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 1.21 by root, Wed Nov 20 14:59:06 2013 UTC vs.
Revision 1.59 by root, Fri Nov 25 23:37:27 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
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 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_DEDUP_STRINGS)) 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}
275
276ecb_inline void
277encode_he (enc_t *enc, HE *he)
278{
279} 375}
280 376
281static void 377static void
282encode_hv (enc_t *enc, HV *hv) 378encode_hv (enc_t *enc, HV *hv)
283{ 379{
290 386
291 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
292 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
293 389
294 if (mg) 390 if (mg)
295 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
296 else 392 else
297 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
298 394
299 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
300 { 396 {
301 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
302 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
303 else 399 else
304 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));
305 401
306 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));
307 } 403 }
308 404
309 if (mg) 405 if (mg)
310 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
311 407
312 --enc->depth; 408 --enc->depth;
313} 409}
314 410
315// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
316static void 412static void
317encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
318{ 414{
319 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
320
321 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
322 && ecb_expect_false (SvREFCNT (sv) > 1))
323 {
324 if (!enc->shareable)
325 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
326
327 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
328
329 if (SvOK (*svp))
330 {
331 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
332 encode_uint (enc, 0x00, SvUV (*svp));
333 return;
334 }
335 else
336 {
337 sv_setuv (*svp, enc->shareable_idx);
338 ++enc->shareable_idx;
339 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
340 }
341 }
342 416
343 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
344 418
345 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
346 { 420 {
353 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
354 ? cbor_tagged_stash 428 ? cbor_tagged_stash
355 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
356 430
357 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
358 GV *method;
359 432
360 if (stash == boolean_stash) 433 if (stash == boolean_stash)
361 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 }
362 else if (stash == error_stash) 438 else if (stash == error_stash)
363 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
364 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
365 { 444 {
366 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
367 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
368 447
369 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
370 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))
371 } 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
372 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
373 { 483 {
374 dSP; 484 dSP;
375 485
376 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
377 // 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
378 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
379 490
380 PUTBACK; 491 PUTBACK;
381 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
394 } 505 }
395 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
396 { 507 {
397 dSP; 508 dSP;
398 509
399 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
400 EXTEND (SP, 2); 513 EXTEND (SP, 2);
401 // 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
402 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
403 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
404 517
409 // catch this surprisingly common error 522 // catch this surprisingly common error
410 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
411 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));
412 525
413 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
414 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
415 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
416 529
417 while (count) 530 while (count)
418 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
419 532
420 PUTBACK; 533 PUTBACK;
441{ 554{
442 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
443 556
444 need (enc, 9); 557 need (enc, 9);
445 558
446 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
447 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
448 //TODO: maybe I32? 561 //TODO: maybe I32?
449 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
450 { 563 {
451 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
452 565
453 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
454 567
455 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
456 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
457 570
458 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
460 } 573 }
461 else 574 else
462 { 575 {
463 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
464 577
465 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
466 579
467 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
468 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
469 582
470 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
479 592
480 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
481 { 594 {
482 STRLEN len; 595 STRLEN len;
483 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
484 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
485 } 598 }
486 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
487 encode_nv (enc, sv); 600 encode_nv (enc, sv);
488 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
489 { 602 {
490 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
491 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
492 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
493 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
494 else 607 else
495 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
496 } 609 }
497 else if (SvROK (sv)) 610 else if (SvROK (sv))
498 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
499 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
500 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
501 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
502 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
503 else 616 else
504 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",
505 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
506} 619}
507 620
508static SV * 621static SV *
509encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
510{ 623{
511 enc_t enc = { }; 624 enc_t enc = { 0 };
512 625
513 enc.cbor = *cbor; 626 enc.cbor = *cbor;
514 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
515 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
516 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
517 630
518 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
519 632
520 if (cbor->flags & F_DEDUP_STRINGS) 633 if (cbor->flags & F_PACK_STRINGS)
521 { 634 {
522 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
523 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
524 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
525 } 638 }
547 CBOR cbor; 660 CBOR cbor;
548 U32 depth; // recursion depth 661 U32 depth; // recursion depth
549 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
550 AV *shareable; 663 AV *shareable;
551 AV *stringref; 664 AV *stringref;
665 SV *decode_tagged;
666 SV *err_sv; // optional sv for error, needs to be freed
552} dec_t; 667} dec_t;
553 668
554#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 669// set dec->err to ERRSV
670ecb_cold static void
671err_errsv (dec_t *dec)
672{
673 if (!dec->err)
674 {
675 dec->err_sv = newSVsv (ERRSV);
555 676
556#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 677 // chop off the trailing \n
678 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
679 *SvEND (dec->err_sv) = 0;
557 680
681 dec->err = SvPVutf8_nolen (dec->err_sv);
682 }
683}
684
685// the following functions are used to reduce code size and help the compiler to optimise
686ecb_cold static void
687err_set (dec_t *dec, const char *reason)
688{
689 if (!dec->err)
690 dec->err = reason;
691}
692
693ecb_cold static void
694err_unexpected_end (dec_t *dec)
695{
696 err_set (dec, "unexpected end of CBOR data");
697}
698
699ecb_cold static void
700err_nesting_exceeded (dec_t *dec)
701{
702 err_set (dec, ERR_NESTING_EXCEEDED);
703}
704
705#define ERR_DO(do) SB do; goto fail; SE
706#define ERR(reason) ERR_DO (err_set (dec, reason))
707#define ERR_ERRSV ERR_DO (err_errsv (dec))
708
709#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR_DO (err_unexpected_end (dec))
710
558#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 711#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
559#define DEC_DEC_DEPTH --dec->depth 712#define DEC_DEC_DEPTH --dec->depth
560 713
561static UV 714static UV
562decode_uint (dec_t *dec) 715decode_uint (dec_t *dec)
563{ 716{
564 switch (*dec->cur & 31) 717 U8 m = *dec->cur & MINOR_MASK;
565 { 718 ++dec->cur;
566 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
567 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
568 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
569 return *dec->cur++ & 31;
570 719
571 case 24: 720 if (ecb_expect_true (m < LENGTH_EXT1))
721 return m;
722 else if (ecb_expect_true (m == LENGTH_EXT1))
723 {
572 WANT (2); 724 WANT (1);
573 dec->cur += 2; 725 dec->cur += 1;
574 return dec->cur[-1]; 726 return dec->cur[-1];
575 727 }
576 case 25: 728 else if (ecb_expect_true (m == LENGTH_EXT2))
729 {
577 WANT (3); 730 WANT (2);
578 dec->cur += 3; 731 dec->cur += 2;
579 return (((UV)dec->cur[-2]) << 8) 732 return (((UV)dec->cur[-2]) << 8)
580 | ((UV)dec->cur[-1]); 733 | ((UV)dec->cur[-1]);
581 734 }
582 case 26: 735 else if (ecb_expect_true (m == LENGTH_EXT4))
736 {
583 WANT (5); 737 WANT (4);
584 dec->cur += 5; 738 dec->cur += 4;
585 return (((UV)dec->cur[-4]) << 24) 739 return (((UV)dec->cur[-4]) << 24)
586 | (((UV)dec->cur[-3]) << 16) 740 | (((UV)dec->cur[-3]) << 16)
587 | (((UV)dec->cur[-2]) << 8) 741 | (((UV)dec->cur[-2]) << 8)
588 | ((UV)dec->cur[-1]); 742 | ((UV)dec->cur[-1]);
589 743 }
590 case 27: 744 else if (ecb_expect_true (m == LENGTH_EXT8))
745 {
591 WANT (9); 746 WANT (8);
592 dec->cur += 9; 747 dec->cur += 8;
748
749 return
750#if UVSIZE < 8
751 0
752#else
593 return (((UV)dec->cur[-8]) << 56) 753 (((UV)dec->cur[-8]) << 56)
594 | (((UV)dec->cur[-7]) << 48) 754 | (((UV)dec->cur[-7]) << 48)
595 | (((UV)dec->cur[-6]) << 40) 755 | (((UV)dec->cur[-6]) << 40)
596 | (((UV)dec->cur[-5]) << 32) 756 | (((UV)dec->cur[-5]) << 32)
757#endif
597 | (((UV)dec->cur[-4]) << 24) 758 | (((UV)dec->cur[-4]) << 24)
598 | (((UV)dec->cur[-3]) << 16) 759 | (((UV)dec->cur[-3]) << 16)
599 | (((UV)dec->cur[-2]) << 8) 760 | (((UV)dec->cur[-2]) << 8)
600 | ((UV)dec->cur[-1]); 761 | ((UV)dec->cur[-1]);
601 762 }
602 default: 763 else
603 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 764 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
604 }
605 765
606fail: 766fail:
607 return 0; 767 return 0;
608} 768}
609 769
614{ 774{
615 AV *av = newAV (); 775 AV *av = newAV ();
616 776
617 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
618 778
619 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
620 { 780 {
621 ++dec->cur; 781 ++dec->cur;
622 782
623 for (;;) 783 for (;;)
624 { 784 {
625 WANT (1); 785 WANT (1);
626 786
627 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
628 { 788 {
629 ++dec->cur; 789 ++dec->cur;
630 break; 790 break;
631 } 791 }
632 792
633 av_push (av, decode_sv (dec)); 793 av_push (av, decode_sv (dec));
634 } 794 }
635 } 795 }
636 else 796 else
637 { 797 {
638 int i, len = decode_uint (dec); 798 UV i, len = decode_uint (dec);
639 799
800 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
640 av_fill (av, len - 1); 801 av_fill (av, len - 1);
641 802
642 for (i = 0; i < len; ++i) 803 for (i = 0; i < len; ++i)
643 AvARRAY (av)[i] = decode_sv (dec); 804 AvARRAY (av)[i] = decode_sv (dec);
644 } 805 }
656decode_he (dec_t *dec, HV *hv) 817decode_he (dec_t *dec, HV *hv)
657{ 818{
658 // for speed reasons, we specialcase single-string 819 // for speed reasons, we specialcase single-string
659 // byte or utf-8 strings as keys, but only when !stringref 820 // byte or utf-8 strings as keys, but only when !stringref
660 821
661 if (expect_true (!dec->stringref)) 822 if (ecb_expect_true (!dec->stringref))
662 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 823 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
663 { 824 {
664 I32 len = decode_uint (dec); 825 STRLEN len = decode_uint (dec);
665 char *key = (char *)dec->cur; 826 char *key = (char *)dec->cur;
666 827
828 WANT (len);
667 dec->cur += len; 829 dec->cur += len;
668 830
669 if (ecb_expect_false (dec->stringref))
670 av_push (dec->stringref, newSVpvn (key, len));
671
672 hv_store (hv, key, len, decode_sv (dec), 0); 831 hv_store (hv, key, len, decode_sv (dec), 0);
673 832
674 return; 833 return;
675 } 834 }
676 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 835 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
677 { 836 {
678 I32 len = decode_uint (dec); 837 STRLEN len = decode_uint (dec);
679 char *key = (char *)dec->cur; 838 char *key = (char *)dec->cur;
680 839
840 WANT (len);
681 dec->cur += len; 841 dec->cur += len;
682 842
683 if (ecb_expect_false (dec->stringref)) 843 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
684 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 844 if (!is_utf8_string (key, len))
845 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
685 846
686 hv_store (hv, key, -len, decode_sv (dec), 0); 847 hv_store (hv, key, -len, decode_sv (dec), 0);
687 848
688 return; 849 return;
689 } 850 }
690 851
691 SV *k = decode_sv (dec); 852 SV *k = decode_sv (dec);
692 SV *v = decode_sv (dec); 853 SV *v = decode_sv (dec);
693 854
855 // we leak memory if uncaught exceptions are thrown by random magical
856 // methods, and this is hopefully the only place where it can happen,
857 // so if there is a chance of an exception, take the very slow path.
858 // since catching exceptions is "undocumented/internal/forbidden" by
859 // the new p5p powers, we need to call out to a perl function :/
860 if (ecb_expect_false (SvAMAGIC (k)))
861 {
862 dSP;
863
864 ENTER; SAVETMPS;
865 PUSHMARK (SP);
866 EXTEND (SP, 3);
867 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
868 PUSHs (sv_2mortal (k));
869 PUSHs (sv_2mortal (v));
870
871 PUTBACK;
872 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
873 SPAGAIN;
874
875 FREETMPS; LEAVE;
876
877 if (SvTRUE (ERRSV))
878 ERR_ERRSV;
879
880 return;
881 }
882
694 hv_store_ent (hv, k, v, 0); 883 hv_store_ent (hv, k, v, 0);
695 SvREFCNT_dec (k); 884 SvREFCNT_dec (k);
885
886fail:
887 ;
696} 888}
697 889
698static SV * 890static SV *
699decode_hv (dec_t *dec) 891decode_hv (dec_t *dec)
700{ 892{
701 HV *hv = newHV (); 893 HV *hv = newHV ();
702 894
703 DEC_INC_DEPTH; 895 DEC_INC_DEPTH;
704 896
705 if ((*dec->cur & 31) == 31) 897 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
706 { 898 {
707 ++dec->cur; 899 ++dec->cur;
708 900
709 for (;;) 901 for (;;)
710 { 902 {
711 WANT (1); 903 WANT (1);
712 904
713 if (*dec->cur == (0xe0 | 31)) 905 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
714 { 906 {
715 ++dec->cur; 907 ++dec->cur;
716 break; 908 break;
717 } 909 }
718 910
719 decode_he (dec, hv); 911 decode_he (dec, hv);
720 } 912 }
721 } 913 }
722 else 914 else
723 { 915 {
724 int pairs = decode_uint (dec); 916 UV pairs = decode_uint (dec);
917
918 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
725 919
726 while (pairs--) 920 while (pairs--)
727 decode_he (dec, hv); 921 decode_he (dec, hv);
728 } 922 }
729 923
739static SV * 933static SV *
740decode_str (dec_t *dec, int utf8) 934decode_str (dec_t *dec, int utf8)
741{ 935{
742 SV *sv = 0; 936 SV *sv = 0;
743 937
744 if ((*dec->cur & 31) == 31) 938 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
745 { 939 {
940 // indefinite length strings
746 ++dec->cur; 941 ++dec->cur;
747 942
943 U8 major = *dec->cur & MAJOR_MISC;
944
748 sv = newSVpvn ("", 0); 945 sv = newSVpvn ("", 0);
749 946
750 // not very fast, and certainly not robust against illegal input
751 for (;;) 947 for (;;)
752 { 948 {
753 WANT (1); 949 WANT (1);
754 950
755 if (*dec->cur == (0xe0 | 31)) 951 if ((*dec->cur - major) > LENGTH_EXT8)
952 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
756 { 953 {
757 ++dec->cur; 954 ++dec->cur;
758 break; 955 break;
759 } 956 }
957 else
958 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
760 959
761 sv_catsv (sv, decode_sv (dec)); 960 STRLEN len = decode_uint (dec);
961
962 WANT (len);
963 sv_catpvn (sv, dec->cur, len);
964 dec->cur += len;
762 } 965 }
763 } 966 }
764 else 967 else
765 { 968 {
766 STRLEN len = decode_uint (dec); 969 STRLEN len = decode_uint (dec);
767 970
768 WANT (len); 971 WANT (len);
769 sv = newSVpvn (dec->cur, len); 972 sv = newSVpvn (dec->cur, len);
770 dec->cur += len; 973 dec->cur += len;
974
975 if (ecb_expect_false (dec->stringref)
976 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
977 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
771 } 978 }
772 979
773 if (utf8) 980 if (utf8)
981 {
982 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
983 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
984 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
985
774 SvUTF8_on (sv); 986 SvUTF8_on (sv);
775 987 }
776 if (ecb_expect_false (dec->stringref)
777 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
778 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
779 988
780 return sv; 989 return sv;
781 990
782fail: 991fail:
783 SvREFCNT_dec (sv); 992 SvREFCNT_dec (sv);
802 sv = newRV_noinc (decode_sv (dec)); 1011 sv = newRV_noinc (decode_sv (dec));
803 break; 1012 break;
804 1013
805 case CBOR_TAG_STRINGREF_NAMESPACE: 1014 case CBOR_TAG_STRINGREF_NAMESPACE:
806 { 1015 {
1016 // do nmot use SAVETMPS/FREETMPS, as these will
1017 // erase mortalised caches, e.g. "shareable"
807 ENTER; SAVETMPS; 1018 ENTER;
808 1019
809 SAVESPTR (dec->stringref); 1020 SAVESPTR (dec->stringref);
810 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1021 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
811 1022
812 sv = decode_sv (dec); 1023 sv = decode_sv (dec);
813 1024
814 FREETMPS; LEAVE; 1025 LEAVE;
815 } 1026 }
816 break; 1027 break;
817 1028
818 case CBOR_TAG_STRINGREF: 1029 case CBOR_TAG_STRINGREF:
819 { 1030 {
820 if ((*dec->cur >> 5) != 0) 1031 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
821 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1032 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
822 1033
823 UV idx = decode_uint (dec); 1034 UV idx = decode_uint (dec);
824 1035
825 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1036 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
832 case CBOR_TAG_VALUE_SHAREABLE: 1043 case CBOR_TAG_VALUE_SHAREABLE:
833 { 1044 {
834 if (ecb_expect_false (!dec->shareable)) 1045 if (ecb_expect_false (!dec->shareable))
835 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1046 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
836 1047
1048 if (dec->cbor.flags & F_ALLOW_CYCLES)
1049 {
837 sv = newSV (0); 1050 sv = newSV (0);
838 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1051 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
839 1052
840 SV *osv = decode_sv (dec); 1053 SV *osv = decode_sv (dec);
841 sv_setsv (sv, osv); 1054 sv_setsv (sv, osv);
842 SvREFCNT_dec_NN (osv); 1055 SvREFCNT_dec_NN (osv);
1056 }
1057 else
1058 {
1059 av_push (dec->shareable, &PL_sv_undef);
1060 int idx = AvFILLp (dec->shareable);
1061 sv = decode_sv (dec);
1062 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1063 }
843 } 1064 }
844 break; 1065 break;
845 1066
846 case CBOR_TAG_VALUE_SHAREDREF: 1067 case CBOR_TAG_VALUE_SHAREDREF:
847 { 1068 {
848 if ((*dec->cur >> 5) != 0) 1069 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
849 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1070 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
850 1071
851 UV idx = decode_uint (dec); 1072 UV idx = decode_uint (dec);
852 1073
853 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1074 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
854 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1075 ERR ("corrupted CBOR data (sharedref index out of bounds)");
855 1076
856 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1077 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1078
1079 if (sv == &PL_sv_undef)
1080 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
857 } 1081 }
858 break; 1082 break;
859 1083
860 case CBOR_TAG_PERL_OBJECT: 1084 case CBOR_TAG_PERL_OBJECT:
861 { 1085 {
876 if (!method) 1100 if (!method)
877 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1101 ERR ("cannot decode perl-object (package does not have a THAW method)");
878 1102
879 dSP; 1103 dSP;
880 1104
881 ENTER; SAVETMPS; PUSHMARK (SP); 1105 ENTER; SAVETMPS;
1106 PUSHMARK (SP);
882 EXTEND (SP, len + 1); 1107 EXTEND (SP, len + 1);
883 // we re-bless the reference to get overload and other niceties right 1108 // we re-bless the reference to get overload and other niceties right
884 PUSHs (*av_fetch (av, 0, 1)); 1109 PUSHs (*av_fetch (av, 0, 1));
885 PUSHs (sv_cbor); 1110 PUSHs (sv_cbor);
886 1111
894 SPAGAIN; 1119 SPAGAIN;
895 1120
896 if (SvTRUE (ERRSV)) 1121 if (SvTRUE (ERRSV))
897 { 1122 {
898 FREETMPS; LEAVE; 1123 FREETMPS; LEAVE;
899 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1124 ERR_ERRSV;
900 } 1125 }
901 1126
902 SvREFCNT_dec (sv); 1127 SvREFCNT_dec (sv);
903 sv = SvREFCNT_inc (POPs); 1128 sv = SvREFCNT_inc (POPs);
904 1129
908 } 1133 }
909 break; 1134 break;
910 1135
911 default: 1136 default:
912 { 1137 {
1138 SV *tag_sv = newSVuv (tag);
1139
913 sv = decode_sv (dec); 1140 sv = decode_sv (dec);
914 1141
1142 dSP;
1143 ENTER; SAVETMPS;
1144 SAVESTACK_POS ();
1145 PUSHMARK (SP);
1146 EXTEND (SP, 2);
1147 PUSHs (tag_sv);
1148 PUSHs (sv);
1149
1150 PUTBACK;
1151 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1152 SPAGAIN;
1153
1154 if (SvTRUE (ERRSV))
1155 {
1156 SvREFCNT_dec (tag_sv);
1157 FREETMPS; LEAVE;
1158 ERR_ERRSV;
1159 }
1160
1161 if (count)
1162 {
1163 SvREFCNT_dec (tag_sv);
1164 SvREFCNT_dec (sv);
1165 sv = SvREFCNT_inc (POPs);
1166 }
1167 else
1168 {
915 AV *av = newAV (); 1169 AV *av = newAV ();
916 av_push (av, newSVuv (tag)); 1170 av_push (av, tag_sv);
917 av_push (av, sv); 1171 av_push (av, sv);
918 1172
919 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1173 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
920 ? cbor_tagged_stash 1174 ? cbor_tagged_stash
921 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1175 : gv_stashpv ("CBOR::XS::Tagged" , 1);
922
923 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1176 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1177 }
1178
1179 PUTBACK;
1180
1181 FREETMPS; LEAVE;
924 } 1182 }
925 break; 1183 break;
926 } 1184 }
927 1185
928 return sv; 1186 return sv;
935static SV * 1193static SV *
936decode_sv (dec_t *dec) 1194decode_sv (dec_t *dec)
937{ 1195{
938 WANT (1); 1196 WANT (1);
939 1197
940 switch (*dec->cur >> 5) 1198 switch (*dec->cur >> MAJOR_SHIFT)
941 { 1199 {
942 case 0: // unsigned int 1200 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
943 return newSVuv (decode_uint (dec)); 1201 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
944 case 1: // negative int 1202 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
945 return newSViv (-1 - (IV)decode_uint (dec)); 1203 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
946 case 2: // octet string 1204 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
947 return decode_str (dec, 0); 1205 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
948 case 3: // utf-8 string 1206 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
949 return decode_str (dec, 1); 1207
950 case 4: // array 1208 case MAJOR_MISC >> MAJOR_SHIFT:
951 return decode_av (dec);
952 case 5: // map
953 return decode_hv (dec);
954 case 6: // tag
955 return decode_tagged (dec);
956 case 7: // misc
957 switch (*dec->cur++ & 31) 1209 switch (*dec->cur++ & MINOR_MASK)
958 { 1210 {
959 case 20: 1211 case SIMPLE_FALSE:
960#if CBOR_SLOW 1212#if CBOR_SLOW
961 types_false = get_bool ("Types::Serialiser::false"); 1213 types_false = get_bool ("Types::Serialiser::false");
962#endif 1214#endif
963 return newSVsv (types_false); 1215 return newSVsv (types_false);
964 case 21: 1216 case SIMPLE_TRUE:
965#if CBOR_SLOW 1217#if CBOR_SLOW
966 types_true = get_bool ("Types::Serialiser::true"); 1218 types_true = get_bool ("Types::Serialiser::true");
967#endif 1219#endif
968 return newSVsv (types_true); 1220 return newSVsv (types_true);
969 case 22: 1221 case SIMPLE_NULL:
970 return newSVsv (&PL_sv_undef); 1222 return newSVsv (&PL_sv_undef);
971 case 23: 1223 case SIMPLE_UNDEF:
972#if CBOR_SLOW 1224#if CBOR_SLOW
973 types_error = get_bool ("Types::Serialiser::error"); 1225 types_error = get_bool ("Types::Serialiser::error");
974#endif 1226#endif
975 return newSVsv (types_error); 1227 return newSVsv (types_error);
976 1228
977 case 25: 1229 case MISC_FLOAT16:
978 { 1230 {
979 WANT (2); 1231 WANT (2);
980 1232
981 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1233 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
982 dec->cur += 2; 1234 dec->cur += 2;
983 1235
984 return newSVnv (ecb_binary16_to_float (fp)); 1236 return newSVnv (ecb_binary16_to_float (fp));
985 } 1237 }
986 1238
987 case 26: 1239 case MISC_FLOAT32:
988 { 1240 {
989 uint32_t fp; 1241 uint32_t fp;
990 WANT (4); 1242 WANT (4);
991 memcpy (&fp, dec->cur, 4); 1243 memcpy (&fp, dec->cur, 4);
992 dec->cur += 4; 1244 dec->cur += 4;
995 fp = ecb_bswap32 (fp); 1247 fp = ecb_bswap32 (fp);
996 1248
997 return newSVnv (ecb_binary32_to_float (fp)); 1249 return newSVnv (ecb_binary32_to_float (fp));
998 } 1250 }
999 1251
1000 case 27: 1252 case MISC_FLOAT64:
1001 { 1253 {
1002 uint64_t fp; 1254 uint64_t fp;
1003 WANT (8); 1255 WANT (8);
1004 memcpy (&fp, dec->cur, 8); 1256 memcpy (&fp, dec->cur, 8);
1005 dec->cur += 8; 1257 dec->cur += 8;
1008 fp = ecb_bswap64 (fp); 1260 fp = ecb_bswap64 (fp);
1009 1261
1010 return newSVnv (ecb_binary64_to_double (fp)); 1262 return newSVnv (ecb_binary64_to_double (fp));
1011 } 1263 }
1012 1264
1013 // 0..19 unassigned 1265 // 0..19 unassigned simple
1014 // 24 reserved + unassigned (reserved values are not encodable) 1266 // 24 reserved + unassigned simple (reserved values are not encodable)
1267 // 28-30 unassigned misc
1268 // 31 break code
1015 default: 1269 default:
1016 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1270 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1017 } 1271 }
1018 1272
1019 break; 1273 break;
1020 } 1274 }
1021 1275
1024} 1278}
1025 1279
1026static SV * 1280static SV *
1027decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1281decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1028{ 1282{
1029 dec_t dec = { }; 1283 dec_t dec = { 0 };
1030 SV *sv; 1284 SV *sv;
1031 STRLEN len; 1285 STRLEN len;
1032 char *data = SvPVbyte (string, len); 1286 char *data = SvPVbyte (string, len);
1033 1287
1034 if (len > cbor->max_size && cbor->max_size) 1288 if (len > cbor->max_size && cbor->max_size)
1048 if (dec.cur != dec.end && !dec.err) 1302 if (dec.cur != dec.end && !dec.err)
1049 dec.err = "garbage after CBOR object"; 1303 dec.err = "garbage after CBOR object";
1050 1304
1051 if (dec.err) 1305 if (dec.err)
1052 { 1306 {
1307 if (dec.shareable)
1308 {
1309 // need to break cyclic links, which would all be in shareable
1310 int i;
1311 SV **svp;
1312
1313 for (i = av_len (dec.shareable) + 1; i--; )
1314 if ((svp = av_fetch (dec.shareable, i, 0)))
1315 sv_setsv (*svp, &PL_sv_undef);
1316 }
1317
1053 SvREFCNT_dec (sv); 1318 SvREFCNT_dec (sv);
1319
1320 if (dec.err_sv)
1321 sv_2mortal (dec.err_sv);
1322
1054 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1323 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1055 } 1324 }
1056 1325
1057 sv = sv_2mortal (sv); 1326 sv = sv_2mortal (sv);
1058 1327
1059 return sv; 1328 return sv;
1060} 1329}
1061 1330
1331/////////////////////////////////////////////////////////////////////////////
1332// incremental parser
1333
1334#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1335
1336// returns 0 for notyet, 1 for success or error
1337static int
1338incr_parse (CBOR *self, SV *cborstr)
1339{
1340 STRLEN cur;
1341 SvPV (cborstr, cur);
1342
1343 while (ecb_expect_true (self->incr_need <= cur))
1344 {
1345 // table of integer count bytes
1346 static I8 incr_len[MINOR_MASK + 1] = {
1347 0, 0, 0, 0, 0, 0, 0, 0,
1348 0, 0, 0, 0, 0, 0, 0, 0,
1349 0, 0, 0, 0, 0, 0, 0, 0,
1350 1, 2, 4, 8,-1,-1,-1,-2
1351 };
1352
1353 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1354 U8 m = *p & MINOR_MASK;
1355 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1356 I8 ilen = incr_len[m];
1357
1358 self->incr_need = self->incr_pos + 1;
1359
1360 if (ecb_expect_false (ilen < 0))
1361 {
1362 if (m != MINOR_INDEF)
1363 return 1; // error
1364
1365 if (*p == (MAJOR_MISC | MINOR_INDEF))
1366 {
1367 if (count >= 0)
1368 return 1; // error
1369
1370 count = 1;
1371 }
1372 else
1373 {
1374 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1375 count = -1;
1376 }
1377 }
1378 else
1379 {
1380 self->incr_need += ilen;
1381 if (ecb_expect_false (self->incr_need > cur))
1382 return 0;
1383
1384 int major = *p >> MAJOR_SHIFT;
1385
1386 switch (major)
1387 {
1388 case MAJOR_TAG >> MAJOR_SHIFT:
1389 ++count; // tags merely prefix another value
1390 break;
1391
1392 case MAJOR_BYTES >> MAJOR_SHIFT:
1393 case MAJOR_TEXT >> MAJOR_SHIFT:
1394 case MAJOR_ARRAY >> MAJOR_SHIFT:
1395 case MAJOR_MAP >> MAJOR_SHIFT:
1396 {
1397 UV len;
1398
1399 if (ecb_expect_false (ilen))
1400 {
1401 len = 0;
1402
1403 do {
1404 len = (len << 8) | *++p;
1405 } while (--ilen);
1406 }
1407 else
1408 len = m;
1409
1410 switch (major)
1411 {
1412 case MAJOR_BYTES >> MAJOR_SHIFT:
1413 case MAJOR_TEXT >> MAJOR_SHIFT:
1414 self->incr_need += len;
1415 if (ecb_expect_false (self->incr_need > cur))
1416 return 0;
1417
1418 break;
1419
1420 case MAJOR_MAP >> MAJOR_SHIFT:
1421 len <<= 1;
1422 case MAJOR_ARRAY >> MAJOR_SHIFT:
1423 if (len)
1424 {
1425 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1426 count = len + 1;
1427 }
1428 break;
1429 }
1430 }
1431 }
1432 }
1433
1434 self->incr_pos = self->incr_need;
1435
1436 if (count > 0)
1437 {
1438 while (!--count)
1439 {
1440 if (!AvFILLp (self->incr_count))
1441 return 1; // done
1442
1443 SvREFCNT_dec_NN (av_pop (self->incr_count));
1444 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1445 }
1446
1447 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1448 }
1449 }
1450
1451 return 0;
1452}
1453
1454
1062///////////////////////////////////////////////////////////////////////////// 1455/////////////////////////////////////////////////////////////////////////////
1063// XS interface functions 1456// XS interface functions
1064 1457
1065MODULE = CBOR::XS PACKAGE = CBOR::XS 1458MODULE = CBOR::XS PACKAGE = CBOR::XS
1066 1459
1074 1467
1075 types_true = get_bool ("Types::Serialiser::true" ); 1468 types_true = get_bool ("Types::Serialiser::true" );
1076 types_false = get_bool ("Types::Serialiser::false"); 1469 types_false = get_bool ("Types::Serialiser::false");
1077 types_error = get_bool ("Types::Serialiser::error"); 1470 types_error = get_bool ("Types::Serialiser::error");
1078 1471
1472 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1473
1079 sv_cbor = newSVpv ("CBOR", 0); 1474 sv_cbor = newSVpv ("CBOR", 0);
1080 SvREADONLY_on (sv_cbor); 1475 SvREADONLY_on (sv_cbor);
1476
1477 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1081} 1478}
1082 1479
1083PROTOTYPES: DISABLE 1480PROTOTYPES: DISABLE
1084 1481
1085void CLONE (...) 1482void CLONE (...)
1104void shrink (CBOR *self, int enable = 1) 1501void shrink (CBOR *self, int enable = 1)
1105 ALIAS: 1502 ALIAS:
1106 shrink = F_SHRINK 1503 shrink = F_SHRINK
1107 allow_unknown = F_ALLOW_UNKNOWN 1504 allow_unknown = F_ALLOW_UNKNOWN
1108 allow_sharing = F_ALLOW_SHARING 1505 allow_sharing = F_ALLOW_SHARING
1506 allow_cycles = F_ALLOW_CYCLES
1109 dedup_strings = F_DEDUP_STRINGS 1507 pack_strings = F_PACK_STRINGS
1508 text_keys = F_TEXT_KEYS
1509 text_strings = F_TEXT_STRINGS
1510 validate_utf8 = F_VALIDATE_UTF8
1110 PPCODE: 1511 PPCODE:
1111{ 1512{
1112 if (enable) 1513 if (enable)
1113 self->flags |= ix; 1514 self->flags |= ix;
1114 else 1515 else
1120void get_shrink (CBOR *self) 1521void get_shrink (CBOR *self)
1121 ALIAS: 1522 ALIAS:
1122 get_shrink = F_SHRINK 1523 get_shrink = F_SHRINK
1123 get_allow_unknown = F_ALLOW_UNKNOWN 1524 get_allow_unknown = F_ALLOW_UNKNOWN
1124 get_allow_sharing = F_ALLOW_SHARING 1525 get_allow_sharing = F_ALLOW_SHARING
1526 get_allow_cycles = F_ALLOW_CYCLES
1125 get_dedup_strings = F_DEDUP_STRINGS 1527 get_pack_strings = F_PACK_STRINGS
1528 get_text_keys = F_TEXT_KEYS
1529 get_text_strings = F_TEXT_STRINGS
1530 get_validate_utf8 = F_VALIDATE_UTF8
1126 PPCODE: 1531 PPCODE:
1127 XPUSHs (boolSV (self->flags & ix)); 1532 XPUSHs (boolSV (self->flags & ix));
1128 1533
1129void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1534void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1130 PPCODE: 1535 PPCODE:
1143 XPUSHs (ST (0)); 1548 XPUSHs (ST (0));
1144 1549
1145int get_max_size (CBOR *self) 1550int get_max_size (CBOR *self)
1146 CODE: 1551 CODE:
1147 RETVAL = self->max_size; 1552 RETVAL = self->max_size;
1553 OUTPUT:
1554 RETVAL
1555
1556void filter (CBOR *self, SV *filter = 0)
1557 PPCODE:
1558 SvREFCNT_dec (self->filter);
1559 self->filter = filter ? newSVsv (filter) : filter;
1560 XPUSHs (ST (0));
1561
1562SV *get_filter (CBOR *self)
1563 CODE:
1564 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1148 OUTPUT: 1565 OUTPUT:
1149 RETVAL 1566 RETVAL
1150 1567
1151void encode (CBOR *self, SV *scalar) 1568void encode (CBOR *self, SV *scalar)
1152 PPCODE: 1569 PPCODE:
1167 EXTEND (SP, 2); 1584 EXTEND (SP, 2);
1168 PUSHs (sv); 1585 PUSHs (sv);
1169 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1586 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1170} 1587}
1171 1588
1589void incr_parse (CBOR *self, SV *cborstr)
1590 ALIAS:
1591 incr_parse_multiple = 1
1592 PPCODE:
1593{
1594 if (SvUTF8 (cborstr))
1595 sv_utf8_downgrade (cborstr, 0);
1596
1597 if (!self->incr_count)
1598 {
1599 self->incr_count = newAV ();
1600 self->incr_pos = 0;
1601 self->incr_need = 1;
1602
1603 av_push (self->incr_count, newSViv (1));
1604 }
1605
1606 do
1607 {
1608 if (!incr_parse (self, cborstr))
1609 {
1610 if (self->incr_need > self->max_size && self->max_size)
1611 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1612 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1613
1614 break;
1615 }
1616
1617 SV *sv;
1618 char *offset;
1619
1620 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1621 XPUSHs (sv);
1622
1623 sv_chop (cborstr, offset);
1624
1625 av_clear (self->incr_count);
1626 av_push (self->incr_count, newSViv (1));
1627
1628 self->incr_pos = 0;
1629 self->incr_need = self->incr_pos + 1;
1630 }
1631 while (ix);
1632}
1633
1634void incr_reset (CBOR *self)
1635 CODE:
1636{
1637 SvREFCNT_dec (self->incr_count);
1638 self->incr_count = 0;
1639}
1640
1641void DESTROY (CBOR *self)
1642 PPCODE:
1643 cbor_free (self);
1644
1172PROTOTYPES: ENABLE 1645PROTOTYPES: ENABLE
1173 1646
1174void encode_cbor (SV *scalar) 1647void encode_cbor (SV *scalar)
1648 ALIAS:
1649 encode_cbor = 0
1650 encode_cbor_sharing = F_ALLOW_SHARING
1175 PPCODE: 1651 PPCODE:
1176{ 1652{
1177 CBOR cbor; 1653 CBOR cbor;
1178 cbor_init (&cbor); 1654 cbor_init (&cbor);
1655 cbor.flags |= ix;
1179 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1656 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1180 XPUSHs (scalar); 1657 XPUSHs (scalar);
1181} 1658}
1182 1659
1183void decode_cbor (SV *cborstr) 1660void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines