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 1.4 by root, Sat Oct 26 21:06:37 2013 UTC vs.
Revision 1.68 by root, Mon Nov 30 18:30:29 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
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
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 // possibly future enhancements: (generic) float, (generic) string
115};
116
14#define F_SHRINK 0x00000200UL 117#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 118#define F_ALLOW_UNKNOWN 0x00000002UL
119#define F_ALLOW_SHARING 0x00000004UL
120#define F_ALLOW_CYCLES 0x00000008UL
121#define F_FORBID_OBJECTS 0x00000010UL
122#define F_PACK_STRINGS 0x00000020UL
123#define F_TEXT_KEYS 0x00000040UL
124#define F_TEXT_STRINGS 0x00000080UL
125#define F_VALIDATE_UTF8 0x00000100UL
16 126
17#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
18 128
19#define SB do { 129#define SB do {
20#define SE } while (0) 130#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 131
33#define IN_RANGE_INC(type,val,beg,end) \ 132#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 133 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 134 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 135
42#else 141#else
43# define CBOR_SLOW 0 142# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 143# define CBOR_STASH cbor_stash
45#endif 144#endif
46 145
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 146static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 147static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 148
50typedef struct { 149typedef struct {
51 U32 flags; 150 U32 flags;
52 U32 max_depth; 151 U32 max_depth;
53 STRLEN max_size; 152 STRLEN max_size;
153 SV *filter;
54 154
55 SV *cb_object; 155 // for the incremental parser
56 HV *cb_sk_object; 156 STRLEN incr_pos; // the current offset into the text
157 STRLEN incr_need; // minimum bytes needed to decode
158 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
57} CBOR; 159} CBOR;
58 160
59INLINE void 161ecb_inline void
60cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
61{ 163{
62 Zero (cbor, 1, CBOR); 164 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 165 cbor->max_depth = 512;
64} 166}
65 167
168ecb_inline void
169cbor_free (CBOR *cbor)
170{
171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
173}
174
66///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
67// utility functions 176// utility functions
68 177
69INLINE SV * 178ecb_inline SV *
70get_bool (const char *name) 179get_bool (const char *name)
71{ 180{
72 SV *sv = get_sv (name, 1); 181 SV *sv = get_sv (name, 1);
73 182
74 SvREADONLY_on (sv); 183 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 184 SvREADONLY_on (SvRV (sv));
76 185
77 return sv; 186 return sv;
78} 187}
79 188
80INLINE void 189ecb_inline void
81shrink (SV *sv) 190shrink (SV *sv)
82{ 191{
83 sv_utf8_downgrade (sv, 1); 192 sv_utf8_downgrade (sv, 1);
84 193
85 if (SvLEN (sv) > SvCUR (sv) + 1) 194 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 199 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 200#endif
92 } 201 }
93} 202}
94 203
95///////////////////////////////////////////////////////////////////////////// 204// minimum length of a string to be registered for stringref
96// fp hell 205ecb_inline int
97 206minimum_string_length (UV idx)
98//TODO 207{
208 return idx <= 23 ? 3
209 : idx <= 0xffU ? 4
210 : idx <= 0xffffU ? 5
211 : idx <= 0xffffffffU ? 7
212 : 11;
213}
99 214
100///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
101// encoder 216// encoder
102 217
103// structure used for encoding CBOR 218// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 221 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 222 char *end; // SvEND (sv)
108 SV *sv; // result scalar 223 SV *sv; // result scalar
109 CBOR cbor; 224 CBOR cbor;
110 U32 depth; // recursion level 225 U32 depth; // recursion level
226 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
227 UV stringref_idx;
228 HV *shareable; // ptr => index, or 0
229 UV shareable_idx;
111} enc_t; 230} enc_t;
112 231
113INLINE void 232ecb_inline void
114need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
115{ 234{
116 if (expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
117 { 236 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 241 }
123} 242}
124 243
125INLINE void 244ecb_inline void
126encode_ch (enc_t *enc, char ch) 245encode_ch (enc_t *enc, char ch)
127{ 246{
128 need (enc, 1); 247 need (enc, 1);
129 *enc->cur++ = ch; 248 *enc->cur++ = ch;
130} 249}
131 250
251// used for tags, intregers, element counts and so on
132static void 252static void
133encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
134{ 254{
135 need (enc, 9); 255 need (enc, 9);
136 256
137 if (len < 24) 257 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
139 else if (len <= 0xff) 259 else if (ecb_expect_true (len <= 0xffU))
140 { 260 {
141 *enc->cur++ = major | 24; 261 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 262 *enc->cur++ = len;
143 } 263 }
144 else if (len <= 0xffff) 264 else if (len <= 0xffffU)
145 { 265 {
146 *enc->cur++ = major | 25; 266 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 268 *enc->cur++ = len;
149 } 269 }
150 else if (len <= 0xffffffff) 270 else if (len <= 0xffffffffU)
151 { 271 {
152 *enc->cur++ = major | 26; 272 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 276 *enc->cur++ = len;
157 } 277 }
158 else 278 else
159 { 279 {
160 *enc->cur++ = major | 27; 280 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 281 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 282 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 283 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 284 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 285 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 288 *enc->cur++ = len;
169 } 289 }
170} 290}
171 291
172static void 292// encodes a perl value into a CBOR integer
293ecb_inline void
294encode_int (enc_t *enc, SV *sv)
295{
296 if (SvIsUV (sv))
297 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
298 else if (SvIVX (sv) >= 0)
299 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
300 else
301 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
302}
303
304ecb_inline void
305encode_tag (enc_t *enc, UV tag)
306{
307 encode_uint (enc, MAJOR_TAG, tag);
308}
309
310// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
311ecb_noinline static void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 313{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 314 STRLEN ulen = len;
315 U8 *p, *pend = (U8 *)str + len;
316
317 for (p = (U8 *)str; p < pend; ++p)
318 ulen += *p >> 7; // count set high bits
319
320 encode_uint (enc, MAJOR_TEXT, ulen);
321
322 need (enc, ulen);
323 for (p = (U8 *)str; p < pend; ++p)
324 if (*p < 0x80)
325 *enc->cur++ = *p;
326 else
327 {
328 *enc->cur++ = 0xc0 + (*p >> 6);
329 *enc->cur++ = 0x80 + (*p & 63);
330 }
331}
332
333ecb_inline void
334encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
335{
336 if (ecb_expect_false (upgrade_utf8))
337 if (!utf8)
338 {
339 encode_str_utf8 (enc, utf8, str, len);
340 return;
341 }
342
343 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 344 need (enc, len);
177 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
178 enc->cur += len; 346 enc->cur += len;
179} 347}
180 348
349ecb_inline void
350encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
351{
352 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
353 {
354 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
355
356 if (SvOK (*svp))
357 {
358 // already registered, use stringref
359 encode_tag (enc, CBOR_TAG_STRINGREF);
360 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
361 return;
362 }
363 else if (len >= minimum_string_length (enc->stringref_idx))
364 {
365 // register only
366 sv_setuv (*svp, enc->stringref_idx);
367 ++enc->stringref_idx;
368 }
369 }
370
371 encode_str (enc, upgrade_utf8, utf8, str, len);
372}
373
374ecb_inline void
375encode_float16 (enc_t *enc, NV nv)
376{
377 need (enc, 1+2);
378
379 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
380
381 uint16_t fp = ecb_float_to_binary16 (nv);
382
383 if (!ecb_big_endian ())
384 fp = ecb_bswap16 (fp);
385
386 memcpy (enc->cur, &fp, 2);
387 enc->cur += 2;
388}
389
390ecb_inline void
391encode_float32 (enc_t *enc, NV nv)
392{
393 need (enc, 1+4);
394
395 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
396
397 uint32_t fp = ecb_float_to_binary32 (nv);
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap32 (fp);
401
402 memcpy (enc->cur, &fp, 4);
403 enc->cur += 4;
404}
405
406ecb_inline void
407encode_float64 (enc_t *enc, NV nv)
408{
409 need (enc, 1+8);
410
411 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
412
413 uint64_t fp = ecb_double_to_binary64 (nv);
414
415 if (!ecb_big_endian ())
416 fp = ecb_bswap64 (fp);
417
418 memcpy (enc->cur, &fp, 8);
419 enc->cur += 8;
420}
421
422ecb_inline void
423encode_forced (enc_t *enc, UV type, SV *sv)
424{
425 switch (type)
426 {
427 case AS_CBOR:
428 {
429 STRLEN len;
430 char *str = SvPVbyte (sv, len);
431
432 need (enc, len);
433 memcpy (enc->cur, str, len);
434 enc->cur += len;
435 }
436 break;
437
438 case AS_BYTES:
439 {
440 STRLEN len;
441 char *str = SvPVbyte (sv, len);
442 encode_strref (enc, 0, 0, str, len);
443 }
444 break;
445
446 case AS_TEXT:
447 {
448 STRLEN len;
449 char *str = SvPVutf8 (sv, len);
450 encode_strref (enc, 1, 1, str, len);
451 }
452 break;
453
454 case AS_INT: encode_int (enc, sv); break;
455 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
456 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
457 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
458
459 default:
460 croak ("encountered malformed CBOR::XS::Tagged object");
461 }
462}
463
181static void encode_sv (enc_t *enc, SV *sv); 464static void encode_sv (enc_t *enc, SV *sv);
182 465
183static void 466static void
184encode_av (enc_t *enc, AV *av) 467encode_av (enc_t *enc, AV *av)
185{ 468{
188 if (enc->depth >= enc->cbor.max_depth) 471 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 472 croak (ERR_NESTING_EXCEEDED);
190 473
191 ++enc->depth; 474 ++enc->depth;
192 475
193 encode_uint (enc, 0x80, len + 1); 476 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 477
478 if (ecb_expect_false (SvMAGICAL (av)))
195 for (i = 0; i <= len; ++i) 479 for (i = 0; i <= len; ++i)
196 { 480 {
197 SV **svp = av_fetch (av, i, 0); 481 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 482 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 483 }
484 else
485 for (i = 0; i <= len; ++i)
486 {
487 SV *sv = AvARRAY (av)[i];
488 encode_sv (enc, sv ? sv : &PL_sv_undef);
489 }
200 490
201 --enc->depth; 491 --enc->depth;
202} 492}
203 493
204static void 494static void
212 ++enc->depth; 502 ++enc->depth;
213 503
214 int pairs = hv_iterinit (hv); 504 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 505 int mg = SvMAGICAL (hv);
216 506
217 if (mg) 507 if (ecb_expect_false (mg))
218 encode_ch (enc, 0xa0 | 31); 508 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 509 else
220 encode_uint (enc, 0xa0, pairs); 510 encode_uint (enc, MAJOR_MAP, pairs);
221 511
222 while ((he = hv_iternext (hv))) 512 while ((he = hv_iternext (hv)))
223 { 513 {
224 if (HeKLEN (he) == HEf_SVKEY) 514 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 515 encode_sv (enc, HeSVKEY (he));
226 else 516 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 517 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 518
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 519 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 520 }
231 521
232 if (mg) 522 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xe0 | 31); 523 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 524
235 --enc->depth; 525 --enc->depth;
236} 526}
237 527
238// encode objects, arrays and special \0=false and \1=true values. 528// encode objects, arrays and special \0=false and \1=true values.
239static void 529static void
240encode_rv (enc_t *enc, SV *sv) 530encode_rv (enc_t *enc, SV *sv)
241{ 531{
242 svtype svt;
243
244 SvGETMAGIC (sv); 532 SvGETMAGIC (sv);
533
245 svt = SvTYPE (sv); 534 svtype svt = SvTYPE (sv);
246 535
247 if (expect_false (SvOBJECT (sv))) 536 if (ecb_expect_false (SvOBJECT (sv)))
248 { 537 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 538 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 539 ? types_boolean_stash
540 : gv_stashpv ("Types::Serialiser::Boolean", 1);
541 HV *error_stash = !CBOR_SLOW || types_error_stash
542 ? types_error_stash
543 : gv_stashpv ("Types::Serialiser::Error", 1);
544 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
545 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 546 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 547
253 if (SvSTASH (sv) == stash) 548 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 549
550 if (stash == boolean_stash)
551 {
552 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
553 return;
554 }
555 else if (stash == error_stash)
556 {
557 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
558 return;
559 }
560 else if (stash == tagged_stash)
561 {
562 if (svt != SVt_PVAV)
563 croak ("encountered CBOR::XS::Tagged object that isn't an array");
564
565 switch (av_len ((AV *)sv))
566 {
567 case 2-1:
568 // actually a tagged value
569 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
570 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
571 break;
572
573 case 3-1:
574 // a forced type [value, type, undef]
575 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
576 break;
577
578 default:
579 croak ("encountered malformed CBOR::XS::Tagged object");
580 }
581
582 return;
583 }
584 }
585
586 if (ecb_expect_false (SvREFCNT (sv) > 1)
587 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
588 {
589 if (ecb_expect_false (!enc->shareable))
590 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
591
592 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
593
594 if (SvOK (*svp))
595 {
596 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
597 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
598 return;
599 }
255 else 600 else
256 { 601 {
257#if 0 //TODO 602 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 603 ++enc->shareable_idx;
604 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
605 }
606 }
607
608 if (ecb_expect_false (SvOBJECT (sv)))
609 {
610 HV *stash = SvSTASH (sv);
611 GV *method;
612
613 if (enc->cbor.flags & F_FORBID_OBJECTS)
614 croak ("encountered object '%s', but forbid_objects is enabled",
615 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
616 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 617 {
618 dSP;
619
620 ENTER; SAVETMPS;
621 PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 622 // we re-bless the reference to get overload and other niceties right
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 623 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 624
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 625 PUTBACK;
626 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 627 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 628 SPAGAIN;
274 629
275 // catch this surprisingly common error 630 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 631 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 632 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 633
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 634 encode_sv (enc, POPs);
283 635
636 PUTBACK;
637
284 FREETMPS; LEAVE; 638 FREETMPS; LEAVE;
285 } 639 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED) 640 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
287 encode_str (enc, "null", 4, 0); 641 {
288 else 642 dSP;
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 643
290 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 644 ENTER; SAVETMPS;
645 PUSHMARK (SP);
646 EXTEND (SP, 2);
647 // we re-bless the reference to get overload and other niceties right
648 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
649 PUSHs (sv_cbor);
650
651 PUTBACK;
652 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
653 SPAGAIN;
654
655 // catch this surprisingly common error
656 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
657 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
658
659 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
660 encode_uint (enc, MAJOR_ARRAY, count + 1);
661 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
662
663 {
664 int i;
665
666 for (i = 0; i < count; ++i)
667 encode_sv (enc, SP[i + 1 - count]);
668
669 SP -= count;
291 } 670 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 671
293 encode_str (enc, "null", 4, 0); 672 PUTBACK;
673
674 FREETMPS; LEAVE;
675 }
294 else 676 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 677 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 678 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 679 }
300 else if (svt == SVt_PVHV) 680 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 681 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 682 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 683 encode_av (enc, (AV *)sv);
304 else if (svt < SVt_PVAV)
305 {
306 STRLEN len = 0;
307 char *pv = svt ? SvPV (sv, len) : 0;
308
309 if (len == 1 && *pv == '1')
310 encode_ch (enc, 0xe0 | 21);
311 else if (len == 1 && *pv == '0')
312 encode_ch (enc, 0xe0 | 20);
313 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314 encode_ch (enc, 0xe0 | 23);
315 else
316 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318 }
319 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320 encode_ch (enc, 0xe0 | 23);
321 else 684 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 685 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 686 encode_tag (enc, CBOR_TAG_INDIRECTION);
687 encode_sv (enc, sv);
688 }
324} 689}
325 690
326static void 691static void
327encode_nv (enc_t *enc, SV *sv) 692encode_nv (enc_t *enc, SV *sv)
328{ 693{
329 double nv = SvNVX (sv); 694 double nv = SvNVX (sv);
330 695
331 need (enc, 9); 696 need (enc, 9);
332 697
333 if (expect_false (nv == (U32)nv)) 698 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 699 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 700 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 701 else if (ecb_expect_false (nv == (float)nv))
337 { 702 encode_float32 (enc, nv);
338 uint32_t fp = ecb_float_to_binary32 (nv);
339
340 *enc->cur++ = 0xe0 | 26;
341
342 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp);
344
345 memcpy (enc->cur, &fp, 4);
346 enc->cur += 4;
347 }
348 else 703 else
349 { 704 encode_float64 (enc, nv);
350 uint64_t fp = ecb_double_to_binary64 (nv);
351
352 *enc->cur++ = 0xe0 | 27;
353
354 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp);
356
357 memcpy (enc->cur, &fp, 8);
358 enc->cur += 8;
359 }
360} 705}
361 706
362static void 707static void
363encode_sv (enc_t *enc, SV *sv) 708encode_sv (enc_t *enc, SV *sv)
364{ 709{
366 711
367 if (SvPOKp (sv)) 712 if (SvPOKp (sv))
368 { 713 {
369 STRLEN len; 714 STRLEN len;
370 char *str = SvPV (sv, len); 715 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 716 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372 } 717 }
373 else if (SvNOKp (sv)) 718 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 719 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 720 else if (SvIOKp (sv))
376 { 721 encode_int (enc, sv);
377 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv));
379 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv));
381 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
383 }
384 else if (SvROK (sv)) 722 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 723 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 724 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 726 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 727 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 728 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 729 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 730 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 731}
394 732
395static SV * 733static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 734encode_cbor (SV *scalar, CBOR *cbor)
397{ 735{
398 enc_t enc; 736 enc_t enc = { 0 };
399 737
400 enc.cbor = *cbor; 738 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 739 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 740 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 741 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 742
406 SvPOK_only (enc.sv); 743 SvPOK_only (enc.sv);
744
745 if (cbor->flags & F_PACK_STRINGS)
746 {
747 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
748 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
749 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
750 }
751
407 encode_sv (&enc, scalar); 752 encode_sv (&enc, scalar);
408 753
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 754 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 755 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 756
425 U8 *end; // end of input string 770 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 771 const char *err; // parse error, if != 0
427 CBOR cbor; 772 CBOR cbor;
428 U32 depth; // recursion depth 773 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 774 U32 maxdepth; // recursion depth limit
775 AV *shareable;
776 AV *stringref;
777 SV *decode_tagged;
778 SV *err_sv; // optional sv for error, needs to be freed
430} dec_t; 779} dec_t;
431 780
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 781// set dec->err to ERRSV
782ecb_cold static void
783err_errsv (dec_t *dec)
784{
785 if (!dec->err)
786 {
787 dec->err_sv = newSVsv (ERRSV);
433 788
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 789 // chop off the trailing \n
790 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
791 *SvEND (dec->err_sv) = 0;
435 792
793 dec->err = SvPVutf8_nolen (dec->err_sv);
794 }
795}
796
797// the following functions are used to reduce code size and help the compiler to optimise
798ecb_cold static void
799err_set (dec_t *dec, const char *reason)
800{
801 if (!dec->err)
802 dec->err = reason;
803}
804
805ecb_cold static void
806err_unexpected_end (dec_t *dec)
807{
808 err_set (dec, "unexpected end of CBOR data");
809}
810
811#define ERR_DO(do) SB do; goto fail; SE
812#define ERR(reason) ERR_DO (err_set (dec, reason))
813#define ERR_ERRSV ERR_DO (err_errsv (dec))
814
815#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
816
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 817#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 818#define DEC_DEC_DEPTH --dec->depth
438 819
439static UV 820static UV
440decode_uint (dec_t *dec) 821decode_uint (dec_t *dec)
441{ 822{
442 switch (*dec->cur & 31) 823 U8 m = *dec->cur & MINOR_MASK;
443 { 824 ++dec->cur;
444 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447 return *dec->cur++ & 31;
448 825
449 case 24: 826 if (ecb_expect_true (m < LENGTH_EXT1))
827 return m;
828 else if (ecb_expect_true (m == LENGTH_EXT1))
829 {
450 WANT (2); 830 WANT (1);
451 dec->cur += 2; 831 dec->cur += 1;
452 return dec->cur[-1]; 832 return dec->cur[-1];
453 833 }
454 case 25: 834 else if (ecb_expect_true (m == LENGTH_EXT2))
835 {
455 WANT (3); 836 WANT (2);
456 dec->cur += 3; 837 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 838 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 839 | ((UV)dec->cur[-1]);
459 840 }
460 case 26: 841 else if (ecb_expect_true (m == LENGTH_EXT4))
842 {
461 WANT (5); 843 WANT (4);
462 dec->cur += 5; 844 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 845 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 846 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 847 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 848 | ((UV)dec->cur[-1]);
467 849 }
468 case 27: 850 else if (ecb_expect_true (m == LENGTH_EXT8))
851 {
469 WANT (9); 852 WANT (8);
470 dec->cur += 9; 853 dec->cur += 8;
854
855 return
856#if UVSIZE < 8
857 0
858#else
471 return (((UV)dec->cur[-8]) << 56) 859 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 860 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 861 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 862 | (((UV)dec->cur[-5]) << 32)
863#endif
475 | (((UV)dec->cur[-4]) << 24) 864 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 865 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 866 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 867 | ((UV)dec->cur[-1]);
479 868 }
480 default: 869 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 870 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 871
484fail: 872fail:
485 return 0; 873 return 0;
486} 874}
487 875
492{ 880{
493 AV *av = newAV (); 881 AV *av = newAV ();
494 882
495 DEC_INC_DEPTH; 883 DEC_INC_DEPTH;
496 884
497 if ((*dec->cur & 31) == 31) 885 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 886 {
499 ++dec->cur; 887 ++dec->cur;
500 888
501 for (;;) 889 for (;;)
502 { 890 {
503 WANT (1); 891 WANT (1);
504 892
505 if (*dec->cur == (0xe0 | 31)) 893 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
506 { 894 {
507 ++dec->cur; 895 ++dec->cur;
508 break; 896 break;
509 } 897 }
510 898
511 av_push (av, decode_sv (dec)); 899 av_push (av, decode_sv (dec));
512 } 900 }
513 } 901 }
514 else 902 else
515 { 903 {
516 int i, len = decode_uint (dec); 904 UV i, len = decode_uint (dec);
517 905
906 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518 av_fill (av, len - 1); 907 av_fill (av, len - 1);
519 908
520 for (i = 0; i < len; ++i) 909 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 910 AvARRAY (av)[i] = decode_sv (dec);
522 } 911 }
523 912
524 DEC_DEC_DEPTH; 913 DEC_DEC_DEPTH;
525 return newRV_noinc ((SV *)av); 914 return newRV_noinc ((SV *)av);
526 915
527fail: 916fail:
528 SvREFCNT_dec (av); 917 SvREFCNT_dec_NN (av);
529 DEC_DEC_DEPTH; 918 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 919 return &PL_sv_undef;
531} 920}
532 921
922static void
923decode_he (dec_t *dec, HV *hv)
924{
925 // for speed reasons, we specialcase single-string
926 // byte or utf-8 strings as keys, but only when !stringref
927
928 if (ecb_expect_true (!dec->stringref))
929 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
930 {
931 STRLEN len = decode_uint (dec);
932 char *key = (char *)dec->cur;
933
934 WANT (len);
935 dec->cur += len;
936
937 hv_store (hv, key, len, decode_sv (dec), 0);
938
939 return;
940 }
941 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
942 {
943 STRLEN len = decode_uint (dec);
944 char *key = (char *)dec->cur;
945
946 WANT (len);
947 dec->cur += len;
948
949 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
950 if (!is_utf8_string (key, len))
951 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
952
953 hv_store (hv, key, -len, decode_sv (dec), 0);
954
955 return;
956 }
957
958 SV *k = decode_sv (dec);
959 SV *v = decode_sv (dec);
960
961 // we leak memory if uncaught exceptions are thrown by random magical
962 // methods, and this is hopefully the only place where it can happen,
963 // so if there is a chance of an exception, take the very slow path.
964 // since catching exceptions is "undocumented/internal/forbidden" by
965 // the new p5p powers, we need to call out to a perl function :/
966 if (ecb_expect_false (SvAMAGIC (k)))
967 {
968 dSP;
969
970 ENTER; SAVETMPS;
971 PUSHMARK (SP);
972 EXTEND (SP, 3);
973 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
974 PUSHs (sv_2mortal (k));
975 PUSHs (sv_2mortal (v));
976
977 PUTBACK;
978 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
979 SPAGAIN;
980
981 FREETMPS; LEAVE;
982
983 if (SvTRUE (ERRSV))
984 ERR_ERRSV;
985
986 return;
987 }
988
989 hv_store_ent (hv, k, v, 0);
990 SvREFCNT_dec_NN (k);
991
992fail:
993 ;
994}
995
533static SV * 996static SV *
534decode_hv (dec_t *dec) 997decode_hv (dec_t *dec)
535{ 998{
536 HV *hv = newHV (); 999 HV *hv = newHV ();
537 1000
538 DEC_INC_DEPTH; 1001 DEC_INC_DEPTH;
539 1002
540 if ((*dec->cur & 31) == 31) 1003 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 1004 {
542 ++dec->cur; 1005 ++dec->cur;
543 1006
544 for (;;) 1007 for (;;)
545 { 1008 {
546 WANT (1); 1009 WANT (1);
547 1010
548 if (*dec->cur == (0xe0 | 31)) 1011 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
549 { 1012 {
550 ++dec->cur; 1013 ++dec->cur;
551 break; 1014 break;
552 } 1015 }
553 1016
554 SV *k = decode_sv (dec); 1017 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 1018 }
559 } 1019 }
560 else 1020 else
561 { 1021 {
562 int len = decode_uint (dec); 1022 UV pairs = decode_uint (dec);
563 1023
1024 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1025
564 while (len--) 1026 while (pairs--)
565 { 1027 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 1028 }
572 1029
573 DEC_DEC_DEPTH; 1030 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 1031 return newRV_noinc ((SV *)hv);
575 1032
576#if 0 1033fail:
577 SV *sv; 1034 SvREFCNT_dec_NN (hv);
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH; 1035 DEC_DEC_DEPTH;
581 decode_ws (dec); 1036 return &PL_sv_undef;
1037}
582 1038
583 for (;;) 1039static SV *
584 { 1040decode_str (dec_t *dec, int utf8)
585 // heuristic: assume that 1041{
586 // a) decode_str + hv_store_ent are abysmally slow. 1042 SV *sv = 0;
587 // b) most hash keys are short, simple ascii text. 1043
588 // => try to "fast-match" such strings to avoid 1044 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
589 // the overhead of decode_str + hv_store_ent.
590 { 1045 {
591 SV *value; 1046 // indefinite length strings
592 char *p = dec->cur; 1047 ++dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594 1048
1049 U8 major = *dec->cur & MAJOR_MISC;
1050
1051 sv = newSVpvn ("", 0);
1052
595 for (;;) 1053 for (;;)
596 { 1054 {
597 // the >= 0x80 is false on most architectures 1055 WANT (1);
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1056
1057 if ((*dec->cur - major) > LENGTH_EXT8)
1058 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 1059 {
600 // slow path, back up and use decode_str 1060 ++dec->cur;
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break; 1061 break;
619 } 1062 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 { 1063 else
685 dSP; 1064 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
686 int count;
687 1065
688 ENTER; SAVETMPS; PUSHMARK (SP); 1066 STRLEN len = decode_uint (dec);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691 1067
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730
731fail:
732 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH;
734 return &PL_sv_undef;
735}
736
737static SV *
738decode_str (dec_t *dec, int utf8)
739{
740 SV *sv;
741
742 if ((*dec->cur & 31) == 31)
743 {
744 ++dec->cur;
745
746 sv = newSVpvn ("", 0);
747
748 // not very fast, and certainly not robust against illegal input
749 for (;;)
750 {
751 WANT (1); 1068 WANT (len);
752 1069 sv_catpvn (sv, dec->cur, len);
753 if (*dec->cur == (0xe0 | 31)) 1070 dec->cur += len;
754 {
755 ++dec->cur;
756 break;
757 }
758
759 SV *sv2 = decode_sv (dec);
760 sv_catsv (sv, sv2);
761 } 1071 }
762 } 1072 }
763 else 1073 else
764 { 1074 {
765 STRLEN len = decode_uint (dec); 1075 STRLEN len = decode_uint (dec);
766 1076
767 WANT (len); 1077 WANT (len);
768 sv = newSVpvn (dec->cur, len); 1078 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 1079 dec->cur += len;
1080
1081 if (ecb_expect_false (dec->stringref)
1082 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1083 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 1084 }
771 1085
772 if (utf8) 1086 if (utf8)
1087 {
1088 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1089 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1090 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1091
773 SvUTF8_on (sv); 1092 SvUTF8_on (sv);
1093 }
774 1094
775 return sv; 1095 return sv;
776 1096
777fail: 1097fail:
1098 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 1099 return &PL_sv_undef;
779} 1100}
780 1101
781static SV * 1102static SV *
782decode_tagged (dec_t *dec) 1103decode_tagged (dec_t *dec)
783{ 1104{
1105 SV *sv = 0;
784 UV tag = decode_uint (dec); 1106 UV tag = decode_uint (dec);
1107
1108 WANT (1);
1109
1110 switch (tag)
1111 {
1112 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 1113 sv = decode_sv (dec);
1114 break;
786 1115
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1116 case CBOR_TAG_INDIRECTION:
1117 sv = newRV_noinc (decode_sv (dec));
1118 break;
1119
1120 case CBOR_TAG_STRINGREF_NAMESPACE:
1121 {
1122 // do not use SAVETMPS/FREETMPS, as these will
1123 // erase mortalised caches, e.g. "shareable"
1124 ENTER;
1125
1126 SAVESPTR (dec->stringref);
1127 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1128
1129 sv = decode_sv (dec);
1130
1131 LEAVE;
1132 }
1133 break;
1134
1135 case CBOR_TAG_STRINGREF:
1136 {
1137 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1138 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1139
1140 UV idx = decode_uint (dec);
1141
1142 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1143 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1144
1145 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1146 }
1147 break;
1148
1149 case CBOR_TAG_VALUE_SHAREABLE:
1150 {
1151 if (ecb_expect_false (!dec->shareable))
1152 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1153
1154 if (dec->cbor.flags & F_ALLOW_CYCLES)
1155 {
1156 sv = newSV (0);
1157 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1158
1159 SV *osv = decode_sv (dec);
1160 sv_setsv (sv, osv);
1161 SvREFCNT_dec_NN (osv);
1162 }
1163 else
1164 {
1165 av_push (dec->shareable, &PL_sv_undef);
1166 int idx = AvFILLp (dec->shareable);
1167 sv = decode_sv (dec);
1168 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1169 }
1170 }
1171 break;
1172
1173 case CBOR_TAG_VALUE_SHAREDREF:
1174 {
1175 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1176 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1177
1178 UV idx = decode_uint (dec);
1179
1180 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1181 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1182
1183 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1184
1185 if (sv == &PL_sv_undef)
1186 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1187 }
1188 break;
1189
1190 case CBOR_TAG_PERL_OBJECT:
1191 {
1192 if (dec->cbor.flags & F_FORBID_OBJECTS)
1193 goto filter;
1194
1195 sv = decode_sv (dec);
1196
1197 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1198 ERR ("corrupted CBOR data (non-array perl object)");
1199
1200 AV *av = (AV *)SvRV (sv);
1201 int len = av_len (av) + 1;
1202 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1203
1204 if (!stash)
1205 ERR ("cannot decode perl-object (package does not exist)");
1206
1207 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1208
1209 if (!method)
1210 ERR ("cannot decode perl-object (package does not have a THAW method)");
1211
1212 dSP;
1213
1214 ENTER; SAVETMPS;
1215 PUSHMARK (SP);
1216 EXTEND (SP, len + 1);
1217 // we re-bless the reference to get overload and other niceties right
1218 PUSHs (*av_fetch (av, 0, 1));
1219 PUSHs (sv_cbor);
1220
1221 int i;
1222
1223 for (i = 1; i < len; ++i)
1224 PUSHs (*av_fetch (av, i, 1));
1225
1226 PUTBACK;
1227 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1228 SPAGAIN;
1229
1230 if (SvTRUE (ERRSV))
1231 {
1232 FREETMPS; LEAVE;
1233 ERR_ERRSV;
1234 }
1235
1236 SvREFCNT_dec_NN (sv);
1237 sv = SvREFCNT_inc (POPs);
1238
1239 PUTBACK;
1240
1241 FREETMPS; LEAVE;
1242 }
1243 break;
1244
1245 default:
1246 filter:
1247 {
1248 SV *tag_sv = newSVuv (tag);
1249
1250 sv = decode_sv (dec);
1251
1252 dSP;
1253 ENTER; SAVETMPS;
1254 PUSHMARK (SP);
1255 EXTEND (SP, 2);
1256 PUSHs (tag_sv);
1257 PUSHs (sv);
1258
1259 PUTBACK;
1260 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1261 SPAGAIN;
1262
1263 if (SvTRUE (ERRSV))
1264 {
1265 SvREFCNT_dec_NN (tag_sv);
1266 FREETMPS; LEAVE;
1267 ERR_ERRSV;
1268 }
1269
1270 if (count)
1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1273 SvREFCNT_dec_NN (sv);
1274 sv = SvREFCNT_inc_NN (TOPs);
1275 SP -= count;
1276 }
1277 else
1278 {
1279 AV *av = newAV ();
1280 av_push (av, tag_sv);
1281 av_push (av, sv);
1282
1283 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1284 ? cbor_tagged_stash
1285 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1286 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1287 }
1288
1289 PUTBACK;
1290
1291 FREETMPS; LEAVE;
1292 }
1293 break;
1294 }
1295
788 return sv; 1296 return sv;
789 1297
790 AV *av = newAV (); 1298fail:
791 av_push (av, newSVuv (tag)); 1299 SvREFCNT_dec (sv);
792 av_push (av, sv); 1300 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1301}
795 1302
796static SV * 1303static SV *
797decode_sv (dec_t *dec) 1304decode_sv (dec_t *dec)
798{ 1305{
799 WANT (1); 1306 WANT (1);
800 1307
801 switch (*dec->cur >> 5) 1308 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1309 {
803 case 0: // unsigned int 1310 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1311 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1312 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1313 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1314 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1315 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1316 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1317
811 return decode_str (dec, 1); 1318 case MAJOR_MISC >> MAJOR_SHIFT:
812 case 4: // array
813 return decode_av (dec);
814 case 5: // map
815 return decode_hv (dec);
816 case 6: // tag
817 return decode_tagged (dec);
818 case 7: // misc
819 switch (*dec->cur++ & 31) 1319 switch (*dec->cur++ & MINOR_MASK)
820 { 1320 {
821 case 20: 1321 case SIMPLE_FALSE:
822#if CBOR_SLOW 1322#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1323 types_false = get_bool ("Types::Serialiser::false");
824#endif 1324#endif
825 return newSVsv (cbor_false); 1325 return newSVsv (types_false);
826 case 21: 1326 case SIMPLE_TRUE:
827#if CBOR_SLOW 1327#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1328 types_true = get_bool ("Types::Serialiser::true");
829#endif 1329#endif
830 return newSVsv (cbor_true); 1330 return newSVsv (types_true);
831 case 22: 1331 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1332 return newSVsv (&PL_sv_undef);
1333 case SIMPLE_UNDEF:
1334#if CBOR_SLOW
1335 types_error = get_bool ("Types::Serialiser::error");
1336#endif
1337 return newSVsv (types_error);
833 1338
834 case 25: 1339 case MISC_FLOAT16:
835 { 1340 {
836 WANT (2); 1341 WANT (2);
837 1342
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1343 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1344 dec->cur += 2;
840 1345
841 return newSVnv (ecb_binary16_to_float (fp)); 1346 return newSVnv (ecb_binary16_to_float (fp));
842 } 1347 }
843 1348
844 case 26: 1349 case MISC_FLOAT32:
845 { 1350 {
846 uint32_t fp; 1351 uint32_t fp;
847 WANT (4); 1352 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1353 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1354 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1357 fp = ecb_bswap32 (fp);
853 1358
854 return newSVnv (ecb_binary32_to_float (fp)); 1359 return newSVnv (ecb_binary32_to_float (fp));
855 } 1360 }
856 1361
857 case 27: 1362 case MISC_FLOAT64:
858 { 1363 {
859 uint64_t fp; 1364 uint64_t fp;
860 WANT (8); 1365 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1366 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1367 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1370 fp = ecb_bswap64 (fp);
866 1371
867 return newSVnv (ecb_binary64_to_double (fp)); 1372 return newSVnv (ecb_binary64_to_double (fp));
868 } 1373 }
869 1374
870 // 0..19 unassigned 1375 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1376 // 24 reserved + unassigned simple (reserved values are not encodable)
1377 // 28-30 unassigned misc
1378 // 31 break code
872 default: 1379 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1380 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1381 }
875 1382
876 break; 1383 break;
877 } 1384 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 1385
935fail: 1386fail:
936 return &PL_sv_undef; 1387 return &PL_sv_undef;
937} 1388}
938 1389
939static SV * 1390static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1391decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1392{
942 dec_t dec; 1393 dec_t dec = { 0 };
943 SV *sv; 1394 SV *sv;
1395 STRLEN len;
1396 char *data = SvPVbyte (string, len);
944 1397
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 1398 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1399 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1400 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1401
980 dec.cbor = *cbor; 1402 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1403 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1404 dec.end = (U8 *)data + len;
983 dec.err = 0;
984 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 1405
989 sv = decode_sv (&dec); 1406 sv = decode_sv (&dec);
990 1407
991 if (offset_return) 1408 if (offset_return)
992 *offset_return = dec.cur; 1409 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1412 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1413 dec.err = "garbage after CBOR object";
997 1414
998 if (dec.err) 1415 if (dec.err)
999 { 1416 {
1417 if (dec.shareable)
1418 {
1419 // need to break cyclic links, which would all be in shareable
1420 int i;
1421 SV **svp;
1422
1423 for (i = av_len (dec.shareable) + 1; i--; )
1424 if ((svp = av_fetch (dec.shareable, i, 0)))
1425 sv_setsv (*svp, &PL_sv_undef);
1426 }
1427
1000 SvREFCNT_dec (sv); 1428 SvREFCNT_dec_NN (sv);
1429
1430 if (dec.err_sv)
1431 sv_2mortal (dec.err_sv);
1432
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1433 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1434 }
1003 1435
1004 sv = sv_2mortal (sv); 1436 sv = sv_2mortal (sv);
1005 1437
1006 return sv; 1438 return sv;
1007} 1439}
1008 1440
1441/////////////////////////////////////////////////////////////////////////////
1442// incremental parser
1443
1444#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1445
1446// returns 0 for notyet, 1 for success or error
1447static int
1448incr_parse (CBOR *self, SV *cborstr)
1449{
1450 STRLEN cur;
1451 SvPV (cborstr, cur);
1452
1453 while (ecb_expect_true (self->incr_need <= cur))
1454 {
1455 // table of integer count bytes
1456 static I8 incr_len[MINOR_MASK + 1] = {
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 0, 0, 0, 0, 0, 0, 0, 0,
1459 0, 0, 0, 0, 0, 0, 0, 0,
1460 1, 2, 4, 8,-1,-1,-1,-2
1461 };
1462
1463 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1464 U8 m = *p & MINOR_MASK;
1465 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1466 I8 ilen = incr_len[m];
1467
1468 self->incr_need = self->incr_pos + 1;
1469
1470 if (ecb_expect_false (ilen < 0))
1471 {
1472 if (m != MINOR_INDEF)
1473 return 1; // error
1474
1475 if (*p == (MAJOR_MISC | MINOR_INDEF))
1476 {
1477 if (count >= 0)
1478 return 1; // error
1479
1480 count = 1;
1481 }
1482 else
1483 {
1484 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1485 count = -1;
1486 }
1487 }
1488 else
1489 {
1490 self->incr_need += ilen;
1491 if (ecb_expect_false (self->incr_need > cur))
1492 return 0;
1493
1494 int major = *p >> MAJOR_SHIFT;
1495
1496 switch (major)
1497 {
1498 case MAJOR_TAG >> MAJOR_SHIFT:
1499 ++count; // tags merely prefix another value
1500 break;
1501
1502 case MAJOR_BYTES >> MAJOR_SHIFT:
1503 case MAJOR_TEXT >> MAJOR_SHIFT:
1504 case MAJOR_ARRAY >> MAJOR_SHIFT:
1505 case MAJOR_MAP >> MAJOR_SHIFT:
1506 {
1507 UV len;
1508
1509 if (ecb_expect_false (ilen))
1510 {
1511 len = 0;
1512
1513 do {
1514 len = (len << 8) | *++p;
1515 } while (--ilen);
1516 }
1517 else
1518 len = m;
1519
1520 switch (major)
1521 {
1522 case MAJOR_BYTES >> MAJOR_SHIFT:
1523 case MAJOR_TEXT >> MAJOR_SHIFT:
1524 self->incr_need += len;
1525 if (ecb_expect_false (self->incr_need > cur))
1526 return 0;
1527
1528 break;
1529
1530 case MAJOR_MAP >> MAJOR_SHIFT:
1531 len <<= 1;
1532 case MAJOR_ARRAY >> MAJOR_SHIFT:
1533 if (len)
1534 {
1535 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1536 count = len + 1;
1537 }
1538 break;
1539 }
1540 }
1541 }
1542 }
1543
1544 self->incr_pos = self->incr_need;
1545
1546 if (count > 0)
1547 {
1548 while (!--count)
1549 {
1550 if (!AvFILLp (self->incr_count))
1551 return 1; // done
1552
1553 SvREFCNT_dec_NN (av_pop (self->incr_count));
1554 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1555 }
1556
1557 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1558 }
1559 }
1560
1561 return 0;
1562}
1563
1564
1009///////////////////////////////////////////////////////////////////////////// 1565/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1566// XS interface functions
1011 1567
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1568MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1569
1014BOOT: 1570BOOT:
1015{ 1571{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1572 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1573 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1574
1019 cbor_true = get_bool ("CBOR::XS::true"); 1575 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1576 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1577
1578 types_true = get_bool ("Types::Serialiser::true" );
1579 types_false = get_bool ("Types::Serialiser::false");
1580 types_error = get_bool ("Types::Serialiser::error");
1581
1582 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1583
1584 sv_cbor = newSVpv ("CBOR", 0);
1585 SvREADONLY_on (sv_cbor);
1586
1587 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1021} 1588}
1022 1589
1023PROTOTYPES: DISABLE 1590PROTOTYPES: DISABLE
1024 1591
1025void CLONE (...) 1592void CLONE (...)
1026 CODE: 1593 CODE:
1027 cbor_stash = 0; 1594 cbor_stash = 0;
1595 cbor_tagged_stash = 0;
1596 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1597 types_boolean_stash = 0;
1029 1598
1030void new (char *klass) 1599void new (char *klass)
1031 PPCODE: 1600 PPCODE:
1032{ 1601{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1602 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1610
1042void shrink (CBOR *self, int enable = 1) 1611void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1612 ALIAS:
1044 shrink = F_SHRINK 1613 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1614 allow_unknown = F_ALLOW_UNKNOWN
1615 allow_sharing = F_ALLOW_SHARING
1616 allow_cycles = F_ALLOW_CYCLES
1617 forbid_objects = F_FORBID_OBJECTS
1618 pack_strings = F_PACK_STRINGS
1619 text_keys = F_TEXT_KEYS
1620 text_strings = F_TEXT_STRINGS
1621 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1622 PPCODE:
1047{ 1623{
1048 if (enable) 1624 if (enable)
1049 self->flags |= ix; 1625 self->flags |= ix;
1050 else 1626 else
1055 1631
1056void get_shrink (CBOR *self) 1632void get_shrink (CBOR *self)
1057 ALIAS: 1633 ALIAS:
1058 get_shrink = F_SHRINK 1634 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1635 get_allow_unknown = F_ALLOW_UNKNOWN
1636 get_allow_sharing = F_ALLOW_SHARING
1637 get_allow_cycles = F_ALLOW_CYCLES
1638 get_forbid_objects = F_FORBID_OBJECTS
1639 get_pack_strings = F_PACK_STRINGS
1640 get_text_keys = F_TEXT_KEYS
1641 get_text_strings = F_TEXT_STRINGS
1642 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1643 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1644 XPUSHs (boolSV (self->flags & ix));
1062 1645
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1646void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1647 PPCODE:
1080 CODE: 1663 CODE:
1081 RETVAL = self->max_size; 1664 RETVAL = self->max_size;
1082 OUTPUT: 1665 OUTPUT:
1083 RETVAL 1666 RETVAL
1084 1667
1085#if 0 //TODO 1668void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1669 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1670 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1671 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1672 XPUSHs (ST (0));
1094}
1095 1673
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1674SV *get_filter (CBOR *self)
1097 PPCODE: 1675 CODE:
1098{ 1676 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1677 OUTPUT:
1100 self->cb_sk_object = newHV (); 1678 RETVAL
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119 1679
1120void encode (CBOR *self, SV *scalar) 1680void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1681 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1682 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1683 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1696 EXTEND (SP, 2);
1137 PUSHs (sv); 1697 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1698 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1699}
1140 1700
1701void incr_parse (CBOR *self, SV *cborstr)
1702 ALIAS:
1703 incr_parse_multiple = 1
1704 PPCODE:
1705{
1706 if (SvUTF8 (cborstr))
1707 sv_utf8_downgrade (cborstr, 0);
1708
1709 if (!self->incr_count)
1710 {
1711 self->incr_count = newAV ();
1712 self->incr_pos = 0;
1713 self->incr_need = 1;
1714
1715 av_push (self->incr_count, newSViv (1));
1716 }
1717
1718 do
1719 {
1720 if (!incr_parse (self, cborstr))
1721 {
1722 if (self->incr_need > self->max_size && self->max_size)
1723 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1724 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1725
1726 break;
1727 }
1728
1729 SV *sv;
1730 char *offset;
1731
1732 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1733 XPUSHs (sv);
1734
1735 sv_chop (cborstr, offset);
1736
1737 av_clear (self->incr_count);
1738 av_push (self->incr_count, newSViv (1));
1739
1740 self->incr_pos = 0;
1741 self->incr_need = self->incr_pos + 1;
1742 }
1743 while (ix);
1744}
1745
1746void incr_reset (CBOR *self)
1747 CODE:
1748{
1749 SvREFCNT_dec (self->incr_count);
1750 self->incr_count = 0;
1751}
1752
1141void DESTROY (CBOR *self) 1753void DESTROY (CBOR *self)
1142 CODE: 1754 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1755 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1756
1146PROTOTYPES: ENABLE 1757PROTOTYPES: ENABLE
1147 1758
1148void encode_cbor (SV *scalar) 1759void encode_cbor (SV *scalar)
1760 ALIAS:
1761 encode_cbor = 0
1762 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1763 PPCODE:
1150{ 1764{
1151 CBOR cbor; 1765 CBOR cbor;
1152 cbor_init (&cbor); 1766 cbor_init (&cbor);
1767 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1768 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1769 XPUSHs (scalar);
1155} 1770}
1156 1771
1157void decode_cbor (SV *cborstr) 1772void decode_cbor (SV *cborstr)
1161 cbor_init (&cbor); 1776 cbor_init (&cbor);
1162 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1777 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1163 XPUSHs (cborstr); 1778 XPUSHs (cborstr);
1164} 1779}
1165 1780
1781#ifdef __AFL_COMPILER
1782
1783void
1784afl_init ()
1785 CODE:
1786 __AFL_INIT ();
1787
1788int
1789afl_loop (unsigned int count = 10000)
1790 CODE:
1791 RETVAL = __AFL_LOOP (count);
1792 OUTPUT:
1793 RETVAL
1794
1795#endif
1796

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines