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 1.29 by root, Thu Nov 28 09:13:12 2013 UTC vs.
Revision 1.58 by root, Fri Nov 25 13:27:29 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))
23#endif 24#endif
24#ifndef SvREFCNT_dec_NN 25#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 27#endif
27 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};
65
28// known tags 66// known tags
29enum cbor_tag 67enum cbor_tag
30{ 68{
31 // extensions 69 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 77
40 // rfc7049 78 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
47 85
48 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52 90
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 96
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
60}; 98};
61 99
62#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
63#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
65#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
66 108
67#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
68 110
69#define SB do { 111#define SB do {
70#define SE } while (0) 112#define SE } while (0)
89typedef struct { 131typedef struct {
90 U32 flags; 132 U32 flags;
91 U32 max_depth; 133 U32 max_depth;
92 STRLEN max_size; 134 STRLEN max_size;
93 SV *filter; 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.
94} CBOR; 141} CBOR;
95 142
96ecb_inline void 143ecb_inline void
97cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
98{ 145{
102 149
103ecb_inline void 150ecb_inline void
104cbor_free (CBOR *cbor) 151cbor_free (CBOR *cbor)
105{ 152{
106 SvREFCNT_dec (cbor->filter); 153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
107} 155}
108 156
109///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
110// utility functions 158// utility functions
111 159
189static void 237static void
190encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
191{ 239{
192 need (enc, 9); 240 need (enc, 9);
193 241
194 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
196 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
197 { 245 {
198 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 247 *enc->cur++ = len;
200 } 248 }
201 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
202 { 250 {
203 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 253 *enc->cur++ = len;
206 } 254 }
207 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
208 { 256 {
209 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 261 *enc->cur++ = len;
214 } 262 }
215 else 263 else
216 { 264 {
217 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
227} 275}
228 276
229ecb_inline void 277ecb_inline void
230encode_tag (enc_t *enc, UV tag) 278encode_tag (enc_t *enc, UV tag)
231{ 279{
232 encode_uint (enc, 0xc0, tag); 280 encode_uint (enc, MAJOR_TAG, tag);
233} 281}
234 282
235static void 283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 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{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 326 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 328
242 if (SvOK (*svp)) 329 if (SvOK (*svp))
243 { 330 {
244 // already registered, use stringref 331 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 332 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 334 return;
248 } 335 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 336 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 337 {
251 // register only 338 // register only
252 sv_setuv (*svp, enc->stringref_idx); 339 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 340 ++enc->stringref_idx;
254 } 341 }
255 } 342 }
256 343
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 344 encode_str (enc, upgrade_utf8, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 345}
262 346
263static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
264 348
265static void 349static void
270 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
272 356
273 ++enc->depth; 357 ++enc->depth;
274 358
275 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 360
361 if (SvMAGICAL (av))
277 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
278 { 363 {
279 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
281 } 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 }
282 373
283 --enc->depth; 374 --enc->depth;
284} 375}
285 376
286static void 377static void
295 386
296 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
298 389
299 if (mg) 390 if (mg)
300 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 392 else
302 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
303 394
304 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
305 { 396 {
306 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
308 else 399 else
309 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));
310 401
311 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));
312 } 403 }
313 404
314 if (mg) 405 if (mg)
315 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 407
317 --enc->depth; 408 --enc->depth;
318} 409}
319 410
320// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
321static void 412static void
322encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
323{ 414{
324 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
325
326 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327 && ecb_expect_false (SvREFCNT (sv) > 1))
328 {
329 if (!enc->shareable)
330 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334 if (SvOK (*svp))
335 {
336 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337 encode_uint (enc, 0x00, SvUV (*svp));
338 return;
339 }
340 else
341 {
342 sv_setuv (*svp, enc->shareable_idx);
343 ++enc->shareable_idx;
344 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345 }
346 }
347 416
348 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
349 418
350 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
351 { 420 {
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 428 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 430
362 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
363 GV *method;
364 432
365 if (stash == boolean_stash) 433 if (stash == boolean_stash)
366 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 }
367 else if (stash == error_stash) 438 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
369 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
370 { 444 {
371 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 447
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 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))
376 } 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
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 483 {
379 dSP; 484 dSP;
380 485
381 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
382 // 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
383 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384 490
385 PUTBACK; 491 PUTBACK;
386 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
399 } 505 }
400 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401 { 507 {
402 dSP; 508 dSP;
403 509
404 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
405 EXTEND (SP, 2); 513 EXTEND (SP, 2);
406 // 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
407 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
409 517
414 // catch this surprisingly common error 522 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416 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));
417 525
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 529
422 while (count) 530 while (count)
423 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
424 532
425 PUTBACK; 533 PUTBACK;
446{ 554{
447 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
448 556
449 need (enc, 9); 557 need (enc, 9);
450 558
451 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 561 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
455 { 563 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
457 565
458 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 567
460 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
462 570
463 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
465 } 573 }
466 else 574 else
467 { 575 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
469 577
470 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 579
472 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
474 582
475 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
484 592
485 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
486 { 594 {
487 STRLEN len; 595 STRLEN len;
488 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490 } 598 }
491 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 600 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
494 { 602 {
495 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 607 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 609 }
502 else if (SvROK (sv)) 610 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 616 else
509 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",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 619}
512 620
513static SV * 621static SV *
514encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
515{ 623{
516 enc_t enc = { }; 624 enc_t enc = { 0 };
517 625
518 enc.cbor = *cbor; 626 enc.cbor = *cbor;
519 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
522 630
523 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
524 632
525 if (cbor->flags & F_ALLOW_STRINGREF) 633 if (cbor->flags & F_PACK_STRINGS)
526 { 634 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 638 }
557 SV *decode_tagged; 665 SV *decode_tagged;
558} dec_t; 666} dec_t;
559 667
560#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
561 669
562#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 ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data")
563 671
564#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)
565#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
566 674
567static UV 675static UV
568decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
569{ 677{
570 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
571 { 679 ++dec->cur;
572 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575 return *dec->cur++ & 31;
576 680
577 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
578 WANT (2); 685 WANT (1);
579 dec->cur += 2; 686 dec->cur += 1;
580 return dec->cur[-1]; 687 return dec->cur[-1];
581 688 }
582 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
583 WANT (3); 691 WANT (2);
584 dec->cur += 3; 692 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
587 695 }
588 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
589 WANT (5); 698 WANT (4);
590 dec->cur += 5; 699 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
595 704 }
596 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
597 WANT (9); 707 WANT (8);
598 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
599 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
603 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
607 723 }
608 default: 724 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 726
612fail: 727fail:
613 return 0; 728 return 0;
614} 729}
615 730
620{ 735{
621 AV *av = newAV (); 736 AV *av = newAV ();
622 737
623 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
624 739
625 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 741 {
627 ++dec->cur; 742 ++dec->cur;
628 743
629 for (;;) 744 for (;;)
630 { 745 {
631 WANT (1); 746 WANT (1);
632 747
633 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 749 {
635 ++dec->cur; 750 ++dec->cur;
636 break; 751 break;
637 } 752 }
638 753
639 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
640 } 755 }
641 } 756 }
642 else 757 else
643 { 758 {
644 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
645 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646 av_fill (av, len - 1); 762 av_fill (av, len - 1);
647 763
648 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
650 } 766 }
663{ 779{
664 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 781 // byte or utf-8 strings as keys, but only when !stringref
666 782
667 if (ecb_expect_true (!dec->stringref)) 783 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 785 {
670 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
671 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
672 788
789 WANT (len);
673 dec->cur += len; 790 dec->cur += len;
674 791
675 if (ecb_expect_false (dec->stringref))
676 av_push (dec->stringref, newSVpvn (key, len));
677
678 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
679 793
680 return; 794 return;
681 } 795 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 797 {
684 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
685 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
686 800
801 WANT (len);
687 dec->cur += len; 802 dec->cur += len;
688 803
689 if (ecb_expect_false (dec->stringref)) 804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 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)");
691 807
692 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
693 809
694 return; 810 return;
695 } 811 }
697 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
699 815
700 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
818
819fail:
820 ;
702} 821}
703 822
704static SV * 823static SV *
705decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
706{ 825{
707 HV *hv = newHV (); 826 HV *hv = newHV ();
708 827
709 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
710 829
711 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 831 {
713 ++dec->cur; 832 ++dec->cur;
714 833
715 for (;;) 834 for (;;)
716 { 835 {
717 WANT (1); 836 WANT (1);
718 837
719 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 839 {
721 ++dec->cur; 840 ++dec->cur;
722 break; 841 break;
723 } 842 }
724 843
725 decode_he (dec, hv); 844 decode_he (dec, hv);
726 } 845 }
727 } 846 }
728 else 847 else
729 { 848 {
730 int pairs = decode_uint (dec); 849 UV pairs = decode_uint (dec);
850
851 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731 852
732 while (pairs--) 853 while (pairs--)
733 decode_he (dec, hv); 854 decode_he (dec, hv);
734 } 855 }
735 856
745static SV * 866static SV *
746decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
747{ 868{
748 SV *sv = 0; 869 SV *sv = 0;
749 870
750 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 872 {
873 // indefinite length strings
752 ++dec->cur; 874 ++dec->cur;
753 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
754 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
755 879
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 880 for (;;)
758 { 881 {
759 WANT (1); 882 WANT (1);
760 883
761 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 886 {
763 ++dec->cur; 887 ++dec->cur;
764 break; 888 break;
765 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 892
767 sv_catsv (sv, decode_sv (dec)); 893 STRLEN len = decode_uint (dec);
894
895 WANT (len);
896 sv_catpvn (sv, dec->cur, len);
897 dec->cur += len;
768 } 898 }
769 } 899 }
770 else 900 else
771 { 901 {
772 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 909 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 910 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 911 }
782 912
783 if (utf8) 913 if (utf8)
914 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918
784 SvUTF8_on (sv); 919 SvUTF8_on (sv);
920 }
785 921
786 return sv; 922 return sv;
787 923
788fail: 924fail:
789 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
808 sv = newRV_noinc (decode_sv (dec)); 944 sv = newRV_noinc (decode_sv (dec));
809 break; 945 break;
810 946
811 case CBOR_TAG_STRINGREF_NAMESPACE: 947 case CBOR_TAG_STRINGREF_NAMESPACE:
812 { 948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
813 ENTER; SAVETMPS; 951 ENTER;
814 952
815 SAVESPTR (dec->stringref); 953 SAVESPTR (dec->stringref);
816 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817 955
818 sv = decode_sv (dec); 956 sv = decode_sv (dec);
819 957
820 FREETMPS; LEAVE; 958 LEAVE;
821 } 959 }
822 break; 960 break;
823 961
824 case CBOR_TAG_STRINGREF: 962 case CBOR_TAG_STRINGREF:
825 { 963 {
826 if ((*dec->cur >> 5) != 0) 964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 966
829 UV idx = decode_uint (dec); 967 UV idx = decode_uint (dec);
830 968
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 case CBOR_TAG_VALUE_SHAREABLE: 976 case CBOR_TAG_VALUE_SHAREABLE:
839 { 977 {
840 if (ecb_expect_false (!dec->shareable)) 978 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
843 sv = newSV (0); 983 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 985
846 SV *osv = decode_sv (dec); 986 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 987 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 988 SvREFCNT_dec_NN (osv);
989 }
990 else
991 {
992 av_push (dec->shareable, &PL_sv_undef);
993 int idx = AvFILLp (dec->shareable);
994 sv = decode_sv (dec);
995 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
996 }
849 } 997 }
850 break; 998 break;
851 999
852 case CBOR_TAG_VALUE_SHAREDREF: 1000 case CBOR_TAG_VALUE_SHAREDREF:
853 { 1001 {
854 if ((*dec->cur >> 5) != 0) 1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 1004
857 UV idx = decode_uint (dec); 1005 UV idx = decode_uint (dec);
858 1006
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 1009
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1011
1012 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 1014 }
864 break; 1015 break;
865 1016
866 case CBOR_TAG_PERL_OBJECT: 1017 case CBOR_TAG_PERL_OBJECT:
867 { 1018 {
882 if (!method) 1033 if (!method)
883 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1034 ERR ("cannot decode perl-object (package does not have a THAW method)");
884 1035
885 dSP; 1036 dSP;
886 1037
887 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
888 EXTEND (SP, len + 1); 1040 EXTEND (SP, len + 1);
889 // we re-bless the reference to get overload and other niceties right 1041 // we re-bless the reference to get overload and other niceties right
890 PUSHs (*av_fetch (av, 0, 1)); 1042 PUSHs (*av_fetch (av, 0, 1));
891 PUSHs (sv_cbor); 1043 PUSHs (sv_cbor);
892 1044
914 } 1066 }
915 break; 1067 break;
916 1068
917 default: 1069 default:
918 { 1070 {
1071 SV *tag_sv = newSVuv (tag);
1072
919 sv = decode_sv (dec); 1073 sv = decode_sv (dec);
920 1074
921 dSP; 1075 dSP;
922 ENTER; SAVETMPS; PUSHMARK (SP); 1076 ENTER; SAVETMPS;
1077 SAVESTACK_POS ();
1078 PUSHMARK (SP);
923 EXTEND (SP, 2); 1079 EXTEND (SP, 2);
924 PUSHs (newSVuv (tag)); 1080 PUSHs (tag_sv);
925 PUSHs (sv); 1081 PUSHs (sv);
926 1082
927 PUTBACK; 1083 PUTBACK;
928 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1084 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929 SPAGAIN; 1085 SPAGAIN;
930 1086
931 if (SvTRUE (ERRSV)) 1087 if (SvTRUE (ERRSV))
932 { 1088 {
1089 SvREFCNT_dec (tag_sv);
933 FREETMPS; LEAVE; 1090 FREETMPS; LEAVE;
934 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1091 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
935 } 1092 }
936 1093
937 if (count) 1094 if (count)
938 { 1095 {
1096 SvREFCNT_dec (tag_sv);
939 SvREFCNT_dec (sv); 1097 SvREFCNT_dec (sv);
940 sv = SvREFCNT_inc (POPs); 1098 sv = SvREFCNT_inc (POPs);
941 } 1099 }
942 else 1100 else
943 { 1101 {
944 AV *av = newAV (); 1102 AV *av = newAV ();
945 av_push (av, newSVuv (tag)); 1103 av_push (av, tag_sv);
946 av_push (av, sv); 1104 av_push (av, sv);
947 1105
948 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1106 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949 ? cbor_tagged_stash 1107 ? cbor_tagged_stash
950 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1108 : gv_stashpv ("CBOR::XS::Tagged" , 1);
968static SV * 1126static SV *
969decode_sv (dec_t *dec) 1127decode_sv (dec_t *dec)
970{ 1128{
971 WANT (1); 1129 WANT (1);
972 1130
973 switch (*dec->cur >> 5) 1131 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1132 {
975 case 0: // unsigned int 1133 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1134 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1135 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1136 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1137 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1138 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1139 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1140
983 case 4: // array 1141 case MAJOR_MISC >> MAJOR_SHIFT:
984 return decode_av (dec);
985 case 5: // map
986 return decode_hv (dec);
987 case 6: // tag
988 return decode_tagged (dec);
989 case 7: // misc
990 switch (*dec->cur++ & 31) 1142 switch (*dec->cur++ & MINOR_MASK)
991 { 1143 {
992 case 20: 1144 case SIMPLE_FALSE:
993#if CBOR_SLOW 1145#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1146 types_false = get_bool ("Types::Serialiser::false");
995#endif 1147#endif
996 return newSVsv (types_false); 1148 return newSVsv (types_false);
997 case 21: 1149 case SIMPLE_TRUE:
998#if CBOR_SLOW 1150#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1151 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1152#endif
1001 return newSVsv (types_true); 1153 return newSVsv (types_true);
1002 case 22: 1154 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1155 return newSVsv (&PL_sv_undef);
1004 case 23: 1156 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1157#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1158 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1159#endif
1008 return newSVsv (types_error); 1160 return newSVsv (types_error);
1009 1161
1010 case 25: 1162 case MISC_FLOAT16:
1011 { 1163 {
1012 WANT (2); 1164 WANT (2);
1013 1165
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1166 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1167 dec->cur += 2;
1016 1168
1017 return newSVnv (ecb_binary16_to_float (fp)); 1169 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1170 }
1019 1171
1020 case 26: 1172 case MISC_FLOAT32:
1021 { 1173 {
1022 uint32_t fp; 1174 uint32_t fp;
1023 WANT (4); 1175 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1176 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1177 dec->cur += 4;
1028 fp = ecb_bswap32 (fp); 1180 fp = ecb_bswap32 (fp);
1029 1181
1030 return newSVnv (ecb_binary32_to_float (fp)); 1182 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1183 }
1032 1184
1033 case 27: 1185 case MISC_FLOAT64:
1034 { 1186 {
1035 uint64_t fp; 1187 uint64_t fp;
1036 WANT (8); 1188 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1189 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1190 dec->cur += 8;
1041 fp = ecb_bswap64 (fp); 1193 fp = ecb_bswap64 (fp);
1042 1194
1043 return newSVnv (ecb_binary64_to_double (fp)); 1195 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1196 }
1045 1197
1046 // 0..19 unassigned 1198 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1199 // 24 reserved + unassigned simple (reserved values are not encodable)
1200 // 28-30 unassigned misc
1201 // 31 break code
1048 default: 1202 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1203 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1204 }
1051 1205
1052 break; 1206 break;
1053 } 1207 }
1054 1208
1057} 1211}
1058 1212
1059static SV * 1213static SV *
1060decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1214decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061{ 1215{
1062 dec_t dec = { }; 1216 dec_t dec = { 0 };
1063 SV *sv; 1217 SV *sv;
1064 STRLEN len; 1218 STRLEN len;
1065 char *data = SvPVbyte (string, len); 1219 char *data = SvPVbyte (string, len);
1066 1220
1067 if (len > cbor->max_size && cbor->max_size) 1221 if (len > cbor->max_size && cbor->max_size)
1081 if (dec.cur != dec.end && !dec.err) 1235 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1236 dec.err = "garbage after CBOR object";
1083 1237
1084 if (dec.err) 1238 if (dec.err)
1085 { 1239 {
1240 if (dec.shareable)
1241 {
1242 // need to break cyclic links, which would all be in shareable
1243 int i;
1244 SV **svp;
1245
1246 for (i = av_len (dec.shareable) + 1; i--; )
1247 if ((svp = av_fetch (dec.shareable, i, 0)))
1248 sv_setsv (*svp, &PL_sv_undef);
1249 }
1250
1086 SvREFCNT_dec (sv); 1251 SvREFCNT_dec (sv);
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1252 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1253 }
1089 1254
1090 sv = sv_2mortal (sv); 1255 sv = sv_2mortal (sv);
1091 1256
1092 return sv; 1257 return sv;
1093} 1258}
1094 1259
1260/////////////////////////////////////////////////////////////////////////////
1261// incremental parser
1262
1263#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1264
1265// returns 0 for notyet, 1 for success or error
1266static int
1267incr_parse (CBOR *self, SV *cborstr)
1268{
1269 STRLEN cur;
1270 SvPV (cborstr, cur);
1271
1272 while (ecb_expect_true (self->incr_need <= cur))
1273 {
1274 // table of integer count bytes
1275 static I8 incr_len[MINOR_MASK + 1] = {
1276 0, 0, 0, 0, 0, 0, 0, 0,
1277 0, 0, 0, 0, 0, 0, 0, 0,
1278 0, 0, 0, 0, 0, 0, 0, 0,
1279 1, 2, 4, 8,-1,-1,-1,-2
1280 };
1281
1282 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1283 U8 m = *p & MINOR_MASK;
1284 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1285 I8 ilen = incr_len[m];
1286
1287 self->incr_need = self->incr_pos + 1;
1288
1289 if (ecb_expect_false (ilen < 0))
1290 {
1291 if (m != MINOR_INDEF)
1292 return 1; // error
1293
1294 if (*p == (MAJOR_MISC | MINOR_INDEF))
1295 {
1296 if (count >= 0)
1297 return 1; // error
1298
1299 count = 1;
1300 }
1301 else
1302 {
1303 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1304 count = -1;
1305 }
1306 }
1307 else
1308 {
1309 self->incr_need += ilen;
1310 if (ecb_expect_false (self->incr_need > cur))
1311 return 0;
1312
1313 int major = *p >> MAJOR_SHIFT;
1314
1315 switch (major)
1316 {
1317 case MAJOR_TAG >> MAJOR_SHIFT:
1318 ++count; // tags merely prefix another value
1319 break;
1320
1321 case MAJOR_BYTES >> MAJOR_SHIFT:
1322 case MAJOR_TEXT >> MAJOR_SHIFT:
1323 case MAJOR_ARRAY >> MAJOR_SHIFT:
1324 case MAJOR_MAP >> MAJOR_SHIFT:
1325 {
1326 UV len;
1327
1328 if (ecb_expect_false (ilen))
1329 {
1330 len = 0;
1331
1332 do {
1333 len = (len << 8) | *++p;
1334 } while (--ilen);
1335 }
1336 else
1337 len = m;
1338
1339 switch (major)
1340 {
1341 case MAJOR_BYTES >> MAJOR_SHIFT:
1342 case MAJOR_TEXT >> MAJOR_SHIFT:
1343 self->incr_need += len;
1344 if (ecb_expect_false (self->incr_need > cur))
1345 return 0;
1346
1347 break;
1348
1349 case MAJOR_MAP >> MAJOR_SHIFT:
1350 len <<= 1;
1351 case MAJOR_ARRAY >> MAJOR_SHIFT:
1352 if (len)
1353 {
1354 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1355 count = len + 1;
1356 }
1357 break;
1358 }
1359 }
1360 }
1361 }
1362
1363 self->incr_pos = self->incr_need;
1364
1365 if (count > 0)
1366 {
1367 while (!--count)
1368 {
1369 if (!AvFILLp (self->incr_count))
1370 return 1; // done
1371
1372 SvREFCNT_dec_NN (av_pop (self->incr_count));
1373 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1374 }
1375
1376 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1377 }
1378 }
1379
1380 return 0;
1381}
1382
1383
1095///////////////////////////////////////////////////////////////////////////// 1384/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1385// XS interface functions
1097 1386
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1387MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1388
1111 1400
1112 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1401 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113 1402
1114 sv_cbor = newSVpv ("CBOR", 0); 1403 sv_cbor = newSVpv ("CBOR", 0);
1115 SvREADONLY_on (sv_cbor); 1404 SvREADONLY_on (sv_cbor);
1405
1406 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116} 1407}
1117 1408
1118PROTOTYPES: DISABLE 1409PROTOTYPES: DISABLE
1119 1410
1120void CLONE (...) 1411void CLONE (...)
1139void shrink (CBOR *self, int enable = 1) 1430void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1431 ALIAS:
1141 shrink = F_SHRINK 1432 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1433 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1434 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1435 allow_cycles = F_ALLOW_CYCLES
1436 pack_strings = F_PACK_STRINGS
1437 text_keys = F_TEXT_KEYS
1438 text_strings = F_TEXT_STRINGS
1439 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1440 PPCODE:
1146{ 1441{
1147 if (enable) 1442 if (enable)
1148 self->flags |= ix; 1443 self->flags |= ix;
1149 else 1444 else
1155void get_shrink (CBOR *self) 1450void get_shrink (CBOR *self)
1156 ALIAS: 1451 ALIAS:
1157 get_shrink = F_SHRINK 1452 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1453 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1454 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1455 get_allow_cycles = F_ALLOW_CYCLES
1456 get_pack_strings = F_PACK_STRINGS
1457 get_text_keys = F_TEXT_KEYS
1458 get_text_strings = F_TEXT_STRINGS
1459 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1460 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1461 XPUSHs (boolSV (self->flags & ix));
1163 1462
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1463void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1464 PPCODE:
1214 EXTEND (SP, 2); 1513 EXTEND (SP, 2);
1215 PUSHs (sv); 1514 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1515 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1516}
1218 1517
1518void incr_parse (CBOR *self, SV *cborstr)
1519 ALIAS:
1520 incr_parse_multiple = 1
1521 PPCODE:
1522{
1523 if (SvUTF8 (cborstr))
1524 sv_utf8_downgrade (cborstr, 0);
1525
1526 if (!self->incr_count)
1527 {
1528 self->incr_count = newAV ();
1529 self->incr_pos = 0;
1530 self->incr_need = 1;
1531
1532 av_push (self->incr_count, newSViv (1));
1533 }
1534
1535 do
1536 {
1537 if (!incr_parse (self, cborstr))
1538 {
1539 if (self->incr_need > self->max_size && self->max_size)
1540 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1541 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1542
1543 break;
1544 }
1545
1546 SV *sv;
1547 char *offset;
1548
1549 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1550 XPUSHs (sv);
1551
1552 sv_chop (cborstr, offset);
1553
1554 av_clear (self->incr_count);
1555 av_push (self->incr_count, newSViv (1));
1556
1557 self->incr_pos = 0;
1558 self->incr_need = self->incr_pos + 1;
1559 }
1560 while (ix);
1561}
1562
1563void incr_reset (CBOR *self)
1564 CODE:
1565{
1566 SvREFCNT_dec (self->incr_count);
1567 self->incr_count = 0;
1568}
1569
1219void DESTROY (CBOR *self) 1570void DESTROY (CBOR *self)
1220 PPCODE: 1571 PPCODE:
1221 cbor_free (self); 1572 cbor_free (self);
1222 1573
1223PROTOTYPES: ENABLE 1574PROTOTYPES: ENABLE
1224 1575
1225void encode_cbor (SV *scalar) 1576void encode_cbor (SV *scalar)
1577 ALIAS:
1578 encode_cbor = 0
1579 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1580 PPCODE:
1227{ 1581{
1228 CBOR cbor; 1582 CBOR cbor;
1229 cbor_init (&cbor); 1583 cbor_init (&cbor);
1584 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1585 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1586 XPUSHs (scalar);
1232} 1587}
1233 1588
1234void decode_cbor (SV *cborstr) 1589void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines