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 1.1 by root, Fri Oct 25 23:09:45 2013 UTC vs.
Revision 1.69 by root, Mon Nov 30 20:38:25 2020 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14#if defined(__BORLANDC__) || defined(_MSC_VER) 16// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 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
17 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
18#define F_SHRINK 0x00000200UL 117#define F_SHRINK 0x00000001UL
19#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
20 126
21#define INIT_SIZE 32 // initial scalar size to be allocated 127#define INIT_SIZE 32 // initial scalar size to be allocated
22 128
23#define SB do { 129#define SB do {
24#define SE } while (0) 130#define SE } while (0)
25
26#if __GNUC__ >= 3
27# define expect(expr,value) __builtin_expect ((expr), (value))
28# define INLINE static inline
29#else
30# define expect(expr,value) (expr)
31# define INLINE static
32#endif
33
34#define expect_false(expr) expect ((expr) != 0, 0)
35#define expect_true(expr) expect ((expr) != 0, 1)
36 131
37#define IN_RANGE_INC(type,val,beg,end) \ 132#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 133 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 134 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 135
46#else 141#else
47# define CBOR_SLOW 0 142# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 143# define CBOR_STASH cbor_stash
49#endif 144#endif
50 145
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 146static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 147static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
53 148
54typedef struct { 149typedef struct {
55 U32 flags; 150 U32 flags;
56 U32 max_depth; 151 U32 max_depth;
57 STRLEN max_size; 152 STRLEN max_size;
153 SV *filter;
58 154
59 SV *cb_object; 155 // for the incremental parser
60 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.
61} CBOR; 159} CBOR;
62 160
63INLINE void 161ecb_inline void
64cbor_init (CBOR *cbor) 162cbor_init (CBOR *cbor)
65{ 163{
66 Zero (cbor, 1, CBOR); 164 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 165 cbor->max_depth = 512;
68} 166}
69 167
168ecb_inline void
169cbor_free (CBOR *cbor)
170{
171 SvREFCNT_dec (cbor->filter);
172 SvREFCNT_dec (cbor->incr_count);
173}
174
70///////////////////////////////////////////////////////////////////////////// 175/////////////////////////////////////////////////////////////////////////////
71// utility functions 176// utility functions
72 177
73INLINE SV * 178ecb_inline SV *
74get_bool (const char *name) 179get_bool (const char *name)
75{ 180{
76 SV *sv = get_sv (name, 1); 181 SV *sv = get_sv (name, 1);
77 182
78 SvREADONLY_on (sv); 183 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 184 SvREADONLY_on (SvRV (sv));
80 185
81 return sv; 186 return sv;
82} 187}
83 188
84INLINE void 189ecb_inline void
85shrink (SV *sv) 190shrink (SV *sv)
86{ 191{
87 sv_utf8_downgrade (sv, 1); 192 sv_utf8_downgrade (sv, 1);
88 193
89 if (SvLEN (sv) > SvCUR (sv) + 1) 194 if (SvLEN (sv) > SvCUR (sv) + 1)
94 SvPV_renew (sv, SvCUR (sv) + 1); 199 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 200#endif
96 } 201 }
97} 202}
98 203
99///////////////////////////////////////////////////////////////////////////// 204// minimum length of a string to be registered for stringref
100// fp hell 205ecb_inline int
101 206minimum_string_length (UV idx)
102//TODO 207{
208 return idx <= 23 ? 3
209 : idx <= 0xffU ? 4
210 : idx <= 0xffffU ? 5
211 : idx <= 0xffffffffU ? 7
212 : 11;
213}
103 214
104///////////////////////////////////////////////////////////////////////////// 215/////////////////////////////////////////////////////////////////////////////
105// encoder 216// encoder
106 217
107// structure used for encoding CBOR 218// structure used for encoding CBOR
110 char *cur; // SvPVX (sv) + current output position 221 char *cur; // SvPVX (sv) + current output position
111 char *end; // SvEND (sv) 222 char *end; // SvEND (sv)
112 SV *sv; // result scalar 223 SV *sv; // result scalar
113 CBOR cbor; 224 CBOR cbor;
114 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;
115} enc_t; 230} enc_t;
116 231
117INLINE void 232ecb_inline void
118need (enc_t *enc, STRLEN len) 233need (enc_t *enc, STRLEN len)
119{ 234{
120 if (expect_false (enc->cur + len >= enc->end)) 235 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
121 { 236 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 237 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 238 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 239 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 240 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 241 }
127} 242}
128 243
129INLINE void 244ecb_inline void
130encode_ch (enc_t *enc, char ch) 245encode_ch (enc_t *enc, char ch)
131{ 246{
132 need (enc, 1); 247 need (enc, 1);
133 *enc->cur++ = ch; 248 *enc->cur++ = ch;
134} 249}
135 250
251// used for tags, intregers, element counts and so on
136static void 252static void
137encode_uint (enc_t *enc, int major, UV len) 253encode_uint (enc_t *enc, int major, UV len)
138{ 254{
139 need (enc, 9); 255 need (enc, 9);
140 256
141 if (len < 24) 257 if (ecb_expect_true (len < LENGTH_EXT1))
142 *enc->cur++ = major | len; 258 *enc->cur++ = major | len;
143 else if (len < 0x100) 259 else if (ecb_expect_true (len <= 0xffU))
144 { 260 {
145 *enc->cur++ = major | 24; 261 *enc->cur++ = major | LENGTH_EXT1;
146 *enc->cur++ = len; 262 *enc->cur++ = len;
147 } 263 }
148 else if (len < 0x10000) 264 else if (len <= 0xffffU)
149 { 265 {
150 *enc->cur++ = major | 25; 266 *enc->cur++ = major | LENGTH_EXT2;
151 *enc->cur++ = len >> 8; 267 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 268 *enc->cur++ = len;
153 } 269 }
154 else if (len < 0x100000000) 270 else if (len <= 0xffffffffU)
155 { 271 {
156 *enc->cur++ = major | 26; 272 *enc->cur++ = major | LENGTH_EXT4;
157 *enc->cur++ = len >> 24; 273 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 274 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 275 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 276 *enc->cur++ = len;
161 } 277 }
162 else if (len) 278 else
163 { 279 {
164 *enc->cur++ = major | 27; 280 *enc->cur++ = major | LENGTH_EXT8;
165 *enc->cur++ = len >> 56; 281 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 282 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 283 *enc->cur++ = len >> 40;
168 *enc->cur++ = len >> 32; 284 *enc->cur++ = len >> 32;
169 *enc->cur++ = len >> 24; 285 *enc->cur++ = len >> 24;
171 *enc->cur++ = len >> 8; 287 *enc->cur++ = len >> 8;
172 *enc->cur++ = len; 288 *enc->cur++ = len;
173 } 289 }
174} 290}
175 291
176static 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
177encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 312encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
178{ 313{
179 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);
180 need (enc, len); 344 need (enc, len);
181 memcpy (enc->cur, str, len); 345 memcpy (enc->cur, str, len);
182 enc->cur += len; 346 enc->cur += len;
183} 347}
184 348
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_bool (enc_t *enc, int istrue)
424{
425 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
426}
427
428ecb_inline void
429encode_forced (enc_t *enc, UV type, SV *sv)
430{
431 switch (type)
432 {
433 case AS_CBOR:
434 {
435 STRLEN len;
436 char *str = SvPVbyte (sv, len);
437
438 need (enc, len);
439 memcpy (enc->cur, str, len);
440 enc->cur += len;
441 }
442 break;
443
444 case AS_BYTES:
445 {
446 STRLEN len;
447 char *str = SvPVbyte (sv, len);
448 encode_strref (enc, 0, 0, str, len);
449 }
450 break;
451
452 case AS_TEXT:
453 {
454 STRLEN len;
455 char *str = SvPVutf8 (sv, len);
456 encode_strref (enc, 1, 1, str, len);
457 }
458 break;
459
460 case AS_INT: encode_int (enc, sv); break;
461
462 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
463 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
464 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
465
466 default:
467 croak ("encountered malformed CBOR::XS::Tagged object");
468 }
469}
470
185static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
186 472
187static void 473static void
188encode_av (enc_t *enc, AV *av) 474encode_av (enc_t *enc, AV *av)
189{ 475{
192 if (enc->depth >= enc->cbor.max_depth) 478 if (enc->depth >= enc->cbor.max_depth)
193 croak (ERR_NESTING_EXCEEDED); 479 croak (ERR_NESTING_EXCEEDED);
194 480
195 ++enc->depth; 481 ++enc->depth;
196 482
197 encode_uint (enc, 0x80, len + 1); 483 encode_uint (enc, MAJOR_ARRAY, len + 1);
198 484
485 if (ecb_expect_false (SvMAGICAL (av)))
199 for (i = 0; i <= len; ++i) 486 for (i = 0; i <= len; ++i)
200 { 487 {
201 SV **svp = av_fetch (av, i, 0); 488 SV **svp = av_fetch (av, i, 0);
202 encode_sv (enc, svp ? *svp : &PL_sv_undef); 489 encode_sv (enc, svp ? *svp : &PL_sv_undef);
203 } 490 }
491 else
492 for (i = 0; i <= len; ++i)
493 {
494 SV *sv = AvARRAY (av)[i];
495 encode_sv (enc, sv ? sv : &PL_sv_undef);
496 }
204 497
205 --enc->depth; 498 --enc->depth;
206} 499}
207 500
208static void 501static void
216 ++enc->depth; 509 ++enc->depth;
217 510
218 int pairs = hv_iterinit (hv); 511 int pairs = hv_iterinit (hv);
219 int mg = SvMAGICAL (hv); 512 int mg = SvMAGICAL (hv);
220 513
221 if (mg) 514 if (ecb_expect_false (mg))
222 encode_ch (enc, 0xa0 | 31); 515 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
223 else 516 else
224 encode_uint (enc, 0xa0, pairs); 517 encode_uint (enc, MAJOR_MAP, pairs);
225 518
226 while ((he = hv_iternext (hv))) 519 while ((he = hv_iternext (hv)))
227 { 520 {
228 if (HeKLEN (he) == HEf_SVKEY) 521 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 522 encode_sv (enc, HeSVKEY (he));
230 else 523 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 524 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 525
233 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 526 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
234 } 527 }
235 528
236 if (mg) 529 if (ecb_expect_false (mg))
237 encode_ch (enc, 0xe0 | 31); 530 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
238 531
239 --enc->depth; 532 --enc->depth;
240} 533}
241 534
242// encode objects, arrays and special \0=false and \1=true values. 535// encode objects, arrays and special \0=false and \1=true values.
243static void 536static void
244encode_rv (enc_t *enc, SV *sv) 537encode_rv (enc_t *enc, SV *sv)
245{ 538{
246 svtype svt;
247
248 SvGETMAGIC (sv); 539 SvGETMAGIC (sv);
540
249 svt = SvTYPE (sv); 541 svtype svt = SvTYPE (sv);
250 542
251 if (expect_false (SvOBJECT (sv))) 543 if (ecb_expect_false (SvOBJECT (sv)))
252 { 544 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 545 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? cbor_boolean_stash 546 ? types_boolean_stash
547 : gv_stashpv ("Types::Serialiser::Boolean", 1);
548 HV *error_stash = !CBOR_SLOW || types_error_stash
549 ? types_error_stash
550 : gv_stashpv ("Types::Serialiser::Error", 1);
551 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
552 ? cbor_tagged_stash
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 553 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 554
257 if (SvSTASH (sv) == stash) 555 HV *stash = SvSTASH (sv);
258 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 556
557 if (stash == boolean_stash)
558 {
559 encode_bool (enc, SvIV (sv));
560 return;
561 }
562 else if (stash == error_stash)
563 {
564 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
565 return;
566 }
567 else if (stash == tagged_stash)
568 {
569 if (svt != SVt_PVAV)
570 croak ("encountered CBOR::XS::Tagged object that isn't an array");
571
572 switch (av_len ((AV *)sv))
573 {
574 case 2-1:
575 // actually a tagged value
576 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
577 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
578 break;
579
580 case 3-1:
581 // a forced type [value, type, undef]
582 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
583 break;
584
585 default:
586 croak ("encountered malformed CBOR::XS::Tagged object");
587 }
588
589 return;
590 }
591 }
592
593 if (ecb_expect_false (SvREFCNT (sv) > 1)
594 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
595 {
596 if (ecb_expect_false (!enc->shareable))
597 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
598
599 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
600
601 if (SvOK (*svp))
602 {
603 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
604 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
605 return;
606 }
259 else 607 else
260 { 608 {
261#if 0 //TODO 609 sv_setuv (*svp, enc->shareable_idx);
262 if (enc->cbor.flags & F_CONV_BLESSED) 610 ++enc->shareable_idx;
611 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
612 }
613 }
614
615 if (ecb_expect_false (SvOBJECT (sv)))
616 {
617 HV *stash = SvSTASH (sv);
618 GV *method;
619
620 if (enc->cbor.flags & F_FORBID_OBJECTS)
621 croak ("encountered object '%s', but forbid_objects is enabled",
622 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
623 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
263 { 624 {
625 dSP;
626
627 ENTER; SAVETMPS;
628 PUSHMARK (SP);
264 // we re-bless the reference to get overload and other niceties right 629 // we re-bless the reference to get overload and other niceties right
265 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
266
267 if (to_cbor)
268 {
269 dSP;
270
271 ENTER; SAVETMPS; PUSHMARK (SP);
272 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 630 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
273 631
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK; 632 PUTBACK;
633 // G_SCALAR ensures that return value is 1
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 634 call_sv ((SV *)GvCV (method), G_SCALAR);
277 SPAGAIN; 635 SPAGAIN;
278 636
279 // catch this surprisingly common error 637 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv) 638 if (SvROK (TOPs) && SvRV (TOPs) == sv)
281 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 639 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
282 640
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv); 641 encode_sv (enc, POPs);
287 642
643 PUTBACK;
644
288 FREETMPS; LEAVE; 645 FREETMPS; LEAVE;
289 } 646 }
290 else if (enc->cbor.flags & F_ALLOW_BLESSED) 647 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
291 encode_str (enc, "null", 4, 0); 648 {
292 else 649 dSP;
293 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 650
294 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 651 ENTER; SAVETMPS;
652 PUSHMARK (SP);
653 EXTEND (SP, 2);
654 // we re-bless the reference to get overload and other niceties right
655 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
656 PUSHs (sv_cbor);
657
658 PUTBACK;
659 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
660 SPAGAIN;
661
662 // catch this surprisingly common error
663 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
664 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
665
666 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
667 encode_uint (enc, MAJOR_ARRAY, count + 1);
668 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
669
670 {
671 int i;
672
673 for (i = 0; i < count; ++i)
674 encode_sv (enc, SP[i + 1 - count]);
675
676 SP -= count;
295 } 677 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED) 678
297 encode_str (enc, "null", 4, 0); 679 PUTBACK;
680
681 FREETMPS; LEAVE;
682 }
298 else 683 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 684 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 685 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 686 }
304 else if (svt == SVt_PVHV) 687 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 688 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 689 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 690 encode_av (enc, (AV *)sv);
308 else if (svt < SVt_PVAV)
309 {
310 STRLEN len = 0;
311 char *pv = svt ? SvPV (sv, len) : 0;
312
313 if (len == 1 && *pv == '1')
314 encode_ch (enc, 0xe0 | 21);
315 else if (len == 1 && *pv == '0')
316 encode_ch (enc, 0xe0 | 20);
317 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
318 encode_ch (enc, 0xe0 | 23);
319 else
320 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
321 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
322 }
323 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
324 encode_ch (enc, 0xe0 | 23);
325 else 691 else
326 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 692 {
327 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 693 encode_tag (enc, CBOR_TAG_INDIRECTION);
694 encode_sv (enc, sv);
695 }
328} 696}
329 697
330static void 698static void
331encode_nv (enc_t *enc, SV *sv) 699encode_nv (enc_t *enc, SV *sv)
332{ 700{
333 double nv = SvNVX (sv); 701 double nv = SvNVX (sv);
334 702
335 need (enc, 9); 703 need (enc, 9);
336 704
337 if (expect_false (nv == (U32)nv)) 705 if (ecb_expect_false (nv == (NV)(U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 706 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
339 //TODO: maybe I32? 707 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 708 else if (ecb_expect_false (nv == (float)nv))
341 { 709 encode_float32 (enc, nv);
342 uint32_t fp = ecb_float_to_binary32 (nv);
343
344 *enc->cur++ = 0xe0 | 26;
345
346 if (!ecb_big_endian ())
347 fp = ecb_bswap32 (fp);
348
349 memcpy (enc->cur, &fp, 4);
350 enc->cur += 4;
351 }
352 else 710 else
353 { 711 encode_float64 (enc, nv);
354 uint64_t fp = ecb_double_to_binary64 (nv);
355
356 *enc->cur++ = 0xe0 | 27;
357
358 if (!ecb_big_endian ())
359 fp = ecb_bswap64 (fp);
360
361 memcpy (enc->cur, &fp, 8);
362 enc->cur += 8;
363 }
364} 712}
365 713
366static void 714static void
367encode_sv (enc_t *enc, SV *sv) 715encode_sv (enc_t *enc, SV *sv)
368{ 716{
370 718
371 if (SvPOKp (sv)) 719 if (SvPOKp (sv))
372 { 720 {
373 STRLEN len; 721 STRLEN len;
374 char *str = SvPV (sv, len); 722 char *str = SvPV (sv, len);
375 encode_str (enc, SvUTF8 (sv), str, len); 723 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
376 } 724 }
377 else if (SvNOKp (sv)) 725 else if (SvNOKp (sv))
378 encode_nv (enc, sv); 726 encode_nv (enc, sv);
379 else if (SvIOKp (sv)) 727 else if (SvIOKp (sv))
380 { 728 encode_int (enc, sv);
381 if (SvIsUV (sv))
382 encode_uint (enc, 0x00, SvUVX (sv));
383 else if (SvIVX (sv) >= 0)
384 encode_uint (enc, 0x00, SvIVX (sv));
385 else
386 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
387 }
388 else if (SvROK (sv)) 729 else if (SvROK (sv))
389 encode_rv (enc, SvRV (sv)); 730 encode_rv (enc, SvRV (sv));
390 else if (!SvOK (sv)) 731 else if (!SvOK (sv))
391 encode_ch (enc, 0xe0 | 22); 732 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
392 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 733 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
393 encode_ch (enc, 0xe0 | 23); 734 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 else 735 else
395 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 736 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
396 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 737 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
397} 738}
398 739
399static SV * 740static SV *
400encode_cbor (SV *scalar, CBOR *cbor) 741encode_cbor (SV *scalar, CBOR *cbor)
401{ 742{
402 enc_t enc; 743 enc_t enc = { 0 };
403 744
404 enc.cbor = *cbor; 745 enc.cbor = *cbor;
405 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
406 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
407 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
408 enc.depth = 0;
409 749
410 SvPOK_only (enc.sv); 750 SvPOK_only (enc.sv);
751
752 if (cbor->flags & F_PACK_STRINGS)
753 {
754 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
755 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
756 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
757 }
758
411 encode_sv (&enc, scalar); 759 encode_sv (&enc, scalar);
412 760
413 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 761 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
414 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 762 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
415 763
429 U8 *end; // end of input string 777 U8 *end; // end of input string
430 const char *err; // parse error, if != 0 778 const char *err; // parse error, if != 0
431 CBOR cbor; 779 CBOR cbor;
432 U32 depth; // recursion depth 780 U32 depth; // recursion depth
433 U32 maxdepth; // recursion depth limit 781 U32 maxdepth; // recursion depth limit
782 AV *shareable;
783 AV *stringref;
784 SV *decode_tagged;
785 SV *err_sv; // optional sv for error, needs to be freed
434} dec_t; 786} dec_t;
435 787
436#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 788// set dec->err to ERRSV
789ecb_cold static void
790err_errsv (dec_t *dec)
791{
792 if (!dec->err)
793 {
794 dec->err_sv = newSVsv (ERRSV);
437 795
438#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 796 // chop off the trailing \n
797 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
798 *SvEND (dec->err_sv) = 0;
439 799
800 dec->err = SvPVutf8_nolen (dec->err_sv);
801 }
802}
803
804// the following functions are used to reduce code size and help the compiler to optimise
805ecb_cold static void
806err_set (dec_t *dec, const char *reason)
807{
808 if (!dec->err)
809 dec->err = reason;
810}
811
812ecb_cold static void
813err_unexpected_end (dec_t *dec)
814{
815 err_set (dec, "unexpected end of CBOR data");
816}
817
818#define ERR_DO(do) SB do; goto fail; SE
819#define ERR(reason) ERR_DO (err_set (dec, reason))
820#define ERR_ERRSV ERR_DO (err_errsv (dec))
821
822#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
823
440#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 824#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
441#define DEC_DEC_DEPTH --dec->depth 825#define DEC_DEC_DEPTH --dec->depth
442 826
443static UV 827static UV
444decode_uint (dec_t *dec) 828decode_uint (dec_t *dec)
445{ 829{
446 switch (*dec->cur & 31) 830 U8 m = *dec->cur & MINOR_MASK;
447 { 831 ++dec->cur;
448 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
449 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
450 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
451 return *dec->cur++ & 31;
452 832
453 case 24: 833 if (ecb_expect_true (m < LENGTH_EXT1))
834 return m;
835 else if (ecb_expect_true (m == LENGTH_EXT1))
836 {
454 WANT (2); 837 WANT (1);
455 dec->cur += 2; 838 dec->cur += 1;
456 return dec->cur[-1]; 839 return dec->cur[-1];
457 840 }
458 case 25: 841 else if (ecb_expect_true (m == LENGTH_EXT2))
842 {
459 WANT (3); 843 WANT (2);
460 dec->cur += 3; 844 dec->cur += 2;
461 return (((UV)dec->cur[-2]) << 8) 845 return (((UV)dec->cur[-2]) << 8)
462 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
463 847 }
464 case 26: 848 else if (ecb_expect_true (m == LENGTH_EXT4))
849 {
465 WANT (5); 850 WANT (4);
466 dec->cur += 5; 851 dec->cur += 4;
467 return (((UV)dec->cur[-4]) << 24) 852 return (((UV)dec->cur[-4]) << 24)
468 | (((UV)dec->cur[-3]) << 16) 853 | (((UV)dec->cur[-3]) << 16)
469 | (((UV)dec->cur[-2]) << 8) 854 | (((UV)dec->cur[-2]) << 8)
470 | ((UV)dec->cur[-1]); 855 | ((UV)dec->cur[-1]);
471 856 }
472 case 27: 857 else if (ecb_expect_true (m == LENGTH_EXT8))
858 {
473 WANT (9); 859 WANT (8);
474 dec->cur += 9; 860 dec->cur += 8;
861
862 return
863#if UVSIZE < 8
864 0
865#else
475 return (((UV)dec->cur[-8]) << 56) 866 (((UV)dec->cur[-8]) << 56)
476 | (((UV)dec->cur[-7]) << 48) 867 | (((UV)dec->cur[-7]) << 48)
477 | (((UV)dec->cur[-6]) << 40) 868 | (((UV)dec->cur[-6]) << 40)
478 | (((UV)dec->cur[-5]) << 32) 869 | (((UV)dec->cur[-5]) << 32)
870#endif
479 | (((UV)dec->cur[-4]) << 24) 871 | (((UV)dec->cur[-4]) << 24)
480 | (((UV)dec->cur[-3]) << 16) 872 | (((UV)dec->cur[-3]) << 16)
481 | (((UV)dec->cur[-2]) << 8) 873 | (((UV)dec->cur[-2]) << 8)
482 | ((UV)dec->cur[-1]); 874 | ((UV)dec->cur[-1]);
483 875 }
484 default: 876 else
485 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 877 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
486 }
487 878
488fail: 879fail:
489 return 0; 880 return 0;
490} 881}
491 882
496{ 887{
497 AV *av = newAV (); 888 AV *av = newAV ();
498 889
499 DEC_INC_DEPTH; 890 DEC_INC_DEPTH;
500 891
501 if ((*dec->cur & 31) == 31) 892 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
502 { 893 {
503 ++dec->cur; 894 ++dec->cur;
504 895
505 for (;;) 896 for (;;)
506 { 897 {
507 WANT (1); 898 WANT (1);
508 899
509 if (*dec->cur == 0xe0 | 31) 900 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
510 { 901 {
511 ++dec->cur; 902 ++dec->cur;
512 break; 903 break;
513 } 904 }
514 905
515 av_push (av, decode_sv (dec)); 906 av_push (av, decode_sv (dec));
516 } 907 }
517 } 908 }
518 else 909 else
519 { 910 {
520 int i, len = decode_uint (dec); 911 UV i, len = decode_uint (dec);
521 912
913 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
522 av_fill (av, len - 1); 914 av_fill (av, len - 1);
523 915
524 for (i = 0; i < len; ++i) 916 for (i = 0; i < len; ++i)
525 AvARRAY (av)[i] = decode_sv (dec); 917 AvARRAY (av)[i] = decode_sv (dec);
526 } 918 }
527 919
528 DEC_DEC_DEPTH; 920 DEC_DEC_DEPTH;
529 return newRV_noinc ((SV *)av); 921 return newRV_noinc ((SV *)av);
530 922
531fail: 923fail:
532 SvREFCNT_dec (av); 924 SvREFCNT_dec_NN (av);
533 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 926 return &PL_sv_undef;
535} 927}
536 928
929static void
930decode_he (dec_t *dec, HV *hv)
931{
932 // for speed reasons, we specialcase single-string
933 // byte or utf-8 strings as keys, but only when !stringref
934
935 if (ecb_expect_true (!dec->stringref))
936 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
937 {
938 STRLEN len = decode_uint (dec);
939 char *key = (char *)dec->cur;
940
941 WANT (len);
942 dec->cur += len;
943
944 hv_store (hv, key, len, decode_sv (dec), 0);
945
946 return;
947 }
948 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
949 {
950 STRLEN len = decode_uint (dec);
951 char *key = (char *)dec->cur;
952
953 WANT (len);
954 dec->cur += len;
955
956 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
957 if (!is_utf8_string (key, len))
958 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
959
960 hv_store (hv, key, -len, decode_sv (dec), 0);
961
962 return;
963 }
964
965 SV *k = decode_sv (dec);
966 SV *v = decode_sv (dec);
967
968 // we leak memory if uncaught exceptions are thrown by random magical
969 // methods, and this is hopefully the only place where it can happen,
970 // so if there is a chance of an exception, take the very slow path.
971 // since catching exceptions is "undocumented/internal/forbidden" by
972 // the new p5p powers, we need to call out to a perl function :/
973 if (ecb_expect_false (SvAMAGIC (k)))
974 {
975 dSP;
976
977 ENTER; SAVETMPS;
978 PUSHMARK (SP);
979 EXTEND (SP, 3);
980 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
981 PUSHs (sv_2mortal (k));
982 PUSHs (sv_2mortal (v));
983
984 PUTBACK;
985 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
986 SPAGAIN;
987
988 FREETMPS; LEAVE;
989
990 if (SvTRUE (ERRSV))
991 ERR_ERRSV;
992
993 return;
994 }
995
996 hv_store_ent (hv, k, v, 0);
997 SvREFCNT_dec_NN (k);
998
999fail:
1000 ;
1001}
1002
537static SV * 1003static SV *
538decode_hv (dec_t *dec) 1004decode_hv (dec_t *dec)
539{ 1005{
540 HV *hv = newHV (); 1006 HV *hv = newHV ();
541 1007
542 DEC_INC_DEPTH; 1008 DEC_INC_DEPTH;
543 1009
544 if ((*dec->cur & 31) == 31) 1010 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
545 { 1011 {
546 ++dec->cur; 1012 ++dec->cur;
547 1013
548 for (;;) 1014 for (;;)
549 { 1015 {
550 WANT (1); 1016 WANT (1);
551 1017
552 if (*dec->cur == 0xe0 | 31) 1018 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
553 { 1019 {
554 ++dec->cur; 1020 ++dec->cur;
555 break; 1021 break;
556 } 1022 }
557 1023
558 SV *k = decode_sv (dec); 1024 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 1025 }
563 } 1026 }
564 else 1027 else
565 { 1028 {
566 int len = decode_uint (dec); 1029 UV pairs = decode_uint (dec);
567 1030
1031 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1032
568 while (len--) 1033 while (pairs--)
569 { 1034 decode_he (dec, hv);
570 SV *k = decode_sv (dec);
571 SV *v = decode_sv (dec);
572
573 hv_store_ent (hv, k, v, 0);
574 }
575 } 1035 }
576 1036
577 DEC_DEC_DEPTH; 1037 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 1038 return newRV_noinc ((SV *)hv);
579 1039
580#if 0 1040fail:
581 SV *sv; 1041 SvREFCNT_dec_NN (hv);
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH; 1042 DEC_DEC_DEPTH;
585 decode_ws (dec); 1043 return &PL_sv_undef;
1044}
586 1045
587 for (;;) 1046static SV *
588 { 1047decode_str (dec_t *dec, int utf8)
589 // heuristic: assume that 1048{
590 // a) decode_str + hv_store_ent are abysmally slow. 1049 SV *sv = 0;
591 // b) most hash keys are short, simple ascii text. 1050
592 // => try to "fast-match" such strings to avoid 1051 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
593 // the overhead of decode_str + hv_store_ent.
594 { 1052 {
595 SV *value; 1053 // indefinite length strings
596 char *p = dec->cur; 1054 ++dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598 1055
1056 U8 major = *dec->cur & MAJOR_MISC;
1057
1058 sv = newSVpvn ("", 0);
1059
599 for (;;) 1060 for (;;)
600 { 1061 {
601 // the >= 0x80 is false on most architectures 1062 WANT (1);
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1063
1064 if ((*dec->cur - major) > LENGTH_EXT8)
1065 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
603 { 1066 {
604 // slow path, back up and use decode_str 1067 ++dec->cur;
605 SV *key = decode_str (dec);
606 if (!key)
607 goto fail;
608
609 decode_ws (dec); EXPECT_CH (':');
610
611 decode_ws (dec);
612 value = decode_sv (dec);
613 if (!value)
614 {
615 SvREFCNT_dec (key);
616 goto fail;
617 }
618
619 hv_store_ent (hv, key, value, 0);
620 SvREFCNT_dec (key);
621
622 break; 1068 break;
623 } 1069 }
624 else if (*p == '"')
625 {
626 // fast path, got a simple key
627 char *key = dec->cur;
628 int len = p - key;
629 dec->cur = p + 1;
630
631 decode_ws (dec); EXPECT_CH (':');
632
633 decode_ws (dec);
634 value = decode_sv (dec);
635 if (!value)
636 goto fail;
637
638 hv_store (hv, key, len, value, 0);
639
640 break;
641 }
642
643 ++p;
644 }
645 }
646
647 decode_ws (dec);
648
649 if (*dec->cur == '}')
650 {
651 ++dec->cur;
652 break;
653 }
654
655 if (*dec->cur != ',')
656 ERR (", or } expected while parsing object/hash");
657
658 ++dec->cur;
659
660 decode_ws (dec);
661
662 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
663 {
664 ++dec->cur;
665 break;
666 }
667 }
668
669 DEC_DEC_DEPTH;
670 sv = newRV_noinc ((SV *)hv);
671
672 // check filter callbacks
673 if (dec->cbor.flags & F_HOOK)
674 {
675 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
676 {
677 HE *cb, *he;
678
679 hv_iterinit (hv);
680 he = hv_iternext (hv);
681 hv_iterinit (hv);
682
683 // the next line creates a mortal sv each time its called.
684 // might want to optimise this for common cases.
685 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
686
687 if (cb)
688 { 1070 else
689 dSP; 1071 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
690 int count;
691 1072
692 ENTER; SAVETMPS; PUSHMARK (SP); 1073 STRLEN len = decode_uint (dec);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695 1074
696 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
697
698 if (count == 1)
699 {
700 sv = newSVsv (POPs);
701 FREETMPS; LEAVE;
702 return sv;
703 }
704
705 SvREFCNT_inc (sv);
706 FREETMPS; LEAVE;
707 }
708 }
709
710 if (dec->cbor.cb_object)
711 {
712 dSP;
713 int count;
714
715 ENTER; SAVETMPS; PUSHMARK (SP);
716 XPUSHs (sv_2mortal (sv));
717
718 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
719
720 if (count == 1)
721 {
722 sv = newSVsv (POPs);
723 FREETMPS; LEAVE;
724 return sv;
725 }
726
727 SvREFCNT_inc (sv);
728 FREETMPS; LEAVE;
729 }
730 }
731
732 return sv;
733#endif
734
735fail:
736 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH;
738 return &PL_sv_undef;
739}
740
741static SV *
742decode_str (dec_t *dec, int utf8)
743{
744 SV *sv;
745
746 if ((*dec->cur & 31) == 31)
747 {
748 ++dec->cur;
749
750 sv = newSVpvn ("", 0);
751
752 // not very fast, and certainly not robust against illegal input
753 for (;;)
754 {
755 WANT (1); 1075 WANT (len);
756 1076 sv_catpvn (sv, dec->cur, len);
757 if (*dec->cur == 0xe0 | 31) 1077 dec->cur += len;
758 {
759 ++dec->cur;
760 break;
761 }
762
763 SV *sv2 = decode_sv (dec);
764 sv_catsv (sv, sv2);
765 } 1078 }
766 } 1079 }
767 else 1080 else
768 { 1081 {
769 STRLEN len = decode_uint (dec); 1082 STRLEN len = decode_uint (dec);
770 1083
771 WANT (len); 1084 WANT (len);
772 sv = newSVpvn (dec->cur, len); 1085 sv = newSVpvn (dec->cur, len);
773 dec->cur += len; 1086 dec->cur += len;
1087
1088 if (ecb_expect_false (dec->stringref)
1089 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1090 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
774 } 1091 }
775 1092
776 if (utf8) 1093 if (utf8)
1094 {
1095 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1096 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1097 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1098
777 SvUTF8_on (sv); 1099 SvUTF8_on (sv);
1100 }
778 1101
779 return sv; 1102 return sv;
780 1103
781fail: 1104fail:
1105 SvREFCNT_dec (sv);
1106 return &PL_sv_undef;
1107}
1108
1109static SV *
1110decode_tagged (dec_t *dec)
1111{
1112 SV *sv = 0;
1113 UV tag = decode_uint (dec);
1114
1115 WANT (1);
1116
1117 switch (tag)
1118 {
1119 case CBOR_TAG_MAGIC:
1120 sv = decode_sv (dec);
1121 break;
1122
1123 case CBOR_TAG_INDIRECTION:
1124 sv = newRV_noinc (decode_sv (dec));
1125 break;
1126
1127 case CBOR_TAG_STRINGREF_NAMESPACE:
1128 {
1129 // do not use SAVETMPS/FREETMPS, as these will
1130 // erase mortalised caches, e.g. "shareable"
1131 ENTER;
1132
1133 SAVESPTR (dec->stringref);
1134 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1135
1136 sv = decode_sv (dec);
1137
1138 LEAVE;
1139 }
1140 break;
1141
1142 case CBOR_TAG_STRINGREF:
1143 {
1144 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1145 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1146
1147 UV idx = decode_uint (dec);
1148
1149 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1150 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1151
1152 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1153 }
1154 break;
1155
1156 case CBOR_TAG_VALUE_SHAREABLE:
1157 {
1158 if (ecb_expect_false (!dec->shareable))
1159 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1160
1161 if (dec->cbor.flags & F_ALLOW_CYCLES)
1162 {
1163 sv = newSV (0);
1164 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1165
1166 SV *osv = decode_sv (dec);
1167 sv_setsv (sv, osv);
1168 SvREFCNT_dec_NN (osv);
1169 }
1170 else
1171 {
1172 av_push (dec->shareable, &PL_sv_undef);
1173 int idx = AvFILLp (dec->shareable);
1174 sv = decode_sv (dec);
1175 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1176 }
1177 }
1178 break;
1179
1180 case CBOR_TAG_VALUE_SHAREDREF:
1181 {
1182 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1183 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1184
1185 UV idx = decode_uint (dec);
1186
1187 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1188 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1189
1190 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1191
1192 if (sv == &PL_sv_undef)
1193 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1194 }
1195 break;
1196
1197 case CBOR_TAG_PERL_OBJECT:
1198 {
1199 if (dec->cbor.flags & F_FORBID_OBJECTS)
1200 goto filter;
1201
1202 sv = decode_sv (dec);
1203
1204 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1205 ERR ("corrupted CBOR data (non-array perl object)");
1206
1207 AV *av = (AV *)SvRV (sv);
1208 int len = av_len (av) + 1;
1209 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1210
1211 if (!stash)
1212 ERR ("cannot decode perl-object (package does not exist)");
1213
1214 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1215
1216 if (!method)
1217 ERR ("cannot decode perl-object (package does not have a THAW method)");
1218
1219 dSP;
1220
1221 ENTER; SAVETMPS;
1222 PUSHMARK (SP);
1223 EXTEND (SP, len + 1);
1224 // we re-bless the reference to get overload and other niceties right
1225 PUSHs (*av_fetch (av, 0, 1));
1226 PUSHs (sv_cbor);
1227
1228 int i;
1229
1230 for (i = 1; i < len; ++i)
1231 PUSHs (*av_fetch (av, i, 1));
1232
1233 PUTBACK;
1234 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1235 SPAGAIN;
1236
1237 if (SvTRUE (ERRSV))
1238 {
1239 FREETMPS; LEAVE;
1240 ERR_ERRSV;
1241 }
1242
1243 SvREFCNT_dec_NN (sv);
1244 sv = SvREFCNT_inc (POPs);
1245
1246 PUTBACK;
1247
1248 FREETMPS; LEAVE;
1249 }
1250 break;
1251
1252 default:
1253 filter:
1254 {
1255 SV *tag_sv = newSVuv (tag);
1256
1257 sv = decode_sv (dec);
1258
1259 dSP;
1260 ENTER; SAVETMPS;
1261 PUSHMARK (SP);
1262 EXTEND (SP, 2);
1263 PUSHs (tag_sv);
1264 PUSHs (sv);
1265
1266 PUTBACK;
1267 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1268 SPAGAIN;
1269
1270 if (SvTRUE (ERRSV))
1271 {
1272 SvREFCNT_dec_NN (tag_sv);
1273 FREETMPS; LEAVE;
1274 ERR_ERRSV;
1275 }
1276
1277 if (count)
1278 {
1279 SvREFCNT_dec_NN (tag_sv);
1280 SvREFCNT_dec_NN (sv);
1281 sv = SvREFCNT_inc_NN (TOPs);
1282 SP -= count;
1283 }
1284 else
1285 {
1286 AV *av = newAV ();
1287 av_push (av, tag_sv);
1288 av_push (av, sv);
1289
1290 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1291 ? cbor_tagged_stash
1292 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1293 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1294 }
1295
1296 PUTBACK;
1297
1298 FREETMPS; LEAVE;
1299 }
1300 break;
1301 }
1302
1303 return sv;
1304
1305fail:
1306 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 1307 return &PL_sv_undef;
783} 1308}
784 1309
785static SV * 1310static SV *
786decode_sv (dec_t *dec) 1311decode_sv (dec_t *dec)
787{ 1312{
788 WANT (1); 1313 WANT (1);
789 1314
790 switch (*dec->cur >> 5) 1315 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1316 {
792 case 0: // unsigned int 1317 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1318 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1319 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1320 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1321 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1322 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1323 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1324
800 return decode_str (dec, 1); 1325 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 abort ();
807 break;
808 case 7: // misc
809 switch (*dec->cur++ & 31) 1326 switch (*dec->cur++ & MINOR_MASK)
810 { 1327 {
811 case 20: 1328 case SIMPLE_FALSE:
812#if CBOR_SLOW 1329#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 1330 types_false = get_bool ("Types::Serialiser::false");
814#endif 1331#endif
815 return newSVsv (cbor_false); 1332 return newSVsv (types_false);
816 case 21: 1333 case SIMPLE_TRUE:
817#if CBOR_SLOW 1334#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 1335 types_true = get_bool ("Types::Serialiser::true");
819#endif 1336#endif
820 return newSVsv (cbor_true); 1337 return newSVsv (types_true);
821 case 22: 1338 case SIMPLE_NULL:
822 return newSVsv (&PL_sv_undef); 1339 return newSVsv (&PL_sv_undef);
1340 case SIMPLE_UNDEF:
1341#if CBOR_SLOW
1342 types_error = get_bool ("Types::Serialiser::error");
1343#endif
1344 return newSVsv (types_error);
823 1345
824 case 25: 1346 case MISC_FLOAT16:
825 // half float
826 abort ();
827 break; 1347 {
1348 WANT (2);
828 1349
1350 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1351 dec->cur += 2;
1352
1353 return newSVnv (ecb_binary16_to_float (fp));
1354 }
1355
829 case 26: 1356 case MISC_FLOAT32:
830 { 1357 {
831 uint32_t fp; 1358 uint32_t fp;
832 WANT (4); 1359 WANT (4);
833 memcpy (&fp, dec->cur, 4); 1360 memcpy (&fp, dec->cur, 4);
834 dec->cur += 4; 1361 dec->cur += 4;
837 fp = ecb_bswap32 (fp); 1364 fp = ecb_bswap32 (fp);
838 1365
839 return newSVnv (ecb_binary32_to_float (fp)); 1366 return newSVnv (ecb_binary32_to_float (fp));
840 } 1367 }
841 1368
842 case 27: 1369 case MISC_FLOAT64:
843 { 1370 {
844 uint64_t fp; 1371 uint64_t fp;
845 WANT (8); 1372 WANT (8);
846 memcpy (&fp, dec->cur, 8); 1373 memcpy (&fp, dec->cur, 8);
847 dec->cur += 8; 1374 dec->cur += 8;
850 fp = ecb_bswap64 (fp); 1377 fp = ecb_bswap64 (fp);
851 1378
852 return newSVnv (ecb_binary64_to_double (fp)); 1379 return newSVnv (ecb_binary64_to_double (fp));
853 } 1380 }
854 1381
855 // 0..19 unassigned 1382 // 0..19 unassigned simple
856 // 24 reserved + unassigned (reserved values are not encodable) 1383 // 24 reserved + unassigned simple (reserved values are not encodable)
1384 // 28-30 unassigned misc
1385 // 31 break code
857 default: 1386 default:
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1387 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
859 } 1388 }
860 1389
861 break; 1390 break;
862 } 1391 }
863#if 0
864 switch (*dec->cur)
865 {
866 //case '"': ++dec->cur; return decode_str (dec);
867 case '[': ++dec->cur; return decode_av (dec);
868 case '{': ++dec->cur; return decode_hv (dec);
869
870 case '-':
871 case '0': case '1': case '2': case '3': case '4':
872 case '5': case '6': case '7': case '8': case '9':
873 //TODO return decode_num (dec);
874
875 case 't':
876 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
877 {
878 dec->cur += 4;
879#if CBOR_SLOW
880 cbor_true = get_bool ("CBOR::XS::true");
881#endif
882 return newSVsv (cbor_true);
883 }
884 else
885 ERR ("'true' expected");
886
887 break;
888
889 case 'f':
890 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
891 {
892 dec->cur += 5;
893#if CBOR_SLOW
894 cbor_false = get_bool ("CBOR::XS::false");
895#endif
896 return newSVsv (cbor_false);
897 }
898 else
899 ERR ("'false' expected");
900
901 break;
902
903 case 'n':
904 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
905 {
906 dec->cur += 4;
907 return newSVsv (&PL_sv_undef);
908 }
909 else
910 ERR ("'null' expected");
911
912 break;
913
914 default:
915 ERR ("malformed CBOR string, neither array, object, number, string or atom");
916 break;
917 }
918#endif
919 1392
920fail: 1393fail:
921 return &PL_sv_undef; 1394 return &PL_sv_undef;
922} 1395}
923 1396
924static SV * 1397static SV *
925decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1398decode_cbor (SV *string, CBOR *cbor, char **offset_return)
926{ 1399{
927 dec_t dec; 1400 dec_t dec = { 0 };
928 SV *sv; 1401 SV *sv;
1402 STRLEN len;
1403 char *data = SvPVbyte (string, len);
929 1404
930 /* work around bugs in 5.10 where manipulating magic values
931 * makes perl ignore the magic in subsequent accesses.
932 * also make a copy of non-PV values, to get them into a clean
933 * state (SvPV should do that, but it's buggy, see below).
934 */
935 /*SvGETMAGIC (string);*/
936 if (SvMAGICAL (string) || !SvPOK (string))
937 string = sv_2mortal (newSVsv (string));
938
939 SvUPGRADE (string, SVt_PV);
940
941 /* work around a bug in perl 5.10, which causes SvCUR to fail an
942 * assertion with -DDEBUGGING, although SvCUR is documented to
943 * return the xpv_cur field which certainly exists after upgrading.
944 * according to nicholas clark, calling SvPOK fixes this.
945 * But it doesn't fix it, so try another workaround, call SvPV_nolen
946 * and hope for the best.
947 * Damnit, SvPV_nolen still trips over yet another assertion. This
948 * assertion business is seriously broken, try yet another workaround
949 * for the broken -DDEBUGGING.
950 */
951 {
952#ifdef DEBUGGING
953 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
954#else
955 STRLEN offset = SvCUR (string);
956#endif
957
958 if (offset > cbor->max_size && cbor->max_size) 1405 if (len > cbor->max_size && cbor->max_size)
959 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1406 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
960 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1407 (unsigned long)len, (unsigned long)cbor->max_size);
961 }
962
963 sv_utf8_downgrade (string, 0);
964 1408
965 dec.cbor = *cbor; 1409 dec.cbor = *cbor;
966 dec.cur = (U8 *)SvPVX (string); 1410 dec.cur = (U8 *)data;
967 dec.end = (U8 *)SvEND (string); 1411 dec.end = (U8 *)data + len;
968 dec.err = 0;
969 dec.depth = 0;
970
971 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
972 ;//TODO dec.cbor.flags |= F_HOOK;
973 1412
974 sv = decode_sv (&dec); 1413 sv = decode_sv (&dec);
975 1414
976 if (offset_return) 1415 if (offset_return)
977 *offset_return = dec.cur; 1416 *offset_return = dec.cur;
978 1417
979 if (!(offset_return || !sv)) 1418 if (!(offset_return || !sv))
980 {
981 if (*dec.cur && !dec.err) 1419 if (dec.cur != dec.end && !dec.err)
982 {
983 dec.err = "garbage after CBOR object"; 1420 dec.err = "garbage after CBOR object";
984 SvREFCNT_dec (sv); 1421
985 sv = 0; 1422 if (dec.err)
1423 {
1424 if (dec.shareable)
986 } 1425 {
987 } 1426 // need to break cyclic links, which would all be in shareable
1427 int i;
1428 SV **svp;
988 1429
989 if (!sv) 1430 for (i = av_len (dec.shareable) + 1; i--; )
1431 if ((svp = av_fetch (dec.shareable, i, 0)))
1432 sv_setsv (*svp, &PL_sv_undef);
1433 }
1434
1435 SvREFCNT_dec_NN (sv);
1436
1437 if (dec.err_sv)
1438 sv_2mortal (dec.err_sv);
1439
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1440 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1441 }
991 1442
992 sv = sv_2mortal (sv); 1443 sv = sv_2mortal (sv);
993 1444
994 return sv; 1445 return sv;
995} 1446}
996 1447
1448/////////////////////////////////////////////////////////////////////////////
1449// incremental parser
1450
1451#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1452
1453// returns 0 for notyet, 1 for success or error
1454static int
1455incr_parse (CBOR *self, SV *cborstr)
1456{
1457 STRLEN cur;
1458 SvPV (cborstr, cur);
1459
1460 while (ecb_expect_true (self->incr_need <= cur))
1461 {
1462 // table of integer count bytes
1463 static I8 incr_len[MINOR_MASK + 1] = {
1464 0, 0, 0, 0, 0, 0, 0, 0,
1465 0, 0, 0, 0, 0, 0, 0, 0,
1466 0, 0, 0, 0, 0, 0, 0, 0,
1467 1, 2, 4, 8,-1,-1,-1,-2
1468 };
1469
1470 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1471 U8 m = *p & MINOR_MASK;
1472 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1473 I8 ilen = incr_len[m];
1474
1475 self->incr_need = self->incr_pos + 1;
1476
1477 if (ecb_expect_false (ilen < 0))
1478 {
1479 if (m != MINOR_INDEF)
1480 return 1; // error
1481
1482 if (*p == (MAJOR_MISC | MINOR_INDEF))
1483 {
1484 if (count >= 0)
1485 return 1; // error
1486
1487 count = 1;
1488 }
1489 else
1490 {
1491 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1492 count = -1;
1493 }
1494 }
1495 else
1496 {
1497 self->incr_need += ilen;
1498 if (ecb_expect_false (self->incr_need > cur))
1499 return 0;
1500
1501 int major = *p >> MAJOR_SHIFT;
1502
1503 switch (major)
1504 {
1505 case MAJOR_TAG >> MAJOR_SHIFT:
1506 ++count; // tags merely prefix another value
1507 break;
1508
1509 case MAJOR_BYTES >> MAJOR_SHIFT:
1510 case MAJOR_TEXT >> MAJOR_SHIFT:
1511 case MAJOR_ARRAY >> MAJOR_SHIFT:
1512 case MAJOR_MAP >> MAJOR_SHIFT:
1513 {
1514 UV len;
1515
1516 if (ecb_expect_false (ilen))
1517 {
1518 len = 0;
1519
1520 do {
1521 len = (len << 8) | *++p;
1522 } while (--ilen);
1523 }
1524 else
1525 len = m;
1526
1527 switch (major)
1528 {
1529 case MAJOR_BYTES >> MAJOR_SHIFT:
1530 case MAJOR_TEXT >> MAJOR_SHIFT:
1531 self->incr_need += len;
1532 if (ecb_expect_false (self->incr_need > cur))
1533 return 0;
1534
1535 break;
1536
1537 case MAJOR_MAP >> MAJOR_SHIFT:
1538 len <<= 1;
1539 case MAJOR_ARRAY >> MAJOR_SHIFT:
1540 if (len)
1541 {
1542 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1543 count = len + 1;
1544 }
1545 break;
1546 }
1547 }
1548 }
1549 }
1550
1551 self->incr_pos = self->incr_need;
1552
1553 if (count > 0)
1554 {
1555 while (!--count)
1556 {
1557 if (!AvFILLp (self->incr_count))
1558 return 1; // done
1559
1560 SvREFCNT_dec_NN (av_pop (self->incr_count));
1561 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1562 }
1563
1564 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1565 }
1566 }
1567
1568 return 0;
1569}
1570
1571
997///////////////////////////////////////////////////////////////////////////// 1572/////////////////////////////////////////////////////////////////////////////
998// XS interface functions 1573// XS interface functions
999 1574
1000MODULE = CBOR::XS PACKAGE = CBOR::XS 1575MODULE = CBOR::XS PACKAGE = CBOR::XS
1001 1576
1002BOOT: 1577BOOT:
1003{ 1578{
1004 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1579 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1005 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1580 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1006 1581
1007 cbor_true = get_bool ("CBOR::XS::true"); 1582 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1008 cbor_false = get_bool ("CBOR::XS::false"); 1583 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1584
1585 types_true = get_bool ("Types::Serialiser::true" );
1586 types_false = get_bool ("Types::Serialiser::false");
1587 types_error = get_bool ("Types::Serialiser::error");
1588
1589 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1590
1591 sv_cbor = newSVpv ("CBOR", 0);
1592 SvREADONLY_on (sv_cbor);
1593
1594 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1009} 1595}
1010 1596
1011PROTOTYPES: DISABLE 1597PROTOTYPES: DISABLE
1012 1598
1013void CLONE (...) 1599void CLONE (...)
1014 CODE: 1600 CODE:
1015 cbor_stash = 0; 1601 cbor_stash = 0;
1602 cbor_tagged_stash = 0;
1603 types_error_stash = 0;
1016 cbor_boolean_stash = 0; 1604 types_boolean_stash = 0;
1017 1605
1018void new (char *klass) 1606void new (char *klass)
1019 PPCODE: 1607 PPCODE:
1020{ 1608{
1021 SV *pv = NEWSV (0, sizeof (CBOR)); 1609 SV *pv = NEWSV (0, sizeof (CBOR));
1029 1617
1030void shrink (CBOR *self, int enable = 1) 1618void shrink (CBOR *self, int enable = 1)
1031 ALIAS: 1619 ALIAS:
1032 shrink = F_SHRINK 1620 shrink = F_SHRINK
1033 allow_unknown = F_ALLOW_UNKNOWN 1621 allow_unknown = F_ALLOW_UNKNOWN
1622 allow_sharing = F_ALLOW_SHARING
1623 allow_cycles = F_ALLOW_CYCLES
1624 forbid_objects = F_FORBID_OBJECTS
1625 pack_strings = F_PACK_STRINGS
1626 text_keys = F_TEXT_KEYS
1627 text_strings = F_TEXT_STRINGS
1628 validate_utf8 = F_VALIDATE_UTF8
1034 PPCODE: 1629 PPCODE:
1035{ 1630{
1036 if (enable) 1631 if (enable)
1037 self->flags |= ix; 1632 self->flags |= ix;
1038 else 1633 else
1043 1638
1044void get_shrink (CBOR *self) 1639void get_shrink (CBOR *self)
1045 ALIAS: 1640 ALIAS:
1046 get_shrink = F_SHRINK 1641 get_shrink = F_SHRINK
1047 get_allow_unknown = F_ALLOW_UNKNOWN 1642 get_allow_unknown = F_ALLOW_UNKNOWN
1643 get_allow_sharing = F_ALLOW_SHARING
1644 get_allow_cycles = F_ALLOW_CYCLES
1645 get_forbid_objects = F_FORBID_OBJECTS
1646 get_pack_strings = F_PACK_STRINGS
1647 get_text_keys = F_TEXT_KEYS
1648 get_text_strings = F_TEXT_STRINGS
1649 get_validate_utf8 = F_VALIDATE_UTF8
1048 PPCODE: 1650 PPCODE:
1049 XPUSHs (boolSV (self->flags & ix)); 1651 XPUSHs (boolSV (self->flags & ix));
1050 1652
1051void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1653void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1052 PPCODE: 1654 PPCODE:
1068 CODE: 1670 CODE:
1069 RETVAL = self->max_size; 1671 RETVAL = self->max_size;
1070 OUTPUT: 1672 OUTPUT:
1071 RETVAL 1673 RETVAL
1072 1674
1073#if 0 //TODO 1675void filter (CBOR *self, SV *filter = 0)
1074
1075void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1076 PPCODE: 1676 PPCODE:
1077{
1078 SvREFCNT_dec (self->cb_object); 1677 SvREFCNT_dec (self->filter);
1079 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1678 self->filter = filter ? newSVsv (filter) : filter;
1080
1081 XPUSHs (ST (0)); 1679 XPUSHs (ST (0));
1082}
1083 1680
1084void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1681SV *get_filter (CBOR *self)
1085 PPCODE: 1682 CODE:
1086{ 1683 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1087 if (!self->cb_sk_object) 1684 OUTPUT:
1088 self->cb_sk_object = newHV (); 1685 RETVAL
1089
1090 if (SvOK (cb))
1091 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1092 else
1093 {
1094 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1095
1096 if (!HvKEYS (self->cb_sk_object))
1097 {
1098 SvREFCNT_dec (self->cb_sk_object);
1099 self->cb_sk_object = 0;
1100 }
1101 }
1102
1103 XPUSHs (ST (0));
1104}
1105
1106#endif
1107 1686
1108void encode (CBOR *self, SV *scalar) 1687void encode (CBOR *self, SV *scalar)
1109 PPCODE: 1688 PPCODE:
1110 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1689 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1111 XPUSHs (scalar); 1690 XPUSHs (scalar);
1124 EXTEND (SP, 2); 1703 EXTEND (SP, 2);
1125 PUSHs (sv); 1704 PUSHs (sv);
1126 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1705 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1127} 1706}
1128 1707
1708void incr_parse (CBOR *self, SV *cborstr)
1709 ALIAS:
1710 incr_parse_multiple = 1
1711 PPCODE:
1712{
1713 if (SvUTF8 (cborstr))
1714 sv_utf8_downgrade (cborstr, 0);
1715
1716 if (!self->incr_count)
1717 {
1718 self->incr_count = newAV ();
1719 self->incr_pos = 0;
1720 self->incr_need = 1;
1721
1722 av_push (self->incr_count, newSViv (1));
1723 }
1724
1725 do
1726 {
1727 if (!incr_parse (self, cborstr))
1728 {
1729 if (self->incr_need > self->max_size && self->max_size)
1730 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1731 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1732
1733 break;
1734 }
1735
1736 SV *sv;
1737 char *offset;
1738
1739 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1740 XPUSHs (sv);
1741
1742 sv_chop (cborstr, offset);
1743
1744 av_clear (self->incr_count);
1745 av_push (self->incr_count, newSViv (1));
1746
1747 self->incr_pos = 0;
1748 self->incr_need = self->incr_pos + 1;
1749 }
1750 while (ix);
1751}
1752
1753void incr_reset (CBOR *self)
1754 CODE:
1755{
1756 SvREFCNT_dec (self->incr_count);
1757 self->incr_count = 0;
1758}
1759
1129void DESTROY (CBOR *self) 1760void DESTROY (CBOR *self)
1130 CODE: 1761 PPCODE:
1131 SvREFCNT_dec (self->cb_sk_object); 1762 cbor_free (self);
1132 SvREFCNT_dec (self->cb_object);
1133 1763
1134PROTOTYPES: ENABLE 1764PROTOTYPES: ENABLE
1135 1765
1136void encode_cbor (SV *scalar) 1766void encode_cbor (SV *scalar)
1767 ALIAS:
1768 encode_cbor = 0
1769 encode_cbor_sharing = F_ALLOW_SHARING
1137 PPCODE: 1770 PPCODE:
1138{ 1771{
1139 CBOR cbor; 1772 CBOR cbor;
1140 cbor_init (&cbor); 1773 cbor_init (&cbor);
1774 cbor.flags |= ix;
1141 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1775 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1142 XPUSHs (scalar); 1776 XPUSHs (scalar);
1143} 1777}
1144 1778
1145void decode_cbor (SV *cborstr) 1779void decode_cbor (SV *cborstr)
1149 cbor_init (&cbor); 1783 cbor_init (&cbor);
1150 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1784 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1151 XPUSHs (cborstr); 1785 XPUSHs (cborstr);
1152} 1786}
1153 1787
1788#ifdef __AFL_COMPILER
1789
1790void
1791afl_init ()
1792 CODE:
1793 __AFL_INIT ();
1794
1795int
1796afl_loop (unsigned int count = 10000)
1797 CODE:
1798 RETVAL = __AFL_LOOP (count);
1799 OUTPUT:
1800 RETVAL
1801
1802#endif
1803

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines