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 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.34 by root, Sat Nov 30 16:19:59 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// known tags, rfc7049 14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known tags
15enum cbor_tag 29enum cbor_tag
16{ 30{
31 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39
40 // rfc7049
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 43 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 44 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 46 CBOR_TAG_BIGFLOAT = 5, // array
23 47
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 48 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 49 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 50 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 52
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 58
35 CBOR_TAG_MAGIC = 55799 59 CBOR_TAG_MAGIC = 55799 // self-describe cbor
36}; 60};
37 61
38#define F_SHRINK 0x00000200UL 62#define F_SHRINK 0x00000001UL
39#define F_ALLOW_UNKNOWN 0x00002000UL 63#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL
65#define F_PACK_STRINGS 0x00000008UL
40 66
41#define INIT_SIZE 32 // initial scalar size to be allocated 67#define INIT_SIZE 32 // initial scalar size to be allocated
42 68
43#define SB do { 69#define SB do {
44#define SE } while (0) 70#define SE } while (0)
55#else 81#else
56# define CBOR_SLOW 0 82# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 83# define CBOR_STASH cbor_stash
58#endif 84#endif
59 85
60static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 86static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
61static SV *cbor_true, *cbor_false; 87static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 88
63typedef struct { 89typedef struct {
64 U32 flags; 90 U32 flags;
65 U32 max_depth; 91 U32 max_depth;
66 STRLEN max_size; 92 STRLEN max_size;
93 SV *filter;
67} CBOR; 94} CBOR;
68 95
69ecb_inline void 96ecb_inline void
70cbor_init (CBOR *cbor) 97cbor_init (CBOR *cbor)
71{ 98{
72 Zero (cbor, 1, CBOR); 99 Zero (cbor, 1, CBOR);
73 cbor->max_depth = 512; 100 cbor->max_depth = 512;
101}
102
103ecb_inline void
104cbor_free (CBOR *cbor)
105{
106 SvREFCNT_dec (cbor->filter);
74} 107}
75 108
76///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
77// utility functions 110// utility functions
78 111
100 SvPV_renew (sv, SvCUR (sv) + 1); 133 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 134#endif
102 } 135 }
103} 136}
104 137
105///////////////////////////////////////////////////////////////////////////// 138// minimum length of a string to be registered for stringref
106// fp hell 139ecb_inline int
107 140minimum_string_length (UV idx)
108//TODO 141{
142 return idx > 23
143 ? idx > 0xffU
144 ? idx > 0xffffU
145 ? idx > 0xffffffffU
146 ? 11
147 : 7
148 : 5
149 : 4
150 : 3;
151}
109 152
110///////////////////////////////////////////////////////////////////////////// 153/////////////////////////////////////////////////////////////////////////////
111// encoder 154// encoder
112 155
113// structure used for encoding CBOR 156// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 159 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 160 char *end; // SvEND (sv)
118 SV *sv; // result scalar 161 SV *sv; // result scalar
119 CBOR cbor; 162 CBOR cbor;
120 U32 depth; // recursion level 163 U32 depth; // recursion level
164 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
165 UV stringref_idx;
166 HV *shareable; // ptr => index, or 0
167 UV shareable_idx;
121} enc_t; 168} enc_t;
122 169
123ecb_inline void 170ecb_inline void
124need (enc_t *enc, STRLEN len) 171need (enc_t *enc, STRLEN len)
125{ 172{
142static void 189static void
143encode_uint (enc_t *enc, int major, UV len) 190encode_uint (enc_t *enc, int major, UV len)
144{ 191{
145 need (enc, 9); 192 need (enc, 9);
146 193
147 if (len < 24) 194 if (ecb_expect_true (len < 24))
148 *enc->cur++ = major | len; 195 *enc->cur++ = major | len;
149 else if (len <= 0xff) 196 else if (ecb_expect_true (len <= 0xff))
150 { 197 {
151 *enc->cur++ = major | 24; 198 *enc->cur++ = major | 24;
152 *enc->cur++ = len; 199 *enc->cur++ = len;
153 } 200 }
154 else if (len <= 0xffff) 201 else if (len <= 0xffff)
177 *enc->cur++ = len >> 8; 224 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 225 *enc->cur++ = len;
179 } 226 }
180} 227}
181 228
182static void 229ecb_inline void
230encode_tag (enc_t *enc, UV tag)
231{
232 encode_uint (enc, 0xc0, tag);
233}
234
235ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 236encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 237{
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 238 encode_uint (enc, utf8 ? 0x60 : 0x40, len);
186 need (enc, len); 239 need (enc, len);
187 memcpy (enc->cur, str, len); 240 memcpy (enc->cur, str, len);
188 enc->cur += len; 241 enc->cur += len;
189} 242}
190 243
244static void
245encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
246{
247 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248 {
249 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250
251 if (SvOK (*svp))
252 {
253 // already registered, use stringref
254 encode_tag (enc, CBOR_TAG_STRINGREF);
255 encode_uint (enc, 0x00, SvUV (*svp));
256 return;
257 }
258 else if (len >= minimum_string_length (enc->stringref_idx))
259 {
260 // register only
261 sv_setuv (*svp, enc->stringref_idx);
262 ++enc->stringref_idx;
263 }
264 }
265
266 encode_str (enc, utf8, str, len);
267}
268
191static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
192 270
193static void 271static void
194encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
195{ 273{
232 while ((he = hv_iternext (hv))) 310 while ((he = hv_iternext (hv)))
233 { 311 {
234 if (HeKLEN (he) == HEf_SVKEY) 312 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 313 encode_sv (enc, HeSVKEY (he));
236 else 314 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 315 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 316
239 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 317 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
240 } 318 }
241 319
242 if (mg) 320 if (mg)
247 325
248// encode objects, arrays and special \0=false and \1=true values. 326// encode objects, arrays and special \0=false and \1=true values.
249static void 327static void
250encode_rv (enc_t *enc, SV *sv) 328encode_rv (enc_t *enc, SV *sv)
251{ 329{
252 svtype svt;
253
254 SvGETMAGIC (sv); 330 SvGETMAGIC (sv);
331
255 svt = SvTYPE (sv); 332 svtype svt = SvTYPE (sv);
256 333
257 if (ecb_expect_false (SvOBJECT (sv))) 334 if (ecb_expect_false (SvOBJECT (sv)))
258 { 335 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 336 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 337 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 338 : gv_stashpv ("Types::Serialiser::Boolean", 1);
339 HV *error_stash = !CBOR_SLOW || types_error_stash
340 ? types_error_stash
341 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 342 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 343 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 344 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 345
346 HV *stash = SvSTASH (sv);
347
266 if (SvSTASH (sv) == boolean_stash) 348 if (stash == boolean_stash)
349 {
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 350 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
351 return;
352 }
353 else if (stash == error_stash)
354 {
355 encode_ch (enc, 0xe0 | 23);
356 return;
357 }
268 else if (SvSTASH (sv) == tagged_stash) 358 else if (stash == tagged_stash)
269 { 359 {
270 if (svt != SVt_PVAV) 360 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 361 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 362
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 363 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 364 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
365
366 return;
367 }
368 }
369
370 if (ecb_expect_false (SvREFCNT (sv) > 1)
371 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
372 {
373 if (!enc->shareable)
374 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
375
376 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
377
378 if (SvOK (*svp))
379 {
380 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
381 encode_uint (enc, 0x00, SvUV (*svp));
382 return;
275 } 383 }
276 else 384 else
277 { 385 {
386 sv_setuv (*svp, enc->shareable_idx);
387 ++enc->shareable_idx;
388 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
389 }
390 }
391
392 if (ecb_expect_false (SvOBJECT (sv)))
393 {
394 HV *stash = SvSTASH (sv);
395 GV *method;
396
397 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
398 {
399 dSP;
400
401 ENTER; SAVETMPS; PUSHMARK (SP);
278 // we re-bless the reference to get overload and other niceties right 402 // we re-bless the reference to get overload and other niceties right
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 403 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 404
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 405 PUTBACK;
406 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 407 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 408 SPAGAIN;
292 409
293 // catch this surprisingly common error 410 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 411 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 412 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 413
297 sv = POPs; 414 encode_sv (enc, POPs);
415
298 PUTBACK; 416 PUTBACK;
299 417
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 418 FREETMPS; LEAVE;
303 } 419 }
420 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
421 {
422 dSP;
423
424 ENTER; SAVETMPS; PUSHMARK (SP);
425 EXTEND (SP, 2);
426 // we re-bless the reference to get overload and other niceties right
427 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
428 PUSHs (sv_cbor);
429
430 PUTBACK;
431 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
432 SPAGAIN;
433
434 // catch this surprisingly common error
435 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
436 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
437
438 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
439 encode_uint (enc, 0x80, count + 1);
440 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
441
442 while (count)
443 encode_sv (enc, SP[1 - count--]);
444
445 PUTBACK;
446
447 FREETMPS; LEAVE;
448 }
304 else 449 else
305 croak ("encountered object '%s', but no TO_CBOR method available on it", 450 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 451 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 452 }
309 else if (svt == SVt_PVHV) 453 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 454 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 455 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 456 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 457 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 458 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 459 encode_tag (enc, CBOR_TAG_INDIRECTION);
460 encode_sv (enc, sv);
461 }
333} 462}
334 463
335static void 464static void
336encode_nv (enc_t *enc, SV *sv) 465encode_nv (enc_t *enc, SV *sv)
337{ 466{
375 504
376 if (SvPOKp (sv)) 505 if (SvPOKp (sv))
377 { 506 {
378 STRLEN len; 507 STRLEN len;
379 char *str = SvPV (sv, len); 508 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 509 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 510 }
382 else if (SvNOKp (sv)) 511 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 512 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 513 else if (SvIOKp (sv))
385 { 514 {
402} 531}
403 532
404static SV * 533static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 534encode_cbor (SV *scalar, CBOR *cbor)
406{ 535{
407 enc_t enc; 536 enc_t enc = { };
408 537
409 enc.cbor = *cbor; 538 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 539 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 540 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 541 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 542
415 SvPOK_only (enc.sv); 543 SvPOK_only (enc.sv);
544
545 if (cbor->flags & F_PACK_STRINGS)
546 {
547 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
548 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
549 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
550 }
551
416 encode_sv (&enc, scalar); 552 encode_sv (&enc, scalar);
417 553
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 554 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 555 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 556
434 U8 *end; // end of input string 570 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 571 const char *err; // parse error, if != 0
436 CBOR cbor; 572 CBOR cbor;
437 U32 depth; // recursion depth 573 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 574 U32 maxdepth; // recursion depth limit
575 AV *shareable;
576 AV *stringref;
577 SV *decode_tagged;
439} dec_t; 578} dec_t;
440 579
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 580#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 581
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 582#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
475 | ((UV)dec->cur[-1]); 614 | ((UV)dec->cur[-1]);
476 615
477 case 27: 616 case 27:
478 WANT (9); 617 WANT (9);
479 dec->cur += 9; 618 dec->cur += 9;
619
620 return
621#if UVSIZE < 8
622 0
623#else
480 return (((UV)dec->cur[-8]) << 56) 624 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 625 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 626 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 627 | (((UV)dec->cur[-5]) << 32)
628#endif
484 | (((UV)dec->cur[-4]) << 24) 629 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 630 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 631 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 632 | ((UV)dec->cur[-1]);
488 633
537 SvREFCNT_dec (av); 682 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 683 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 684 return &PL_sv_undef;
540} 685}
541 686
687static void
688decode_he (dec_t *dec, HV *hv)
689{
690 // for speed reasons, we specialcase single-string
691 // byte or utf-8 strings as keys, but only when !stringref
692
693 if (ecb_expect_true (!dec->stringref))
694 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
695 {
696 I32 len = decode_uint (dec);
697 char *key = (char *)dec->cur;
698
699 dec->cur += len;
700
701 if (ecb_expect_false (dec->stringref))
702 av_push (dec->stringref, newSVpvn (key, len));
703
704 hv_store (hv, key, len, decode_sv (dec), 0);
705
706 return;
707 }
708 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
709 {
710 I32 len = decode_uint (dec);
711 char *key = (char *)dec->cur;
712
713 dec->cur += len;
714
715 if (ecb_expect_false (dec->stringref))
716 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
717
718 hv_store (hv, key, -len, decode_sv (dec), 0);
719
720 return;
721 }
722
723 SV *k = decode_sv (dec);
724 SV *v = decode_sv (dec);
725
726 hv_store_ent (hv, k, v, 0);
727 SvREFCNT_dec (k);
728}
729
542static SV * 730static SV *
543decode_hv (dec_t *dec) 731decode_hv (dec_t *dec)
544{ 732{
545 HV *hv = newHV (); 733 HV *hv = newHV ();
546 734
558 { 746 {
559 ++dec->cur; 747 ++dec->cur;
560 break; 748 break;
561 } 749 }
562 750
563 SV *k = decode_sv (dec); 751 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 752 }
568 } 753 }
569 else 754 else
570 { 755 {
571 int len = decode_uint (dec); 756 int pairs = decode_uint (dec);
572 757
573 while (len--) 758 while (pairs--)
574 { 759 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 760 }
581 761
582 DEC_DEC_DEPTH; 762 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 763 return newRV_noinc ((SV *)hv);
584 764
593{ 773{
594 SV *sv = 0; 774 SV *sv = 0;
595 775
596 if ((*dec->cur & 31) == 31) 776 if ((*dec->cur & 31) == 31)
597 { 777 {
778 // indefinite length strings
598 ++dec->cur; 779 ++dec->cur;
599 780
781 unsigned char major = *dec->cur & 0xe0;
782
600 sv = newSVpvn ("", 0); 783 sv = newSVpvn ("", 0);
601 784
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 785 for (;;)
604 { 786 {
605 WANT (1); 787 WANT (1);
606 788
789 if ((*dec->cur ^ major) >= 31)
607 if (*dec->cur == (0xe0 | 31)) 790 if (*dec->cur == (0xe0 | 31))
608 { 791 {
609 ++dec->cur; 792 ++dec->cur;
610 break; 793 break;
611 } 794 }
795 else
796 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 797
613 sv_catsv (sv, decode_sv (dec)); 798 STRLEN len = decode_uint (dec);
799
800 WANT (len);
801 sv_catpvn (sv, dec->cur, len);
802 dec->cur += len;
614 } 803 }
615 } 804 }
616 else 805 else
617 { 806 {
618 STRLEN len = decode_uint (dec); 807 STRLEN len = decode_uint (dec);
619 808
620 WANT (len); 809 WANT (len);
621 sv = newSVpvn (dec->cur, len); 810 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 811 dec->cur += len;
812
813 if (ecb_expect_false (dec->stringref)
814 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
815 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 816 }
624 817
625 if (utf8) 818 if (utf8)
626 SvUTF8_on (sv); 819 SvUTF8_on (sv);
627 820
633} 826}
634 827
635static SV * 828static SV *
636decode_tagged (dec_t *dec) 829decode_tagged (dec_t *dec)
637{ 830{
831 SV *sv = 0;
638 UV tag = decode_uint (dec); 832 UV tag = decode_uint (dec);
833
834 WANT (1);
835
836 switch (tag)
837 {
838 case CBOR_TAG_MAGIC:
839 sv = decode_sv (dec);
840 break;
841
842 case CBOR_TAG_INDIRECTION:
843 sv = newRV_noinc (decode_sv (dec));
844 break;
845
846 case CBOR_TAG_STRINGREF_NAMESPACE:
847 {
848 ENTER; SAVETMPS;
849
850 SAVESPTR (dec->stringref);
851 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
852
853 sv = decode_sv (dec);
854
855 FREETMPS; LEAVE;
856 }
857 break;
858
859 case CBOR_TAG_STRINGREF:
860 {
861 if ((*dec->cur >> 5) != 0)
862 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
863
864 UV idx = decode_uint (dec);
865
866 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
867 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
868
869 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
870 }
871 break;
872
873 case CBOR_TAG_VALUE_SHAREABLE:
874 {
875 if (ecb_expect_false (!dec->shareable))
876 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
877
878 sv = newSV (0);
879 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
880
639 SV *sv = decode_sv (dec); 881 SV *osv = decode_sv (dec);
882 sv_setsv (sv, osv);
883 SvREFCNT_dec_NN (osv);
884 }
885 break;
640 886
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 887 case CBOR_TAG_VALUE_SHAREDREF:
888 {
889 if ((*dec->cur >> 5) != 0)
890 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
891
892 UV idx = decode_uint (dec);
893
894 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
895 ERR ("corrupted CBOR data (sharedref index out of bounds)");
896
897 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
898 }
899 break;
900
901 case CBOR_TAG_PERL_OBJECT:
902 {
903 sv = decode_sv (dec);
904
905 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
906 ERR ("corrupted CBOR data (non-array perl object)");
907
908 AV *av = (AV *)SvRV (sv);
909 int len = av_len (av) + 1;
910 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
911
912 if (!stash)
913 ERR ("cannot decode perl-object (package does not exist)");
914
915 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
916
917 if (!method)
918 ERR ("cannot decode perl-object (package does not have a THAW method)");
919
920 dSP;
921
922 ENTER; SAVETMPS; PUSHMARK (SP);
923 EXTEND (SP, len + 1);
924 // we re-bless the reference to get overload and other niceties right
925 PUSHs (*av_fetch (av, 0, 1));
926 PUSHs (sv_cbor);
927
928 int i;
929
930 for (i = 1; i < len; ++i)
931 PUSHs (*av_fetch (av, i, 1));
932
933 PUTBACK;
934 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
935 SPAGAIN;
936
937 if (SvTRUE (ERRSV))
938 {
939 FREETMPS; LEAVE;
940 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
941 }
942
943 SvREFCNT_dec (sv);
944 sv = SvREFCNT_inc (POPs);
945
946 PUTBACK;
947
948 FREETMPS; LEAVE;
949 }
950 break;
951
952 default:
953 {
954 sv = decode_sv (dec);
955
956 dSP;
957 ENTER; SAVETMPS; PUSHMARK (SP);
958 EXTEND (SP, 2);
959 PUSHs (newSVuv (tag));
960 PUSHs (sv);
961
962 PUTBACK;
963 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
964 SPAGAIN;
965
966 if (SvTRUE (ERRSV))
967 {
968 FREETMPS; LEAVE;
969 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
970 }
971
972 if (count)
973 {
974 SvREFCNT_dec (sv);
975 sv = SvREFCNT_inc (POPs);
976 }
977 else
978 {
979 AV *av = newAV ();
980 av_push (av, newSVuv (tag));
981 av_push (av, sv);
982
983 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
984 ? cbor_tagged_stash
985 : gv_stashpv ("CBOR::XS::Tagged" , 1);
986 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
987 }
988
989 PUTBACK;
990
991 FREETMPS; LEAVE;
992 }
993 break;
994 }
995
642 return sv; 996 return sv;
643 997
644 AV *av = newAV (); 998fail:
645 av_push (av, newSVuv (tag)); 999 SvREFCNT_dec (sv);
646 av_push (av, sv); 1000 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 1001}
654 1002
655static SV * 1003static SV *
656decode_sv (dec_t *dec) 1004decode_sv (dec_t *dec)
657{ 1005{
676 case 7: // misc 1024 case 7: // misc
677 switch (*dec->cur++ & 31) 1025 switch (*dec->cur++ & 31)
678 { 1026 {
679 case 20: 1027 case 20:
680#if CBOR_SLOW 1028#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1029 types_false = get_bool ("Types::Serialiser::false");
682#endif 1030#endif
683 return newSVsv (cbor_false); 1031 return newSVsv (types_false);
684 case 21: 1032 case 21:
685#if CBOR_SLOW 1033#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1034 types_true = get_bool ("Types::Serialiser::true");
687#endif 1035#endif
688 return newSVsv (cbor_true); 1036 return newSVsv (types_true);
689 case 22: 1037 case 22:
690 return newSVsv (&PL_sv_undef); 1038 return newSVsv (&PL_sv_undef);
1039 case 23:
1040#if CBOR_SLOW
1041 types_error = get_bool ("Types::Serialiser::error");
1042#endif
1043 return newSVsv (types_error);
691 1044
692 case 25: 1045 case 25:
693 { 1046 {
694 WANT (2); 1047 WANT (2);
695 1048
739} 1092}
740 1093
741static SV * 1094static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1095decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1096{
744 dec_t dec; 1097 dec_t dec = { };
745 SV *sv; 1098 SV *sv;
1099 STRLEN len;
1100 char *data = SvPVbyte (string, len);
746 1101
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1102 if (len > cbor->max_size && cbor->max_size)
776 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1103 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1104 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1105
782 dec.cbor = *cbor; 1106 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1107 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1108 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1109
788 sv = decode_sv (&dec); 1110 sv = decode_sv (&dec);
789 1111
790 if (offset_return) 1112 if (offset_return)
791 *offset_return = dec.cur; 1113 *offset_return = dec.cur;
795 dec.err = "garbage after CBOR object"; 1117 dec.err = "garbage after CBOR object";
796 1118
797 if (dec.err) 1119 if (dec.err)
798 { 1120 {
799 SvREFCNT_dec (sv); 1121 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1122 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1123 }
802 1124
803 sv = sv_2mortal (sv); 1125 sv = sv_2mortal (sv);
804 1126
805 return sv; 1127 return sv;
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1133MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1134
813BOOT: 1135BOOT:
814{ 1136{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1137 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1138 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1139
819 cbor_true = get_bool ("CBOR::XS::true"); 1140 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1141 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1142
1143 types_true = get_bool ("Types::Serialiser::true" );
1144 types_false = get_bool ("Types::Serialiser::false");
1145 types_error = get_bool ("Types::Serialiser::error");
1146
1147 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1148
1149 sv_cbor = newSVpv ("CBOR", 0);
1150 SvREADONLY_on (sv_cbor);
821} 1151}
822 1152
823PROTOTYPES: DISABLE 1153PROTOTYPES: DISABLE
824 1154
825void CLONE (...) 1155void CLONE (...)
826 CODE: 1156 CODE:
827 cbor_stash = 0; 1157 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1158 cbor_tagged_stash = 0;
1159 types_error_stash = 0;
1160 types_boolean_stash = 0;
830 1161
831void new (char *klass) 1162void new (char *klass)
832 PPCODE: 1163 PPCODE:
833{ 1164{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1165 SV *pv = NEWSV (0, sizeof (CBOR));
842 1173
843void shrink (CBOR *self, int enable = 1) 1174void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1175 ALIAS:
845 shrink = F_SHRINK 1176 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1177 allow_unknown = F_ALLOW_UNKNOWN
1178 allow_sharing = F_ALLOW_SHARING
1179 pack_strings = F_PACK_STRINGS
847 PPCODE: 1180 PPCODE:
848{ 1181{
849 if (enable) 1182 if (enable)
850 self->flags |= ix; 1183 self->flags |= ix;
851 else 1184 else
856 1189
857void get_shrink (CBOR *self) 1190void get_shrink (CBOR *self)
858 ALIAS: 1191 ALIAS:
859 get_shrink = F_SHRINK 1192 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1193 get_allow_unknown = F_ALLOW_UNKNOWN
1194 get_allow_sharing = F_ALLOW_SHARING
1195 get_pack_strings = F_PACK_STRINGS
861 PPCODE: 1196 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1197 XPUSHs (boolSV (self->flags & ix));
863 1198
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1199void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1200 PPCODE:
881 CODE: 1216 CODE:
882 RETVAL = self->max_size; 1217 RETVAL = self->max_size;
883 OUTPUT: 1218 OUTPUT:
884 RETVAL 1219 RETVAL
885 1220
886#if 0 //TODO 1221void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1222 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1223 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1224 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1225 XPUSHs (ST (0));
895}
896 1226
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1227SV *get_filter (CBOR *self)
898 PPCODE: 1228 CODE:
899{ 1229 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1230 OUTPUT:
901 self->cb_sk_object = newHV (); 1231 RETVAL
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1232
921void encode (CBOR *self, SV *scalar) 1233void encode (CBOR *self, SV *scalar)
922 PPCODE: 1234 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1235 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1236 XPUSHs (scalar);
937 EXTEND (SP, 2); 1249 EXTEND (SP, 2);
938 PUSHs (sv); 1250 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1251 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1252}
941 1253
942#if 0
943
944void DESTROY (CBOR *self) 1254void DESTROY (CBOR *self)
945 CODE: 1255 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1256 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1257
951PROTOTYPES: ENABLE 1258PROTOTYPES: ENABLE
952 1259
953void encode_cbor (SV *scalar) 1260void encode_cbor (SV *scalar)
954 PPCODE: 1261 PPCODE:

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines