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 1.16 by root, Tue Oct 29 20:59:16 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"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
24 69
25// known tags 70// known tags
26enum cbor_tag 71enum cbor_tag
27{ 72{
28 // inofficial extensions (pending iana registration) 73 // extensions
29 CBOR_TAG_PERL_OBJECT = 256, 74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
31 81
32 // rfc7049 82 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
39 89
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 94
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 100
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 102};
53 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_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
54#define F_SHRINK 0x00000200UL 116#define F_SHRINK 0x00000001UL
55#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
56 125
57#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
58 127
59#define SB do { 128#define SB do {
60#define SE } while (0) 129#define SE } while (0)
72# define CBOR_SLOW 0 141# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 142# define CBOR_STASH cbor_stash
74#endif 143#endif
75 144
76static 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::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 146static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 147
79typedef struct { 148typedef struct {
80 U32 flags; 149 U32 flags;
81 U32 max_depth; 150 U32 max_depth;
82 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.
83} CBOR; 158} CBOR;
84 159
85ecb_inline void 160ecb_inline void
86cbor_init (CBOR *cbor) 161cbor_init (CBOR *cbor)
87{ 162{
88 Zero (cbor, 1, CBOR); 163 Zero (cbor, 1, CBOR);
89 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);
90} 172}
91 173
92///////////////////////////////////////////////////////////////////////////// 174/////////////////////////////////////////////////////////////////////////////
93// utility functions 175// utility functions
94 176
116 SvPV_renew (sv, SvCUR (sv) + 1); 198 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 199#endif
118 } 200 }
119} 201}
120 202
121///////////////////////////////////////////////////////////////////////////// 203// minimum length of a string to be registered for stringref
122// fp hell 204ecb_inline int
123 205minimum_string_length (UV idx)
124//TODO 206{
207 return idx <= 23 ? 3
208 : idx <= 0xffU ? 4
209 : idx <= 0xffffU ? 5
210 : idx <= 0xffffffffU ? 7
211 : 11;
212}
125 213
126///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
127// encoder 215// encoder
128 216
129// structure used for encoding CBOR 217// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 220 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 221 char *end; // SvEND (sv)
134 SV *sv; // result scalar 222 SV *sv; // result scalar
135 CBOR cbor; 223 CBOR cbor;
136 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;
137} enc_t; 229} enc_t;
138 230
139ecb_inline void 231ecb_inline void
140need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
141{ 233{
142 if (ecb_expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
143 { 235 {
144 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
145 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
146 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
147 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
153{ 245{
154 need (enc, 1); 246 need (enc, 1);
155 *enc->cur++ = ch; 247 *enc->cur++ = ch;
156} 248}
157 249
250// used for tags, intregers, element counts and so on
158static void 251static void
159encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
160{ 253{
161 need (enc, 9); 254 need (enc, 9);
162 255
163 if (len < 24) 256 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 257 *enc->cur++ = major | len;
165 else if (len <= 0xff) 258 else if (ecb_expect_true (len <= 0xffU))
166 { 259 {
167 *enc->cur++ = major | 24; 260 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 261 *enc->cur++ = len;
169 } 262 }
170 else if (len <= 0xffff) 263 else if (len <= 0xffffU)
171 { 264 {
172 *enc->cur++ = major | 25; 265 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 266 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 267 *enc->cur++ = len;
175 } 268 }
176 else if (len <= 0xffffffff) 269 else if (len <= 0xffffffffU)
177 { 270 {
178 *enc->cur++ = major | 26; 271 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 272 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 273 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 274 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 275 *enc->cur++ = len;
183 } 276 }
184 else 277 else
185 { 278 {
186 *enc->cur++ = major | 27; 279 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 280 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 281 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 282 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 283 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 284 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 287 *enc->cur++ = len;
195 } 288 }
196} 289}
197 290
198static 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
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 311encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 312{
201 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);
202 need (enc, len); 343 need (enc, len);
203 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
204 enc->cur += len; 345 enc->cur += len;
205} 346}
206 347
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
207static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
208 463
209static void 464static void
210encode_av (enc_t *enc, AV *av) 465encode_av (enc_t *enc, AV *av)
211{ 466{
214 if (enc->depth >= enc->cbor.max_depth) 469 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 470 croak (ERR_NESTING_EXCEEDED);
216 471
217 ++enc->depth; 472 ++enc->depth;
218 473
219 encode_uint (enc, 0x80, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 475
476 if (ecb_expect_false (SvMAGICAL (av)))
221 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
222 { 478 {
223 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 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 }
226 488
227 --enc->depth; 489 --enc->depth;
228} 490}
229 491
230static void 492static void
238 ++enc->depth; 500 ++enc->depth;
239 501
240 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
242 504
243 if (mg) 505 if (ecb_expect_false (mg))
244 encode_ch (enc, 0xa0 | 31); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 507 else
246 encode_uint (enc, 0xa0, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
247 509
248 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
249 { 511 {
250 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
252 else 514 else
253 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));
254 516
255 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));
256 } 518 }
257 519
258 if (mg) 520 if (ecb_expect_false (mg))
259 encode_ch (enc, 0xe0 | 31); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 522
261 --enc->depth; 523 --enc->depth;
262} 524}
263 525
264// encode objects, arrays and special \0=false and \1=true values. 526// encode objects, arrays and special \0=false and \1=true values.
265static void 527static void
266encode_rv (enc_t *enc, SV *sv) 528encode_rv (enc_t *enc, SV *sv)
267{ 529{
268 svtype svt;
269
270 SvGETMAGIC (sv); 530 SvGETMAGIC (sv);
531
271 svt = SvTYPE (sv); 532 svtype svt = SvTYPE (sv);
272 533
273 if (ecb_expect_false (SvOBJECT (sv))) 534 if (ecb_expect_false (SvOBJECT (sv)))
274 { 535 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 536 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 537 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 542 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 543 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 544 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 545
285 HV *stash = SvSTASH (sv); 546 HV *stash = SvSTASH (sv);
286 GV *method;
287 547
288 if (stash == boolean_stash) 548 if (stash == boolean_stash)
289 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 }
290 else if (stash == error_stash) 553 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 554 {
555 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
556 return;
557 }
292 else if (stash == tagged_stash) 558 else if (stash == tagged_stash)
293 { 559 {
294 if (svt != SVt_PVAV) 560 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 567 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 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))
299 } 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))));
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 615 {
302 dSP; 616 dSP;
303 617
304 ENTER; SAVETMPS; PUSHMARK (SP); 618 ENTER; SAVETMPS;
619 PUSHMARK (SP);
305 // 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
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 621 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307 622
308 PUTBACK; 623 PUTBACK;
309 // G_SCALAR ensures that return value is 1 624 // G_SCALAR ensures that return value is 1
322 } 637 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 { 639 {
325 dSP; 640 dSP;
326 641
327 ENTER; SAVETMPS; PUSHMARK (SP); 642 ENTER; SAVETMPS;
643 PUSHMARK (SP);
328 EXTEND (SP, 2); 644 EXTEND (SP, 2);
329 // 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
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor); 647 PUSHs (sv_cbor);
332 648
336 652
337 // catch this surprisingly common error 653 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 655 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 656
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 658 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 660
345 while (count) 661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
346 encode_sv (enc, SP[1 - count--]); 665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
668 }
347 669
348 PUTBACK; 670 PUTBACK;
349 671
350 FREETMPS; LEAVE; 672 FREETMPS; LEAVE;
351 } 673 }
355 } 677 }
356 else if (svt == SVt_PVHV) 678 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 679 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 680 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 681 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 682 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 683 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 684 encode_tag (enc, CBOR_TAG_INDIRECTION);
685 encode_sv (enc, sv);
686 }
380} 687}
381 688
382static void 689static void
383encode_nv (enc_t *enc, SV *sv) 690encode_nv (enc_t *enc, SV *sv)
384{ 691{
385 double nv = SvNVX (sv); 692 double nv = SvNVX (sv);
386 693
387 need (enc, 9); 694 need (enc, 9);
388 695
389 if (ecb_expect_false (nv == (U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 698 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
393 { 700 encode_float32 (enc, nv);
394 uint32_t fp = ecb_float_to_binary32 (nv);
395
396 *enc->cur++ = 0xe0 | 26;
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403 }
404 else 701 else
405 { 702 encode_float64 (enc, nv);
406 uint64_t fp = ecb_double_to_binary64 (nv);
407
408 *enc->cur++ = 0xe0 | 27;
409
410 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp);
412
413 memcpy (enc->cur, &fp, 8);
414 enc->cur += 8;
415 }
416} 703}
417 704
418static void 705static void
419encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
420{ 707{
422 709
423 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
424 { 711 {
425 STRLEN len; 712 STRLEN len;
426 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
428 } 715 }
429 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 717 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
432 { 719 encode_int (e,v sv);
433 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv));
435 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv));
437 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
439 }
440 else if (SvROK (sv)) 720 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 726 else
447 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",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 728 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 729}
450 730
451static SV * 731static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
453{ 733{
454 enc_t enc; 734 enc_t enc = { 0 };
455 735
456 enc.cbor = *cbor; 736 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 740
462 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
463 encode_sv (&enc, scalar); 750 encode_sv (&enc, scalar);
464 751
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 752 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *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
467 754
481 U8 *end; // end of input string 768 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 769 const char *err; // parse error, if != 0
483 CBOR cbor; 770 CBOR cbor;
484 U32 depth; // recursion depth 771 U32 depth; // recursion depth
485 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
486} dec_t; 777} dec_t;
487 778
488#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);
489 786
490#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;
491 790
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
492#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)
493#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
494 817
495static UV 818static UV
496decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
497{ 820{
498 switch (*dec->cur & 31) 821 U8 m = *dec->cur & MINOR_MASK;
499 { 822 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 823
505 case 24: 824 if (ecb_expect_true (m < LENGTH_EXT1))
825 return m;
826 else if (ecb_expect_true (m == LENGTH_EXT1))
827 {
506 WANT (2); 828 WANT (1);
507 dec->cur += 2; 829 dec->cur += 1;
508 return dec->cur[-1]; 830 return dec->cur[-1];
509 831 }
510 case 25: 832 else if (ecb_expect_true (m == LENGTH_EXT2))
833 {
511 WANT (3); 834 WANT (2);
512 dec->cur += 3; 835 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 836 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 837 | ((UV)dec->cur[-1]);
515 838 }
516 case 26: 839 else if (ecb_expect_true (m == LENGTH_EXT4))
840 {
517 WANT (5); 841 WANT (4);
518 dec->cur += 5; 842 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 843 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 844 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 845 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
523 847 }
524 case 27: 848 else if (ecb_expect_true (m == LENGTH_EXT8))
849 {
525 WANT (9); 850 WANT (8);
526 dec->cur += 9; 851 dec->cur += 8;
852
853 return
854#if UVSIZE < 8
855 0
856#else
527 return (((UV)dec->cur[-8]) << 56) 857 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 858 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 859 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 860 | (((UV)dec->cur[-5]) << 32)
861#endif
531 | (((UV)dec->cur[-4]) << 24) 862 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 863 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 864 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 865 | ((UV)dec->cur[-1]);
535 866 }
536 default: 867 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 868 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 869
540fail: 870fail:
541 return 0; 871 return 0;
542} 872}
543 873
548{ 878{
549 AV *av = newAV (); 879 AV *av = newAV ();
550 880
551 DEC_INC_DEPTH; 881 DEC_INC_DEPTH;
552 882
553 if ((*dec->cur & 31) == 31) 883 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 884 {
555 ++dec->cur; 885 ++dec->cur;
556 886
557 for (;;) 887 for (;;)
558 { 888 {
559 WANT (1); 889 WANT (1);
560 890
561 if (*dec->cur == (0xe0 | 31)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
562 { 892 {
563 ++dec->cur; 893 ++dec->cur;
564 break; 894 break;
565 } 895 }
566 896
567 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
568 } 898 }
569 } 899 }
570 else 900 else
571 { 901 {
572 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
573 903
904 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 905 av_fill (av, len - 1);
575 906
576 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 908 AvARRAY (av)[i] = decode_sv (dec);
578 } 909 }
579 910
580 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
581 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
582 913
583fail: 914fail:
584 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
585 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
586 return &PL_sv_undef; 917 return &PL_sv_undef;
587} 918}
588 919
589static void 920static void
590decode_he (dec_t *dec, HV *hv) 921decode_he (dec_t *dec, HV *hv)
591{ 922{
592 // for speed reasons, we specialcase single-string 923 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 924 // byte or utf-8 strings as keys, but only when !stringref
594 925
595 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 926 if (ecb_expect_true (!dec->stringref))
927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
596 { 928 {
597 I32 len = decode_uint (dec); 929 STRLEN len = decode_uint (dec);
598 char *key = (char *)dec->cur; 930 char *key = (char *)dec->cur;
599 931
932 WANT (len);
600 dec->cur += len; 933 dec->cur += len;
601 934
602 hv_store (hv, key, len, decode_sv (dec), 0); 935 hv_store (hv, key, len, decode_sv (dec), 0);
936
937 return;
603 } 938 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 939 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 940 {
606 I32 len = decode_uint (dec); 941 STRLEN len = decode_uint (dec);
607 char *key = (char *)dec->cur; 942 char *key = (char *)dec->cur;
608 943
944 WANT (len);
609 dec->cur += len; 945 dec->cur += len;
610 946
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 951 hv_store (hv, key, -len, decode_sv (dec), 0);
952
953 return;
612 } 954 }
613 else 955
614 {
615 SV *k = decode_sv (dec); 956 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 957 SV *v = decode_sv (dec);
617 958
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
618 hv_store_ent (hv, k, v, 0); 987 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 988 SvREFCNT_dec_NN (k);
620 } 989
990fail:
991 ;
621} 992}
622 993
623static SV * 994static SV *
624decode_hv (dec_t *dec) 995decode_hv (dec_t *dec)
625{ 996{
626 HV *hv = newHV (); 997 HV *hv = newHV ();
627 998
628 DEC_INC_DEPTH; 999 DEC_INC_DEPTH;
629 1000
630 if ((*dec->cur & 31) == 31) 1001 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 1002 {
632 ++dec->cur; 1003 ++dec->cur;
633 1004
634 for (;;) 1005 for (;;)
635 { 1006 {
636 WANT (1); 1007 WANT (1);
637 1008
638 if (*dec->cur == (0xe0 | 31)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
639 { 1010 {
640 ++dec->cur; 1011 ++dec->cur;
641 break; 1012 break;
642 } 1013 }
643 1014
644 decode_he (dec, hv); 1015 decode_he (dec, hv);
645 } 1016 }
646 } 1017 }
647 else 1018 else
648 { 1019 {
649 int pairs = decode_uint (dec); 1020 UV pairs = decode_uint (dec);
1021
1022 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
650 1023
651 while (pairs--) 1024 while (pairs--)
652 decode_he (dec, hv); 1025 decode_he (dec, hv);
653 } 1026 }
654 1027
655 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
656 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
657 1030
658fail: 1031fail:
659 SvREFCNT_dec (hv); 1032 SvREFCNT_dec_NN (hv);
660 DEC_DEC_DEPTH; 1033 DEC_DEC_DEPTH;
661 return &PL_sv_undef; 1034 return &PL_sv_undef;
662} 1035}
663 1036
664static SV * 1037static SV *
665decode_str (dec_t *dec, int utf8) 1038decode_str (dec_t *dec, int utf8)
666{ 1039{
667 SV *sv = 0; 1040 SV *sv = 0;
668 1041
669 if ((*dec->cur & 31) == 31) 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
670 { 1043 {
1044 // indefinite length strings
671 ++dec->cur; 1045 ++dec->cur;
672 1046
1047 U8 major = *dec->cur & MAJOR_MISC;
1048
673 sv = newSVpvn ("", 0); 1049 sv = newSVpvn ("", 0);
674 1050
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 1051 for (;;)
677 { 1052 {
678 WANT (1); 1053 WANT (1);
679 1054
680 if (*dec->cur == (0xe0 | 31)) 1055 if ((*dec->cur - major) > LENGTH_EXT8)
1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 1057 {
682 ++dec->cur; 1058 ++dec->cur;
683 break; 1059 break;
684 } 1060 }
1061 else
1062 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 1063
686 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;
687 } 1069 }
688 } 1070 }
689 else 1071 else
690 { 1072 {
691 STRLEN len = decode_uint (dec); 1073 STRLEN len = decode_uint (dec);
692 1074
693 WANT (len); 1075 WANT (len);
694 sv = newSVpvn (dec->cur, len); 1076 sv = newSVpvn (dec->cur, len);
695 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));
696 } 1082 }
697 1083
698 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
699 SvUTF8_on (sv); 1090 SvUTF8_on (sv);
1091 }
700 1092
701 return sv; 1093 return sv;
702 1094
703fail: 1095fail:
704 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
706} 1098}
707 1099
708static SV * 1100static SV *
709decode_tagged (dec_t *dec) 1101decode_tagged (dec_t *dec)
710{ 1102{
1103 SV *sv = 0;
711 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:
712 SV *sv = decode_sv (dec); 1111 sv = decode_sv (dec);
1112 break;
713 1113
714 if (tag == CBOR_TAG_MAGIC) 1114 case CBOR_TAG_INDIRECTION:
715 return sv; 1115 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 1116 break;
717 { 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
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)"); 1196 ERR ("corrupted CBOR data (non-array perl object)");
720 1197
721 AV *av = (AV *)SvRV (sv); 1198 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1; 1199 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1200 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724 1201
725 if (!stash) 1202 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)"); 1203 ERR ("cannot decode perl-object (package does not exist)");
727 1204
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1205 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 1206
730 if (!method) 1207 if (!method)
731 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)");
732 1209
733 dSP; 1210 dSP;
734 1211
735 ENTER; SAVETMPS; PUSHMARK (SP); 1212 ENTER; SAVETMPS;
1213 PUSHMARK (SP);
736 EXTEND (SP, len + 1); 1214 EXTEND (SP, len + 1);
737 // 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
738 PUSHs (*av_fetch (av, 0, 1)); 1216 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor); 1217 PUSHs (sv_cbor);
740 1218
741 int i; 1219 int i;
742 1220
743 for (i = 1; i < len; ++i) 1221 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1)); 1222 PUSHs (*av_fetch (av, i, 1));
745 1223
746 PUTBACK; 1224 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR); 1225 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN; 1226 SPAGAIN;
749 1227
1228 if (SvTRUE (ERRSV))
1229 {
1230 FREETMPS; LEAVE;
1231 ERR_ERRSV;
1232 }
1233
750 SvREFCNT_dec (sv); 1234 SvREFCNT_dec_NN (sv);
751 sv = SvREFCNT_inc (POPs); 1235 sv = SvREFCNT_inc (POPs);
752 1236
753 PUTBACK; 1237 PUTBACK;
754 1238
755 FREETMPS; LEAVE; 1239 FREETMPS; LEAVE;
1240 }
1241 break;
756 1242
757 return sv; 1243 default:
758 } 1244 filter:
759 else 1245 {
760 { 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 {
761 AV *av = newAV (); 1277 AV *av = newAV ();
762 av_push (av, newSVuv (tag)); 1278 av_push (av, tag_sv);
763 av_push (av, sv); 1279 av_push (av, sv);
764 1280
765 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
766 ? cbor_tagged_stash 1282 ? cbor_tagged_stash
767 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769 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;
770 } 1292 }
1293
1294 return sv;
771 1295
772fail: 1296fail:
773 SvREFCNT_dec (sv); 1297 SvREFCNT_dec (sv);
774 return &PL_sv_undef; 1298 return &PL_sv_undef;
775} 1299}
777static SV * 1301static SV *
778decode_sv (dec_t *dec) 1302decode_sv (dec_t *dec)
779{ 1303{
780 WANT (1); 1304 WANT (1);
781 1305
782 switch (*dec->cur >> 5) 1306 switch (*dec->cur >> MAJOR_SHIFT)
783 { 1307 {
784 case 0: // unsigned int 1308 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
785 return newSVuv (decode_uint (dec)); 1309 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
786 case 1: // negative int 1310 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
787 return newSViv (-1 - (IV)decode_uint (dec)); 1311 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
788 case 2: // octet string 1312 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
789 return decode_str (dec, 0); 1313 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
790 case 3: // utf-8 string 1314 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
791 return decode_str (dec, 1); 1315
792 case 4: // array 1316 case MAJOR_MISC >> MAJOR_SHIFT:
793 return decode_av (dec);
794 case 5: // map
795 return decode_hv (dec);
796 case 6: // tag
797 return decode_tagged (dec);
798 case 7: // misc
799 switch (*dec->cur++ & 31) 1317 switch (*dec->cur++ & MINOR_MASK)
800 { 1318 {
801 case 20: 1319 case SIMPLE_FALSE:
802#if CBOR_SLOW 1320#if CBOR_SLOW
803 types_false = get_bool ("Types::Serialiser::false"); 1321 types_false = get_bool ("Types::Serialiser::false");
804#endif 1322#endif
805 return newSVsv (types_false); 1323 return newSVsv (types_false);
806 case 21: 1324 case SIMPLE_TRUE:
807#if CBOR_SLOW 1325#if CBOR_SLOW
808 types_true = get_bool ("Types::Serialiser::true"); 1326 types_true = get_bool ("Types::Serialiser::true");
809#endif 1327#endif
810 return newSVsv (types_true); 1328 return newSVsv (types_true);
811 case 22: 1329 case SIMPLE_NULL:
812 return newSVsv (&PL_sv_undef); 1330 return newSVsv (&PL_sv_undef);
813 case 23: 1331 case SIMPLE_UNDEF:
814#if CBOR_SLOW 1332#if CBOR_SLOW
815 types_error = get_bool ("Types::Serialiser::error"); 1333 types_error = get_bool ("Types::Serialiser::error");
816#endif 1334#endif
817 return newSVsv (types_error); 1335 return newSVsv (types_error);
818 1336
819 case 25: 1337 case MISC_FLOAT16:
820 { 1338 {
821 WANT (2); 1339 WANT (2);
822 1340
823 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1341 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
824 dec->cur += 2; 1342 dec->cur += 2;
825 1343
826 return newSVnv (ecb_binary16_to_float (fp)); 1344 return newSVnv (ecb_binary16_to_float (fp));
827 } 1345 }
828 1346
829 case 26: 1347 case MISC_FLOAT32:
830 { 1348 {
831 uint32_t fp; 1349 uint32_t fp;
832 WANT (4); 1350 WANT (4);
833 memcpy (&fp, dec->cur, 4); 1351 memcpy (&fp, dec->cur, 4);
834 dec->cur += 4; 1352 dec->cur += 4;
837 fp = ecb_bswap32 (fp); 1355 fp = ecb_bswap32 (fp);
838 1356
839 return newSVnv (ecb_binary32_to_float (fp)); 1357 return newSVnv (ecb_binary32_to_float (fp));
840 } 1358 }
841 1359
842 case 27: 1360 case MISC_FLOAT64:
843 { 1361 {
844 uint64_t fp; 1362 uint64_t fp;
845 WANT (8); 1363 WANT (8);
846 memcpy (&fp, dec->cur, 8); 1364 memcpy (&fp, dec->cur, 8);
847 dec->cur += 8; 1365 dec->cur += 8;
850 fp = ecb_bswap64 (fp); 1368 fp = ecb_bswap64 (fp);
851 1369
852 return newSVnv (ecb_binary64_to_double (fp)); 1370 return newSVnv (ecb_binary64_to_double (fp));
853 } 1371 }
854 1372
855 // 0..19 unassigned 1373 // 0..19 unassigned simple
856 // 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
857 default: 1377 default:
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1378 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
859 } 1379 }
860 1380
861 break; 1381 break;
862 } 1382 }
863 1383
866} 1386}
867 1387
868static SV * 1388static SV *
869decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1389decode_cbor (SV *string, CBOR *cbor, char **offset_return)
870{ 1390{
871 dec_t dec; 1391 dec_t dec = { 0 };
872 SV *sv; 1392 SV *sv;
873 STRLEN len; 1393 STRLEN len;
874 char *data = SvPVbyte (string, len); 1394 char *data = SvPVbyte (string, len);
875 1395
876 if (len > cbor->max_size && cbor->max_size) 1396 if (len > cbor->max_size && cbor->max_size)
878 (unsigned long)len, (unsigned long)cbor->max_size); 1398 (unsigned long)len, (unsigned long)cbor->max_size);
879 1399
880 dec.cbor = *cbor; 1400 dec.cbor = *cbor;
881 dec.cur = (U8 *)data; 1401 dec.cur = (U8 *)data;
882 dec.end = (U8 *)data + len; 1402 dec.end = (U8 *)data + len;
883 dec.err = 0;
884 dec.depth = 0;
885 1403
886 sv = decode_sv (&dec); 1404 sv = decode_sv (&dec);
887 1405
888 if (offset_return) 1406 if (offset_return)
889 *offset_return = dec.cur; 1407 *offset_return = dec.cur;
892 if (dec.cur != dec.end && !dec.err) 1410 if (dec.cur != dec.end && !dec.err)
893 dec.err = "garbage after CBOR object"; 1411 dec.err = "garbage after CBOR object";
894 1412
895 if (dec.err) 1413 if (dec.err)
896 { 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
897 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
898 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (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);
899 } 1432 }
900 1433
901 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
902 1435
903 return sv; 1436 return sv;
904} 1437}
905 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
906///////////////////////////////////////////////////////////////////////////// 1563/////////////////////////////////////////////////////////////////////////////
907// XS interface functions 1564// XS interface functions
908 1565
909MODULE = CBOR::XS PACKAGE = CBOR::XS 1566MODULE = CBOR::XS PACKAGE = CBOR::XS
910 1567
918 1575
919 types_true = get_bool ("Types::Serialiser::true" ); 1576 types_true = get_bool ("Types::Serialiser::true" );
920 types_false = get_bool ("Types::Serialiser::false"); 1577 types_false = get_bool ("Types::Serialiser::false");
921 types_error = get_bool ("Types::Serialiser::error"); 1578 types_error = get_bool ("Types::Serialiser::error");
922 1579
1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1581
923 sv_cbor = newSVpv ("CBOR", 0); 1582 sv_cbor = newSVpv ("CBOR", 0);
924 SvREADONLY_on (sv_cbor); 1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
925} 1586}
926 1587
927PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
928 1589
929void CLONE (...) 1590void CLONE (...)
947 1608
948void shrink (CBOR *self, int enable = 1) 1609void shrink (CBOR *self, int enable = 1)
949 ALIAS: 1610 ALIAS:
950 shrink = F_SHRINK 1611 shrink = F_SHRINK
951 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
952 PPCODE: 1620 PPCODE:
953{ 1621{
954 if (enable) 1622 if (enable)
955 self->flags |= ix; 1623 self->flags |= ix;
956 else 1624 else
961 1629
962void get_shrink (CBOR *self) 1630void get_shrink (CBOR *self)
963 ALIAS: 1631 ALIAS:
964 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
965 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
966 PPCODE: 1641 PPCODE:
967 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
968 1643
969void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
970 PPCODE: 1645 PPCODE:
983 XPUSHs (ST (0)); 1658 XPUSHs (ST (0));
984 1659
985int get_max_size (CBOR *self) 1660int get_max_size (CBOR *self)
986 CODE: 1661 CODE:
987 RETVAL = self->max_size; 1662 RETVAL = self->max_size;
1663 OUTPUT:
1664 RETVAL
1665
1666void filter (CBOR *self, SV *filter = 0)
1667 PPCODE:
1668 SvREFCNT_dec (self->filter);
1669 self->filter = filter ? newSVsv (filter) : filter;
1670 XPUSHs (ST (0));
1671
1672SV *get_filter (CBOR *self)
1673 CODE:
1674 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
988 OUTPUT: 1675 OUTPUT:
989 RETVAL 1676 RETVAL
990 1677
991void encode (CBOR *self, SV *scalar) 1678void encode (CBOR *self, SV *scalar)
992 PPCODE: 1679 PPCODE:
1007 EXTEND (SP, 2); 1694 EXTEND (SP, 2);
1008 PUSHs (sv); 1695 PUSHs (sv);
1009 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1696 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1010} 1697}
1011 1698
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}
1750
1751void DESTROY (CBOR *self)
1752 PPCODE:
1753 cbor_free (self);
1754
1012PROTOTYPES: ENABLE 1755PROTOTYPES: ENABLE
1013 1756
1014void encode_cbor (SV *scalar) 1757void encode_cbor (SV *scalar)
1758 ALIAS:
1759 encode_cbor = 0
1760 encode_cbor_sharing = F_ALLOW_SHARING
1015 PPCODE: 1761 PPCODE:
1016{ 1762{
1017 CBOR cbor; 1763 CBOR cbor;
1018 cbor_init (&cbor); 1764 cbor_init (&cbor);
1765 cbor.flags |= ix;
1019 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1766 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1020 XPUSHs (scalar); 1767 XPUSHs (scalar);
1021} 1768}
1022 1769
1023void decode_cbor (SV *cborstr) 1770void decode_cbor (SV *cborstr)
1027 cbor_init (&cbor); 1774 cbor_init (&cbor);
1028 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1029 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1030} 1777}
1031 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