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 1.79 by root, Fri Sep 8 06:05:01 2023 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// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0
34#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len)))
35
36// known major and minor types
37enum cbor_type
38{
39 MAJOR_SHIFT = 5,
40 MINOR_MASK = 0x1f,
41
42 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
43 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
44 MAJOR_BYTES = 2 << MAJOR_SHIFT,
45 MAJOR_TEXT = 3 << MAJOR_SHIFT,
46 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
47 MAJOR_MAP = 5 << MAJOR_SHIFT,
48 MAJOR_TAG = 6 << MAJOR_SHIFT,
49 MAJOR_MISC = 7 << MAJOR_SHIFT,
50
51 // INT/STRING/ARRAY/MAP subtypes
52 LENGTH_EXT1 = 24,
53 LENGTH_EXT2 = 25,
54 LENGTH_EXT4 = 26,
55 LENGTH_EXT8 = 27,
56
57 // SIMPLE types (effectively MISC subtypes)
58 SIMPLE_FALSE = 20,
59 SIMPLE_TRUE = 21,
60 SIMPLE_NULL = 22,
61 SIMPLE_UNDEF = 23,
62
63 // MISC subtype (unused)
64 MISC_EXT1 = 24,
65 MISC_FLOAT16 = 25,
66 MISC_FLOAT32 = 26,
67 MISC_FLOAT64 = 27,
68
69 // BYTES/TEXT/ARRAY/MAP
70 MINOR_INDEF = 31,
71};
13 72
14// known tags 73// known tags
15enum cbor_tag 74enum cbor_tag
16{ 75{
17 // inofficial extensions (pending iana registration) 76 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 77 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 78 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
79 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
80 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
81 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
82 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
83 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 84
21 // rfc7049 85 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 86 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 87 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 88 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 89 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 90 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 91 CBOR_TAG_BIGFLOAT = 5, // array
28 92
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 93 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 94 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 95 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 96 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 97
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 98 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 99 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 100 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 101 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 102 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 103
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 104 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 105};
42 106
107// known forced types, also hardcoded in CBOR.pm
108enum
109{
110 AS_CBOR = 0,
111 AS_INT = 1,
112 AS_BYTES = 2,
113 AS_TEXT = 3,
114 AS_FLOAT16 = 4,
115 AS_FLOAT32 = 5,
116 AS_FLOAT64 = 6,
117 AS_MAP = 7,
118 // possibly future enhancements: (generic) float, (generic) string
119};
120
43#define F_SHRINK 0x00000200UL 121#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 122#define F_ALLOW_UNKNOWN 0x00000002UL
123#define F_ALLOW_SHARING 0x00000004UL
124#define F_ALLOW_CYCLES 0x00000008UL
125#define F_FORBID_OBJECTS 0x00000010UL
126#define F_PACK_STRINGS 0x00000020UL
127#define F_TEXT_KEYS 0x00000040UL
128#define F_TEXT_STRINGS 0x00000080UL
129#define F_VALIDATE_UTF8 0x00000100UL
45 130
46#define INIT_SIZE 32 // initial scalar size to be allocated 131#define INIT_SIZE 32 // initial scalar size to be allocated
47 132
48#define SB do { 133#define SB do {
49#define SE } while (0) 134#define SE } while (0)
61# define CBOR_SLOW 0 146# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 147# define CBOR_STASH cbor_stash
63#endif 148#endif
64 149
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 150static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 151static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 152
68typedef struct { 153typedef struct {
69 U32 flags; 154 U32 flags;
70 U32 max_depth; 155 U32 max_depth;
71 STRLEN max_size; 156 STRLEN max_size;
157 SV *filter;
158
159 // for the incremental parser
160 STRLEN incr_pos; // the current offset into the text
161 STRLEN incr_need; // minimum bytes needed to decode
162 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 163} CBOR;
73 164
74ecb_inline void 165ecb_inline void
75cbor_init (CBOR *cbor) 166cbor_init (CBOR *cbor)
76{ 167{
77 Zero (cbor, 1, CBOR); 168 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 169 cbor->max_depth = 512;
170}
171
172ecb_inline void
173cbor_free (CBOR *cbor)
174{
175 SvREFCNT_dec (cbor->filter);
176 SvREFCNT_dec (cbor->incr_count);
79} 177}
80 178
81///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
82// utility functions 180// utility functions
83 181
105 SvPV_renew (sv, SvCUR (sv) + 1); 203 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 204#endif
107 } 205 }
108} 206}
109 207
110///////////////////////////////////////////////////////////////////////////// 208// minimum length of a string to be registered for stringref
111// fp hell 209ecb_inline STRLEN
112 210minimum_string_length (UV idx)
113//TODO 211{
212 return idx <= 23 ? 3
213 : idx <= 0xffU ? 4
214 : idx <= 0xffffU ? 5
215 : idx <= 0xffffffffU ? 7
216 : 11;
217}
114 218
115///////////////////////////////////////////////////////////////////////////// 219/////////////////////////////////////////////////////////////////////////////
116// encoder 220// encoder
117 221
118// structure used for encoding CBOR 222// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 225 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 226 char *end; // SvEND (sv)
123 SV *sv; // result scalar 227 SV *sv; // result scalar
124 CBOR cbor; 228 CBOR cbor;
125 U32 depth; // recursion level 229 U32 depth; // recursion level
230 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
231 UV stringref_idx;
232 HV *shareable; // ptr => index, or 0
233 UV shareable_idx;
126} enc_t; 234} enc_t;
127 235
128ecb_inline void 236ecb_inline void
129need (enc_t *enc, STRLEN len) 237need (enc_t *enc, STRLEN len)
130{ 238{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 239 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 240 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 241 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 242 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 243 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 244 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
137 } 245 }
138} 246}
139 247
248static void encode_sv (enc_t *enc, SV *sv);
249
140ecb_inline void 250ecb_inline void
141encode_ch (enc_t *enc, char ch) 251encode_ch (enc_t *enc, char ch)
142{ 252{
143 need (enc, 1); 253 need (enc, 1);
144 *enc->cur++ = ch; 254 *enc->cur++ = ch;
145} 255}
146 256
257// used for tags, intregers, element counts and so on
147static void 258static void
148encode_uint (enc_t *enc, int major, UV len) 259encode_uint (enc_t *enc, int major, UV len)
149{ 260{
150 need (enc, 9); 261 need (enc, 9);
151 262
152 if (len < 24) 263 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 264 *enc->cur++ = major | len;
154 else if (len <= 0xff) 265 else if (ecb_expect_true (len <= 0xffU))
155 { 266 {
156 *enc->cur++ = major | 24; 267 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 268 *enc->cur++ = len;
158 } 269 }
159 else if (len <= 0xffff) 270 else if (len <= 0xffffU)
160 { 271 {
161 *enc->cur++ = major | 25; 272 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 274 *enc->cur++ = len;
164 } 275 }
165 else if (len <= 0xffffffff) 276 else if (len <= 0xffffffffU)
166 { 277 {
167 *enc->cur++ = major | 26; 278 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 279 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 280 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 281 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 282 *enc->cur++ = len;
172 } 283 }
173 else 284 else
174 { 285 {
175 *enc->cur++ = major | 27; 286 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 287 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 288 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 289 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 290 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 291 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 293 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 294 *enc->cur++ = len;
184 } 295 }
185} 296}
186 297
187static void 298// encodes a perl value into a CBOR integer
299ecb_inline void
300encode_int (enc_t *enc, SV *sv)
301{
302 if (SvIsUV (sv))
303 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
304 else if (SvIVX (sv) >= 0)
305 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
306 else
307 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
308}
309
310ecb_inline void
311encode_tag (enc_t *enc, UV tag)
312{
313 encode_uint (enc, MAJOR_TAG, tag);
314}
315
316// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
317ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 318encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 319{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 320 STRLEN ulen = len;
321 U8 *p, *pend = (U8 *)str + len;
322
323 for (p = (U8 *)str; p < pend; ++p)
324 ulen += *p >> 7; // count set high bits
325
326 encode_uint (enc, MAJOR_TEXT, ulen);
327
328 need (enc, ulen);
329 for (p = (U8 *)str; p < pend; ++p)
330 if (*p < 0x80)
331 *enc->cur++ = *p;
332 else
333 {
334 *enc->cur++ = 0xc0 + (*p >> 6);
335 *enc->cur++ = 0x80 + (*p & 63);
336 }
337}
338
339ecb_inline void
340encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
341{
342 if (ecb_expect_false (upgrade_utf8))
343 if (!utf8)
344 {
345 encode_str_utf8 (enc, utf8, str, len);
346 return;
347 }
348
349 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 350 need (enc, len);
192 memcpy (enc->cur, str, len); 351 memcpy (enc->cur, str, len);
193 enc->cur += len; 352 enc->cur += len;
194} 353}
195 354
196static void encode_sv (enc_t *enc, SV *sv); 355ecb_inline void
356encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
357{
358 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
359 {
360 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
361
362 if (SvOK (*svp))
363 {
364 // already registered, use stringref
365 encode_tag (enc, CBOR_TAG_STRINGREF);
366 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
367 return;
368 }
369 else if (len >= minimum_string_length (enc->stringref_idx))
370 {
371 // register only
372 sv_setuv (*svp, enc->stringref_idx);
373 ++enc->stringref_idx;
374 }
375 }
376
377 encode_str (enc, upgrade_utf8, utf8, str, len);
378}
379
380ecb_inline void
381encode_float16 (enc_t *enc, NV nv)
382{
383 need (enc, 1+2);
384
385 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
386
387 uint16_t fp = ecb_float_to_binary16 (nv);
388
389 if (!ecb_big_endian ())
390 fp = ecb_bswap16 (fp);
391
392 memcpy (enc->cur, &fp, 2);
393 enc->cur += 2;
394}
395
396ecb_inline void
397encode_float32 (enc_t *enc, NV nv)
398{
399 need (enc, 1+4);
400
401 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
402
403 uint32_t fp = ecb_float_to_binary32 (nv);
404
405 if (!ecb_big_endian ())
406 fp = ecb_bswap32 (fp);
407
408 memcpy (enc->cur, &fp, 4);
409 enc->cur += 4;
410}
411
412ecb_inline void
413encode_float64 (enc_t *enc, NV nv)
414{
415 need (enc, 1+8);
416
417 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
418
419 uint64_t fp = ecb_double_to_binary64 (nv);
420
421 if (!ecb_big_endian ())
422 fp = ecb_bswap64 (fp);
423
424 memcpy (enc->cur, &fp, 8);
425 enc->cur += 8;
426}
427
428ecb_inline void
429encode_bool (enc_t *enc, int istrue)
430{
431 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
432}
433
434// encodes an arrayref containing key-value pairs as CBOR map
435ecb_inline void
436encode_array_as_map (enc_t *enc, SV *sv)
437{
438 if (enc->depth >= enc->cbor.max_depth)
439 croak (ERR_NESTING_EXCEEDED);
440
441 ++enc->depth;
442
443 // as_map does error checking for us, but we re-check in case
444 // things have changed.
445
446 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
447 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
448
449 AV *av = (AV *)SvRV (sv);
450 int i, len = av_len (av);
451
452 if (!(len & 1))
453 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
454
455 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
456
457 for (i = 0; i <= len; ++i)
458 {
459 SV **svp = av_fetch (av, i, 0);
460 encode_sv (enc, svp ? *svp : &PL_sv_undef);
461 }
462
463 --enc->depth;
464}
465
466ecb_inline void
467encode_forced (enc_t *enc, UV type, SV *sv)
468{
469 switch (type)
470 {
471 case AS_CBOR:
472 {
473 STRLEN len;
474 char *str = SvPVbyte (sv, len);
475
476 need (enc, len);
477 memcpy (enc->cur, str, len);
478 enc->cur += len;
479 }
480 break;
481
482 case AS_BYTES:
483 {
484 STRLEN len;
485 char *str = SvPVbyte (sv, len);
486 encode_strref (enc, 0, 0, str, len);
487 }
488 break;
489
490 case AS_TEXT:
491 {
492 STRLEN len;
493 char *str = SvPVutf8 (sv, len);
494 encode_strref (enc, 1, 1, str, len);
495 }
496 break;
497
498 case AS_INT: encode_int (enc, sv); break;
499
500 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
501 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
502 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
503
504 case AS_MAP: encode_array_as_map (enc, sv); break;
505
506 default:
507 croak ("encountered malformed CBOR::XS::Tagged object");
508 }
509}
197 510
198static void 511static void
199encode_av (enc_t *enc, AV *av) 512encode_av (enc_t *enc, AV *av)
200{ 513{
201 int i, len = av_len (av); 514 int i, len = av_len (av);
203 if (enc->depth >= enc->cbor.max_depth) 516 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 517 croak (ERR_NESTING_EXCEEDED);
205 518
206 ++enc->depth; 519 ++enc->depth;
207 520
208 encode_uint (enc, 0x80, len + 1); 521 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 522
523 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 524 for (i = 0; i <= len; ++i)
211 { 525 {
212 SV **svp = av_fetch (av, i, 0); 526 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 527 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 528 }
529 else
530 for (i = 0; i <= len; ++i)
531 {
532 SV *sv = AvARRAY (av)[i];
533 encode_sv (enc, sv ? sv : &PL_sv_undef);
534 }
215 535
216 --enc->depth; 536 --enc->depth;
217} 537}
218 538
219static void 539static void
227 ++enc->depth; 547 ++enc->depth;
228 548
229 int pairs = hv_iterinit (hv); 549 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 550 int mg = SvMAGICAL (hv);
231 551
232 if (mg) 552 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 553 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 554 else
235 encode_uint (enc, 0xa0, pairs); 555 encode_uint (enc, MAJOR_MAP, pairs);
236 556
237 while ((he = hv_iternext (hv))) 557 while ((he = hv_iternext (hv)))
238 { 558 {
239 if (HeKLEN (he) == HEf_SVKEY) 559 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 560 encode_sv (enc, HeSVKEY (he));
241 else 561 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 562 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 563
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 564 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 565 }
246 566
247 if (mg) 567 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 568 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 569
250 --enc->depth; 570 --enc->depth;
251} 571}
252 572
253// encode objects, arrays and special \0=false and \1=true values. 573// encode objects, arrays and special \0=false and \1=true values.
254static void 574static void
255encode_rv (enc_t *enc, SV *sv) 575encode_rv (enc_t *enc, SV *sv)
256{ 576{
257 svtype svt;
258
259 SvGETMAGIC (sv); 577 SvGETMAGIC (sv);
578
260 svt = SvTYPE (sv); 579 svtype svt = SvTYPE (sv);
261 580
262 if (ecb_expect_false (SvOBJECT (sv))) 581 if (ecb_expect_false (SvOBJECT (sv)))
263 { 582 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 583 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 584 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 589 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 590 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 591 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 592
274 HV *stash = SvSTASH (sv); 593 HV *stash = SvSTASH (sv);
275 GV *method;
276 594
277 if (stash == boolean_stash) 595 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 596 {
597 encode_bool (enc, SvIV (sv));
598 return;
599 }
279 else if (stash == error_stash) 600 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 601 {
602 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
603 return;
604 }
281 else if (stash == tagged_stash) 605 else if (stash == tagged_stash)
282 { 606 {
283 if (svt != SVt_PVAV) 607 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 608 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 609
610 switch (av_len ((AV *)sv))
611 {
612 case 2-1:
613 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 614 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 615 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
616 break;
617
618 case 3-1:
619 // a forced type [value, type, undef]
620 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
621 break;
622
623 default:
624 croak ("encountered malformed CBOR::XS::Tagged object");
625 }
626
627 return;
628 }
629 }
630
631 if (ecb_expect_false (SvREFCNT (sv) > 1)
632 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
633 {
634 if (ecb_expect_false (!enc->shareable))
635 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
636
637 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
638
639 if (SvOK (*svp))
288 } 640 {
641 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
642 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
643 return;
644 }
645 else
646 {
647 sv_setuv (*svp, enc->shareable_idx);
648 ++enc->shareable_idx;
649 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
650 }
651 }
652
653 if (ecb_expect_false (SvOBJECT (sv)))
654 {
655 HV *stash = SvSTASH (sv);
656 GV *method;
657
658 if (enc->cbor.flags & F_FORBID_OBJECTS)
659 croak ("encountered object '%s', but forbid_objects is enabled",
660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 661 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 662 {
291 dSP; 663 dSP;
292 664
293 ENTER; SAVETMPS; PUSHMARK (SP); 665 ENTER; SAVETMPS;
666 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 667 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 668 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 669
297 PUTBACK; 670 PUTBACK;
298 // G_SCALAR ensures that return value is 1 671 // G_SCALAR ensures that return value is 1
311 } 684 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 685 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 686 {
314 dSP; 687 dSP;
315 688
316 ENTER; SAVETMPS; PUSHMARK (SP); 689 ENTER; SAVETMPS;
690 PUSHMARK (SP);
317 EXTEND (SP, 2); 691 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 692 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 693 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 694 PUSHs (sv_cbor);
321 695
325 699
326 // catch this surprisingly common error 700 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 701 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)); 702 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 703
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 704 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 705 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 706 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 707
334 while (count) 708 {
709 int i;
710
711 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 712 encode_sv (enc, SP[i + 1 - count]);
713
714 SP -= count;
715 }
336 716
337 PUTBACK; 717 PUTBACK;
338 718
339 FREETMPS; LEAVE; 719 FREETMPS; LEAVE;
340 } 720 }
344 } 724 }
345 else if (svt == SVt_PVHV) 725 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 726 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 727 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 728 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 729 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 730 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 731 encode_tag (enc, CBOR_TAG_INDIRECTION);
732 encode_sv (enc, sv);
733 }
369} 734}
370 735
371static void 736static void
372encode_nv (enc_t *enc, SV *sv) 737encode_nv (enc_t *enc, SV *sv)
373{ 738{
374 double nv = SvNVX (sv); 739 double nv = SvNVX (sv);
375 740
376 need (enc, 9); 741 need (enc, 9);
377 742
378 if (ecb_expect_false (nv == (U32)nv)) 743 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 744 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 745 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 746 else if (ecb_expect_false (nv == (float)nv))
382 { 747 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 748 else
394 { 749 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} 750}
406 751
407static void 752static void
408encode_sv (enc_t *enc, SV *sv) 753encode_sv (enc_t *enc, SV *sv)
409{ 754{
411 756
412 if (SvPOKp (sv)) 757 if (SvPOKp (sv))
413 { 758 {
414 STRLEN len; 759 STRLEN len;
415 char *str = SvPV (sv, len); 760 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 761 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 762 }
418 else if (SvNOKp (sv)) 763 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 764 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 765 else if (SvIOKp (sv))
421 { 766 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)) 767 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 768 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 769 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 770 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 771 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 772 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 773 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 774 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 775 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 776}
439 777
440static SV * 778static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 779encode_cbor (SV *scalar, CBOR *cbor)
442{ 780{
443 enc_t enc; 781 enc_t enc = { 0 };
444 782
445 enc.cbor = *cbor; 783 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 784 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 785 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 786 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 787
451 SvPOK_only (enc.sv); 788 SvPOK_only (enc.sv);
789
790 if (cbor->flags & F_PACK_STRINGS)
791 {
792 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
793 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
794 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
795 }
796
452 encode_sv (&enc, scalar); 797 encode_sv (&enc, scalar);
453 798
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 799 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 800 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 801
470 U8 *end; // end of input string 815 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 816 const char *err; // parse error, if != 0
472 CBOR cbor; 817 CBOR cbor;
473 U32 depth; // recursion depth 818 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 819 U32 maxdepth; // recursion depth limit
820 AV *shareable;
821 AV *stringref;
822 SV *decode_tagged;
823 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 824} dec_t;
476 825
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 826// set dec->err to ERRSV
827ecb_cold static void
828err_errsv (dec_t *dec)
829{
830 if (!dec->err)
831 {
832 dec->err_sv = newSVsv (ERRSV);
478 833
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 834 // chop off the trailing \n
835 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
836 *SvEND (dec->err_sv) = 0;
480 837
838 dec->err = SvPVutf8_nolen (dec->err_sv);
839 }
840}
841
842// the following functions are used to reduce code size and help the compiler to optimise
843ecb_cold static void
844err_set (dec_t *dec, const char *reason)
845{
846 if (!dec->err)
847 dec->err = reason;
848}
849
850ecb_cold static void
851err_unexpected_end (dec_t *dec)
852{
853 err_set (dec, "unexpected end of CBOR data");
854}
855
856#define ERR_DO(do) SB do; goto fail; SE
857#define ERR(reason) ERR_DO (err_set (dec, reason))
858#define ERR_ERRSV ERR_DO (err_errsv (dec))
859
860#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
861
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 862#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
482#define DEC_DEC_DEPTH --dec->depth 863#define DEC_DEC_DEPTH --dec->depth
483 864
484static UV 865static UV
485decode_uint (dec_t *dec) 866decode_uint (dec_t *dec)
486{ 867{
487 switch (*dec->cur & 31) 868 U8 m = *dec->cur & MINOR_MASK;
488 { 869 ++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;
493 870
494 case 24: 871 if (ecb_expect_true (m < LENGTH_EXT1))
872 return m;
873 else if (ecb_expect_true (m == LENGTH_EXT1))
874 {
495 WANT (2); 875 WANT (1);
496 dec->cur += 2; 876 dec->cur += 1;
497 return dec->cur[-1]; 877 return dec->cur[-1];
498 878 }
499 case 25: 879 else if (ecb_expect_true (m == LENGTH_EXT2))
880 {
500 WANT (3); 881 WANT (2);
501 dec->cur += 3; 882 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 883 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 884 | ((UV)dec->cur[-1]);
504 885 }
505 case 26: 886 else if (ecb_expect_true (m == LENGTH_EXT4))
887 {
506 WANT (5); 888 WANT (4);
507 dec->cur += 5; 889 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 890 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 891 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 892 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 893 | ((UV)dec->cur[-1]);
512 894 }
513 case 27: 895 else if (ecb_expect_true (m == LENGTH_EXT8))
896 {
514 WANT (9); 897 WANT (8);
515 dec->cur += 9; 898 dec->cur += 8;
899
900 return
901#if UVSIZE < 8
902 0
903#else
516 return (((UV)dec->cur[-8]) << 56) 904 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 905 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 906 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 907 | (((UV)dec->cur[-5]) << 32)
908#endif
520 | (((UV)dec->cur[-4]) << 24) 909 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 910 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 911 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 912 | ((UV)dec->cur[-1]);
524 913 }
525 default: 914 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 915 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 916
529fail: 917fail:
530 return 0; 918 return 0;
531} 919}
532 920
537{ 925{
538 AV *av = newAV (); 926 AV *av = newAV ();
539 927
540 DEC_INC_DEPTH; 928 DEC_INC_DEPTH;
541 929
542 if ((*dec->cur & 31) == 31) 930 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 931 {
544 ++dec->cur; 932 ++dec->cur;
545 933
546 for (;;) 934 for (;;)
547 { 935 {
548 WANT (1); 936 WANT (1);
549 937
550 if (*dec->cur == (0xe0 | 31)) 938 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 939 {
552 ++dec->cur; 940 ++dec->cur;
553 break; 941 break;
554 } 942 }
555 943
556 av_push (av, decode_sv (dec)); 944 av_push (av, decode_sv (dec));
557 } 945 }
558 } 946 }
559 else 947 else
560 { 948 {
561 int i, len = decode_uint (dec); 949 UV i, len = decode_uint (dec);
562 950
951 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); 952 av_fill (av, len - 1);
564 953
565 for (i = 0; i < len; ++i) 954 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 955 AvARRAY (av)[i] = decode_sv (dec);
567 } 956 }
568 957
569 DEC_DEC_DEPTH; 958 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 959 return newRV_noinc ((SV *)av);
571 960
572fail: 961fail:
573 SvREFCNT_dec (av); 962 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 963 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 964 return &PL_sv_undef;
576} 965}
577 966
967static void
968decode_he (dec_t *dec, HV *hv)
969{
970 // for speed reasons, we specialcase single-string
971 // byte or utf-8 strings as keys, but only when !stringref
972
973 if (ecb_expect_true (!dec->stringref))
974 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
975 {
976 STRLEN len = decode_uint (dec);
977 char *key = (char *)dec->cur;
978
979 WANT (len);
980 dec->cur += len;
981
982 hv_store (hv, key, len, decode_sv (dec), 0);
983
984 return;
985 }
986 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
987 {
988 STRLEN len = decode_uint (dec);
989 char *key = (char *)dec->cur;
990
991 WANT (len);
992 dec->cur += len;
993
994 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
995 if (!cbor_is_utf8_string ((U8 *)key, len))
996 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
997
998 hv_store (hv, key, -len, decode_sv (dec), 0);
999
1000 return;
1001 }
1002
1003 SV *k = decode_sv (dec);
1004 SV *v = decode_sv (dec);
1005
1006 // we leak memory if uncaught exceptions are thrown by random magical
1007 // methods, and this is hopefully the only place where it can happen,
1008 // so if there is a chance of an exception, take the very slow path.
1009 // since catching exceptions is "undocumented/internal/forbidden" by
1010 // the new p5p powers, we need to call out to a perl function :/
1011 if (ecb_expect_false (SvAMAGIC (k)))
1012 {
1013 dSP;
1014
1015 ENTER; SAVETMPS;
1016 PUSHMARK (SP);
1017 EXTEND (SP, 3);
1018 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1019 PUSHs (sv_2mortal (k));
1020 PUSHs (sv_2mortal (v));
1021
1022 PUTBACK;
1023 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1024 SPAGAIN;
1025
1026 FREETMPS; LEAVE;
1027
1028 if (SvTRUE (ERRSV))
1029 ERR_ERRSV;
1030
1031 return;
1032 }
1033
1034 hv_store_ent (hv, k, v, 0);
1035 SvREFCNT_dec_NN (k);
1036
1037fail:
1038 ;
1039}
1040
578static SV * 1041static SV *
579decode_hv (dec_t *dec) 1042decode_hv (dec_t *dec)
580{ 1043{
581 HV *hv = newHV (); 1044 HV *hv = newHV ();
582 1045
583 DEC_INC_DEPTH; 1046 DEC_INC_DEPTH;
584 1047
585 if ((*dec->cur & 31) == 31) 1048 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1049 {
587 ++dec->cur; 1050 ++dec->cur;
588 1051
589 for (;;) 1052 for (;;)
590 { 1053 {
591 WANT (1); 1054 WANT (1);
592 1055
593 if (*dec->cur == (0xe0 | 31)) 1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1057 {
595 ++dec->cur; 1058 ++dec->cur;
596 break; 1059 break;
597 } 1060 }
598 1061
599 SV *k = decode_sv (dec); 1062 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 } 1063 }
604 } 1064 }
605 else 1065 else
606 { 1066 {
607 int len = decode_uint (dec); 1067 UV pairs = decode_uint (dec);
608 1068
1069 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1070
609 while (len--) 1071 while (pairs--)
610 { 1072 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 } 1073 }
617 1074
618 DEC_DEC_DEPTH; 1075 DEC_DEC_DEPTH;
619 return newRV_noinc ((SV *)hv); 1076 return newRV_noinc ((SV *)hv);
620 1077
621fail: 1078fail:
622 SvREFCNT_dec (hv); 1079 SvREFCNT_dec_NN (hv);
623 DEC_DEC_DEPTH; 1080 DEC_DEC_DEPTH;
624 return &PL_sv_undef; 1081 return &PL_sv_undef;
625} 1082}
626 1083
627static SV * 1084static SV *
628decode_str (dec_t *dec, int utf8) 1085decode_str (dec_t *dec, int utf8)
629{ 1086{
630 SV *sv = 0; 1087 SV *sv = 0;
631 1088
632 if ((*dec->cur & 31) == 31) 1089 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
633 { 1090 {
1091 // indefinite length strings
634 ++dec->cur; 1092 ++dec->cur;
635 1093
1094 U8 major = *dec->cur & MAJOR_MISC;
1095
636 sv = newSVpvn ("", 0); 1096 sv = newSVpvn ("", 0);
637 1097
638 // not very fast, and certainly not robust against illegal input
639 for (;;) 1098 for (;;)
640 { 1099 {
641 WANT (1); 1100 WANT (1);
642 1101
643 if (*dec->cur == (0xe0 | 31)) 1102 if ((*dec->cur - major) > LENGTH_EXT8)
1103 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
644 { 1104 {
645 ++dec->cur; 1105 ++dec->cur;
646 break; 1106 break;
647 } 1107 }
1108 else
1109 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
648 1110
649 sv_catsv (sv, decode_sv (dec)); 1111 STRLEN len = decode_uint (dec);
1112
1113 WANT (len);
1114 sv_catpvn (sv, dec->cur, len);
1115 dec->cur += len;
650 } 1116 }
651 } 1117 }
652 else 1118 else
653 { 1119 {
654 STRLEN len = decode_uint (dec); 1120 STRLEN len = decode_uint (dec);
655 1121
656 WANT (len); 1122 WANT (len);
657 sv = newSVpvn (dec->cur, len); 1123 sv = newSVpvn (dec->cur, len);
658 dec->cur += len; 1124 dec->cur += len;
1125
1126 if (ecb_expect_false (dec->stringref)
1127 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1128 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
659 } 1129 }
660 1130
661 if (utf8) 1131 if (utf8)
1132 {
1133 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1134 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1135 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1136
662 SvUTF8_on (sv); 1137 SvUTF8_on (sv);
1138 }
663 1139
664 return sv; 1140 return sv;
665 1141
666fail: 1142fail:
667 SvREFCNT_dec (sv); 1143 SvREFCNT_dec (sv);
669} 1145}
670 1146
671static SV * 1147static SV *
672decode_tagged (dec_t *dec) 1148decode_tagged (dec_t *dec)
673{ 1149{
1150 SV *sv = 0;
674 UV tag = decode_uint (dec); 1151 UV tag = decode_uint (dec);
1152
1153 WANT (1);
1154
1155 switch (tag)
1156 {
1157 case CBOR_TAG_MAGIC:
675 SV *sv = decode_sv (dec); 1158 sv = decode_sv (dec);
1159 break;
676 1160
677 if (tag == CBOR_TAG_MAGIC) 1161 case CBOR_TAG_INDIRECTION:
678 return sv; 1162 sv = newRV_noinc (decode_sv (dec));
679 else if (tag == CBOR_TAG_PERL_OBJECT) 1163 break;
680 { 1164
1165 case CBOR_TAG_STRINGREF_NAMESPACE:
1166 {
1167 // do not use SAVETMPS/FREETMPS, as these will
1168 // erase mortalised caches, e.g. "shareable"
1169 ENTER;
1170
1171 SAVESPTR (dec->stringref);
1172 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1173
1174 sv = decode_sv (dec);
1175
1176 LEAVE;
1177 }
1178 break;
1179
1180 case CBOR_TAG_STRINGREF:
1181 {
1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1183 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1184
1185 UV idx = decode_uint (dec);
1186
1187 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1188 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1189
1190 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1191 }
1192 break;
1193
1194 case CBOR_TAG_VALUE_SHAREABLE:
1195 {
1196 if (ecb_expect_false (!dec->shareable))
1197 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1198
1199 if (dec->cbor.flags & F_ALLOW_CYCLES)
1200 {
1201 // if cycles are allowed, then we store an AV as value
1202 // while it is being decoded, and gather unresolved
1203 // references in it, to be re4solved after decoding.
1204 int idx, i;
1205 AV *av = newAV ();
1206 av_push (dec->shareable, (SV *)av);
1207 idx = AvFILLp (dec->shareable);
1208
1209 sv = decode_sv (dec);
1210
1211 // the AV now contains \undef for all unresolved references,
1212 // so we fix them up here.
1213 for (i = 0; i <= AvFILLp (av); ++i)
1214 SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
1215
1216 // now replace the AV by a reference to the completed value
1217 SvREFCNT_dec_NN ((SV *)av);
1218 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1219 }
1220 else
1221 {
1222 av_push (dec->shareable, &PL_sv_undef);
1223 int idx = AvFILLp (dec->shareable);
1224 sv = decode_sv (dec);
1225 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1226 }
1227 }
1228 break;
1229
1230 case CBOR_TAG_VALUE_SHAREDREF:
1231 {
1232 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1233 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1234
1235 UV idx = decode_uint (dec);
1236
1237 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1238 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1239
1240 sv = AvARRAY (dec->shareable)[idx];
1241
1242 // reference to cycle, we create a new \undef and use that, and also
1243 // registerr it in the AV for later fixing
1244 if (SvTYPE (sv) == SVt_PVAV)
1245 {
1246 AV *av = (AV *)sv;
1247 sv = newRV_noinc (&PL_sv_undef);
1248 av_push (av, SvREFCNT_inc_NN (sv));
1249 }
1250 else if (sv == &PL_sv_undef) // not yet decoded, but cycles not allowed
1251 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1252 else // we decoded the object earlier, no cycle
1253 sv = newSVsv (sv);
1254 }
1255 break;
1256
1257 case CBOR_TAG_PERL_OBJECT:
1258 {
1259 if (dec->cbor.flags & F_FORBID_OBJECTS)
1260 goto filter;
1261
1262 sv = decode_sv (dec);
1263
681 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1264 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
682 ERR ("corrupted CBOR data (non-array perl object)"); 1265 ERR ("corrupted CBOR data (non-array perl object)");
683 1266
684 AV *av = (AV *)SvRV (sv); 1267 AV *av = (AV *)SvRV (sv);
685 int len = av_len (av) + 1; 1268 int len = av_len (av) + 1;
686 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1269 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
687 1270
688 if (!stash) 1271 if (!stash)
689 ERR ("cannot decode perl-object (package does not exist)"); 1272 ERR ("cannot decode perl-object (package does not exist)");
690 1273
691 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1274 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
692 1275
693 if (!method) 1276 if (!method)
694 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1277 ERR ("cannot decode perl-object (package does not have a THAW method)");
695 1278
696 dSP; 1279 dSP;
697 1280
698 ENTER; SAVETMPS; PUSHMARK (SP); 1281 ENTER; SAVETMPS;
1282 PUSHMARK (SP);
699 EXTEND (SP, len + 1); 1283 EXTEND (SP, len + 1);
700 // we re-bless the reference to get overload and other niceties right 1284 // we re-bless the reference to get overload and other niceties right
701 PUSHs (*av_fetch (av, 0, 1)); 1285 PUSHs (*av_fetch (av, 0, 1));
702 PUSHs (sv_cbor); 1286 PUSHs (sv_cbor);
703 1287
704 int i; 1288 int i;
705 1289
706 for (i = 1; i < len; ++i) 1290 for (i = 1; i < len; ++i)
707 PUSHs (*av_fetch (av, i, 1)); 1291 PUSHs (*av_fetch (av, i, 1));
708 1292
709 PUTBACK; 1293 PUTBACK;
710 call_sv ((SV *)GvCV (method), G_SCALAR); 1294 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
711 SPAGAIN; 1295 SPAGAIN;
712 1296
1297 if (SvTRUE (ERRSV))
1298 {
1299 FREETMPS; LEAVE;
1300 ERR_ERRSV;
1301 }
1302
1303 SvREFCNT_dec_NN (sv);
713 sv = SvREFCNT_inc (POPs); 1304 sv = SvREFCNT_inc (POPs);
714 1305
715 PUTBACK; 1306 PUTBACK;
716 1307
717 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1309 }
1310 break;
718 1311
719 return sv; 1312 default:
720 } 1313 filter:
721 else 1314 {
722 { 1315 SV *tag_sv = newSVuv (tag);
1316
1317 sv = decode_sv (dec);
1318
1319 dSP;
1320 ENTER; SAVETMPS;
1321 PUSHMARK (SP);
1322 EXTEND (SP, 2);
1323 PUSHs (tag_sv);
1324 PUSHs (sv);
1325
1326 PUTBACK;
1327 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1328 SPAGAIN;
1329
1330 if (SvTRUE (ERRSV))
1331 {
1332 SvREFCNT_dec_NN (tag_sv);
1333 FREETMPS; LEAVE;
1334 ERR_ERRSV;
1335 }
1336
1337 if (count)
1338 {
1339 SvREFCNT_dec_NN (tag_sv);
1340 SvREFCNT_dec_NN (sv);
1341 sv = SvREFCNT_inc_NN (TOPs);
1342 SP -= count;
1343 }
1344 else
1345 {
723 AV *av = newAV (); 1346 AV *av = newAV ();
724 av_push (av, newSVuv (tag)); 1347 av_push (av, tag_sv);
725 av_push (av, sv); 1348 av_push (av, sv);
726 1349
727 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1350 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
728 ? cbor_tagged_stash 1351 ? cbor_tagged_stash
729 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 : gv_stashpv ("CBOR::XS::Tagged" , 1);
730
731 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1353 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1354 }
1355
1356 PUTBACK;
1357
1358 FREETMPS; LEAVE;
1359 }
1360 break;
732 } 1361 }
1362
1363 return sv;
733 1364
734fail: 1365fail:
735 SvREFCNT_dec (sv); 1366 SvREFCNT_dec (sv);
736 return &PL_sv_undef; 1367 return &PL_sv_undef;
737} 1368}
739static SV * 1370static SV *
740decode_sv (dec_t *dec) 1371decode_sv (dec_t *dec)
741{ 1372{
742 WANT (1); 1373 WANT (1);
743 1374
744 switch (*dec->cur >> 5) 1375 switch (*dec->cur >> MAJOR_SHIFT)
745 { 1376 {
746 case 0: // unsigned int 1377 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
747 return newSVuv (decode_uint (dec)); 1378 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
748 case 1: // negative int 1379 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
749 return newSViv (-1 - (IV)decode_uint (dec)); 1380 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
750 case 2: // octet string 1381 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
751 return decode_str (dec, 0); 1382 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
752 case 3: // utf-8 string 1383 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
753 return decode_str (dec, 1); 1384
754 case 4: // array 1385 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) 1386 switch (*dec->cur++ & MINOR_MASK)
762 { 1387 {
763 case 20: 1388 case SIMPLE_FALSE:
764#if CBOR_SLOW 1389#if CBOR_SLOW
765 types_false = get_bool ("Types::Serialiser::false"); 1390 types_false = get_bool ("Types::Serialiser::false");
766#endif 1391#endif
767 return newSVsv (types_false); 1392 return newSVsv (types_false);
768 case 21: 1393 case SIMPLE_TRUE:
769#if CBOR_SLOW 1394#if CBOR_SLOW
770 types_true = get_bool ("Types::Serialiser::true"); 1395 types_true = get_bool ("Types::Serialiser::true");
771#endif 1396#endif
772 return newSVsv (types_true); 1397 return newSVsv (types_true);
773 case 22: 1398 case SIMPLE_NULL:
774 return newSVsv (&PL_sv_undef); 1399 return newSVsv (&PL_sv_undef);
775 case 23: 1400 case SIMPLE_UNDEF:
776#if CBOR_SLOW 1401#if CBOR_SLOW
777 types_error = get_bool ("Types::Serialiser::error"); 1402 types_error = get_bool ("Types::Serialiser::error");
778#endif 1403#endif
779 return newSVsv (types_error); 1404 return newSVsv (types_error);
780 1405
781 case 25: 1406 case MISC_FLOAT16:
782 { 1407 {
783 WANT (2); 1408 WANT (2);
784 1409
785 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1410 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
786 dec->cur += 2; 1411 dec->cur += 2;
787 1412
788 return newSVnv (ecb_binary16_to_float (fp)); 1413 return newSVnv (ecb_binary16_to_float (fp));
789 } 1414 }
790 1415
791 case 26: 1416 case MISC_FLOAT32:
792 { 1417 {
793 uint32_t fp; 1418 uint32_t fp;
794 WANT (4); 1419 WANT (4);
795 memcpy (&fp, dec->cur, 4); 1420 memcpy (&fp, dec->cur, 4);
796 dec->cur += 4; 1421 dec->cur += 4;
799 fp = ecb_bswap32 (fp); 1424 fp = ecb_bswap32 (fp);
800 1425
801 return newSVnv (ecb_binary32_to_float (fp)); 1426 return newSVnv (ecb_binary32_to_float (fp));
802 } 1427 }
803 1428
804 case 27: 1429 case MISC_FLOAT64:
805 { 1430 {
806 uint64_t fp; 1431 uint64_t fp;
807 WANT (8); 1432 WANT (8);
808 memcpy (&fp, dec->cur, 8); 1433 memcpy (&fp, dec->cur, 8);
809 dec->cur += 8; 1434 dec->cur += 8;
812 fp = ecb_bswap64 (fp); 1437 fp = ecb_bswap64 (fp);
813 1438
814 return newSVnv (ecb_binary64_to_double (fp)); 1439 return newSVnv (ecb_binary64_to_double (fp));
815 } 1440 }
816 1441
817 // 0..19 unassigned 1442 // 0..19 unassigned simple
818 // 24 reserved + unassigned (reserved values are not encodable) 1443 // 24 reserved + unassigned simple (reserved values are not encodable)
1444 // 28-30 unassigned misc
1445 // 31 break code
819 default: 1446 default:
820 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1447 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
821 } 1448 }
822 1449
823 break; 1450 break;
824 } 1451 }
825 1452
828} 1455}
829 1456
830static SV * 1457static SV *
831decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1458decode_cbor (SV *string, CBOR *cbor, char **offset_return)
832{ 1459{
833 dec_t dec; 1460 dec_t dec = { 0 };
834 SV *sv; 1461 SV *sv;
1462 STRLEN len;
1463 char *data = SvPVbyte (string, len);
835 1464
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) 1465 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", 1466 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); 1467 (unsigned long)len, (unsigned long)cbor->max_size);
867 }
868
869 sv_utf8_downgrade (string, 0);
870 1468
871 dec.cbor = *cbor; 1469 dec.cbor = *cbor;
872 dec.cur = (U8 *)SvPVX (string); 1470 dec.cur = (U8 *)data;
873 dec.end = (U8 *)SvEND (string); 1471 dec.end = (U8 *)data + len;
874 dec.err = 0;
875 dec.depth = 0;
876 1472
877 sv = decode_sv (&dec); 1473 sv = decode_sv (&dec);
878 1474
879 if (offset_return) 1475 if (offset_return)
880 *offset_return = dec.cur; 1476 *offset_return = dec.cur;
883 if (dec.cur != dec.end && !dec.err) 1479 if (dec.cur != dec.end && !dec.err)
884 dec.err = "garbage after CBOR object"; 1480 dec.err = "garbage after CBOR object";
885 1481
886 if (dec.err) 1482 if (dec.err)
887 { 1483 {
1484 if (dec.shareable)
1485 {
1486 // need to break cyclic links, which would all be in shareable
1487 int i;
1488 SV **svp;
1489
1490 for (i = av_len (dec.shareable) + 1; i--; )
1491 if ((svp = av_fetch (dec.shareable, i, 0)))
1492 sv_setsv (*svp, &PL_sv_undef);
1493 }
1494
888 SvREFCNT_dec (sv); 1495 SvREFCNT_dec_NN (sv);
1496
1497 if (dec.err_sv)
1498 sv_2mortal (dec.err_sv);
1499
889 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1500 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
890 } 1501 }
891 1502
892 sv = sv_2mortal (sv); 1503 sv = sv_2mortal (sv);
893 1504
894 return sv; 1505 return sv;
895} 1506}
896 1507
1508/////////////////////////////////////////////////////////////////////////////
1509// incremental parser
1510
1511#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1512
1513// returns 0 for notyet, 1 for success or error
1514static int
1515incr_parse (CBOR *self, SV *cborstr)
1516{
1517 STRLEN cur;
1518 SvPV (cborstr, cur);
1519
1520 while (ecb_expect_true (self->incr_need <= cur))
1521 {
1522 // table of integer count bytes
1523 static I8 incr_len[MINOR_MASK + 1] = {
1524 0, 0, 0, 0, 0, 0, 0, 0,
1525 0, 0, 0, 0, 0, 0, 0, 0,
1526 0, 0, 0, 0, 0, 0, 0, 0,
1527 1, 2, 4, 8,-1,-1,-1,-2
1528 };
1529
1530 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1531 U8 m = *p & MINOR_MASK;
1532 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1533 I8 ilen = incr_len[m];
1534
1535 self->incr_need = self->incr_pos + 1;
1536
1537 if (ecb_expect_false (ilen < 0))
1538 {
1539 if (m != MINOR_INDEF)
1540 return 1; // error
1541
1542 if (*p == (MAJOR_MISC | MINOR_INDEF))
1543 {
1544 if (count >= 0)
1545 return 1; // error
1546
1547 count = 1;
1548 }
1549 else
1550 {
1551 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1552 count = -1;
1553 }
1554 }
1555 else
1556 {
1557 self->incr_need += ilen;
1558 if (ecb_expect_false (self->incr_need > cur))
1559 return 0;
1560
1561 int major = *p >> MAJOR_SHIFT;
1562
1563 switch (major)
1564 {
1565 case MAJOR_TAG >> MAJOR_SHIFT:
1566 ++count; // tags merely prefix another value
1567 break;
1568
1569 case MAJOR_BYTES >> MAJOR_SHIFT:
1570 case MAJOR_TEXT >> MAJOR_SHIFT:
1571 case MAJOR_ARRAY >> MAJOR_SHIFT:
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 {
1574 UV len;
1575
1576 if (ecb_expect_false (ilen))
1577 {
1578 len = 0;
1579
1580 do {
1581 len = (len << 8) | *++p;
1582 } while (--ilen);
1583 }
1584 else
1585 len = m;
1586
1587 switch (major)
1588 {
1589 case MAJOR_BYTES >> MAJOR_SHIFT:
1590 case MAJOR_TEXT >> MAJOR_SHIFT:
1591 self->incr_need += len;
1592 if (ecb_expect_false (self->incr_need > cur))
1593 return 0;
1594
1595 break;
1596
1597 case MAJOR_MAP >> MAJOR_SHIFT:
1598 len <<= 1;
1599 /* FALLTHROUGH */
1600 case MAJOR_ARRAY >> MAJOR_SHIFT:
1601 if (len)
1602 {
1603 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1604 count = len + 1;
1605 }
1606 break;
1607 }
1608 }
1609 }
1610 }
1611
1612 self->incr_pos = self->incr_need;
1613
1614 if (count > 0)
1615 {
1616 while (!--count)
1617 {
1618 if (!AvFILLp (self->incr_count))
1619 return 1; // done
1620
1621 SvREFCNT_dec_NN (av_pop (self->incr_count));
1622 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1623 }
1624
1625 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1626 }
1627 }
1628
1629 return 0;
1630}
1631
1632
897///////////////////////////////////////////////////////////////////////////// 1633/////////////////////////////////////////////////////////////////////////////
898// XS interface functions 1634// XS interface functions
899 1635
900MODULE = CBOR::XS PACKAGE = CBOR::XS 1636MODULE = CBOR::XS PACKAGE = CBOR::XS
901 1637
909 1645
910 types_true = get_bool ("Types::Serialiser::true" ); 1646 types_true = get_bool ("Types::Serialiser::true" );
911 types_false = get_bool ("Types::Serialiser::false"); 1647 types_false = get_bool ("Types::Serialiser::false");
912 types_error = get_bool ("Types::Serialiser::error"); 1648 types_error = get_bool ("Types::Serialiser::error");
913 1649
1650 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1651
914 sv_cbor = newSVpv ("CBOR", 0); 1652 sv_cbor = newSVpv ("CBOR", 0);
915 SvREADONLY_on (sv_cbor); 1653 SvREADONLY_on (sv_cbor);
1654
1655 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
916} 1656}
917 1657
918PROTOTYPES: DISABLE 1658PROTOTYPES: DISABLE
919 1659
920void CLONE (...) 1660void CLONE (...)
938 1678
939void shrink (CBOR *self, int enable = 1) 1679void shrink (CBOR *self, int enable = 1)
940 ALIAS: 1680 ALIAS:
941 shrink = F_SHRINK 1681 shrink = F_SHRINK
942 allow_unknown = F_ALLOW_UNKNOWN 1682 allow_unknown = F_ALLOW_UNKNOWN
1683 allow_sharing = F_ALLOW_SHARING
1684 allow_cycles = F_ALLOW_CYCLES
1685 forbid_objects = F_FORBID_OBJECTS
1686 pack_strings = F_PACK_STRINGS
1687 text_keys = F_TEXT_KEYS
1688 text_strings = F_TEXT_STRINGS
1689 validate_utf8 = F_VALIDATE_UTF8
943 PPCODE: 1690 PPCODE:
944{ 1691{
945 if (enable) 1692 if (enable)
946 self->flags |= ix; 1693 self->flags |= ix;
947 else 1694 else
952 1699
953void get_shrink (CBOR *self) 1700void get_shrink (CBOR *self)
954 ALIAS: 1701 ALIAS:
955 get_shrink = F_SHRINK 1702 get_shrink = F_SHRINK
956 get_allow_unknown = F_ALLOW_UNKNOWN 1703 get_allow_unknown = F_ALLOW_UNKNOWN
1704 get_allow_sharing = F_ALLOW_SHARING
1705 get_allow_cycles = F_ALLOW_CYCLES
1706 get_forbid_objects = F_FORBID_OBJECTS
1707 get_pack_strings = F_PACK_STRINGS
1708 get_text_keys = F_TEXT_KEYS
1709 get_text_strings = F_TEXT_STRINGS
1710 get_validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1711 PPCODE:
958 XPUSHs (boolSV (self->flags & ix)); 1712 XPUSHs (boolSV (self->flags & ix));
959 1713
960void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1714void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
961 PPCODE: 1715 PPCODE:
977 CODE: 1731 CODE:
978 RETVAL = self->max_size; 1732 RETVAL = self->max_size;
979 OUTPUT: 1733 OUTPUT:
980 RETVAL 1734 RETVAL
981 1735
982#if 0 //TODO 1736void filter (CBOR *self, SV *filter = 0)
983
984void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
985 PPCODE: 1737 PPCODE:
986{
987 SvREFCNT_dec (self->cb_object); 1738 SvREFCNT_dec (self->filter);
988 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1739 self->filter = filter ? newSVsv (filter) : filter;
989
990 XPUSHs (ST (0)); 1740 XPUSHs (ST (0));
991}
992 1741
993void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1742SV *get_filter (CBOR *self)
994 PPCODE: 1743 CODE:
995{ 1744 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
996 if (!self->cb_sk_object) 1745 OUTPUT:
997 self->cb_sk_object = newHV (); 1746 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 1747
1017void encode (CBOR *self, SV *scalar) 1748void encode (CBOR *self, SV *scalar)
1018 PPCODE: 1749 PPCODE:
1019 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1750 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1020 XPUSHs (scalar); 1751 XPUSHs (scalar);
1033 EXTEND (SP, 2); 1764 EXTEND (SP, 2);
1034 PUSHs (sv); 1765 PUSHs (sv);
1035 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1766 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1036} 1767}
1037 1768
1038#if 0 1769void incr_parse (CBOR *self, SV *cborstr)
1770 ALIAS:
1771 incr_parse_multiple = 1
1772 PPCODE:
1773{
1774 if (SvUTF8 (cborstr))
1775 sv_utf8_downgrade (cborstr, 0);
1776
1777 if (!self->incr_count)
1778 {
1779 self->incr_count = newAV ();
1780 self->incr_pos = 0;
1781 self->incr_need = 1;
1782
1783 av_push (self->incr_count, newSViv (1));
1784 }
1785
1786 do
1787 {
1788 if (!incr_parse (self, cborstr))
1789 {
1790 if (self->incr_need > self->max_size && self->max_size)
1791 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1792 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1793
1794 break;
1795 }
1796
1797 SV *sv;
1798 char *offset;
1799
1800 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1801 XPUSHs (sv);
1802
1803 sv_chop (cborstr, offset);
1804
1805 av_clear (self->incr_count);
1806 av_push (self->incr_count, newSViv (1));
1807
1808 self->incr_pos = 0;
1809 self->incr_need = self->incr_pos + 1;
1810 }
1811 while (ix);
1812}
1813
1814void incr_reset (CBOR *self)
1815 CODE:
1816{
1817 SvREFCNT_dec (self->incr_count);
1818 self->incr_count = 0;
1819}
1039 1820
1040void DESTROY (CBOR *self) 1821void DESTROY (CBOR *self)
1041 CODE: 1822 PPCODE:
1042 SvREFCNT_dec (self->cb_sk_object); 1823 cbor_free (self);
1043 SvREFCNT_dec (self->cb_object);
1044
1045#endif
1046 1824
1047PROTOTYPES: ENABLE 1825PROTOTYPES: ENABLE
1048 1826
1049void encode_cbor (SV *scalar) 1827void encode_cbor (SV *scalar)
1828 ALIAS:
1829 encode_cbor = 0
1830 encode_cbor_sharing = F_ALLOW_SHARING
1050 PPCODE: 1831 PPCODE:
1051{ 1832{
1052 CBOR cbor; 1833 CBOR cbor;
1053 cbor_init (&cbor); 1834 cbor_init (&cbor);
1835 cbor.flags |= ix;
1054 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1836 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1055 XPUSHs (scalar); 1837 XPUSHs (scalar);
1056} 1838}
1057 1839
1058void decode_cbor (SV *cborstr) 1840void decode_cbor (SV *cborstr)
1062 cbor_init (&cbor); 1844 cbor_init (&cbor);
1063 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1845 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1064 XPUSHs (cborstr); 1846 XPUSHs (cborstr);
1065} 1847}
1066 1848
1849#ifdef __AFL_COMPILER
1850
1851void
1852afl_init ()
1853 CODE:
1854 __AFL_INIT ();
1855
1856int
1857afl_loop (unsigned int count = 10000)
1858 CODE:
1859 RETVAL = __AFL_LOOP (count);
1860 OUTPUT:
1861 RETVAL
1862
1863#endif
1864

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines