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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.66 by root, Sun Nov 29 21:32:01 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"
15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
13 69
14// known tags 70// known tags
15enum cbor_tag 71enum cbor_tag
16{ 72{
17 // inofficial extensions (pending iana registration) 73 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 81
21 // rfc7049 82 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
28 89
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 94
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 100
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 102};
42 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_BYTES = 1,
109 AS_TEXT = 2,
110 AS_FLOAT16 = 3,
111 AS_FLOAT32 = 4,
112 AS_FLOAT64 = 5,
113 // possibly future enhancements: float, integer
114};
115
43#define F_SHRINK 0x00000200UL 116#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 117#define F_ALLOW_UNKNOWN 0x00000002UL
118#define F_ALLOW_SHARING 0x00000004UL
119#define F_ALLOW_CYCLES 0x00000008UL
120#define F_FORBID_OBJECTS 0x00000010UL
121#define F_PACK_STRINGS 0x00000020UL
122#define F_TEXT_KEYS 0x00000040UL
123#define F_TEXT_STRINGS 0x00000080UL
124#define F_VALIDATE_UTF8 0x00000100UL
45 125
46#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
47 127
48#define SB do { 128#define SB do {
49#define SE } while (0) 129#define SE } while (0)
61# define CBOR_SLOW 0 141# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 142# define CBOR_STASH cbor_stash
63#endif 143#endif
64 144
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 145static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 146static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 147
68typedef struct { 148typedef struct {
69 U32 flags; 149 U32 flags;
70 U32 max_depth; 150 U32 max_depth;
71 STRLEN max_size; 151 STRLEN max_size;
152 SV *filter;
153
154 // for the incremental parser
155 STRLEN incr_pos; // the current offset into the text
156 STRLEN incr_need; // minimum bytes needed to decode
157 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 158} CBOR;
73 159
74ecb_inline void 160ecb_inline void
75cbor_init (CBOR *cbor) 161cbor_init (CBOR *cbor)
76{ 162{
77 Zero (cbor, 1, CBOR); 163 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 164 cbor->max_depth = 512;
165}
166
167ecb_inline void
168cbor_free (CBOR *cbor)
169{
170 SvREFCNT_dec (cbor->filter);
171 SvREFCNT_dec (cbor->incr_count);
79} 172}
80 173
81///////////////////////////////////////////////////////////////////////////// 174/////////////////////////////////////////////////////////////////////////////
82// utility functions 175// utility functions
83 176
105 SvPV_renew (sv, SvCUR (sv) + 1); 198 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 199#endif
107 } 200 }
108} 201}
109 202
110///////////////////////////////////////////////////////////////////////////// 203// minimum length of a string to be registered for stringref
111// fp hell 204ecb_inline int
112 205minimum_string_length (UV idx)
113//TODO 206{
207 return idx <= 23 ? 3
208 : idx <= 0xffU ? 4
209 : idx <= 0xffffU ? 5
210 : idx <= 0xffffffffU ? 7
211 : 11;
212}
114 213
115///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
116// encoder 215// encoder
117 216
118// structure used for encoding CBOR 217// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 220 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 221 char *end; // SvEND (sv)
123 SV *sv; // result scalar 222 SV *sv; // result scalar
124 CBOR cbor; 223 CBOR cbor;
125 U32 depth; // recursion level 224 U32 depth; // recursion level
225 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
226 UV stringref_idx;
227 HV *shareable; // ptr => index, or 0
228 UV shareable_idx;
126} enc_t; 229} enc_t;
127 230
128ecb_inline void 231ecb_inline void
129need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
130{ 233{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 235 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142{ 245{
143 need (enc, 1); 246 need (enc, 1);
144 *enc->cur++ = ch; 247 *enc->cur++ = ch;
145} 248}
146 249
250// used for tags, intregers, element counts and so on
147static void 251static void
148encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
149{ 253{
150 need (enc, 9); 254 need (enc, 9);
151 255
152 if (len < 24) 256 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 257 *enc->cur++ = major | len;
154 else if (len <= 0xff) 258 else if (ecb_expect_true (len <= 0xffU))
155 { 259 {
156 *enc->cur++ = major | 24; 260 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 261 *enc->cur++ = len;
158 } 262 }
159 else if (len <= 0xffff) 263 else if (len <= 0xffffU)
160 { 264 {
161 *enc->cur++ = major | 25; 265 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 266 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 267 *enc->cur++ = len;
164 } 268 }
165 else if (len <= 0xffffffff) 269 else if (len <= 0xffffffffU)
166 { 270 {
167 *enc->cur++ = major | 26; 271 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 272 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 273 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 274 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 275 *enc->cur++ = len;
172 } 276 }
173 else 277 else
174 { 278 {
175 *enc->cur++ = major | 27; 279 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 280 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 281 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 282 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 283 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 284 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 287 *enc->cur++ = len;
184 } 288 }
185} 289}
186 290
187static void 291// encodes a perl value into a CBOR integer
292ecb_inline
293encode_int (enc_t *enc, SV *sv)
294{
295 if (SvIsUV (sv))
296 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
297 else if (SvIVX (sv) >= 0)
298 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
299 else
300 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
301}
302
303ecb_inline void
304encode_tag (enc_t *enc, UV tag)
305{
306 encode_uint (enc, MAJOR_TAG, tag);
307}
308
309// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
310ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 311encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 312{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 313 STRLEN ulen = len;
314 U8 *p, *pend = (U8 *)str + len;
315
316 for (p = (U8 *)str; p < pend; ++p)
317 ulen += *p >> 7; // count set high bits
318
319 encode_uint (enc, MAJOR_TEXT, ulen);
320
321 need (enc, ulen);
322 for (p = (U8 *)str; p < pend; ++p)
323 if (*p < 0x80)
324 *enc->cur++ = *p;
325 else
326 {
327 *enc->cur++ = 0xc0 + (*p >> 6);
328 *enc->cur++ = 0x80 + (*p & 63);
329 }
330}
331
332ecb_inline void
333encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
334{
335 if (ecb_expect_false (upgrade_utf8))
336 if (!utf8)
337 {
338 encode_str_utf8 (enc, utf8, str, len);
339 return;
340 }
341
342 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 343 need (enc, len);
192 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
193 enc->cur += len; 345 enc->cur += len;
194} 346}
195 347
348ecb_inline void
349encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
350{
351 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
352 {
353 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
354
355 if (SvOK (*svp))
356 {
357 // already registered, use stringref
358 encode_tag (enc, CBOR_TAG_STRINGREF);
359 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
360 return;
361 }
362 else if (len >= minimum_string_length (enc->stringref_idx))
363 {
364 // register only
365 sv_setuv (*svp, enc->stringref_idx);
366 ++enc->stringref_idx;
367 }
368 }
369
370 encode_str (enc, upgrade_utf8, utf8, str, len);
371}
372
373ecb_inline void
374encode_float16 (enc_t *enc, NV nv)
375{
376 need (enc, 1+2);
377
378 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
379
380 uint16_t fp = ecb_float_to_binary16 (nv);
381
382 if (!ecb_big_endian ())
383 fp = ecb_bswap16 (fp);
384
385 memcpy (enc->cur, &fp, 2);
386 enc->cur += 2;
387}
388
389ecb_inline void
390encode_float32 (enc_t *enc, NV nv)
391{
392 need (enc, 1+4);
393
394 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
395
396 uint32_t fp = ecb_float_to_binary32 (nv);
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403}
404
405ecb_inline void
406encode_float64 (enc_t *enc, NV nv)
407{
408 need (enc, 1+8);
409
410 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
411
412 uint64_t fp = ecb_double_to_binary64 (nv);
413
414 if (!ecb_big_endian ())
415 fp = ecb_bswap64 (fp);
416
417 memcpy (enc->cur, &fp, 8);
418 enc->cur += 8;
419}
420
421ecb_inline void
422encode_forced (enc_t *enc, UV type, SV *sv)
423{
424 switch (type)
425 {
426 case AS_CBOR:
427 {
428 STRLEN len;
429 char *str = SvPVbyte (sv, len);
430
431 need (enc, len);
432 memcpy (enc->cur, str, len);
433 enc->cur += len;
434 }
435 break;
436
437 case AS_BYTES:
438 {
439 STRLEN len;
440 char *str = SvPVbyte (sv, len);
441 encode_strref (enc, 0, 0, str, len);
442 }
443 break;
444
445 case AS_TEXT:
446 {
447 STRLEN len;
448 char *str = SvPVutf8 (sv, len);
449 encode_strref (enc, 1, 1, str, len);
450 }
451 break;
452
453 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
454 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
455 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
456
457 default:
458 croak ("encountered malformed CBOR::XS::Tagged object");
459 }
460}
461
196static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
197 463
198static void 464static void
199encode_av (enc_t *enc, AV *av) 465encode_av (enc_t *enc, AV *av)
200{ 466{
203 if (enc->depth >= enc->cbor.max_depth) 469 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 470 croak (ERR_NESTING_EXCEEDED);
205 471
206 ++enc->depth; 472 ++enc->depth;
207 473
208 encode_uint (enc, 0x80, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 475
476 if (ecb_expect_false (SvMAGICAL (av)))
210 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
211 { 478 {
212 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 481 }
482 else
483 for (i = 0; i <= len; ++i)
484 {
485 SV *sv = AvARRAY (av)[i];
486 encode_sv (enc, sv ? sv : &PL_sv_undef);
487 }
215 488
216 --enc->depth; 489 --enc->depth;
217} 490}
218 491
219static void 492static void
227 ++enc->depth; 500 ++enc->depth;
228 501
229 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
231 504
232 if (mg) 505 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xa0 | 31); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 507 else
235 encode_uint (enc, 0xa0, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
236 509
237 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
238 { 511 {
239 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
241 else 514 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 515 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 516
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 517 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 518 }
246 519
247 if (mg) 520 if (ecb_expect_false (mg))
248 encode_ch (enc, 0xe0 | 31); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 522
250 --enc->depth; 523 --enc->depth;
251} 524}
252 525
253// encode objects, arrays and special \0=false and \1=true values. 526// encode objects, arrays and special \0=false and \1=true values.
254static void 527static void
255encode_rv (enc_t *enc, SV *sv) 528encode_rv (enc_t *enc, SV *sv)
256{ 529{
257 svtype svt;
258
259 SvGETMAGIC (sv); 530 SvGETMAGIC (sv);
531
260 svt = SvTYPE (sv); 532 svtype svt = SvTYPE (sv);
261 533
262 if (ecb_expect_false (SvOBJECT (sv))) 534 if (ecb_expect_false (SvOBJECT (sv)))
263 { 535 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 536 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 537 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 542 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 543 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 544 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 545
274 HV *stash = SvSTASH (sv); 546 HV *stash = SvSTASH (sv);
275 GV *method;
276 547
277 if (stash == boolean_stash) 548 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 549 {
550 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
551 return;
552 }
279 else if (stash == error_stash) 553 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 554 {
555 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
556 return;
557 }
281 else if (stash == tagged_stash) 558 else if (stash == tagged_stash)
282 { 559 {
283 if (svt != SVt_PVAV) 560 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 567 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 568 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
569 break;
570
571 case 3-1:
572 // a forced type [value, type, undef]
573 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
574 break;
575
576 default:
577 croak ("encountered malformed CBOR::XS::Tagged object");
578 }
579
580 return;
581 }
582 }
583
584 if (ecb_expect_false (SvREFCNT (sv) > 1)
585 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
586 {
587 if (ecb_expect_false (!enc->shareable))
588 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
589
590 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
591
592 if (SvOK (*svp))
288 } 593 {
594 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
595 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
596 return;
597 }
598 else
599 {
600 sv_setuv (*svp, enc->shareable_idx);
601 ++enc->shareable_idx;
602 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
603 }
604 }
605
606 if (ecb_expect_false (SvOBJECT (sv)))
607 {
608 HV *stash = SvSTASH (sv);
609 GV *method;
610
611 if (enc->cbor.flags & F_FORBID_OBJECTS)
612 croak ("encountered object '%s', but forbid_objects is enabled",
613 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 615 {
291 dSP; 616 dSP;
292 617
293 ENTER; SAVETMPS; PUSHMARK (SP); 618 ENTER; SAVETMPS;
619 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 620 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 621 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 622
297 PUTBACK; 623 PUTBACK;
298 // G_SCALAR ensures that return value is 1 624 // G_SCALAR ensures that return value is 1
311 } 637 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 639 {
314 dSP; 640 dSP;
315 641
316 ENTER; SAVETMPS; PUSHMARK (SP); 642 ENTER; SAVETMPS;
643 PUSHMARK (SP);
317 EXTEND (SP, 2); 644 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 645 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 647 PUSHs (sv_cbor);
321 648
325 652
326 // catch this surprisingly common error 653 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 655 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 656
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 658 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 660
334 while (count) 661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
335 encode_sv (enc, SP[1 - count--]); 665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
668 }
336 669
337 PUTBACK; 670 PUTBACK;
338 671
339 FREETMPS; LEAVE; 672 FREETMPS; LEAVE;
340 } 673 }
344 } 677 }
345 else if (svt == SVt_PVHV) 678 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 679 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 680 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 681 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 682 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 683 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 684 encode_tag (enc, CBOR_TAG_INDIRECTION);
685 encode_sv (enc, sv);
686 }
369} 687}
370 688
371static void 689static void
372encode_nv (enc_t *enc, SV *sv) 690encode_nv (enc_t *enc, SV *sv)
373{ 691{
374 double nv = SvNVX (sv); 692 double nv = SvNVX (sv);
375 693
376 need (enc, 9); 694 need (enc, 9);
377 695
378 if (ecb_expect_false (nv == (U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 698 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
382 { 700 encode_float32 (enc, nv);
383 uint32_t fp = ecb_float_to_binary32 (nv);
384
385 *enc->cur++ = 0xe0 | 26;
386
387 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp);
389
390 memcpy (enc->cur, &fp, 4);
391 enc->cur += 4;
392 }
393 else 701 else
394 { 702 encode_float64 (enc, nv);
395 uint64_t fp = ecb_double_to_binary64 (nv);
396
397 *enc->cur++ = 0xe0 | 27;
398
399 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp);
401
402 memcpy (enc->cur, &fp, 8);
403 enc->cur += 8;
404 }
405} 703}
406 704
407static void 705static void
408encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
409{ 707{
411 709
412 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
413 { 711 {
414 STRLEN len; 712 STRLEN len;
415 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
417 } 715 }
418 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 717 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
421 { 719 encode_int (e,v sv);
422 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv));
424 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv));
426 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
428 }
429 else if (SvROK (sv)) 720 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 726 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 727 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 728 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 729}
439 730
440static SV * 731static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
442{ 733{
443 enc_t enc; 734 enc_t enc = { 0 };
444 735
445 enc.cbor = *cbor; 736 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 740
451 SvPOK_only (enc.sv); 741 SvPOK_only (enc.sv);
742
743 if (cbor->flags & F_PACK_STRINGS)
744 {
745 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
746 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
747 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
748 }
749
452 encode_sv (&enc, scalar); 750 encode_sv (&enc, scalar);
453 751
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 752 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 753 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 754
470 U8 *end; // end of input string 768 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 769 const char *err; // parse error, if != 0
472 CBOR cbor; 770 CBOR cbor;
473 U32 depth; // recursion depth 771 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 772 U32 maxdepth; // recursion depth limit
773 AV *shareable;
774 AV *stringref;
775 SV *decode_tagged;
776 SV *err_sv; // optional sv for error, needs to be freed
475} dec_t; 777} dec_t;
476 778
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 779// set dec->err to ERRSV
780ecb_cold static void
781err_errsv (dec_t *dec)
782{
783 if (!dec->err)
784 {
785 dec->err_sv = newSVsv (ERRSV);
478 786
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 787 // chop off the trailing \n
788 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
789 *SvEND (dec->err_sv) = 0;
480 790
791 dec->err = SvPVutf8_nolen (dec->err_sv);
792 }
793}
794
795// the following functions are used to reduce code size and help the compiler to optimise
796ecb_cold static void
797err_set (dec_t *dec, const char *reason)
798{
799 if (!dec->err)
800 dec->err = reason;
801}
802
803ecb_cold static void
804err_unexpected_end (dec_t *dec)
805{
806 err_set (dec, "unexpected end of CBOR data");
807}
808
809#define ERR_DO(do) SB do; goto fail; SE
810#define ERR(reason) ERR_DO (err_set (dec, reason))
811#define ERR_ERRSV ERR_DO (err_errsv (dec))
812
813#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
814
481#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 815#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
482#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
483 817
484static UV 818static UV
485decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
486{ 820{
487 switch (*dec->cur & 31) 821 U8 m = *dec->cur & MINOR_MASK;
488 { 822 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 823
494 case 24: 824 if (ecb_expect_true (m < LENGTH_EXT1))
825 return m;
826 else if (ecb_expect_true (m == LENGTH_EXT1))
827 {
495 WANT (2); 828 WANT (1);
496 dec->cur += 2; 829 dec->cur += 1;
497 return dec->cur[-1]; 830 return dec->cur[-1];
498 831 }
499 case 25: 832 else if (ecb_expect_true (m == LENGTH_EXT2))
833 {
500 WANT (3); 834 WANT (2);
501 dec->cur += 3; 835 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 836 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 837 | ((UV)dec->cur[-1]);
504 838 }
505 case 26: 839 else if (ecb_expect_true (m == LENGTH_EXT4))
840 {
506 WANT (5); 841 WANT (4);
507 dec->cur += 5; 842 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 843 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 844 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 845 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
512 847 }
513 case 27: 848 else if (ecb_expect_true (m == LENGTH_EXT8))
849 {
514 WANT (9); 850 WANT (8);
515 dec->cur += 9; 851 dec->cur += 8;
852
853 return
854#if UVSIZE < 8
855 0
856#else
516 return (((UV)dec->cur[-8]) << 56) 857 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 858 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 859 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 860 | (((UV)dec->cur[-5]) << 32)
861#endif
520 | (((UV)dec->cur[-4]) << 24) 862 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 863 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 864 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 865 | ((UV)dec->cur[-1]);
524 866 }
525 default: 867 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 868 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 869
529fail: 870fail:
530 return 0; 871 return 0;
531} 872}
532 873
537{ 878{
538 AV *av = newAV (); 879 AV *av = newAV ();
539 880
540 DEC_INC_DEPTH; 881 DEC_INC_DEPTH;
541 882
542 if ((*dec->cur & 31) == 31) 883 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 884 {
544 ++dec->cur; 885 ++dec->cur;
545 886
546 for (;;) 887 for (;;)
547 { 888 {
548 WANT (1); 889 WANT (1);
549 890
550 if (*dec->cur == (0xe0 | 31)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
551 { 892 {
552 ++dec->cur; 893 ++dec->cur;
553 break; 894 break;
554 } 895 }
555 896
556 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
557 } 898 }
558 } 899 }
559 else 900 else
560 { 901 {
561 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
562 903
904 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 905 av_fill (av, len - 1);
564 906
565 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 908 AvARRAY (av)[i] = decode_sv (dec);
567 } 909 }
568 910
569 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
570 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
571 913
572fail: 914fail:
573 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
574 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 917 return &PL_sv_undef;
576} 918}
577 919
920static void
921decode_he (dec_t *dec, HV *hv)
922{
923 // for speed reasons, we specialcase single-string
924 // byte or utf-8 strings as keys, but only when !stringref
925
926 if (ecb_expect_true (!dec->stringref))
927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
928 {
929 STRLEN len = decode_uint (dec);
930 char *key = (char *)dec->cur;
931
932 WANT (len);
933 dec->cur += len;
934
935 hv_store (hv, key, len, decode_sv (dec), 0);
936
937 return;
938 }
939 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
940 {
941 STRLEN len = decode_uint (dec);
942 char *key = (char *)dec->cur;
943
944 WANT (len);
945 dec->cur += len;
946
947 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
948 if (!is_utf8_string (key, len))
949 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
950
951 hv_store (hv, key, -len, decode_sv (dec), 0);
952
953 return;
954 }
955
956 SV *k = decode_sv (dec);
957 SV *v = decode_sv (dec);
958
959 // we leak memory if uncaught exceptions are thrown by random magical
960 // methods, and this is hopefully the only place where it can happen,
961 // so if there is a chance of an exception, take the very slow path.
962 // since catching exceptions is "undocumented/internal/forbidden" by
963 // the new p5p powers, we need to call out to a perl function :/
964 if (ecb_expect_false (SvAMAGIC (k)))
965 {
966 dSP;
967
968 ENTER; SAVETMPS;
969 PUSHMARK (SP);
970 EXTEND (SP, 3);
971 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
972 PUSHs (sv_2mortal (k));
973 PUSHs (sv_2mortal (v));
974
975 PUTBACK;
976 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
977 SPAGAIN;
978
979 FREETMPS; LEAVE;
980
981 if (SvTRUE (ERRSV))
982 ERR_ERRSV;
983
984 return;
985 }
986
987 hv_store_ent (hv, k, v, 0);
988 SvREFCNT_dec_NN (k);
989
990fail:
991 ;
992}
993
578static SV * 994static SV *
579decode_hv (dec_t *dec) 995decode_hv (dec_t *dec)
580{ 996{
581 HV *hv = newHV (); 997 HV *hv = newHV ();
582 998
583 DEC_INC_DEPTH; 999 DEC_INC_DEPTH;
584 1000
585 if ((*dec->cur & 31) == 31) 1001 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 1002 {
587 ++dec->cur; 1003 ++dec->cur;
588 1004
589 for (;;) 1005 for (;;)
590 { 1006 {
591 WANT (1); 1007 WANT (1);
592 1008
593 if (*dec->cur == (0xe0 | 31)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
594 { 1010 {
595 ++dec->cur; 1011 ++dec->cur;
596 break; 1012 break;
597 } 1013 }
598 1014
599 SV *k = decode_sv (dec); 1015 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 1016 }
605 } 1017 }
606 else 1018 else
607 { 1019 {
608 int len = decode_uint (dec); 1020 UV pairs = decode_uint (dec);
609 1021
1022 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1023
610 while (len--) 1024 while (pairs--)
611 { 1025 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 1026 }
619 1027
620 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
622 1030
623fail: 1031fail:
624 SvREFCNT_dec (hv); 1032 SvREFCNT_dec_NN (hv);
625 DEC_DEC_DEPTH; 1033 DEC_DEC_DEPTH;
626 return &PL_sv_undef; 1034 return &PL_sv_undef;
627} 1035}
628 1036
629static SV * 1037static SV *
630decode_str (dec_t *dec, int utf8) 1038decode_str (dec_t *dec, int utf8)
631{ 1039{
632 SV *sv = 0; 1040 SV *sv = 0;
633 1041
634 if ((*dec->cur & 31) == 31) 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
635 { 1043 {
1044 // indefinite length strings
636 ++dec->cur; 1045 ++dec->cur;
637 1046
1047 U8 major = *dec->cur & MAJOR_MISC;
1048
638 sv = newSVpvn ("", 0); 1049 sv = newSVpvn ("", 0);
639 1050
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 1051 for (;;)
642 { 1052 {
643 WANT (1); 1053 WANT (1);
644 1054
645 if (*dec->cur == (0xe0 | 31)) 1055 if ((*dec->cur - major) > LENGTH_EXT8)
1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 1057 {
647 ++dec->cur; 1058 ++dec->cur;
648 break; 1059 break;
649 } 1060 }
1061 else
1062 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 1063
651 sv_catsv (sv, decode_sv (dec)); 1064 STRLEN len = decode_uint (dec);
1065
1066 WANT (len);
1067 sv_catpvn (sv, dec->cur, len);
1068 dec->cur += len;
652 } 1069 }
653 } 1070 }
654 else 1071 else
655 { 1072 {
656 STRLEN len = decode_uint (dec); 1073 STRLEN len = decode_uint (dec);
657 1074
658 WANT (len); 1075 WANT (len);
659 sv = newSVpvn (dec->cur, len); 1076 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 1077 dec->cur += len;
1078
1079 if (ecb_expect_false (dec->stringref)
1080 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1081 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 1082 }
662 1083
663 if (utf8) 1084 if (utf8)
1085 {
1086 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1087 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1088 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1089
664 SvUTF8_on (sv); 1090 SvUTF8_on (sv);
1091 }
665 1092
666 return sv; 1093 return sv;
667 1094
668fail: 1095fail:
669 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
671} 1098}
672 1099
673static SV * 1100static SV *
674decode_tagged (dec_t *dec) 1101decode_tagged (dec_t *dec)
675{ 1102{
1103 SV *sv = 0;
676 UV tag = decode_uint (dec); 1104 UV tag = decode_uint (dec);
1105
1106 WANT (1);
1107
1108 switch (tag)
1109 {
1110 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 1111 sv = decode_sv (dec);
1112 break;
678 1113
679 if (tag == CBOR_TAG_MAGIC) 1114 case CBOR_TAG_INDIRECTION:
680 return sv; 1115 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 1116 break;
682 { 1117
1118 case CBOR_TAG_STRINGREF_NAMESPACE:
1119 {
1120 // do not use SAVETMPS/FREETMPS, as these will
1121 // erase mortalised caches, e.g. "shareable"
1122 ENTER;
1123
1124 SAVESPTR (dec->stringref);
1125 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1126
1127 sv = decode_sv (dec);
1128
1129 LEAVE;
1130 }
1131 break;
1132
1133 case CBOR_TAG_STRINGREF:
1134 {
1135 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1136 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1137
1138 UV idx = decode_uint (dec);
1139
1140 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1141 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1142
1143 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1144 }
1145 break;
1146
1147 case CBOR_TAG_VALUE_SHAREABLE:
1148 {
1149 if (ecb_expect_false (!dec->shareable))
1150 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1151
1152 if (dec->cbor.flags & F_ALLOW_CYCLES)
1153 {
1154 sv = newSV (0);
1155 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1156
1157 SV *osv = decode_sv (dec);
1158 sv_setsv (sv, osv);
1159 SvREFCNT_dec_NN (osv);
1160 }
1161 else
1162 {
1163 av_push (dec->shareable, &PL_sv_undef);
1164 int idx = AvFILLp (dec->shareable);
1165 sv = decode_sv (dec);
1166 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1167 }
1168 }
1169 break;
1170
1171 case CBOR_TAG_VALUE_SHAREDREF:
1172 {
1173 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1174 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1175
1176 UV idx = decode_uint (dec);
1177
1178 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1179 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1180
1181 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1182
1183 if (sv == &PL_sv_undef)
1184 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1185 }
1186 break;
1187
1188 case CBOR_TAG_PERL_OBJECT:
1189 {
1190 if (dec->cbor.flags & F_FORBID_OBJECTS)
1191 goto filter;
1192
1193 sv = decode_sv (dec);
1194
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 1196 ERR ("corrupted CBOR data (non-array perl object)");
685 1197
686 AV *av = (AV *)SvRV (sv); 1198 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 1199 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1200 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 1201
690 if (!stash) 1202 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 1203 ERR ("cannot decode perl-object (package does not exist)");
692 1204
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1205 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 1206
695 if (!method) 1207 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1208 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 1209
698 dSP; 1210 dSP;
699 1211
700 ENTER; SAVETMPS; PUSHMARK (SP); 1212 ENTER; SAVETMPS;
1213 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1214 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1215 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1216 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1217 PUSHs (sv_cbor);
705 1218
706 int i; 1219 int i;
707 1220
708 for (i = 1; i < len; ++i) 1221 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1222 PUSHs (*av_fetch (av, i, 1));
710 1223
711 PUTBACK; 1224 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1225 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1226 SPAGAIN;
714 1227
1228 if (SvTRUE (ERRSV))
1229 {
1230 FREETMPS; LEAVE;
1231 ERR_ERRSV;
1232 }
1233
1234 SvREFCNT_dec_NN (sv);
715 sv = SvREFCNT_inc (POPs); 1235 sv = SvREFCNT_inc (POPs);
716 1236
717 PUTBACK; 1237 PUTBACK;
718 1238
719 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1240 }
1241 break;
720 1242
721 return sv; 1243 default:
722 } 1244 filter:
723 else 1245 {
724 { 1246 SV *tag_sv = newSVuv (tag);
1247
1248 sv = decode_sv (dec);
1249
1250 dSP;
1251 ENTER; SAVETMPS;
1252 PUSHMARK (SP);
1253 EXTEND (SP, 2);
1254 PUSHs (tag_sv);
1255 PUSHs (sv);
1256
1257 PUTBACK;
1258 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1259 SPAGAIN;
1260
1261 if (SvTRUE (ERRSV))
1262 {
1263 SvREFCNT_dec_NN (tag_sv);
1264 FREETMPS; LEAVE;
1265 ERR_ERRSV;
1266 }
1267
1268 if (count)
1269 {
1270 SvREFCNT_dec_NN (tag_sv);
1271 SvREFCNT_dec_NN (sv);
1272 sv = SvREFCNT_inc_NN (TOPs);
1273 SP -= count;
1274 }
1275 else
1276 {
725 AV *av = newAV (); 1277 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1278 av_push (av, tag_sv);
727 av_push (av, sv); 1279 av_push (av, sv);
728 1280
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1282 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1284 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1285 }
1286
1287 PUTBACK;
1288
1289 FREETMPS; LEAVE;
1290 }
1291 break;
734 } 1292 }
1293
1294 return sv;
735 1295
736fail: 1296fail:
737 SvREFCNT_dec (sv); 1297 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1298 return &PL_sv_undef;
739} 1299}
741static SV * 1301static SV *
742decode_sv (dec_t *dec) 1302decode_sv (dec_t *dec)
743{ 1303{
744 WANT (1); 1304 WANT (1);
745 1305
746 switch (*dec->cur >> 5) 1306 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1307 {
748 case 0: // unsigned int 1308 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1309 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1310 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1311 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1312 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1313 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1314 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1315
756 case 4: // array 1316 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1317 switch (*dec->cur++ & MINOR_MASK)
764 { 1318 {
765 case 20: 1319 case SIMPLE_FALSE:
766#if CBOR_SLOW 1320#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1321 types_false = get_bool ("Types::Serialiser::false");
768#endif 1322#endif
769 return newSVsv (types_false); 1323 return newSVsv (types_false);
770 case 21: 1324 case SIMPLE_TRUE:
771#if CBOR_SLOW 1325#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1326 types_true = get_bool ("Types::Serialiser::true");
773#endif 1327#endif
774 return newSVsv (types_true); 1328 return newSVsv (types_true);
775 case 22: 1329 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1330 return newSVsv (&PL_sv_undef);
777 case 23: 1331 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1332#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1333 types_error = get_bool ("Types::Serialiser::error");
780#endif 1334#endif
781 return newSVsv (types_error); 1335 return newSVsv (types_error);
782 1336
783 case 25: 1337 case MISC_FLOAT16:
784 { 1338 {
785 WANT (2); 1339 WANT (2);
786 1340
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1341 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1342 dec->cur += 2;
789 1343
790 return newSVnv (ecb_binary16_to_float (fp)); 1344 return newSVnv (ecb_binary16_to_float (fp));
791 } 1345 }
792 1346
793 case 26: 1347 case MISC_FLOAT32:
794 { 1348 {
795 uint32_t fp; 1349 uint32_t fp;
796 WANT (4); 1350 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1351 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1352 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1355 fp = ecb_bswap32 (fp);
802 1356
803 return newSVnv (ecb_binary32_to_float (fp)); 1357 return newSVnv (ecb_binary32_to_float (fp));
804 } 1358 }
805 1359
806 case 27: 1360 case MISC_FLOAT64:
807 { 1361 {
808 uint64_t fp; 1362 uint64_t fp;
809 WANT (8); 1363 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1364 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1365 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1368 fp = ecb_bswap64 (fp);
815 1369
816 return newSVnv (ecb_binary64_to_double (fp)); 1370 return newSVnv (ecb_binary64_to_double (fp));
817 } 1371 }
818 1372
819 // 0..19 unassigned 1373 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1374 // 24 reserved + unassigned simple (reserved values are not encodable)
1375 // 28-30 unassigned misc
1376 // 31 break code
821 default: 1377 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1378 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1379 }
824 1380
825 break; 1381 break;
826 } 1382 }
827 1383
830} 1386}
831 1387
832static SV * 1388static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1389decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1390{
835 dec_t dec; 1391 dec_t dec = { 0 };
836 SV *sv; 1392 SV *sv;
1393 STRLEN len;
1394 char *data = SvPVbyte (string, len);
837 1395
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1396 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1397 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1398 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1399
873 dec.cbor = *cbor; 1400 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1401 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1402 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1403
879 sv = decode_sv (&dec); 1404 sv = decode_sv (&dec);
880 1405
881 if (offset_return) 1406 if (offset_return)
882 *offset_return = dec.cur; 1407 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1410 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1411 dec.err = "garbage after CBOR object";
887 1412
888 if (dec.err) 1413 if (dec.err)
889 { 1414 {
1415 if (dec.shareable)
1416 {
1417 // need to break cyclic links, which would all be in shareable
1418 int i;
1419 SV **svp;
1420
1421 for (i = av_len (dec.shareable) + 1; i--; )
1422 if ((svp = av_fetch (dec.shareable, i, 0)))
1423 sv_setsv (*svp, &PL_sv_undef);
1424 }
1425
890 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1431 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1432 }
893 1433
894 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
895 1435
896 return sv; 1436 return sv;
897} 1437}
898 1438
1439/////////////////////////////////////////////////////////////////////////////
1440// incremental parser
1441
1442#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1443
1444// returns 0 for notyet, 1 for success or error
1445static int
1446incr_parse (CBOR *self, SV *cborstr)
1447{
1448 STRLEN cur;
1449 SvPV (cborstr, cur);
1450
1451 while (ecb_expect_true (self->incr_need <= cur))
1452 {
1453 // table of integer count bytes
1454 static I8 incr_len[MINOR_MASK + 1] = {
1455 0, 0, 0, 0, 0, 0, 0, 0,
1456 0, 0, 0, 0, 0, 0, 0, 0,
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 1, 2, 4, 8,-1,-1,-1,-2
1459 };
1460
1461 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1462 U8 m = *p & MINOR_MASK;
1463 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1464 I8 ilen = incr_len[m];
1465
1466 self->incr_need = self->incr_pos + 1;
1467
1468 if (ecb_expect_false (ilen < 0))
1469 {
1470 if (m != MINOR_INDEF)
1471 return 1; // error
1472
1473 if (*p == (MAJOR_MISC | MINOR_INDEF))
1474 {
1475 if (count >= 0)
1476 return 1; // error
1477
1478 count = 1;
1479 }
1480 else
1481 {
1482 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1483 count = -1;
1484 }
1485 }
1486 else
1487 {
1488 self->incr_need += ilen;
1489 if (ecb_expect_false (self->incr_need > cur))
1490 return 0;
1491
1492 int major = *p >> MAJOR_SHIFT;
1493
1494 switch (major)
1495 {
1496 case MAJOR_TAG >> MAJOR_SHIFT:
1497 ++count; // tags merely prefix another value
1498 break;
1499
1500 case MAJOR_BYTES >> MAJOR_SHIFT:
1501 case MAJOR_TEXT >> MAJOR_SHIFT:
1502 case MAJOR_ARRAY >> MAJOR_SHIFT:
1503 case MAJOR_MAP >> MAJOR_SHIFT:
1504 {
1505 UV len;
1506
1507 if (ecb_expect_false (ilen))
1508 {
1509 len = 0;
1510
1511 do {
1512 len = (len << 8) | *++p;
1513 } while (--ilen);
1514 }
1515 else
1516 len = m;
1517
1518 switch (major)
1519 {
1520 case MAJOR_BYTES >> MAJOR_SHIFT:
1521 case MAJOR_TEXT >> MAJOR_SHIFT:
1522 self->incr_need += len;
1523 if (ecb_expect_false (self->incr_need > cur))
1524 return 0;
1525
1526 break;
1527
1528 case MAJOR_MAP >> MAJOR_SHIFT:
1529 len <<= 1;
1530 case MAJOR_ARRAY >> MAJOR_SHIFT:
1531 if (len)
1532 {
1533 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1534 count = len + 1;
1535 }
1536 break;
1537 }
1538 }
1539 }
1540 }
1541
1542 self->incr_pos = self->incr_need;
1543
1544 if (count > 0)
1545 {
1546 while (!--count)
1547 {
1548 if (!AvFILLp (self->incr_count))
1549 return 1; // done
1550
1551 SvREFCNT_dec_NN (av_pop (self->incr_count));
1552 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1553 }
1554
1555 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1556 }
1557 }
1558
1559 return 0;
1560}
1561
1562
899///////////////////////////////////////////////////////////////////////////// 1563/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1564// XS interface functions
901 1565
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1566MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1567
911 1575
912 types_true = get_bool ("Types::Serialiser::true" ); 1576 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1577 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1578 types_error = get_bool ("Types::Serialiser::error");
915 1579
1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1581
916 sv_cbor = newSVpv ("CBOR", 0); 1582 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
918} 1586}
919 1587
920PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
921 1589
922void CLONE (...) 1590void CLONE (...)
940 1608
941void shrink (CBOR *self, int enable = 1) 1609void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1610 ALIAS:
943 shrink = F_SHRINK 1611 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1612 allow_unknown = F_ALLOW_UNKNOWN
1613 allow_sharing = F_ALLOW_SHARING
1614 allow_cycles = F_ALLOW_CYCLES
1615 forbid_objects = F_FORBID_OBJECTS
1616 pack_strings = F_PACK_STRINGS
1617 text_keys = F_TEXT_KEYS
1618 text_strings = F_TEXT_STRINGS
1619 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1620 PPCODE:
946{ 1621{
947 if (enable) 1622 if (enable)
948 self->flags |= ix; 1623 self->flags |= ix;
949 else 1624 else
954 1629
955void get_shrink (CBOR *self) 1630void get_shrink (CBOR *self)
956 ALIAS: 1631 ALIAS:
957 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1633 get_allow_unknown = F_ALLOW_UNKNOWN
1634 get_allow_sharing = F_ALLOW_SHARING
1635 get_allow_cycles = F_ALLOW_CYCLES
1636 get_forbid_objects = F_FORBID_OBJECTS
1637 get_pack_strings = F_PACK_STRINGS
1638 get_text_keys = F_TEXT_KEYS
1639 get_text_strings = F_TEXT_STRINGS
1640 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1641 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
961 1643
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1645 PPCODE:
979 CODE: 1661 CODE:
980 RETVAL = self->max_size; 1662 RETVAL = self->max_size;
981 OUTPUT: 1663 OUTPUT:
982 RETVAL 1664 RETVAL
983 1665
984#if 0 //TODO 1666void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1667 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1668 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1669 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1670 XPUSHs (ST (0));
993}
994 1671
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1672SV *get_filter (CBOR *self)
996 PPCODE: 1673 CODE:
997{ 1674 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1675 OUTPUT:
999 self->cb_sk_object = newHV (); 1676 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1677
1019void encode (CBOR *self, SV *scalar) 1678void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1679 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1680 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1681 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1694 EXTEND (SP, 2);
1036 PUSHs (sv); 1695 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1696 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1697}
1039 1698
1040#if 0 1699void incr_parse (CBOR *self, SV *cborstr)
1700 ALIAS:
1701 incr_parse_multiple = 1
1702 PPCODE:
1703{
1704 if (SvUTF8 (cborstr))
1705 sv_utf8_downgrade (cborstr, 0);
1706
1707 if (!self->incr_count)
1708 {
1709 self->incr_count = newAV ();
1710 self->incr_pos = 0;
1711 self->incr_need = 1;
1712
1713 av_push (self->incr_count, newSViv (1));
1714 }
1715
1716 do
1717 {
1718 if (!incr_parse (self, cborstr))
1719 {
1720 if (self->incr_need > self->max_size && self->max_size)
1721 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1722 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1723
1724 break;
1725 }
1726
1727 SV *sv;
1728 char *offset;
1729
1730 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1731 XPUSHs (sv);
1732
1733 sv_chop (cborstr, offset);
1734
1735 av_clear (self->incr_count);
1736 av_push (self->incr_count, newSViv (1));
1737
1738 self->incr_pos = 0;
1739 self->incr_need = self->incr_pos + 1;
1740 }
1741 while (ix);
1742}
1743
1744void incr_reset (CBOR *self)
1745 CODE:
1746{
1747 SvREFCNT_dec (self->incr_count);
1748 self->incr_count = 0;
1749}
1041 1750
1042void DESTROY (CBOR *self) 1751void DESTROY (CBOR *self)
1043 CODE: 1752 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1753 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1754
1049PROTOTYPES: ENABLE 1755PROTOTYPES: ENABLE
1050 1756
1051void encode_cbor (SV *scalar) 1757void encode_cbor (SV *scalar)
1758 ALIAS:
1759 encode_cbor = 0
1760 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1761 PPCODE:
1053{ 1762{
1054 CBOR cbor; 1763 CBOR cbor;
1055 cbor_init (&cbor); 1764 cbor_init (&cbor);
1765 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1766 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1767 XPUSHs (scalar);
1058} 1768}
1059 1769
1060void decode_cbor (SV *cborstr) 1770void decode_cbor (SV *cborstr)
1064 cbor_init (&cbor); 1774 cbor_init (&cbor);
1065 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1066 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1067} 1777}
1068 1778
1779#ifdef __AFL_COMPILER
1780
1781void
1782afl_init ()
1783 CODE:
1784 __AFL_INIT ();
1785
1786int
1787afl_loop (unsigned int count = 10000)
1788 CODE:
1789 RETVAL = __AFL_LOOP (count);
1790 OUTPUT:
1791 RETVAL
1792
1793#endif
1794

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines