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 1.80 by root, Fri Sep 8 20:03:06 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_ALLOW_WEAK_CYCLES 0x00000010UL
126#define F_FORBID_OBJECTS 0x00000020UL
127#define F_PACK_STRINGS 0x00000040UL
128#define F_TEXT_KEYS 0x00000080UL
129#define F_TEXT_STRINGS 0x00000100UL
130#define F_VALIDATE_UTF8 0x00000200UL
45 131
46#define INIT_SIZE 32 // initial scalar size to be allocated 132#define INIT_SIZE 32 // initial scalar size to be allocated
47 133
48#define SB do { 134#define SB do {
49#define SE } while (0) 135#define SE } while (0)
61# define CBOR_SLOW 0 147# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 148# define CBOR_STASH cbor_stash
63#endif 149#endif
64 150
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 151static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 152static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 153
68typedef struct { 154typedef struct {
69 U32 flags; 155 U32 flags;
70 U32 max_depth; 156 U32 max_depth;
71 STRLEN max_size; 157 STRLEN max_size;
158 SV *filter;
159
160 // for the incremental parser
161 STRLEN incr_pos; // the current offset into the text
162 STRLEN incr_need; // minimum bytes needed to decode
163 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 164} CBOR;
73 165
74ecb_inline void 166ecb_inline void
75cbor_init (CBOR *cbor) 167cbor_init (CBOR *cbor)
76{ 168{
77 Zero (cbor, 1, CBOR); 169 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 170 cbor->max_depth = 512;
171}
172
173ecb_inline void
174cbor_free (CBOR *cbor)
175{
176 SvREFCNT_dec (cbor->filter);
177 SvREFCNT_dec (cbor->incr_count);
79} 178}
80 179
81///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
82// utility functions 181// utility functions
83 182
105 SvPV_renew (sv, SvCUR (sv) + 1); 204 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 205#endif
107 } 206 }
108} 207}
109 208
110///////////////////////////////////////////////////////////////////////////// 209// minimum length of a string to be registered for stringref
111// fp hell 210ecb_inline STRLEN
112 211minimum_string_length (UV idx)
113//TODO 212{
213 return idx <= 23 ? 3
214 : idx <= 0xffU ? 4
215 : idx <= 0xffffU ? 5
216 : idx <= 0xffffffffU ? 7
217 : 11;
218}
114 219
115///////////////////////////////////////////////////////////////////////////// 220/////////////////////////////////////////////////////////////////////////////
116// encoder 221// encoder
117 222
118// structure used for encoding CBOR 223// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 226 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 227 char *end; // SvEND (sv)
123 SV *sv; // result scalar 228 SV *sv; // result scalar
124 CBOR cbor; 229 CBOR cbor;
125 U32 depth; // recursion level 230 U32 depth; // recursion level
231 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
232 UV stringref_idx;
233 HV *shareable; // ptr => index, or 0
234 UV shareable_idx;
126} enc_t; 235} enc_t;
127 236
128ecb_inline void 237ecb_inline void
129need (enc_t *enc, STRLEN len) 238need (enc_t *enc, STRLEN len)
130{ 239{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 240 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 241 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 242 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 243 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 244 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 245 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
137 } 246 }
138} 247}
139 248
249static void encode_sv (enc_t *enc, SV *sv);
250
140ecb_inline void 251ecb_inline void
141encode_ch (enc_t *enc, char ch) 252encode_ch (enc_t *enc, char ch)
142{ 253{
143 need (enc, 1); 254 need (enc, 1);
144 *enc->cur++ = ch; 255 *enc->cur++ = ch;
145} 256}
146 257
258// used for tags, intregers, element counts and so on
147static void 259static void
148encode_uint (enc_t *enc, int major, UV len) 260encode_uint (enc_t *enc, int major, UV len)
149{ 261{
150 need (enc, 9); 262 need (enc, 9);
151 263
152 if (len < 24) 264 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 265 *enc->cur++ = major | len;
154 else if (len <= 0xff) 266 else if (ecb_expect_true (len <= 0xffU))
155 { 267 {
156 *enc->cur++ = major | 24; 268 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 269 *enc->cur++ = len;
158 } 270 }
159 else if (len <= 0xffff) 271 else if (len <= 0xffffU)
160 { 272 {
161 *enc->cur++ = major | 25; 273 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 274 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 275 *enc->cur++ = len;
164 } 276 }
165 else if (len <= 0xffffffff) 277 else if (len <= 0xffffffffU)
166 { 278 {
167 *enc->cur++ = major | 26; 279 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 280 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 281 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 282 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 283 *enc->cur++ = len;
172 } 284 }
173 else 285 else
174 { 286 {
175 *enc->cur++ = major | 27; 287 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 288 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 289 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 290 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 291 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 292 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 294 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 295 *enc->cur++ = len;
184 } 296 }
185} 297}
186 298
187static void 299// encodes a perl value into a CBOR integer
300ecb_inline void
301encode_int (enc_t *enc, SV *sv)
302{
303 if (SvIsUV (sv))
304 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
305 else if (SvIVX (sv) >= 0)
306 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
307 else
308 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
309}
310
311ecb_inline void
312encode_tag (enc_t *enc, UV tag)
313{
314 encode_uint (enc, MAJOR_TAG, tag);
315}
316
317// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
318ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 319encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 320{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 321 STRLEN ulen = len;
322 U8 *p, *pend = (U8 *)str + len;
323
324 for (p = (U8 *)str; p < pend; ++p)
325 ulen += *p >> 7; // count set high bits
326
327 encode_uint (enc, MAJOR_TEXT, ulen);
328
329 need (enc, ulen);
330 for (p = (U8 *)str; p < pend; ++p)
331 if (*p < 0x80)
332 *enc->cur++ = *p;
333 else
334 {
335 *enc->cur++ = 0xc0 + (*p >> 6);
336 *enc->cur++ = 0x80 + (*p & 63);
337 }
338}
339
340ecb_inline void
341encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
342{
343 if (ecb_expect_false (upgrade_utf8))
344 if (!utf8)
345 {
346 encode_str_utf8 (enc, utf8, str, len);
347 return;
348 }
349
350 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 351 need (enc, len);
192 memcpy (enc->cur, str, len); 352 memcpy (enc->cur, str, len);
193 enc->cur += len; 353 enc->cur += len;
194} 354}
195 355
196static void encode_sv (enc_t *enc, SV *sv); 356ecb_inline void
357encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
358{
359 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
360 {
361 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
362
363 if (SvOK (*svp))
364 {
365 // already registered, use stringref
366 encode_tag (enc, CBOR_TAG_STRINGREF);
367 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
368 return;
369 }
370 else if (len >= minimum_string_length (enc->stringref_idx))
371 {
372 // register only
373 sv_setuv (*svp, enc->stringref_idx);
374 ++enc->stringref_idx;
375 }
376 }
377
378 encode_str (enc, upgrade_utf8, utf8, str, len);
379}
380
381ecb_inline void
382encode_float16 (enc_t *enc, NV nv)
383{
384 need (enc, 1+2);
385
386 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
387
388 uint16_t fp = ecb_float_to_binary16 (nv);
389
390 if (!ecb_big_endian ())
391 fp = ecb_bswap16 (fp);
392
393 memcpy (enc->cur, &fp, 2);
394 enc->cur += 2;
395}
396
397ecb_inline void
398encode_float32 (enc_t *enc, NV nv)
399{
400 need (enc, 1+4);
401
402 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
403
404 uint32_t fp = ecb_float_to_binary32 (nv);
405
406 if (!ecb_big_endian ())
407 fp = ecb_bswap32 (fp);
408
409 memcpy (enc->cur, &fp, 4);
410 enc->cur += 4;
411}
412
413ecb_inline void
414encode_float64 (enc_t *enc, NV nv)
415{
416 need (enc, 1+8);
417
418 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
419
420 uint64_t fp = ecb_double_to_binary64 (nv);
421
422 if (!ecb_big_endian ())
423 fp = ecb_bswap64 (fp);
424
425 memcpy (enc->cur, &fp, 8);
426 enc->cur += 8;
427}
428
429ecb_inline void
430encode_bool (enc_t *enc, int istrue)
431{
432 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
433}
434
435// encodes an arrayref containing key-value pairs as CBOR map
436ecb_inline void
437encode_array_as_map (enc_t *enc, SV *sv)
438{
439 if (enc->depth >= enc->cbor.max_depth)
440 croak (ERR_NESTING_EXCEEDED);
441
442 ++enc->depth;
443
444 // as_map does error checking for us, but we re-check in case
445 // things have changed.
446
447 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
448 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
449
450 AV *av = (AV *)SvRV (sv);
451 int i, len = av_len (av);
452
453 if (!(len & 1))
454 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
455
456 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
457
458 for (i = 0; i <= len; ++i)
459 {
460 SV **svp = av_fetch (av, i, 0);
461 encode_sv (enc, svp ? *svp : &PL_sv_undef);
462 }
463
464 --enc->depth;
465}
466
467ecb_inline void
468encode_forced (enc_t *enc, UV type, SV *sv)
469{
470 switch (type)
471 {
472 case AS_CBOR:
473 {
474 STRLEN len;
475 char *str = SvPVbyte (sv, len);
476
477 need (enc, len);
478 memcpy (enc->cur, str, len);
479 enc->cur += len;
480 }
481 break;
482
483 case AS_BYTES:
484 {
485 STRLEN len;
486 char *str = SvPVbyte (sv, len);
487 encode_strref (enc, 0, 0, str, len);
488 }
489 break;
490
491 case AS_TEXT:
492 {
493 STRLEN len;
494 char *str = SvPVutf8 (sv, len);
495 encode_strref (enc, 1, 1, str, len);
496 }
497 break;
498
499 case AS_INT: encode_int (enc, sv); break;
500
501 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
502 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
503 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
504
505 case AS_MAP: encode_array_as_map (enc, sv); break;
506
507 default:
508 croak ("encountered malformed CBOR::XS::Tagged object");
509 }
510}
197 511
198static void 512static void
199encode_av (enc_t *enc, AV *av) 513encode_av (enc_t *enc, AV *av)
200{ 514{
201 int i, len = av_len (av); 515 int i, len = av_len (av);
203 if (enc->depth >= enc->cbor.max_depth) 517 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 518 croak (ERR_NESTING_EXCEEDED);
205 519
206 ++enc->depth; 520 ++enc->depth;
207 521
208 encode_uint (enc, 0x80, len + 1); 522 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 523
524 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 525 for (i = 0; i <= len; ++i)
211 { 526 {
212 SV **svp = av_fetch (av, i, 0); 527 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 528 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 529 }
530 else
531 for (i = 0; i <= len; ++i)
532 {
533 SV *sv = AvARRAY (av)[i];
534 encode_sv (enc, sv ? sv : &PL_sv_undef);
535 }
215 536
216 --enc->depth; 537 --enc->depth;
217} 538}
218 539
219static void 540static void
227 ++enc->depth; 548 ++enc->depth;
228 549
229 int pairs = hv_iterinit (hv); 550 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 551 int mg = SvMAGICAL (hv);
231 552
232 if (mg) 553 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 554 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 555 else
235 encode_uint (enc, 0xa0, pairs); 556 encode_uint (enc, MAJOR_MAP, pairs);
236 557
237 while ((he = hv_iternext (hv))) 558 while ((he = hv_iternext (hv)))
238 { 559 {
239 if (HeKLEN (he) == HEf_SVKEY) 560 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 561 encode_sv (enc, HeSVKEY (he));
241 else 562 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 563 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 564
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 565 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 566 }
246 567
247 if (mg) 568 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 569 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 570
250 --enc->depth; 571 --enc->depth;
251} 572}
252 573
253// encode objects, arrays and special \0=false and \1=true values. 574// encode objects, arrays and special \0=false and \1=true values.
254static void 575static void
255encode_rv (enc_t *enc, SV *sv) 576encode_rv (enc_t *enc, SV *sv)
256{ 577{
257 svtype svt;
258
259 SvGETMAGIC (sv); 578 SvGETMAGIC (sv);
579
260 svt = SvTYPE (sv); 580 svtype svt = SvTYPE (sv);
261 581
262 if (ecb_expect_false (SvOBJECT (sv))) 582 if (ecb_expect_false (SvOBJECT (sv)))
263 { 583 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 584 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 585 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 590 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 591 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 592 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 593
274 HV *stash = SvSTASH (sv); 594 HV *stash = SvSTASH (sv);
275 GV *method;
276 595
277 if (stash == boolean_stash) 596 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 597 {
598 encode_bool (enc, SvIV (sv));
599 return;
600 }
279 else if (stash == error_stash) 601 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 602 {
603 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
604 return;
605 }
281 else if (stash == tagged_stash) 606 else if (stash == tagged_stash)
282 { 607 {
283 if (svt != SVt_PVAV) 608 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 609 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 610
611 switch (av_len ((AV *)sv))
612 {
613 case 2-1:
614 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 615 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 616 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
617 break;
618
619 case 3-1:
620 // a forced type [value, type, undef]
621 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
622 break;
623
624 default:
625 croak ("encountered malformed CBOR::XS::Tagged object");
626 }
627
628 return;
629 }
630 }
631
632 if (ecb_expect_false (SvREFCNT (sv) > 1)
633 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
634 {
635 if (ecb_expect_false (!enc->shareable))
636 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
637
638 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
639
640 if (SvOK (*svp))
288 } 641 {
642 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
643 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
644 return;
645 }
646 else
647 {
648 sv_setuv (*svp, enc->shareable_idx);
649 ++enc->shareable_idx;
650 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
651 }
652 }
653
654 if (ecb_expect_false (SvOBJECT (sv)))
655 {
656 HV *stash = SvSTASH (sv);
657 GV *method;
658
659 if (enc->cbor.flags & F_FORBID_OBJECTS)
660 croak ("encountered object '%s', but forbid_objects is enabled",
661 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 662 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 663 {
291 dSP; 664 dSP;
292 665
293 ENTER; SAVETMPS; PUSHMARK (SP); 666 ENTER; SAVETMPS;
667 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 668 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 669 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 670
297 PUTBACK; 671 PUTBACK;
298 // G_SCALAR ensures that return value is 1 672 // G_SCALAR ensures that return value is 1
311 } 685 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 686 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 687 {
314 dSP; 688 dSP;
315 689
316 ENTER; SAVETMPS; PUSHMARK (SP); 690 ENTER; SAVETMPS;
691 PUSHMARK (SP);
317 EXTEND (SP, 2); 692 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 693 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 694 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 695 PUSHs (sv_cbor);
321 696
325 700
326 // catch this surprisingly common error 701 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 702 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)); 703 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 704
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 705 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 706 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 707 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 708
334 while (count) 709 {
710 int i;
711
712 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 713 encode_sv (enc, SP[i + 1 - count]);
714
715 SP -= count;
716 }
336 717
337 PUTBACK; 718 PUTBACK;
338 719
339 FREETMPS; LEAVE; 720 FREETMPS; LEAVE;
340 } 721 }
344 } 725 }
345 else if (svt == SVt_PVHV) 726 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 727 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 728 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 729 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 730 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 731 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 732 encode_tag (enc, CBOR_TAG_INDIRECTION);
733 encode_sv (enc, sv);
734 }
369} 735}
370 736
371static void 737static void
372encode_nv (enc_t *enc, SV *sv) 738encode_nv (enc_t *enc, SV *sv)
373{ 739{
374 double nv = SvNVX (sv); 740 double nv = SvNVX (sv);
375 741
376 need (enc, 9); 742 need (enc, 9);
377 743
378 if (ecb_expect_false (nv == (U32)nv)) 744 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 745 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 746 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 747 else if (ecb_expect_false (nv == (float)nv))
382 { 748 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 749 else
394 { 750 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} 751}
406 752
407static void 753static void
408encode_sv (enc_t *enc, SV *sv) 754encode_sv (enc_t *enc, SV *sv)
409{ 755{
411 757
412 if (SvPOKp (sv)) 758 if (SvPOKp (sv))
413 { 759 {
414 STRLEN len; 760 STRLEN len;
415 char *str = SvPV (sv, len); 761 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 762 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 763 }
418 else if (SvNOKp (sv)) 764 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 765 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 766 else if (SvIOKp (sv))
421 { 767 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)) 768 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 769 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 770 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 771 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 772 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 773 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 774 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 775 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 776 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 777}
439 778
440static SV * 779static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 780encode_cbor (SV *scalar, CBOR *cbor)
442{ 781{
443 enc_t enc; 782 enc_t enc = { 0 };
444 783
445 enc.cbor = *cbor; 784 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 785 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 786 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 787 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 788
451 SvPOK_only (enc.sv); 789 SvPOK_only (enc.sv);
790
791 if (cbor->flags & F_PACK_STRINGS)
792 {
793 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
794 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
795 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
796 }
797
452 encode_sv (&enc, scalar); 798 encode_sv (&enc, scalar);
453 799
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 800 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 801 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 802
470 U8 *end; // end of input string 816 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 817 const char *err; // parse error, if != 0
472 CBOR cbor; 818 CBOR cbor;
473 U32 depth; // recursion depth 819 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 820 U32 maxdepth; // recursion depth limit
821 AV *shareable;
822 AV *stringref;
823 SV *decode_tagged;
824 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 825} dec_t;
476 826
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 827// set dec->err to ERRSV
828ecb_cold static void
829err_errsv (dec_t *dec)
830{
831 if (!dec->err)
832 {
833 dec->err_sv = newSVsv (ERRSV);
478 834
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 835 // chop off the trailing \n
836 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
837 *SvEND (dec->err_sv) = 0;
480 838
839 dec->err = SvPVutf8_nolen (dec->err_sv);
840 }
841}
842
843// the following functions are used to reduce code size and help the compiler to optimise
844ecb_cold static void
845err_set (dec_t *dec, const char *reason)
846{
847 if (!dec->err)
848 dec->err = reason;
849}
850
851ecb_cold static void
852err_unexpected_end (dec_t *dec)
853{
854 err_set (dec, "unexpected end of CBOR data");
855}
856
857#define ERR_DO(do) SB do; goto fail; SE
858#define ERR(reason) ERR_DO (err_set (dec, reason))
859#define ERR_ERRSV ERR_DO (err_errsv (dec))
860
861#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
862
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 863#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 864#define DEC_DEC_DEPTH --dec->depth
483 865
484static UV 866static UV
485decode_uint (dec_t *dec) 867decode_uint (dec_t *dec)
486{ 868{
487 switch (*dec->cur & 31) 869 U8 m = *dec->cur & MINOR_MASK;
488 { 870 ++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;
493 871
494 case 24: 872 if (ecb_expect_true (m < LENGTH_EXT1))
873 return m;
874 else if (ecb_expect_true (m == LENGTH_EXT1))
875 {
495 WANT (2); 876 WANT (1);
496 dec->cur += 2; 877 dec->cur += 1;
497 return dec->cur[-1]; 878 return dec->cur[-1];
498 879 }
499 case 25: 880 else if (ecb_expect_true (m == LENGTH_EXT2))
881 {
500 WANT (3); 882 WANT (2);
501 dec->cur += 3; 883 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 884 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 885 | ((UV)dec->cur[-1]);
504 886 }
505 case 26: 887 else if (ecb_expect_true (m == LENGTH_EXT4))
888 {
506 WANT (5); 889 WANT (4);
507 dec->cur += 5; 890 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 891 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 892 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 893 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 894 | ((UV)dec->cur[-1]);
512 895 }
513 case 27: 896 else if (ecb_expect_true (m == LENGTH_EXT8))
897 {
514 WANT (9); 898 WANT (8);
515 dec->cur += 9; 899 dec->cur += 8;
900
901 return
902#if UVSIZE < 8
903 0
904#else
516 return (((UV)dec->cur[-8]) << 56) 905 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 906 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 907 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 908 | (((UV)dec->cur[-5]) << 32)
909#endif
520 | (((UV)dec->cur[-4]) << 24) 910 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 911 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 912 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 913 | ((UV)dec->cur[-1]);
524 914 }
525 default: 915 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 916 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 917
529fail: 918fail:
530 return 0; 919 return 0;
531} 920}
532 921
537{ 926{
538 AV *av = newAV (); 927 AV *av = newAV ();
539 928
540 DEC_INC_DEPTH; 929 DEC_INC_DEPTH;
541 930
542 if ((*dec->cur & 31) == 31) 931 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 932 {
544 ++dec->cur; 933 ++dec->cur;
545 934
546 for (;;) 935 for (;;)
547 { 936 {
548 WANT (1); 937 WANT (1);
549 938
550 if (*dec->cur == (0xe0 | 31)) 939 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 940 {
552 ++dec->cur; 941 ++dec->cur;
553 break; 942 break;
554 } 943 }
555 944
556 av_push (av, decode_sv (dec)); 945 av_push (av, decode_sv (dec));
557 } 946 }
558 } 947 }
559 else 948 else
560 { 949 {
561 int i, len = decode_uint (dec); 950 UV i, len = decode_uint (dec);
562 951
952 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); 953 av_fill (av, len - 1);
564 954
565 for (i = 0; i < len; ++i) 955 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 956 AvARRAY (av)[i] = decode_sv (dec);
567 } 957 }
568 958
569 DEC_DEC_DEPTH; 959 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 960 return newRV_noinc ((SV *)av);
571 961
572fail: 962fail:
573 SvREFCNT_dec (av); 963 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 964 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 965 return &PL_sv_undef;
576} 966}
577 967
968static void
969decode_he (dec_t *dec, HV *hv)
970{
971 // for speed reasons, we specialcase single-string
972 // byte or utf-8 strings as keys, but only when !stringref
973
974 if (ecb_expect_true (!dec->stringref))
975 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
976 {
977 STRLEN len = decode_uint (dec);
978 char *key = (char *)dec->cur;
979
980 WANT (len);
981 dec->cur += len;
982
983 hv_store (hv, key, len, decode_sv (dec), 0);
984
985 return;
986 }
987 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
988 {
989 STRLEN len = decode_uint (dec);
990 char *key = (char *)dec->cur;
991
992 WANT (len);
993 dec->cur += len;
994
995 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
996 if (!cbor_is_utf8_string ((U8 *)key, len))
997 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
998
999 hv_store (hv, key, -len, decode_sv (dec), 0);
1000
1001 return;
1002 }
1003
1004 SV *k = decode_sv (dec);
1005 SV *v = decode_sv (dec);
1006
1007 // we leak memory if uncaught exceptions are thrown by random magical
1008 // methods, and this is hopefully the only place where it can happen,
1009 // so if there is a chance of an exception, take the very slow path.
1010 // since catching exceptions is "undocumented/internal/forbidden" by
1011 // the new p5p powers, we need to call out to a perl function :/
1012 if (ecb_expect_false (SvAMAGIC (k)))
1013 {
1014 dSP;
1015
1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
1018 EXTEND (SP, 3);
1019 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1020 PUSHs (sv_2mortal (k));
1021 PUSHs (sv_2mortal (v));
1022
1023 PUTBACK;
1024 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1025 SPAGAIN;
1026
1027 FREETMPS; LEAVE;
1028
1029 if (SvTRUE (ERRSV))
1030 ERR_ERRSV;
1031
1032 return;
1033 }
1034
1035 hv_store_ent (hv, k, v, 0);
1036 SvREFCNT_dec_NN (k);
1037
1038fail:
1039 ;
1040}
1041
578static SV * 1042static SV *
579decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
580{ 1044{
581 HV *hv = newHV (); 1045 HV *hv = newHV ();
582 1046
583 DEC_INC_DEPTH; 1047 DEC_INC_DEPTH;
584 1048
585 if ((*dec->cur & 31) == 31) 1049 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1050 {
587 ++dec->cur; 1051 ++dec->cur;
588 1052
589 for (;;) 1053 for (;;)
590 { 1054 {
591 WANT (1); 1055 WANT (1);
592 1056
593 if (*dec->cur == (0xe0 | 31)) 1057 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1058 {
595 ++dec->cur; 1059 ++dec->cur;
596 break; 1060 break;
597 } 1061 }
598 1062
599 SV *k = decode_sv (dec); 1063 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 } 1064 }
604 } 1065 }
605 else 1066 else
606 { 1067 {
607 int len = decode_uint (dec); 1068 UV pairs = decode_uint (dec);
608 1069
1070 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1071
609 while (len--) 1072 while (pairs--)
610 { 1073 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 } 1074 }
617 1075
618 DEC_DEC_DEPTH; 1076 DEC_DEC_DEPTH;
619 return newRV_noinc ((SV *)hv); 1077 return newRV_noinc ((SV *)hv);
620 1078
621fail: 1079fail:
622 SvREFCNT_dec (hv); 1080 SvREFCNT_dec_NN (hv);
623 DEC_DEC_DEPTH; 1081 DEC_DEC_DEPTH;
624 return &PL_sv_undef; 1082 return &PL_sv_undef;
625} 1083}
626 1084
627static SV * 1085static SV *
628decode_str (dec_t *dec, int utf8) 1086decode_str (dec_t *dec, int utf8)
629{ 1087{
630 SV *sv = 0; 1088 SV *sv = 0;
631 1089
632 if ((*dec->cur & 31) == 31) 1090 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
633 { 1091 {
1092 // indefinite length strings
634 ++dec->cur; 1093 ++dec->cur;
635 1094
1095 U8 major = *dec->cur & MAJOR_MISC;
1096
636 sv = newSVpvn ("", 0); 1097 sv = newSVpvn ("", 0);
637 1098
638 // not very fast, and certainly not robust against illegal input
639 for (;;) 1099 for (;;)
640 { 1100 {
641 WANT (1); 1101 WANT (1);
642 1102
643 if (*dec->cur == (0xe0 | 31)) 1103 if ((*dec->cur - major) > LENGTH_EXT8)
1104 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
644 { 1105 {
645 ++dec->cur; 1106 ++dec->cur;
646 break; 1107 break;
647 } 1108 }
1109 else
1110 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
648 1111
649 sv_catsv (sv, decode_sv (dec)); 1112 STRLEN len = decode_uint (dec);
1113
1114 WANT (len);
1115 sv_catpvn (sv, dec->cur, len);
1116 dec->cur += len;
650 } 1117 }
651 } 1118 }
652 else 1119 else
653 { 1120 {
654 STRLEN len = decode_uint (dec); 1121 STRLEN len = decode_uint (dec);
655 1122
656 WANT (len); 1123 WANT (len);
657 sv = newSVpvn (dec->cur, len); 1124 sv = newSVpvn (dec->cur, len);
658 dec->cur += len; 1125 dec->cur += len;
1126
1127 if (ecb_expect_false (dec->stringref)
1128 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1129 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
659 } 1130 }
660 1131
661 if (utf8) 1132 if (utf8)
1133 {
1134 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1135 if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv)))
1136 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1137
662 SvUTF8_on (sv); 1138 SvUTF8_on (sv);
1139 }
663 1140
664 return sv; 1141 return sv;
665 1142
666fail: 1143fail:
667 SvREFCNT_dec (sv); 1144 SvREFCNT_dec (sv);
669} 1146}
670 1147
671static SV * 1148static SV *
672decode_tagged (dec_t *dec) 1149decode_tagged (dec_t *dec)
673{ 1150{
1151 SV *sv = 0;
674 UV tag = decode_uint (dec); 1152 UV tag = decode_uint (dec);
1153
1154 WANT (1);
1155
1156 switch (tag)
1157 {
1158 case CBOR_TAG_MAGIC:
675 SV *sv = decode_sv (dec); 1159 sv = decode_sv (dec);
1160 break;
676 1161
677 if (tag == CBOR_TAG_MAGIC) 1162 case CBOR_TAG_INDIRECTION:
678 return sv; 1163 sv = newRV_noinc (decode_sv (dec));
679 else if (tag == CBOR_TAG_PERL_OBJECT) 1164 break;
680 { 1165
1166 case CBOR_TAG_STRINGREF_NAMESPACE:
1167 {
1168 // do not use SAVETMPS/FREETMPS, as these will
1169 // erase mortalised caches, e.g. "shareable"
1170 ENTER;
1171
1172 SAVESPTR (dec->stringref);
1173 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1174
1175 sv = decode_sv (dec);
1176
1177 LEAVE;
1178 }
1179 break;
1180
1181 case CBOR_TAG_STRINGREF:
1182 {
1183 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1184 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1185
1186 UV idx = decode_uint (dec);
1187
1188 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1189 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1190
1191 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1192 }
1193 break;
1194
1195 case CBOR_TAG_VALUE_SHAREABLE:
1196 {
1197 if (ecb_expect_false (!dec->shareable))
1198 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1199
1200 if (ecb_expect_false (dec->cbor.flags & (F_ALLOW_CYCLES | F_ALLOW_WEAK_CYCLES)))
1201 {
1202 // if cycles are allowed, then we store an AV as value
1203 // while it is being decoded, and gather unresolved
1204 // references in it, to be re4solved after decoding.
1205 int idx, i;
1206 AV *av = newAV ();
1207 av_push (dec->shareable, (SV *)av);
1208 idx = AvFILLp (dec->shareable);
1209
1210 sv = decode_sv (dec);
1211
1212 // the AV now contains \undef for all unresolved references,
1213 // so we fix them up here.
1214 for (i = 0; i <= AvFILLp (av); ++i)
1215 SvRV_set (AvARRAY (av)[i], SvREFCNT_inc_NN (SvRV (sv)));
1216
1217 // weaken all recursive references
1218 if (dec->cbor.flags & F_ALLOW_WEAK_CYCLES)
1219 for (i = 0; i <= AvFILLp (av); ++i)
1220 sv_rvweaken (AvARRAY (av)[i]);
1221
1222 // now replace the AV by a reference to the completed value
1223 SvREFCNT_dec_NN ((SV *)av);
1224 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1225 }
1226 else
1227 {
1228 av_push (dec->shareable, &PL_sv_undef);
1229 int idx = AvFILLp (dec->shareable);
1230 sv = decode_sv (dec);
1231 AvARRAY (dec->shareable)[idx] = SvREFCNT_inc_NN (sv);
1232 }
1233 }
1234 break;
1235
1236 case CBOR_TAG_VALUE_SHAREDREF:
1237 {
1238 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1239 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1240
1241 UV idx = decode_uint (dec);
1242
1243 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1244 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1245
1246 sv = AvARRAY (dec->shareable)[idx];
1247
1248 // reference to cycle, we create a new \undef and use that, and also
1249 // registerr it in the AV for later fixing
1250 if (ecb_expect_false (SvTYPE (sv) == SVt_PVAV))
1251 {
1252 AV *av = (AV *)sv;
1253 sv = newRV_noinc (&PL_sv_undef);
1254 av_push (av, SvREFCNT_inc_NN (sv));
1255 }
1256 else if (ecb_expect_false (sv == &PL_sv_undef)) // not yet decoded, but cycles not allowed
1257 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1258 else // we decoded the object earlier, no cycle
1259 sv = newSVsv (sv);
1260 }
1261 break;
1262
1263 case CBOR_TAG_PERL_OBJECT:
1264 {
1265 if (dec->cbor.flags & F_FORBID_OBJECTS)
1266 goto filter;
1267
1268 sv = decode_sv (dec);
1269
681 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1270 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
682 ERR ("corrupted CBOR data (non-array perl object)"); 1271 ERR ("corrupted CBOR data (non-array perl object)");
683 1272
684 AV *av = (AV *)SvRV (sv); 1273 AV *av = (AV *)SvRV (sv);
685 int len = av_len (av) + 1; 1274 int len = av_len (av) + 1;
686 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1275 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
687 1276
688 if (!stash) 1277 if (!stash)
689 ERR ("cannot decode perl-object (package does not exist)"); 1278 ERR ("cannot decode perl-object (package does not exist)");
690 1279
691 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1280 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
692 1281
693 if (!method) 1282 if (!method)
694 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1283 ERR ("cannot decode perl-object (package does not have a THAW method)");
695 1284
696 dSP; 1285 dSP;
697 1286
698 ENTER; SAVETMPS; PUSHMARK (SP); 1287 ENTER; SAVETMPS;
1288 PUSHMARK (SP);
699 EXTEND (SP, len + 1); 1289 EXTEND (SP, len + 1);
700 // we re-bless the reference to get overload and other niceties right 1290 // we re-bless the reference to get overload and other niceties right
701 PUSHs (*av_fetch (av, 0, 1)); 1291 PUSHs (*av_fetch (av, 0, 1));
702 PUSHs (sv_cbor); 1292 PUSHs (sv_cbor);
703 1293
704 int i; 1294 int i;
705 1295
706 for (i = 1; i < len; ++i) 1296 for (i = 1; i < len; ++i)
707 PUSHs (*av_fetch (av, i, 1)); 1297 PUSHs (*av_fetch (av, i, 1));
708 1298
709 PUTBACK; 1299 PUTBACK;
710 call_sv ((SV *)GvCV (method), G_SCALAR); 1300 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
711 SPAGAIN; 1301 SPAGAIN;
712 1302
1303 if (SvTRUE (ERRSV))
1304 {
1305 FREETMPS; LEAVE;
1306 ERR_ERRSV;
1307 }
1308
1309 SvREFCNT_dec_NN (sv);
713 sv = SvREFCNT_inc (POPs); 1310 sv = SvREFCNT_inc (POPs);
714 1311
715 PUTBACK; 1312 PUTBACK;
716 1313
717 FREETMPS; LEAVE; 1314 FREETMPS; LEAVE;
1315 }
1316 break;
718 1317
719 return sv; 1318 default:
720 } 1319 filter:
721 else 1320 {
722 { 1321 SV *tag_sv = newSVuv (tag);
1322
1323 sv = decode_sv (dec);
1324
1325 dSP;
1326 ENTER; SAVETMPS;
1327 PUSHMARK (SP);
1328 EXTEND (SP, 2);
1329 PUSHs (tag_sv);
1330 PUSHs (sv);
1331
1332 PUTBACK;
1333 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1334 SPAGAIN;
1335
1336 if (SvTRUE (ERRSV))
1337 {
1338 SvREFCNT_dec_NN (tag_sv);
1339 FREETMPS; LEAVE;
1340 ERR_ERRSV;
1341 }
1342
1343 if (count)
1344 {
1345 SvREFCNT_dec_NN (tag_sv);
1346 SvREFCNT_dec_NN (sv);
1347 sv = SvREFCNT_inc_NN (TOPs);
1348 SP -= count;
1349 }
1350 else
1351 {
723 AV *av = newAV (); 1352 AV *av = newAV ();
724 av_push (av, newSVuv (tag)); 1353 av_push (av, tag_sv);
725 av_push (av, sv); 1354 av_push (av, sv);
726 1355
727 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1356 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
728 ? cbor_tagged_stash 1357 ? cbor_tagged_stash
729 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1358 : gv_stashpv ("CBOR::XS::Tagged" , 1);
730
731 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1359 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1360 }
1361
1362 PUTBACK;
1363
1364 FREETMPS; LEAVE;
1365 }
1366 break;
732 } 1367 }
1368
1369 return sv;
733 1370
734fail: 1371fail:
735 SvREFCNT_dec (sv); 1372 SvREFCNT_dec (sv);
736 return &PL_sv_undef; 1373 return &PL_sv_undef;
737} 1374}
739static SV * 1376static SV *
740decode_sv (dec_t *dec) 1377decode_sv (dec_t *dec)
741{ 1378{
742 WANT (1); 1379 WANT (1);
743 1380
744 switch (*dec->cur >> 5) 1381 switch (*dec->cur >> MAJOR_SHIFT)
745 { 1382 {
746 case 0: // unsigned int 1383 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
747 return newSVuv (decode_uint (dec)); 1384 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
748 case 1: // negative int 1385 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
749 return newSViv (-1 - (IV)decode_uint (dec)); 1386 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
750 case 2: // octet string 1387 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
751 return decode_str (dec, 0); 1388 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
752 case 3: // utf-8 string 1389 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
753 return decode_str (dec, 1); 1390
754 case 4: // array 1391 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) 1392 switch (*dec->cur++ & MINOR_MASK)
762 { 1393 {
763 case 20: 1394 case SIMPLE_FALSE:
764#if CBOR_SLOW 1395#if CBOR_SLOW
765 types_false = get_bool ("Types::Serialiser::false"); 1396 types_false = get_bool ("Types::Serialiser::false");
766#endif 1397#endif
767 return newSVsv (types_false); 1398 return newSVsv (types_false);
768 case 21: 1399 case SIMPLE_TRUE:
769#if CBOR_SLOW 1400#if CBOR_SLOW
770 types_true = get_bool ("Types::Serialiser::true"); 1401 types_true = get_bool ("Types::Serialiser::true");
771#endif 1402#endif
772 return newSVsv (types_true); 1403 return newSVsv (types_true);
773 case 22: 1404 case SIMPLE_NULL:
774 return newSVsv (&PL_sv_undef); 1405 return newSVsv (&PL_sv_undef);
775 case 23: 1406 case SIMPLE_UNDEF:
776#if CBOR_SLOW 1407#if CBOR_SLOW
777 types_error = get_bool ("Types::Serialiser::error"); 1408 types_error = get_bool ("Types::Serialiser::error");
778#endif 1409#endif
779 return newSVsv (types_error); 1410 return newSVsv (types_error);
780 1411
781 case 25: 1412 case MISC_FLOAT16:
782 { 1413 {
783 WANT (2); 1414 WANT (2);
784 1415
785 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1416 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
786 dec->cur += 2; 1417 dec->cur += 2;
787 1418
788 return newSVnv (ecb_binary16_to_float (fp)); 1419 return newSVnv (ecb_binary16_to_float (fp));
789 } 1420 }
790 1421
791 case 26: 1422 case MISC_FLOAT32:
792 { 1423 {
793 uint32_t fp; 1424 uint32_t fp;
794 WANT (4); 1425 WANT (4);
795 memcpy (&fp, dec->cur, 4); 1426 memcpy (&fp, dec->cur, 4);
796 dec->cur += 4; 1427 dec->cur += 4;
799 fp = ecb_bswap32 (fp); 1430 fp = ecb_bswap32 (fp);
800 1431
801 return newSVnv (ecb_binary32_to_float (fp)); 1432 return newSVnv (ecb_binary32_to_float (fp));
802 } 1433 }
803 1434
804 case 27: 1435 case MISC_FLOAT64:
805 { 1436 {
806 uint64_t fp; 1437 uint64_t fp;
807 WANT (8); 1438 WANT (8);
808 memcpy (&fp, dec->cur, 8); 1439 memcpy (&fp, dec->cur, 8);
809 dec->cur += 8; 1440 dec->cur += 8;
812 fp = ecb_bswap64 (fp); 1443 fp = ecb_bswap64 (fp);
813 1444
814 return newSVnv (ecb_binary64_to_double (fp)); 1445 return newSVnv (ecb_binary64_to_double (fp));
815 } 1446 }
816 1447
817 // 0..19 unassigned 1448 // 0..19 unassigned simple
818 // 24 reserved + unassigned (reserved values are not encodable) 1449 // 24 reserved + unassigned simple (reserved values are not encodable)
1450 // 28-30 unassigned misc
1451 // 31 break code
819 default: 1452 default:
820 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1453 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
821 } 1454 }
822 1455
823 break; 1456 break;
824 } 1457 }
825 1458
828} 1461}
829 1462
830static SV * 1463static SV *
831decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1464decode_cbor (SV *string, CBOR *cbor, char **offset_return)
832{ 1465{
833 dec_t dec; 1466 dec_t dec = { 0 };
834 SV *sv; 1467 SV *sv;
1468 STRLEN len;
1469 char *data = SvPVbyte (string, len);
835 1470
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) 1471 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", 1472 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); 1473 (unsigned long)len, (unsigned long)cbor->max_size);
867 }
868
869 sv_utf8_downgrade (string, 0);
870 1474
871 dec.cbor = *cbor; 1475 dec.cbor = *cbor;
872 dec.cur = (U8 *)SvPVX (string); 1476 dec.cur = (U8 *)data;
873 dec.end = (U8 *)SvEND (string); 1477 dec.end = (U8 *)data + len;
874 dec.err = 0;
875 dec.depth = 0;
876 1478
877 sv = decode_sv (&dec); 1479 sv = decode_sv (&dec);
878 1480
879 if (offset_return) 1481 if (offset_return)
880 *offset_return = dec.cur; 1482 *offset_return = dec.cur;
883 if (dec.cur != dec.end && !dec.err) 1485 if (dec.cur != dec.end && !dec.err)
884 dec.err = "garbage after CBOR object"; 1486 dec.err = "garbage after CBOR object";
885 1487
886 if (dec.err) 1488 if (dec.err)
887 { 1489 {
1490 if (dec.shareable)
1491 {
1492 // need to break cyclic links, which would all be in shareable
1493 int i;
1494 SV **svp;
1495
1496 for (i = av_len (dec.shareable) + 1; i--; )
1497 if ((svp = av_fetch (dec.shareable, i, 0)))
1498 sv_setsv (*svp, &PL_sv_undef);
1499 }
1500
888 SvREFCNT_dec (sv); 1501 SvREFCNT_dec_NN (sv);
1502
1503 if (dec.err_sv)
1504 sv_2mortal (dec.err_sv);
1505
889 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1506 croak ("%s, at offset %ld (octet 0x%02x)", dec.err, (long)(dec.cur - (U8 *)data), (int)(uint8_t)*dec.cur);
890 } 1507 }
891 1508
892 sv = sv_2mortal (sv); 1509 sv = sv_2mortal (sv);
893 1510
894 return sv; 1511 return sv;
895} 1512}
896 1513
1514/////////////////////////////////////////////////////////////////////////////
1515// incremental parser
1516
1517#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1518
1519// returns 0 for notyet, 1 for success or error
1520static int
1521incr_parse (CBOR *self, SV *cborstr)
1522{
1523 STRLEN cur;
1524 SvPV (cborstr, cur);
1525
1526 while (ecb_expect_true (self->incr_need <= cur))
1527 {
1528 // table of integer count bytes
1529 static I8 incr_len[MINOR_MASK + 1] = {
1530 0, 0, 0, 0, 0, 0, 0, 0,
1531 0, 0, 0, 0, 0, 0, 0, 0,
1532 0, 0, 0, 0, 0, 0, 0, 0,
1533 1, 2, 4, 8,-1,-1,-1,-2
1534 };
1535
1536 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1537 U8 m = *p & MINOR_MASK;
1538 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1539 I8 ilen = incr_len[m];
1540
1541 self->incr_need = self->incr_pos + 1;
1542
1543 if (ecb_expect_false (ilen < 0))
1544 {
1545 if (m != MINOR_INDEF)
1546 return 1; // error
1547
1548 if (*p == (MAJOR_MISC | MINOR_INDEF))
1549 {
1550 if (count >= 0)
1551 return 1; // error
1552
1553 count = 1;
1554 }
1555 else
1556 {
1557 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1558 count = -1;
1559 }
1560 }
1561 else
1562 {
1563 self->incr_need += ilen;
1564 if (ecb_expect_false (self->incr_need > cur))
1565 return 0;
1566
1567 int major = *p >> MAJOR_SHIFT;
1568
1569 switch (major)
1570 {
1571 case MAJOR_TAG >> MAJOR_SHIFT:
1572 ++count; // tags merely prefix another value
1573 break;
1574
1575 case MAJOR_BYTES >> MAJOR_SHIFT:
1576 case MAJOR_TEXT >> MAJOR_SHIFT:
1577 case MAJOR_ARRAY >> MAJOR_SHIFT:
1578 case MAJOR_MAP >> MAJOR_SHIFT:
1579 {
1580 UV len;
1581
1582 if (ecb_expect_false (ilen))
1583 {
1584 len = 0;
1585
1586 do {
1587 len = (len << 8) | *++p;
1588 } while (--ilen);
1589 }
1590 else
1591 len = m;
1592
1593 switch (major)
1594 {
1595 case MAJOR_BYTES >> MAJOR_SHIFT:
1596 case MAJOR_TEXT >> MAJOR_SHIFT:
1597 self->incr_need += len;
1598 if (ecb_expect_false (self->incr_need > cur))
1599 return 0;
1600
1601 break;
1602
1603 case MAJOR_MAP >> MAJOR_SHIFT:
1604 len <<= 1;
1605 /* FALLTHROUGH */
1606 case MAJOR_ARRAY >> MAJOR_SHIFT:
1607 if (len)
1608 {
1609 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1610 count = len + 1;
1611 }
1612 break;
1613 }
1614 }
1615 }
1616 }
1617
1618 self->incr_pos = self->incr_need;
1619
1620 if (count > 0)
1621 {
1622 while (!--count)
1623 {
1624 if (!AvFILLp (self->incr_count))
1625 return 1; // done
1626
1627 SvREFCNT_dec_NN (av_pop (self->incr_count));
1628 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1629 }
1630
1631 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1632 }
1633 }
1634
1635 return 0;
1636}
1637
1638
897///////////////////////////////////////////////////////////////////////////// 1639/////////////////////////////////////////////////////////////////////////////
898// XS interface functions 1640// XS interface functions
899 1641
900MODULE = CBOR::XS PACKAGE = CBOR::XS 1642MODULE = CBOR::XS PACKAGE = CBOR::XS
901 1643
909 1651
910 types_true = get_bool ("Types::Serialiser::true" ); 1652 types_true = get_bool ("Types::Serialiser::true" );
911 types_false = get_bool ("Types::Serialiser::false"); 1653 types_false = get_bool ("Types::Serialiser::false");
912 types_error = get_bool ("Types::Serialiser::error"); 1654 types_error = get_bool ("Types::Serialiser::error");
913 1655
1656 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1657
914 sv_cbor = newSVpv ("CBOR", 0); 1658 sv_cbor = newSVpv ("CBOR", 0);
915 SvREADONLY_on (sv_cbor); 1659 SvREADONLY_on (sv_cbor);
1660
1661 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
916} 1662}
917 1663
918PROTOTYPES: DISABLE 1664PROTOTYPES: DISABLE
919 1665
920void CLONE (...) 1666void CLONE (...)
936 ))); 1682 )));
937} 1683}
938 1684
939void shrink (CBOR *self, int enable = 1) 1685void shrink (CBOR *self, int enable = 1)
940 ALIAS: 1686 ALIAS:
941 shrink = F_SHRINK 1687 shrink = F_SHRINK
942 allow_unknown = F_ALLOW_UNKNOWN 1688 allow_unknown = F_ALLOW_UNKNOWN
1689 allow_sharing = F_ALLOW_SHARING
1690 allow_cycles = F_ALLOW_CYCLES
1691 allow_weak_cycles = F_ALLOW_WEAK_CYCLES
1692 forbid_objects = F_FORBID_OBJECTS
1693 pack_strings = F_PACK_STRINGS
1694 text_keys = F_TEXT_KEYS
1695 text_strings = F_TEXT_STRINGS
1696 validate_utf8 = F_VALIDATE_UTF8
943 PPCODE: 1697 PPCODE:
944{ 1698{
945 if (enable) 1699 if (enable)
946 self->flags |= ix; 1700 self->flags |= ix;
947 else 1701 else
950 XPUSHs (ST (0)); 1704 XPUSHs (ST (0));
951} 1705}
952 1706
953void get_shrink (CBOR *self) 1707void get_shrink (CBOR *self)
954 ALIAS: 1708 ALIAS:
955 get_shrink = F_SHRINK 1709 get_shrink = F_SHRINK
956 get_allow_unknown = F_ALLOW_UNKNOWN 1710 get_allow_unknown = F_ALLOW_UNKNOWN
1711 get_allow_sharing = F_ALLOW_SHARING
1712 get_allow_cycles = F_ALLOW_CYCLES
1713 get_allow_weak_cycles = F_ALLOW_WEAK_CYCLES
1714 get_forbid_objects = F_FORBID_OBJECTS
1715 get_pack_strings = F_PACK_STRINGS
1716 get_text_keys = F_TEXT_KEYS
1717 get_text_strings = F_TEXT_STRINGS
1718 get_validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1719 PPCODE:
958 XPUSHs (boolSV (self->flags & ix)); 1720 XPUSHs (boolSV (self->flags & ix));
959 1721
960void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1722void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
961 PPCODE: 1723 PPCODE:
977 CODE: 1739 CODE:
978 RETVAL = self->max_size; 1740 RETVAL = self->max_size;
979 OUTPUT: 1741 OUTPUT:
980 RETVAL 1742 RETVAL
981 1743
982#if 0 //TODO 1744void filter (CBOR *self, SV *filter = 0)
983
984void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
985 PPCODE: 1745 PPCODE:
986{
987 SvREFCNT_dec (self->cb_object); 1746 SvREFCNT_dec (self->filter);
988 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1747 self->filter = filter ? newSVsv (filter) : filter;
989
990 XPUSHs (ST (0)); 1748 XPUSHs (ST (0));
991}
992 1749
993void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1750SV *get_filter (CBOR *self)
994 PPCODE: 1751 CODE:
995{ 1752 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
996 if (!self->cb_sk_object) 1753 OUTPUT:
997 self->cb_sk_object = newHV (); 1754 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 1755
1017void encode (CBOR *self, SV *scalar) 1756void encode (CBOR *self, SV *scalar)
1018 PPCODE: 1757 PPCODE:
1019 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1758 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1020 XPUSHs (scalar); 1759 XPUSHs (scalar);
1033 EXTEND (SP, 2); 1772 EXTEND (SP, 2);
1034 PUSHs (sv); 1773 PUSHs (sv);
1035 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1774 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1036} 1775}
1037 1776
1038#if 0 1777void incr_parse (CBOR *self, SV *cborstr)
1778 ALIAS:
1779 incr_parse_multiple = 1
1780 PPCODE:
1781{
1782 if (SvUTF8 (cborstr))
1783 sv_utf8_downgrade (cborstr, 0);
1784
1785 if (!self->incr_count)
1786 {
1787 self->incr_count = newAV ();
1788 self->incr_pos = 0;
1789 self->incr_need = 1;
1790
1791 av_push (self->incr_count, newSViv (1));
1792 }
1793
1794 do
1795 {
1796 if (!incr_parse (self, cborstr))
1797 {
1798 if (self->incr_need > self->max_size && self->max_size)
1799 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1800 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1801
1802 break;
1803 }
1804
1805 SV *sv;
1806 char *offset;
1807
1808 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1809 XPUSHs (sv);
1810
1811 sv_chop (cborstr, offset);
1812
1813 av_clear (self->incr_count);
1814 av_push (self->incr_count, newSViv (1));
1815
1816 self->incr_pos = 0;
1817 self->incr_need = self->incr_pos + 1;
1818 }
1819 while (ix);
1820}
1821
1822void incr_reset (CBOR *self)
1823 CODE:
1824{
1825 SvREFCNT_dec (self->incr_count);
1826 self->incr_count = 0;
1827}
1039 1828
1040void DESTROY (CBOR *self) 1829void DESTROY (CBOR *self)
1041 CODE: 1830 PPCODE:
1042 SvREFCNT_dec (self->cb_sk_object); 1831 cbor_free (self);
1043 SvREFCNT_dec (self->cb_object);
1044
1045#endif
1046 1832
1047PROTOTYPES: ENABLE 1833PROTOTYPES: ENABLE
1048 1834
1049void encode_cbor (SV *scalar) 1835void encode_cbor (SV *scalar)
1836 ALIAS:
1837 encode_cbor = 0
1838 encode_cbor_sharing = F_ALLOW_SHARING
1050 PPCODE: 1839 PPCODE:
1051{ 1840{
1052 CBOR cbor; 1841 CBOR cbor;
1053 cbor_init (&cbor); 1842 cbor_init (&cbor);
1843 cbor.flags |= ix;
1054 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1844 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1055 XPUSHs (scalar); 1845 XPUSHs (scalar);
1056} 1846}
1057 1847
1058void decode_cbor (SV *cborstr) 1848void decode_cbor (SV *cborstr)
1062 cbor_init (&cbor); 1852 cbor_init (&cbor);
1063 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1853 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1064 XPUSHs (cborstr); 1854 XPUSHs (cborstr);
1065} 1855}
1066 1856
1857#ifdef __AFL_COMPILER
1858
1859void
1860afl_init ()
1861 CODE:
1862 __AFL_INIT ();
1863
1864int
1865afl_loop (unsigned int count = 10000)
1866 CODE:
1867 RETVAL = __AFL_LOOP (count);
1868 OUTPUT:
1869 RETVAL
1870
1871#endif
1872

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines