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 1.3 by root, Sat Oct 26 11:08:34 2013 UTC vs.
Revision 1.17 by root, Tue Oct 29 22:04:52 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 = 256,
30 CBOR_TAG_GENERIC_OBJECT = 257,
31
32 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array
39
40 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor
52};
53
14#define F_SHRINK 0x00000200UL 54#define F_SHRINK 0x00000200UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 55#define F_ALLOW_UNKNOWN 0x00002000UL
16 56
17#define INIT_SIZE 32 // initial scalar size to be allocated 57#define INIT_SIZE 32 // initial scalar size to be allocated
18 58
19#define SB do { 59#define SB do {
20#define SE } while (0) 60#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 61
33#define IN_RANGE_INC(type,val,beg,end) \ 62#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 63 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 64 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 65
42#else 71#else
43# define CBOR_SLOW 0 72# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 73# define CBOR_STASH cbor_stash
45#endif 74#endif
46 75
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 77static SV *types_true, *types_false, *types_error, *sv_cbor;
49 78
50typedef struct { 79typedef struct {
51 U32 flags; 80 U32 flags;
52 U32 max_depth; 81 U32 max_depth;
53 STRLEN max_size; 82 STRLEN max_size;
54
55 SV *cb_object;
56 HV *cb_sk_object;
57} CBOR; 83} CBOR;
58 84
59INLINE void 85ecb_inline void
60cbor_init (CBOR *cbor) 86cbor_init (CBOR *cbor)
61{ 87{
62 Zero (cbor, 1, CBOR); 88 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 89 cbor->max_depth = 512;
64} 90}
65 91
66///////////////////////////////////////////////////////////////////////////// 92/////////////////////////////////////////////////////////////////////////////
67// utility functions 93// utility functions
68 94
69INLINE SV * 95ecb_inline SV *
70get_bool (const char *name) 96get_bool (const char *name)
71{ 97{
72 SV *sv = get_sv (name, 1); 98 SV *sv = get_sv (name, 1);
73 99
74 SvREADONLY_on (sv); 100 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 101 SvREADONLY_on (SvRV (sv));
76 102
77 return sv; 103 return sv;
78} 104}
79 105
80INLINE void 106ecb_inline void
81shrink (SV *sv) 107shrink (SV *sv)
82{ 108{
83 sv_utf8_downgrade (sv, 1); 109 sv_utf8_downgrade (sv, 1);
84 110
85 if (SvLEN (sv) > SvCUR (sv) + 1) 111 if (SvLEN (sv) > SvCUR (sv) + 1)
108 SV *sv; // result scalar 134 SV *sv; // result scalar
109 CBOR cbor; 135 CBOR cbor;
110 U32 depth; // recursion level 136 U32 depth; // recursion level
111} enc_t; 137} enc_t;
112 138
113INLINE void 139ecb_inline void
114need (enc_t *enc, STRLEN len) 140need (enc_t *enc, STRLEN len)
115{ 141{
116 if (expect_false (enc->cur + len >= enc->end)) 142 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 143 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 144 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 145 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 146 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 147 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 148 }
123} 149}
124 150
125INLINE void 151ecb_inline void
126encode_ch (enc_t *enc, char ch) 152encode_ch (enc_t *enc, char ch)
127{ 153{
128 need (enc, 1); 154 need (enc, 1);
129 *enc->cur++ = ch; 155 *enc->cur++ = ch;
130} 156}
134{ 160{
135 need (enc, 9); 161 need (enc, 9);
136 162
137 if (len < 24) 163 if (len < 24)
138 *enc->cur++ = major | len; 164 *enc->cur++ = major | len;
139 else if (len < 0x100) 165 else if (len <= 0xff)
140 { 166 {
141 *enc->cur++ = major | 24; 167 *enc->cur++ = major | 24;
142 *enc->cur++ = len; 168 *enc->cur++ = len;
143 } 169 }
144 else if (len < 0x10000) 170 else if (len <= 0xffff)
145 { 171 {
146 *enc->cur++ = major | 25; 172 *enc->cur++ = major | 25;
147 *enc->cur++ = len >> 8; 173 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 174 *enc->cur++ = len;
149 } 175 }
150 else if (len < 0x100000000) 176 else if (len <= 0xffffffff)
151 { 177 {
152 *enc->cur++ = major | 26; 178 *enc->cur++ = major | 26;
153 *enc->cur++ = len >> 24; 179 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 180 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 181 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 182 *enc->cur++ = len;
157 } 183 }
158 else if (len) 184 else
159 { 185 {
160 *enc->cur++ = major | 27; 186 *enc->cur++ = major | 27;
161 *enc->cur++ = len >> 56; 187 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 188 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 189 *enc->cur++ = len >> 40;
224 if (HeKLEN (he) == HEf_SVKEY) 250 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 251 encode_sv (enc, HeSVKEY (he));
226 else 252 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 254
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 255 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 256 }
231 257
232 if (mg) 258 if (mg)
233 encode_ch (enc, 0xe0 | 31); 259 encode_ch (enc, 0xe0 | 31);
234 260
242 svtype svt; 268 svtype svt;
243 269
244 SvGETMAGIC (sv); 270 SvGETMAGIC (sv);
245 svt = SvTYPE (sv); 271 svt = SvTYPE (sv);
246 272
247 if (expect_false (SvOBJECT (sv))) 273 if (ecb_expect_false (SvOBJECT (sv)))
248 { 274 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 276 ? types_boolean_stash
277 : gv_stashpv ("Types::Serialiser::Boolean", 1);
278 HV *error_stash = !CBOR_SLOW || types_error_stash
279 ? types_error_stash
280 : gv_stashpv ("Types::Serialiser::Error", 1);
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 284
253 if (SvSTASH (sv) == stash) 285 HV *stash = SvSTASH (sv);
286 GV *method;
287
288 if (stash == boolean_stash)
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 289 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
290 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23);
292 else if (stash == tagged_stash)
293 {
294 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
299 }
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 {
302 dSP;
303
304 ENTER; SAVETMPS; PUSHMARK (SP);
305 // we re-bless the reference to get overload and other niceties right
306 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
307
308 PUTBACK;
309 // G_SCALAR ensures that return value is 1
310 call_sv ((SV *)GvCV (method), G_SCALAR);
311 SPAGAIN;
312
313 // catch this surprisingly common error
314 if (SvROK (TOPs) && SvRV (TOPs) == sv)
315 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
316
317 encode_sv (enc, POPs);
318
319 PUTBACK;
320
321 FREETMPS; LEAVE;
322 }
323 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
324 {
325 dSP;
326
327 ENTER; SAVETMPS; PUSHMARK (SP);
328 EXTEND (SP, 2);
329 // we re-bless the reference to get overload and other niceties right
330 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
331 PUSHs (sv_cbor);
332
333 PUTBACK;
334 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
335 SPAGAIN;
336
337 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344
345 while (count)
346 encode_sv (enc, SP[1 - count--]);
347
348 PUTBACK;
349
350 FREETMPS; LEAVE;
351 }
255 else 352 else
256 {
257#if 0 //TODO
258 if (enc->cbor.flags & F_CONV_BLESSED)
259 {
260 // we re-bless the reference to get overload and other niceties right
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
269
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK;
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
273 SPAGAIN;
274
275 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
278
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv);
283
284 FREETMPS; LEAVE;
285 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED)
287 encode_str (enc, "null", 4, 0);
288 else
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 353 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
290 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED)
293 encode_str (enc, "null", 4, 0);
294 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 354 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 355 }
300 else if (svt == SVt_PVHV) 356 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 357 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 358 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 359 encode_av (enc, (AV *)sv);
328{ 384{
329 double nv = SvNVX (sv); 385 double nv = SvNVX (sv);
330 386
331 need (enc, 9); 387 need (enc, 9);
332 388
333 if (expect_false (nv == (U32)nv)) 389 if (ecb_expect_false (nv == (U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 390 encode_uint (enc, 0x00, (U32)nv);
335 //TODO: maybe I32? 391 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 392 else if (ecb_expect_false (nv == (float)nv))
337 { 393 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 394 uint32_t fp = ecb_float_to_binary32 (nv);
339 395
340 *enc->cur++ = 0xe0 | 26; 396 *enc->cur++ = 0xe0 | 26;
341 397
429 U32 maxdepth; // recursion depth limit 485 U32 maxdepth; // recursion depth limit
430} dec_t; 486} dec_t;
431 487
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 488#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
433 489
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 490#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
435 491
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 492#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 493#define DEC_DEC_DEPTH --dec->depth
438 494
439static UV 495static UV
528 SvREFCNT_dec (av); 584 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 585 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 586 return &PL_sv_undef;
531} 587}
532 588
589static void
590decode_he (dec_t *dec, HV *hv)
591{
592 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys.
594
595 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
596 {
597 I32 len = decode_uint (dec);
598 char *key = (char *)dec->cur;
599
600 dec->cur += len;
601
602 hv_store (hv, key, len, decode_sv (dec), 0);
603 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
605 {
606 I32 len = decode_uint (dec);
607 char *key = (char *)dec->cur;
608
609 dec->cur += len;
610
611 hv_store (hv, key, -len, decode_sv (dec), 0);
612 }
613 else
614 {
615 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec);
617
618 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k);
620 }
621}
622
533static SV * 623static SV *
534decode_hv (dec_t *dec) 624decode_hv (dec_t *dec)
535{ 625{
536 HV *hv = newHV (); 626 HV *hv = newHV ();
537 627
549 { 639 {
550 ++dec->cur; 640 ++dec->cur;
551 break; 641 break;
552 } 642 }
553 643
554 SV *k = decode_sv (dec); 644 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 645 }
559 } 646 }
560 else 647 else
561 { 648 {
562 int len = decode_uint (dec); 649 int pairs = decode_uint (dec);
563 650
564 while (len--) 651 while (pairs--)
565 { 652 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 653 }
572 654
573 DEC_DEC_DEPTH; 655 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 656 return newRV_noinc ((SV *)hv);
575
576#if 0
577 SV *sv;
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH;
581 decode_ws (dec);
582
583 for (;;)
584 {
585 // heuristic: assume that
586 // a) decode_str + hv_store_ent are abysmally slow.
587 // b) most hash keys are short, simple ascii text.
588 // => try to "fast-match" such strings to avoid
589 // the overhead of decode_str + hv_store_ent.
590 {
591 SV *value;
592 char *p = dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594
595 for (;;)
596 {
597 // the >= 0x80 is false on most architectures
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599 {
600 // slow path, back up and use decode_str
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break;
619 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 {
685 dSP;
686 int count;
687
688 ENTER; SAVETMPS; PUSHMARK (SP);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730 657
731fail: 658fail:
732 SvREFCNT_dec (hv); 659 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 660 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 661 return &PL_sv_undef;
735} 662}
736 663
737static SV * 664static SV *
738decode_str (dec_t *dec, int utf8) 665decode_str (dec_t *dec, int utf8)
739{ 666{
740 SV *sv; 667 SV *sv = 0;
741 668
742 if ((*dec->cur & 31) == 31) 669 if ((*dec->cur & 31) == 31)
743 { 670 {
744 ++dec->cur; 671 ++dec->cur;
745 672
754 { 681 {
755 ++dec->cur; 682 ++dec->cur;
756 break; 683 break;
757 } 684 }
758 685
759 SV *sv2 = decode_sv (dec);
760 sv_catsv (sv, sv2); 686 sv_catsv (sv, decode_sv (dec));
761 } 687 }
762 } 688 }
763 else 689 else
764 { 690 {
765 STRLEN len = decode_uint (dec); 691 STRLEN len = decode_uint (dec);
773 SvUTF8_on (sv); 699 SvUTF8_on (sv);
774 700
775 return sv; 701 return sv;
776 702
777fail: 703fail:
704 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 705 return &PL_sv_undef;
779} 706}
780 707
781static SV * 708static SV *
782decode_tagged (dec_t *dec) 709decode_tagged (dec_t *dec)
783{ 710{
784 UV tag = decode_uint (dec); 711 UV tag = decode_uint (dec);
785 SV *sv = decode_sv (dec); 712 SV *sv = decode_sv (dec);
786 713
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 714 if (tag == CBOR_TAG_MAGIC)
788 return sv; 715 return sv;
716 else if (tag == CBOR_TAG_PERL_OBJECT)
717 {
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)");
789 720
721 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724
725 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)");
727
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729
730 if (!method)
731 ERR ("cannot decode perl-object (package does not have a THAW method)");
732
733 dSP;
734
735 ENTER; SAVETMPS; PUSHMARK (SP);
736 EXTEND (SP, len + 1);
737 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor);
740
741 int i;
742
743 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1));
745
746 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN;
749
750 if (SvTRUE (ERRSV))
751 {
752 FREETMPS; LEAVE;
753 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
754 }
755
756 SvREFCNT_dec (sv);
757 sv = SvREFCNT_inc (POPs);
758
759 PUTBACK;
760
761 FREETMPS; LEAVE;
762
763 return sv;
764 }
765 else
766 {
790 AV *av = newAV (); 767 AV *av = newAV ();
791 av_push (av, newSVuv (tag)); 768 av_push (av, newSVuv (tag));
792 av_push (av, sv); 769 av_push (av, sv);
793 return newRV_noinc ((SV *)av); 770
771 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
772 ? cbor_tagged_stash
773 : gv_stashpv ("CBOR::XS::Tagged" , 1);
774
775 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
776 }
777
778fail:
779 SvREFCNT_dec (sv);
780 return &PL_sv_undef;
794} 781}
795 782
796static SV * 783static SV *
797decode_sv (dec_t *dec) 784decode_sv (dec_t *dec)
798{ 785{
799 WANT (1); 786 WANT (1);
800 787
801 switch (*dec->cur >> 5) 788 switch (*dec->cur >> 5)
802 { 789 {
803 case 0: // unsigned int 790 case 0: // unsigned int
804 //TODO: 64 bit values on 3 2bit perls
805 return newSVuv (decode_uint (dec)); 791 return newSVuv (decode_uint (dec));
806 case 1: // negative int 792 case 1: // negative int
807 return newSViv (-1 - (IV)decode_uint (dec)); 793 return newSViv (-1 - (IV)decode_uint (dec));
808 case 2: // octet string 794 case 2: // octet string
809 return decode_str (dec, 0); 795 return decode_str (dec, 0);
818 case 7: // misc 804 case 7: // misc
819 switch (*dec->cur++ & 31) 805 switch (*dec->cur++ & 31)
820 { 806 {
821 case 20: 807 case 20:
822#if CBOR_SLOW 808#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 809 types_false = get_bool ("Types::Serialiser::false");
824#endif 810#endif
825 return newSVsv (cbor_false); 811 return newSVsv (types_false);
826 case 21: 812 case 21:
827#if CBOR_SLOW 813#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 814 types_true = get_bool ("Types::Serialiser::true");
829#endif 815#endif
830 return newSVsv (cbor_true); 816 return newSVsv (types_true);
831 case 22: 817 case 22:
832 return newSVsv (&PL_sv_undef); 818 return newSVsv (&PL_sv_undef);
819 case 23:
820#if CBOR_SLOW
821 types_error = get_bool ("Types::Serialiser::error");
822#endif
823 return newSVsv (types_error);
833 824
834 case 25: 825 case 25:
835 { 826 {
836 WANT (2); 827 WANT (2);
837 828
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 864 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
874 } 865 }
875 866
876 break; 867 break;
877 } 868 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 869
935fail: 870fail:
936 return &PL_sv_undef; 871 return &PL_sv_undef;
937} 872}
938 873
939static SV * 874static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 875decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 876{
942 dec_t dec; 877 dec_t dec;
943 SV *sv; 878 SV *sv;
879 STRLEN len;
880 char *data = SvPVbyte (string, len);
944 881
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 882 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 883 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 884 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 885
980 dec.cbor = *cbor; 886 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 887 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 888 dec.end = (U8 *)data + len;
983 dec.err = 0; 889 dec.err = 0;
984 dec.depth = 0; 890 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 891
989 sv = decode_sv (&dec); 892 sv = decode_sv (&dec);
990 893
991 if (offset_return) 894 if (offset_return)
992 *offset_return = dec.cur; 895 *offset_return = dec.cur;
996 dec.err = "garbage after CBOR object"; 899 dec.err = "garbage after CBOR object";
997 900
998 if (dec.err) 901 if (dec.err)
999 { 902 {
1000 SvREFCNT_dec (sv); 903 SvREFCNT_dec (sv);
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 904 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 905 }
1003 906
1004 sv = sv_2mortal (sv); 907 sv = sv_2mortal (sv);
1005 908
1006 return sv; 909 return sv;
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 915MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 916
1014BOOT: 917BOOT:
1015{ 918{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 919 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 920 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 921
1019 cbor_true = get_bool ("CBOR::XS::true"); 922 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 923 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
924
925 types_true = get_bool ("Types::Serialiser::true" );
926 types_false = get_bool ("Types::Serialiser::false");
927 types_error = get_bool ("Types::Serialiser::error");
928
929 sv_cbor = newSVpv ("CBOR", 0);
930 SvREADONLY_on (sv_cbor);
1021} 931}
1022 932
1023PROTOTYPES: DISABLE 933PROTOTYPES: DISABLE
1024 934
1025void CLONE (...) 935void CLONE (...)
1026 CODE: 936 CODE:
1027 cbor_stash = 0; 937 cbor_stash = 0;
938 cbor_tagged_stash = 0;
939 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 940 types_boolean_stash = 0;
1029 941
1030void new (char *klass) 942void new (char *klass)
1031 PPCODE: 943 PPCODE:
1032{ 944{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 945 SV *pv = NEWSV (0, sizeof (CBOR));
1080 CODE: 992 CODE:
1081 RETVAL = self->max_size; 993 RETVAL = self->max_size;
1082 OUTPUT: 994 OUTPUT:
1083 RETVAL 995 RETVAL
1084 996
1085#if 0 //TODO
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1092
1093 XPUSHs (ST (0));
1094}
1095
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
1097 PPCODE:
1098{
1099 if (!self->cb_sk_object)
1100 self->cb_sk_object = newHV ();
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119
1120void encode (CBOR *self, SV *scalar) 997void encode (CBOR *self, SV *scalar)
1121 PPCODE: 998 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 999 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1000 XPUSHs (scalar);
1124 1001
1136 EXTEND (SP, 2); 1013 EXTEND (SP, 2);
1137 PUSHs (sv); 1014 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1015 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1016}
1140 1017
1141void DESTROY (CBOR *self)
1142 CODE:
1143 SvREFCNT_dec (self->cb_sk_object);
1144 SvREFCNT_dec (self->cb_object);
1145
1146PROTOTYPES: ENABLE 1018PROTOTYPES: ENABLE
1147 1019
1148void encode_cbor (SV *scalar) 1020void encode_cbor (SV *scalar)
1149 PPCODE: 1021 PPCODE:
1150{ 1022{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines