if (!method)
731 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1255 ERR ("cannot decode perl-object (package does not have a THAW method)");
732 1256
733 dSP; 1257 dSP;
734 1258
735 ENTER; SAVETMPS; PUSHMARK (SP); 1259 ENTER; SAVETMPS;
1260 PUSHMARK (SP);
736 EXTEND (SP, len + 1); 1261 EXTEND (SP, len + 1);
737 // we re-bless the reference to get overload and other niceties right 1262 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1)); 1263 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor); 1264 PUSHs (sv_cbor);
740 1265
741 int i; 1266 int i;
742 1267
743 for (i = 1; i < len; ++i) 1268 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1)); 1269 PUSHs (*av_fetch (av, i, 1));
745 1270
746 PUTBACK; 1271 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR); 1272 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN; 1273 SPAGAIN;
749 1274
1275 if (SvTRUE (ERRSV))
1276 {
1277 FREETMPS; LEAVE;
1278 ERR_ERRSV;
1279 }
1280
750 SvREFCNT_dec (sv); 1281 SvREFCNT_dec_NN (sv);
751 sv = SvREFCNT_inc (POPs); 1282 sv = SvREFCNT_inc (POPs);
752 1283
753 PUTBACK; 1284 PUTBACK;
754 1285
755 FREETMPS; LEAVE; 1286 FREETMPS; LEAVE;
1287 }
1288 break;
756 1289
757 return sv; 1290 default:
758 } 1291 filter:
759 else 1292 {
760 { 1293 SV *tag_sv = newSVuv (tag);
1294
1295 sv = decode_sv (dec);
1296
1297 dSP;
1298 ENTER; SAVETMPS;
1299 PUSHMARK (SP);
1300 EXTEND (SP, 2);
1301 PUSHs (tag_sv);
1302 PUSHs (sv);
1303
1304 PUTBACK;
1305 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1306 SPAGAIN;
1307
1308 if (SvTRUE (ERRSV))
1309 {
1310 SvREFCNT_dec_NN (tag_sv);
1311 FREETMPS; LEAVE;
1312 ERR_ERRSV;
1313 }
1314
1315 if (count)
1316 {
1317 SvREFCNT_dec_NN (tag_sv);
1318 SvREFCNT_dec_NN (sv);
1319 sv = SvREFCNT_inc_NN (TOPs);
1320 SP -= count;
1321 }
1322 else
1323 {
761 AV *av = newAV (); 1324 AV *av = newAV ();
762 av_push (av, newSVuv (tag)); 1325 av_push (av, tag_sv);
763 av_push (av, sv); 1326 av_push (av, sv);
764 1327
765 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1328 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
766 ? cbor_tagged_stash 1329 ? cbor_tagged_stash
767 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1330 : gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1331 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1332 }
1333
1334 PUTBACK;
1335
1336 FREETMPS; LEAVE;
1337 }
1338 break;
770 } 1339 }
1340
1341 return sv;
771 1342
772fail: 1343fail:
773 SvREFCNT_dec (sv); 1344 SvREFCNT_dec (sv);
774 return &PL_sv_undef; 1345 return &PL_sv_undef;
775} 1346}
777static SV * 1348static SV *
778decode_sv (dec_t *dec) 1349decode_sv (dec_t *dec)
779{ 1350{
780 WANT (1); 1351 WANT (1);
781 1352
782 switch (*dec->cur >> 5) 1353 switch (*dec->cur >> MAJOR_SHIFT)
783 { 1354 {
784 case 0: // unsigned int 1355 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
785 return newSVuv (decode_uint (dec)); 1356 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
786 case 1: // negative int 1357 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
787 return newSViv (-1 - (IV)decode_uint (dec)); 1358 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
788 case 2: // octet string 1359 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
789 return decode_str (dec, 0); 1360 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
790 case 3: // utf-8 string 1361 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
791 return decode_str (dec, 1); 1362
792 case 4: // array 1363 case MAJOR_MISC >> MAJOR_SHIFT:
793 return decode_av (dec);
794 case 5: // map
795 return decode_hv (dec);
796 case 6: // tag
797 return decode_tagged (dec);
798 case 7: // misc
799 switch (*dec->cur++ & 31) 1364 switch (*dec->cur++ & MINOR_MASK)
800 { 1365 {
801 case 20: 1366 case SIMPLE_FALSE:
802#if CBOR_SLOW 1367#if CBOR_SLOW
803 types_false = get_bool ("Types::Serialiser::false"); 1368 types_false = get_bool ("Types::Serialiser::false");
804#endif 1369#endif
805 return newSVsv (types_false); 1370 return newSVsv (types_false);
806 case 21: 1371 case SIMPLE_TRUE:
807#if CBOR_SLOW 1372#if CBOR_SLOW
808 types_true = get_bool ("Types::Serialiser::true"); 1373 types_true = get_bool ("Types::Serialiser::true");
809#endif 1374#endif
810 return newSVsv (types_true); 1375 return newSVsv (types_true);
811 case 22: 1376 case SIMPLE_NULL:
812 return newSVsv (&PL_sv_undef); 1377 return newSVsv (&PL_sv_undef);
813 case 23: 1378 case SIMPLE_UNDEF:
814#if CBOR_SLOW 1379#if CBOR_SLOW
815 types_error = get_bool ("Types::Serialiser::error"); 1380 types_error = get_bool ("Types::Serialiser::error");
816#endif 1381#endif
817 return newSVsv (types_error); 1382 return newSVsv (types_error);
818 1383
819 case 25: 1384 case MISC_FLOAT16:
820 { 1385 {
821 WANT (2); 1386 WANT (2);
822 1387
823 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1388 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
824 dec->cur += 2; 1389 dec->cur += 2;
825 1390
826 return newSVnv (ecb_binary16_to_float (fp)); 1391 return newSVnv (ecb_binary16_to_float (fp));
827 } 1392 }
828 1393
829 case 26: 1394 case MISC_FLOAT32:
830 { 1395 {
831 uint32_t fp; 1396 uint32_t fp;
832 WANT (4); 1397 WANT (4);
833 memcpy (&fp, dec->cur, 4); 1398 memcpy (&fp, dec->cur, 4);
834 dec->cur += 4; 1399 dec->cur += 4;
837 fp = ecb_bswap32 (fp); 1402 fp = ecb_bswap32 (fp);
838 1403
839 return newSVnv (ecb_binary32_to_float (fp)); 1404 return newSVnv (ecb_binary32_to_float (fp));
840 } 1405 }
841 1406
842 case 27: 1407 case MISC_FLOAT64:
843 { 1408 {
844 uint64_t fp; 1409 uint64_t fp;
845 WANT (8); 1410 WANT (8);
846 memcpy (&fp, dec->cur, 8); 1411 memcpy (&fp, dec->cur, 8);
847 dec->cur += 8; 1412 dec->cur += 8;
850 fp = ecb_bswap64 (fp); 1415 fp = ecb_bswap64 (fp);
851 1416
852 return newSVnv (ecb_binary64_to_double (fp)); 1417 return newSVnv (ecb_binary64_to_double (fp));
853 } 1418 }
854 1419
855 // 0..19 unassigned 1420 // 0..19 unassigned simple
856 // 24 reserved + unassigned (reserved values are not encodable) 1421 // 24 reserved + unassigned simple (reserved values are not encodable)
1422 // 28-30 unassigned misc
1423 // 31 break code
857 default: 1424 default:
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1425 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
859 } 1426 }
860 1427
861 break; 1428 break;
862 } 1429 }
863 1430
866} 1433}
867 1434
868static SV * 1435static SV *
869decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1436decode_cbor (SV *string, CBOR *cbor, char **offset_return)
870{ 1437{
871 dec_t dec; 1438 dec_t dec = { 0 };
872 SV *sv; 1439 SV *sv;
873 STRLEN len; 1440 STRLEN len;
874 char *data = SvPVbyte (string, len); 1441 char *data = SvPVbyte (string, len);
875 1442
876 if (len > cbor->max_size && cbor->max_size) 1443 if (len > cbor->max_size && cbor->max_size)
878 (unsigned long)len, (unsigned long)cbor->max_size); 1445 (unsigned long)len, (unsigned long)cbor->max_size);
879 1446
880 dec.cbor = *cbor; 1447 dec.cbor = *cbor;
881 dec.cur = (U8 *)data; 1448 dec.cur = (U8 *)data;
882 dec.end = (U8 *)data + len; 1449 dec.end = (U8 *)data + len;
883 dec.err = 0;
884 dec.depth = 0;
885 1450
886 sv = decode_sv (&dec); 1451 sv = decode_sv (&dec);
887 1452
888 if (offset_return) 1453 if (offset_return)
889 *offset_return = dec.cur; 1454 *offset_return = dec.cur;
892 if (dec.cur != dec.end && !dec.err) 1457 if (dec.cur != dec.end && !dec.err)
893 dec.err = "garbage after CBOR object"; 1458 dec.err = "garbage after CBOR object";
894 1459
895 if (dec.err) 1460 if (dec.err)
896 { 1461 {
1462 if (dec.shareable)
1463 {
1464 // need to break cyclic links, which would all be in shareable
1465 int i;
1466 SV **svp;
1467
1468 for (i = av_len (dec.shareable) + 1; i--; )
1469 if ((svp = av_fetch (dec.shareable, i, 0)))
1470 sv_setsv (*svp, &PL_sv_undef);
1471 }
1472
897 SvREFCNT_dec (sv); 1473 SvREFCNT_dec_NN (sv);
1474
1475 if (dec.err_sv)
1476 sv_2mortal (dec.err_sv);
1477
898 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1478 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
899 } 1479 }
900 1480
901 sv = sv_2mortal (sv); 1481 sv = sv_2mortal (sv);
902 1482
903 return sv; 1483 return sv;
904} 1484}
905 1485
1486/////////////////////////////////////////////////////////////////////////////
1487// incremental parser
1488
1489#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1490
1491// returns 0 for notyet, 1 for success or error
1492static int
1493incr_parse (CBOR *self, SV *cborstr)
1494{
1495 STRLEN cur;
1496 SvPV (cborstr, cur);
1497
1498 while (ecb_expect_true (self->incr_need <= cur))
1499 {
1500 // table of integer count bytes
1501 static I8 incr_len[MINOR_MASK + 1] = {
1502 0, 0, 0, 0, 0, 0, 0, 0,
1503 0, 0, 0, 0, 0, 0, 0, 0,
1504 0, 0, 0, 0, 0, 0, 0, 0,
1505 1, 2, 4, 8,-1,-1,-1,-2
1506 };
1507
1508 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1509 U8 m = *p & MINOR_MASK;
1510 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1511 I8 ilen = incr_len[m];
1512
1513 self->incr_need = self->incr_pos + 1;
1514
1515 if (ecb_expect_false (ilen < 0))
1516 {
1517 if (m != MINOR_INDEF)
1518 return 1; // error
1519
1520 if (*p == (MAJOR_MISC | MINOR_INDEF))
1521 {
1522 if (count >= 0)
1523 return 1; // error
1524
1525 count = 1;
1526 }
1527 else
1528 {
1529 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1530 count = -1;
1531 }
1532 }
1533 else
1534 {
1535 self->incr_need += ilen;
1536 if (ecb_expect_false (self->incr_need > cur))
1537 return 0;
1538
1539 int major = *p >> MAJOR_SHIFT;
1540
1541 switch (major)
1542 {
1543 case MAJOR_TAG >> MAJOR_SHIFT:
1544 ++count; // tags merely prefix another value
1545 break;
1546
1547 case MAJOR_BYTES >> MAJOR_SHIFT:
1548 case MAJOR_TEXT >> MAJOR_SHIFT:
1549 case MAJOR_ARRAY >> MAJOR_SHIFT:
1550 case MAJOR_MAP >> MAJOR_SHIFT:
1551 {
1552 UV len;
1553
1554 if (ecb_expect_false (ilen))
1555 {
1556 len = 0;
1557
1558 do {
1559 len = (len << 8) | *++p;
1560 } while (--ilen);
1561 }
1562 else
1563 len = m;
1564
1565 switch (major)
1566 {
1567 case MAJOR_BYTES >> MAJOR_SHIFT:
1568 case MAJOR_TEXT >> MAJOR_SHIFT:
1569 self->incr_need += len;
1570 if (ecb_expect_false (self->incr_need > cur))
1571 return 0;
1572
1573 break;
1574
1575 case MAJOR_MAP >> MAJOR_SHIFT:
1576 len <<= 1;
1577 /* FALLTHROUGH */
1578 case MAJOR_ARRAY >> MAJOR_SHIFT:
1579 if (len)
1580 {
1581 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1582 count = len + 1;
1583 }
1584 break;
1585 }
1586 }
1587 }
1588 }
1589
1590 self->incr_pos = self->incr_need;
1591
1592 if (count > 0)
1593 {
1594 while (!--count)
1595 {
1596 if (!AvFILLp (self->incr_count))
1597 return 1; // done
1598
1599 SvREFCNT_dec_NN (av_pop (self->incr_count));
1600 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1601 }
1602
1603 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1604 }
1605 }
1606
1607 return 0;
1608}
1609
1610
906///////////////////////////////////////////////////////////////////////////// 1611/////////////////////////////////////////////////////////////////////////////
907// XS interface functions 1612// XS interface functions
908 1613
909MODULE = CBOR::XS PACKAGE = CBOR::XS 1614MODULE = CBOR::XS PACKAGE = CBOR::XS
910 1615
918 1623
919 types_true = get_bool ("Types::Serialiser::true" ); 1624 types_true = get_bool ("Types::Serialiser::true" );
920 types_false = get_bool ("Types::Serialiser::false"); 1625 types_false = get_bool ("Types::Serialiser::false");
921 types_error = get_bool ("Types::Serialiser::error"); 1626 types_error = get_bool ("Types::Serialiser::error");
922 1627
1628 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1629
923 sv_cbor = newSVpv ("CBOR", 0); 1630 sv_cbor = newSVpv ("CBOR", 0);
924 SvREADONLY_on (sv_cbor); 1631 SvREADONLY_on (sv_cbor);
1632
1633 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
925} 1634}
926 1635
927PROTOTYPES: DISABLE 1636PROTOTYPES: DISABLE
928 1637
929void CLONE (...) 1638void CLONE (...)
947 1656
948void shrink (CBOR *self, int enable = 1) 1657void shrink (CBOR *self, int enable = 1)
949 ALIAS: 1658 ALIAS:
950 shrink = F_SHRINK 1659 shrink = F_SHRINK
951 allow_unknown = F_ALLOW_UNKNOWN 1660 allow_unknown = F_ALLOW_UNKNOWN
1661 allow_sharing = F_ALLOW_SHARING
1662 allow_cycles = F_ALLOW_CYCLES
1663 forbid_objects = F_FORBID_OBJECTS
1664 pack_strings = F_PACK_STRINGS
1665 text_keys = F_TEXT_KEYS
1666 text_strings = F_TEXT_STRINGS
1667 validate_utf8 = F_VALIDATE_UTF8
952 PPCODE: 1668 PPCODE:
953{ 1669{
954 if (enable) 1670 if (enable)
955 self->flags |= ix; 1671 self->flags |= ix;
956 else 1672 else
961 1677
962void get_shrink (CBOR *self) 1678void get_shrink (CBOR *self)
963 ALIAS: 1679 ALIAS:
964 get_shrink = F_SHRINK 1680 get_shrink = F_SHRINK
965 get_allow_unknown = F_ALLOW_UNKNOWN 1681 get_allow_unknown = F_ALLOW_UNKNOWN
1682 get_allow_sharing = F_ALLOW_SHARING
1683 get_allow_cycles = F_ALLOW_CYCLES
1684 get_forbid_objects = F_FORBID_OBJECTS
1685 get_pack_strings = F_PACK_STRINGS
1686 get_text_keys = F_TEXT_KEYS
1687 get_text_strings = F_TEXT_STRINGS
1688 get_validate_utf8 = F_VALIDATE_UTF8
966 PPCODE: 1689 PPCODE:
967 XPUSHs (boolSV (self->flags & ix)); 1690 XPUSHs (boolSV (self->flags & ix));
968 1691
969void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1692void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
970 PPCODE: 1693 PPCODE:
983 XPUSHs (ST (0)); 1706 XPUSHs (ST (0));
984 1707
985int get_max_size (CBOR *self) 1708int get_max_size (CBOR *self)
986 CODE: 1709 CODE:
987 RETVAL = self->max_size; 1710 RETVAL = self->max_size;
1711 OUTPUT:
1712 RETVAL
1713
1714void filter (CBOR *self, SV *filter = 0)
1715 PPCODE:
1716 SvREFCNT_dec (self->filter);
1717 self->filter = filter ? newSVsv (filter) : filter;
1718 XPUSHs (ST (0));
1719
1720SV *get_filter (CBOR *self)
1721 CODE:
1722 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
988 OUTPUT: 1723 OUTPUT:
989 RETVAL 1724 RETVAL
990 1725
991void encode (CBOR *self, SV *scalar) 1726void encode (CBOR *self, SV *scalar)
992 PPCODE: 1727 PPCODE:
1007 EXTEND (SP, 2); 1742 EXTEND (SP, 2);
1008 PUSHs (sv); 1743 PUSHs (sv);
1009 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1744 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1010} 1745}
1011 1746
1747void incr_parse (CBOR *self, SV *cborstr)
1748 ALIAS:
1749 incr_parse_multiple = 1
1750 PPCODE:
1751{
1752 if (SvUTF8 (cborstr))
1753 sv_utf8_downgrade (cborstr, 0);
1754
1755 if (!self->incr_count)
1756 {
1757 self->incr_count = newAV ();
1758 self->incr_pos = 0;
1759 self->incr_need = 1;
1760
1761 av_push (self->incr_count, newSViv (1));
1762 }
1763
1764 do
1765 {
1766 if (!incr_parse (self, cborstr))
1767 {
1768 if (self->incr_need > self->max_size && self->max_size)
1769 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1770 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1771
1772 break;
1773 }
1774
1775 SV *sv;
1776 char *offset;
1777
1778 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1779 XPUSHs (sv);
1780
1781 sv_chop (cborstr, offset);
1782
1783 av_clear (self->incr_count);
1784 av_push (self->incr_count, newSViv (1));
1785
1786 self->incr_pos = 0;
1787 self->incr_need = self->incr_pos + 1;
1788 }
1789 while (ix);
1790}
1791
1792void incr_reset (CBOR *self)
1793 CODE:
1794{
1795 SvREFCNT_dec (self->incr_count);
1796 self->incr_count = 0;
1797}
1798
1799void DESTROY (CBOR *self)
1800 PPCODE:
1801 cbor_free (self);
1802
1012PROTOTYPES: ENABLE 1803PROTOTYPES: ENABLE
1013 1804
1014void encode_cbor (SV *scalar) 1805void encode_cbor (SV *scalar)
1806 ALIAS:
1807 encode_cbor = 0
1808 encode_cbor_sharing = F_ALLOW_SHARING
1015 PPCODE: 1809 PPCODE:
1016{ 1810{
1017 CBOR cbor; 1811 CBOR cbor;
1018 cbor_init (&cbor); 1812 cbor_init (&cbor);
1813 cbor.flags |= ix;
1019 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1814 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1020 XPUSHs (scalar); 1815 XPUSHs (scalar);
1021} 1816}
1022 1817
1023void decode_cbor (SV *cborstr) 1818void decode_cbor (SV *cborstr)
1027 cbor_init (&cbor); 1822 cbor_init (&cbor);
1028 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1823 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1029 XPUSHs (cborstr); 1824 XPUSHs (cborstr);
1030} 1825}
1031 1826
1827#ifdef __AFL_COMPILER
1828
1829void
1830afl_init ()
1831 CODE:
1832 __AFL_INIT ();
1833
1834int
1835afl_loop (unsigned int count = 10000)
1836 CODE:
1837 RETVAL = __AFL_LOOP (count);
1838 OUTPUT:
1839 RETVAL
1840
1841#endif
1842

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines