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 1.11 by root, Sun Oct 27 22:35:15 2013 UTC vs.
Revision 1.69 by root, Mon Nov 30 20:38:25 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_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 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_bool (enc_t *enc, int istrue)
424{
425 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
426}
427
428ecb_inline void
429encode_forced (enc_t *enc, UV type, SV *sv)
430{
431 switch (type)
432 {
433 case AS_CBOR:
434 {
435 STRLEN len;
436 char *str = SvPVbyte (sv, len);
437
438 need (enc, len);
439 memcpy (enc->cur, str, len);
440 enc->cur += len;
441 }
442 break;
443
444 case AS_BYTES:
445 {
446 STRLEN len;
447 char *str = SvPVbyte (sv, len);
448 encode_strref (enc, 0, 0, str, len);
449 }
450 break;
451
452 case AS_TEXT:
453 {
454 STRLEN len;
455 char *str = SvPVutf8 (sv, len);
456 encode_strref (enc, 1, 1, str, len);
457 }
458 break;
459
460 case AS_INT: encode_int (enc, sv); break;
461
462 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
463 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
464 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
465
466 default:
467 croak ("encountered malformed CBOR::XS::Tagged object");
468 }
469}
470
196static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
197 472
198static void 473static void
199encode_av (enc_t *enc, AV *av) 474encode_av (enc_t *enc, AV *av)
200{ 475{
203 if (enc->depth >= enc->cbor.max_depth) 478 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 479 croak (ERR_NESTING_EXCEEDED);
205 480
206 ++enc->depth; 481 ++enc->depth;
207 482
208 encode_uint (enc, 0x80, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 484
485 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
211 { 487 {
212 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 490 }
491 else
492 for (i = 0; i <= len; ++i)
493 {
494 SV *sv = AvARRAY (av)[i];
495 encode_sv (enc, sv ? sv : &PL_sv_undef);
496 }
215 497
216 --enc->depth; 498 --enc->depth;
217} 499}
218 500
219static void 501static void
227 ++enc->depth; 509 ++enc->depth;
228 510
229 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
231 513
232 if (mg) 514 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 516 else
235 encode_uint (enc, 0xa0, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
236 518
237 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
238 { 520 {
239 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
241 else 523 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 524 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 525
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 526 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 527 }
246 528
247 if (mg) 529 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 531
250 --enc->depth; 532 --enc->depth;
251} 533}
252 534
253// encode objects, arrays and special \0=false and \1=true values. 535// encode objects, arrays and special \0=false and \1=true values.
254static void 536static void
255encode_rv (enc_t *enc, SV *sv) 537encode_rv (enc_t *enc, SV *sv)
256{ 538{
257 svtype svt;
258
259 SvGETMAGIC (sv); 539 SvGETMAGIC (sv);
540
260 svt = SvTYPE (sv); 541 svtype svt = SvTYPE (sv);
261 542
262 if (ecb_expect_false (SvOBJECT (sv))) 543 if (ecb_expect_false (SvOBJECT (sv)))
263 { 544 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 545 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 546 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 551 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 552 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 553 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 554
274 HV *stash = SvSTASH (sv); 555 HV *stash = SvSTASH (sv);
275 GV *method;
276 556
277 if (stash == boolean_stash) 557 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 558 {
559 encode_bool (enc, SvIV (sv));
560 return;
561 }
279 else if (stash == error_stash) 562 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 563 {
564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
565 return;
566 }
281 else if (stash == tagged_stash) 567 else if (stash == tagged_stash)
282 { 568 {
283 if (svt != SVt_PVAV) 569 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 576 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 577 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
578 break;
579
580 case 3-1:
581 // a forced type [value, type, undef]
582 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
583 break;
584
585 default:
586 croak ("encountered malformed CBOR::XS::Tagged object");
587 }
588
589 return;
590 }
591 }
592
593 if (ecb_expect_false (SvREFCNT (sv) > 1)
594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
595 {
596 if (ecb_expect_false (!enc->shareable))
597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
598
599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
600
601 if (SvOK (*svp))
288 } 602 {
603 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
604 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
605 return;
606 }
607 else
608 {
609 sv_setuv (*svp, enc->shareable_idx);
610 ++enc->shareable_idx;
611 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
612 }
613 }
614
615 if (ecb_expect_false (SvOBJECT (sv)))
616 {
617 HV *stash = SvSTASH (sv);
618 GV *method;
619
620 if (enc->cbor.flags & F_FORBID_OBJECTS)
621 croak ("encountered object '%s', but forbid_objects is enabled",
622 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 624 {
291 dSP; 625 dSP;
292 626
293 ENTER; SAVETMPS; PUSHMARK (SP); 627 ENTER; SAVETMPS;
628 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 629 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 630 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 631
297 PUTBACK; 632 PUTBACK;
298 // G_SCALAR ensures that return value is 1 633 // G_SCALAR ensures that return value is 1
311 } 646 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 648 {
314 dSP; 649 dSP;
315 650
316 ENTER; SAVETMPS; PUSHMARK (SP); 651 ENTER; SAVETMPS;
652 PUSHMARK (SP);
317 EXTEND (SP, 2); 653 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 654 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 656 PUSHs (sv_cbor);
321 657
325 661
326 // catch this surprisingly common error 662 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 663 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)); 664 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 665
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 667 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 669
334 while (count) 670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
677 }
336 678
337 PUTBACK; 679 PUTBACK;
338 680
339 FREETMPS; LEAVE; 681 FREETMPS; LEAVE;
340 } 682 }
344 } 686 }
345 else if (svt == SVt_PVHV) 687 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 688 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 689 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 690 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 691 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 692 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 693 encode_tag (enc, CBOR_TAG_INDIRECTION);
694 encode_sv (enc, sv);
695 }
369} 696}
370 697
371static void 698static void
372encode_nv (enc_t *enc, SV *sv) 699encode_nv (enc_t *enc, SV *sv)
373{ 700{
374 double nv = SvNVX (sv); 701 double nv = SvNVX (sv);
375 702
376 need (enc, 9); 703 need (enc, 9);
377 704
378 if (ecb_expect_false (nv == (U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 707 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
382 { 709 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 710 else
394 { 711 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} 712}
406 713
407static void 714static void
408encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
409{ 716{
411 718
412 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
413 { 720 {
414 STRLEN len; 721 STRLEN len;
415 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 724 }
418 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 726 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
421 { 728 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)) 729 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 734 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 735 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 736 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 737 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 738}
439 739
440static SV * 740static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
442{ 742{
443 enc_t enc; 743 enc_t enc = { 0 };
444 744
445 enc.cbor = *cbor; 745 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 749
451 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
751
752 if (cbor->flags & F_PACK_STRINGS)
753 {
754 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
755 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
756 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
757 }
758
452 encode_sv (&enc, scalar); 759 encode_sv (&enc, scalar);
453 760
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 761 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 762 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 763
470 U8 *end; // end of input string 777 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 778 const char *err; // parse error, if != 0
472 CBOR cbor; 779 CBOR cbor;
473 U32 depth; // recursion depth 780 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
782 AV *shareable;
783 AV *stringref;
784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 786} dec_t;
476 787
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 788// set dec->err to ERRSV
789ecb_cold static void
790err_errsv (dec_t *dec)
791{
792 if (!dec->err)
793 {
794 dec->err_sv = newSVsv (ERRSV);
478 795
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 796 // chop off the trailing \n
797 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
798 *SvEND (dec->err_sv) = 0;
480 799
800 dec->err = SvPVutf8_nolen (dec->err_sv);
801 }
802}
803
804// the following functions are used to reduce code size and help the compiler to optimise
805ecb_cold static void
806err_set (dec_t *dec, const char *reason)
807{
808 if (!dec->err)
809 dec->err = reason;
810}
811
812ecb_cold static void
813err_unexpected_end (dec_t *dec)
814{
815 err_set (dec, "unexpected end of CBOR data");
816}
817
818#define ERR_DO(do) SB do; goto fail; SE
819#define ERR(reason) ERR_DO (err_set (dec, reason))
820#define ERR_ERRSV ERR_DO (err_errsv (dec))
821
822#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
823
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 824#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 825#define DEC_DEC_DEPTH --dec->depth
483 826
484static UV 827static UV
485decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
486{ 829{
487 switch (*dec->cur & 31) 830 U8 m = *dec->cur & MINOR_MASK;
488 { 831 ++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;
493 832
494 case 24: 833 if (ecb_expect_true (m < LENGTH_EXT1))
834 return m;
835 else if (ecb_expect_true (m == LENGTH_EXT1))
836 {
495 WANT (2); 837 WANT (1);
496 dec->cur += 2; 838 dec->cur += 1;
497 return dec->cur[-1]; 839 return dec->cur[-1];
498 840 }
499 case 25: 841 else if (ecb_expect_true (m == LENGTH_EXT2))
842 {
500 WANT (3); 843 WANT (2);
501 dec->cur += 3; 844 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 845 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
504 847 }
505 case 26: 848 else if (ecb_expect_true (m == LENGTH_EXT4))
849 {
506 WANT (5); 850 WANT (4);
507 dec->cur += 5; 851 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 852 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 853 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 854 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 855 | ((UV)dec->cur[-1]);
512 856 }
513 case 27: 857 else if (ecb_expect_true (m == LENGTH_EXT8))
858 {
514 WANT (9); 859 WANT (8);
515 dec->cur += 9; 860 dec->cur += 8;
861
862 return
863#if UVSIZE < 8
864 0
865#else
516 return (((UV)dec->cur[-8]) << 56) 866 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 867 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 868 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 869 | (((UV)dec->cur[-5]) << 32)
870#endif
520 | (((UV)dec->cur[-4]) << 24) 871 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 872 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 873 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 874 | ((UV)dec->cur[-1]);
524 875 }
525 default: 876 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 877 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 878
529fail: 879fail:
530 return 0; 880 return 0;
531} 881}
532 882
537{ 887{
538 AV *av = newAV (); 888 AV *av = newAV ();
539 889
540 DEC_INC_DEPTH; 890 DEC_INC_DEPTH;
541 891
542 if ((*dec->cur & 31) == 31) 892 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 893 {
544 ++dec->cur; 894 ++dec->cur;
545 895
546 for (;;) 896 for (;;)
547 { 897 {
548 WANT (1); 898 WANT (1);
549 899
550 if (*dec->cur == (0xe0 | 31)) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 901 {
552 ++dec->cur; 902 ++dec->cur;
553 break; 903 break;
554 } 904 }
555 905
556 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
557 } 907 }
558 } 908 }
559 else 909 else
560 { 910 {
561 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
562 912
913 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); 914 av_fill (av, len - 1);
564 915
565 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 917 AvARRAY (av)[i] = decode_sv (dec);
567 } 918 }
568 919
569 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
571 922
572fail: 923fail:
573 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 926 return &PL_sv_undef;
576} 927}
577 928
929static void
930decode_he (dec_t *dec, HV *hv)
931{
932 // for speed reasons, we specialcase single-string
933 // byte or utf-8 strings as keys, but only when !stringref
934
935 if (ecb_expect_true (!dec->stringref))
936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
937 {
938 STRLEN len = decode_uint (dec);
939 char *key = (char *)dec->cur;
940
941 WANT (len);
942 dec->cur += len;
943
944 hv_store (hv, key, len, decode_sv (dec), 0);
945
946 return;
947 }
948 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
949 {
950 STRLEN len = decode_uint (dec);
951 char *key = (char *)dec->cur;
952
953 WANT (len);
954 dec->cur += len;
955
956 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
957 if (!is_utf8_string (key, len))
958 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
959
960 hv_store (hv, key, -len, decode_sv (dec), 0);
961
962 return;
963 }
964
965 SV *k = decode_sv (dec);
966 SV *v = decode_sv (dec);
967
968 // we leak memory if uncaught exceptions are thrown by random magical
969 // methods, and this is hopefully the only place where it can happen,
970 // so if there is a chance of an exception, take the very slow path.
971 // since catching exceptions is "undocumented/internal/forbidden" by
972 // the new p5p powers, we need to call out to a perl function :/
973 if (ecb_expect_false (SvAMAGIC (k)))
974 {
975 dSP;
976
977 ENTER; SAVETMPS;
978 PUSHMARK (SP);
979 EXTEND (SP, 3);
980 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
981 PUSHs (sv_2mortal (k));
982 PUSHs (sv_2mortal (v));
983
984 PUTBACK;
985 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
986 SPAGAIN;
987
988 FREETMPS; LEAVE;
989
990 if (SvTRUE (ERRSV))
991 ERR_ERRSV;
992
993 return;
994 }
995
996 hv_store_ent (hv, k, v, 0);
997 SvREFCNT_dec_NN (k);
998
999fail:
1000 ;
1001}
1002
578static SV * 1003static SV *
579decode_hv (dec_t *dec) 1004decode_hv (dec_t *dec)
580{ 1005{
581 HV *hv = newHV (); 1006 HV *hv = newHV ();
582 1007
583 DEC_INC_DEPTH; 1008 DEC_INC_DEPTH;
584 1009
585 if ((*dec->cur & 31) == 31) 1010 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1011 {
587 ++dec->cur; 1012 ++dec->cur;
588 1013
589 for (;;) 1014 for (;;)
590 { 1015 {
591 WANT (1); 1016 WANT (1);
592 1017
593 if (*dec->cur == (0xe0 | 31)) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1019 {
595 ++dec->cur; 1020 ++dec->cur;
596 break; 1021 break;
597 } 1022 }
598 1023
599 SV *k = decode_sv (dec); 1024 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 } 1025 }
604 } 1026 }
605 else 1027 else
606 { 1028 {
607 int len = decode_uint (dec); 1029 UV pairs = decode_uint (dec);
608 1030
1031 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1032
609 while (len--) 1033 while (pairs--)
610 { 1034 decode_he (dec, hv);
611 SV *k = decode_sv (dec);
612 SV *v = decode_sv (dec);
613
614 hv_store_ent (hv, k, v, 0);
615 }
616 } 1035 }
617 1036
618 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
619 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
620 1039
621fail: 1040fail:
622 SvREFCNT_dec (hv); 1041 SvREFCNT_dec_NN (hv);
623 DEC_DEC_DEPTH; 1042 DEC_DEC_DEPTH;
624 return &PL_sv_undef; 1043 return &PL_sv_undef;
625} 1044}
626 1045
627static SV * 1046static SV *
628decode_str (dec_t *dec, int utf8) 1047decode_str (dec_t *dec, int utf8)
629{ 1048{
630 SV *sv = 0; 1049 SV *sv = 0;
631 1050
632 if ((*dec->cur & 31) == 31) 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
633 { 1052 {
1053 // indefinite length strings
634 ++dec->cur; 1054 ++dec->cur;
635 1055
1056 U8 major = *dec->cur & MAJOR_MISC;
1057
636 sv = newSVpvn ("", 0); 1058 sv = newSVpvn ("", 0);
637 1059
638 // not very fast, and certainly not robust against illegal input
639 for (;;) 1060 for (;;)
640 { 1061 {
641 WANT (1); 1062 WANT (1);
642 1063
643 if (*dec->cur == (0xe0 | 31)) 1064 if ((*dec->cur - major) > LENGTH_EXT8)
1065 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
644 { 1066 {
645 ++dec->cur; 1067 ++dec->cur;
646 break; 1068 break;
647 } 1069 }
1070 else
1071 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
648 1072
649 sv_catsv (sv, decode_sv (dec)); 1073 STRLEN len = decode_uint (dec);
1074
1075 WANT (len);
1076 sv_catpvn (sv, dec->cur, len);
1077 dec->cur += len;
650 } 1078 }
651 } 1079 }
652 else 1080 else
653 { 1081 {
654 STRLEN len = decode_uint (dec); 1082 STRLEN len = decode_uint (dec);
655 1083
656 WANT (len); 1084 WANT (len);
657 sv = newSVpvn (dec->cur, len); 1085 sv = newSVpvn (dec->cur, len);
658 dec->cur += len; 1086 dec->cur += len;
1087
1088 if (ecb_expect_false (dec->stringref)
1089 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1090 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
659 } 1091 }
660 1092
661 if (utf8) 1093 if (utf8)
1094 {
1095 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1096 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1097 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1098
662 SvUTF8_on (sv); 1099 SvUTF8_on (sv);
1100 }
663 1101
664 return sv; 1102 return sv;
665 1103
666fail: 1104fail:
667 SvREFCNT_dec (sv); 1105 SvREFCNT_dec (sv);
669} 1107}
670 1108
671static SV * 1109static SV *
672decode_tagged (dec_t *dec) 1110decode_tagged (dec_t *dec)
673{ 1111{
1112 SV *sv = 0;
674 UV tag = decode_uint (dec); 1113 UV tag = decode_uint (dec);
1114
1115 WANT (1);
1116
1117 switch (tag)
1118 {
1119 case CBOR_TAG_MAGIC:
675 SV *sv = decode_sv (dec); 1120 sv = decode_sv (dec);
1121 break;
676 1122
677 if (tag == CBOR_TAG_MAGIC) 1123 case CBOR_TAG_INDIRECTION:
678 return sv; 1124 sv = newRV_noinc (decode_sv (dec));
679 else if (tag == CBOR_TAG_PERL_OBJECT) 1125 break;
680 { 1126
1127 case CBOR_TAG_STRINGREF_NAMESPACE:
1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
1131 ENTER;
1132
1133 SAVESPTR (dec->stringref);
1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1135
1136 sv = decode_sv (dec);
1137
1138 LEAVE;
1139 }
1140 break;
1141
1142 case CBOR_TAG_STRINGREF:
1143 {
1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1146
1147 UV idx = decode_uint (dec);
1148
1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1150 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1151
1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1153 }
1154 break;
1155
1156 case CBOR_TAG_VALUE_SHAREABLE:
1157 {
1158 if (ecb_expect_false (!dec->shareable))
1159 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1160
1161 if (dec->cbor.flags & F_ALLOW_CYCLES)
1162 {
1163 sv = newSV (0);
1164 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1165
1166 SV *osv = decode_sv (dec);
1167 sv_setsv (sv, osv);
1168 SvREFCNT_dec_NN (osv);
1169 }
1170 else
1171 {
1172 av_push (dec->shareable, &PL_sv_undef);
1173 int idx = AvFILLp (dec->shareable);
1174 sv = decode_sv (dec);
1175 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1176 }
1177 }
1178 break;
1179
1180 case CBOR_TAG_VALUE_SHAREDREF:
1181 {
1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1184
1185 UV idx = decode_uint (dec);
1186
1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1189
1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1191
1192 if (sv == &PL_sv_undef)
1193 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1194 }
1195 break;
1196
1197 case CBOR_TAG_PERL_OBJECT:
1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
1202 sv = decode_sv (dec);
1203
681 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
682 ERR ("corrupted CBOR data (non-array perl object)"); 1205 ERR ("corrupted CBOR data (non-array perl object)");
683 1206
684 AV *av = (AV *)SvRV (sv); 1207 AV *av = (AV *)SvRV (sv);
685 int len = av_len (av) + 1; 1208 int len = av_len (av) + 1;
686 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1209 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
687 1210
688 if (!stash) 1211 if (!stash)
689 ERR ("cannot decode perl-object (package does not exist)"); 1212 ERR ("cannot decode perl-object (package does not exist)");
690 1213
691 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1214 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
692 1215
693 if (!method) 1216 if (!method)
694 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1217 ERR ("cannot decode perl-object (package does not have a THAW method)");
695 1218
696 dSP; 1219 dSP;
697 1220
698 ENTER; SAVETMPS; PUSHMARK (SP); 1221 ENTER; SAVETMPS;
1222 PUSHMARK (SP);
699 EXTEND (SP, len + 1); 1223 EXTEND (SP, len + 1);
700 // we re-bless the reference to get overload and other niceties right 1224 // we re-bless the reference to get overload and other niceties right
701 PUSHs (*av_fetch (av, 0, 1)); 1225 PUSHs (*av_fetch (av, 0, 1));
702 PUSHs (sv_cbor); 1226 PUSHs (sv_cbor);
703 1227
704 int i; 1228 int i;
705 1229
706 for (i = 1; i < len; ++i) 1230 for (i = 1; i < len; ++i)
707 PUSHs (*av_fetch (av, i, 1)); 1231 PUSHs (*av_fetch (av, i, 1));
708 1232
709 PUTBACK; 1233 PUTBACK;
710 call_sv ((SV *)GvCV (method), G_SCALAR); 1234 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
711 SPAGAIN; 1235 SPAGAIN;
712 1236
1237 if (SvTRUE (ERRSV))
1238 {
1239 FREETMPS; LEAVE;
1240 ERR_ERRSV;
1241 }
1242
1243 SvREFCNT_dec_NN (sv);
713 sv = SvREFCNT_inc (POPs); 1244 sv = SvREFCNT_inc (POPs);
714 1245
715 PUTBACK; 1246 PUTBACK;
716 1247
717 FREETMPS; LEAVE; 1248 FREETMPS; LEAVE;
1249 }
1250 break;
718 1251
719 return sv; 1252 default:
720 } 1253 filter:
721 else 1254 {
722 { 1255 SV *tag_sv = newSVuv (tag);
1256
1257 sv = decode_sv (dec);
1258
1259 dSP;
1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
1262 EXTEND (SP, 2);
1263 PUSHs (tag_sv);
1264 PUSHs (sv);
1265
1266 PUTBACK;
1267 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1268 SPAGAIN;
1269
1270 if (SvTRUE (ERRSV))
1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1273 FREETMPS; LEAVE;
1274 ERR_ERRSV;
1275 }
1276
1277 if (count)
1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1280 SvREFCNT_dec_NN (sv);
1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1283 }
1284 else
1285 {
723 AV *av = newAV (); 1286 AV *av = newAV ();
724 av_push (av, newSVuv (tag)); 1287 av_push (av, tag_sv);
725 av_push (av, sv); 1288 av_push (av, sv);
726 1289
727 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
728 ? cbor_tagged_stash 1291 ? cbor_tagged_stash
729 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
730
731 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1293 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1294 }
1295
1296 PUTBACK;
1297
1298 FREETMPS; LEAVE;
1299 }
1300 break;
732 } 1301 }
1302
1303 return sv;
733 1304
734fail: 1305fail:
735 SvREFCNT_dec (sv); 1306 SvREFCNT_dec (sv);
736 return &PL_sv_undef; 1307 return &PL_sv_undef;
737} 1308}
739static SV * 1310static SV *
740decode_sv (dec_t *dec) 1311decode_sv (dec_t *dec)
741{ 1312{
742 WANT (1); 1313 WANT (1);
743 1314
744 switch (*dec->cur >> 5) 1315 switch (*dec->cur >> MAJOR_SHIFT)
745 { 1316 {
746 case 0: // unsigned int 1317 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
747 return newSVuv (decode_uint (dec)); 1318 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
748 case 1: // negative int 1319 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
749 return newSViv (-1 - (IV)decode_uint (dec)); 1320 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
750 case 2: // octet string 1321 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
751 return decode_str (dec, 0); 1322 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
752 case 3: // utf-8 string 1323 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
753 return decode_str (dec, 1); 1324
754 case 4: // array 1325 case MAJOR_MISC >> MAJOR_SHIFT:
755 return decode_av (dec);
756 case 5: // map
757 return decode_hv (dec);
758 case 6: // tag
759 return decode_tagged (dec);
760 case 7: // misc
761 switch (*dec->cur++ & 31) 1326 switch (*dec->cur++ & MINOR_MASK)
762 { 1327 {
763 case 20: 1328 case SIMPLE_FALSE:
764#if CBOR_SLOW 1329#if CBOR_SLOW
765 types_false = get_bool ("Types::Serialiser::false"); 1330 types_false = get_bool ("Types::Serialiser::false");
766#endif 1331#endif
767 return newSVsv (types_false); 1332 return newSVsv (types_false);
768 case 21: 1333 case SIMPLE_TRUE:
769#if CBOR_SLOW 1334#if CBOR_SLOW
770 types_true = get_bool ("Types::Serialiser::true"); 1335 types_true = get_bool ("Types::Serialiser::true");
771#endif 1336#endif
772 return newSVsv (types_true); 1337 return newSVsv (types_true);
773 case 22: 1338 case SIMPLE_NULL:
774 return newSVsv (&PL_sv_undef); 1339 return newSVsv (&PL_sv_undef);
775 case 23: 1340 case SIMPLE_UNDEF:
776#if CBOR_SLOW 1341#if CBOR_SLOW
777 types_error = get_bool ("Types::Serialiser::error"); 1342 types_error = get_bool ("Types::Serialiser::error");
778#endif 1343#endif
779 return newSVsv (types_error); 1344 return newSVsv (types_error);
780 1345
781 case 25: 1346 case MISC_FLOAT16:
782 { 1347 {
783 WANT (2); 1348 WANT (2);
784 1349
785 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1350 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
786 dec->cur += 2; 1351 dec->cur += 2;
787 1352
788 return newSVnv (ecb_binary16_to_float (fp)); 1353 return newSVnv (ecb_binary16_to_float (fp));
789 } 1354 }
790 1355
791 case 26: 1356 case MISC_FLOAT32:
792 { 1357 {
793 uint32_t fp; 1358 uint32_t fp;
794 WANT (4); 1359 WANT (4);
795 memcpy (&fp, dec->cur, 4); 1360 memcpy (&fp, dec->cur, 4);
796 dec->cur += 4; 1361 dec->cur += 4;
799 fp = ecb_bswap32 (fp); 1364 fp = ecb_bswap32 (fp);
800 1365
801 return newSVnv (ecb_binary32_to_float (fp)); 1366 return newSVnv (ecb_binary32_to_float (fp));
802 } 1367 }
803 1368
804 case 27: 1369 case MISC_FLOAT64:
805 { 1370 {
806 uint64_t fp; 1371 uint64_t fp;
807 WANT (8); 1372 WANT (8);
808 memcpy (&fp, dec->cur, 8); 1373 memcpy (&fp, dec->cur, 8);
809 dec->cur += 8; 1374 dec->cur += 8;
812 fp = ecb_bswap64 (fp); 1377 fp = ecb_bswap64 (fp);
813 1378
814 return newSVnv (ecb_binary64_to_double (fp)); 1379 return newSVnv (ecb_binary64_to_double (fp));
815 } 1380 }
816 1381
817 // 0..19 unassigned 1382 // 0..19 unassigned simple
818 // 24 reserved + unassigned (reserved values are not encodable) 1383 // 24 reserved + unassigned simple (reserved values are not encodable)
1384 // 28-30 unassigned misc
1385 // 31 break code
819 default: 1386 default:
820 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1387 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
821 } 1388 }
822 1389
823 break; 1390 break;
824 } 1391 }
825 1392
828} 1395}
829 1396
830static SV * 1397static SV *
831decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1398decode_cbor (SV *string, CBOR *cbor, char **offset_return)
832{ 1399{
833 dec_t dec; 1400 dec_t dec = { 0 };
834 SV *sv; 1401 SV *sv;
1402 STRLEN len;
1403 char *data = SvPVbyte (string, len);
835 1404
836 /* work around bugs in 5.10 where manipulating magic values
837 * makes perl ignore the magic in subsequent accesses.
838 * also make a copy of non-PV values, to get them into a clean
839 * state (SvPV should do that, but it's buggy, see below).
840 */
841 /*SvGETMAGIC (string);*/
842 if (SvMAGICAL (string) || !SvPOK (string))
843 string = sv_2mortal (newSVsv (string));
844
845 SvUPGRADE (string, SVt_PV);
846
847 /* work around a bug in perl 5.10, which causes SvCUR to fail an
848 * assertion with -DDEBUGGING, although SvCUR is documented to
849 * return the xpv_cur field which certainly exists after upgrading.
850 * according to nicholas clark, calling SvPOK fixes this.
851 * But it doesn't fix it, so try another workaround, call SvPV_nolen
852 * and hope for the best.
853 * Damnit, SvPV_nolen still trips over yet another assertion. This
854 * assertion business is seriously broken, try yet another workaround
855 * for the broken -DDEBUGGING.
856 */
857 {
858#ifdef DEBUGGING
859 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
860#else
861 STRLEN offset = SvCUR (string);
862#endif
863
864 if (offset > cbor->max_size && cbor->max_size) 1405 if (len > cbor->max_size && cbor->max_size)
865 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1406 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
866 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1407 (unsigned long)len, (unsigned long)cbor->max_size);
867 }
868
869 sv_utf8_downgrade (string, 0);
870 1408
871 dec.cbor = *cbor; 1409 dec.cbor = *cbor;
872 dec.cur = (U8 *)SvPVX (string); 1410 dec.cur = (U8 *)data;
873 dec.end = (U8 *)SvEND (string); 1411 dec.end = (U8 *)data + len;
874 dec.err = 0;
875 dec.depth = 0;
876 1412
877 sv = decode_sv (&dec); 1413 sv = decode_sv (&dec);
878 1414
879 if (offset_return) 1415 if (offset_return)
880 *offset_return = dec.cur; 1416 *offset_return = dec.cur;
883 if (dec.cur != dec.end && !dec.err) 1419 if (dec.cur != dec.end && !dec.err)
884 dec.err = "garbage after CBOR object"; 1420 dec.err = "garbage after CBOR object";
885 1421
886 if (dec.err) 1422 if (dec.err)
887 { 1423 {
1424 if (dec.shareable)
1425 {
1426 // need to break cyclic links, which would all be in shareable
1427 int i;
1428 SV **svp;
1429
1430 for (i = av_len (dec.shareable) + 1; i--; )
1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1432 sv_setsv (*svp, &PL_sv_undef);
1433 }
1434
888 SvREFCNT_dec (sv); 1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
889 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1440 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
890 } 1441 }
891 1442
892 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
893 1444
894 return sv; 1445 return sv;
895} 1446}
896 1447
1448/////////////////////////////////////////////////////////////////////////////
1449// incremental parser
1450
1451#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1452
1453// returns 0 for notyet, 1 for success or error
1454static int
1455incr_parse (CBOR *self, SV *cborstr)
1456{
1457 STRLEN cur;
1458 SvPV (cborstr, cur);
1459
1460 while (ecb_expect_true (self->incr_need <= cur))
1461 {
1462 // table of integer count bytes
1463 static I8 incr_len[MINOR_MASK + 1] = {
1464 0, 0, 0, 0, 0, 0, 0, 0,
1465 0, 0, 0, 0, 0, 0, 0, 0,
1466 0, 0, 0, 0, 0, 0, 0, 0,
1467 1, 2, 4, 8,-1,-1,-1,-2
1468 };
1469
1470 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1471 U8 m = *p & MINOR_MASK;
1472 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1473 I8 ilen = incr_len[m];
1474
1475 self->incr_need = self->incr_pos + 1;
1476
1477 if (ecb_expect_false (ilen < 0))
1478 {
1479 if (m != MINOR_INDEF)
1480 return 1; // error
1481
1482 if (*p == (MAJOR_MISC | MINOR_INDEF))
1483 {
1484 if (count >= 0)
1485 return 1; // error
1486
1487 count = 1;
1488 }
1489 else
1490 {
1491 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1492 count = -1;
1493 }
1494 }
1495 else
1496 {
1497 self->incr_need += ilen;
1498 if (ecb_expect_false (self->incr_need > cur))
1499 return 0;
1500
1501 int major = *p >> MAJOR_SHIFT;
1502
1503 switch (major)
1504 {
1505 case MAJOR_TAG >> MAJOR_SHIFT:
1506 ++count; // tags merely prefix another value
1507 break;
1508
1509 case MAJOR_BYTES >> MAJOR_SHIFT:
1510 case MAJOR_TEXT >> MAJOR_SHIFT:
1511 case MAJOR_ARRAY >> MAJOR_SHIFT:
1512 case MAJOR_MAP >> MAJOR_SHIFT:
1513 {
1514 UV len;
1515
1516 if (ecb_expect_false (ilen))
1517 {
1518 len = 0;
1519
1520 do {
1521 len = (len << 8) | *++p;
1522 } while (--ilen);
1523 }
1524 else
1525 len = m;
1526
1527 switch (major)
1528 {
1529 case MAJOR_BYTES >> MAJOR_SHIFT:
1530 case MAJOR_TEXT >> MAJOR_SHIFT:
1531 self->incr_need += len;
1532 if (ecb_expect_false (self->incr_need > cur))
1533 return 0;
1534
1535 break;
1536
1537 case MAJOR_MAP >> MAJOR_SHIFT:
1538 len <<= 1;
1539 case MAJOR_ARRAY >> MAJOR_SHIFT:
1540 if (len)
1541 {
1542 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1543 count = len + 1;
1544 }
1545 break;
1546 }
1547 }
1548 }
1549 }
1550
1551 self->incr_pos = self->incr_need;
1552
1553 if (count > 0)
1554 {
1555 while (!--count)
1556 {
1557 if (!AvFILLp (self->incr_count))
1558 return 1; // done
1559
1560 SvREFCNT_dec_NN (av_pop (self->incr_count));
1561 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1562 }
1563
1564 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1565 }
1566 }
1567
1568 return 0;
1569}
1570
1571
897///////////////////////////////////////////////////////////////////////////// 1572/////////////////////////////////////////////////////////////////////////////
898// XS interface functions 1573// XS interface functions
899 1574
900MODULE = CBOR::XS PACKAGE = CBOR::XS 1575MODULE = CBOR::XS PACKAGE = CBOR::XS
901 1576
909 1584
910 types_true = get_bool ("Types::Serialiser::true" ); 1585 types_true = get_bool ("Types::Serialiser::true" );
911 types_false = get_bool ("Types::Serialiser::false"); 1586 types_false = get_bool ("Types::Serialiser::false");
912 types_error = get_bool ("Types::Serialiser::error"); 1587 types_error = get_bool ("Types::Serialiser::error");
913 1588
1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1590
914 sv_cbor = newSVpv ("CBOR", 0); 1591 sv_cbor = newSVpv ("CBOR", 0);
915 SvREADONLY_on (sv_cbor); 1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
916} 1595}
917 1596
918PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
919 1598
920void CLONE (...) 1599void CLONE (...)
938 1617
939void shrink (CBOR *self, int enable = 1) 1618void shrink (CBOR *self, int enable = 1)
940 ALIAS: 1619 ALIAS:
941 shrink = F_SHRINK 1620 shrink = F_SHRINK
942 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1622 allow_sharing = F_ALLOW_SHARING
1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1627 text_strings = F_TEXT_STRINGS
1628 validate_utf8 = F_VALIDATE_UTF8
943 PPCODE: 1629 PPCODE:
944{ 1630{
945 if (enable) 1631 if (enable)
946 self->flags |= ix; 1632 self->flags |= ix;
947 else 1633 else
952 1638
953void get_shrink (CBOR *self) 1639void get_shrink (CBOR *self)
954 ALIAS: 1640 ALIAS:
955 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
956 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1643 get_allow_sharing = F_ALLOW_SHARING
1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1646 get_pack_strings = F_PACK_STRINGS
1647 get_text_keys = F_TEXT_KEYS
1648 get_text_strings = F_TEXT_STRINGS
1649 get_validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1650 PPCODE:
958 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
959 1652
960void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
961 PPCODE: 1654 PPCODE:
977 CODE: 1670 CODE:
978 RETVAL = self->max_size; 1671 RETVAL = self->max_size;
979 OUTPUT: 1672 OUTPUT:
980 RETVAL 1673 RETVAL
981 1674
982#if 0 //TODO 1675void filter (CBOR *self, SV *filter = 0)
983
984void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
985 PPCODE: 1676 PPCODE:
986{
987 SvREFCNT_dec (self->cb_object); 1677 SvREFCNT_dec (self->filter);
988 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1678 self->filter = filter ? newSVsv (filter) : filter;
989
990 XPUSHs (ST (0)); 1679 XPUSHs (ST (0));
991}
992 1680
993void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1681SV *get_filter (CBOR *self)
994 PPCODE: 1682 CODE:
995{ 1683 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
996 if (!self->cb_sk_object) 1684 OUTPUT:
997 self->cb_sk_object = newHV (); 1685 RETVAL
998
999 if (SvOK (cb))
1000 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1001 else
1002 {
1003 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1004
1005 if (!HvKEYS (self->cb_sk_object))
1006 {
1007 SvREFCNT_dec (self->cb_sk_object);
1008 self->cb_sk_object = 0;
1009 }
1010 }
1011
1012 XPUSHs (ST (0));
1013}
1014
1015#endif
1016 1686
1017void encode (CBOR *self, SV *scalar) 1687void encode (CBOR *self, SV *scalar)
1018 PPCODE: 1688 PPCODE:
1019 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1689 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1020 XPUSHs (scalar); 1690 XPUSHs (scalar);
1033 EXTEND (SP, 2); 1703 EXTEND (SP, 2);
1034 PUSHs (sv); 1704 PUSHs (sv);
1035 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1705 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1036} 1706}
1037 1707
1038#if 0 1708void incr_parse (CBOR *self, SV *cborstr)
1709 ALIAS:
1710 incr_parse_multiple = 1
1711 PPCODE:
1712{
1713 if (SvUTF8 (cborstr))
1714 sv_utf8_downgrade (cborstr, 0);
1715
1716 if (!self->incr_count)
1717 {
1718 self->incr_count = newAV ();
1719 self->incr_pos = 0;
1720 self->incr_need = 1;
1721
1722 av_push (self->incr_count, newSViv (1));
1723 }
1724
1725 do
1726 {
1727 if (!incr_parse (self, cborstr))
1728 {
1729 if (self->incr_need > self->max_size && self->max_size)
1730 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1731 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1732
1733 break;
1734 }
1735
1736 SV *sv;
1737 char *offset;
1738
1739 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1740 XPUSHs (sv);
1741
1742 sv_chop (cborstr, offset);
1743
1744 av_clear (self->incr_count);
1745 av_push (self->incr_count, newSViv (1));
1746
1747 self->incr_pos = 0;
1748 self->incr_need = self->incr_pos + 1;
1749 }
1750 while (ix);
1751}
1752
1753void incr_reset (CBOR *self)
1754 CODE:
1755{
1756 SvREFCNT_dec (self->incr_count);
1757 self->incr_count = 0;
1758}
1039 1759
1040void DESTROY (CBOR *self) 1760void DESTROY (CBOR *self)
1041 CODE: 1761 PPCODE:
1042 SvREFCNT_dec (self->cb_sk_object); 1762 cbor_free (self);
1043 SvREFCNT_dec (self->cb_object);
1044
1045#endif
1046 1763
1047PROTOTYPES: ENABLE 1764PROTOTYPES: ENABLE
1048 1765
1049void encode_cbor (SV *scalar) 1766void encode_cbor (SV *scalar)
1767 ALIAS:
1768 encode_cbor = 0
1769 encode_cbor_sharing = F_ALLOW_SHARING
1050 PPCODE: 1770 PPCODE:
1051{ 1771{
1052 CBOR cbor; 1772 CBOR cbor;
1053 cbor_init (&cbor); 1773 cbor_init (&cbor);
1774 cbor.flags |= ix;
1054 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1775 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1055 XPUSHs (scalar); 1776 XPUSHs (scalar);
1056} 1777}
1057 1778
1058void decode_cbor (SV *cborstr) 1779void decode_cbor (SV *cborstr)
1062 cbor_init (&cbor); 1783 cbor_init (&cbor);
1063 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1064 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1065} 1786}
1066 1787
1788#ifdef __AFL_COMPILER
1789
1790void
1791afl_init ()
1792 CODE:
1793 __AFL_INIT ();
1794
1795int
1796afl_loop (unsigned int count = 10000)
1797 CODE:
1798 RETVAL = __AFL_LOOP (count);
1799 OUTPUT:
1800 RETVAL
1801
1802#endif
1803

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines