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 1.17 by root, Tue Oct 29 22:04:52 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
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;
990 printf ("len %d\n", len);//D
608 991
992 WANT (len);
609 dec->cur += len; 993 dec->cur += len;
610 994
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 999 hv_store (hv, key, -len, decode_sv (dec), 0);
1000
1001 return;
612 } 1002 }
613 else 1003
614 {
615 SV *k = decode_sv (dec); 1004 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 1005 SV *v = decode_sv (dec);
617 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
618 hv_store_ent (hv, k, v, 0); 1035 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 1036 SvREFCNT_dec_NN (k);
620 } 1037
1038fail:
1039 ;
621} 1040}
622 1041
623static SV * 1042static SV *
624decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
625{ 1044{
626 HV *hv = newHV (); 1045 HV *hv = newHV ();
627 1046
628 DEC_INC_DEPTH; 1047 DEC_INC_DEPTH;
629 1048
630 if ((*dec->cur & 31) == 31) 1049 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 1050 {
632 ++dec->cur; 1051 ++dec->cur;
633 1052
634 for (;;) 1053 for (;;)
635 { 1054 {
636 WANT (1); 1055 WANT (1);
637 1056
638 if (*dec->cur == (0xe0 | 31)) 1057 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
639 { 1058 {
640 ++dec->cur; 1059 ++dec->cur;
641 break; 1060 break;
642 } 1061 }
643 1062
644 decode_he (dec, hv); 1063 decode_he (dec, hv);
645 } 1064 }
646 } 1065 }
647 else 1066 else
648 { 1067 {
649 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
650 1071
651 while (pairs--) 1072 while (pairs--)
652 decode_he (dec, hv); 1073 decode_he (dec, hv);
653 } 1074 }
654 1075
655 DEC_DEC_DEPTH; 1076 DEC_DEC_DEPTH;
656 return newRV_noinc ((SV *)hv); 1077 return newRV_noinc ((SV *)hv);
657 1078
658fail: 1079fail:
659 SvREFCNT_dec (hv); 1080 SvREFCNT_dec_NN (hv);
660 DEC_DEC_DEPTH; 1081 DEC_DEC_DEPTH;
661 return &PL_sv_undef; 1082 return &PL_sv_undef;
662} 1083}
663 1084
664static SV * 1085static SV *
665decode_str (dec_t *dec, int utf8) 1086decode_str (dec_t *dec, int utf8)
666{ 1087{
667 SV *sv = 0; 1088 SV *sv = 0;
668 1089
669 if ((*dec->cur & 31) == 31) 1090 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
670 { 1091 {
1092 // indefinite length strings
671 ++dec->cur; 1093 ++dec->cur;
672 1094
1095 U8 major = *dec->cur & MAJOR_MISC;
1096
673 sv = newSVpvn ("", 0); 1097 sv = newSVpvn ("", 0);
674 1098
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 1099 for (;;)
677 { 1100 {
678 WANT (1); 1101 WANT (1);
679 1102
680 if (*dec->cur == (0xe0 | 31)) 1103 if ((*dec->cur - major) > LENGTH_EXT8)
1104 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 1105 {
682 ++dec->cur; 1106 ++dec->cur;
683 break; 1107 break;
684 } 1108 }
1109 else
1110 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 1111
686 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;
687 } 1117 }
688 } 1118 }
689 else 1119 else
690 { 1120 {
691 STRLEN len = decode_uint (dec); 1121 STRLEN len = decode_uint (dec);
692 1122
693 WANT (len); 1123 WANT (len);
694 sv = newSVpvn (dec->cur, len); 1124 sv = newSVpvn (dec->cur, len);
695 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));
696 } 1130 }
697 1131
698 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
699 SvUTF8_on (sv); 1138 SvUTF8_on (sv);
1139 }
700 1140
701 return sv; 1141 return sv;
702 1142
703fail: 1143fail:
704 SvREFCNT_dec (sv); 1144 SvREFCNT_dec (sv);
706} 1146}
707 1147
708static SV * 1148static SV *
709decode_tagged (dec_t *dec) 1149decode_tagged (dec_t *dec)
710{ 1150{
1151 SV *sv = 0;
711 UV tag = decode_uint (dec); 1152 UV tag = decode_uint (dec);
1153
1154 WANT (1);
1155
1156 switch (tag)
1157 {
1158 case CBOR_TAG_MAGIC:
712 SV *sv = decode_sv (dec); 1159 sv = decode_sv (dec);
1160 break;
713 1161
714 if (tag == CBOR_TAG_MAGIC) 1162 case CBOR_TAG_INDIRECTION:
715 return sv; 1163 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 1164 break;
717 { 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;
1194
1195 case CBOR_TAG_VALUE_SHAREABLE:
1196 {
1197 if (ecb_expect_false (!dec->shareable))
1198 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1199
1200 if (dec->cbor.flags & F_ALLOW_CYCLES)
1201 {
1202 sv = newSV (0);
1203 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1204
1205 SV *osv = decode_sv (dec);
1206 sv_setsv (sv, osv);
1207 SvREFCNT_dec_NN (osv);
1208 }
1209 else
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;
1218
1219 case CBOR_TAG_VALUE_SHAREDREF:
1220 {
1221 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1222 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1223
1224 UV idx = decode_uint (dec);
1225
1226 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1227 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1228
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");
1233 }
1234 break;
1235
1236 case CBOR_TAG_PERL_OBJECT:
1237 {
1238 if (dec->cbor.flags & F_FORBID_OBJECTS)
1239 goto filter;
1240
1241 sv = decode_sv (dec);
1242
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1243 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)"); 1244 ERR ("corrupted CBOR data (non-array perl object)");
720 1245
721 AV *av = (AV *)SvRV (sv); 1246 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1; 1247 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1248 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724 1249
725 if (!stash) 1250 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)"); 1251 ERR ("cannot decode perl-object (package does not exist)");
727 1252
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1253 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 1254
730 if (!method) 1255 if (!method)
731 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)");
732 1257
733 dSP; 1258 dSP;
734 1259
735 ENTER; SAVETMPS; PUSHMARK (SP); 1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
736 EXTEND (SP, len + 1); 1262 EXTEND (SP, len + 1);
737 // 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
738 PUSHs (*av_fetch (av, 0, 1)); 1264 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor); 1265 PUSHs (sv_cbor);
740 1266
741 int i; 1267 int i;
742 1268
743 for (i = 1; i < len; ++i) 1269 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1)); 1270 PUSHs (*av_fetch (av, i, 1));
745 1271
746 PUTBACK; 1272 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); 1273 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN; 1274 SPAGAIN;
749 1275
750 if (SvTRUE (ERRSV)) 1276 if (SvTRUE (ERRSV))
751 { 1277 {
1278 FREETMPS; LEAVE;
1279 ERR_ERRSV;
1280 }
1281
1282 SvREFCNT_dec_NN (sv);
1283 sv = SvREFCNT_inc (POPs);
1284
1285 PUTBACK;
1286
752 FREETMPS; LEAVE; 1287 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1288 }
1289 break;
1290
1291 default:
1292 filter:
754 } 1293 {
1294 SV *tag_sv = newSVuv (tag);
755 1295
756 SvREFCNT_dec (sv); 1296 sv = decode_sv (dec);
757 sv = SvREFCNT_inc (POPs);
758 1297
1298 dSP;
1299 ENTER; SAVETMPS;
1300 PUSHMARK (SP);
1301 EXTEND (SP, 2);
1302 PUSHs (tag_sv);
1303 PUSHs (sv);
1304
759 PUTBACK; 1305 PUTBACK;
1306 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1307 SPAGAIN;
760 1308
1309 if (SvTRUE (ERRSV))
1310 {
1311 SvREFCNT_dec_NN (tag_sv);
761 FREETMPS; LEAVE; 1312 FREETMPS; LEAVE;
1313 ERR_ERRSV;
1314 }
762 1315
763 return sv; 1316 if (count)
764 } 1317 {
765 else 1318 SvREFCNT_dec_NN (tag_sv);
766 { 1319 SvREFCNT_dec_NN (sv);
1320 sv = SvREFCNT_inc_NN (TOPs);
1321 SP -= count;
1322 }
1323 else
1324 {
767 AV *av = newAV (); 1325 AV *av = newAV ();
768 av_push (av, newSVuv (tag)); 1326 av_push (av, tag_sv);
769 av_push (av, sv); 1327 av_push (av, sv);
770 1328
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1329 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash 1330 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1331 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774
775 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1332 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1333 }
1334
1335 PUTBACK;
1336
1337 FREETMPS; LEAVE;
1338 }
1339 break;
776 } 1340 }
1341
1342 return sv;
777 1343
778fail: 1344fail:
779 SvREFCNT_dec (sv); 1345 SvREFCNT_dec (sv);
780 return &PL_sv_undef; 1346 return &PL_sv_undef;
781} 1347}
783static SV * 1349static SV *
784decode_sv (dec_t *dec) 1350decode_sv (dec_t *dec)
785{ 1351{
786 WANT (1); 1352 WANT (1);
787 1353
788 switch (*dec->cur >> 5) 1354 switch (*dec->cur >> MAJOR_SHIFT)
789 { 1355 {
790 case 0: // unsigned int 1356 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
791 return newSVuv (decode_uint (dec)); 1357 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
792 case 1: // negative int 1358 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
793 return newSViv (-1 - (IV)decode_uint (dec)); 1359 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
794 case 2: // octet string 1360 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
795 return decode_str (dec, 0); 1361 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
796 case 3: // utf-8 string 1362 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
797 return decode_str (dec, 1); 1363
798 case 4: // array 1364 case MAJOR_MISC >> MAJOR_SHIFT:
799 return decode_av (dec);
800 case 5: // map
801 return decode_hv (dec);
802 case 6: // tag
803 return decode_tagged (dec);
804 case 7: // misc
805 switch (*dec->cur++ & 31) 1365 switch (*dec->cur++ & MINOR_MASK)
806 { 1366 {
807 case 20: 1367 case SIMPLE_FALSE:
808#if CBOR_SLOW 1368#if CBOR_SLOW
809 types_false = get_bool ("Types::Serialiser::false"); 1369 types_false = get_bool ("Types::Serialiser::false");
810#endif 1370#endif
811 return newSVsv (types_false); 1371 return newSVsv (types_false);
812 case 21: 1372 case SIMPLE_TRUE:
813#if CBOR_SLOW 1373#if CBOR_SLOW
814 types_true = get_bool ("Types::Serialiser::true"); 1374 types_true = get_bool ("Types::Serialiser::true");
815#endif 1375#endif
816 return newSVsv (types_true); 1376 return newSVsv (types_true);
817 case 22: 1377 case SIMPLE_NULL:
818 return newSVsv (&PL_sv_undef); 1378 return newSVsv (&PL_sv_undef);
819 case 23: 1379 case SIMPLE_UNDEF:
820#if CBOR_SLOW 1380#if CBOR_SLOW
821 types_error = get_bool ("Types::Serialiser::error"); 1381 types_error = get_bool ("Types::Serialiser::error");
822#endif 1382#endif
823 return newSVsv (types_error); 1383 return newSVsv (types_error);
824 1384
825 case 25: 1385 case MISC_FLOAT16:
826 { 1386 {
827 WANT (2); 1387 WANT (2);
828 1388
829 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1389 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
830 dec->cur += 2; 1390 dec->cur += 2;
831 1391
832 return newSVnv (ecb_binary16_to_float (fp)); 1392 return newSVnv (ecb_binary16_to_float (fp));
833 } 1393 }
834 1394
835 case 26: 1395 case MISC_FLOAT32:
836 { 1396 {
837 uint32_t fp; 1397 uint32_t fp;
838 WANT (4); 1398 WANT (4);
839 memcpy (&fp, dec->cur, 4); 1399 memcpy (&fp, dec->cur, 4);
840 dec->cur += 4; 1400 dec->cur += 4;
843 fp = ecb_bswap32 (fp); 1403 fp = ecb_bswap32 (fp);
844 1404
845 return newSVnv (ecb_binary32_to_float (fp)); 1405 return newSVnv (ecb_binary32_to_float (fp));
846 } 1406 }
847 1407
848 case 27: 1408 case MISC_FLOAT64:
849 { 1409 {
850 uint64_t fp; 1410 uint64_t fp;
851 WANT (8); 1411 WANT (8);
852 memcpy (&fp, dec->cur, 8); 1412 memcpy (&fp, dec->cur, 8);
853 dec->cur += 8; 1413 dec->cur += 8;
856 fp = ecb_bswap64 (fp); 1416 fp = ecb_bswap64 (fp);
857 1417
858 return newSVnv (ecb_binary64_to_double (fp)); 1418 return newSVnv (ecb_binary64_to_double (fp));
859 } 1419 }
860 1420
861 // 0..19 unassigned 1421 // 0..19 unassigned simple
862 // 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
863 default: 1425 default:
864 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1426 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
865 } 1427 }
866 1428
867 break; 1429 break;
868 } 1430 }
869 1431
872} 1434}
873 1435
874static SV * 1436static SV *
875decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1437decode_cbor (SV *string, CBOR *cbor, char **offset_return)
876{ 1438{
877 dec_t dec; 1439 dec_t dec = { 0 };
878 SV *sv; 1440 SV *sv;
879 STRLEN len; 1441 STRLEN len;
880 char *data = SvPVbyte (string, len); 1442 char *data = SvPVbyte (string, len);
881 1443
882 if (len > cbor->max_size && cbor->max_size) 1444 if (len > cbor->max_size && cbor->max_size)
884 (unsigned long)len, (unsigned long)cbor->max_size); 1446 (unsigned long)len, (unsigned long)cbor->max_size);
885 1447
886 dec.cbor = *cbor; 1448 dec.cbor = *cbor;
887 dec.cur = (U8 *)data; 1449 dec.cur = (U8 *)data;
888 dec.end = (U8 *)data + len; 1450 dec.end = (U8 *)data + len;
889 dec.err = 0;
890 dec.depth = 0;
891 1451
892 sv = decode_sv (&dec); 1452 sv = decode_sv (&dec);
893 1453
894 if (offset_return) 1454 if (offset_return)
895 *offset_return = dec.cur; 1455 *offset_return = dec.cur;
898 if (dec.cur != dec.end && !dec.err) 1458 if (dec.cur != dec.end && !dec.err)
899 dec.err = "garbage after CBOR object"; 1459 dec.err = "garbage after CBOR object";
900 1460
901 if (dec.err) 1461 if (dec.err)
902 { 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
903 SvREFCNT_dec (sv); 1474 SvREFCNT_dec_NN (sv);
1475
1476 if (dec.err_sv)
1477 sv_2mortal (dec.err_sv);
1478
904 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);
905 } 1480 }
906 1481
907 sv = sv_2mortal (sv); 1482 sv = sv_2mortal (sv);
908 1483
909 return sv; 1484 return sv;
910} 1485}
911 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
912///////////////////////////////////////////////////////////////////////////// 1612/////////////////////////////////////////////////////////////////////////////
913// XS interface functions 1613// XS interface functions
914 1614
915MODULE = CBOR::XS PACKAGE = CBOR::XS 1615MODULE = CBOR::XS PACKAGE = CBOR::XS
916 1616
924 1624
925 types_true = get_bool ("Types::Serialiser::true" ); 1625 types_true = get_bool ("Types::Serialiser::true" );
926 types_false = get_bool ("Types::Serialiser::false"); 1626 types_false = get_bool ("Types::Serialiser::false");
927 types_error = get_bool ("Types::Serialiser::error"); 1627 types_error = get_bool ("Types::Serialiser::error");
928 1628
1629 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1630
929 sv_cbor = newSVpv ("CBOR", 0); 1631 sv_cbor = newSVpv ("CBOR", 0);
930 SvREADONLY_on (sv_cbor); 1632 SvREADONLY_on (sv_cbor);
1633
1634 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
931} 1635}
932 1636
933PROTOTYPES: DISABLE 1637PROTOTYPES: DISABLE
934 1638
935void CLONE (...) 1639void CLONE (...)
953 1657
954void shrink (CBOR *self, int enable = 1) 1658void shrink (CBOR *self, int enable = 1)
955 ALIAS: 1659 ALIAS:
956 shrink = F_SHRINK 1660 shrink = F_SHRINK
957 allow_unknown = F_ALLOW_UNKNOWN 1661 allow_unknown = F_ALLOW_UNKNOWN
1662 allow_sharing = F_ALLOW_SHARING
1663 allow_cycles = F_ALLOW_CYCLES
1664 forbid_objects = F_FORBID_OBJECTS
1665 pack_strings = F_PACK_STRINGS
1666 text_keys = F_TEXT_KEYS
1667 text_strings = F_TEXT_STRINGS
1668 validate_utf8 = F_VALIDATE_UTF8
958 PPCODE: 1669 PPCODE:
959{ 1670{
960 if (enable) 1671 if (enable)
961 self->flags |= ix; 1672 self->flags |= ix;
962 else 1673 else
967 1678
968void get_shrink (CBOR *self) 1679void get_shrink (CBOR *self)
969 ALIAS: 1680 ALIAS:
970 get_shrink = F_SHRINK 1681 get_shrink = F_SHRINK
971 get_allow_unknown = F_ALLOW_UNKNOWN 1682 get_allow_unknown = F_ALLOW_UNKNOWN
1683 get_allow_sharing = F_ALLOW_SHARING
1684 get_allow_cycles = F_ALLOW_CYCLES
1685 get_forbid_objects = F_FORBID_OBJECTS
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
972 PPCODE: 1690 PPCODE:
973 XPUSHs (boolSV (self->flags & ix)); 1691 XPUSHs (boolSV (self->flags & ix));
974 1692
975void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1693void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
976 PPCODE: 1694 PPCODE:
989 XPUSHs (ST (0)); 1707 XPUSHs (ST (0));
990 1708
991int get_max_size (CBOR *self) 1709int get_max_size (CBOR *self)
992 CODE: 1710 CODE:
993 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);
994 OUTPUT: 1724 OUTPUT:
995 RETVAL 1725 RETVAL
996 1726
997void encode (CBOR *self, SV *scalar) 1727void encode (CBOR *self, SV *scalar)
998 PPCODE: 1728 PPCODE:
1013 EXTEND (SP, 2); 1743 EXTEND (SP, 2);
1014 PUSHs (sv); 1744 PUSHs (sv);
1015 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1745 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1016} 1746}
1017 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
1018PROTOTYPES: ENABLE 1804PROTOTYPES: ENABLE
1019 1805
1020void encode_cbor (SV *scalar) 1806void encode_cbor (SV *scalar)
1807 ALIAS:
1808 encode_cbor = 0
1809 encode_cbor_sharing = F_ALLOW_SHARING
1021 PPCODE: 1810 PPCODE:
1022{ 1811{
1023 CBOR cbor; 1812 CBOR cbor;
1024 cbor_init (&cbor); 1813 cbor_init (&cbor);
1814 cbor.flags |= ix;
1025 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1815 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1026 XPUSHs (scalar); 1816 XPUSHs (scalar);
1027} 1817}
1028 1818
1029void decode_cbor (SV *cborstr) 1819void decode_cbor (SV *cborstr)
1033 cbor_init (&cbor); 1823 cbor_init (&cbor);
1034 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1824 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1035 XPUSHs (cborstr); 1825 XPUSHs (cborstr);
1036} 1826}
1037 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