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 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.71 by root, Fri Mar 19 17:30:27 2021 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
24 69
25// known tags 70// known tags
26enum cbor_tag 71enum cbor_tag
27{ 72{
28 // inofficial extensions (pending iana registration) 73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing 77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 81
37 // rfc7049 82 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 85 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 88 CBOR_TAG_BIGFLOAT = 5, // array
44 89
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 90 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 91 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 92 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 94
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 100
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 102};
58 103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_INT = 1,
109 AS_BYTES = 2,
110 AS_TEXT = 3,
111 AS_FLOAT16 = 4,
112 AS_FLOAT32 = 5,
113 AS_FLOAT64 = 6,
114 AS_MAP = 7,
115 // possibly future enhancements: (generic) float, (generic) string
116};
117
59#define F_SHRINK 0x00000001UL 118#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 119#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 120#define F_ALLOW_SHARING 0x00000004UL
121#define F_ALLOW_CYCLES 0x00000008UL
122#define F_FORBID_OBJECTS 0x00000010UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 123#define F_PACK_STRINGS 0x00000020UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 124#define F_TEXT_KEYS 0x00000040UL
125#define F_TEXT_STRINGS 0x00000080UL
126#define F_VALIDATE_UTF8 0x00000100UL
64 127
65#define INIT_SIZE 32 // initial scalar size to be allocated 128#define INIT_SIZE 32 // initial scalar size to be allocated
66 129
67#define SB do { 130#define SB do {
68#define SE } while (0) 131#define SE } while (0)
80# define CBOR_SLOW 0 143# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 144# define CBOR_STASH cbor_stash
82#endif 145#endif
83 146
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 147static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 148static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 149
87typedef struct { 150typedef struct {
88 U32 flags; 151 U32 flags;
89 U32 max_depth; 152 U32 max_depth;
90 STRLEN max_size; 153 STRLEN max_size;
154 SV *filter;
155
156 // for the incremental parser
157 STRLEN incr_pos; // the current offset into the text
158 STRLEN incr_need; // minimum bytes needed to decode
159 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 160} CBOR;
92 161
93ecb_inline void 162ecb_inline void
94cbor_init (CBOR *cbor) 163cbor_init (CBOR *cbor)
95{ 164{
96 Zero (cbor, 1, CBOR); 165 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 166 cbor->max_depth = 512;
167}
168
169ecb_inline void
170cbor_free (CBOR *cbor)
171{
172 SvREFCNT_dec (cbor->filter);
173 SvREFCNT_dec (cbor->incr_count);
98} 174}
99 175
100///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
101// utility functions 177// utility functions
102 178
124 SvPV_renew (sv, SvCUR (sv) + 1); 200 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 201#endif
126 } 202 }
127} 203}
128 204
205// minimum length of a string to be registered for stringref
206ecb_inline STRLEN
207minimum_string_length (UV idx)
208{
209 return idx <= 23 ? 3
210 : idx <= 0xffU ? 4
211 : idx <= 0xffffU ? 5
212 : idx <= 0xffffffffU ? 7
213 : 11;
214}
215
129///////////////////////////////////////////////////////////////////////////// 216/////////////////////////////////////////////////////////////////////////////
130// encoder 217// encoder
131 218
132// structure used for encoding CBOR 219// structure used for encoding CBOR
133typedef struct 220typedef struct
135 char *cur; // SvPVX (sv) + current output position 222 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 223 char *end; // SvEND (sv)
137 SV *sv; // result scalar 224 SV *sv; // result scalar
138 CBOR cbor; 225 CBOR cbor;
139 U32 depth; // recursion level 226 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 227 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
228 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 229 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 230 UV shareable_idx;
143} enc_t; 231} enc_t;
144 232
145ecb_inline void 233ecb_inline void
146need (enc_t *enc, STRLEN len) 234need (enc_t *enc, STRLEN len)
147{ 235{
148 if (ecb_expect_false (enc->cur + len >= enc->end)) 236 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
149 { 237 {
150 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 238 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
151 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 239 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
152 enc->cur = SvPVX (enc->sv) + cur; 240 enc->cur = SvPVX (enc->sv) + cur;
153 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 241 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
154 } 242 }
155} 243}
156 244
245static void encode_sv (enc_t *enc, SV *sv);
246
157ecb_inline void 247ecb_inline void
158encode_ch (enc_t *enc, char ch) 248encode_ch (enc_t *enc, char ch)
159{ 249{
160 need (enc, 1); 250 need (enc, 1);
161 *enc->cur++ = ch; 251 *enc->cur++ = ch;
162} 252}
163 253
254// used for tags, intregers, element counts and so on
164static void 255static void
165encode_uint (enc_t *enc, int major, UV len) 256encode_uint (enc_t *enc, int major, UV len)
166{ 257{
167 need (enc, 9); 258 need (enc, 9);
168 259
169 if (len < 24) 260 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 261 *enc->cur++ = major | len;
171 else if (len <= 0xff) 262 else if (ecb_expect_true (len <= 0xffU))
172 { 263 {
173 *enc->cur++ = major | 24; 264 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 265 *enc->cur++ = len;
175 } 266 }
176 else if (len <= 0xffff) 267 else if (len <= 0xffffU)
177 { 268 {
178 *enc->cur++ = major | 25; 269 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 271 *enc->cur++ = len;
181 } 272 }
182 else if (len <= 0xffffffff) 273 else if (len <= 0xffffffffU)
183 { 274 {
184 *enc->cur++ = major | 26; 275 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 276 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 277 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 278 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 279 *enc->cur++ = len;
189 } 280 }
190 else 281 else
191 { 282 {
192 *enc->cur++ = major | 27; 283 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 284 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 285 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 286 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 287 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 288 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 290 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 291 *enc->cur++ = len;
201 } 292 }
202} 293}
203 294
204static void 295// encodes a perl value into a CBOR integer
296ecb_inline void
297encode_int (enc_t *enc, SV *sv)
298{
299 if (SvIsUV (sv))
300 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
301 else if (SvIVX (sv) >= 0)
302 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
303 else
304 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
305}
306
307ecb_inline void
308encode_tag (enc_t *enc, UV tag)
309{
310 encode_uint (enc, MAJOR_TAG, tag);
311}
312
313// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
314ecb_noinline static void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 315encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 316{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 317 STRLEN ulen = len;
318 U8 *p, *pend = (U8 *)str + len;
319
320 for (p = (U8 *)str; p < pend; ++p)
321 ulen += *p >> 7; // count set high bits
322
323 encode_uint (enc, MAJOR_TEXT, ulen);
324
325 need (enc, ulen);
326 for (p = (U8 *)str; p < pend; ++p)
327 if (*p < 0x80)
328 *enc->cur++ = *p;
329 else
330 {
331 *enc->cur++ = 0xc0 + (*p >> 6);
332 *enc->cur++ = 0x80 + (*p & 63);
333 }
334}
335
336ecb_inline void
337encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
338{
339 if (ecb_expect_false (upgrade_utf8))
340 if (!utf8)
341 {
342 encode_str_utf8 (enc, utf8, str, len);
343 return;
344 }
345
346 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 347 need (enc, len);
209 memcpy (enc->cur, str, len); 348 memcpy (enc->cur, str, len);
210 enc->cur += len; 349 enc->cur += len;
211} 350}
212 351
213ecb_inline void 352ecb_inline void
214encode_tag (enc_t *enc, UV tag) 353encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
215{ 354{
216 encode_uint (enc, 0xc0, tag); 355 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
217} 356 {
357 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
218 358
219static int
220encode_sharable2 (enc_t *enc, SV *sv)
221{
222 if (!enc->sharable)
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ());
224
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 359 if (SvOK (*svp))
228 { 360 {
361 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 362 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 363 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 364 return;
233 } 365 }
234 else 366 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 367 {
368 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 369 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 370 ++enc->stringref_idx;
238 371 }
239 return 0;
240 } 372 }
241}
242 373
374 encode_str (enc, upgrade_utf8, utf8, str, len);
375}
376
243ecb_inline int 377ecb_inline void
378encode_float16 (enc_t *enc, NV nv)
379{
380 need (enc, 1+2);
381
382 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
383
384 uint16_t fp = ecb_float_to_binary16 (nv);
385
386 if (!ecb_big_endian ())
387 fp = ecb_bswap16 (fp);
388
389 memcpy (enc->cur, &fp, 2);
390 enc->cur += 2;
391}
392
393ecb_inline void
394encode_float32 (enc_t *enc, NV nv)
395{
396 need (enc, 1+4);
397
398 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
399
400 uint32_t fp = ecb_float_to_binary32 (nv);
401
402 if (!ecb_big_endian ())
403 fp = ecb_bswap32 (fp);
404
405 memcpy (enc->cur, &fp, 4);
406 enc->cur += 4;
407}
408
409ecb_inline void
410encode_float64 (enc_t *enc, NV nv)
411{
412 need (enc, 1+8);
413
414 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
415
416 uint64_t fp = ecb_double_to_binary64 (nv);
417
418 if (!ecb_big_endian ())
419 fp = ecb_bswap64 (fp);
420
421 memcpy (enc->cur, &fp, 8);
422 enc->cur += 8;
423}
424
425ecb_inline void
426encode_bool (enc_t *enc, int istrue)
427{
428 encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
429}
430
431// encodes an arrayref containing key-value pairs as CBOR map
432ecb_inline void
244encode_sharable (enc_t *enc, SV *sv) 433encode_array_as_map (enc_t *enc, SV *sv)
245{ 434{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) 435 if (enc->depth >= enc->cbor.max_depth)
247 && ecb_expect_false (SvREFCNT (sv) > 1)) 436 croak (ERR_NESTING_EXCEEDED);
248 return encode_sharable2 (enc, sv);
249 437
250 return 0; 438 ++enc->depth;
251}
252 439
253static void encode_sv (enc_t *enc, SV *sv); 440 // as_map does error checking for us, but we re-check in case
441 // things have changed.
442
443 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
444 croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)");
445
446 AV *av = (AV *)SvRV (sv);
447 int i, len = av_len (av);
448
449 if (!(len & 1))
450 croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)");
451
452 encode_uint (enc, MAJOR_MAP, (len + 1) >> 1);
453
454 for (i = 0; i <= len; ++i)
455 {
456 SV **svp = av_fetch (av, i, 0);
457 encode_sv (enc, svp ? *svp : &PL_sv_undef);
458 }
459
460 --enc->depth;
461}
462
463ecb_inline void
464encode_forced (enc_t *enc, UV type, SV *sv)
465{
466 switch (type)
467 {
468 case AS_CBOR:
469 {
470 STRLEN len;
471 char *str = SvPVbyte (sv, len);
472
473 need (enc, len);
474 memcpy (enc->cur, str, len);
475 enc->cur += len;
476 }
477 break;
478
479 case AS_BYTES:
480 {
481 STRLEN len;
482 char *str = SvPVbyte (sv, len);
483 encode_strref (enc, 0, 0, str, len);
484 }
485 break;
486
487 case AS_TEXT:
488 {
489 STRLEN len;
490 char *str = SvPVutf8 (sv, len);
491 encode_strref (enc, 1, 1, str, len);
492 }
493 break;
494
495 case AS_INT: encode_int (enc, sv); break;
496
497 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
498 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
499 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
500
501 case AS_MAP: encode_array_as_map (enc, sv); break;
502
503 default:
504 croak ("encountered malformed CBOR::XS::Tagged object");
505 }
506}
254 507
255static void 508static void
256encode_av (enc_t *enc, AV *av) 509encode_av (enc_t *enc, AV *av)
257{ 510{
258 int i, len = av_len (av); 511 int i, len = av_len (av);
260 if (enc->depth >= enc->cbor.max_depth) 513 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 514 croak (ERR_NESTING_EXCEEDED);
262 515
263 ++enc->depth; 516 ++enc->depth;
264 517
265 encode_uint (enc, 0x80, len + 1); 518 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 519
520 if (ecb_expect_false (SvMAGICAL (av)))
267 for (i = 0; i <= len; ++i) 521 for (i = 0; i <= len; ++i)
268 { 522 {
269 SV **svp = av_fetch (av, i, 0); 523 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 524 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 525 }
526 else
527 for (i = 0; i <= len; ++i)
528 {
529 SV *sv = AvARRAY (av)[i];
530 encode_sv (enc, sv ? sv : &PL_sv_undef);
531 }
272 532
273 --enc->depth; 533 --enc->depth;
274} 534}
275 535
276static void 536static void
284 ++enc->depth; 544 ++enc->depth;
285 545
286 int pairs = hv_iterinit (hv); 546 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 547 int mg = SvMAGICAL (hv);
288 548
289 if (mg) 549 if (ecb_expect_false (mg))
290 encode_ch (enc, 0xa0 | 31); 550 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 551 else
292 encode_uint (enc, 0xa0, pairs); 552 encode_uint (enc, MAJOR_MAP, pairs);
293 553
294 while ((he = hv_iternext (hv))) 554 while ((he = hv_iternext (hv)))
295 { 555 {
296 if (HeKLEN (he) == HEf_SVKEY) 556 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 557 encode_sv (enc, HeSVKEY (he));
298 else 558 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 559 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 560
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 561 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 562 }
303 563
304 if (mg) 564 if (ecb_expect_false (mg))
305 encode_ch (enc, 0xe0 | 31); 565 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 566
307 --enc->depth; 567 --enc->depth;
308} 568}
309 569
310// encode objects, arrays and special \0=false and \1=true values. 570// encode objects, arrays and special \0=false and \1=true values.
311static void 571static void
312encode_rv (enc_t *enc, SV *sv) 572encode_rv (enc_t *enc, SV *sv)
313{ 573{
314 svtype svt;
315
316 SvGETMAGIC (sv); 574 SvGETMAGIC (sv);
575
317 svt = SvTYPE (sv); 576 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 577
322 if (ecb_expect_false (SvOBJECT (sv))) 578 if (ecb_expect_false (SvOBJECT (sv)))
323 { 579 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 580 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 581 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 586 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 587 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 588 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 589
334 HV *stash = SvSTASH (sv); 590 HV *stash = SvSTASH (sv);
335 GV *method;
336 591
337 if (stash == boolean_stash) 592 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 593 {
594 encode_bool (enc, SvIV (sv));
595 return;
596 }
339 else if (stash == error_stash) 597 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 598 {
599 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
600 return;
601 }
341 else if (stash == tagged_stash) 602 else if (stash == tagged_stash)
342 { 603 {
343 if (svt != SVt_PVAV) 604 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 605 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 606
607 switch (av_len ((AV *)sv))
608 {
609 case 2-1:
610 // actually a tagged value
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 611 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 612 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
613 break;
614
615 case 3-1:
616 // a forced type [value, type, undef]
617 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
618 break;
619
620 default:
621 croak ("encountered malformed CBOR::XS::Tagged object");
622 }
623
624 return;
625 }
626 }
627
628 if (ecb_expect_false (SvREFCNT (sv) > 1)
629 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
630 {
631 if (ecb_expect_false (!enc->shareable))
632 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
633
634 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
635
636 if (SvOK (*svp))
348 } 637 {
638 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
639 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
640 return;
641 }
642 else
643 {
644 sv_setuv (*svp, enc->shareable_idx);
645 ++enc->shareable_idx;
646 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
647 }
648 }
649
650 if (ecb_expect_false (SvOBJECT (sv)))
651 {
652 HV *stash = SvSTASH (sv);
653 GV *method;
654
655 if (enc->cbor.flags & F_FORBID_OBJECTS)
656 croak ("encountered object '%s', but forbid_objects is enabled",
657 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 658 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 659 {
351 dSP; 660 dSP;
352 661
353 ENTER; SAVETMPS; PUSHMARK (SP); 662 ENTER; SAVETMPS;
663 PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 664 // we re-bless the reference to get overload and other niceties right
355 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 665 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
356 666
357 PUTBACK; 667 PUTBACK;
358 // G_SCALAR ensures that return value is 1 668 // G_SCALAR ensures that return value is 1
371 } 681 }
372 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 682 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
373 { 683 {
374 dSP; 684 dSP;
375 685
376 ENTER; SAVETMPS; PUSHMARK (SP); 686 ENTER; SAVETMPS;
687 PUSHMARK (SP);
377 EXTEND (SP, 2); 688 EXTEND (SP, 2);
378 // we re-bless the reference to get overload and other niceties right 689 // we re-bless the reference to get overload and other niceties right
379 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 690 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
380 PUSHs (sv_cbor); 691 PUSHs (sv_cbor);
381 692
386 // catch this surprisingly common error 697 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 698 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 699 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 700
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 701 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 702 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 703 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 704
394 while (count) 705 {
706 int i;
707
708 for (i = 0; i < count; ++i)
395 encode_sv (enc, SP[1 - count--]); 709 encode_sv (enc, SP[i + 1 - count]);
710
711 SP -= count;
712 }
396 713
397 PUTBACK; 714 PUTBACK;
398 715
399 FREETMPS; LEAVE; 716 FREETMPS; LEAVE;
400 } 717 }
418{ 735{
419 double nv = SvNVX (sv); 736 double nv = SvNVX (sv);
420 737
421 need (enc, 9); 738 need (enc, 9);
422 739
423 if (ecb_expect_false (nv == (U32)nv)) 740 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 741 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 742 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 743 else if (ecb_expect_false (nv == (float)nv))
427 { 744 encode_float32 (enc, nv);
428 uint32_t fp = ecb_float_to_binary32 (nv);
429
430 *enc->cur++ = 0xe0 | 26;
431
432 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp);
434
435 memcpy (enc->cur, &fp, 4);
436 enc->cur += 4;
437 }
438 else 745 else
439 { 746 encode_float64 (enc, nv);
440 uint64_t fp = ecb_double_to_binary64 (nv);
441
442 *enc->cur++ = 0xe0 | 27;
443
444 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp);
446
447 memcpy (enc->cur, &fp, 8);
448 enc->cur += 8;
449 }
450} 747}
451 748
452static void 749static void
453encode_sv (enc_t *enc, SV *sv) 750encode_sv (enc_t *enc, SV *sv)
454{ 751{
455 SvGETMAGIC (sv); 752 SvGETMAGIC (sv);
456 753
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 754 if (SvPOKp (sv))
461 { 755 {
462 STRLEN len; 756 STRLEN len;
463 char *str = SvPV (sv, len); 757 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 758 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
465 } 759 }
466 else if (SvNOKp (sv)) 760 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 761 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 762 else if (SvIOKp (sv))
469 { 763 encode_int (enc, sv);
470 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv));
472 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv));
474 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
476 }
477 else if (SvROK (sv)) 764 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 765 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 766 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 767 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 768 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 769 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 770 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 771 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 772 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 773}
487 774
488static SV * 775static SV *
489encode_cbor (SV *scalar, CBOR *cbor) 776encode_cbor (SV *scalar, CBOR *cbor)
490{ 777{
491 enc_t enc = { }; 778 enc_t enc = { 0 };
492 779
493 enc.cbor = *cbor; 780 enc.cbor = *cbor;
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 781 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 782 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 783 enc.end = SvEND (enc.sv);
497 784
498 SvPOK_only (enc.sv); 785 SvPOK_only (enc.sv);
786
787 if (cbor->flags & F_PACK_STRINGS)
788 {
789 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
790 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
791 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
792 }
793
499 encode_sv (&enc, scalar); 794 encode_sv (&enc, scalar);
500 795
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 796 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 797 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 798
517 U8 *end; // end of input string 812 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 813 const char *err; // parse error, if != 0
519 CBOR cbor; 814 CBOR cbor;
520 U32 depth; // recursion depth 815 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 816 U32 maxdepth; // recursion depth limit
522 AV *sharable; 817 AV *shareable;
818 AV *stringref;
819 SV *decode_tagged;
820 SV *err_sv; // optional sv for error, needs to be freed
523} dec_t; 821} dec_t;
524 822
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 823// set dec->err to ERRSV
824ecb_cold static void
825err_errsv (dec_t *dec)
826{
827 if (!dec->err)
828 {
829 dec->err_sv = newSVsv (ERRSV);
526 830
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 831 // chop off the trailing \n
832 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
833 *SvEND (dec->err_sv) = 0;
528 834
835 dec->err = SvPVutf8_nolen (dec->err_sv);
836 }
837}
838
839// the following functions are used to reduce code size and help the compiler to optimise
840ecb_cold static void
841err_set (dec_t *dec, const char *reason)
842{
843 if (!dec->err)
844 dec->err = reason;
845}
846
847ecb_cold static void
848err_unexpected_end (dec_t *dec)
849{
850 err_set (dec, "unexpected end of CBOR data");
851}
852
853#define ERR_DO(do) SB do; goto fail; SE
854#define ERR(reason) ERR_DO (err_set (dec, reason))
855#define ERR_ERRSV ERR_DO (err_errsv (dec))
856
857#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
858
529#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 859#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
530#define DEC_DEC_DEPTH --dec->depth 860#define DEC_DEC_DEPTH --dec->depth
531 861
532static UV 862static UV
533decode_uint (dec_t *dec) 863decode_uint (dec_t *dec)
534{ 864{
535 switch (*dec->cur & 31) 865 U8 m = *dec->cur & MINOR_MASK;
536 { 866 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 867
542 case 24: 868 if (ecb_expect_true (m < LENGTH_EXT1))
869 return m;
870 else if (ecb_expect_true (m == LENGTH_EXT1))
871 {
543 WANT (2); 872 WANT (1);
544 dec->cur += 2; 873 dec->cur += 1;
545 return dec->cur[-1]; 874 return dec->cur[-1];
546 875 }
547 case 25: 876 else if (ecb_expect_true (m == LENGTH_EXT2))
877 {
548 WANT (3); 878 WANT (2);
549 dec->cur += 3; 879 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 880 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 881 | ((UV)dec->cur[-1]);
552 882 }
553 case 26: 883 else if (ecb_expect_true (m == LENGTH_EXT4))
884 {
554 WANT (5); 885 WANT (4);
555 dec->cur += 5; 886 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 887 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 888 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 889 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 890 | ((UV)dec->cur[-1]);
560 891 }
561 case 27: 892 else if (ecb_expect_true (m == LENGTH_EXT8))
893 {
562 WANT (9); 894 WANT (8);
563 dec->cur += 9; 895 dec->cur += 8;
896
897 return
898#if UVSIZE < 8
899 0
900#else
564 return (((UV)dec->cur[-8]) << 56) 901 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 902 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 903 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 904 | (((UV)dec->cur[-5]) << 32)
905#endif
568 | (((UV)dec->cur[-4]) << 24) 906 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 907 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 908 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 909 | ((UV)dec->cur[-1]);
572 910 }
573 default: 911 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 912 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 913
577fail: 914fail:
578 return 0; 915 return 0;
579} 916}
580 917
585{ 922{
586 AV *av = newAV (); 923 AV *av = newAV ();
587 924
588 DEC_INC_DEPTH; 925 DEC_INC_DEPTH;
589 926
590 if ((*dec->cur & 31) == 31) 927 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 928 {
592 ++dec->cur; 929 ++dec->cur;
593 930
594 for (;;) 931 for (;;)
595 { 932 {
596 WANT (1); 933 WANT (1);
597 934
598 if (*dec->cur == (0xe0 | 31)) 935 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
599 { 936 {
600 ++dec->cur; 937 ++dec->cur;
601 break; 938 break;
602 } 939 }
603 940
604 av_push (av, decode_sv (dec)); 941 av_push (av, decode_sv (dec));
605 } 942 }
606 } 943 }
607 else 944 else
608 { 945 {
609 int i, len = decode_uint (dec); 946 UV i, len = decode_uint (dec);
610 947
948 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 949 av_fill (av, len - 1);
612 950
613 for (i = 0; i < len; ++i) 951 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 952 AvARRAY (av)[i] = decode_sv (dec);
615 } 953 }
616 954
617 DEC_DEC_DEPTH; 955 DEC_DEC_DEPTH;
618 return newRV_noinc ((SV *)av); 956 return newRV_noinc ((SV *)av);
619 957
620fail: 958fail:
621 SvREFCNT_dec (av); 959 SvREFCNT_dec_NN (av);
622 DEC_DEC_DEPTH; 960 DEC_DEC_DEPTH;
623 return &PL_sv_undef; 961 return &PL_sv_undef;
624} 962}
625 963
626static void 964static void
627decode_he (dec_t *dec, HV *hv) 965decode_he (dec_t *dec, HV *hv)
628{ 966{
629 // for speed reasons, we specialcase single-string 967 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 968 // byte or utf-8 strings as keys, but only when !stringref
631 969
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 970 if (ecb_expect_true (!dec->stringref))
971 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 972 {
634 I32 len = decode_uint (dec); 973 STRLEN len = decode_uint (dec);
635 char *key = (char *)dec->cur; 974 char *key = (char *)dec->cur;
636 975
976 WANT (len);
637 dec->cur += len; 977 dec->cur += len;
638 978
639 hv_store (hv, key, len, decode_sv (dec), 0); 979 hv_store (hv, key, len, decode_sv (dec), 0);
980
981 return;
640 } 982 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 983 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 984 {
643 I32 len = decode_uint (dec); 985 STRLEN len = decode_uint (dec);
644 char *key = (char *)dec->cur; 986 char *key = (char *)dec->cur;
645 987
988 WANT (len);
646 dec->cur += len; 989 dec->cur += len;
647 990
991 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
992 if (!is_utf8_string ((U8 *)key, len))
993 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
994
648 hv_store (hv, key, -len, decode_sv (dec), 0); 995 hv_store (hv, key, -len, decode_sv (dec), 0);
996
997 return;
649 } 998 }
650 else 999
651 {
652 SV *k = decode_sv (dec); 1000 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 1001 SV *v = decode_sv (dec);
654 1002
1003 // we leak memory if uncaught exceptions are thrown by random magical
1004 // methods, and this is hopefully the only place where it can happen,
1005 // so if there is a chance of an exception, take the very slow path.
1006 // since catching exceptions is "undocumented/internal/forbidden" by
1007 // the new p5p powers, we need to call out to a perl function :/
1008 if (ecb_expect_false (SvAMAGIC (k)))
1009 {
1010 dSP;
1011
1012 ENTER; SAVETMPS;
1013 PUSHMARK (SP);
1014 EXTEND (SP, 3);
1015 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
1016 PUSHs (sv_2mortal (k));
1017 PUSHs (sv_2mortal (v));
1018
1019 PUTBACK;
1020 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
1021 SPAGAIN;
1022
1023 FREETMPS; LEAVE;
1024
1025 if (SvTRUE (ERRSV))
1026 ERR_ERRSV;
1027
1028 return;
1029 }
1030
655 hv_store_ent (hv, k, v, 0); 1031 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 1032 SvREFCNT_dec_NN (k);
657 } 1033
1034fail:
1035 ;
658} 1036}
659 1037
660static SV * 1038static SV *
661decode_hv (dec_t *dec) 1039decode_hv (dec_t *dec)
662{ 1040{
663 HV *hv = newHV (); 1041 HV *hv = newHV ();
664 1042
665 DEC_INC_DEPTH; 1043 DEC_INC_DEPTH;
666 1044
667 if ((*dec->cur & 31) == 31) 1045 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 1046 {
669 ++dec->cur; 1047 ++dec->cur;
670 1048
671 for (;;) 1049 for (;;)
672 { 1050 {
673 WANT (1); 1051 WANT (1);
674 1052
675 if (*dec->cur == (0xe0 | 31)) 1053 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
676 { 1054 {
677 ++dec->cur; 1055 ++dec->cur;
678 break; 1056 break;
679 } 1057 }
680 1058
681 decode_he (dec, hv); 1059 decode_he (dec, hv);
682 } 1060 }
683 } 1061 }
684 else 1062 else
685 { 1063 {
686 int pairs = decode_uint (dec); 1064 UV pairs = decode_uint (dec);
1065
1066 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
687 1067
688 while (pairs--) 1068 while (pairs--)
689 decode_he (dec, hv); 1069 decode_he (dec, hv);
690 } 1070 }
691 1071
692 DEC_DEC_DEPTH; 1072 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)hv); 1073 return newRV_noinc ((SV *)hv);
694 1074
695fail: 1075fail:
696 SvREFCNT_dec (hv); 1076 SvREFCNT_dec_NN (hv);
697 DEC_DEC_DEPTH; 1077 DEC_DEC_DEPTH;
698 return &PL_sv_undef; 1078 return &PL_sv_undef;
699} 1079}
700 1080
701static SV * 1081static SV *
702decode_str (dec_t *dec, int utf8) 1082decode_str (dec_t *dec, int utf8)
703{ 1083{
704 SV *sv = 0; 1084 SV *sv = 0;
705 1085
706 if ((*dec->cur & 31) == 31) 1086 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
707 { 1087 {
1088 // indefinite length strings
708 ++dec->cur; 1089 ++dec->cur;
709 1090
1091 U8 major = *dec->cur & MAJOR_MISC;
1092
710 sv = newSVpvn ("", 0); 1093 sv = newSVpvn ("", 0);
711 1094
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 1095 for (;;)
714 { 1096 {
715 WANT (1); 1097 WANT (1);
716 1098
717 if (*dec->cur == (0xe0 | 31)) 1099 if ((*dec->cur - major) > LENGTH_EXT8)
1100 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 1101 {
719 ++dec->cur; 1102 ++dec->cur;
720 break; 1103 break;
721 } 1104 }
1105 else
1106 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 1107
723 sv_catsv (sv, decode_sv (dec)); 1108 STRLEN len = decode_uint (dec);
1109
1110 WANT (len);
1111 sv_catpvn (sv, dec->cur, len);
1112 dec->cur += len;
724 } 1113 }
725 } 1114 }
726 else 1115 else
727 { 1116 {
728 STRLEN len = decode_uint (dec); 1117 STRLEN len = decode_uint (dec);
729 1118
730 WANT (len); 1119 WANT (len);
731 sv = newSVpvn (dec->cur, len); 1120 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 1121 dec->cur += len;
1122
1123 if (ecb_expect_false (dec->stringref)
1124 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1125 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 1126 }
734 1127
735 if (utf8) 1128 if (utf8)
1129 {
1130 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1131 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1132 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1133
736 SvUTF8_on (sv); 1134 SvUTF8_on (sv);
1135 }
737 1136
738 return sv; 1137 return sv;
739 1138
740fail: 1139fail:
741 SvREFCNT_dec (sv); 1140 SvREFCNT_dec (sv);
743} 1142}
744 1143
745static SV * 1144static SV *
746decode_tagged (dec_t *dec) 1145decode_tagged (dec_t *dec)
747{ 1146{
1147 SV *sv = 0;
748 UV tag = decode_uint (dec); 1148 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 1149
1150 WANT (1);
750 1151
751 switch (tag) 1152 switch (tag)
752 { 1153 {
753 case CBOR_TAG_MAGIC: 1154 case CBOR_TAG_MAGIC:
754 return sv; 1155 sv = decode_sv (dec);
1156 break;
755 1157
756 case CBOR_TAG_INDIRECTION: 1158 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 1159 sv = newRV_noinc (decode_sv (dec));
1160 break;
758 1161
1162 case CBOR_TAG_STRINGREF_NAMESPACE:
1163 {
1164 // do not use SAVETMPS/FREETMPS, as these will
1165 // erase mortalised caches, e.g. "shareable"
1166 ENTER;
1167
1168 SAVESPTR (dec->stringref);
1169 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1170
1171 sv = decode_sv (dec);
1172
1173 LEAVE;
1174 }
1175 break;
1176
1177 case CBOR_TAG_STRINGREF:
1178 {
1179 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1180 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1181
1182 UV idx = decode_uint (dec);
1183
1184 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1185 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1186
1187 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1188 }
1189 break;
1190
759 case CBOR_TAG_VALUE_SHARABLE: 1191 case CBOR_TAG_VALUE_SHAREABLE:
1192 {
760 if (ecb_expect_false (!dec->sharable)) 1193 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 1194 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 1195
1196 if (dec->cbor.flags & F_ALLOW_CYCLES)
1197 {
1198 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 1199 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 1200
765 return sv; 1201 SV *osv = decode_sv (dec);
1202 sv_setsv (sv, osv);
1203 SvREFCNT_dec_NN (osv);
1204 }
1205 else
1206 {
1207 av_push (dec->shareable, &PL_sv_undef);
1208 int idx = AvFILLp (dec->shareable);
1209 sv = decode_sv (dec);
1210 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1211 }
1212 }
1213 break;
766 1214
767 case CBOR_TAG_VALUE_SHAREDREF: 1215 case CBOR_TAG_VALUE_SHAREDREF:
768 { 1216 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 1217 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 1218 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 1219
1220 UV idx = decode_uint (dec);
1221
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 1222 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1223 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 1224
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 1225 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1226
1227 if (sv == &PL_sv_undef)
1228 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 1229 }
1230 break;
779 1231
780 case CBOR_TAG_PERL_OBJECT: 1232 case CBOR_TAG_PERL_OBJECT:
781 { 1233 {
1234 if (dec->cbor.flags & F_FORBID_OBJECTS)
1235 goto filter;
1236
1237 sv = decode_sv (dec);
1238
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1239 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 1240 ERR ("corrupted CBOR data (non-array perl object)");
784 1241
785 AV *av = (AV *)SvRV (sv); 1242 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 1243 int len = av_len (av) + 1;
794 if (!method) 1251 if (!method)
795 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1252 ERR ("cannot decode perl-object (package does not have a THAW method)");
796 1253
797 dSP; 1254 dSP;
798 1255
799 ENTER; SAVETMPS; PUSHMARK (SP); 1256 ENTER; SAVETMPS;
1257 PUSHMARK (SP);
800 EXTEND (SP, len + 1); 1258 EXTEND (SP, len + 1);
801 // we re-bless the reference to get overload and other niceties right 1259 // we re-bless the reference to get overload and other niceties right
802 PUSHs (*av_fetch (av, 0, 1)); 1260 PUSHs (*av_fetch (av, 0, 1));
803 PUSHs (sv_cbor); 1261 PUSHs (sv_cbor);
804 1262
812 SPAGAIN; 1270 SPAGAIN;
813 1271
814 if (SvTRUE (ERRSV)) 1272 if (SvTRUE (ERRSV))
815 { 1273 {
816 FREETMPS; LEAVE; 1274 FREETMPS; LEAVE;
817 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1275 ERR_ERRSV;
818 } 1276 }
819 1277
820 SvREFCNT_dec (sv); 1278 SvREFCNT_dec_NN (sv);
821 sv = SvREFCNT_inc (POPs); 1279 sv = SvREFCNT_inc (POPs);
822 1280
823 PUTBACK; 1281 PUTBACK;
824 1282
825 FREETMPS; LEAVE; 1283 FREETMPS; LEAVE;
826
827 return sv;
828 } 1284 }
1285 break;
829 1286
830 default: 1287 default:
1288 filter:
1289 {
1290 SV *tag_sv = newSVuv (tag);
1291
1292 sv = decode_sv (dec);
1293
1294 dSP;
1295 ENTER; SAVETMPS;
1296 PUSHMARK (SP);
1297 EXTEND (SP, 2);
1298 PUSHs (tag_sv);
1299 PUSHs (sv);
1300
1301 PUTBACK;
1302 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1303 SPAGAIN;
1304
1305 if (SvTRUE (ERRSV))
831 { 1306 {
1307 SvREFCNT_dec_NN (tag_sv);
1308 FREETMPS; LEAVE;
1309 ERR_ERRSV;
1310 }
1311
1312 if (count)
1313 {
1314 SvREFCNT_dec_NN (tag_sv);
1315 SvREFCNT_dec_NN (sv);
1316 sv = SvREFCNT_inc_NN (TOPs);
1317 SP -= count;
1318 }
1319 else
1320 {
832 AV *av = newAV (); 1321 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1322 av_push (av, tag_sv);
834 av_push (av, sv); 1323 av_push (av, sv);
835 1324
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1325 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1326 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1327 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1328 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1329 }
1330
1331 PUTBACK;
1332
1333 FREETMPS; LEAVE;
1334 }
1335 break;
842 } 1336 }
1337
1338 return sv;
843 1339
844fail: 1340fail:
845 SvREFCNT_dec (sv); 1341 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1342 return &PL_sv_undef;
847} 1343}
849static SV * 1345static SV *
850decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
851{ 1347{
852 WANT (1); 1348 WANT (1);
853 1349
854 switch (*dec->cur >> 5) 1350 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1351 {
856 case 0: // unsigned int 1352 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1353 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1354 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1355 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1356 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1357 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1358 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1359
864 case 4: // array 1360 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1361 switch (*dec->cur++ & MINOR_MASK)
872 { 1362 {
873 case 20: 1363 case SIMPLE_FALSE:
874#if CBOR_SLOW 1364#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1365 types_false = get_bool ("Types::Serialiser::false");
876#endif 1366#endif
877 return newSVsv (types_false); 1367 return newSVsv (types_false);
878 case 21: 1368 case SIMPLE_TRUE:
879#if CBOR_SLOW 1369#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1370 types_true = get_bool ("Types::Serialiser::true");
881#endif 1371#endif
882 return newSVsv (types_true); 1372 return newSVsv (types_true);
883 case 22: 1373 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1374 return newSVsv (&PL_sv_undef);
885 case 23: 1375 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1376#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1377 types_error = get_bool ("Types::Serialiser::error");
888#endif 1378#endif
889 return newSVsv (types_error); 1379 return newSVsv (types_error);
890 1380
891 case 25: 1381 case MISC_FLOAT16:
892 { 1382 {
893 WANT (2); 1383 WANT (2);
894 1384
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1385 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1386 dec->cur += 2;
897 1387
898 return newSVnv (ecb_binary16_to_float (fp)); 1388 return newSVnv (ecb_binary16_to_float (fp));
899 } 1389 }
900 1390
901 case 26: 1391 case MISC_FLOAT32:
902 { 1392 {
903 uint32_t fp; 1393 uint32_t fp;
904 WANT (4); 1394 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1395 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1396 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1399 fp = ecb_bswap32 (fp);
910 1400
911 return newSVnv (ecb_binary32_to_float (fp)); 1401 return newSVnv (ecb_binary32_to_float (fp));
912 } 1402 }
913 1403
914 case 27: 1404 case MISC_FLOAT64:
915 { 1405 {
916 uint64_t fp; 1406 uint64_t fp;
917 WANT (8); 1407 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1408 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1409 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1412 fp = ecb_bswap64 (fp);
923 1413
924 return newSVnv (ecb_binary64_to_double (fp)); 1414 return newSVnv (ecb_binary64_to_double (fp));
925 } 1415 }
926 1416
927 // 0..19 unassigned 1417 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1418 // 24 reserved + unassigned simple (reserved values are not encodable)
1419 // 28-30 unassigned misc
1420 // 31 break code
929 default: 1421 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1422 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1423 }
932 1424
933 break; 1425 break;
934 } 1426 }
935 1427
938} 1430}
939 1431
940static SV * 1432static SV *
941decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1433decode_cbor (SV *string, CBOR *cbor, char **offset_return)
942{ 1434{
943 dec_t dec = { }; 1435 dec_t dec = { 0 };
944 SV *sv; 1436 SV *sv;
945 STRLEN len; 1437 STRLEN len;
946 char *data = SvPVbyte (string, len); 1438 char *data = SvPVbyte (string, len);
947 1439
948 if (len > cbor->max_size && cbor->max_size) 1440 if (len > cbor->max_size && cbor->max_size)
962 if (dec.cur != dec.end && !dec.err) 1454 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1455 dec.err = "garbage after CBOR object";
964 1456
965 if (dec.err) 1457 if (dec.err)
966 { 1458 {
1459 if (dec.shareable)
1460 {
1461 // need to break cyclic links, which would all be in shareable
1462 int i;
1463 SV **svp;
1464
1465 for (i = av_len (dec.shareable) + 1; i--; )
1466 if ((svp = av_fetch (dec.shareable, i, 0)))
1467 sv_setsv (*svp, &PL_sv_undef);
1468 }
1469
967 SvREFCNT_dec (sv); 1470 SvREFCNT_dec_NN (sv);
1471
1472 if (dec.err_sv)
1473 sv_2mortal (dec.err_sv);
1474
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1475 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1476 }
970 1477
971 sv = sv_2mortal (sv); 1478 sv = sv_2mortal (sv);
972 1479
973 return sv; 1480 return sv;
974} 1481}
975 1482
1483/////////////////////////////////////////////////////////////////////////////
1484// incremental parser
1485
1486#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1487
1488// returns 0 for notyet, 1 for success or error
1489static int
1490incr_parse (CBOR *self, SV *cborstr)
1491{
1492 STRLEN cur;
1493 SvPV (cborstr, cur);
1494
1495 while (ecb_expect_true (self->incr_need <= cur))
1496 {
1497 // table of integer count bytes
1498 static I8 incr_len[MINOR_MASK + 1] = {
1499 0, 0, 0, 0, 0, 0, 0, 0,
1500 0, 0, 0, 0, 0, 0, 0, 0,
1501 0, 0, 0, 0, 0, 0, 0, 0,
1502 1, 2, 4, 8,-1,-1,-1,-2
1503 };
1504
1505 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1506 U8 m = *p & MINOR_MASK;
1507 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1508 I8 ilen = incr_len[m];
1509
1510 self->incr_need = self->incr_pos + 1;
1511
1512 if (ecb_expect_false (ilen < 0))
1513 {
1514 if (m != MINOR_INDEF)
1515 return 1; // error
1516
1517 if (*p == (MAJOR_MISC | MINOR_INDEF))
1518 {
1519 if (count >= 0)
1520 return 1; // error
1521
1522 count = 1;
1523 }
1524 else
1525 {
1526 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1527 count = -1;
1528 }
1529 }
1530 else
1531 {
1532 self->incr_need += ilen;
1533 if (ecb_expect_false (self->incr_need > cur))
1534 return 0;
1535
1536 int major = *p >> MAJOR_SHIFT;
1537
1538 switch (major)
1539 {
1540 case MAJOR_TAG >> MAJOR_SHIFT:
1541 ++count; // tags merely prefix another value
1542 break;
1543
1544 case MAJOR_BYTES >> MAJOR_SHIFT:
1545 case MAJOR_TEXT >> MAJOR_SHIFT:
1546 case MAJOR_ARRAY >> MAJOR_SHIFT:
1547 case MAJOR_MAP >> MAJOR_SHIFT:
1548 {
1549 UV len;
1550
1551 if (ecb_expect_false (ilen))
1552 {
1553 len = 0;
1554
1555 do {
1556 len = (len << 8) | *++p;
1557 } while (--ilen);
1558 }
1559 else
1560 len = m;
1561
1562 switch (major)
1563 {
1564 case MAJOR_BYTES >> MAJOR_SHIFT:
1565 case MAJOR_TEXT >> MAJOR_SHIFT:
1566 self->incr_need += len;
1567 if (ecb_expect_false (self->incr_need > cur))
1568 return 0;
1569
1570 break;
1571
1572 case MAJOR_MAP >> MAJOR_SHIFT:
1573 len <<= 1;
1574 /* FALLTHROUGH */
1575 case MAJOR_ARRAY >> MAJOR_SHIFT:
1576 if (len)
1577 {
1578 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1579 count = len + 1;
1580 }
1581 break;
1582 }
1583 }
1584 }
1585 }
1586
1587 self->incr_pos = self->incr_need;
1588
1589 if (count > 0)
1590 {
1591 while (!--count)
1592 {
1593 if (!AvFILLp (self->incr_count))
1594 return 1; // done
1595
1596 SvREFCNT_dec_NN (av_pop (self->incr_count));
1597 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1598 }
1599
1600 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1601 }
1602 }
1603
1604 return 0;
1605}
1606
1607
976///////////////////////////////////////////////////////////////////////////// 1608/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1609// XS interface functions
978 1610
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1611MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1612
988 1620
989 types_true = get_bool ("Types::Serialiser::true" ); 1621 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1622 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1623 types_error = get_bool ("Types::Serialiser::error");
992 1624
1625 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1626
993 sv_cbor = newSVpv ("CBOR", 0); 1627 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1628 SvREADONLY_on (sv_cbor);
1629
1630 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
995} 1631}
996 1632
997PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
998 1634
999void CLONE (...) 1635void CLONE (...)
1018void shrink (CBOR *self, int enable = 1) 1654void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1655 ALIAS:
1020 shrink = F_SHRINK 1656 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1657 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1658 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1659 allow_cycles = F_ALLOW_CYCLES
1660 forbid_objects = F_FORBID_OBJECTS
1024 dedup_strings = F_DEDUP_STRINGS 1661 pack_strings = F_PACK_STRINGS
1662 text_keys = F_TEXT_KEYS
1663 text_strings = F_TEXT_STRINGS
1664 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1665 PPCODE:
1026{ 1666{
1027 if (enable) 1667 if (enable)
1028 self->flags |= ix; 1668 self->flags |= ix;
1029 else 1669 else
1035void get_shrink (CBOR *self) 1675void get_shrink (CBOR *self)
1036 ALIAS: 1676 ALIAS:
1037 get_shrink = F_SHRINK 1677 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1678 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1679 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1680 get_allow_cycles = F_ALLOW_CYCLES
1681 get_forbid_objects = F_FORBID_OBJECTS
1041 get_dedup_strings = F_DEDUP_STRINGS 1682 get_pack_strings = F_PACK_STRINGS
1683 get_text_keys = F_TEXT_KEYS
1684 get_text_strings = F_TEXT_STRINGS
1685 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1686 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1687 XPUSHs (boolSV (self->flags & ix));
1044 1688
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1689void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1690 PPCODE:
1059 XPUSHs (ST (0)); 1703 XPUSHs (ST (0));
1060 1704
1061int get_max_size (CBOR *self) 1705int get_max_size (CBOR *self)
1062 CODE: 1706 CODE:
1063 RETVAL = self->max_size; 1707 RETVAL = self->max_size;
1708 OUTPUT:
1709 RETVAL
1710
1711void filter (CBOR *self, SV *filter = 0)
1712 PPCODE:
1713 SvREFCNT_dec (self->filter);
1714 self->filter = filter ? newSVsv (filter) : filter;
1715 XPUSHs (ST (0));
1716
1717SV *get_filter (CBOR *self)
1718 CODE:
1719 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1720 OUTPUT:
1065 RETVAL 1721 RETVAL
1066 1722
1067void encode (CBOR *self, SV *scalar) 1723void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1724 PPCODE:
1083 EXTEND (SP, 2); 1739 EXTEND (SP, 2);
1084 PUSHs (sv); 1740 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1741 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1742}
1087 1743
1744void incr_parse (CBOR *self, SV *cborstr)
1745 ALIAS:
1746 incr_parse_multiple = 1
1747 PPCODE:
1748{
1749 if (SvUTF8 (cborstr))
1750 sv_utf8_downgrade (cborstr, 0);
1751
1752 if (!self->incr_count)
1753 {
1754 self->incr_count = newAV ();
1755 self->incr_pos = 0;
1756 self->incr_need = 1;
1757
1758 av_push (self->incr_count, newSViv (1));
1759 }
1760
1761 do
1762 {
1763 if (!incr_parse (self, cborstr))
1764 {
1765 if (self->incr_need > self->max_size && self->max_size)
1766 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1767 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1768
1769 break;
1770 }
1771
1772 SV *sv;
1773 char *offset;
1774
1775 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1776 XPUSHs (sv);
1777
1778 sv_chop (cborstr, offset);
1779
1780 av_clear (self->incr_count);
1781 av_push (self->incr_count, newSViv (1));
1782
1783 self->incr_pos = 0;
1784 self->incr_need = self->incr_pos + 1;
1785 }
1786 while (ix);
1787}
1788
1789void incr_reset (CBOR *self)
1790 CODE:
1791{
1792 SvREFCNT_dec (self->incr_count);
1793 self->incr_count = 0;
1794}
1795
1796void DESTROY (CBOR *self)
1797 PPCODE:
1798 cbor_free (self);
1799
1088PROTOTYPES: ENABLE 1800PROTOTYPES: ENABLE
1089 1801
1090void encode_cbor (SV *scalar) 1802void encode_cbor (SV *scalar)
1803 ALIAS:
1804 encode_cbor = 0
1805 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1806 PPCODE:
1092{ 1807{
1093 CBOR cbor; 1808 CBOR cbor;
1094 cbor_init (&cbor); 1809 cbor_init (&cbor);
1810 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1811 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1812 XPUSHs (scalar);
1097} 1813}
1098 1814
1099void decode_cbor (SV *cborstr) 1815void decode_cbor (SV *cborstr)
1103 cbor_init (&cbor); 1819 cbor_init (&cbor);
1104 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1820 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1105 XPUSHs (cborstr); 1821 XPUSHs (cborstr);
1106} 1822}
1107 1823
1824#ifdef __AFL_COMPILER
1825
1826void
1827afl_init ()
1828 CODE:
1829 __AFL_INIT ();
1830
1831int
1832afl_loop (unsigned int count = 10000)
1833 CODE:
1834 RETVAL = __AFL_LOOP (count);
1835 OUTPUT:
1836 RETVAL
1837
1838#endif
1839

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines