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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.67 by root, Sun Nov 29 21:35:06 2020 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"
15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
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// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
13 69
14// known tags 70// known tags
15enum cbor_tag 71enum cbor_tag
16{ 72{
17 // inofficial extensions (pending iana registration) 73 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 81
21 // rfc7049 82 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
28 89
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 94
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 100
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 102};
42 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_BYTES = 1,
109 AS_TEXT = 2,
110 AS_INT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 // possibly future enhancements: (generic) float, (generic) string
115};
116
43#define F_SHRINK 0x00000200UL 117#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
119#define F_ALLOW_SHARING 0x00000004UL
120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
125#define F_VALIDATE_UTF8 0x00000100UL
45 126
46#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
47 128
48#define SB do { 129#define SB do {
49#define SE } while (0) 130#define SE } while (0)
61# define CBOR_SLOW 0 142# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 143# define CBOR_STASH cbor_stash
63#endif 144#endif
64 145
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 146static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 147static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 148
68typedef struct { 149typedef struct {
69 U32 flags; 150 U32 flags;
70 U32 max_depth; 151 U32 max_depth;
71 STRLEN max_size; 152 STRLEN max_size;
153 SV *filter;
154
155 // for the incremental parser
156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 159} CBOR;
73 160
74ecb_inline void 161ecb_inline void
75cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
76{ 163{
77 Zero (cbor, 1, CBOR); 164 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 165 cbor->max_depth = 512;
166}
167
168ecb_inline void
169cbor_free (CBOR *cbor)
170{
171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
79} 173}
80 174
81///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
82// utility functions 176// utility functions
83 177
105 SvPV_renew (sv, SvCUR (sv) + 1); 199 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 200#endif
107 } 201 }
108} 202}
109 203
110///////////////////////////////////////////////////////////////////////////// 204// minimum length of a string to be registered for stringref
111// fp hell 205ecb_inline int
112 206minimum_string_length (UV idx)
113//TODO 207{
208 return idx <= 23 ? 3
209 : idx <= 0xffU ? 4
210 : idx <= 0xffffU ? 5
211 : idx <= 0xffffffffU ? 7
212 : 11;
213}
114 214
115///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
116// encoder 216// encoder
117 217
118// structure used for encoding CBOR 218// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 221 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 222 char *end; // SvEND (sv)
123 SV *sv; // result scalar 223 SV *sv; // result scalar
124 CBOR cbor; 224 CBOR cbor;
125 U32 depth; // recursion level 225 U32 depth; // recursion level
226 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
227 UV stringref_idx;
228 HV *shareable; // ptr => index, or 0
229 UV shareable_idx;
126} enc_t; 230} enc_t;
127 231
128ecb_inline void 232ecb_inline void
129need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
130{ 234{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 236 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142{ 246{
143 need (enc, 1); 247 need (enc, 1);
144 *enc->cur++ = ch; 248 *enc->cur++ = ch;
145} 249}
146 250
251// used for tags, intregers, element counts and so on
147static void 252static void
148encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
149{ 254{
150 need (enc, 9); 255 need (enc, 9);
151 256
152 if (len < 24) 257 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
154 else if (len <= 0xff) 259 else if (ecb_expect_true (len <= 0xffU))
155 { 260 {
156 *enc->cur++ = major | 24; 261 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 262 *enc->cur++ = len;
158 } 263 }
159 else if (len <= 0xffff) 264 else if (len <= 0xffffU)
160 { 265 {
161 *enc->cur++ = major | 25; 266 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 268 *enc->cur++ = len;
164 } 269 }
165 else if (len <= 0xffffffff) 270 else if (len <= 0xffffffffU)
166 { 271 {
167 *enc->cur++ = major | 26; 272 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 276 *enc->cur++ = len;
172 } 277 }
173 else 278 else
174 { 279 {
175 *enc->cur++ = major | 27; 280 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 281 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 282 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 283 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 284 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 285 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 288 *enc->cur++ = len;
184 } 289 }
185} 290}
186 291
187static void 292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
304ecb_inline void
305encode_tag (enc_t *enc, UV tag)
306{
307 encode_uint (enc, MAJOR_TAG, tag);
308}
309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 313{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
333ecb_inline void
334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 344 need (enc, len);
192 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
193 enc->cur += len; 346 enc->cur += len;
194} 347}
195 348
349ecb_inline void
350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
351{
352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
353 {
354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
355
356 if (SvOK (*svp))
357 {
358 // already registered, use stringref
359 encode_tag (enc, CBOR_TAG_STRINGREF);
360 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
361 return;
362 }
363 else if (len >= minimum_string_length (enc->stringref_idx))
364 {
365 // register only
366 sv_setuv (*svp, enc->stringref_idx);
367 ++enc->stringref_idx;
368 }
369 }
370
371 encode_str (enc, upgrade_utf8, utf8, str, len);
372}
373
374ecb_inline void
375encode_float16 (enc_t *enc, NV nv)
376{
377 need (enc, 1+2);
378
379 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
380
381 uint16_t fp = ecb_float_to_binary16 (nv);
382
383 if (!ecb_big_endian ())
384 fp = ecb_bswap16 (fp);
385
386 memcpy (enc->cur, &fp, 2);
387 enc->cur += 2;
388}
389
390ecb_inline void
391encode_float32 (enc_t *enc, NV nv)
392{
393 need (enc, 1+4);
394
395 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
396
397 uint32_t fp = ecb_float_to_binary32 (nv);
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap32 (fp);
401
402 memcpy (enc->cur, &fp, 4);
403 enc->cur += 4;
404}
405
406ecb_inline void
407encode_float64 (enc_t *enc, NV nv)
408{
409 need (enc, 1+8);
410
411 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
412
413 uint64_t fp = ecb_double_to_binary64 (nv);
414
415 if (!ecb_big_endian ())
416 fp = ecb_bswap64 (fp);
417
418 memcpy (enc->cur, &fp, 8);
419 enc->cur += 8;
420}
421
422ecb_inline void
423encode_forced (enc_t *enc, UV type, SV *sv)
424{
425 switch (type)
426 {
427 case AS_CBOR:
428 {
429 STRLEN len;
430 char *str = SvPVbyte (sv, len);
431
432 need (enc, len);
433 memcpy (enc->cur, str, len);
434 enc->cur += len;
435 }
436 break;
437
438 case AS_BYTES:
439 {
440 STRLEN len;
441 char *str = SvPVbyte (sv, len);
442 encode_strref (enc, 0, 0, str, len);
443 }
444 break;
445
446 case AS_TEXT:
447 {
448 STRLEN len;
449 char *str = SvPVutf8 (sv, len);
450 encode_strref (enc, 1, 1, str, len);
451 }
452 break;
453
454 case AS_INT: encode_int (enc, sv); break;
455 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
456 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
457 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
458
459 default:
460 croak ("encountered malformed CBOR::XS::Tagged object");
461 }
462}
463
196static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
197 465
198static void 466static void
199encode_av (enc_t *enc, AV *av) 467encode_av (enc_t *enc, AV *av)
200{ 468{
203 if (enc->depth >= enc->cbor.max_depth) 471 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 472 croak (ERR_NESTING_EXCEEDED);
205 473
206 ++enc->depth; 474 ++enc->depth;
207 475
208 encode_uint (enc, 0x80, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 477
478 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
211 { 480 {
212 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 483 }
484 else
485 for (i = 0; i <= len; ++i)
486 {
487 SV *sv = AvARRAY (av)[i];
488 encode_sv (enc, sv ? sv : &PL_sv_undef);
489 }
215 490
216 --enc->depth; 491 --enc->depth;
217} 492}
218 493
219static void 494static void
227 ++enc->depth; 502 ++enc->depth;
228 503
229 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
231 506
232 if (mg) 507 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 509 else
235 encode_uint (enc, 0xa0, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
236 511
237 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
238 { 513 {
239 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
241 else 516 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 517 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 518
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 519 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 520 }
246 521
247 if (mg) 522 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 524
250 --enc->depth; 525 --enc->depth;
251} 526}
252 527
253// encode objects, arrays and special \0=false and \1=true values. 528// encode objects, arrays and special \0=false and \1=true values.
254static void 529static void
255encode_rv (enc_t *enc, SV *sv) 530encode_rv (enc_t *enc, SV *sv)
256{ 531{
257 svtype svt;
258
259 SvGETMAGIC (sv); 532 SvGETMAGIC (sv);
533
260 svt = SvTYPE (sv); 534 svtype svt = SvTYPE (sv);
261 535
262 if (ecb_expect_false (SvOBJECT (sv))) 536 if (ecb_expect_false (SvOBJECT (sv)))
263 { 537 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 538 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 539 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 544 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 545 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 546 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 547
274 HV *stash = SvSTASH (sv); 548 HV *stash = SvSTASH (sv);
275 GV *method;
276 549
277 if (stash == boolean_stash) 550 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 551 {
552 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
553 return;
554 }
279 else if (stash == error_stash) 555 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 556 {
557 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
558 return;
559 }
281 else if (stash == tagged_stash) 560 else if (stash == tagged_stash)
282 { 561 {
283 if (svt != SVt_PVAV) 562 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 569 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 570 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
571 break;
572
573 case 3-1:
574 // a forced type [value, type, undef]
575 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
576 break;
577
578 default:
579 croak ("encountered malformed CBOR::XS::Tagged object");
580 }
581
582 return;
583 }
584 }
585
586 if (ecb_expect_false (SvREFCNT (sv) > 1)
587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
588 {
589 if (ecb_expect_false (!enc->shareable))
590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
591
592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
593
594 if (SvOK (*svp))
288 } 595 {
596 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
597 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
598 return;
599 }
600 else
601 {
602 sv_setuv (*svp, enc->shareable_idx);
603 ++enc->shareable_idx;
604 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
605 }
606 }
607
608 if (ecb_expect_false (SvOBJECT (sv)))
609 {
610 HV *stash = SvSTASH (sv);
611 GV *method;
612
613 if (enc->cbor.flags & F_FORBID_OBJECTS)
614 croak ("encountered object '%s', but forbid_objects is enabled",
615 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 617 {
291 dSP; 618 dSP;
292 619
293 ENTER; SAVETMPS; PUSHMARK (SP); 620 ENTER; SAVETMPS;
621 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 622 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 623 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 624
297 PUTBACK; 625 PUTBACK;
298 // G_SCALAR ensures that return value is 1 626 // G_SCALAR ensures that return value is 1
311 } 639 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 641 {
314 dSP; 642 dSP;
315 643
316 ENTER; SAVETMPS; PUSHMARK (SP); 644 ENTER; SAVETMPS;
645 PUSHMARK (SP);
317 EXTEND (SP, 2); 646 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 647 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 649 PUSHs (sv_cbor);
321 650
325 654
326 // catch this surprisingly common error 655 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 657 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 658
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 660 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 662
334 while (count) 663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
670 }
336 671
337 PUTBACK; 672 PUTBACK;
338 673
339 FREETMPS; LEAVE; 674 FREETMPS; LEAVE;
340 } 675 }
344 } 679 }
345 else if (svt == SVt_PVHV) 680 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 681 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 682 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 683 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 684 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 685 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 686 encode_tag (enc, CBOR_TAG_INDIRECTION);
687 encode_sv (enc, sv);
688 }
369} 689}
370 690
371static void 691static void
372encode_nv (enc_t *enc, SV *sv) 692encode_nv (enc_t *enc, SV *sv)
373{ 693{
374 double nv = SvNVX (sv); 694 double nv = SvNVX (sv);
375 695
376 need (enc, 9); 696 need (enc, 9);
377 697
378 if (ecb_expect_false (nv == (U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 700 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
382 { 702 encode_float32 (enc, nv);
383 uint32_t fp = ecb_float_to_binary32 (nv);
384
385 *enc->cur++ = 0xe0 | 26;
386
387 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp);
389
390 memcpy (enc->cur, &fp, 4);
391 enc->cur += 4;
392 }
393 else 703 else
394 { 704 encode_float64 (enc, nv);
395 uint64_t fp = ecb_double_to_binary64 (nv);
396
397 *enc->cur++ = 0xe0 | 27;
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp);
401
402 memcpy (enc->cur, &fp, 8);
403 enc->cur += 8;
404 }
405} 705}
406 706
407static void 707static void
408encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
409{ 709{
411 711
412 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
413 { 713 {
414 STRLEN len; 714 STRLEN len;
415 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 717 }
418 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 719 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
421 { 721 encode_int (enc, sv);
422 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv));
424 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv));
426 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
428 }
429 else if (SvROK (sv)) 722 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 727 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 728 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 729 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 730 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 731}
439 732
440static SV * 733static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
442{ 735{
443 enc_t enc; 736 enc_t enc = { 0 };
444 737
445 enc.cbor = *cbor; 738 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 742
451 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
744
745 if (cbor->flags & F_PACK_STRINGS)
746 {
747 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
748 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
749 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
750 }
751
452 encode_sv (&enc, scalar); 752 encode_sv (&enc, scalar);
453 753
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 754 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 755 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 756
470 U8 *end; // end of input string 770 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 771 const char *err; // parse error, if != 0
472 CBOR cbor; 772 CBOR cbor;
473 U32 depth; // recursion depth 773 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
775 AV *shareable;
776 AV *stringref;
777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 779} dec_t;
476 780
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 781// set dec->err to ERRSV
782ecb_cold static void
783err_errsv (dec_t *dec)
784{
785 if (!dec->err)
786 {
787 dec->err_sv = newSVsv (ERRSV);
478 788
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 789 // chop off the trailing \n
790 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
791 *SvEND (dec->err_sv) = 0;
480 792
793 dec->err = SvPVutf8_nolen (dec->err_sv);
794 }
795}
796
797// the following functions are used to reduce code size and help the compiler to optimise
798ecb_cold static void
799err_set (dec_t *dec, const char *reason)
800{
801 if (!dec->err)
802 dec->err = reason;
803}
804
805ecb_cold static void
806err_unexpected_end (dec_t *dec)
807{
808 err_set (dec, "unexpected end of CBOR data");
809}
810
811#define ERR_DO(do) SB do; goto fail; SE
812#define ERR(reason) ERR_DO (err_set (dec, reason))
813#define ERR_ERRSV ERR_DO (err_errsv (dec))
814
815#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
816
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 817#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
482#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
483 819
484static UV 820static UV
485decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
486{ 822{
487 switch (*dec->cur & 31) 823 U8 m = *dec->cur & MINOR_MASK;
488 { 824 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 825
494 case 24: 826 if (ecb_expect_true (m < LENGTH_EXT1))
827 return m;
828 else if (ecb_expect_true (m == LENGTH_EXT1))
829 {
495 WANT (2); 830 WANT (1);
496 dec->cur += 2; 831 dec->cur += 1;
497 return dec->cur[-1]; 832 return dec->cur[-1];
498 833 }
499 case 25: 834 else if (ecb_expect_true (m == LENGTH_EXT2))
835 {
500 WANT (3); 836 WANT (2);
501 dec->cur += 3; 837 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 838 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 839 | ((UV)dec->cur[-1]);
504 840 }
505 case 26: 841 else if (ecb_expect_true (m == LENGTH_EXT4))
842 {
506 WANT (5); 843 WANT (4);
507 dec->cur += 5; 844 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 845 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 846 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 847 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 848 | ((UV)dec->cur[-1]);
512 849 }
513 case 27: 850 else if (ecb_expect_true (m == LENGTH_EXT8))
851 {
514 WANT (9); 852 WANT (8);
515 dec->cur += 9; 853 dec->cur += 8;
854
855 return
856#if UVSIZE < 8
857 0
858#else
516 return (((UV)dec->cur[-8]) << 56) 859 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 860 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 861 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 862 | (((UV)dec->cur[-5]) << 32)
863#endif
520 | (((UV)dec->cur[-4]) << 24) 864 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 865 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 866 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 867 | ((UV)dec->cur[-1]);
524 868 }
525 default: 869 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 870 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 871
529fail: 872fail:
530 return 0; 873 return 0;
531} 874}
532 875
537{ 880{
538 AV *av = newAV (); 881 AV *av = newAV ();
539 882
540 DEC_INC_DEPTH; 883 DEC_INC_DEPTH;
541 884
542 if ((*dec->cur & 31) == 31) 885 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 886 {
544 ++dec->cur; 887 ++dec->cur;
545 888
546 for (;;) 889 for (;;)
547 { 890 {
548 WANT (1); 891 WANT (1);
549 892
550 if (*dec->cur == (0xe0 | 31)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 894 {
552 ++dec->cur; 895 ++dec->cur;
553 break; 896 break;
554 } 897 }
555 898
556 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
557 } 900 }
558 } 901 }
559 else 902 else
560 { 903 {
561 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
562 905
906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 907 av_fill (av, len - 1);
564 908
565 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 910 AvARRAY (av)[i] = decode_sv (dec);
567 } 911 }
568 912
569 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
571 915
572fail: 916fail:
573 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 919 return &PL_sv_undef;
576} 920}
577 921
922static void
923decode_he (dec_t *dec, HV *hv)
924{
925 // for speed reasons, we specialcase single-string
926 // byte or utf-8 strings as keys, but only when !stringref
927
928 if (ecb_expect_true (!dec->stringref))
929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
930 {
931 STRLEN len = decode_uint (dec);
932 char *key = (char *)dec->cur;
933
934 WANT (len);
935 dec->cur += len;
936
937 hv_store (hv, key, len, decode_sv (dec), 0);
938
939 return;
940 }
941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
942 {
943 STRLEN len = decode_uint (dec);
944 char *key = (char *)dec->cur;
945
946 WANT (len);
947 dec->cur += len;
948
949 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
950 if (!is_utf8_string (key, len))
951 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
952
953 hv_store (hv, key, -len, decode_sv (dec), 0);
954
955 return;
956 }
957
958 SV *k = decode_sv (dec);
959 SV *v = decode_sv (dec);
960
961 // we leak memory if uncaught exceptions are thrown by random magical
962 // methods, and this is hopefully the only place where it can happen,
963 // so if there is a chance of an exception, take the very slow path.
964 // since catching exceptions is "undocumented/internal/forbidden" by
965 // the new p5p powers, we need to call out to a perl function :/
966 if (ecb_expect_false (SvAMAGIC (k)))
967 {
968 dSP;
969
970 ENTER; SAVETMPS;
971 PUSHMARK (SP);
972 EXTEND (SP, 3);
973 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
974 PUSHs (sv_2mortal (k));
975 PUSHs (sv_2mortal (v));
976
977 PUTBACK;
978 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
979 SPAGAIN;
980
981 FREETMPS; LEAVE;
982
983 if (SvTRUE (ERRSV))
984 ERR_ERRSV;
985
986 return;
987 }
988
989 hv_store_ent (hv, k, v, 0);
990 SvREFCNT_dec_NN (k);
991
992fail:
993 ;
994}
995
578static SV * 996static SV *
579decode_hv (dec_t *dec) 997decode_hv (dec_t *dec)
580{ 998{
581 HV *hv = newHV (); 999 HV *hv = newHV ();
582 1000
583 DEC_INC_DEPTH; 1001 DEC_INC_DEPTH;
584 1002
585 if ((*dec->cur & 31) == 31) 1003 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1004 {
587 ++dec->cur; 1005 ++dec->cur;
588 1006
589 for (;;) 1007 for (;;)
590 { 1008 {
591 WANT (1); 1009 WANT (1);
592 1010
593 if (*dec->cur == (0xe0 | 31)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1012 {
595 ++dec->cur; 1013 ++dec->cur;
596 break; 1014 break;
597 } 1015 }
598 1016
599 SV *k = decode_sv (dec); 1017 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 1018 }
605 } 1019 }
606 else 1020 else
607 { 1021 {
608 int len = decode_uint (dec); 1022 UV pairs = decode_uint (dec);
609 1023
1024 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1025
610 while (len--) 1026 while (pairs--)
611 { 1027 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 1028 }
619 1029
620 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
622 1032
623fail: 1033fail:
624 SvREFCNT_dec (hv); 1034 SvREFCNT_dec_NN (hv);
625 DEC_DEC_DEPTH; 1035 DEC_DEC_DEPTH;
626 return &PL_sv_undef; 1036 return &PL_sv_undef;
627} 1037}
628 1038
629static SV * 1039static SV *
630decode_str (dec_t *dec, int utf8) 1040decode_str (dec_t *dec, int utf8)
631{ 1041{
632 SV *sv = 0; 1042 SV *sv = 0;
633 1043
634 if ((*dec->cur & 31) == 31) 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
635 { 1045 {
1046 // indefinite length strings
636 ++dec->cur; 1047 ++dec->cur;
637 1048
1049 U8 major = *dec->cur & MAJOR_MISC;
1050
638 sv = newSVpvn ("", 0); 1051 sv = newSVpvn ("", 0);
639 1052
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 1053 for (;;)
642 { 1054 {
643 WANT (1); 1055 WANT (1);
644 1056
645 if (*dec->cur == (0xe0 | 31)) 1057 if ((*dec->cur - major) > LENGTH_EXT8)
1058 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 1059 {
647 ++dec->cur; 1060 ++dec->cur;
648 break; 1061 break;
649 } 1062 }
1063 else
1064 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 1065
651 sv_catsv (sv, decode_sv (dec)); 1066 STRLEN len = decode_uint (dec);
1067
1068 WANT (len);
1069 sv_catpvn (sv, dec->cur, len);
1070 dec->cur += len;
652 } 1071 }
653 } 1072 }
654 else 1073 else
655 { 1074 {
656 STRLEN len = decode_uint (dec); 1075 STRLEN len = decode_uint (dec);
657 1076
658 WANT (len); 1077 WANT (len);
659 sv = newSVpvn (dec->cur, len); 1078 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 1079 dec->cur += len;
1080
1081 if (ecb_expect_false (dec->stringref)
1082 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1083 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 1084 }
662 1085
663 if (utf8) 1086 if (utf8)
1087 {
1088 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1089 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1090 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1091
664 SvUTF8_on (sv); 1092 SvUTF8_on (sv);
1093 }
665 1094
666 return sv; 1095 return sv;
667 1096
668fail: 1097fail:
669 SvREFCNT_dec (sv); 1098 SvREFCNT_dec (sv);
671} 1100}
672 1101
673static SV * 1102static SV *
674decode_tagged (dec_t *dec) 1103decode_tagged (dec_t *dec)
675{ 1104{
1105 SV *sv = 0;
676 UV tag = decode_uint (dec); 1106 UV tag = decode_uint (dec);
1107
1108 WANT (1);
1109
1110 switch (tag)
1111 {
1112 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 1113 sv = decode_sv (dec);
1114 break;
678 1115
679 if (tag == CBOR_TAG_MAGIC) 1116 case CBOR_TAG_INDIRECTION:
680 return sv; 1117 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 1118 break;
682 { 1119
1120 case CBOR_TAG_STRINGREF_NAMESPACE:
1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
1124 ENTER;
1125
1126 SAVESPTR (dec->stringref);
1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1128
1129 sv = decode_sv (dec);
1130
1131 LEAVE;
1132 }
1133 break;
1134
1135 case CBOR_TAG_STRINGREF:
1136 {
1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1139
1140 UV idx = decode_uint (dec);
1141
1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1143 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1144
1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1146 }
1147 break;
1148
1149 case CBOR_TAG_VALUE_SHAREABLE:
1150 {
1151 if (ecb_expect_false (!dec->shareable))
1152 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1153
1154 if (dec->cbor.flags & F_ALLOW_CYCLES)
1155 {
1156 sv = newSV (0);
1157 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1158
1159 SV *osv = decode_sv (dec);
1160 sv_setsv (sv, osv);
1161 SvREFCNT_dec_NN (osv);
1162 }
1163 else
1164 {
1165 av_push (dec->shareable, &PL_sv_undef);
1166 int idx = AvFILLp (dec->shareable);
1167 sv = decode_sv (dec);
1168 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1169 }
1170 }
1171 break;
1172
1173 case CBOR_TAG_VALUE_SHAREDREF:
1174 {
1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1177
1178 UV idx = decode_uint (dec);
1179
1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1182
1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1184
1185 if (sv == &PL_sv_undef)
1186 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1187 }
1188 break;
1189
1190 case CBOR_TAG_PERL_OBJECT:
1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
1195 sv = decode_sv (dec);
1196
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 1198 ERR ("corrupted CBOR data (non-array perl object)");
685 1199
686 AV *av = (AV *)SvRV (sv); 1200 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 1201 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1202 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 1203
690 if (!stash) 1204 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 1205 ERR ("cannot decode perl-object (package does not exist)");
692 1206
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1207 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 1208
695 if (!method) 1209 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1210 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 1211
698 dSP; 1212 dSP;
699 1213
700 ENTER; SAVETMPS; PUSHMARK (SP); 1214 ENTER; SAVETMPS;
1215 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1216 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1217 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1218 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1219 PUSHs (sv_cbor);
705 1220
706 int i; 1221 int i;
707 1222
708 for (i = 1; i < len; ++i) 1223 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1224 PUSHs (*av_fetch (av, i, 1));
710 1225
711 PUTBACK; 1226 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1227 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1228 SPAGAIN;
714 1229
1230 if (SvTRUE (ERRSV))
1231 {
1232 FREETMPS; LEAVE;
1233 ERR_ERRSV;
1234 }
1235
1236 SvREFCNT_dec_NN (sv);
715 sv = SvREFCNT_inc (POPs); 1237 sv = SvREFCNT_inc (POPs);
716 1238
717 PUTBACK; 1239 PUTBACK;
718 1240
719 FREETMPS; LEAVE; 1241 FREETMPS; LEAVE;
1242 }
1243 break;
720 1244
721 return sv; 1245 default:
722 } 1246 filter:
723 else 1247 {
724 { 1248 SV *tag_sv = newSVuv (tag);
1249
1250 sv = decode_sv (dec);
1251
1252 dSP;
1253 ENTER; SAVETMPS;
1254 PUSHMARK (SP);
1255 EXTEND (SP, 2);
1256 PUSHs (tag_sv);
1257 PUSHs (sv);
1258
1259 PUTBACK;
1260 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1261 SPAGAIN;
1262
1263 if (SvTRUE (ERRSV))
1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1266 FREETMPS; LEAVE;
1267 ERR_ERRSV;
1268 }
1269
1270 if (count)
1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1273 SvREFCNT_dec_NN (sv);
1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1276 }
1277 else
1278 {
725 AV *av = newAV (); 1279 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1280 av_push (av, tag_sv);
727 av_push (av, sv); 1281 av_push (av, sv);
728 1282
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1284 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1286 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1287 }
1288
1289 PUTBACK;
1290
1291 FREETMPS; LEAVE;
1292 }
1293 break;
734 } 1294 }
1295
1296 return sv;
735 1297
736fail: 1298fail:
737 SvREFCNT_dec (sv); 1299 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1300 return &PL_sv_undef;
739} 1301}
741static SV * 1303static SV *
742decode_sv (dec_t *dec) 1304decode_sv (dec_t *dec)
743{ 1305{
744 WANT (1); 1306 WANT (1);
745 1307
746 switch (*dec->cur >> 5) 1308 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1309 {
748 case 0: // unsigned int 1310 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1311 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1312 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1313 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1314 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1315 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1316 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1317
756 case 4: // array 1318 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1319 switch (*dec->cur++ & MINOR_MASK)
764 { 1320 {
765 case 20: 1321 case SIMPLE_FALSE:
766#if CBOR_SLOW 1322#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1323 types_false = get_bool ("Types::Serialiser::false");
768#endif 1324#endif
769 return newSVsv (types_false); 1325 return newSVsv (types_false);
770 case 21: 1326 case SIMPLE_TRUE:
771#if CBOR_SLOW 1327#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1328 types_true = get_bool ("Types::Serialiser::true");
773#endif 1329#endif
774 return newSVsv (types_true); 1330 return newSVsv (types_true);
775 case 22: 1331 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1332 return newSVsv (&PL_sv_undef);
777 case 23: 1333 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1334#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1335 types_error = get_bool ("Types::Serialiser::error");
780#endif 1336#endif
781 return newSVsv (types_error); 1337 return newSVsv (types_error);
782 1338
783 case 25: 1339 case MISC_FLOAT16:
784 { 1340 {
785 WANT (2); 1341 WANT (2);
786 1342
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1343 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1344 dec->cur += 2;
789 1345
790 return newSVnv (ecb_binary16_to_float (fp)); 1346 return newSVnv (ecb_binary16_to_float (fp));
791 } 1347 }
792 1348
793 case 26: 1349 case MISC_FLOAT32:
794 { 1350 {
795 uint32_t fp; 1351 uint32_t fp;
796 WANT (4); 1352 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1353 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1354 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1357 fp = ecb_bswap32 (fp);
802 1358
803 return newSVnv (ecb_binary32_to_float (fp)); 1359 return newSVnv (ecb_binary32_to_float (fp));
804 } 1360 }
805 1361
806 case 27: 1362 case MISC_FLOAT64:
807 { 1363 {
808 uint64_t fp; 1364 uint64_t fp;
809 WANT (8); 1365 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1366 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1367 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1370 fp = ecb_bswap64 (fp);
815 1371
816 return newSVnv (ecb_binary64_to_double (fp)); 1372 return newSVnv (ecb_binary64_to_double (fp));
817 } 1373 }
818 1374
819 // 0..19 unassigned 1375 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1376 // 24 reserved + unassigned simple (reserved values are not encodable)
1377 // 28-30 unassigned misc
1378 // 31 break code
821 default: 1379 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1380 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1381 }
824 1382
825 break; 1383 break;
826 } 1384 }
827 1385
830} 1388}
831 1389
832static SV * 1390static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1391decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1392{
835 dec_t dec; 1393 dec_t dec = { 0 };
836 SV *sv; 1394 SV *sv;
1395 STRLEN len;
1396 char *data = SvPVbyte (string, len);
837 1397
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1398 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1399 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1400 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1401
873 dec.cbor = *cbor; 1402 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1403 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1404 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1405
879 sv = decode_sv (&dec); 1406 sv = decode_sv (&dec);
880 1407
881 if (offset_return) 1408 if (offset_return)
882 *offset_return = dec.cur; 1409 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1412 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1413 dec.err = "garbage after CBOR object";
887 1414
888 if (dec.err) 1415 if (dec.err)
889 { 1416 {
1417 if (dec.shareable)
1418 {
1419 // need to break cyclic links, which would all be in shareable
1420 int i;
1421 SV **svp;
1422
1423 for (i = av_len (dec.shareable) + 1; i--; )
1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1425 sv_setsv (*svp, &PL_sv_undef);
1426 }
1427
890 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1433 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1434 }
893 1435
894 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
895 1437
896 return sv; 1438 return sv;
897} 1439}
898 1440
1441/////////////////////////////////////////////////////////////////////////////
1442// incremental parser
1443
1444#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1445
1446// returns 0 for notyet, 1 for success or error
1447static int
1448incr_parse (CBOR *self, SV *cborstr)
1449{
1450 STRLEN cur;
1451 SvPV (cborstr, cur);
1452
1453 while (ecb_expect_true (self->incr_need <= cur))
1454 {
1455 // table of integer count bytes
1456 static I8 incr_len[MINOR_MASK + 1] = {
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 0, 0, 0, 0, 0, 0, 0, 0,
1459 0, 0, 0, 0, 0, 0, 0, 0,
1460 1, 2, 4, 8,-1,-1,-1,-2
1461 };
1462
1463 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1464 U8 m = *p & MINOR_MASK;
1465 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1466 I8 ilen = incr_len[m];
1467
1468 self->incr_need = self->incr_pos + 1;
1469
1470 if (ecb_expect_false (ilen < 0))
1471 {
1472 if (m != MINOR_INDEF)
1473 return 1; // error
1474
1475 if (*p == (MAJOR_MISC | MINOR_INDEF))
1476 {
1477 if (count >= 0)
1478 return 1; // error
1479
1480 count = 1;
1481 }
1482 else
1483 {
1484 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1485 count = -1;
1486 }
1487 }
1488 else
1489 {
1490 self->incr_need += ilen;
1491 if (ecb_expect_false (self->incr_need > cur))
1492 return 0;
1493
1494 int major = *p >> MAJOR_SHIFT;
1495
1496 switch (major)
1497 {
1498 case MAJOR_TAG >> MAJOR_SHIFT:
1499 ++count; // tags merely prefix another value
1500 break;
1501
1502 case MAJOR_BYTES >> MAJOR_SHIFT:
1503 case MAJOR_TEXT >> MAJOR_SHIFT:
1504 case MAJOR_ARRAY >> MAJOR_SHIFT:
1505 case MAJOR_MAP >> MAJOR_SHIFT:
1506 {
1507 UV len;
1508
1509 if (ecb_expect_false (ilen))
1510 {
1511 len = 0;
1512
1513 do {
1514 len = (len << 8) | *++p;
1515 } while (--ilen);
1516 }
1517 else
1518 len = m;
1519
1520 switch (major)
1521 {
1522 case MAJOR_BYTES >> MAJOR_SHIFT:
1523 case MAJOR_TEXT >> MAJOR_SHIFT:
1524 self->incr_need += len;
1525 if (ecb_expect_false (self->incr_need > cur))
1526 return 0;
1527
1528 break;
1529
1530 case MAJOR_MAP >> MAJOR_SHIFT:
1531 len <<= 1;
1532 case MAJOR_ARRAY >> MAJOR_SHIFT:
1533 if (len)
1534 {
1535 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1536 count = len + 1;
1537 }
1538 break;
1539 }
1540 }
1541 }
1542 }
1543
1544 self->incr_pos = self->incr_need;
1545
1546 if (count > 0)
1547 {
1548 while (!--count)
1549 {
1550 if (!AvFILLp (self->incr_count))
1551 return 1; // done
1552
1553 SvREFCNT_dec_NN (av_pop (self->incr_count));
1554 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1555 }
1556
1557 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1558 }
1559 }
1560
1561 return 0;
1562}
1563
1564
899///////////////////////////////////////////////////////////////////////////// 1565/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1566// XS interface functions
901 1567
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1568MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1569
911 1577
912 types_true = get_bool ("Types::Serialiser::true" ); 1578 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1579 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1580 types_error = get_bool ("Types::Serialiser::error");
915 1581
1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1583
916 sv_cbor = newSVpv ("CBOR", 0); 1584 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
918} 1588}
919 1589
920PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
921 1591
922void CLONE (...) 1592void CLONE (...)
940 1610
941void shrink (CBOR *self, int enable = 1) 1611void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1612 ALIAS:
943 shrink = F_SHRINK 1613 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1615 allow_sharing = F_ALLOW_SHARING
1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1620 text_strings = F_TEXT_STRINGS
1621 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1622 PPCODE:
946{ 1623{
947 if (enable) 1624 if (enable)
948 self->flags |= ix; 1625 self->flags |= ix;
949 else 1626 else
954 1631
955void get_shrink (CBOR *self) 1632void get_shrink (CBOR *self)
956 ALIAS: 1633 ALIAS:
957 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1636 get_allow_sharing = F_ALLOW_SHARING
1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1639 get_pack_strings = F_PACK_STRINGS
1640 get_text_keys = F_TEXT_KEYS
1641 get_text_strings = F_TEXT_STRINGS
1642 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1643 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
961 1645
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1647 PPCODE:
979 CODE: 1663 CODE:
980 RETVAL = self->max_size; 1664 RETVAL = self->max_size;
981 OUTPUT: 1665 OUTPUT:
982 RETVAL 1666 RETVAL
983 1667
984#if 0 //TODO 1668void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1669 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1670 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1671 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1672 XPUSHs (ST (0));
993}
994 1673
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1674SV *get_filter (CBOR *self)
996 PPCODE: 1675 CODE:
997{ 1676 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1677 OUTPUT:
999 self->cb_sk_object = newHV (); 1678 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1679
1019void encode (CBOR *self, SV *scalar) 1680void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1681 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1682 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1683 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1696 EXTEND (SP, 2);
1036 PUSHs (sv); 1697 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1698 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1699}
1039 1700
1040#if 0 1701void incr_parse (CBOR *self, SV *cborstr)
1702 ALIAS:
1703 incr_parse_multiple = 1
1704 PPCODE:
1705{
1706 if (SvUTF8 (cborstr))
1707 sv_utf8_downgrade (cborstr, 0);
1708
1709 if (!self->incr_count)
1710 {
1711 self->incr_count = newAV ();
1712 self->incr_pos = 0;
1713 self->incr_need = 1;
1714
1715 av_push (self->incr_count, newSViv (1));
1716 }
1717
1718 do
1719 {
1720 if (!incr_parse (self, cborstr))
1721 {
1722 if (self->incr_need > self->max_size && self->max_size)
1723 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1724 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1725
1726 break;
1727 }
1728
1729 SV *sv;
1730 char *offset;
1731
1732 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1733 XPUSHs (sv);
1734
1735 sv_chop (cborstr, offset);
1736
1737 av_clear (self->incr_count);
1738 av_push (self->incr_count, newSViv (1));
1739
1740 self->incr_pos = 0;
1741 self->incr_need = self->incr_pos + 1;
1742 }
1743 while (ix);
1744}
1745
1746void incr_reset (CBOR *self)
1747 CODE:
1748{
1749 SvREFCNT_dec (self->incr_count);
1750 self->incr_count = 0;
1751}
1041 1752
1042void DESTROY (CBOR *self) 1753void DESTROY (CBOR *self)
1043 CODE: 1754 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1755 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1756
1049PROTOTYPES: ENABLE 1757PROTOTYPES: ENABLE
1050 1758
1051void encode_cbor (SV *scalar) 1759void encode_cbor (SV *scalar)
1760 ALIAS:
1761 encode_cbor = 0
1762 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1763 PPCODE:
1053{ 1764{
1054 CBOR cbor; 1765 CBOR cbor;
1055 cbor_init (&cbor); 1766 cbor_init (&cbor);
1767 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1768 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1769 XPUSHs (scalar);
1058} 1770}
1059 1771
1060void decode_cbor (SV *cborstr) 1772void decode_cbor (SV *cborstr)
1064 cbor_init (&cbor); 1776 cbor_init (&cbor);
1065 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1066 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
1067} 1779}
1068 1780
1781#ifdef __AFL_COMPILER
1782
1783void
1784afl_init ()
1785 CODE:
1786 __AFL_INIT ();
1787
1788int
1789afl_loop (unsigned int count = 10000)
1790 CODE:
1791 RETVAL = __AFL_LOOP (count);
1792 OUTPUT:
1793 RETVAL
1794
1795#endif
1796

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines