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 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.20 by root, Wed Nov 20 11:06:42 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// 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
25// known tags
26enum cbor_tag
27{
28 // inofficial extensions (pending iana registration)
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // 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
36
37 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array
44
45 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor
57};
58
14#define F_SHRINK 0x00000200UL 59#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 60#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO
62#define F_DEDUP_STRINGS 0x00000008UL //TODO
63#define F_DEDUP_KEYS 0x00000010UL //TODO
16 64
17#define INIT_SIZE 32 // initial scalar size to be allocated 65#define INIT_SIZE 32 // initial scalar size to be allocated
18 66
19#define SB do { 67#define SB do {
20#define SE } while (0) 68#define SE } while (0)
… …
31#else 79#else
32# define CBOR_SLOW 0 80# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 81# define CBOR_STASH cbor_stash
34#endif 82#endif
35 83
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 85static SV *types_true, *types_false, *types_error, *sv_cbor;
38 86
39typedef struct { 87typedef struct {
40 U32 flags; 88 U32 flags;
41 U32 max_depth; 89 U32 max_depth;
42 STRLEN max_size; 90 STRLEN max_size;
43
44 SV *cb_object;
45 HV *cb_sk_object;
46} CBOR; 91} CBOR;
47 92
48ecb_inline void 93ecb_inline void
49cbor_init (CBOR *cbor) 94cbor_init (CBOR *cbor)
50{ 95{
… …
80#endif 125#endif
81 } 126 }
82} 127}
83 128
84///////////////////////////////////////////////////////////////////////////// 129/////////////////////////////////////////////////////////////////////////////
85// fp hell
86
87//TODO
88
89/////////////////////////////////////////////////////////////////////////////
90// encoder 130// encoder
91 131
92// structure used for encoding CBOR 132// structure used for encoding CBOR
93typedef struct 133typedef struct
94{ 134{
95 char *cur; // SvPVX (sv) + current output position 135 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 136 char *end; // SvEND (sv)
97 SV *sv; // result scalar 137 SV *sv; // result scalar
98 CBOR cbor; 138 CBOR cbor;
99 U32 depth; // recursion level 139 U32 depth; // recursion level
140 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
141 UV stringref_idx;
142 HV *shareable; // ptr => index, or 0
143 UV shareable_idx;
100} enc_t; 144} enc_t;
101 145
102ecb_inline void 146ecb_inline void
103need (enc_t *enc, STRLEN len) 147need (enc_t *enc, STRLEN len)
104{ 148{
… …
165 need (enc, len); 209 need (enc, len);
166 memcpy (enc->cur, str, len); 210 memcpy (enc->cur, str, len);
167 enc->cur += len; 211 enc->cur += len;
168} 212}
169 213
214ecb_inline void
215encode_tag (enc_t *enc, UV tag)
216{
217 encode_uint (enc, 0xc0, tag);
218}
219
170static void encode_sv (enc_t *enc, SV *sv); 220static void encode_sv (enc_t *enc, SV *sv);
171 221
172static void 222static void
173encode_av (enc_t *enc, AV *av) 223encode_av (enc_t *enc, AV *av)
174{ 224{
… …
186 SV **svp = av_fetch (av, i, 0); 236 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 237 encode_sv (enc, svp ? *svp : &PL_sv_undef);
188 } 238 }
189 239
190 --enc->depth; 240 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
191} 250}
192 251
193static void 252static void
194encode_hv (enc_t *enc, HV *hv) 253encode_hv (enc_t *enc, HV *hv)
195{ 254{
… …
208 else 267 else
209 encode_uint (enc, 0xa0, pairs); 268 encode_uint (enc, 0xa0, pairs);
210 269
211 while ((he = hv_iternext (hv))) 270 while ((he = hv_iternext (hv)))
212 { 271 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
213 if (HeKLEN (he) == HEf_SVKEY) 276 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
215 else 293 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 294 encode_he (enc, he);
217 295
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 296 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 297 }
220 298
221 if (mg) 299 if (mg)
… …
226 304
227// encode objects, arrays and special \0=false and \1=true values. 305// encode objects, arrays and special \0=false and \1=true values.
228static void 306static void
229encode_rv (enc_t *enc, SV *sv) 307encode_rv (enc_t *enc, SV *sv)
230{ 308{
231 svtype svt;
232
233 SvGETMAGIC (sv); 309 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332
234 svt = SvTYPE (sv); 333 svtype svt = SvTYPE (sv);
235 334
236 if (ecb_expect_false (SvOBJECT (sv))) 335 if (ecb_expect_false (SvOBJECT (sv)))
237 { 336 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 337 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 338 ? types_boolean_stash
339 : gv_stashpv ("Types::Serialiser::Boolean", 1);
340 HV *error_stash = !CBOR_SLOW || types_error_stash
341 ? types_error_stash
342 : gv_stashpv ("Types::Serialiser::Error", 1);
343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 345 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 346
242 if (SvSTASH (sv) == stash) 347 HV *stash = SvSTASH (sv);
348 GV *method;
349
350 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 351 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
352 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23);
354 else if (stash == tagged_stash)
355 {
356 if (svt != SVt_PVAV)
357 croak ("encountered CBOR::XS::Tagged object that isn't an array");
358
359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
361 }
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 {
364 dSP;
365
366 ENTER; SAVETMPS; PUSHMARK (SP);
367 // we re-bless the reference to get overload and other niceties right
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369
370 PUTBACK;
371 // G_SCALAR ensures that return value is 1
372 call_sv ((SV *)GvCV (method), G_SCALAR);
373 SPAGAIN;
374
375 // catch this surprisingly common error
376 if (SvROK (TOPs) && SvRV (TOPs) == sv)
377 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
378
379 encode_sv (enc, POPs);
380
381 PUTBACK;
382
383 FREETMPS; LEAVE;
384 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 {
387 dSP;
388
389 ENTER; SAVETMPS; PUSHMARK (SP);
390 EXTEND (SP, 2);
391 // we re-bless the reference to get overload and other niceties right
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor);
394
395 PUTBACK;
396 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
397 SPAGAIN;
398
399 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406
407 while (count)
408 encode_sv (enc, SP[1 - count--]);
409
410 PUTBACK;
411
412 FREETMPS; LEAVE;
413 }
244 else 414 else
245 {
246#if 0 //TODO
247 if (enc->cbor.flags & F_CONV_BLESSED)
248 {
249 // we re-bless the reference to get overload and other niceties right
250 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252 if (to_cbor)
253 {
254 dSP;
255
256 ENTER; SAVETMPS; PUSHMARK (SP);
257 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
258
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK;
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
262 SPAGAIN;
263
264 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv)
266 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
267
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv);
272
273 FREETMPS; LEAVE;
274 }
275 else if (enc->cbor.flags & F_ALLOW_BLESSED)
276 encode_str (enc, "null", 4, 0);
277 else
278 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 415 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
279 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED)
282 encode_str (enc, "null", 4, 0);
283 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 416 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 417 }
289 else if (svt == SVt_PVHV) 418 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 419 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 420 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 421 encode_av (enc, (AV *)sv);
293 else if (svt < SVt_PVAV)
294 {
295 STRLEN len = 0;
296 char *pv = svt ? SvPV (sv, len) : 0;
297
298 if (len == 1 && *pv == '1')
299 encode_ch (enc, 0xe0 | 21);
300 else if (len == 1 && *pv == '0')
301 encode_ch (enc, 0xe0 | 20);
302 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303 encode_ch (enc, 0xe0 | 23);
304 else
305 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309 encode_ch (enc, 0xe0 | 23);
310 else 422 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 423 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 424 encode_tag (enc, CBOR_TAG_INDIRECTION);
425 encode_sv (enc, sv);
426 }
313} 427}
314 428
315static void 429static void
316encode_nv (enc_t *enc, SV *sv) 430encode_nv (enc_t *enc, SV *sv)
317{ 431{
… …
382} 496}
383 497
384static SV * 498static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 499encode_cbor (SV *scalar, CBOR *cbor)
386{ 500{
387 enc_t enc; 501 enc_t enc = { };
388 502
389 enc.cbor = *cbor; 503 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 505 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 506 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 507
395 SvPOK_only (enc.sv); 508 SvPOK_only (enc.sv);
509
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))
511 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 }
516
396 encode_sv (&enc, scalar); 517 encode_sv (&enc, scalar);
397 518
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 519 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 520 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 521
… …
414 U8 *end; // end of input string 535 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 536 const char *err; // parse error, if != 0
416 CBOR cbor; 537 CBOR cbor;
417 U32 depth; // recursion depth 538 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 539 U32 maxdepth; // recursion depth limit
540 AV *shareable;
541 AV *stringref;
419} dec_t; 542} dec_t;
420 543
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 544#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422 545
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 546#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
… …
517 SvREFCNT_dec (av); 640 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 641 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 642 return &PL_sv_undef;
520} 643}
521 644
645static void
646decode_he (dec_t *dec, HV *hv)
647{
648 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys.
650
651 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
652 {
653 I32 len = decode_uint (dec);
654 char *key = (char *)dec->cur;
655
656 dec->cur += len;
657
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0);
662 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
664 {
665 I32 len = decode_uint (dec);
666 char *key = (char *)dec->cur;
667
668 dec->cur += len;
669
670 if (ecb_expect_false (dec->stringref))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
672
673 hv_store (hv, key, -len, decode_sv (dec), 0);
674 }
675 else
676 {
677 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec);
679
680 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k);
682 }
683}
684
522static SV * 685static SV *
523decode_hv (dec_t *dec) 686decode_hv (dec_t *dec)
524{ 687{
525 HV *hv = newHV (); 688 HV *hv = newHV ();
526 689
… …
538 { 701 {
539 ++dec->cur; 702 ++dec->cur;
540 break; 703 break;
541 } 704 }
542 705
543 SV *k = decode_sv (dec); 706 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 707 }
548 } 708 }
549 else 709 else
550 { 710 {
551 int len = decode_uint (dec); 711 int pairs = decode_uint (dec);
552 712
553 while (len--) 713 while (pairs--)
554 { 714 decode_he (dec, hv);
555 SV *k = decode_sv (dec);
556 SV *v = decode_sv (dec);
557
558 hv_store_ent (hv, k, v, 0);
559 }
560 } 715 }
561 716
562 DEC_DEC_DEPTH; 717 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 718 return newRV_noinc ((SV *)hv);
564
565#if 0
566 SV *sv;
567 HV *hv = newHV ();
568
569 DEC_INC_DEPTH;
570 decode_ws (dec);
571
572 for (;;)
573 {
574 // heuristic: assume that
575 // a) decode_str + hv_store_ent are abysmally slow.
576 // b) most hash keys are short, simple ascii text.
577 // => try to "fast-match" such strings to avoid
578 // the overhead of decode_str + hv_store_ent.
579 {
580 SV *value;
581 char *p = dec->cur;
582 char *e = p + 24; // only try up to 24 bytes
583
584 for (;;)
585 {
586 // the >= 0x80 is false on most architectures
587 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
588 {
589 // slow path, back up and use decode_str
590 SV *key = decode_str (dec);
591 if (!key)
592 goto fail;
593
594 decode_ws (dec); EXPECT_CH (':');
595
596 decode_ws (dec);
597 value = decode_sv (dec);
598 if (!value)
599 {
600 SvREFCNT_dec (key);
601 goto fail;
602 }
603
604 hv_store_ent (hv, key, value, 0);
605 SvREFCNT_dec (key);
606
607 break;
608 }
609 else if (*p == '"')
610 {
611 // fast path, got a simple key
612 char *key = dec->cur;
613 int len = p - key;
614 dec->cur = p + 1;
615
616 decode_ws (dec); EXPECT_CH (':');
617
618 decode_ws (dec);
619 value = decode_sv (dec);
620 if (!value)
621 goto fail;
622
623 hv_store (hv, key, len, value, 0);
624
625 break;
626 }
627
628 ++p;
629 }
630 }
631
632 decode_ws (dec);
633
634 if (*dec->cur == '}')
635 {
636 ++dec->cur;
637 break;
638 }
639
640 if (*dec->cur != ',')
641 ERR (", or } expected while parsing object/hash");
642
643 ++dec->cur;
644
645 decode_ws (dec);
646
647 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648 {
649 ++dec->cur;
650 break;
651 }
652 }
653
654 DEC_DEC_DEPTH;
655 sv = newRV_noinc ((SV *)hv);
656
657 // check filter callbacks
658 if (dec->cbor.flags & F_HOOK)
659 {
660 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661 {
662 HE *cb, *he;
663
664 hv_iterinit (hv);
665 he = hv_iternext (hv);
666 hv_iterinit (hv);
667
668 // the next line creates a mortal sv each time its called.
669 // might want to optimise this for common cases.
670 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672 if (cb)
673 {
674 dSP;
675 int count;
676
677 ENTER; SAVETMPS; PUSHMARK (SP);
678 XPUSHs (HeVAL (he));
679 sv_2mortal (sv);
680
681 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683 if (count == 1)
684 {
685 sv = newSVsv (POPs);
686 FREETMPS; LEAVE;
687 return sv;
688 }
689
690 SvREFCNT_inc (sv);
691 FREETMPS; LEAVE;
692 }
693 }
694
695 if (dec->cbor.cb_object)
696 {
697 dSP;
698 int count;
699
700 ENTER; SAVETMPS; PUSHMARK (SP);
701 XPUSHs (sv_2mortal (sv));
702
703 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705 if (count == 1)
706 {
707 sv = newSVsv (POPs);
708 FREETMPS; LEAVE;
709 return sv;
710 }
711
712 SvREFCNT_inc (sv);
713 FREETMPS; LEAVE;
714 }
715 }
716
717 return sv;
718#endif
719 719
720fail: 720fail:
721 SvREFCNT_dec (hv); 721 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 722 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 723 return &PL_sv_undef;
724} 724}
725 725
726static SV * 726static SV *
727decode_str (dec_t *dec, int utf8) 727decode_str (dec_t *dec, int utf8)
728{ 728{
729 SV *sv; 729 SV *sv = 0;
730 730
731 if ((*dec->cur & 31) == 31) 731 if ((*dec->cur & 31) == 31)
732 { 732 {
733 ++dec->cur; 733 ++dec->cur;
734 734
… …
743 { 743 {
744 ++dec->cur; 744 ++dec->cur;
745 break; 745 break;
746 } 746 }
747 747
748 SV *sv2 = decode_sv (dec);
749 sv_catsv (sv, sv2); 748 sv_catsv (sv, decode_sv (dec));
750 } 749 }
751 } 750 }
752 else 751 else
753 { 752 {
754 STRLEN len = decode_uint (dec); 753 STRLEN len = decode_uint (dec);
… …
759 } 758 }
760 759
761 if (utf8) 760 if (utf8)
762 SvUTF8_on (sv); 761 SvUTF8_on (sv);
763 762
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765
764 return sv; 766 return sv;
765 767
766fail: 768fail:
769 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 770 return &PL_sv_undef;
768} 771}
769 772
770static SV * 773static SV *
771decode_tagged (dec_t *dec) 774decode_tagged (dec_t *dec)
772{ 775{
776 SV *sv = 0;
773 UV tag = decode_uint (dec); 777 UV tag = decode_uint (dec);
778
779 WANT (1);
780
781 switch (tag)
782 {
783 case CBOR_TAG_MAGIC:
784 sv = decode_sv (dec);
785 break;
786
787 case CBOR_TAG_INDIRECTION:
788 sv = newRV_noinc (decode_sv (dec));
789 break;
790
791 case CBOR_TAG_STRINGREF_NAMESPACE:
792 {
793 ENTER; SAVETMPS;
794
795 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797
798 sv = decode_sv (dec);
799
800 FREETMPS; LEAVE;
801 }
802 break;
803
804 case CBOR_TAG_STRINGREF:
805 {
806 if ((*dec->cur >> 5) != 0)
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808
809 UV idx = decode_uint (dec);
810
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
812 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
813
814 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
815 }
816 break;
817
818 case CBOR_TAG_VALUE_SHAREABLE:
819 {
820 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822
823 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825
774 SV *sv = decode_sv (dec); 826 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (osv);
829 }
830 break;
775 831
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 832 case CBOR_TAG_VALUE_SHAREDREF:
833 {
834 if ((*dec->cur >> 5) != 0)
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836
837 UV idx = decode_uint (dec);
838
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
843 }
844 break;
845
846 case CBOR_TAG_PERL_OBJECT:
847 {
848 sv = decode_sv (dec);
849
850 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851 ERR ("corrupted CBOR data (non-array perl object)");
852
853 AV *av = (AV *)SvRV (sv);
854 int len = av_len (av) + 1;
855 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
856
857 if (!stash)
858 ERR ("cannot decode perl-object (package does not exist)");
859
860 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
861
862 if (!method)
863 ERR ("cannot decode perl-object (package does not have a THAW method)");
864
865 dSP;
866
867 ENTER; SAVETMPS; PUSHMARK (SP);
868 EXTEND (SP, len + 1);
869 // we re-bless the reference to get overload and other niceties right
870 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor);
872
873 int i;
874
875 for (i = 1; i < len; ++i)
876 PUSHs (*av_fetch (av, i, 1));
877
878 PUTBACK;
879 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
880 SPAGAIN;
881
882 if (SvTRUE (ERRSV))
883 {
884 FREETMPS; LEAVE;
885 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
886 }
887
888 SvREFCNT_dec (sv);
889 sv = SvREFCNT_inc (POPs);
890
891 PUTBACK;
892
893 FREETMPS; LEAVE;
894 }
895 break;
896
897 default:
898 {
899 sv = decode_sv (dec);
900
901 AV *av = newAV ();
902 av_push (av, newSVuv (tag));
903 av_push (av, sv);
904
905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906 ? cbor_tagged_stash
907 : gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
910 }
911 break;
912 }
913
777 return sv; 914 return sv;
778 915
779 AV *av = newAV (); 916fail:
780 av_push (av, newSVuv (tag)); 917 SvREFCNT_dec (sv);
781 av_push (av, sv); 918 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 919}
784 920
785static SV * 921static SV *
786decode_sv (dec_t *dec) 922decode_sv (dec_t *dec)
787{ 923{
788 WANT (1); 924 WANT (1);
789 925
790 switch (*dec->cur >> 5) 926 switch (*dec->cur >> 5)
791 { 927 {
792 case 0: // unsigned int 928 case 0: // unsigned int
793 //TODO: 64 bit values on 3 2bit perls
794 return newSVuv (decode_uint (dec)); 929 return newSVuv (decode_uint (dec));
795 case 1: // negative int 930 case 1: // negative int
796 return newSViv (-1 - (IV)decode_uint (dec)); 931 return newSViv (-1 - (IV)decode_uint (dec));
797 case 2: // octet string 932 case 2: // octet string
798 return decode_str (dec, 0); 933 return decode_str (dec, 0);
… …
807 case 7: // misc 942 case 7: // misc
808 switch (*dec->cur++ & 31) 943 switch (*dec->cur++ & 31)
809 { 944 {
810 case 20: 945 case 20:
811#if CBOR_SLOW 946#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 947 types_false = get_bool ("Types::Serialiser::false");
813#endif 948#endif
814 return newSVsv (cbor_false); 949 return newSVsv (types_false);
815 case 21: 950 case 21:
816#if CBOR_SLOW 951#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 952 types_true = get_bool ("Types::Serialiser::true");
818#endif 953#endif
819 return newSVsv (cbor_true); 954 return newSVsv (types_true);
820 case 22: 955 case 22:
821 return newSVsv (&PL_sv_undef); 956 return newSVsv (&PL_sv_undef);
957 case 23:
958#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error");
960#endif
961 return newSVsv (types_error);
822 962
823 case 25: 963 case 25:
824 { 964 {
825 WANT (2); 965 WANT (2);
826 966
… …
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1002 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863 } 1003 }
864 1004
865 break; 1005 break;
866 } 1006 }
867#if 0
868 switch (*dec->cur)
869 {
870 //case '"': ++dec->cur; return decode_str (dec);
871 case '[': ++dec->cur; return decode_av (dec);
872 case '{': ++dec->cur; return decode_hv (dec);
873
874 case '-':
875 case '0': case '1': case '2': case '3': case '4':
876 case '5': case '6': case '7': case '8': case '9':
877 //TODO return decode_num (dec);
878
879 case 't':
880 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881 {
882 dec->cur += 4;
883#if CBOR_SLOW
884 cbor_true = get_bool ("CBOR::XS::true");
885#endif
886 return newSVsv (cbor_true);
887 }
888 else
889 ERR ("'true' expected");
890
891 break;
892
893 case 'f':
894 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895 {
896 dec->cur += 5;
897#if CBOR_SLOW
898 cbor_false = get_bool ("CBOR::XS::false");
899#endif
900 return newSVsv (cbor_false);
901 }
902 else
903 ERR ("'false' expected");
904
905 break;
906
907 case 'n':
908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909 {
910 dec->cur += 4;
911 return newSVsv (&PL_sv_undef);
912 }
913 else
914 ERR ("'null' expected");
915
916 break;
917
918 default:
919 ERR ("malformed CBOR string, neither array, object, number, string or atom");
920 break;
921 }
922#endif
923 1007
924fail: 1008fail:
925 return &PL_sv_undef; 1009 return &PL_sv_undef;
926} 1010}
927 1011
928static SV * 1012static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1013decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1014{
931 dec_t dec; 1015 dec_t dec = { };
932 SV *sv; 1016 SV *sv;
1017 STRLEN len;
1018 char *data = SvPVbyte (string, len);
933 1019
934 /* work around bugs in 5.10 where manipulating magic values
935 * makes perl ignore the magic in subsequent accesses.
936 * also make a copy of non-PV values, to get them into a clean
937 * state (SvPV should do that, but it's buggy, see below).
938 */
939 /*SvGETMAGIC (string);*/
940 if (SvMAGICAL (string) || !SvPOK (string))
941 string = sv_2mortal (newSVsv (string));
942
943 SvUPGRADE (string, SVt_PV);
944
945 /* work around a bug in perl 5.10, which causes SvCUR to fail an
946 * assertion with -DDEBUGGING, although SvCUR is documented to
947 * return the xpv_cur field which certainly exists after upgrading.
948 * according to nicholas clark, calling SvPOK fixes this.
949 * But it doesn't fix it, so try another workaround, call SvPV_nolen
950 * and hope for the best.
951 * Damnit, SvPV_nolen still trips over yet another assertion. This
952 * assertion business is seriously broken, try yet another workaround
953 * for the broken -DDEBUGGING.
954 */
955 {
956#ifdef DEBUGGING
957 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958#else
959 STRLEN offset = SvCUR (string);
960#endif
961
962 if (offset > cbor->max_size && cbor->max_size) 1020 if (len > cbor->max_size && cbor->max_size)
963 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1021 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1022 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1023
969 dec.cbor = *cbor; 1024 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1025 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1026 dec.end = (U8 *)data + len;
972 dec.err = 0;
973 dec.depth = 0;
974
975 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976 ;//TODO dec.cbor.flags |= F_HOOK;
977 1027
978 sv = decode_sv (&dec); 1028 sv = decode_sv (&dec);
979 1029
980 if (offset_return) 1030 if (offset_return)
981 *offset_return = dec.cur; 1031 *offset_return = dec.cur;
… …
985 dec.err = "garbage after CBOR object"; 1035 dec.err = "garbage after CBOR object";
986 1036
987 if (dec.err) 1037 if (dec.err)
988 { 1038 {
989 SvREFCNT_dec (sv); 1039 SvREFCNT_dec (sv);
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1041 }
992 1042
993 sv = sv_2mortal (sv); 1043 sv = sv_2mortal (sv);
994 1044
995 return sv; 1045 return sv;
… …
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1051MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1052
1003BOOT: 1053BOOT:
1004{ 1054{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1055 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1056 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1057
1008 cbor_true = get_bool ("CBOR::XS::true"); 1058 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1059 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1060
1061 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error");
1064
1065 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor);
1010} 1067}
1011 1068
1012PROTOTYPES: DISABLE 1069PROTOTYPES: DISABLE
1013 1070
1014void CLONE (...) 1071void CLONE (...)
1015 CODE: 1072 CODE:
1016 cbor_stash = 0; 1073 cbor_stash = 0;
1074 cbor_tagged_stash = 0;
1075 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1076 types_boolean_stash = 0;
1018 1077
1019void new (char *klass) 1078void new (char *klass)
1020 PPCODE: 1079 PPCODE:
1021{ 1080{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1081 SV *pv = NEWSV (0, sizeof (CBOR));
… …
1030 1089
1031void shrink (CBOR *self, int enable = 1) 1090void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1091 ALIAS:
1033 shrink = F_SHRINK 1092 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1093 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS
1096 dedup_strings = F_DEDUP_STRINGS
1035 PPCODE: 1097 PPCODE:
1036{ 1098{
1037 if (enable) 1099 if (enable)
1038 self->flags |= ix; 1100 self->flags |= ix;
1039 else 1101 else
… …
1044 1106
1045void get_shrink (CBOR *self) 1107void get_shrink (CBOR *self)
1046 ALIAS: 1108 ALIAS:
1047 get_shrink = F_SHRINK 1109 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1110 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS
1113 get_dedup_strings = F_DEDUP_STRINGS
1049 PPCODE: 1114 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1115 XPUSHs (boolSV (self->flags & ix));
1051 1116
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1118 PPCODE:
… …
1068int get_max_size (CBOR *self) 1133int get_max_size (CBOR *self)
1069 CODE: 1134 CODE:
1070 RETVAL = self->max_size; 1135 RETVAL = self->max_size;
1071 OUTPUT: 1136 OUTPUT:
1072 RETVAL 1137 RETVAL
1073
1074#if 0 //TODO
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1081
1082 XPUSHs (ST (0));
1083}
1084
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
1086 PPCODE:
1087{
1088 if (!self->cb_sk_object)
1089 self->cb_sk_object = newHV ();
1090
1091 if (SvOK (cb))
1092 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093 else
1094 {
1095 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097 if (!HvKEYS (self->cb_sk_object))
1098 {
1099 SvREFCNT_dec (self->cb_sk_object);
1100 self->cb_sk_object = 0;
1101 }
1102 }
1103
1104 XPUSHs (ST (0));
1105}
1106
1107#endif
1108 1138
1109void encode (CBOR *self, SV *scalar) 1139void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1140 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1141 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1142 XPUSHs (scalar);
… …
1125 EXTEND (SP, 2); 1155 EXTEND (SP, 2);
1126 PUSHs (sv); 1156 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1158}
1129 1159
1130void DESTROY (CBOR *self)
1131 CODE:
1132 SvREFCNT_dec (self->cb_sk_object);
1133 SvREFCNT_dec (self->cb_object);
1134
1135PROTOTYPES: ENABLE 1160PROTOTYPES: ENABLE
1136 1161
1137void encode_cbor (SV *scalar) 1162void encode_cbor (SV *scalar)
1138 PPCODE: 1163 PPCODE:
1139{ 1164{

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines