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 1.26 by root, Fri Nov 22 09:40:13 2013 UTC vs.
Revision 1.56 by root, Fri Nov 25 11:33:03 2016 UTC

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 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 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 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
62#define F_ALLOW_STRINGREF 0x00000008UL //TODO 103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
63 108
64#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
65 110
66#define SB do { 111#define SB do {
67#define SE } while (0) 112#define SE } while (0)
79# define CBOR_SLOW 0 124# define CBOR_SLOW 0
80# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
81#endif 126#endif
82 127
83static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
84static SV *types_true, *types_false, *types_error, *sv_cbor; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
85 130
86typedef struct { 131typedef struct {
87 U32 flags; 132 U32 flags;
88 U32 max_depth; 133 U32 max_depth;
89 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
90} CBOR; 141} CBOR;
91 142
92ecb_inline void 143ecb_inline void
93cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
94{ 145{
95 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
96 cbor->max_depth = 512; 147 cbor->max_depth = 512;
148}
149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
97} 155}
98 156
99///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
100// utility functions 158// utility functions
101 159
131{ 189{
132 return idx > 23 190 return idx > 23
133 ? idx > 0xffU 191 ? idx > 0xffU
134 ? idx > 0xffffU 192 ? idx > 0xffffU
135 ? idx > 0xffffffffU 193 ? idx > 0xffffffffU
194 ? 11
136 ? 7 195 : 7
137 : 6
138 : 5 196 : 5
139 : 4 197 : 4
140 : 3; 198 : 3;
141} 199}
142 200
179static void 237static void
180encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
181{ 239{
182 need (enc, 9); 240 need (enc, 9);
183 241
184 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
185 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
186 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
187 { 245 {
188 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
189 *enc->cur++ = len; 247 *enc->cur++ = len;
190 } 248 }
191 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
192 { 250 {
193 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
194 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
195 *enc->cur++ = len; 253 *enc->cur++ = len;
196 } 254 }
197 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
198 { 256 {
199 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
200 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
201 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
202 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
203 *enc->cur++ = len; 261 *enc->cur++ = len;
204 } 262 }
205 else 263 else
206 { 264 {
207 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
208 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
209 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
210 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
211 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
212 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
217} 275}
218 276
219ecb_inline void 277ecb_inline void
220encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
221{ 279{
222 encode_uint (enc, 0xc0, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
223} 281}
224 282
225static void 283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
226encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
227{ 286{
287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
317 need (enc, len);
318 memcpy (enc->cur, str, len);
319 enc->cur += len;
320}
321
322ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{
228 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
229 { 326 {
230 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
231 328
232 if (SvOK (*svp)) 329 if (SvOK (*svp))
233 { 330 {
234 // already registered, use stringref 331 // already registered, use stringref
235 encode_tag (enc, CBOR_TAG_STRINGREF); 332 encode_tag (enc, CBOR_TAG_STRINGREF);
236 encode_uint (enc, 0x00, SvUV (*svp)); 333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
237 return; 334 return;
238 } 335 }
239 else if (len >= minimum_string_length (enc->stringref_idx)) 336 else if (len >= minimum_string_length (enc->stringref_idx))
240 { 337 {
241 // register only 338 // register only
242 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
243 ++enc->stringref_idx; 340 ++enc->stringref_idx;
244 } 341 }
245 } 342 }
246 343
247 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
248 need (enc, len);
249 memcpy (enc->cur, str, len);
250 enc->cur += len;
251} 345}
252 346
253static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
254 348
255static void 349static void
260 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
262 356
263 ++enc->depth; 357 ++enc->depth;
264 358
265 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 360
361 if (SvMAGICAL (av))
267 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
268 { 363 {
269 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
271 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
272 373
273 --enc->depth; 374 --enc->depth;
274} 375}
275 376
276static void 377static void
285 386
286 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
288 389
289 if (mg) 390 if (mg)
290 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 392 else
292 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
293 394
294 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
295 { 396 {
296 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
298 else 399 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 401
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 403 }
303 404
304 if (mg) 405 if (mg)
305 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 407
307 --enc->depth; 408 --enc->depth;
308} 409}
309 410
310// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
311static void 412static void
312encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
313{ 414{
314 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
315
316 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
317 && ecb_expect_false (SvREFCNT (sv) > 1))
318 {
319 if (!enc->shareable)
320 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
321
322 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
323
324 if (SvOK (*svp))
325 {
326 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
327 encode_uint (enc, 0x00, SvUV (*svp));
328 return;
329 }
330 else
331 {
332 sv_setuv (*svp, enc->shareable_idx);
333 ++enc->shareable_idx;
334 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
335 }
336 }
337 416
338 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
339 418
340 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
341 { 420 {
348 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
349 ? cbor_tagged_stash 428 ? cbor_tagged_stash
350 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
351 430
352 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
353 GV *method;
354 432
355 if (stash == boolean_stash) 433 if (stash == boolean_stash)
356 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
357 else if (stash == error_stash) 438 else if (stash == error_stash)
358 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
359 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
360 { 444 {
361 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
362 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
363 447
364 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
365 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
366 } 464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
469 else
470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
367 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
368 { 483 {
369 dSP; 484 dSP;
370 485
371 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
372 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
373 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
374 490
375 PUTBACK; 491 PUTBACK;
376 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
389 } 505 }
390 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
391 { 507 {
392 dSP; 508 dSP;
393 509
394 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
395 EXTEND (SP, 2); 513 EXTEND (SP, 2);
396 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
397 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
398 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
399 517
404 // catch this surprisingly common error 522 // catch this surprisingly common error
405 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
406 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
407 525
408 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
409 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
410 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
411 529
412 while (count) 530 while (count)
413 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
414 532
415 PUTBACK; 533 PUTBACK;
436{ 554{
437 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
438 556
439 need (enc, 9); 557 need (enc, 9);
440 558
441 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
442 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
443 //TODO: maybe I32? 561 //TODO: maybe I32?
444 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
445 { 563 {
446 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
447 565
448 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
449 567
450 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
451 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
452 570
453 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
455 } 573 }
456 else 574 else
457 { 575 {
458 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
459 577
460 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
461 579
462 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
463 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
464 582
465 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
474 592
475 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
476 { 594 {
477 STRLEN len; 595 STRLEN len;
478 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
479 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
480 } 598 }
481 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
482 encode_nv (enc, sv); 600 encode_nv (enc, sv);
483 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
484 { 602 {
485 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
486 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
487 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
488 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
489 else 607 else
490 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
491 } 609 }
492 else if (SvROK (sv)) 610 else if (SvROK (sv))
493 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
494 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
495 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
496 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
497 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
498 else 616 else
499 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
500 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
501} 619}
502 620
503static SV * 621static SV *
504encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
505{ 623{
506 enc_t enc = { }; 624 enc_t enc = { 0 };
507 625
508 enc.cbor = *cbor; 626 enc.cbor = *cbor;
509 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
510 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
511 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
512 630
513 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
514 632
515 if (cbor->flags & F_ALLOW_STRINGREF) 633 if (cbor->flags & F_PACK_STRINGS)
516 { 634 {
517 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
518 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
519 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
520 } 638 }
542 CBOR cbor; 660 CBOR cbor;
543 U32 depth; // recursion depth 661 U32 depth; // recursion depth
544 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
545 AV *shareable; 663 AV *shareable;
546 AV *stringref; 664 AV *stringref;
665 SV *decode_tagged;
547} dec_t; 666} dec_t;
548 667
549#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
550 669
551#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
552 671
553#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
554#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
555 674
556static UV 675static UV
557decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
558{ 677{
559 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
560 { 679 ++dec->cur;
561 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
562 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
563 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
564 return *dec->cur++ & 31;
565 680
566 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
567 WANT (2); 685 WANT (1);
568 dec->cur += 2; 686 dec->cur += 1;
569 return dec->cur[-1]; 687 return dec->cur[-1];
570 688 }
571 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
572 WANT (3); 691 WANT (2);
573 dec->cur += 3; 692 dec->cur += 2;
574 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
575 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
576 695 }
577 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
578 WANT (5); 698 WANT (4);
579 dec->cur += 5; 699 dec->cur += 4;
580 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
581 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
582 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
583 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
584 704 }
585 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
586 WANT (9); 707 WANT (8);
587 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
588 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
589 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
590 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
591 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
592 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
593 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
594 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
595 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
596 723 }
597 default: 724 else
598 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
599 }
600 726
601fail: 727fail:
602 return 0; 728 return 0;
603} 729}
604 730
609{ 735{
610 AV *av = newAV (); 736 AV *av = newAV ();
611 737
612 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
613 739
614 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
615 { 741 {
616 ++dec->cur; 742 ++dec->cur;
617 743
618 for (;;) 744 for (;;)
619 { 745 {
620 WANT (1); 746 WANT (1);
621 747
622 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
623 { 749 {
624 ++dec->cur; 750 ++dec->cur;
625 break; 751 break;
626 } 752 }
627 753
630 } 756 }
631 else 757 else
632 { 758 {
633 int i, len = decode_uint (dec); 759 int i, len = decode_uint (dec);
634 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
635 av_fill (av, len - 1); 762 av_fill (av, len - 1);
636 763
637 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
638 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
639 } 766 }
652{ 779{
653 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
654 // byte or utf-8 strings as keys, but only when !stringref 781 // byte or utf-8 strings as keys, but only when !stringref
655 782
656 if (ecb_expect_true (!dec->stringref)) 783 if (ecb_expect_true (!dec->stringref))
657 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
658 { 785 {
659 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
660 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
661 788
789 WANT (len);
662 dec->cur += len; 790 dec->cur += len;
663 791
664 if (ecb_expect_false (dec->stringref))
665 av_push (dec->stringref, newSVpvn (key, len));
666
667 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
668 793
669 return; 794 return;
670 } 795 }
671 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
672 { 797 {
673 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
674 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
675 800
801 WANT (len);
676 dec->cur += len; 802 dec->cur += len;
677 803
678 if (ecb_expect_false (dec->stringref)) 804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
679 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
680 807
681 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
682 809
683 return; 810 return;
684 } 811 }
686 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
687 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
688 815
689 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
690 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
818
819fail:
820 ;
691} 821}
692 822
693static SV * 823static SV *
694decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
695{ 825{
696 HV *hv = newHV (); 826 HV *hv = newHV ();
697 827
698 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
699 829
700 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
701 { 831 {
702 ++dec->cur; 832 ++dec->cur;
703 833
704 for (;;) 834 for (;;)
705 { 835 {
706 WANT (1); 836 WANT (1);
707 837
708 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
709 { 839 {
710 ++dec->cur; 840 ++dec->cur;
711 break; 841 break;
712 } 842 }
713 843
734static SV * 864static SV *
735decode_str (dec_t *dec, int utf8) 865decode_str (dec_t *dec, int utf8)
736{ 866{
737 SV *sv = 0; 867 SV *sv = 0;
738 868
739 if ((*dec->cur & 31) == 31) 869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
740 { 870 {
871 // indefinite length strings
741 ++dec->cur; 872 ++dec->cur;
742 873
874 U8 major = *dec->cur & MAJOR_MISC;
875
743 sv = newSVpvn ("", 0); 876 sv = newSVpvn ("", 0);
744 877
745 // not very fast, and certainly not robust against illegal input
746 for (;;) 878 for (;;)
747 { 879 {
748 WANT (1); 880 WANT (1);
749 881
750 if (*dec->cur == (0xe0 | 31)) 882 if ((*dec->cur - major) > LENGTH_EXT8)
883 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
751 { 884 {
752 ++dec->cur; 885 ++dec->cur;
753 break; 886 break;
754 } 887 }
888 else
889 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
755 890
756 sv_catsv (sv, decode_sv (dec)); 891 STRLEN len = decode_uint (dec);
892
893 WANT (len);
894 sv_catpvn (sv, dec->cur, len);
895 dec->cur += len;
757 } 896 }
758 } 897 }
759 else 898 else
760 { 899 {
761 STRLEN len = decode_uint (dec); 900 STRLEN len = decode_uint (dec);
768 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 907 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
769 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 908 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 909 }
771 910
772 if (utf8) 911 if (utf8)
912 {
913 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
914 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
915 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
916
773 SvUTF8_on (sv); 917 SvUTF8_on (sv);
918 }
774 919
775 return sv; 920 return sv;
776 921
777fail: 922fail:
778 SvREFCNT_dec (sv); 923 SvREFCNT_dec (sv);
797 sv = newRV_noinc (decode_sv (dec)); 942 sv = newRV_noinc (decode_sv (dec));
798 break; 943 break;
799 944
800 case CBOR_TAG_STRINGREF_NAMESPACE: 945 case CBOR_TAG_STRINGREF_NAMESPACE:
801 { 946 {
947 // do nmot use SAVETMPS/FREETMPS, as these will
948 // erase mortalised caches, e.g. "shareable"
802 ENTER; SAVETMPS; 949 ENTER;
803 950
804 SAVESPTR (dec->stringref); 951 SAVESPTR (dec->stringref);
805 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 952 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
806 953
807 sv = decode_sv (dec); 954 sv = decode_sv (dec);
808 955
809 FREETMPS; LEAVE; 956 LEAVE;
810 } 957 }
811 break; 958 break;
812 959
813 case CBOR_TAG_STRINGREF: 960 case CBOR_TAG_STRINGREF:
814 { 961 {
815 if ((*dec->cur >> 5) != 0) 962 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
816 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 963 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
817 964
818 UV idx = decode_uint (dec); 965 UV idx = decode_uint (dec);
819 966
820 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 967 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
827 case CBOR_TAG_VALUE_SHAREABLE: 974 case CBOR_TAG_VALUE_SHAREABLE:
828 { 975 {
829 if (ecb_expect_false (!dec->shareable)) 976 if (ecb_expect_false (!dec->shareable))
830 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 977 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
831 978
979 if (dec->cbor.flags & F_ALLOW_CYCLES)
980 {
832 sv = newSV (0); 981 sv = newSV (0);
833 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 982 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
834 983
835 SV *osv = decode_sv (dec); 984 SV *osv = decode_sv (dec);
836 sv_setsv (sv, osv); 985 sv_setsv (sv, osv);
837 SvREFCNT_dec_NN (osv); 986 SvREFCNT_dec_NN (osv);
987 }
988 else
989 {
990 av_push (dec->shareable, &PL_sv_undef);
991 int idx = AvFILLp (dec->shareable);
992 sv = decode_sv (dec);
993 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
994 }
838 } 995 }
839 break; 996 break;
840 997
841 case CBOR_TAG_VALUE_SHAREDREF: 998 case CBOR_TAG_VALUE_SHAREDREF:
842 { 999 {
843 if ((*dec->cur >> 5) != 0) 1000 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
844 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1001 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
845 1002
846 UV idx = decode_uint (dec); 1003 UV idx = decode_uint (dec);
847 1004
848 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1005 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
849 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1006 ERR ("corrupted CBOR data (sharedref index out of bounds)");
850 1007
851 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1008 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1009
1010 if (sv == &PL_sv_undef)
1011 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
852 } 1012 }
853 break; 1013 break;
854 1014
855 case CBOR_TAG_PERL_OBJECT: 1015 case CBOR_TAG_PERL_OBJECT:
856 { 1016 {
871 if (!method) 1031 if (!method)
872 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1032 ERR ("cannot decode perl-object (package does not have a THAW method)");
873 1033
874 dSP; 1034 dSP;
875 1035
876 ENTER; SAVETMPS; PUSHMARK (SP); 1036 ENTER; SAVETMPS;
1037 PUSHMARK (SP);
877 EXTEND (SP, len + 1); 1038 EXTEND (SP, len + 1);
878 // we re-bless the reference to get overload and other niceties right 1039 // we re-bless the reference to get overload and other niceties right
879 PUSHs (*av_fetch (av, 0, 1)); 1040 PUSHs (*av_fetch (av, 0, 1));
880 PUSHs (sv_cbor); 1041 PUSHs (sv_cbor);
881 1042
905 1066
906 default: 1067 default:
907 { 1068 {
908 sv = decode_sv (dec); 1069 sv = decode_sv (dec);
909 1070
1071 dSP;
1072 ENTER; SAVETMPS;
1073 SAVESTACK_POS ();
1074 PUSHMARK (SP);
1075 EXTEND (SP, 2);
1076 PUSHs (newSVuv (tag));
1077 PUSHs (sv);
1078
1079 PUTBACK;
1080 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1081 SPAGAIN;
1082
1083 if (SvTRUE (ERRSV))
1084 {
1085 FREETMPS; LEAVE;
1086 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1087 }
1088
1089 if (count)
1090 {
1091 SvREFCNT_dec (sv);
1092 sv = SvREFCNT_inc (POPs);
1093 }
1094 else
1095 {
910 AV *av = newAV (); 1096 AV *av = newAV ();
911 av_push (av, newSVuv (tag)); 1097 av_push (av, newSVuv (tag));
912 av_push (av, sv); 1098 av_push (av, sv);
913 1099
914 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1100 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
915 ? cbor_tagged_stash 1101 ? cbor_tagged_stash
916 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1102 : gv_stashpv ("CBOR::XS::Tagged" , 1);
917
918 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1103 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1104 }
1105
1106 PUTBACK;
1107
1108 FREETMPS; LEAVE;
919 } 1109 }
920 break; 1110 break;
921 } 1111 }
922 1112
923 return sv; 1113 return sv;
930static SV * 1120static SV *
931decode_sv (dec_t *dec) 1121decode_sv (dec_t *dec)
932{ 1122{
933 WANT (1); 1123 WANT (1);
934 1124
935 switch (*dec->cur >> 5) 1125 switch (*dec->cur >> MAJOR_SHIFT)
936 { 1126 {
937 case 0: // unsigned int 1127 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
938 return newSVuv (decode_uint (dec)); 1128 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
939 case 1: // negative int 1129 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
940 return newSViv (-1 - (IV)decode_uint (dec)); 1130 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
941 case 2: // octet string 1131 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
942 return decode_str (dec, 0); 1132 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
943 case 3: // utf-8 string 1133 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
944 return decode_str (dec, 1); 1134
945 case 4: // array 1135 case MAJOR_MISC >> MAJOR_SHIFT:
946 return decode_av (dec);
947 case 5: // map
948 return decode_hv (dec);
949 case 6: // tag
950 return decode_tagged (dec);
951 case 7: // misc
952 switch (*dec->cur++ & 31) 1136 switch (*dec->cur++ & MINOR_MASK)
953 { 1137 {
954 case 20: 1138 case SIMPLE_FALSE:
955#if CBOR_SLOW 1139#if CBOR_SLOW
956 types_false = get_bool ("Types::Serialiser::false"); 1140 types_false = get_bool ("Types::Serialiser::false");
957#endif 1141#endif
958 return newSVsv (types_false); 1142 return newSVsv (types_false);
959 case 21: 1143 case SIMPLE_TRUE:
960#if CBOR_SLOW 1144#if CBOR_SLOW
961 types_true = get_bool ("Types::Serialiser::true"); 1145 types_true = get_bool ("Types::Serialiser::true");
962#endif 1146#endif
963 return newSVsv (types_true); 1147 return newSVsv (types_true);
964 case 22: 1148 case SIMPLE_NULL:
965 return newSVsv (&PL_sv_undef); 1149 return newSVsv (&PL_sv_undef);
966 case 23: 1150 case SIMPLE_UNDEF:
967#if CBOR_SLOW 1151#if CBOR_SLOW
968 types_error = get_bool ("Types::Serialiser::error"); 1152 types_error = get_bool ("Types::Serialiser::error");
969#endif 1153#endif
970 return newSVsv (types_error); 1154 return newSVsv (types_error);
971 1155
972 case 25: 1156 case MISC_FLOAT16:
973 { 1157 {
974 WANT (2); 1158 WANT (2);
975 1159
976 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1160 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
977 dec->cur += 2; 1161 dec->cur += 2;
978 1162
979 return newSVnv (ecb_binary16_to_float (fp)); 1163 return newSVnv (ecb_binary16_to_float (fp));
980 } 1164 }
981 1165
982 case 26: 1166 case MISC_FLOAT32:
983 { 1167 {
984 uint32_t fp; 1168 uint32_t fp;
985 WANT (4); 1169 WANT (4);
986 memcpy (&fp, dec->cur, 4); 1170 memcpy (&fp, dec->cur, 4);
987 dec->cur += 4; 1171 dec->cur += 4;
990 fp = ecb_bswap32 (fp); 1174 fp = ecb_bswap32 (fp);
991 1175
992 return newSVnv (ecb_binary32_to_float (fp)); 1176 return newSVnv (ecb_binary32_to_float (fp));
993 } 1177 }
994 1178
995 case 27: 1179 case MISC_FLOAT64:
996 { 1180 {
997 uint64_t fp; 1181 uint64_t fp;
998 WANT (8); 1182 WANT (8);
999 memcpy (&fp, dec->cur, 8); 1183 memcpy (&fp, dec->cur, 8);
1000 dec->cur += 8; 1184 dec->cur += 8;
1003 fp = ecb_bswap64 (fp); 1187 fp = ecb_bswap64 (fp);
1004 1188
1005 return newSVnv (ecb_binary64_to_double (fp)); 1189 return newSVnv (ecb_binary64_to_double (fp));
1006 } 1190 }
1007 1191
1008 // 0..19 unassigned 1192 // 0..19 unassigned simple
1009 // 24 reserved + unassigned (reserved values are not encodable) 1193 // 24 reserved + unassigned simple (reserved values are not encodable)
1194 // 28-30 unassigned misc
1195 // 31 break code
1010 default: 1196 default:
1011 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1197 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1012 } 1198 }
1013 1199
1014 break; 1200 break;
1015 } 1201 }
1016 1202
1019} 1205}
1020 1206
1021static SV * 1207static SV *
1022decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1208decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1023{ 1209{
1024 dec_t dec = { }; 1210 dec_t dec = { 0 };
1025 SV *sv; 1211 SV *sv;
1026 STRLEN len; 1212 STRLEN len;
1027 char *data = SvPVbyte (string, len); 1213 char *data = SvPVbyte (string, len);
1028 1214
1029 if (len > cbor->max_size && cbor->max_size) 1215 if (len > cbor->max_size && cbor->max_size)
1043 if (dec.cur != dec.end && !dec.err) 1229 if (dec.cur != dec.end && !dec.err)
1044 dec.err = "garbage after CBOR object"; 1230 dec.err = "garbage after CBOR object";
1045 1231
1046 if (dec.err) 1232 if (dec.err)
1047 { 1233 {
1234 if (dec.shareable)
1235 {
1236 // need to break cyclic links, which whould all be in shareable
1237 int i;
1238 SV **svp;
1239
1240 for (i = av_len (dec.shareable) + 1; i--; )
1241 if ((svp = av_fetch (dec.shareable, i, 0)))
1242 sv_setsv (*svp, &PL_sv_undef);
1243 }
1244
1048 SvREFCNT_dec (sv); 1245 SvREFCNT_dec (sv);
1049 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1246 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1050 } 1247 }
1051 1248
1052 sv = sv_2mortal (sv); 1249 sv = sv_2mortal (sv);
1053 1250
1054 return sv; 1251 return sv;
1055} 1252}
1056 1253
1254/////////////////////////////////////////////////////////////////////////////
1255// incremental parser
1256
1257#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1258
1259// returns 0 for notyet, 1 for success or error
1260static int
1261incr_parse (CBOR *self, SV *cborstr)
1262{
1263 STRLEN cur;
1264 SvPV (cborstr, cur);
1265
1266 while (ecb_expect_true (self->incr_need <= cur))
1267 {
1268 // table of integer count bytes
1269 static I8 incr_len[MINOR_MASK + 1] = {
1270 0, 0, 0, 0, 0, 0, 0, 0,
1271 0, 0, 0, 0, 0, 0, 0, 0,
1272 0, 0, 0, 0, 0, 0, 0, 0,
1273 1, 2, 4, 8,-1,-1,-1,-2
1274 };
1275
1276 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1277 U8 m = *p & MINOR_MASK;
1278 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1279 I8 ilen = incr_len[m];
1280
1281 self->incr_need = self->incr_pos + 1;
1282
1283 if (ecb_expect_false (ilen < 0))
1284 {
1285 if (m != MINOR_INDEF)
1286 return 1; // error
1287
1288 if (*p == (MAJOR_MISC | MINOR_INDEF))
1289 {
1290 if (count >= 0)
1291 return 1; // error
1292
1293 count = 1;
1294 }
1295 else
1296 {
1297 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1298 count = -1;
1299 }
1300 }
1301 else
1302 {
1303 self->incr_need += ilen;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 int major = *p >> MAJOR_SHIFT;
1308
1309 switch (major)
1310 {
1311 case MAJOR_TAG >> MAJOR_SHIFT:
1312 ++count; // tags merely prefix another value
1313 break;
1314
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 case MAJOR_ARRAY >> MAJOR_SHIFT:
1318 case MAJOR_MAP >> MAJOR_SHIFT:
1319 {
1320 UV len;
1321
1322 if (ecb_expect_false (ilen))
1323 {
1324 len = 0;
1325
1326 do {
1327 len = (len << 8) | *++p;
1328 } while (--ilen);
1329 }
1330 else
1331 len = m;
1332
1333 switch (major)
1334 {
1335 case MAJOR_BYTES >> MAJOR_SHIFT:
1336 case MAJOR_TEXT >> MAJOR_SHIFT:
1337 self->incr_need += len;
1338 if (ecb_expect_false (self->incr_need > cur))
1339 return 0;
1340
1341 break;
1342
1343 case MAJOR_MAP >> MAJOR_SHIFT:
1344 len <<= 1;
1345 case MAJOR_ARRAY >> MAJOR_SHIFT:
1346 if (len)
1347 {
1348 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1349 count = len + 1;
1350 }
1351 break;
1352 }
1353 }
1354 }
1355 }
1356
1357 self->incr_pos = self->incr_need;
1358
1359 if (count > 0)
1360 {
1361 while (!--count)
1362 {
1363 if (!AvFILLp (self->incr_count))
1364 return 1; // done
1365
1366 SvREFCNT_dec_NN (av_pop (self->incr_count));
1367 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1368 }
1369
1370 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1371 }
1372 }
1373
1374 return 0;
1375}
1376
1377
1057///////////////////////////////////////////////////////////////////////////// 1378/////////////////////////////////////////////////////////////////////////////
1058// XS interface functions 1379// XS interface functions
1059 1380
1060MODULE = CBOR::XS PACKAGE = CBOR::XS 1381MODULE = CBOR::XS PACKAGE = CBOR::XS
1061 1382
1069 1390
1070 types_true = get_bool ("Types::Serialiser::true" ); 1391 types_true = get_bool ("Types::Serialiser::true" );
1071 types_false = get_bool ("Types::Serialiser::false"); 1392 types_false = get_bool ("Types::Serialiser::false");
1072 types_error = get_bool ("Types::Serialiser::error"); 1393 types_error = get_bool ("Types::Serialiser::error");
1073 1394
1395 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1396
1074 sv_cbor = newSVpv ("CBOR", 0); 1397 sv_cbor = newSVpv ("CBOR", 0);
1075 SvREADONLY_on (sv_cbor); 1398 SvREADONLY_on (sv_cbor);
1399
1400 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1076} 1401}
1077 1402
1078PROTOTYPES: DISABLE 1403PROTOTYPES: DISABLE
1079 1404
1080void CLONE (...) 1405void CLONE (...)
1099void shrink (CBOR *self, int enable = 1) 1424void shrink (CBOR *self, int enable = 1)
1100 ALIAS: 1425 ALIAS:
1101 shrink = F_SHRINK 1426 shrink = F_SHRINK
1102 allow_unknown = F_ALLOW_UNKNOWN 1427 allow_unknown = F_ALLOW_UNKNOWN
1103 allow_sharing = F_ALLOW_SHARING 1428 allow_sharing = F_ALLOW_SHARING
1104 allow_stringref = F_ALLOW_STRINGREF 1429 allow_cycles = F_ALLOW_CYCLES
1430 pack_strings = F_PACK_STRINGS
1431 text_keys = F_TEXT_KEYS
1432 text_strings = F_TEXT_STRINGS
1433 validate_utf8 = F_VALIDATE_UTF8
1105 PPCODE: 1434 PPCODE:
1106{ 1435{
1107 if (enable) 1436 if (enable)
1108 self->flags |= ix; 1437 self->flags |= ix;
1109 else 1438 else
1115void get_shrink (CBOR *self) 1444void get_shrink (CBOR *self)
1116 ALIAS: 1445 ALIAS:
1117 get_shrink = F_SHRINK 1446 get_shrink = F_SHRINK
1118 get_allow_unknown = F_ALLOW_UNKNOWN 1447 get_allow_unknown = F_ALLOW_UNKNOWN
1119 get_allow_sharing = F_ALLOW_SHARING 1448 get_allow_sharing = F_ALLOW_SHARING
1120 get_allow_stringref = F_ALLOW_STRINGREF 1449 get_allow_cycles = F_ALLOW_CYCLES
1450 get_pack_strings = F_PACK_STRINGS
1451 get_text_keys = F_TEXT_KEYS
1452 get_text_strings = F_TEXT_STRINGS
1453 get_validate_utf8 = F_VALIDATE_UTF8
1121 PPCODE: 1454 PPCODE:
1122 XPUSHs (boolSV (self->flags & ix)); 1455 XPUSHs (boolSV (self->flags & ix));
1123 1456
1124void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1457void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1125 PPCODE: 1458 PPCODE:
1138 XPUSHs (ST (0)); 1471 XPUSHs (ST (0));
1139 1472
1140int get_max_size (CBOR *self) 1473int get_max_size (CBOR *self)
1141 CODE: 1474 CODE:
1142 RETVAL = self->max_size; 1475 RETVAL = self->max_size;
1476 OUTPUT:
1477 RETVAL
1478
1479void filter (CBOR *self, SV *filter = 0)
1480 PPCODE:
1481 SvREFCNT_dec (self->filter);
1482 self->filter = filter ? newSVsv (filter) : filter;
1483 XPUSHs (ST (0));
1484
1485SV *get_filter (CBOR *self)
1486 CODE:
1487 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1143 OUTPUT: 1488 OUTPUT:
1144 RETVAL 1489 RETVAL
1145 1490
1146void encode (CBOR *self, SV *scalar) 1491void encode (CBOR *self, SV *scalar)
1147 PPCODE: 1492 PPCODE:
1162 EXTEND (SP, 2); 1507 EXTEND (SP, 2);
1163 PUSHs (sv); 1508 PUSHs (sv);
1164 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1509 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1165} 1510}
1166 1511
1512void incr_parse (CBOR *self, SV *cborstr)
1513 ALIAS:
1514 incr_parse_multiple = 1
1515 PPCODE:
1516{
1517 if (SvUTF8 (cborstr))
1518 sv_utf8_downgrade (cborstr, 0);
1519
1520 if (!self->incr_count)
1521 {
1522 self->incr_count = newAV ();
1523 self->incr_pos = 0;
1524 self->incr_need = 1;
1525
1526 av_push (self->incr_count, newSViv (1));
1527 }
1528
1529 do
1530 {
1531 if (!incr_parse (self, cborstr))
1532 {
1533 if (self->incr_need > self->max_size && self->max_size)
1534 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1535 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1536
1537 break;
1538 }
1539
1540 SV *sv;
1541 char *offset;
1542
1543 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1544 XPUSHs (sv);
1545
1546 sv_chop (cborstr, offset);
1547
1548 av_clear (self->incr_count);
1549 av_push (self->incr_count, newSViv (1));
1550
1551 self->incr_pos = 0;
1552 self->incr_need = self->incr_pos + 1;
1553 }
1554 while (ix);
1555}
1556
1557void incr_reset (CBOR *self)
1558 CODE:
1559{
1560 SvREFCNT_dec (self->incr_count);
1561 self->incr_count = 0;
1562}
1563
1564void DESTROY (CBOR *self)
1565 PPCODE:
1566 cbor_free (self);
1567
1167PROTOTYPES: ENABLE 1568PROTOTYPES: ENABLE
1168 1569
1169void encode_cbor (SV *scalar) 1570void encode_cbor (SV *scalar)
1571 ALIAS:
1572 encode_cbor = 0
1573 encode_cbor_sharing = F_ALLOW_SHARING
1170 PPCODE: 1574 PPCODE:
1171{ 1575{
1172 CBOR cbor; 1576 CBOR cbor;
1173 cbor_init (&cbor); 1577 cbor_init (&cbor);
1578 cbor.flags |= ix;
1174 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1579 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1175 XPUSHs (scalar); 1580 XPUSHs (scalar);
1176} 1581}
1177 1582
1178void decode_cbor (SV *cborstr) 1583void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines