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 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.70 by root, Fri Dec 4 02:57:14 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"
13 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};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
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 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
14#define F_SHRINK 0x00000200UL 118#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
120#define F_ALLOW_SHARING 0x00000004UL
121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
123#define F_PACK_STRINGS 0x00000020UL
124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
16 127
17#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
18 129
19#define SB do { 130#define SB do {
20#define SE } while (0) 131#define SE } while (0)
31#else 142#else
32# define CBOR_SLOW 0 143# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
34#endif 145#endif
35 146
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 147static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 149
39typedef struct { 150typedef struct {
40 U32 flags; 151 U32 flags;
41 U32 max_depth; 152 U32 max_depth;
42 STRLEN max_size; 153 STRLEN max_size;
154 SV *filter;
43 155
44 SV *cb_object; 156 // for the incremental parser
45 HV *cb_sk_object; 157 STRLEN incr_pos; // the current offset into the text
158 STRLEN incr_need; // minimum bytes needed to decode
159 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
46} CBOR; 160} CBOR;
47 161
48ecb_inline void 162ecb_inline void
49cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
50{ 164{
51 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 166 cbor->max_depth = 512;
167}
168
169ecb_inline void
170cbor_free (CBOR *cbor)
171{
172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
53} 174}
54 175
55///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
56// utility functions 177// utility functions
57 178
79 SvPV_renew (sv, SvCUR (sv) + 1); 200 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 201#endif
81 } 202 }
82} 203}
83 204
84///////////////////////////////////////////////////////////////////////////// 205// minimum length of a string to be registered for stringref
85// fp hell 206ecb_inline int
86 207minimum_string_length (UV idx)
87//TODO 208{
209 return idx <= 23 ? 3
210 : idx <= 0xffU ? 4
211 : idx <= 0xffffU ? 5
212 : idx <= 0xffffffffU ? 7
213 : 11;
214}
88 215
89///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
90// encoder 217// encoder
91 218
92// structure used for encoding CBOR 219// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 222 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 223 char *end; // SvEND (sv)
97 SV *sv; // result scalar 224 SV *sv; // result scalar
98 CBOR cbor; 225 CBOR cbor;
99 U32 depth; // recursion level 226 U32 depth; // recursion level
227 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
228 UV stringref_idx;
229 HV *shareable; // ptr => index, or 0
230 UV shareable_idx;
100} enc_t; 231} enc_t;
101 232
102ecb_inline void 233ecb_inline void
103need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
104{ 235{
105 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
106 { 237 {
107 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
108 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
109 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
110 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
111 } 242 }
112} 243}
113 244
245static void encode_sv (enc_t *enc, SV *sv);
246
114ecb_inline void 247ecb_inline void
115encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
116{ 249{
117 need (enc, 1); 250 need (enc, 1);
118 *enc->cur++ = ch; 251 *enc->cur++ = ch;
119} 252}
120 253
254// used for tags, intregers, element counts and so on
121static void 255static void
122encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
123{ 257{
124 need (enc, 9); 258 need (enc, 9);
125 259
126 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
128 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
129 { 263 {
130 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 265 *enc->cur++ = len;
132 } 266 }
133 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
134 { 268 {
135 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 271 *enc->cur++ = len;
138 } 272 }
139 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
140 { 274 {
141 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 279 *enc->cur++ = len;
146 } 280 }
147 else 281 else
148 { 282 {
149 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 291 *enc->cur++ = len;
158 } 292 }
159} 293}
160 294
161static void 295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
307ecb_inline void
308encode_tag (enc_t *enc, UV tag)
309{
310 encode_uint (enc, MAJOR_TAG, tag);
311}
312
313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 316{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
336ecb_inline void
337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
165 need (enc, len); 347 need (enc, len);
166 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
167 enc->cur += len; 349 enc->cur += len;
168} 350}
169 351
170static void encode_sv (enc_t *enc, SV *sv); 352ecb_inline void
353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
354{
355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
356 {
357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
358
359 if (SvOK (*svp))
360 {
361 // already registered, use stringref
362 encode_tag (enc, CBOR_TAG_STRINGREF);
363 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
364 return;
365 }
366 else if (len >= minimum_string_length (enc->stringref_idx))
367 {
368 // register only
369 sv_setuv (*svp, enc->stringref_idx);
370 ++enc->stringref_idx;
371 }
372 }
373
374 encode_str (enc, upgrade_utf8, utf8, str, len);
375}
376
377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
171 507
172static void 508static void
173encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
174{ 510{
175 int i, len = av_len (av); 511 int i, len = av_len (av);
177 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
179 515
180 ++enc->depth; 516 ++enc->depth;
181 517
182 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 519
520 if (ecb_expect_false (SvMAGICAL (av)))
184 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
185 { 522 {
186 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
188 } 525 }
526 else
527 for (i = 0; i <= len; ++i)
528 {
529 SV *sv = AvARRAY (av)[i];
530 encode_sv (enc, sv ? sv : &PL_sv_undef);
531 }
189 532
190 --enc->depth; 533 --enc->depth;
191} 534}
192 535
193static void 536static void
201 ++enc->depth; 544 ++enc->depth;
202 545
203 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
205 548
206 if (mg) 549 if (ecb_expect_false (mg))
207 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 551 else
209 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
210 553
211 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
212 { 555 {
213 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
215 else 558 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217 560
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 562 }
220 563
221 if (mg) 564 if (ecb_expect_false (mg))
222 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 566
224 --enc->depth; 567 --enc->depth;
225} 568}
226 569
227// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
228static void 571static void
229encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
230{ 573{
231 svtype svt;
232
233 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
575
234 svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
235 577
236 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
237 { 579 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 580 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 581 ? types_boolean_stash
582 : gv_stashpv ("Types::Serialiser::Boolean", 1);
583 HV *error_stash = !CBOR_SLOW || types_error_stash
584 ? types_error_stash
585 : gv_stashpv ("Types::Serialiser::Error", 1);
586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
587 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 589
242 if (SvSTASH (sv) == stash) 590 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 591
592 if (stash == boolean_stash)
593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
597 else if (stash == error_stash)
598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
602 else if (stash == tagged_stash)
603 {
604 if (svt != SVt_PVAV)
605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
623
624 return;
625 }
626 }
627
628 if (ecb_expect_false (SvREFCNT (sv) > 1)
629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
630 {
631 if (ecb_expect_false (!enc->shareable))
632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
633
634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
635
636 if (SvOK (*svp))
637 {
638 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
639 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
640 return;
641 }
244 else 642 else
245 { 643 {
246#if 0 //TODO 644 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 645 ++enc->shareable_idx;
646 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
647 }
648 }
649
650 if (ecb_expect_false (SvOBJECT (sv)))
651 {
652 HV *stash = SvSTASH (sv);
653 GV *method;
654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 659 {
660 dSP;
661
662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
249 // we re-bless the reference to get overload and other niceties right 664 // we re-bless the reference to get overload and other niceties right
250 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252 if (to_cbor)
253 {
254 dSP;
255
256 ENTER; SAVETMPS; PUSHMARK (SP);
257 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 666
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 667 PUTBACK;
668 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 669 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 670 SPAGAIN;
263 671
264 // catch this surprisingly common error 672 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 673 if (SvROK (TOPs) && SvRV (TOPs) == sv)
266 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 674 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 675
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 676 encode_sv (enc, POPs);
272 677
678 PUTBACK;
679
273 FREETMPS; LEAVE; 680 FREETMPS; LEAVE;
274 } 681 }
275 else if (enc->cbor.flags & F_ALLOW_BLESSED) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
276 encode_str (enc, "null", 4, 0); 683 {
277 else 684 dSP;
278 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 685
279 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
688 EXTEND (SP, 2);
689 // we re-bless the reference to get overload and other niceties right
690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
691 PUSHs (sv_cbor);
692
693 PUTBACK;
694 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
695 SPAGAIN;
696
697 // catch this surprisingly common error
698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
700
701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
702 encode_uint (enc, MAJOR_ARRAY, count + 1);
703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
704
705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
280 } 712 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 713
282 encode_str (enc, "null", 4, 0); 714 PUTBACK;
715
716 FREETMPS; LEAVE;
717 }
283 else 718 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 719 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 720 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 721 }
289 else if (svt == SVt_PVHV) 722 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 723 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 724 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 725 encode_av (enc, (AV *)sv);
293 else if (svt < SVt_PVAV)
294 {
295 STRLEN len = 0;
296 char *pv = svt ? SvPV (sv, len) : 0;
297
298 if (len == 1 && *pv == '1')
299 encode_ch (enc, 0xe0 | 21);
300 else if (len == 1 && *pv == '0')
301 encode_ch (enc, 0xe0 | 20);
302 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303 encode_ch (enc, 0xe0 | 23);
304 else
305 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309 encode_ch (enc, 0xe0 | 23);
310 else 726 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 727 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 728 encode_tag (enc, CBOR_TAG_INDIRECTION);
729 encode_sv (enc, sv);
730 }
313} 731}
314 732
315static void 733static void
316encode_nv (enc_t *enc, SV *sv) 734encode_nv (enc_t *enc, SV *sv)
317{ 735{
318 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
319 737
320 need (enc, 9); 738 need (enc, 9);
321 739
322 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 742 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
326 { 744 encode_float32 (enc, nv);
327 uint32_t fp = ecb_float_to_binary32 (nv);
328
329 *enc->cur++ = 0xe0 | 26;
330
331 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp);
333
334 memcpy (enc->cur, &fp, 4);
335 enc->cur += 4;
336 }
337 else 745 else
338 { 746 encode_float64 (enc, nv);
339 uint64_t fp = ecb_double_to_binary64 (nv);
340
341 *enc->cur++ = 0xe0 | 27;
342
343 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp);
345
346 memcpy (enc->cur, &fp, 8);
347 enc->cur += 8;
348 }
349} 747}
350 748
351static void 749static void
352encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
353{ 751{
355 753
356 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
357 { 755 {
358 STRLEN len; 756 STRLEN len;
359 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
361 } 759 }
362 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 761 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
365 { 763 encode_int (enc, sv);
366 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv));
368 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv));
370 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
372 }
373 else if (SvROK (sv)) 764 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 770 else
380 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 771 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 773}
383 774
384static SV * 775static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
386{ 777{
387 enc_t enc; 778 enc_t enc = { 0 };
388 779
389 enc.cbor = *cbor; 780 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 784
395 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
786
787 if (cbor->flags & F_PACK_STRINGS)
788 {
789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
792 }
793
396 encode_sv (&enc, scalar); 794 encode_sv (&enc, scalar);
397 795
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 796 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 797 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 798
414 U8 *end; // end of input string 812 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 813 const char *err; // parse error, if != 0
416 CBOR cbor; 814 CBOR cbor;
417 U32 depth; // recursion depth 815 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
817 AV *shareable;
818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
419} dec_t; 821} dec_t;
420 822
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
422 830
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
424 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
425#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
426#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
427 861
428static UV 862static UV
429decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
430{ 864{
431 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
432 { 866 ++dec->cur;
433 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436 return *dec->cur++ & 31;
437 867
438 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
439 WANT (2); 872 WANT (1);
440 dec->cur += 2; 873 dec->cur += 1;
441 return dec->cur[-1]; 874 return dec->cur[-1];
442 875 }
443 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
444 WANT (3); 878 WANT (2);
445 dec->cur += 3; 879 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
448 882 }
449 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
450 WANT (5); 885 WANT (4);
451 dec->cur += 5; 886 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
456 891 }
457 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
458 WANT (9); 894 WANT (8);
459 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
460 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
464 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
468 910 }
469 default: 911 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 913
473fail: 914fail:
474 return 0; 915 return 0;
475} 916}
476 917
481{ 922{
482 AV *av = newAV (); 923 AV *av = newAV ();
483 924
484 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
485 926
486 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 928 {
488 ++dec->cur; 929 ++dec->cur;
489 930
490 for (;;) 931 for (;;)
491 { 932 {
492 WANT (1); 933 WANT (1);
493 934
494 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
495 { 936 {
496 ++dec->cur; 937 ++dec->cur;
497 break; 938 break;
498 } 939 }
499 940
500 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
501 } 942 }
502 } 943 }
503 else 944 else
504 { 945 {
505 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
506 947
948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
507 av_fill (av, len - 1); 949 av_fill (av, len - 1);
508 950
509 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
511 } 953 }
512 954
513 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
514 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
515 957
516fail: 958fail:
517 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
518 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 961 return &PL_sv_undef;
520} 962}
521 963
964static void
965decode_he (dec_t *dec, HV *hv)
966{
967 // for speed reasons, we specialcase single-string
968 // byte or utf-8 strings as keys, but only when !stringref
969
970 if (ecb_expect_true (!dec->stringref))
971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
972 {
973 STRLEN len = decode_uint (dec);
974 char *key = (char *)dec->cur;
975
976 WANT (len);
977 dec->cur += len;
978
979 hv_store (hv, key, len, decode_sv (dec), 0);
980
981 return;
982 }
983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
984 {
985 STRLEN len = decode_uint (dec);
986 char *key = (char *)dec->cur;
987
988 WANT (len);
989 dec->cur += len;
990
991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
992 if (!is_utf8_string (key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
994
995 hv_store (hv, key, -len, decode_sv (dec), 0);
996
997 return;
998 }
999
1000 SV *k = decode_sv (dec);
1001 SV *v = decode_sv (dec);
1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
1031 hv_store_ent (hv, k, v, 0);
1032 SvREFCNT_dec_NN (k);
1033
1034fail:
1035 ;
1036}
1037
522static SV * 1038static SV *
523decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
524{ 1040{
525 HV *hv = newHV (); 1041 HV *hv = newHV ();
526 1042
527 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
528 1044
529 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 1046 {
531 ++dec->cur; 1047 ++dec->cur;
532 1048
533 for (;;) 1049 for (;;)
534 { 1050 {
535 WANT (1); 1051 WANT (1);
536 1052
537 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
538 { 1054 {
539 ++dec->cur; 1055 ++dec->cur;
540 break; 1056 break;
541 } 1057 }
542 1058
543 SV *k = decode_sv (dec); 1059 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 1060 }
548 } 1061 }
549 else 1062 else
550 { 1063 {
551 int len = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
552 1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1067
553 while (len--) 1068 while (pairs--)
554 { 1069 decode_he (dec, hv);
555 SV *k = decode_sv (dec);
556 SV *v = decode_sv (dec);
557
558 hv_store_ent (hv, k, v, 0);
559 }
560 } 1070 }
561 1071
562 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
564 1074
565#if 0 1075fail:
566 SV *sv; 1076 SvREFCNT_dec_NN (hv);
567 HV *hv = newHV ();
568
569 DEC_INC_DEPTH; 1077 DEC_DEC_DEPTH;
570 decode_ws (dec); 1078 return &PL_sv_undef;
1079}
571 1080
572 for (;;) 1081static SV *
573 { 1082decode_str (dec_t *dec, int utf8)
574 // heuristic: assume that 1083{
575 // a) decode_str + hv_store_ent are abysmally slow. 1084 SV *sv = 0;
576 // b) most hash keys are short, simple ascii text. 1085
577 // => try to "fast-match" such strings to avoid 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
578 // the overhead of decode_str + hv_store_ent.
579 { 1087 {
580 SV *value; 1088 // indefinite length strings
581 char *p = dec->cur; 1089 ++dec->cur;
582 char *e = p + 24; // only try up to 24 bytes
583 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
1093 sv = newSVpvn ("", 0);
1094
584 for (;;) 1095 for (;;)
585 { 1096 {
586 // the >= 0x80 is false on most architectures 1097 WANT (1);
587 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1098
1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
588 { 1101 {
589 // slow path, back up and use decode_str 1102 ++dec->cur;
590 SV *key = decode_str (dec);
591 if (!key)
592 goto fail;
593
594 decode_ws (dec); EXPECT_CH (':');
595
596 decode_ws (dec);
597 value = decode_sv (dec);
598 if (!value)
599 {
600 SvREFCNT_dec (key);
601 goto fail;
602 }
603
604 hv_store_ent (hv, key, value, 0);
605 SvREFCNT_dec (key);
606
607 break; 1103 break;
608 } 1104 }
609 else if (*p == '"')
610 {
611 // fast path, got a simple key
612 char *key = dec->cur;
613 int len = p - key;
614 dec->cur = p + 1;
615
616 decode_ws (dec); EXPECT_CH (':');
617
618 decode_ws (dec);
619 value = decode_sv (dec);
620 if (!value)
621 goto fail;
622
623 hv_store (hv, key, len, value, 0);
624
625 break;
626 }
627
628 ++p;
629 }
630 }
631
632 decode_ws (dec);
633
634 if (*dec->cur == '}')
635 {
636 ++dec->cur;
637 break;
638 }
639
640 if (*dec->cur != ',')
641 ERR (", or } expected while parsing object/hash");
642
643 ++dec->cur;
644
645 decode_ws (dec);
646
647 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648 {
649 ++dec->cur;
650 break;
651 }
652 }
653
654 DEC_DEC_DEPTH;
655 sv = newRV_noinc ((SV *)hv);
656
657 // check filter callbacks
658 if (dec->cbor.flags & F_HOOK)
659 {
660 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661 {
662 HE *cb, *he;
663
664 hv_iterinit (hv);
665 he = hv_iternext (hv);
666 hv_iterinit (hv);
667
668 // the next line creates a mortal sv each time its called.
669 // might want to optimise this for common cases.
670 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672 if (cb)
673 { 1105 else
674 dSP; 1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
675 int count;
676 1107
677 ENTER; SAVETMPS; PUSHMARK (SP); 1108 STRLEN len = decode_uint (dec);
678 XPUSHs (HeVAL (he));
679 sv_2mortal (sv);
680 1109
681 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683 if (count == 1)
684 {
685 sv = newSVsv (POPs);
686 FREETMPS; LEAVE;
687 return sv;
688 }
689
690 SvREFCNT_inc (sv);
691 FREETMPS; LEAVE;
692 }
693 }
694
695 if (dec->cbor.cb_object)
696 {
697 dSP;
698 int count;
699
700 ENTER; SAVETMPS; PUSHMARK (SP);
701 XPUSHs (sv_2mortal (sv));
702
703 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705 if (count == 1)
706 {
707 sv = newSVsv (POPs);
708 FREETMPS; LEAVE;
709 return sv;
710 }
711
712 SvREFCNT_inc (sv);
713 FREETMPS; LEAVE;
714 }
715 }
716
717 return sv;
718#endif
719
720fail:
721 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH;
723 return &PL_sv_undef;
724}
725
726static SV *
727decode_str (dec_t *dec, int utf8)
728{
729 SV *sv;
730
731 if ((*dec->cur & 31) == 31)
732 {
733 ++dec->cur;
734
735 sv = newSVpvn ("", 0);
736
737 // not very fast, and certainly not robust against illegal input
738 for (;;)
739 {
740 WANT (1); 1110 WANT (len);
741 1111 sv_catpvn (sv, dec->cur, len);
742 if (*dec->cur == (0xe0 | 31)) 1112 dec->cur += len;
743 {
744 ++dec->cur;
745 break;
746 }
747
748 SV *sv2 = decode_sv (dec);
749 sv_catsv (sv, sv2);
750 } 1113 }
751 } 1114 }
752 else 1115 else
753 { 1116 {
754 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
755 1118
756 WANT (len); 1119 WANT (len);
757 sv = newSVpvn (dec->cur, len); 1120 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 1121 dec->cur += len;
1122
1123 if (ecb_expect_false (dec->stringref)
1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 1126 }
760 1127
761 if (utf8) 1128 if (utf8)
1129 {
1130 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1131 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1132 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1133
762 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
763 1136
764 return sv; 1137 return sv;
765 1138
766fail: 1139fail:
1140 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 1141 return &PL_sv_undef;
768} 1142}
769 1143
770static SV * 1144static SV *
771decode_tagged (dec_t *dec) 1145decode_tagged (dec_t *dec)
772{ 1146{
1147 SV *sv = 0;
773 UV tag = decode_uint (dec); 1148 UV tag = decode_uint (dec);
1149
1150 WANT (1);
1151
1152 switch (tag)
1153 {
1154 case CBOR_TAG_MAGIC:
774 SV *sv = decode_sv (dec); 1155 sv = decode_sv (dec);
1156 break;
775 1157
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1158 case CBOR_TAG_INDIRECTION:
1159 sv = newRV_noinc (decode_sv (dec));
1160 break;
1161
1162 case CBOR_TAG_STRINGREF_NAMESPACE:
1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
1166 ENTER;
1167
1168 SAVESPTR (dec->stringref);
1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1170
1171 sv = decode_sv (dec);
1172
1173 LEAVE;
1174 }
1175 break;
1176
1177 case CBOR_TAG_STRINGREF:
1178 {
1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1181
1182 UV idx = decode_uint (dec);
1183
1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1186
1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1188 }
1189 break;
1190
1191 case CBOR_TAG_VALUE_SHAREABLE:
1192 {
1193 if (ecb_expect_false (!dec->shareable))
1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
1198 sv = newSV (0);
1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1200
1201 SV *osv = decode_sv (dec);
1202 sv_setsv (sv, osv);
1203 SvREFCNT_dec_NN (osv);
1204 }
1205 else
1206 {
1207 av_push (dec->shareable, &PL_sv_undef);
1208 int idx = AvFILLp (dec->shareable);
1209 sv = decode_sv (dec);
1210 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1211 }
1212 }
1213 break;
1214
1215 case CBOR_TAG_VALUE_SHAREDREF:
1216 {
1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1219
1220 UV idx = decode_uint (dec);
1221
1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1224
1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1226
1227 if (sv == &PL_sv_undef)
1228 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1229 }
1230 break;
1231
1232 case CBOR_TAG_PERL_OBJECT:
1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
1237 sv = decode_sv (dec);
1238
1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1240 ERR ("corrupted CBOR data (non-array perl object)");
1241
1242 AV *av = (AV *)SvRV (sv);
1243 int len = av_len (av) + 1;
1244 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1245
1246 if (!stash)
1247 ERR ("cannot decode perl-object (package does not exist)");
1248
1249 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1250
1251 if (!method)
1252 ERR ("cannot decode perl-object (package does not have a THAW method)");
1253
1254 dSP;
1255
1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
1258 EXTEND (SP, len + 1);
1259 // we re-bless the reference to get overload and other niceties right
1260 PUSHs (*av_fetch (av, 0, 1));
1261 PUSHs (sv_cbor);
1262
1263 int i;
1264
1265 for (i = 1; i < len; ++i)
1266 PUSHs (*av_fetch (av, i, 1));
1267
1268 PUTBACK;
1269 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1270 SPAGAIN;
1271
1272 if (SvTRUE (ERRSV))
1273 {
1274 FREETMPS; LEAVE;
1275 ERR_ERRSV;
1276 }
1277
1278 SvREFCNT_dec_NN (sv);
1279 sv = SvREFCNT_inc (POPs);
1280
1281 PUTBACK;
1282
1283 FREETMPS; LEAVE;
1284 }
1285 break;
1286
1287 default:
1288 filter:
1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1292 sv = decode_sv (dec);
1293
1294 dSP;
1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1297 EXTEND (SP, 2);
1298 PUSHs (tag_sv);
1299 PUSHs (sv);
1300
1301 PUTBACK;
1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1303 SPAGAIN;
1304
1305 if (SvTRUE (ERRSV))
1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1308 FREETMPS; LEAVE;
1309 ERR_ERRSV;
1310 }
1311
1312 if (count)
1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1315 SvREFCNT_dec_NN (sv);
1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1318 }
1319 else
1320 {
1321 AV *av = newAV ();
1322 av_push (av, tag_sv);
1323 av_push (av, sv);
1324
1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1326 ? cbor_tagged_stash
1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1328 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1329 }
1330
1331 PUTBACK;
1332
1333 FREETMPS; LEAVE;
1334 }
1335 break;
1336 }
1337
777 return sv; 1338 return sv;
778 1339
779 AV *av = newAV (); 1340fail:
780 av_push (av, newSVuv (tag)); 1341 SvREFCNT_dec (sv);
781 av_push (av, sv); 1342 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1343}
784 1344
785static SV * 1345static SV *
786decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
787{ 1347{
788 WANT (1); 1348 WANT (1);
789 1349
790 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1351 {
792 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1359
800 return decode_str (dec, 1); 1360 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 return decode_tagged (dec);
807 case 7: // misc
808 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
809 { 1362 {
810 case 20: 1363 case SIMPLE_FALSE:
811#if CBOR_SLOW 1364#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
813#endif 1366#endif
814 return newSVsv (cbor_false); 1367 return newSVsv (types_false);
815 case 21: 1368 case SIMPLE_TRUE:
816#if CBOR_SLOW 1369#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
818#endif 1371#endif
819 return newSVsv (cbor_true); 1372 return newSVsv (types_true);
820 case 22: 1373 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
1375 case SIMPLE_UNDEF:
1376#if CBOR_SLOW
1377 types_error = get_bool ("Types::Serialiser::error");
1378#endif
1379 return newSVsv (types_error);
822 1380
823 case 25: 1381 case MISC_FLOAT16:
824 { 1382 {
825 WANT (2); 1383 WANT (2);
826 1384
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1386 dec->cur += 2;
829 1387
830 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
831 } 1389 }
832 1390
833 case 26: 1391 case MISC_FLOAT32:
834 { 1392 {
835 uint32_t fp; 1393 uint32_t fp;
836 WANT (4); 1394 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1396 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
842 1400
843 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
844 } 1402 }
845 1403
846 case 27: 1404 case MISC_FLOAT64:
847 { 1405 {
848 uint64_t fp; 1406 uint64_t fp;
849 WANT (8); 1407 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1409 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
855 1413
856 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
857 } 1415 }
858 1416
859 // 0..19 unassigned 1417 // 0..19 unassigned simple
860 // 24 reserved + unassigned (reserved values are not encodable) 1418 // 24 reserved + unassigned simple (reserved values are not encodable)
1419 // 28-30 unassigned misc
1420 // 31 break code
861 default: 1421 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
863 } 1423 }
864 1424
865 break; 1425 break;
866 } 1426 }
867#if 0
868 switch (*dec->cur)
869 {
870 //case '"': ++dec->cur; return decode_str (dec);
871 case '[': ++dec->cur; return decode_av (dec);
872 case '{': ++dec->cur; return decode_hv (dec);
873
874 case '-':
875 case '0': case '1': case '2': case '3': case '4':
876 case '5': case '6': case '7': case '8': case '9':
877 //TODO return decode_num (dec);
878
879 case 't':
880 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881 {
882 dec->cur += 4;
883#if CBOR_SLOW
884 cbor_true = get_bool ("CBOR::XS::true");
885#endif
886 return newSVsv (cbor_true);
887 }
888 else
889 ERR ("'true' expected");
890
891 break;
892
893 case 'f':
894 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895 {
896 dec->cur += 5;
897#if CBOR_SLOW
898 cbor_false = get_bool ("CBOR::XS::false");
899#endif
900 return newSVsv (cbor_false);
901 }
902 else
903 ERR ("'false' expected");
904
905 break;
906
907 case 'n':
908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909 {
910 dec->cur += 4;
911 return newSVsv (&PL_sv_undef);
912 }
913 else
914 ERR ("'null' expected");
915
916 break;
917
918 default:
919 ERR ("malformed CBOR string, neither array, object, number, string or atom");
920 break;
921 }
922#endif
923 1427
924fail: 1428fail:
925 return &PL_sv_undef; 1429 return &PL_sv_undef;
926} 1430}
927 1431
928static SV * 1432static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1434{
931 dec_t dec; 1435 dec_t dec = { 0 };
932 SV *sv; 1436 SV *sv;
1437 STRLEN len;
1438 char *data = SvPVbyte (string, len);
933 1439
934 /* work around bugs in 5.10 where manipulating magic values
935 * makes perl ignore the magic in subsequent accesses.
936 * also make a copy of non-PV values, to get them into a clean
937 * state (SvPV should do that, but it's buggy, see below).
938 */
939 /*SvGETMAGIC (string);*/
940 if (SvMAGICAL (string) || !SvPOK (string))
941 string = sv_2mortal (newSVsv (string));
942
943 SvUPGRADE (string, SVt_PV);
944
945 /* work around a bug in perl 5.10, which causes SvCUR to fail an
946 * assertion with -DDEBUGGING, although SvCUR is documented to
947 * return the xpv_cur field which certainly exists after upgrading.
948 * according to nicholas clark, calling SvPOK fixes this.
949 * But it doesn't fix it, so try another workaround, call SvPV_nolen
950 * and hope for the best.
951 * Damnit, SvPV_nolen still trips over yet another assertion. This
952 * assertion business is seriously broken, try yet another workaround
953 * for the broken -DDEBUGGING.
954 */
955 {
956#ifdef DEBUGGING
957 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958#else
959 STRLEN offset = SvCUR (string);
960#endif
961
962 if (offset > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
963 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1441 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1442 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1443
969 dec.cbor = *cbor; 1444 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1445 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1446 dec.end = (U8 *)data + len;
972 dec.err = 0;
973 dec.depth = 0;
974
975 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976 ;//TODO dec.cbor.flags |= F_HOOK;
977 1447
978 sv = decode_sv (&dec); 1448 sv = decode_sv (&dec);
979 1449
980 if (offset_return) 1450 if (offset_return)
981 *offset_return = dec.cur; 1451 *offset_return = dec.cur;
984 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
985 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
986 1456
987 if (dec.err) 1457 if (dec.err)
988 { 1458 {
1459 if (dec.shareable)
1460 {
1461 // need to break cyclic links, which would all be in shareable
1462 int i;
1463 SV **svp;
1464
1465 for (i = av_len (dec.shareable) + 1; i--; )
1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1467 sv_setsv (*svp, &PL_sv_undef);
1468 }
1469
989 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1476 }
992 1477
993 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
994 1479
995 return sv; 1480 return sv;
996} 1481}
997 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 case MAJOR_ARRAY >> MAJOR_SHIFT:
1575 if (len)
1576 {
1577 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1578 count = len + 1;
1579 }
1580 break;
1581 }
1582 }
1583 }
1584 }
1585
1586 self->incr_pos = self->incr_need;
1587
1588 if (count > 0)
1589 {
1590 while (!--count)
1591 {
1592 if (!AvFILLp (self->incr_count))
1593 return 1; // done
1594
1595 SvREFCNT_dec_NN (av_pop (self->incr_count));
1596 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1597 }
1598
1599 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1600 }
1601 }
1602
1603 return 0;
1604}
1605
1606
998///////////////////////////////////////////////////////////////////////////// 1607/////////////////////////////////////////////////////////////////////////////
999// XS interface functions 1608// XS interface functions
1000 1609
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1610MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1611
1003BOOT: 1612BOOT:
1004{ 1613{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1614 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1615 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1616
1008 cbor_true = get_bool ("CBOR::XS::true"); 1617 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1618 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1619
1620 types_true = get_bool ("Types::Serialiser::true" );
1621 types_false = get_bool ("Types::Serialiser::false");
1622 types_error = get_bool ("Types::Serialiser::error");
1623
1624 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1625
1626 sv_cbor = newSVpv ("CBOR", 0);
1627 SvREADONLY_on (sv_cbor);
1628
1629 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1010} 1630}
1011 1631
1012PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1013 1633
1014void CLONE (...) 1634void CLONE (...)
1015 CODE: 1635 CODE:
1016 cbor_stash = 0; 1636 cbor_stash = 0;
1637 cbor_tagged_stash = 0;
1638 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1639 types_boolean_stash = 0;
1018 1640
1019void new (char *klass) 1641void new (char *klass)
1020 PPCODE: 1642 PPCODE:
1021{ 1643{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1644 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1652
1031void shrink (CBOR *self, int enable = 1) 1653void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1654 ALIAS:
1033 shrink = F_SHRINK 1655 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1656 allow_unknown = F_ALLOW_UNKNOWN
1657 allow_sharing = F_ALLOW_SHARING
1658 allow_cycles = F_ALLOW_CYCLES
1659 forbid_objects = F_FORBID_OBJECTS
1660 pack_strings = F_PACK_STRINGS
1661 text_keys = F_TEXT_KEYS
1662 text_strings = F_TEXT_STRINGS
1663 validate_utf8 = F_VALIDATE_UTF8
1035 PPCODE: 1664 PPCODE:
1036{ 1665{
1037 if (enable) 1666 if (enable)
1038 self->flags |= ix; 1667 self->flags |= ix;
1039 else 1668 else
1044 1673
1045void get_shrink (CBOR *self) 1674void get_shrink (CBOR *self)
1046 ALIAS: 1675 ALIAS:
1047 get_shrink = F_SHRINK 1676 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1677 get_allow_unknown = F_ALLOW_UNKNOWN
1678 get_allow_sharing = F_ALLOW_SHARING
1679 get_allow_cycles = F_ALLOW_CYCLES
1680 get_forbid_objects = F_FORBID_OBJECTS
1681 get_pack_strings = F_PACK_STRINGS
1682 get_text_keys = F_TEXT_KEYS
1683 get_text_strings = F_TEXT_STRINGS
1684 get_validate_utf8 = F_VALIDATE_UTF8
1049 PPCODE: 1685 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1686 XPUSHs (boolSV (self->flags & ix));
1051 1687
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1688void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1689 PPCODE:
1069 CODE: 1705 CODE:
1070 RETVAL = self->max_size; 1706 RETVAL = self->max_size;
1071 OUTPUT: 1707 OUTPUT:
1072 RETVAL 1708 RETVAL
1073 1709
1074#if 0 //TODO 1710void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1711 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1712 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1713 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1714 XPUSHs (ST (0));
1083}
1084 1715
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1716SV *get_filter (CBOR *self)
1086 PPCODE: 1717 CODE:
1087{ 1718 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1719 OUTPUT:
1089 self->cb_sk_object = newHV (); 1720 RETVAL
1090
1091 if (SvOK (cb))
1092 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093 else
1094 {
1095 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097 if (!HvKEYS (self->cb_sk_object))
1098 {
1099 SvREFCNT_dec (self->cb_sk_object);
1100 self->cb_sk_object = 0;
1101 }
1102 }
1103
1104 XPUSHs (ST (0));
1105}
1106
1107#endif
1108 1721
1109void encode (CBOR *self, SV *scalar) 1722void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1723 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1724 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1725 XPUSHs (scalar);
1125 EXTEND (SP, 2); 1738 EXTEND (SP, 2);
1126 PUSHs (sv); 1739 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1740 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1741}
1129 1742
1743void incr_parse (CBOR *self, SV *cborstr)
1744 ALIAS:
1745 incr_parse_multiple = 1
1746 PPCODE:
1747{
1748 if (SvUTF8 (cborstr))
1749 sv_utf8_downgrade (cborstr, 0);
1750
1751 if (!self->incr_count)
1752 {
1753 self->incr_count = newAV ();
1754 self->incr_pos = 0;
1755 self->incr_need = 1;
1756
1757 av_push (self->incr_count, newSViv (1));
1758 }
1759
1760 do
1761 {
1762 if (!incr_parse (self, cborstr))
1763 {
1764 if (self->incr_need > self->max_size && self->max_size)
1765 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1766 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1767
1768 break;
1769 }
1770
1771 SV *sv;
1772 char *offset;
1773
1774 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1775 XPUSHs (sv);
1776
1777 sv_chop (cborstr, offset);
1778
1779 av_clear (self->incr_count);
1780 av_push (self->incr_count, newSViv (1));
1781
1782 self->incr_pos = 0;
1783 self->incr_need = self->incr_pos + 1;
1784 }
1785 while (ix);
1786}
1787
1788void incr_reset (CBOR *self)
1789 CODE:
1790{
1791 SvREFCNT_dec (self->incr_count);
1792 self->incr_count = 0;
1793}
1794
1130void DESTROY (CBOR *self) 1795void DESTROY (CBOR *self)
1131 CODE: 1796 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1797 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1798
1135PROTOTYPES: ENABLE 1799PROTOTYPES: ENABLE
1136 1800
1137void encode_cbor (SV *scalar) 1801void encode_cbor (SV *scalar)
1802 ALIAS:
1803 encode_cbor = 0
1804 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1805 PPCODE:
1139{ 1806{
1140 CBOR cbor; 1807 CBOR cbor;
1141 cbor_init (&cbor); 1808 cbor_init (&cbor);
1809 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1810 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1811 XPUSHs (scalar);
1144} 1812}
1145 1813
1146void decode_cbor (SV *cborstr) 1814void decode_cbor (SV *cborstr)
1150 cbor_init (&cbor); 1818 cbor_init (&cbor);
1151 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1819 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1152 XPUSHs (cborstr); 1820 XPUSHs (cborstr);
1153} 1821}
1154 1822
1823#ifdef __AFL_COMPILER
1824
1825void
1826afl_init ()
1827 CODE:
1828 __AFL_INIT ();
1829
1830int
1831afl_loop (unsigned int count = 10000)
1832 CODE:
1833 RETVAL = __AFL_LOOP (count);
1834 OUTPUT:
1835 RETVAL
1836
1837#endif
1838

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines