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 1.17 by root, Tue Oct 29 22:04:52 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"
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
29 CBOR_TAG_PERL_OBJECT = 256, 74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 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
31 81
32 // rfc7049 82 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
39 89
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 94
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 100
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 102};
53 103
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
54#define F_SHRINK 0x00000200UL 118#define F_SHRINK 0x00000001UL
55#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
56 127
57#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
58 129
59#define SB do { 130#define SB do {
60#define SE } while (0) 131#define SE } while (0)
72# define CBOR_SLOW 0 143# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
74#endif 145#endif
75 146
76static 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::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 149
79typedef struct { 150typedef struct {
80 U32 flags; 151 U32 flags;
81 U32 max_depth; 152 U32 max_depth;
82 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.
83} CBOR; 160} CBOR;
84 161
85ecb_inline void 162ecb_inline void
86cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
87{ 164{
88 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
89 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);
90} 174}
91 175
92///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
93// utility functions 177// utility functions
94 178
116 SvPV_renew (sv, SvCUR (sv) + 1); 200 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 201#endif
118 } 202 }
119} 203}
120 204
121///////////////////////////////////////////////////////////////////////////// 205// minimum length of a string to be registered for stringref
122// fp hell 206ecb_inline STRLEN
123 207minimum_string_length (UV idx)
124//TODO 208{
209 return idx <= 23 ? 3
210 : idx <= 0xffU ? 4
211 : idx <= 0xffffU ? 5
212 : idx <= 0xffffffffU ? 7
213 : 11;
214}
125 215
126///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
127// encoder 217// encoder
128 218
129// structure used for encoding CBOR 219// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 222 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 223 char *end; // SvEND (sv)
134 SV *sv; // result scalar 224 SV *sv; // result scalar
135 CBOR cbor; 225 CBOR cbor;
136 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;
137} enc_t; 231} enc_t;
138 232
139ecb_inline void 233ecb_inline void
140need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
141{ 235{
142 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
143 { 237 {
144 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
145 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
146 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
147 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
148 } 242 }
149} 243}
150 244
245static void encode_sv (enc_t *enc, SV *sv);
246
151ecb_inline void 247ecb_inline void
152encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
153{ 249{
154 need (enc, 1); 250 need (enc, 1);
155 *enc->cur++ = ch; 251 *enc->cur++ = ch;
156} 252}
157 253
254// used for tags, intregers, element counts and so on
158static void 255static void
159encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
160{ 257{
161 need (enc, 9); 258 need (enc, 9);
162 259
163 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
165 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
166 { 263 {
167 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 265 *enc->cur++ = len;
169 } 266 }
170 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
171 { 268 {
172 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 271 *enc->cur++ = len;
175 } 272 }
176 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
177 { 274 {
178 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 279 *enc->cur++ = len;
183 } 280 }
184 else 281 else
185 { 282 {
186 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 291 *enc->cur++ = len;
195 } 292 }
196} 293}
197 294
198static 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
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 316{
201 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);
202 need (enc, len); 347 need (enc, len);
203 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
204 enc->cur += len; 349 enc->cur += len;
205} 350}
206 351
207static 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}
208 507
209static void 508static void
210encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
211{ 510{
212 int i, len = av_len (av); 511 int i, len = av_len (av);
214 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
216 515
217 ++enc->depth; 516 ++enc->depth;
218 517
219 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 519
520 if (ecb_expect_false (SvMAGICAL (av)))
221 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
222 { 522 {
223 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 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 }
226 532
227 --enc->depth; 533 --enc->depth;
228} 534}
229 535
230static void 536static void
238 ++enc->depth; 544 ++enc->depth;
239 545
240 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
242 548
243 if (mg) 549 if (ecb_expect_false (mg))
244 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 551 else
246 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
247 553
248 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
249 { 555 {
250 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
252 else 558 else
253 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));
254 560
255 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));
256 } 562 }
257 563
258 if (mg) 564 if (ecb_expect_false (mg))
259 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 566
261 --enc->depth; 567 --enc->depth;
262} 568}
263 569
264// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
265static void 571static void
266encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
267{ 573{
268 svtype svt;
269
270 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
575
271 svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
272 577
273 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
274 { 579 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 580 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 581 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 587 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 589
285 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
286 GV *method;
287 591
288 if (stash == boolean_stash) 592 if (stash == boolean_stash)
289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
290 else if (stash == error_stash) 597 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
292 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
293 { 603 {
294 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 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))
299 } 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))));
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 659 {
302 dSP; 660 dSP;
303 661
304 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
305 // 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
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307 666
308 PUTBACK; 667 PUTBACK;
309 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
322 } 681 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 { 683 {
325 dSP; 684 dSP;
326 685
327 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
328 EXTEND (SP, 2); 688 EXTEND (SP, 2);
329 // 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
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
332 692
336 696
337 // catch this surprisingly common error 697 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 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));
340 700
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 704
345 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
346 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
347 713
348 PUTBACK; 714 PUTBACK;
349 715
350 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
351 } 717 }
355 } 721 }
356 else if (svt == SVt_PVHV) 722 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 723 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 724 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 725 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 726 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 727 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 728 encode_tag (enc, CBOR_TAG_INDIRECTION);
729 encode_sv (enc, sv);
730 }
380} 731}
381 732
382static void 733static void
383encode_nv (enc_t *enc, SV *sv) 734encode_nv (enc_t *enc, SV *sv)
384{ 735{
385 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
386 737
387 need (enc, 9); 738 need (enc, 9);
388 739
389 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 742 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
393 { 744 encode_float32 (enc, nv);
394 uint32_t fp = ecb_float_to_binary32 (nv);
395
396 *enc->cur++ = 0xe0 | 26;
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403 }
404 else 745 else
405 { 746 encode_float64 (enc, nv);
406 uint64_t fp = ecb_double_to_binary64 (nv);
407
408 *enc->cur++ = 0xe0 | 27;
409
410 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp);
412
413 memcpy (enc->cur, &fp, 8);
414 enc->cur += 8;
415 }
416} 747}
417 748
418static void 749static void
419encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
420{ 751{
422 753
423 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
424 { 755 {
425 STRLEN len; 756 STRLEN len;
426 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428 } 759 }
429 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 761 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
432 { 763 encode_int (enc, sv);
433 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv));
435 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv));
437 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
439 }
440 else if (SvROK (sv)) 764 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 770 else
447 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",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 773}
450 774
451static SV * 775static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
453{ 777{
454 enc_t enc; 778 enc_t enc = { 0 };
455 779
456 enc.cbor = *cbor; 780 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 784
462 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
463 encode_sv (&enc, scalar); 794 encode_sv (&enc, scalar);
464 795
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 796 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *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
467 798
481 U8 *end; // end of input string 812 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 813 const char *err; // parse error, if != 0
483 CBOR cbor; 814 CBOR cbor;
484 U32 depth; // recursion depth 815 U32 depth; // recursion depth
485 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
486} dec_t; 821} dec_t;
487 822
488#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);
489 830
490#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;
491 834
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
492#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)
493#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
494 861
495static UV 862static UV
496decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
497{ 864{
498 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
499 { 866 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 867
505 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
506 WANT (2); 872 WANT (1);
507 dec->cur += 2; 873 dec->cur += 1;
508 return dec->cur[-1]; 874 return dec->cur[-1];
509 875 }
510 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
511 WANT (3); 878 WANT (2);
512 dec->cur += 3; 879 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
515 882 }
516 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
517 WANT (5); 885 WANT (4);
518 dec->cur += 5; 886 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
523 891 }
524 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
525 WANT (9); 894 WANT (8);
526 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
527 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
531 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
535 910 }
536 default: 911 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 913
540fail: 914fail:
541 return 0; 915 return 0;
542} 916}
543 917
548{ 922{
549 AV *av = newAV (); 923 AV *av = newAV ();
550 924
551 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
552 926
553 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 928 {
555 ++dec->cur; 929 ++dec->cur;
556 930
557 for (;;) 931 for (;;)
558 { 932 {
559 WANT (1); 933 WANT (1);
560 934
561 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
562 { 936 {
563 ++dec->cur; 937 ++dec->cur;
564 break; 938 break;
565 } 939 }
566 940
567 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
568 } 942 }
569 } 943 }
570 else 944 else
571 { 945 {
572 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
573 947
948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 949 av_fill (av, len - 1);
575 950
576 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
578 } 953 }
579 954
580 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
581 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
582 957
583fail: 958fail:
584 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
585 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
586 return &PL_sv_undef; 961 return &PL_sv_undef;
587} 962}
588 963
589static void 964static void
590decode_he (dec_t *dec, HV *hv) 965decode_he (dec_t *dec, HV *hv)
591{ 966{
592 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 968 // byte or utf-8 strings as keys, but only when !stringref
594 969
595 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))
596 { 972 {
597 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
598 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
599 975
976 WANT (len);
600 dec->cur += len; 977 dec->cur += len;
601 978
602 hv_store (hv, key, len, decode_sv (dec), 0); 979 hv_store (hv, key, len, decode_sv (dec), 0);
980
981 return;
603 } 982 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 984 {
606 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
607 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
608 987
988 WANT (len);
609 dec->cur += len; 989 dec->cur += len;
610 990
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
996
997 return;
612 } 998 }
613 else 999
614 {
615 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
617 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
618 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
620 } 1033
1034fail:
1035 ;
621} 1036}
622 1037
623static SV * 1038static SV *
624decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
625{ 1040{
626 HV *hv = newHV (); 1041 HV *hv = newHV ();
627 1042
628 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
629 1044
630 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 1046 {
632 ++dec->cur; 1047 ++dec->cur;
633 1048
634 for (;;) 1049 for (;;)
635 { 1050 {
636 WANT (1); 1051 WANT (1);
637 1052
638 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
639 { 1054 {
640 ++dec->cur; 1055 ++dec->cur;
641 break; 1056 break;
642 } 1057 }
643 1058
644 decode_he (dec, hv); 1059 decode_he (dec, hv);
645 } 1060 }
646 } 1061 }
647 else 1062 else
648 { 1063 {
649 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
650 1067
651 while (pairs--) 1068 while (pairs--)
652 decode_he (dec, hv); 1069 decode_he (dec, hv);
653 } 1070 }
654 1071
655 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
656 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
657 1074
658fail: 1075fail:
659 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
660 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
661 return &PL_sv_undef; 1078 return &PL_sv_undef;
662} 1079}
663 1080
664static SV * 1081static SV *
665decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
666{ 1083{
667 SV *sv = 0; 1084 SV *sv = 0;
668 1085
669 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
670 { 1087 {
1088 // indefinite length strings
671 ++dec->cur; 1089 ++dec->cur;
672 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
673 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
674 1094
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 1095 for (;;)
677 { 1096 {
678 WANT (1); 1097 WANT (1);
679 1098
680 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 1101 {
682 ++dec->cur; 1102 ++dec->cur;
683 break; 1103 break;
684 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 1107
686 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;
687 } 1113 }
688 } 1114 }
689 else 1115 else
690 { 1116 {
691 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
692 1118
693 WANT (len); 1119 WANT (len);
694 sv = newSVpvn (dec->cur, len); 1120 sv = newSVpvn (dec->cur, len);
695 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));
696 } 1126 }
697 1127
698 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
699 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
700 1136
701 return sv; 1137 return sv;
702 1138
703fail: 1139fail:
704 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
706} 1142}
707 1143
708static SV * 1144static SV *
709decode_tagged (dec_t *dec) 1145decode_tagged (dec_t *dec)
710{ 1146{
1147 SV *sv = 0;
711 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:
712 SV *sv = decode_sv (dec); 1155 sv = decode_sv (dec);
1156 break;
713 1157
714 if (tag == CBOR_TAG_MAGIC) 1158 case CBOR_TAG_INDIRECTION:
715 return sv; 1159 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 1160 break;
717 { 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
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
720 1241
721 AV *av = (AV *)SvRV (sv); 1242 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1; 1243 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1244 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724 1245
725 if (!stash) 1246 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)"); 1247 ERR ("cannot decode perl-object (package does not exist)");
727 1248
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1249 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 1250
730 if (!method) 1251 if (!method)
731 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)");
732 1253
733 dSP; 1254 dSP;
734 1255
735 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
736 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
737 // 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
738 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
740 1262
741 int i; 1263 int i;
742 1264
743 for (i = 1; i < len; ++i) 1265 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1)); 1266 PUSHs (*av_fetch (av, i, 1));
745 1267
746 PUTBACK; 1268 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); 1269 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN; 1270 SPAGAIN;
749 1271
750 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
751 { 1273 {
1274 FREETMPS; LEAVE;
1275 ERR_ERRSV;
1276 }
1277
1278 SvREFCNT_dec_NN (sv);
1279 sv = SvREFCNT_inc (POPs);
1280
1281 PUTBACK;
1282
752 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1284 }
1285 break;
1286
1287 default:
1288 filter:
754 } 1289 {
1290 SV *tag_sv = newSVuv (tag);
755 1291
756 SvREFCNT_dec (sv); 1292 sv = decode_sv (dec);
757 sv = SvREFCNT_inc (POPs);
758 1293
1294 dSP;
1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1297 EXTEND (SP, 2);
1298 PUSHs (tag_sv);
1299 PUSHs (sv);
1300
759 PUTBACK; 1301 PUTBACK;
1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1303 SPAGAIN;
760 1304
1305 if (SvTRUE (ERRSV))
1306 {
1307 SvREFCNT_dec_NN (tag_sv);
761 FREETMPS; LEAVE; 1308 FREETMPS; LEAVE;
1309 ERR_ERRSV;
1310 }
762 1311
763 return sv; 1312 if (count)
764 } 1313 {
765 else 1314 SvREFCNT_dec_NN (tag_sv);
766 { 1315 SvREFCNT_dec_NN (sv);
1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1318 }
1319 else
1320 {
767 AV *av = newAV (); 1321 AV *av = newAV ();
768 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
769 av_push (av, sv); 1323 av_push (av, sv);
770 1324
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774
775 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;
776 } 1336 }
1337
1338 return sv;
777 1339
778fail: 1340fail:
779 SvREFCNT_dec (sv); 1341 SvREFCNT_dec (sv);
780 return &PL_sv_undef; 1342 return &PL_sv_undef;
781} 1343}
783static SV * 1345static SV *
784decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
785{ 1347{
786 WANT (1); 1348 WANT (1);
787 1349
788 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
789 { 1351 {
790 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
791 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
792 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
793 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
794 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
795 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
796 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
797 return decode_str (dec, 1); 1359
798 case 4: // array 1360 case MAJOR_MISC >> MAJOR_SHIFT:
799 return decode_av (dec);
800 case 5: // map
801 return decode_hv (dec);
802 case 6: // tag
803 return decode_tagged (dec);
804 case 7: // misc
805 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
806 { 1362 {
807 case 20: 1363 case SIMPLE_FALSE:
808#if CBOR_SLOW 1364#if CBOR_SLOW
809 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
810#endif 1366#endif
811 return newSVsv (types_false); 1367 return newSVsv (types_false);
812 case 21: 1368 case SIMPLE_TRUE:
813#if CBOR_SLOW 1369#if CBOR_SLOW
814 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
815#endif 1371#endif
816 return newSVsv (types_true); 1372 return newSVsv (types_true);
817 case 22: 1373 case SIMPLE_NULL:
818 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
819 case 23: 1375 case SIMPLE_UNDEF:
820#if CBOR_SLOW 1376#if CBOR_SLOW
821 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
822#endif 1378#endif
823 return newSVsv (types_error); 1379 return newSVsv (types_error);
824 1380
825 case 25: 1381 case MISC_FLOAT16:
826 { 1382 {
827 WANT (2); 1383 WANT (2);
828 1384
829 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
830 dec->cur += 2; 1386 dec->cur += 2;
831 1387
832 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
833 } 1389 }
834 1390
835 case 26: 1391 case MISC_FLOAT32:
836 { 1392 {
837 uint32_t fp; 1393 uint32_t fp;
838 WANT (4); 1394 WANT (4);
839 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
840 dec->cur += 4; 1396 dec->cur += 4;
843 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
844 1400
845 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
846 } 1402 }
847 1403
848 case 27: 1404 case MISC_FLOAT64:
849 { 1405 {
850 uint64_t fp; 1406 uint64_t fp;
851 WANT (8); 1407 WANT (8);
852 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
853 dec->cur += 8; 1409 dec->cur += 8;
856 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
857 1413
858 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
859 } 1415 }
860 1416
861 // 0..19 unassigned 1417 // 0..19 unassigned simple
862 // 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
863 default: 1421 default:
864 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
865 } 1423 }
866 1424
867 break; 1425 break;
868 } 1426 }
869 1427
872} 1430}
873 1431
874static SV * 1432static SV *
875decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
876{ 1434{
877 dec_t dec; 1435 dec_t dec = { 0 };
878 SV *sv; 1436 SV *sv;
879 STRLEN len; 1437 STRLEN len;
880 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
881 1439
882 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
884 (unsigned long)len, (unsigned long)cbor->max_size); 1442 (unsigned long)len, (unsigned long)cbor->max_size);
885 1443
886 dec.cbor = *cbor; 1444 dec.cbor = *cbor;
887 dec.cur = (U8 *)data; 1445 dec.cur = (U8 *)data;
888 dec.end = (U8 *)data + len; 1446 dec.end = (U8 *)data + len;
889 dec.err = 0;
890 dec.depth = 0;
891 1447
892 sv = decode_sv (&dec); 1448 sv = decode_sv (&dec);
893 1449
894 if (offset_return) 1450 if (offset_return)
895 *offset_return = dec.cur; 1451 *offset_return = dec.cur;
898 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
899 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
900 1456
901 if (dec.err) 1457 if (dec.err)
902 { 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
903 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
904 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);
905 } 1476 }
906 1477
907 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
908 1479
909 return sv; 1480 return sv;
910} 1481}
911 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
912///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
913// XS interface functions 1609// XS interface functions
914 1610
915MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
916 1612
924 1620
925 types_true = get_bool ("Types::Serialiser::true" ); 1621 types_true = get_bool ("Types::Serialiser::true" );
926 types_false = get_bool ("Types::Serialiser::false"); 1622 types_false = get_bool ("Types::Serialiser::false");
927 types_error = get_bool ("Types::Serialiser::error"); 1623 types_error = get_bool ("Types::Serialiser::error");
928 1624
1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1626
929 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
930 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
931} 1631}
932 1632
933PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
934 1634
935void CLONE (...) 1635void CLONE (...)
953 1653
954void shrink (CBOR *self, int enable = 1) 1654void shrink (CBOR *self, int enable = 1)
955 ALIAS: 1655 ALIAS:
956 shrink = F_SHRINK 1656 shrink = F_SHRINK
957 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
958 PPCODE: 1665 PPCODE:
959{ 1666{
960 if (enable) 1667 if (enable)
961 self->flags |= ix; 1668 self->flags |= ix;
962 else 1669 else
967 1674
968void get_shrink (CBOR *self) 1675void get_shrink (CBOR *self)
969 ALIAS: 1676 ALIAS:
970 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
971 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
972 PPCODE: 1686 PPCODE:
973 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
974 1688
975void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
976 PPCODE: 1690 PPCODE:
989 XPUSHs (ST (0)); 1703 XPUSHs (ST (0));
990 1704
991int get_max_size (CBOR *self) 1705int get_max_size (CBOR *self)
992 CODE: 1706 CODE:
993 RETVAL = self->max_size; 1707 RETVAL = self->max_size;
1708 OUTPUT:
1709 RETVAL
1710
1711void filter (CBOR *self, SV *filter = 0)
1712 PPCODE:
1713 SvREFCNT_dec (self->filter);
1714 self->filter = filter ? newSVsv (filter) : filter;
1715 XPUSHs (ST (0));
1716
1717SV *get_filter (CBOR *self)
1718 CODE:
1719 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
994 OUTPUT: 1720 OUTPUT:
995 RETVAL 1721 RETVAL
996 1722
997void encode (CBOR *self, SV *scalar) 1723void encode (CBOR *self, SV *scalar)
998 PPCODE: 1724 PPCODE:
1013 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1014 PUSHs (sv); 1740 PUSHs (sv);
1015 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1016} 1742}
1017 1743
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}
1795
1796void DESTROY (CBOR *self)
1797 PPCODE:
1798 cbor_free (self);
1799
1018PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
1019 1801
1020void encode_cbor (SV *scalar) 1802void encode_cbor (SV *scalar)
1803 ALIAS:
1804 encode_cbor = 0
1805 encode_cbor_sharing = F_ALLOW_SHARING
1021 PPCODE: 1806 PPCODE:
1022{ 1807{
1023 CBOR cbor; 1808 CBOR cbor;
1024 cbor_init (&cbor); 1809 cbor_init (&cbor);
1810 cbor.flags |= ix;
1025 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1811 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1026 XPUSHs (scalar); 1812 XPUSHs (scalar);
1027} 1813}
1028 1814
1029void decode_cbor (SV *cborstr) 1815void decode_cbor (SV *cborstr)
1033 cbor_init (&cbor); 1819 cbor_init (&cbor);
1034 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1035 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1036} 1822}
1037 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