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 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.70 by root, Fri Dec 4 02:57:14 2020 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 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};
24 69
25// known tags 70// known tags
26enum cbor_tag 71enum cbor_tag
27{ 72{
28 // inofficial extensions (pending iana registration) 73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 81
37 // rfc7049 82 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
44 89
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 94
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 100
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 102};
58 103
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
59#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 120#define F_ALLOW_SHARING 0x00000004UL
121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 123#define F_PACK_STRINGS 0x00000020UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
64 127
65#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
66 129
67#define SB do { 130#define SB do {
68#define SE } while (0) 131#define SE } while (0)
80# define CBOR_SLOW 0 143# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
82#endif 145#endif
83 146
84static 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::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 149
87typedef struct { 150typedef struct {
88 U32 flags; 151 U32 flags;
89 U32 max_depth; 152 U32 max_depth;
90 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.
91} CBOR; 160} CBOR;
92 161
93ecb_inline void 162ecb_inline void
94cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
95{ 164{
96 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
97 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);
98} 174}
99 175
100///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
101// utility functions 177// utility functions
102 178
122 SvPV_shrink_to_cur (sv); 198 SvPV_shrink_to_cur (sv);
123#elif defined (SvPV_renew) 199#elif defined (SvPV_renew)
124 SvPV_renew (sv, SvCUR (sv) + 1); 200 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 201#endif
126 } 202 }
203}
204
205// minimum length of a string to be registered for stringref
206ecb_inline int
207minimum_string_length (UV idx)
208{
209 return idx <= 23 ? 3
210 : idx <= 0xffU ? 4
211 : idx <= 0xffffU ? 5
212 : idx <= 0xffffffffU ? 7
213 : 11;
127} 214}
128 215
129///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
130// encoder 217// encoder
131 218
144} enc_t; 231} enc_t;
145 232
146ecb_inline void 233ecb_inline void
147need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
148{ 235{
149 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150 { 237 {
151 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
154 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
155 } 242 }
156} 243}
157 244
245static void encode_sv (enc_t *enc, SV *sv);
246
158ecb_inline void 247ecb_inline void
159encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
160{ 249{
161 need (enc, 1); 250 need (enc, 1);
162 *enc->cur++ = ch; 251 *enc->cur++ = ch;
163} 252}
164 253
254// used for tags, intregers, element counts and so on
165static void 255static void
166encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
167{ 257{
168 need (enc, 9); 258 need (enc, 9);
169 259
170 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
171 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
172 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
173 { 263 {
174 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
175 *enc->cur++ = len; 265 *enc->cur++ = len;
176 } 266 }
177 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
178 { 268 {
179 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
180 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
181 *enc->cur++ = len; 271 *enc->cur++ = len;
182 } 272 }
183 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
184 { 274 {
185 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
186 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
187 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
188 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
189 *enc->cur++ = len; 279 *enc->cur++ = len;
190 } 280 }
191 else 281 else
192 { 282 {
193 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
194 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
195 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
196 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
197 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
198 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
200 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
201 *enc->cur++ = len; 291 *enc->cur++ = len;
202 } 292 }
203} 293}
204 294
205static 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
206encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207{ 316{
208 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);
209 need (enc, len); 347 need (enc, len);
210 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
211 enc->cur += len; 349 enc->cur += len;
212} 350}
213 351
214ecb_inline void 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
215encode_tag (enc_t *enc, UV tag) 378encode_float16 (enc_t *enc, NV nv)
216{ 379{
217 encode_uint (enc, 0xc0, tag); 380 need (enc, 1+2);
218}
219 381
220static void encode_sv (enc_t *enc, SV *sv); 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}
221 507
222static void 508static void
223encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
224{ 510{
225 int i, len = av_len (av); 511 int i, len = av_len (av);
227 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
228 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
229 515
230 ++enc->depth; 516 ++enc->depth;
231 517
232 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
233 519
520 if (ecb_expect_false (SvMAGICAL (av)))
234 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
235 { 522 {
236 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
237 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
238 } 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 }
239 532
240 --enc->depth; 533 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250} 534}
251 535
252static void 536static void
253encode_hv (enc_t *enc, HV *hv) 537encode_hv (enc_t *enc, HV *hv)
254{ 538{
260 ++enc->depth; 544 ++enc->depth;
261 545
262 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
263 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
264 548
265 if (mg) 549 if (ecb_expect_false (mg))
266 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267 else 551 else
268 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
269 553
270 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
271 { 555 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
276 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/ 557 encode_sv (enc, HeSVKEY (he));
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
293 else 558 else
294 encode_he (enc, he); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295 560
296 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));
297 } 562 }
298 563
299 if (mg) 564 if (ecb_expect_false (mg))
300 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301 566
302 --enc->depth; 567 --enc->depth;
303} 568}
304 569
305// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
306static void 571static void
307encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
308{ 573{
309 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332 575
333 svtype svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
334 577
335 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
336 { 579 {
343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344 ? cbor_tagged_stash 587 ? cbor_tagged_stash
345 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
346 589
347 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
348 GV *method;
349 591
350 if (stash == boolean_stash) 592 if (stash == boolean_stash)
351 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
352 else if (stash == error_stash) 597 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
354 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
355 { 603 {
356 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
357 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
358 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360 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))
361 } 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))));
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 { 659 {
364 dSP; 660 dSP;
365 661
366 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
367 // 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
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369 666
370 PUTBACK; 667 PUTBACK;
371 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
384 } 681 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 { 683 {
387 dSP; 684 dSP;
388 685
389 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
390 EXTEND (SP, 2); 688 EXTEND (SP, 2);
391 // 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
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
394 692
399 // catch this surprisingly common error 697 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 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));
402 700
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406 704
407 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
408 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
409 713
410 PUTBACK; 714 PUTBACK;
411 715
412 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
413 } 717 }
431{ 735{
432 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
433 737
434 need (enc, 9); 738 need (enc, 9);
435 739
436 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
437 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438 //TODO: maybe I32? 742 //TODO: maybe I32?
439 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
440 { 744 encode_float32 (enc, nv);
441 uint32_t fp = ecb_float_to_binary32 (nv);
442
443 *enc->cur++ = 0xe0 | 26;
444
445 if (!ecb_big_endian ())
446 fp = ecb_bswap32 (fp);
447
448 memcpy (enc->cur, &fp, 4);
449 enc->cur += 4;
450 }
451 else 745 else
452 { 746 encode_float64 (enc, nv);
453 uint64_t fp = ecb_double_to_binary64 (nv);
454
455 *enc->cur++ = 0xe0 | 27;
456
457 if (!ecb_big_endian ())
458 fp = ecb_bswap64 (fp);
459
460 memcpy (enc->cur, &fp, 8);
461 enc->cur += 8;
462 }
463} 747}
464 748
465static void 749static void
466encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
467{ 751{
469 753
470 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
471 { 755 {
472 STRLEN len; 756 STRLEN len;
473 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
474 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475 } 759 }
476 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
477 encode_nv (enc, sv); 761 encode_nv (enc, sv);
478 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
479 { 763 encode_int (enc, sv);
480 if (SvIsUV (sv))
481 encode_uint (enc, 0x00, SvUVX (sv));
482 else if (SvIVX (sv) >= 0)
483 encode_uint (enc, 0x00, SvIVX (sv));
484 else
485 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
486 }
487 else if (SvROK (sv)) 764 else if (SvROK (sv))
488 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
489 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
490 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493 else 770 else
494 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",
495 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496} 773}
497 774
498static SV * 775static SV *
499encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
500{ 777{
501 enc_t enc = { }; 778 enc_t enc = { 0 };
502 779
503 enc.cbor = *cbor; 780 enc.cbor = *cbor;
504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
506 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
507 784
508 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
509 786
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)) 787 if (cbor->flags & F_PACK_STRINGS)
511 { 788 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 } 792 }
537 CBOR cbor; 814 CBOR cbor;
538 U32 depth; // recursion depth 815 U32 depth; // recursion depth
539 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
540 AV *shareable; 817 AV *shareable;
541 AV *stringref; 818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
542} dec_t; 821} dec_t;
543 822
544#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);
545 830
546#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;
547 834
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
548#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)
549#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
550 861
551static UV 862static UV
552decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
553{ 864{
554 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
555 { 866 ++dec->cur;
556 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559 return *dec->cur++ & 31;
560 867
561 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
562 WANT (2); 872 WANT (1);
563 dec->cur += 2; 873 dec->cur += 1;
564 return dec->cur[-1]; 874 return dec->cur[-1];
565 875 }
566 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
567 WANT (3); 878 WANT (2);
568 dec->cur += 3; 879 dec->cur += 2;
569 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
570 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
571 882 }
572 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
573 WANT (5); 885 WANT (4);
574 dec->cur += 5; 886 dec->cur += 4;
575 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
576 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
577 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
578 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
579 891 }
580 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
581 WANT (9); 894 WANT (8);
582 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
583 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
584 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
585 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
586 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
587 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
588 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
589 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
590 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
591 910 }
592 default: 911 else
593 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594 }
595 913
596fail: 914fail:
597 return 0; 915 return 0;
598} 916}
599 917
604{ 922{
605 AV *av = newAV (); 923 AV *av = newAV ();
606 924
607 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
608 926
609 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610 { 928 {
611 ++dec->cur; 929 ++dec->cur;
612 930
613 for (;;) 931 for (;;)
614 { 932 {
615 WANT (1); 933 WANT (1);
616 934
617 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
618 { 936 {
619 ++dec->cur; 937 ++dec->cur;
620 break; 938 break;
621 } 939 }
622 940
623 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
624 } 942 }
625 } 943 }
626 else 944 else
627 { 945 {
628 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
629 947
948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630 av_fill (av, len - 1); 949 av_fill (av, len - 1);
631 950
632 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
633 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
634 } 953 }
635 954
636 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
637 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
638 957
639fail: 958fail:
640 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
641 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
642 return &PL_sv_undef; 961 return &PL_sv_undef;
643} 962}
644 963
645static void 964static void
646decode_he (dec_t *dec, HV *hv) 965decode_he (dec_t *dec, HV *hv)
647{ 966{
648 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys. 968 // byte or utf-8 strings as keys, but only when !stringref
650 969
651 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 970 if (ecb_expect_true (!dec->stringref))
971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652 { 972 {
653 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
654 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
655 975
976 WANT (len);
656 dec->cur += len; 977 dec->cur += len;
657 978
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0); 979 hv_store (hv, key, len, decode_sv (dec), 0);
980
981 return;
662 } 982 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664 { 984 {
665 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
666 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
667 987
988 WANT (len);
668 dec->cur += len; 989 dec->cur += len;
669 990
670 if (ecb_expect_false (dec->stringref)) 991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 992 if (!is_utf8_string (key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672 994
673 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
996
997 return;
674 } 998 }
675 else 999
676 {
677 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
679 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
680 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
682 } 1033
1034fail:
1035 ;
683} 1036}
684 1037
685static SV * 1038static SV *
686decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
687{ 1040{
688 HV *hv = newHV (); 1041 HV *hv = newHV ();
689 1042
690 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
691 1044
692 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693 { 1046 {
694 ++dec->cur; 1047 ++dec->cur;
695 1048
696 for (;;) 1049 for (;;)
697 { 1050 {
698 WANT (1); 1051 WANT (1);
699 1052
700 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
701 { 1054 {
702 ++dec->cur; 1055 ++dec->cur;
703 break; 1056 break;
704 } 1057 }
705 1058
706 decode_he (dec, hv); 1059 decode_he (dec, hv);
707 } 1060 }
708 } 1061 }
709 else 1062 else
710 { 1063 {
711 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
712 1067
713 while (pairs--) 1068 while (pairs--)
714 decode_he (dec, hv); 1069 decode_he (dec, hv);
715 } 1070 }
716 1071
717 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
718 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
719 1074
720fail: 1075fail:
721 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
722 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 1078 return &PL_sv_undef;
724} 1079}
725 1080
726static SV * 1081static SV *
727decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
728{ 1083{
729 SV *sv = 0; 1084 SV *sv = 0;
730 1085
731 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
732 { 1087 {
1088 // indefinite length strings
733 ++dec->cur; 1089 ++dec->cur;
734 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
735 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
736 1094
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 1095 for (;;)
739 { 1096 {
740 WANT (1); 1097 WANT (1);
741 1098
742 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 1101 {
744 ++dec->cur; 1102 ++dec->cur;
745 break; 1103 break;
746 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 1107
748 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;
749 } 1113 }
750 } 1114 }
751 else 1115 else
752 { 1116 {
753 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
754 1118
755 WANT (len); 1119 WANT (len);
756 sv = newSVpvn (dec->cur, len); 1120 sv = newSVpvn (dec->cur, len);
757 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));
758 } 1126 }
759 1127
760 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
761 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
762 1135 }
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765 1136
766 return sv; 1137 return sv;
767 1138
768fail: 1139fail:
769 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
788 sv = newRV_noinc (decode_sv (dec)); 1159 sv = newRV_noinc (decode_sv (dec));
789 break; 1160 break;
790 1161
791 case CBOR_TAG_STRINGREF_NAMESPACE: 1162 case CBOR_TAG_STRINGREF_NAMESPACE:
792 { 1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
793 ENTER; SAVETMPS; 1166 ENTER;
794 1167
795 SAVESPTR (dec->stringref); 1168 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797 1170
798 sv = decode_sv (dec); 1171 sv = decode_sv (dec);
799 1172
800 FREETMPS; LEAVE; 1173 LEAVE;
801 } 1174 }
802 break; 1175 break;
803 1176
804 case CBOR_TAG_STRINGREF: 1177 case CBOR_TAG_STRINGREF:
805 { 1178 {
806 if ((*dec->cur >> 5) != 0) 1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808 1181
809 UV idx = decode_uint (dec); 1182 UV idx = decode_uint (dec);
810 1183
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
812 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813 1186
814 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815 } 1188 }
816 break; 1189 break;
818 case CBOR_TAG_VALUE_SHAREABLE: 1191 case CBOR_TAG_VALUE_SHAREABLE:
819 { 1192 {
820 if (ecb_expect_false (!dec->shareable)) 1193 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822 1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
823 sv = newSV (0); 1198 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825 1200
826 SV *osv = decode_sv (dec); 1201 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv); 1202 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (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 }
829 } 1212 }
830 break; 1213 break;
831 1214
832 case CBOR_TAG_VALUE_SHAREDREF: 1215 case CBOR_TAG_VALUE_SHAREDREF:
833 { 1216 {
834 if ((*dec->cur >> 5) != 0) 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836 1219
837 UV idx = decode_uint (dec); 1220 UV idx = decode_uint (dec);
838 1221
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841 1224
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 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");
843 } 1229 }
844 break; 1230 break;
845 1231
846 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
847 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
848 sv = decode_sv (dec); 1237 sv = decode_sv (dec);
849 1238
850 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
852 1241
862 if (!method) 1251 if (!method)
863 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)");
864 1253
865 dSP; 1254 dSP;
866 1255
867 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
868 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
869 // 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
870 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
872 1262
880 SPAGAIN; 1270 SPAGAIN;
881 1271
882 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
883 { 1273 {
884 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
885 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
886 } 1276 }
887 1277
888 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
889 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
890 1280
891 PUTBACK; 1281 PUTBACK;
892 1282
893 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
894 } 1284 }
895 break; 1285 break;
896 1286
897 default: 1287 default:
1288 filter:
898 { 1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
899 sv = decode_sv (dec); 1292 sv = decode_sv (dec);
900 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 {
901 AV *av = newAV (); 1321 AV *av = newAV ();
902 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
903 av_push (av, sv); 1323 av_push (av, sv);
904 1324
905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
907 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909 sv = 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;
910 } 1334 }
911 break; 1335 break;
912 } 1336 }
913 1337
914 return sv; 1338 return sv;
921static SV * 1345static SV *
922decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
923{ 1347{
924 WANT (1); 1348 WANT (1);
925 1349
926 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
927 { 1351 {
928 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935 return decode_str (dec, 1); 1359
936 case 4: // array 1360 case MAJOR_MISC >> MAJOR_SHIFT:
937 return decode_av (dec);
938 case 5: // map
939 return decode_hv (dec);
940 case 6: // tag
941 return decode_tagged (dec);
942 case 7: // misc
943 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
944 { 1362 {
945 case 20: 1363 case SIMPLE_FALSE:
946#if CBOR_SLOW 1364#if CBOR_SLOW
947 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
948#endif 1366#endif
949 return newSVsv (types_false); 1367 return newSVsv (types_false);
950 case 21: 1368 case SIMPLE_TRUE:
951#if CBOR_SLOW 1369#if CBOR_SLOW
952 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
953#endif 1371#endif
954 return newSVsv (types_true); 1372 return newSVsv (types_true);
955 case 22: 1373 case SIMPLE_NULL:
956 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
957 case 23: 1375 case SIMPLE_UNDEF:
958#if CBOR_SLOW 1376#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
960#endif 1378#endif
961 return newSVsv (types_error); 1379 return newSVsv (types_error);
962 1380
963 case 25: 1381 case MISC_FLOAT16:
964 { 1382 {
965 WANT (2); 1383 WANT (2);
966 1384
967 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968 dec->cur += 2; 1386 dec->cur += 2;
969 1387
970 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
971 } 1389 }
972 1390
973 case 26: 1391 case MISC_FLOAT32:
974 { 1392 {
975 uint32_t fp; 1393 uint32_t fp;
976 WANT (4); 1394 WANT (4);
977 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
978 dec->cur += 4; 1396 dec->cur += 4;
981 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
982 1400
983 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
984 } 1402 }
985 1403
986 case 27: 1404 case MISC_FLOAT64:
987 { 1405 {
988 uint64_t fp; 1406 uint64_t fp;
989 WANT (8); 1407 WANT (8);
990 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
991 dec->cur += 8; 1409 dec->cur += 8;
994 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
995 1413
996 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
997 } 1415 }
998 1416
999 // 0..19 unassigned 1417 // 0..19 unassigned simple
1000 // 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
1001 default: 1421 default:
1002 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003 } 1423 }
1004 1424
1005 break; 1425 break;
1006 } 1426 }
1007 1427
1010} 1430}
1011 1431
1012static SV * 1432static SV *
1013decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014{ 1434{
1015 dec_t dec = { }; 1435 dec_t dec = { 0 };
1016 SV *sv; 1436 SV *sv;
1017 STRLEN len; 1437 STRLEN len;
1018 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
1019 1439
1020 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
1034 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
1035 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
1036 1456
1037 if (dec.err) 1457 if (dec.err)
1038 { 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
1039 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (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);
1041 } 1476 }
1042 1477
1043 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
1044 1479
1045 return sv; 1480 return sv;
1046} 1481}
1047 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 case MAJOR_ARRAY >> MAJOR_SHIFT:
1575 if (len)
1576 {
1577 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1578 count = len + 1;
1579 }
1580 break;
1581 }
1582 }
1583 }
1584 }
1585
1586 self->incr_pos = self->incr_need;
1587
1588 if (count > 0)
1589 {
1590 while (!--count)
1591 {
1592 if (!AvFILLp (self->incr_count))
1593 return 1; // done
1594
1595 SvREFCNT_dec_NN (av_pop (self->incr_count));
1596 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1597 }
1598
1599 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1600 }
1601 }
1602
1603 return 0;
1604}
1605
1606
1048///////////////////////////////////////////////////////////////////////////// 1607/////////////////////////////////////////////////////////////////////////////
1049// XS interface functions 1608// XS interface functions
1050 1609
1051MODULE = CBOR::XS PACKAGE = CBOR::XS 1610MODULE = CBOR::XS PACKAGE = CBOR::XS
1052 1611
1060 1619
1061 types_true = get_bool ("Types::Serialiser::true" ); 1620 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false"); 1621 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error"); 1622 types_error = get_bool ("Types::Serialiser::error");
1064 1623
1624 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1625
1065 sv_cbor = newSVpv ("CBOR", 0); 1626 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor); 1627 SvREADONLY_on (sv_cbor);
1628
1629 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067} 1630}
1068 1631
1069PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1070 1633
1071void CLONE (...) 1634void CLONE (...)
1090void shrink (CBOR *self, int enable = 1) 1653void shrink (CBOR *self, int enable = 1)
1091 ALIAS: 1654 ALIAS:
1092 shrink = F_SHRINK 1655 shrink = F_SHRINK
1093 allow_unknown = F_ALLOW_UNKNOWN 1656 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING 1657 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS 1658 allow_cycles = F_ALLOW_CYCLES
1659 forbid_objects = F_FORBID_OBJECTS
1096 dedup_strings = F_DEDUP_STRINGS 1660 pack_strings = F_PACK_STRINGS
1661 text_keys = F_TEXT_KEYS
1662 text_strings = F_TEXT_STRINGS
1663 validate_utf8 = F_VALIDATE_UTF8
1097 PPCODE: 1664 PPCODE:
1098{ 1665{
1099 if (enable) 1666 if (enable)
1100 self->flags |= ix; 1667 self->flags |= ix;
1101 else 1668 else
1107void get_shrink (CBOR *self) 1674void get_shrink (CBOR *self)
1108 ALIAS: 1675 ALIAS:
1109 get_shrink = F_SHRINK 1676 get_shrink = F_SHRINK
1110 get_allow_unknown = F_ALLOW_UNKNOWN 1677 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING 1678 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS 1679 get_allow_cycles = F_ALLOW_CYCLES
1680 get_forbid_objects = F_FORBID_OBJECTS
1113 get_dedup_strings = F_DEDUP_STRINGS 1681 get_pack_strings = F_PACK_STRINGS
1682 get_text_keys = F_TEXT_KEYS
1683 get_text_strings = F_TEXT_STRINGS
1684 get_validate_utf8 = F_VALIDATE_UTF8
1114 PPCODE: 1685 PPCODE:
1115 XPUSHs (boolSV (self->flags & ix)); 1686 XPUSHs (boolSV (self->flags & ix));
1116 1687
1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1688void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118 PPCODE: 1689 PPCODE:
1131 XPUSHs (ST (0)); 1702 XPUSHs (ST (0));
1132 1703
1133int get_max_size (CBOR *self) 1704int get_max_size (CBOR *self)
1134 CODE: 1705 CODE:
1135 RETVAL = self->max_size; 1706 RETVAL = self->max_size;
1707 OUTPUT:
1708 RETVAL
1709
1710void filter (CBOR *self, SV *filter = 0)
1711 PPCODE:
1712 SvREFCNT_dec (self->filter);
1713 self->filter = filter ? newSVsv (filter) : filter;
1714 XPUSHs (ST (0));
1715
1716SV *get_filter (CBOR *self)
1717 CODE:
1718 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136 OUTPUT: 1719 OUTPUT:
1137 RETVAL 1720 RETVAL
1138 1721
1139void encode (CBOR *self, SV *scalar) 1722void encode (CBOR *self, SV *scalar)
1140 PPCODE: 1723 PPCODE:
1155 EXTEND (SP, 2); 1738 EXTEND (SP, 2);
1156 PUSHs (sv); 1739 PUSHs (sv);
1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1740 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158} 1741}
1159 1742
1743void incr_parse (CBOR *self, SV *cborstr)
1744 ALIAS:
1745 incr_parse_multiple = 1
1746 PPCODE:
1747{
1748 if (SvUTF8 (cborstr))
1749 sv_utf8_downgrade (cborstr, 0);
1750
1751 if (!self->incr_count)
1752 {
1753 self->incr_count = newAV ();
1754 self->incr_pos = 0;
1755 self->incr_need = 1;
1756
1757 av_push (self->incr_count, newSViv (1));
1758 }
1759
1760 do
1761 {
1762 if (!incr_parse (self, cborstr))
1763 {
1764 if (self->incr_need > self->max_size && self->max_size)
1765 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1766 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1767
1768 break;
1769 }
1770
1771 SV *sv;
1772 char *offset;
1773
1774 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1775 XPUSHs (sv);
1776
1777 sv_chop (cborstr, offset);
1778
1779 av_clear (self->incr_count);
1780 av_push (self->incr_count, newSViv (1));
1781
1782 self->incr_pos = 0;
1783 self->incr_need = self->incr_pos + 1;
1784 }
1785 while (ix);
1786}
1787
1788void incr_reset (CBOR *self)
1789 CODE:
1790{
1791 SvREFCNT_dec (self->incr_count);
1792 self->incr_count = 0;
1793}
1794
1795void DESTROY (CBOR *self)
1796 PPCODE:
1797 cbor_free (self);
1798
1160PROTOTYPES: ENABLE 1799PROTOTYPES: ENABLE
1161 1800
1162void encode_cbor (SV *scalar) 1801void encode_cbor (SV *scalar)
1802 ALIAS:
1803 encode_cbor = 0
1804 encode_cbor_sharing = F_ALLOW_SHARING
1163 PPCODE: 1805 PPCODE:
1164{ 1806{
1165 CBOR cbor; 1807 CBOR cbor;
1166 cbor_init (&cbor); 1808 cbor_init (&cbor);
1809 cbor.flags |= ix;
1167 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1810 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168 XPUSHs (scalar); 1811 XPUSHs (scalar);
1169} 1812}
1170 1813
1171void decode_cbor (SV *cborstr) 1814void decode_cbor (SV *cborstr)
1175 cbor_init (&cbor); 1818 cbor_init (&cbor);
1176 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1819 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1177 XPUSHs (cborstr); 1820 XPUSHs (cborstr);
1178} 1821}
1179 1822
1823#ifdef __AFL_COMPILER
1824
1825void
1826afl_init ()
1827 CODE:
1828 __AFL_INIT ();
1829
1830int
1831afl_loop (unsigned int count = 10000)
1832 CODE:
1833 RETVAL = __AFL_LOOP (count);
1834 OUTPUT:
1835 RETVAL
1836
1837#endif
1838

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines