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 1.1 by root, Fri Oct 25 23:09:45 2013 UTC vs.
Revision 1.16 by root, Tue Oct 29 20:59:16 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#if defined(__BORLANDC__) || defined(_MSC_VER) 14// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 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};
17 53
18#define F_SHRINK 0x00000200UL 54#define F_SHRINK 0x00000200UL
19#define F_ALLOW_UNKNOWN 0x00002000UL 55#define F_ALLOW_UNKNOWN 0x00002000UL
20 56
21#define INIT_SIZE 32 // initial scalar size to be allocated 57#define INIT_SIZE 32 // initial scalar size to be allocated
22 58
23#define SB do { 59#define SB do {
24#define SE } while (0) 60#define SE } while (0)
25
26#if __GNUC__ >= 3
27# define expect(expr,value) __builtin_expect ((expr), (value))
28# define INLINE static inline
29#else
30# define expect(expr,value) (expr)
31# define INLINE static
32#endif
33
34#define expect_false(expr) expect ((expr) != 0, 0)
35#define expect_true(expr) expect ((expr) != 0, 1)
36 61
37#define IN_RANGE_INC(type,val,beg,end) \ 62#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 63 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 64 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 65
46#else 71#else
47# define CBOR_SLOW 0 72# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 73# define CBOR_STASH cbor_stash
49#endif 74#endif
50 75
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 77static SV *types_true, *types_false, *types_error, *sv_cbor;
53 78
54typedef struct { 79typedef struct {
55 U32 flags; 80 U32 flags;
56 U32 max_depth; 81 U32 max_depth;
57 STRLEN max_size; 82 STRLEN max_size;
58
59 SV *cb_object;
60 HV *cb_sk_object;
61} CBOR; 83} CBOR;
62 84
63INLINE void 85ecb_inline void
64cbor_init (CBOR *cbor) 86cbor_init (CBOR *cbor)
65{ 87{
66 Zero (cbor, 1, CBOR); 88 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 89 cbor->max_depth = 512;
68} 90}
69 91
70///////////////////////////////////////////////////////////////////////////// 92/////////////////////////////////////////////////////////////////////////////
71// utility functions 93// utility functions
72 94
73INLINE SV * 95ecb_inline SV *
74get_bool (const char *name) 96get_bool (const char *name)
75{ 97{
76 SV *sv = get_sv (name, 1); 98 SV *sv = get_sv (name, 1);
77 99
78 SvREADONLY_on (sv); 100 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 101 SvREADONLY_on (SvRV (sv));
80 102
81 return sv; 103 return sv;
82} 104}
83 105
84INLINE void 106ecb_inline void
85shrink (SV *sv) 107shrink (SV *sv)
86{ 108{
87 sv_utf8_downgrade (sv, 1); 109 sv_utf8_downgrade (sv, 1);
88 110
89 if (SvLEN (sv) > SvCUR (sv) + 1) 111 if (SvLEN (sv) > SvCUR (sv) + 1)
112 SV *sv; // result scalar 134 SV *sv; // result scalar
113 CBOR cbor; 135 CBOR cbor;
114 U32 depth; // recursion level 136 U32 depth; // recursion level
115} enc_t; 137} enc_t;
116 138
117INLINE void 139ecb_inline void
118need (enc_t *enc, STRLEN len) 140need (enc_t *enc, STRLEN len)
119{ 141{
120 if (expect_false (enc->cur + len >= enc->end)) 142 if (ecb_expect_false (enc->cur + len >= enc->end))
121 { 143 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 144 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 145 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 146 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 147 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 148 }
127} 149}
128 150
129INLINE void 151ecb_inline void
130encode_ch (enc_t *enc, char ch) 152encode_ch (enc_t *enc, char ch)
131{ 153{
132 need (enc, 1); 154 need (enc, 1);
133 *enc->cur++ = ch; 155 *enc->cur++ = ch;
134} 156}
138{ 160{
139 need (enc, 9); 161 need (enc, 9);
140 162
141 if (len < 24) 163 if (len < 24)
142 *enc->cur++ = major | len; 164 *enc->cur++ = major | len;
143 else if (len < 0x100) 165 else if (len <= 0xff)
144 { 166 {
145 *enc->cur++ = major | 24; 167 *enc->cur++ = major | 24;
146 *enc->cur++ = len; 168 *enc->cur++ = len;
147 } 169 }
148 else if (len < 0x10000) 170 else if (len <= 0xffff)
149 { 171 {
150 *enc->cur++ = major | 25; 172 *enc->cur++ = major | 25;
151 *enc->cur++ = len >> 8; 173 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 174 *enc->cur++ = len;
153 } 175 }
154 else if (len < 0x100000000) 176 else if (len <= 0xffffffff)
155 { 177 {
156 *enc->cur++ = major | 26; 178 *enc->cur++ = major | 26;
157 *enc->cur++ = len >> 24; 179 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 180 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 181 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 182 *enc->cur++ = len;
161 } 183 }
162 else if (len) 184 else
163 { 185 {
164 *enc->cur++ = major | 27; 186 *enc->cur++ = major | 27;
165 *enc->cur++ = len >> 56; 187 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 188 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 189 *enc->cur++ = len >> 40;
228 if (HeKLEN (he) == HEf_SVKEY) 250 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 251 encode_sv (enc, HeSVKEY (he));
230 else 252 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 254
233 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));
234 } 256 }
235 257
236 if (mg) 258 if (mg)
237 encode_ch (enc, 0xe0 | 31); 259 encode_ch (enc, 0xe0 | 31);
238 260
246 svtype svt; 268 svtype svt;
247 269
248 SvGETMAGIC (sv); 270 SvGETMAGIC (sv);
249 svt = SvTYPE (sv); 271 svt = SvTYPE (sv);
250 272
251 if (expect_false (SvOBJECT (sv))) 273 if (ecb_expect_false (SvOBJECT (sv)))
252 { 274 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? 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
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 284
257 if (SvSTASH (sv) == stash) 285 HV *stash = SvSTASH (sv);
286 GV *method;
287
288 if (stash == boolean_stash)
258 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 }
259 else 352 else
260 {
261#if 0 //TODO
262 if (enc->cbor.flags & F_CONV_BLESSED)
263 {
264 // we re-bless the reference to get overload and other niceties right
265 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
266
267 if (to_cbor)
268 {
269 dSP;
270
271 ENTER; SAVETMPS; PUSHMARK (SP);
272 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
273
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK;
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR);
277 SPAGAIN;
278
279 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv)
281 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
282
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv);
287
288 FREETMPS; LEAVE;
289 }
290 else if (enc->cbor.flags & F_ALLOW_BLESSED)
291 encode_str (enc, "null", 4, 0);
292 else
293 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",
294 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
295 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED)
297 encode_str (enc, "null", 4, 0);
298 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 354 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 355 }
304 else if (svt == SVt_PVHV) 356 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 357 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 358 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 359 encode_av (enc, (AV *)sv);
332{ 384{
333 double nv = SvNVX (sv); 385 double nv = SvNVX (sv);
334 386
335 need (enc, 9); 387 need (enc, 9);
336 388
337 if (expect_false (nv == (U32)nv)) 389 if (ecb_expect_false (nv == (U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 390 encode_uint (enc, 0x00, (U32)nv);
339 //TODO: maybe I32? 391 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 392 else if (ecb_expect_false (nv == (float)nv))
341 { 393 {
342 uint32_t fp = ecb_float_to_binary32 (nv); 394 uint32_t fp = ecb_float_to_binary32 (nv);
343 395
344 *enc->cur++ = 0xe0 | 26; 396 *enc->cur++ = 0xe0 | 26;
345 397
433 U32 maxdepth; // recursion depth limit 485 U32 maxdepth; // recursion depth limit
434} dec_t; 486} dec_t;
435 487
436#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
437 489
438#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")
439 491
440#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)
441#define DEC_DEC_DEPTH --dec->depth 493#define DEC_DEC_DEPTH --dec->depth
442 494
443static UV 495static UV
504 556
505 for (;;) 557 for (;;)
506 { 558 {
507 WANT (1); 559 WANT (1);
508 560
509 if (*dec->cur == 0xe0 | 31) 561 if (*dec->cur == (0xe0 | 31))
510 { 562 {
511 ++dec->cur; 563 ++dec->cur;
512 break; 564 break;
513 } 565 }
514 566
532 SvREFCNT_dec (av); 584 SvREFCNT_dec (av);
533 DEC_DEC_DEPTH; 585 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 586 return &PL_sv_undef;
535} 587}
536 588
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
537static SV * 623static SV *
538decode_hv (dec_t *dec) 624decode_hv (dec_t *dec)
539{ 625{
540 HV *hv = newHV (); 626 HV *hv = newHV ();
541 627
547 633
548 for (;;) 634 for (;;)
549 { 635 {
550 WANT (1); 636 WANT (1);
551 637
552 if (*dec->cur == 0xe0 | 31) 638 if (*dec->cur == (0xe0 | 31))
553 { 639 {
554 ++dec->cur; 640 ++dec->cur;
555 break; 641 break;
556 } 642 }
557 643
558 SV *k = decode_sv (dec); 644 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 645 }
563 } 646 }
564 else 647 else
565 { 648 {
566 int len = decode_uint (dec); 649 int pairs = decode_uint (dec);
567 650
568 while (len--) 651 while (pairs--)
569 { 652 decode_he (dec, hv);
570 SV *k = decode_sv (dec);
571 SV *v = decode_sv (dec);
572
573 hv_store_ent (hv, k, v, 0);
574 }
575 } 653 }
576 654
577 DEC_DEC_DEPTH; 655 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 656 return newRV_noinc ((SV *)hv);
579
580#if 0
581 SV *sv;
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH;
585 decode_ws (dec);
586
587 for (;;)
588 {
589 // heuristic: assume that
590 // a) decode_str + hv_store_ent are abysmally slow.
591 // b) most hash keys are short, simple ascii text.
592 // => try to "fast-match" such strings to avoid
593 // the overhead of decode_str + hv_store_ent.
594 {
595 SV *value;
596 char *p = dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598
599 for (;;)
600 {
601 // the >= 0x80 is false on most architectures
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
603 {
604 // slow path, back up and use decode_str
605 SV *key = decode_str (dec);
606 if (!key)
607 goto fail;
608
609 decode_ws (dec); EXPECT_CH (':');
610
611 decode_ws (dec);
612 value = decode_sv (dec);
613 if (!value)
614 {
615 SvREFCNT_dec (key);
616 goto fail;
617 }
618
619 hv_store_ent (hv, key, value, 0);
620 SvREFCNT_dec (key);
621
622 break;
623 }
624 else if (*p == '"')
625 {
626 // fast path, got a simple key
627 char *key = dec->cur;
628 int len = p - key;
629 dec->cur = p + 1;
630
631 decode_ws (dec); EXPECT_CH (':');
632
633 decode_ws (dec);
634 value = decode_sv (dec);
635 if (!value)
636 goto fail;
637
638 hv_store (hv, key, len, value, 0);
639
640 break;
641 }
642
643 ++p;
644 }
645 }
646
647 decode_ws (dec);
648
649 if (*dec->cur == '}')
650 {
651 ++dec->cur;
652 break;
653 }
654
655 if (*dec->cur != ',')
656 ERR (", or } expected while parsing object/hash");
657
658 ++dec->cur;
659
660 decode_ws (dec);
661
662 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
663 {
664 ++dec->cur;
665 break;
666 }
667 }
668
669 DEC_DEC_DEPTH;
670 sv = newRV_noinc ((SV *)hv);
671
672 // check filter callbacks
673 if (dec->cbor.flags & F_HOOK)
674 {
675 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
676 {
677 HE *cb, *he;
678
679 hv_iterinit (hv);
680 he = hv_iternext (hv);
681 hv_iterinit (hv);
682
683 // the next line creates a mortal sv each time its called.
684 // might want to optimise this for common cases.
685 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
686
687 if (cb)
688 {
689 dSP;
690 int count;
691
692 ENTER; SAVETMPS; PUSHMARK (SP);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695
696 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
697
698 if (count == 1)
699 {
700 sv = newSVsv (POPs);
701 FREETMPS; LEAVE;
702 return sv;
703 }
704
705 SvREFCNT_inc (sv);
706 FREETMPS; LEAVE;
707 }
708 }
709
710 if (dec->cbor.cb_object)
711 {
712 dSP;
713 int count;
714
715 ENTER; SAVETMPS; PUSHMARK (SP);
716 XPUSHs (sv_2mortal (sv));
717
718 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
719
720 if (count == 1)
721 {
722 sv = newSVsv (POPs);
723 FREETMPS; LEAVE;
724 return sv;
725 }
726
727 SvREFCNT_inc (sv);
728 FREETMPS; LEAVE;
729 }
730 }
731
732 return sv;
733#endif
734 657
735fail: 658fail:
736 SvREFCNT_dec (hv); 659 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH; 660 DEC_DEC_DEPTH;
738 return &PL_sv_undef; 661 return &PL_sv_undef;
739} 662}
740 663
741static SV * 664static SV *
742decode_str (dec_t *dec, int utf8) 665decode_str (dec_t *dec, int utf8)
743{ 666{
744 SV *sv; 667 SV *sv = 0;
745 668
746 if ((*dec->cur & 31) == 31) 669 if ((*dec->cur & 31) == 31)
747 { 670 {
748 ++dec->cur; 671 ++dec->cur;
749 672
752 // not very fast, and certainly not robust against illegal input 675 // not very fast, and certainly not robust against illegal input
753 for (;;) 676 for (;;)
754 { 677 {
755 WANT (1); 678 WANT (1);
756 679
757 if (*dec->cur == 0xe0 | 31) 680 if (*dec->cur == (0xe0 | 31))
758 { 681 {
759 ++dec->cur; 682 ++dec->cur;
760 break; 683 break;
761 } 684 }
762 685
763 SV *sv2 = decode_sv (dec);
764 sv_catsv (sv, sv2); 686 sv_catsv (sv, decode_sv (dec));
765 } 687 }
766 } 688 }
767 else 689 else
768 { 690 {
769 STRLEN len = decode_uint (dec); 691 STRLEN len = decode_uint (dec);
777 SvUTF8_on (sv); 699 SvUTF8_on (sv);
778 700
779 return sv; 701 return sv;
780 702
781fail: 703fail:
704 SvREFCNT_dec (sv);
705 return &PL_sv_undef;
706}
707
708static SV *
709decode_tagged (dec_t *dec)
710{
711 UV tag = decode_uint (dec);
712 SV *sv = decode_sv (dec);
713
714 if (tag == CBOR_TAG_MAGIC)
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)");
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);
748 SPAGAIN;
749
750 SvREFCNT_dec (sv);
751 sv = SvREFCNT_inc (POPs);
752
753 PUTBACK;
754
755 FREETMPS; LEAVE;
756
757 return sv;
758 }
759 else
760 {
761 AV *av = newAV ();
762 av_push (av, newSVuv (tag));
763 av_push (av, sv);
764
765 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
766 ? cbor_tagged_stash
767 : gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
770 }
771
772fail:
773 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 774 return &PL_sv_undef;
783} 775}
784 776
785static SV * 777static SV *
786decode_sv (dec_t *dec) 778decode_sv (dec_t *dec)
788 WANT (1); 780 WANT (1);
789 781
790 switch (*dec->cur >> 5) 782 switch (*dec->cur >> 5)
791 { 783 {
792 case 0: // unsigned int 784 case 0: // unsigned int
793 //TODO: 64 bit values on 3 2bit perls
794 return newSVuv (decode_uint (dec)); 785 return newSVuv (decode_uint (dec));
795 case 1: // negative int 786 case 1: // negative int
796 return newSViv (-1 - (IV)decode_uint (dec)); 787 return newSViv (-1 - (IV)decode_uint (dec));
797 case 2: // octet string 788 case 2: // octet string
798 return decode_str (dec, 0); 789 return decode_str (dec, 0);
801 case 4: // array 792 case 4: // array
802 return decode_av (dec); 793 return decode_av (dec);
803 case 5: // map 794 case 5: // map
804 return decode_hv (dec); 795 return decode_hv (dec);
805 case 6: // tag 796 case 6: // tag
806 abort (); 797 return decode_tagged (dec);
807 break;
808 case 7: // misc 798 case 7: // misc
809 switch (*dec->cur++ & 31) 799 switch (*dec->cur++ & 31)
810 { 800 {
811 case 20: 801 case 20:
812#if CBOR_SLOW 802#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 803 types_false = get_bool ("Types::Serialiser::false");
814#endif 804#endif
815 return newSVsv (cbor_false); 805 return newSVsv (types_false);
816 case 21: 806 case 21:
817#if CBOR_SLOW 807#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 808 types_true = get_bool ("Types::Serialiser::true");
819#endif 809#endif
820 return newSVsv (cbor_true); 810 return newSVsv (types_true);
821 case 22: 811 case 22:
822 return newSVsv (&PL_sv_undef); 812 return newSVsv (&PL_sv_undef);
813 case 23:
814#if CBOR_SLOW
815 types_error = get_bool ("Types::Serialiser::error");
816#endif
817 return newSVsv (types_error);
823 818
824 case 25: 819 case 25:
825 // half float
826 abort ();
827 break; 820 {
821 WANT (2);
822
823 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
824 dec->cur += 2;
825
826 return newSVnv (ecb_binary16_to_float (fp));
827 }
828 828
829 case 26: 829 case 26:
830 { 830 {
831 uint32_t fp; 831 uint32_t fp;
832 WANT (4); 832 WANT (4);
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
859 } 859 }
860 860
861 break; 861 break;
862 } 862 }
863#if 0
864 switch (*dec->cur)
865 {
866 //case '"': ++dec->cur; return decode_str (dec);
867 case '[': ++dec->cur; return decode_av (dec);
868 case '{': ++dec->cur; return decode_hv (dec);
869
870 case '-':
871 case '0': case '1': case '2': case '3': case '4':
872 case '5': case '6': case '7': case '8': case '9':
873 //TODO return decode_num (dec);
874
875 case 't':
876 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
877 {
878 dec->cur += 4;
879#if CBOR_SLOW
880 cbor_true = get_bool ("CBOR::XS::true");
881#endif
882 return newSVsv (cbor_true);
883 }
884 else
885 ERR ("'true' expected");
886
887 break;
888
889 case 'f':
890 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
891 {
892 dec->cur += 5;
893#if CBOR_SLOW
894 cbor_false = get_bool ("CBOR::XS::false");
895#endif
896 return newSVsv (cbor_false);
897 }
898 else
899 ERR ("'false' expected");
900
901 break;
902
903 case 'n':
904 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
905 {
906 dec->cur += 4;
907 return newSVsv (&PL_sv_undef);
908 }
909 else
910 ERR ("'null' expected");
911
912 break;
913
914 default:
915 ERR ("malformed CBOR string, neither array, object, number, string or atom");
916 break;
917 }
918#endif
919 863
920fail: 864fail:
921 return &PL_sv_undef; 865 return &PL_sv_undef;
922} 866}
923 867
924static SV * 868static SV *
925decode_cbor (SV *string, CBOR *cbor, char **offset_return) 869decode_cbor (SV *string, CBOR *cbor, char **offset_return)
926{ 870{
927 dec_t dec; 871 dec_t dec;
928 SV *sv; 872 SV *sv;
873 STRLEN len;
874 char *data = SvPVbyte (string, len);
929 875
930 /* work around bugs in 5.10 where manipulating magic values
931 * makes perl ignore the magic in subsequent accesses.
932 * also make a copy of non-PV values, to get them into a clean
933 * state (SvPV should do that, but it's buggy, see below).
934 */
935 /*SvGETMAGIC (string);*/
936 if (SvMAGICAL (string) || !SvPOK (string))
937 string = sv_2mortal (newSVsv (string));
938
939 SvUPGRADE (string, SVt_PV);
940
941 /* work around a bug in perl 5.10, which causes SvCUR to fail an
942 * assertion with -DDEBUGGING, although SvCUR is documented to
943 * return the xpv_cur field which certainly exists after upgrading.
944 * according to nicholas clark, calling SvPOK fixes this.
945 * But it doesn't fix it, so try another workaround, call SvPV_nolen
946 * and hope for the best.
947 * Damnit, SvPV_nolen still trips over yet another assertion. This
948 * assertion business is seriously broken, try yet another workaround
949 * for the broken -DDEBUGGING.
950 */
951 {
952#ifdef DEBUGGING
953 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
954#else
955 STRLEN offset = SvCUR (string);
956#endif
957
958 if (offset > cbor->max_size && cbor->max_size) 876 if (len > cbor->max_size && cbor->max_size)
959 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 877 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
960 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 878 (unsigned long)len, (unsigned long)cbor->max_size);
961 }
962
963 sv_utf8_downgrade (string, 0);
964 879
965 dec.cbor = *cbor; 880 dec.cbor = *cbor;
966 dec.cur = (U8 *)SvPVX (string); 881 dec.cur = (U8 *)data;
967 dec.end = (U8 *)SvEND (string); 882 dec.end = (U8 *)data + len;
968 dec.err = 0; 883 dec.err = 0;
969 dec.depth = 0; 884 dec.depth = 0;
970 885
971 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
972 ;//TODO dec.cbor.flags |= F_HOOK;
973
974 sv = decode_sv (&dec); 886 sv = decode_sv (&dec);
975 887
976 if (offset_return) 888 if (offset_return)
977 *offset_return = dec.cur; 889 *offset_return = dec.cur;
978 890
979 if (!(offset_return || !sv)) 891 if (!(offset_return || !sv))
980 {
981 if (*dec.cur && !dec.err) 892 if (dec.cur != dec.end && !dec.err)
982 {
983 dec.err = "garbage after CBOR object"; 893 dec.err = "garbage after CBOR object";
894
895 if (dec.err)
896 {
984 SvREFCNT_dec (sv); 897 SvREFCNT_dec (sv);
985 sv = 0;
986 }
987 }
988
989 if (!sv)
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 898 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
899 }
991 900
992 sv = sv_2mortal (sv); 901 sv = sv_2mortal (sv);
993 902
994 return sv; 903 return sv;
995} 904}
1000MODULE = CBOR::XS PACKAGE = CBOR::XS 909MODULE = CBOR::XS PACKAGE = CBOR::XS
1001 910
1002BOOT: 911BOOT:
1003{ 912{
1004 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 913 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1005 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 914 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1006 915
1007 cbor_true = get_bool ("CBOR::XS::true"); 916 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1008 cbor_false = get_bool ("CBOR::XS::false"); 917 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
918
919 types_true = get_bool ("Types::Serialiser::true" );
920 types_false = get_bool ("Types::Serialiser::false");
921 types_error = get_bool ("Types::Serialiser::error");
922
923 sv_cbor = newSVpv ("CBOR", 0);
924 SvREADONLY_on (sv_cbor);
1009} 925}
1010 926
1011PROTOTYPES: DISABLE 927PROTOTYPES: DISABLE
1012 928
1013void CLONE (...) 929void CLONE (...)
1014 CODE: 930 CODE:
1015 cbor_stash = 0; 931 cbor_stash = 0;
932 cbor_tagged_stash = 0;
933 types_error_stash = 0;
1016 cbor_boolean_stash = 0; 934 types_boolean_stash = 0;
1017 935
1018void new (char *klass) 936void new (char *klass)
1019 PPCODE: 937 PPCODE:
1020{ 938{
1021 SV *pv = NEWSV (0, sizeof (CBOR)); 939 SV *pv = NEWSV (0, sizeof (CBOR));
1068 CODE: 986 CODE:
1069 RETVAL = self->max_size; 987 RETVAL = self->max_size;
1070 OUTPUT: 988 OUTPUT:
1071 RETVAL 989 RETVAL
1072 990
1073#if 0 //TODO
1074
1075void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1076 PPCODE:
1077{
1078 SvREFCNT_dec (self->cb_object);
1079 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1080
1081 XPUSHs (ST (0));
1082}
1083
1084void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef)
1085 PPCODE:
1086{
1087 if (!self->cb_sk_object)
1088 self->cb_sk_object = newHV ();
1089
1090 if (SvOK (cb))
1091 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1092 else
1093 {
1094 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1095
1096 if (!HvKEYS (self->cb_sk_object))
1097 {
1098 SvREFCNT_dec (self->cb_sk_object);
1099 self->cb_sk_object = 0;
1100 }
1101 }
1102
1103 XPUSHs (ST (0));
1104}
1105
1106#endif
1107
1108void encode (CBOR *self, SV *scalar) 991void encode (CBOR *self, SV *scalar)
1109 PPCODE: 992 PPCODE:
1110 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 993 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1111 XPUSHs (scalar); 994 XPUSHs (scalar);
1112 995
1124 EXTEND (SP, 2); 1007 EXTEND (SP, 2);
1125 PUSHs (sv); 1008 PUSHs (sv);
1126 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1009 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1127} 1010}
1128 1011
1129void DESTROY (CBOR *self)
1130 CODE:
1131 SvREFCNT_dec (self->cb_sk_object);
1132 SvREFCNT_dec (self->cb_object);
1133
1134PROTOTYPES: ENABLE 1012PROTOTYPES: ENABLE
1135 1013
1136void encode_cbor (SV *scalar) 1014void encode_cbor (SV *scalar)
1137 PPCODE: 1015 PPCODE:
1138{ 1016{

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines