ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.71 by root, Fri Mar 19 17:30:27 2021 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
13 69
14// known tags 70// known tags
15enum cbor_tag 71enum cbor_tag
16{ 72{
17 // inofficial extensions (pending iana registration) 73 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 81
21 // rfc7049 82 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
28 89
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 94
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 100
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 102};
42 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
43#define F_SHRINK 0x00000200UL 118#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
120#define F_ALLOW_SHARING 0x00000004UL
121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
123#define F_PACK_STRINGS 0x00000020UL
124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
45 127
46#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
47 129
48#define SB do { 130#define SB do {
49#define SE } while (0) 131#define SE } while (0)
61# define CBOR_SLOW 0 143# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
63#endif 145#endif
64 146
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 147static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 149
68typedef struct { 150typedef struct {
69 U32 flags; 151 U32 flags;
70 U32 max_depth; 152 U32 max_depth;
71 STRLEN max_size; 153 STRLEN max_size;
154 SV *filter;
155
156 // for the incremental parser
157 STRLEN incr_pos; // the current offset into the text
158 STRLEN incr_need; // minimum bytes needed to decode
159 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 160} CBOR;
73 161
74ecb_inline void 162ecb_inline void
75cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
76{ 164{
77 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 166 cbor->max_depth = 512;
167}
168
169ecb_inline void
170cbor_free (CBOR *cbor)
171{
172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
79} 174}
80 175
81///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
82// utility functions 177// utility functions
83 178
105 SvPV_renew (sv, SvCUR (sv) + 1); 200 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 201#endif
107 } 202 }
108} 203}
109 204
110///////////////////////////////////////////////////////////////////////////// 205// minimum length of a string to be registered for stringref
111// fp hell 206ecb_inline STRLEN
112 207minimum_string_length (UV idx)
113//TODO 208{
209 return idx <= 23 ? 3
210 : idx <= 0xffU ? 4
211 : idx <= 0xffffU ? 5
212 : idx <= 0xffffffffU ? 7
213 : 11;
214}
114 215
115///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
116// encoder 217// encoder
117 218
118// structure used for encoding CBOR 219// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 222 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 223 char *end; // SvEND (sv)
123 SV *sv; // result scalar 224 SV *sv; // result scalar
124 CBOR cbor; 225 CBOR cbor;
125 U32 depth; // recursion level 226 U32 depth; // recursion level
227 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
228 UV stringref_idx;
229 HV *shareable; // ptr => index, or 0
230 UV shareable_idx;
126} enc_t; 231} enc_t;
127 232
128ecb_inline void 233ecb_inline void
129need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
130{ 235{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 237 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
137 } 242 }
138} 243}
139 244
245static void encode_sv (enc_t *enc, SV *sv);
246
140ecb_inline void 247ecb_inline void
141encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
142{ 249{
143 need (enc, 1); 250 need (enc, 1);
144 *enc->cur++ = ch; 251 *enc->cur++ = ch;
145} 252}
146 253
254// used for tags, intregers, element counts and so on
147static void 255static void
148encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
149{ 257{
150 need (enc, 9); 258 need (enc, 9);
151 259
152 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
154 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
155 { 263 {
156 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 265 *enc->cur++ = len;
158 } 266 }
159 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
160 { 268 {
161 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 271 *enc->cur++ = len;
164 } 272 }
165 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
166 { 274 {
167 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 279 *enc->cur++ = len;
172 } 280 }
173 else 281 else
174 { 282 {
175 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 291 *enc->cur++ = len;
184 } 292 }
185} 293}
186 294
187static void 295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
307ecb_inline void
308encode_tag (enc_t *enc, UV tag)
309{
310 encode_uint (enc, MAJOR_TAG, tag);
311}
312
313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 316{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
336ecb_inline void
337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 347 need (enc, len);
192 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
193 enc->cur += len; 349 enc->cur += len;
194} 350}
195 351
196static void encode_sv (enc_t *enc, SV *sv); 352ecb_inline void
353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
354{
355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
356 {
357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
358
359 if (SvOK (*svp))
360 {
361 // already registered, use stringref
362 encode_tag (enc, CBOR_TAG_STRINGREF);
363 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
364 return;
365 }
366 else if (len >= minimum_string_length (enc->stringref_idx))
367 {
368 // register only
369 sv_setuv (*svp, enc->stringref_idx);
370 ++enc->stringref_idx;
371 }
372 }
373
374 encode_str (enc, upgrade_utf8, utf8, str, len);
375}
376
377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
433encode_array_as_map (enc_t *enc, SV *sv)
434{
435 if (enc->depth >= enc->cbor.max_depth)
436 croak (ERR_NESTING_EXCEEDED);
437
438 ++enc->depth;
439
440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
197 507
198static void 508static void
199encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
200{ 510{
201 int i, len = av_len (av); 511 int i, len = av_len (av);
203 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
205 515
206 ++enc->depth; 516 ++enc->depth;
207 517
208 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 519
520 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
211 { 522 {
212 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 525 }
526 else
527 for (i = 0; i <= len; ++i)
528 {
529 SV *sv = AvARRAY (av)[i];
530 encode_sv (enc, sv ? sv : &PL_sv_undef);
531 }
215 532
216 --enc->depth; 533 --enc->depth;
217} 534}
218 535
219static void 536static void
227 ++enc->depth; 544 ++enc->depth;
228 545
229 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
231 548
232 if (mg) 549 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 551 else
235 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
236 553
237 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
238 { 555 {
239 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
241 else 558 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 560
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 562 }
246 563
247 if (mg) 564 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 566
250 --enc->depth; 567 --enc->depth;
251} 568}
252 569
253// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
254static void 571static void
255encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
256{ 573{
257 svtype svt;
258
259 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
575
260 svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
261 577
262 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
263 { 579 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 580 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 581 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 587 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 589
274 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
275 GV *method;
276 591
277 if (stash == boolean_stash) 592 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
279 else if (stash == error_stash) 597 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
281 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
282 { 603 {
283 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
623
624 return;
625 }
626 }
627
628 if (ecb_expect_false (SvREFCNT (sv) > 1)
629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
630 {
631 if (ecb_expect_false (!enc->shareable))
632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
633
634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
635
636 if (SvOK (*svp))
288 } 637 {
638 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
639 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
640 return;
641 }
642 else
643 {
644 sv_setuv (*svp, enc->shareable_idx);
645 ++enc->shareable_idx;
646 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
647 }
648 }
649
650 if (ecb_expect_false (SvOBJECT (sv)))
651 {
652 HV *stash = SvSTASH (sv);
653 GV *method;
654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 659 {
291 dSP; 660 dSP;
292 661
293 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 664 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 666
297 PUTBACK; 667 PUTBACK;
298 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
311 } 681 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 683 {
314 dSP; 684 dSP;
315 685
316 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
317 EXTEND (SP, 2); 688 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
321 692
325 696
326 // catch this surprisingly common error 697 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 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)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 700
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 704
334 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
336 713
337 PUTBACK; 714 PUTBACK;
338 715
339 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
340 } 717 }
344 } 721 }
345 else if (svt == SVt_PVHV) 722 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 723 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 724 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 725 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 726 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 727 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 728 encode_tag (enc, CBOR_TAG_INDIRECTION);
729 encode_sv (enc, sv);
730 }
369} 731}
370 732
371static void 733static void
372encode_nv (enc_t *enc, SV *sv) 734encode_nv (enc_t *enc, SV *sv)
373{ 735{
374 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
375 737
376 need (enc, 9); 738 need (enc, 9);
377 739
378 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 742 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
382 { 744 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 745 else
394 { 746 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} 747}
406 748
407static void 749static void
408encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
409{ 751{
411 753
412 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
413 { 755 {
414 STRLEN len; 756 STRLEN len;
415 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 759 }
418 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 761 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
421 { 763 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)) 764 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 770 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 771 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 773}
439 774
440static SV * 775static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
442{ 777{
443 enc_t enc; 778 enc_t enc = { 0 };
444 779
445 enc.cbor = *cbor; 780 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 784
451 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
786
787 if (cbor->flags & F_PACK_STRINGS)
788 {
789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
792 }
793
452 encode_sv (&enc, scalar); 794 encode_sv (&enc, scalar);
453 795
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 796 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 797 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 798
470 U8 *end; // end of input string 812 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 813 const char *err; // parse error, if != 0
472 CBOR cbor; 814 CBOR cbor;
473 U32 depth; // recursion depth 815 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
817 AV *shareable;
818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 821} dec_t;
476 822
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
478 830
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
480 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
482#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
483 861
484static UV 862static UV
485decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
486{ 864{
487 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
488 { 866 ++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;
493 867
494 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
495 WANT (2); 872 WANT (1);
496 dec->cur += 2; 873 dec->cur += 1;
497 return dec->cur[-1]; 874 return dec->cur[-1];
498 875 }
499 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
500 WANT (3); 878 WANT (2);
501 dec->cur += 3; 879 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
504 882 }
505 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
506 WANT (5); 885 WANT (4);
507 dec->cur += 5; 886 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
512 891 }
513 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
514 WANT (9); 894 WANT (8);
515 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
516 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
520 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
524 910 }
525 default: 911 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 913
529fail: 914fail:
530 return 0; 915 return 0;
531} 916}
532 917
537{ 922{
538 AV *av = newAV (); 923 AV *av = newAV ();
539 924
540 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
541 926
542 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 928 {
544 ++dec->cur; 929 ++dec->cur;
545 930
546 for (;;) 931 for (;;)
547 { 932 {
548 WANT (1); 933 WANT (1);
549 934
550 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 936 {
552 ++dec->cur; 937 ++dec->cur;
553 break; 938 break;
554 } 939 }
555 940
556 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
557 } 942 }
558 } 943 }
559 else 944 else
560 { 945 {
561 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
562 947
948 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); 949 av_fill (av, len - 1);
564 950
565 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
567 } 953 }
568 954
569 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
571 957
572fail: 958fail:
573 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 961 return &PL_sv_undef;
576} 962}
577 963
964static void
965decode_he (dec_t *dec, HV *hv)
966{
967 // for speed reasons, we specialcase single-string
968 // byte or utf-8 strings as keys, but only when !stringref
969
970 if (ecb_expect_true (!dec->stringref))
971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
972 {
973 STRLEN len = decode_uint (dec);
974 char *key = (char *)dec->cur;
975
976 WANT (len);
977 dec->cur += len;
978
979 hv_store (hv, key, len, decode_sv (dec), 0);
980
981 return;
982 }
983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
984 {
985 STRLEN len = decode_uint (dec);
986 char *key = (char *)dec->cur;
987
988 WANT (len);
989 dec->cur += len;
990
991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
992 if (!is_utf8_string ((U8 *)key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
994
995 hv_store (hv, key, -len, decode_sv (dec), 0);
996
997 return;
998 }
999
1000 SV *k = decode_sv (dec);
1001 SV *v = decode_sv (dec);
1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
1031 hv_store_ent (hv, k, v, 0);
1032 SvREFCNT_dec_NN (k);
1033
1034fail:
1035 ;
1036}
1037
578static SV * 1038static SV *
579decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
580{ 1040{
581 HV *hv = newHV (); 1041 HV *hv = newHV ();
582 1042
583 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
584 1044
585 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1046 {
587 ++dec->cur; 1047 ++dec->cur;
588 1048
589 for (;;) 1049 for (;;)
590 { 1050 {
591 WANT (1); 1051 WANT (1);
592 1052
593 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1054 {
595 ++dec->cur; 1055 ++dec->cur;
596 break; 1056 break;
597 } 1057 }
598 1058
599 SV *k = decode_sv (dec); 1059 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 1060 }
605 } 1061 }
606 else 1062 else
607 { 1063 {
608 int len = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
609 1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1067
610 while (len--) 1068 while (pairs--)
611 { 1069 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 1070 }
619 1071
620 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
622 1074
623fail: 1075fail:
624 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
625 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
626 return &PL_sv_undef; 1078 return &PL_sv_undef;
627} 1079}
628 1080
629static SV * 1081static SV *
630decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
631{ 1083{
632 SV *sv = 0; 1084 SV *sv = 0;
633 1085
634 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
635 { 1087 {
1088 // indefinite length strings
636 ++dec->cur; 1089 ++dec->cur;
637 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
638 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
639 1094
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 1095 for (;;)
642 { 1096 {
643 WANT (1); 1097 WANT (1);
644 1098
645 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 1101 {
647 ++dec->cur; 1102 ++dec->cur;
648 break; 1103 break;
649 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 1107
651 sv_catsv (sv, decode_sv (dec)); 1108 STRLEN len = decode_uint (dec);
1109
1110 WANT (len);
1111 sv_catpvn (sv, dec->cur, len);
1112 dec->cur += len;
652 } 1113 }
653 } 1114 }
654 else 1115 else
655 { 1116 {
656 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
657 1118
658 WANT (len); 1119 WANT (len);
659 sv = newSVpvn (dec->cur, len); 1120 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 1121 dec->cur += len;
1122
1123 if (ecb_expect_false (dec->stringref)
1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 1126 }
662 1127
663 if (utf8) 1128 if (utf8)
1129 {
1130 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1131 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1132 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1133
664 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
665 1136
666 return sv; 1137 return sv;
667 1138
668fail: 1139fail:
669 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
671} 1142}
672 1143
673static SV * 1144static SV *
674decode_tagged (dec_t *dec) 1145decode_tagged (dec_t *dec)
675{ 1146{
1147 SV *sv = 0;
676 UV tag = decode_uint (dec); 1148 UV tag = decode_uint (dec);
1149
1150 WANT (1);
1151
1152 switch (tag)
1153 {
1154 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 1155 sv = decode_sv (dec);
1156 break;
678 1157
679 if (tag == CBOR_TAG_MAGIC) 1158 case CBOR_TAG_INDIRECTION:
680 return sv; 1159 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 1160 break;
682 { 1161
1162 case CBOR_TAG_STRINGREF_NAMESPACE:
1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
1166 ENTER;
1167
1168 SAVESPTR (dec->stringref);
1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1170
1171 sv = decode_sv (dec);
1172
1173 LEAVE;
1174 }
1175 break;
1176
1177 case CBOR_TAG_STRINGREF:
1178 {
1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1181
1182 UV idx = decode_uint (dec);
1183
1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1186
1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1188 }
1189 break;
1190
1191 case CBOR_TAG_VALUE_SHAREABLE:
1192 {
1193 if (ecb_expect_false (!dec->shareable))
1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
1198 sv = newSV (0);
1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1200
1201 SV *osv = decode_sv (dec);
1202 sv_setsv (sv, osv);
1203 SvREFCNT_dec_NN (osv);
1204 }
1205 else
1206 {
1207 av_push (dec->shareable, &PL_sv_undef);
1208 int idx = AvFILLp (dec->shareable);
1209 sv = decode_sv (dec);
1210 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1211 }
1212 }
1213 break;
1214
1215 case CBOR_TAG_VALUE_SHAREDREF:
1216 {
1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1219
1220 UV idx = decode_uint (dec);
1221
1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1224
1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1226
1227 if (sv == &PL_sv_undef)
1228 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1229 }
1230 break;
1231
1232 case CBOR_TAG_PERL_OBJECT:
1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
1237 sv = decode_sv (dec);
1238
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
685 1241
686 AV *av = (AV *)SvRV (sv); 1242 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 1243 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1244 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 1245
690 if (!stash) 1246 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 1247 ERR ("cannot decode perl-object (package does not exist)");
692 1248
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1249 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 1250
695 if (!method) 1251 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1252 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 1253
698 dSP; 1254 dSP;
699 1255
700 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1259 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
705 1262
706 int i; 1263 int i;
707 1264
708 for (i = 1; i < len; ++i) 1265 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1266 PUSHs (*av_fetch (av, i, 1));
710 1267
711 PUTBACK; 1268 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1269 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1270 SPAGAIN;
714 1271
1272 if (SvTRUE (ERRSV))
1273 {
1274 FREETMPS; LEAVE;
1275 ERR_ERRSV;
1276 }
1277
1278 SvREFCNT_dec_NN (sv);
715 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
716 1280
717 PUTBACK; 1281 PUTBACK;
718 1282
719 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
1284 }
1285 break;
720 1286
721 return sv; 1287 default:
722 } 1288 filter:
723 else 1289 {
724 { 1290 SV *tag_sv = newSVuv (tag);
1291
1292 sv = decode_sv (dec);
1293
1294 dSP;
1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1297 EXTEND (SP, 2);
1298 PUSHs (tag_sv);
1299 PUSHs (sv);
1300
1301 PUTBACK;
1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1303 SPAGAIN;
1304
1305 if (SvTRUE (ERRSV))
1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1308 FREETMPS; LEAVE;
1309 ERR_ERRSV;
1310 }
1311
1312 if (count)
1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1315 SvREFCNT_dec_NN (sv);
1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1318 }
1319 else
1320 {
725 AV *av = newAV (); 1321 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
727 av_push (av, sv); 1323 av_push (av, sv);
728 1324
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1328 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1329 }
1330
1331 PUTBACK;
1332
1333 FREETMPS; LEAVE;
1334 }
1335 break;
734 } 1336 }
1337
1338 return sv;
735 1339
736fail: 1340fail:
737 SvREFCNT_dec (sv); 1341 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1342 return &PL_sv_undef;
739} 1343}
741static SV * 1345static SV *
742decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
743{ 1347{
744 WANT (1); 1348 WANT (1);
745 1349
746 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1351 {
748 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1359
756 case 4: // array 1360 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
764 { 1362 {
765 case 20: 1363 case SIMPLE_FALSE:
766#if CBOR_SLOW 1364#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
768#endif 1366#endif
769 return newSVsv (types_false); 1367 return newSVsv (types_false);
770 case 21: 1368 case SIMPLE_TRUE:
771#if CBOR_SLOW 1369#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
773#endif 1371#endif
774 return newSVsv (types_true); 1372 return newSVsv (types_true);
775 case 22: 1373 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
777 case 23: 1375 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1376#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
780#endif 1378#endif
781 return newSVsv (types_error); 1379 return newSVsv (types_error);
782 1380
783 case 25: 1381 case MISC_FLOAT16:
784 { 1382 {
785 WANT (2); 1383 WANT (2);
786 1384
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1386 dec->cur += 2;
789 1387
790 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
791 } 1389 }
792 1390
793 case 26: 1391 case MISC_FLOAT32:
794 { 1392 {
795 uint32_t fp; 1393 uint32_t fp;
796 WANT (4); 1394 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1396 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
802 1400
803 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
804 } 1402 }
805 1403
806 case 27: 1404 case MISC_FLOAT64:
807 { 1405 {
808 uint64_t fp; 1406 uint64_t fp;
809 WANT (8); 1407 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1409 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
815 1413
816 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
817 } 1415 }
818 1416
819 // 0..19 unassigned 1417 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1418 // 24 reserved + unassigned simple (reserved values are not encodable)
1419 // 28-30 unassigned misc
1420 // 31 break code
821 default: 1421 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1423 }
824 1424
825 break; 1425 break;
826 } 1426 }
827 1427
830} 1430}
831 1431
832static SV * 1432static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1434{
835 dec_t dec; 1435 dec_t dec = { 0 };
836 SV *sv; 1436 SV *sv;
1437 STRLEN len;
1438 char *data = SvPVbyte (string, len);
837 1439
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1441 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1442 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1443
873 dec.cbor = *cbor; 1444 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1445 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1446 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1447
879 sv = decode_sv (&dec); 1448 sv = decode_sv (&dec);
880 1449
881 if (offset_return) 1450 if (offset_return)
882 *offset_return = dec.cur; 1451 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
887 1456
888 if (dec.err) 1457 if (dec.err)
889 { 1458 {
1459 if (dec.shareable)
1460 {
1461 // need to break cyclic links, which would all be in shareable
1462 int i;
1463 SV **svp;
1464
1465 for (i = av_len (dec.shareable) + 1; i--; )
1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1467 sv_setsv (*svp, &PL_sv_undef);
1468 }
1469
890 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1476 }
893 1477
894 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
895 1479
896 return sv; 1480 return sv;
897} 1481}
898 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 /* FALLTHROUGH */
1575 case MAJOR_ARRAY >> MAJOR_SHIFT:
1576 if (len)
1577 {
1578 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1579 count = len + 1;
1580 }
1581 break;
1582 }
1583 }
1584 }
1585 }
1586
1587 self->incr_pos = self->incr_need;
1588
1589 if (count > 0)
1590 {
1591 while (!--count)
1592 {
1593 if (!AvFILLp (self->incr_count))
1594 return 1; // done
1595
1596 SvREFCNT_dec_NN (av_pop (self->incr_count));
1597 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1598 }
1599
1600 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1601 }
1602 }
1603
1604 return 0;
1605}
1606
1607
899///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1609// XS interface functions
901 1610
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1612
911 1620
912 types_true = get_bool ("Types::Serialiser::true" ); 1621 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1622 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1623 types_error = get_bool ("Types::Serialiser::error");
915 1624
1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1626
916 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
918} 1631}
919 1632
920PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
921 1634
922void CLONE (...) 1635void CLONE (...)
940 1653
941void shrink (CBOR *self, int enable = 1) 1654void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1655 ALIAS:
943 shrink = F_SHRINK 1656 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1657 allow_unknown = F_ALLOW_UNKNOWN
1658 allow_sharing = F_ALLOW_SHARING
1659 allow_cycles = F_ALLOW_CYCLES
1660 forbid_objects = F_FORBID_OBJECTS
1661 pack_strings = F_PACK_STRINGS
1662 text_keys = F_TEXT_KEYS
1663 text_strings = F_TEXT_STRINGS
1664 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1665 PPCODE:
946{ 1666{
947 if (enable) 1667 if (enable)
948 self->flags |= ix; 1668 self->flags |= ix;
949 else 1669 else
954 1674
955void get_shrink (CBOR *self) 1675void get_shrink (CBOR *self)
956 ALIAS: 1676 ALIAS:
957 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1678 get_allow_unknown = F_ALLOW_UNKNOWN
1679 get_allow_sharing = F_ALLOW_SHARING
1680 get_allow_cycles = F_ALLOW_CYCLES
1681 get_forbid_objects = F_FORBID_OBJECTS
1682 get_pack_strings = F_PACK_STRINGS
1683 get_text_keys = F_TEXT_KEYS
1684 get_text_strings = F_TEXT_STRINGS
1685 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1686 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
961 1688
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1690 PPCODE:
979 CODE: 1706 CODE:
980 RETVAL = self->max_size; 1707 RETVAL = self->max_size;
981 OUTPUT: 1708 OUTPUT:
982 RETVAL 1709 RETVAL
983 1710
984#if 0 //TODO 1711void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1712 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1713 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1714 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1715 XPUSHs (ST (0));
993}
994 1716
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1717SV *get_filter (CBOR *self)
996 PPCODE: 1718 CODE:
997{ 1719 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1720 OUTPUT:
999 self->cb_sk_object = newHV (); 1721 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1722
1019void encode (CBOR *self, SV *scalar) 1723void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1724 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1725 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1726 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1036 PUSHs (sv); 1740 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1742}
1039 1743
1040#if 0 1744void incr_parse (CBOR *self, SV *cborstr)
1745 ALIAS:
1746 incr_parse_multiple = 1
1747 PPCODE:
1748{
1749 if (SvUTF8 (cborstr))
1750 sv_utf8_downgrade (cborstr, 0);
1751
1752 if (!self->incr_count)
1753 {
1754 self->incr_count = newAV ();
1755 self->incr_pos = 0;
1756 self->incr_need = 1;
1757
1758 av_push (self->incr_count, newSViv (1));
1759 }
1760
1761 do
1762 {
1763 if (!incr_parse (self, cborstr))
1764 {
1765 if (self->incr_need > self->max_size && self->max_size)
1766 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1767 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1768
1769 break;
1770 }
1771
1772 SV *sv;
1773 char *offset;
1774
1775 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1776 XPUSHs (sv);
1777
1778 sv_chop (cborstr, offset);
1779
1780 av_clear (self->incr_count);
1781 av_push (self->incr_count, newSViv (1));
1782
1783 self->incr_pos = 0;
1784 self->incr_need = self->incr_pos + 1;
1785 }
1786 while (ix);
1787}
1788
1789void incr_reset (CBOR *self)
1790 CODE:
1791{
1792 SvREFCNT_dec (self->incr_count);
1793 self->incr_count = 0;
1794}
1041 1795
1042void DESTROY (CBOR *self) 1796void DESTROY (CBOR *self)
1043 CODE: 1797 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1798 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1799
1049PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
1050 1801
1051void encode_cbor (SV *scalar) 1802void encode_cbor (SV *scalar)
1803 ALIAS:
1804 encode_cbor = 0
1805 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1806 PPCODE:
1053{ 1807{
1054 CBOR cbor; 1808 CBOR cbor;
1055 cbor_init (&cbor); 1809 cbor_init (&cbor);
1810 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1811 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1812 XPUSHs (scalar);
1058} 1813}
1059 1814
1060void decode_cbor (SV *cborstr) 1815void decode_cbor (SV *cborstr)
1064 cbor_init (&cbor); 1819 cbor_init (&cbor);
1065 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1066 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1067} 1822}
1068 1823
1824#ifdef __AFL_COMPILER
1825
1826void
1827afl_init ()
1828 CODE:
1829 __AFL_INIT ();
1830
1831int
1832afl_loop (unsigned int count = 10000)
1833 CODE:
1834 RETVAL = __AFL_LOOP (count);
1835 OUTPUT:
1836 RETVAL
1837
1838#endif
1839

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines