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 1.61 by root, Sat Nov 26 02:10:19 2016 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_dec_NN
27# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
28#endif
29
30// known major and minor types
31enum cbor_type
32{
33 MAJOR_SHIFT = 5,
34 MINOR_MASK = 0x1f,
35
36 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
37 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
38 MAJOR_BYTES = 2 << MAJOR_SHIFT,
39 MAJOR_TEXT = 3 << MAJOR_SHIFT,
40 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
41 MAJOR_MAP = 5 << MAJOR_SHIFT,
42 MAJOR_TAG = 6 << MAJOR_SHIFT,
43 MAJOR_MISC = 7 << MAJOR_SHIFT,
44
45 // INT/STRING/ARRAY/MAP subtypes
46 LENGTH_EXT1 = 24,
47 LENGTH_EXT2 = 25,
48 LENGTH_EXT4 = 26,
49 LENGTH_EXT8 = 27,
50
51 // SIMPLE types (effectively MISC subtypes)
52 SIMPLE_FALSE = 20,
53 SIMPLE_TRUE = 21,
54 SIMPLE_NULL = 22,
55 SIMPLE_UNDEF = 23,
56
57 // MISC subtype (unused)
58 MISC_EXT1 = 24,
59 MISC_FLOAT16 = 25,
60 MISC_FLOAT32 = 26,
61 MISC_FLOAT64 = 27,
62
63 // BYTES/TEXT/ARRAY/MAP
64 MINOR_INDEF = 31,
65};
66
67// known tags
68enum cbor_tag
69{
70 // extensions
71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
72 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
73 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
74 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
76 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
77 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
78
79 // rfc7049
80 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
81 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
82 CBOR_TAG_POS_BIGNUM = 2, // byte string
83 CBOR_TAG_NEG_BIGNUM = 3, // byte string
84 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
85 CBOR_TAG_BIGFLOAT = 5, // array
86
87 CBOR_TAG_CONV_B64U = 21, // base64url, any
88 CBOR_TAG_CONV_B64 = 22, // base64, any
89 CBOR_TAG_CONV_HEX = 23, // base16, any
90 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
91
92 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
93 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
94 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
95 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
96 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
97
98 CBOR_TAG_MAGIC = 55799, // self-describe cbor
99};
100
14#define F_SHRINK 0x00000200UL 101#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 102#define F_ALLOW_UNKNOWN 0x00000002UL
103#define F_ALLOW_SHARING 0x00000004UL
104#define F_ALLOW_CYCLES 0x00000008UL
105#define F_FORBID_OBJECTS 0x00000010UL
106#define F_PACK_STRINGS 0x00000020UL
107#define F_TEXT_KEYS 0x00000040UL
108#define F_TEXT_STRINGS 0x00000080UL
109#define F_VALIDATE_UTF8 0x00000100UL
16 110
17#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
18 112
19#define SB do { 113#define SB do {
20#define SE } while (0) 114#define SE } while (0)
31#else 125#else
32# define CBOR_SLOW 0 126# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 127# define CBOR_STASH cbor_stash
34#endif 128#endif
35 129
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 130static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 131static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 132
39typedef struct { 133typedef struct {
40 U32 flags; 134 U32 flags;
41 U32 max_depth; 135 U32 max_depth;
42 STRLEN max_size; 136 STRLEN max_size;
137 SV *filter;
43 138
44 SV *cb_object; 139 // for the incremental parser
45 HV *cb_sk_object; 140 STRLEN incr_pos; // the current offset into the text
141 STRLEN incr_need; // minimum bytes needed to decode
142 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
46} CBOR; 143} CBOR;
47 144
48ecb_inline void 145ecb_inline void
49cbor_init (CBOR *cbor) 146cbor_init (CBOR *cbor)
50{ 147{
51 Zero (cbor, 1, CBOR); 148 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 149 cbor->max_depth = 512;
150}
151
152ecb_inline void
153cbor_free (CBOR *cbor)
154{
155 SvREFCNT_dec (cbor->filter);
156 SvREFCNT_dec (cbor->incr_count);
53} 157}
54 158
55///////////////////////////////////////////////////////////////////////////// 159/////////////////////////////////////////////////////////////////////////////
56// utility functions 160// utility functions
57 161
79 SvPV_renew (sv, SvCUR (sv) + 1); 183 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 184#endif
81 } 185 }
82} 186}
83 187
84///////////////////////////////////////////////////////////////////////////// 188// minimum length of a string to be registered for stringref
85// fp hell 189ecb_inline int
190minimum_string_length (UV idx)
191{
86 192
87//TODO 193 return idx <= 23 ? 3
194 : idx <= 0xffU ? 4
195 : idx <= 0xffffU ? 5
196 : idx <= 0xffffffffU ? 7
197 : 11;
198}
88 199
89///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
90// encoder 201// encoder
91 202
92// structure used for encoding CBOR 203// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
97 SV *sv; // result scalar 208 SV *sv; // result scalar
98 CBOR cbor; 209 CBOR cbor;
99 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
100} enc_t; 215} enc_t;
101 216
102ecb_inline void 217ecb_inline void
103need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
104{ 219{
105 if (ecb_expect_false (enc->cur + len >= enc->end)) 220 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
106 { 221 {
107 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
108 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
109 enc->cur = SvPVX (enc->sv) + cur; 224 enc->cur = SvPVX (enc->sv) + cur;
110 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
121static void 236static void
122encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
123{ 238{
124 need (enc, 9); 239 need (enc, 9);
125 240
126 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
128 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
129 { 244 {
130 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 246 *enc->cur++ = len;
132 } 247 }
133 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
134 { 249 {
135 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 252 *enc->cur++ = len;
138 } 253 }
139 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
140 { 255 {
141 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 260 *enc->cur++ = len;
146 } 261 }
147 else 262 else
148 { 263 {
149 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 272 *enc->cur++ = len;
158 } 273 }
159} 274}
160 275
161static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
283ecb_noinline static void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 284encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 285{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 286 STRLEN ulen = len;
287 U8 *p, *pend = (U8 *)str + len;
288
289 for (p = (U8 *)str; p < pend; ++p)
290 ulen += *p >> 7; // count set high bits
291
292 encode_uint (enc, MAJOR_TEXT, ulen);
293
294 need (enc, ulen);
295 for (p = (U8 *)str; p < pend; ++p)
296 if (*p < 0x80)
297 *enc->cur++ = *p;
298 else
299 {
300 *enc->cur++ = 0xc0 + (*p >> 6);
301 *enc->cur++ = 0x80 + (*p & 63);
302 }
303}
304
305ecb_inline void
306encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
307{
308 if (ecb_expect_false (upgrade_utf8))
309 if (!utf8)
310 {
311 encode_str_utf8 (enc, utf8, str, len);
312 return;
313 }
314
315 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
165 need (enc, len); 316 need (enc, len);
166 memcpy (enc->cur, str, len); 317 memcpy (enc->cur, str, len);
167 enc->cur += len; 318 enc->cur += len;
168} 319}
169 320
321ecb_inline void
322encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
323{
324 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
325 {
326 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
327
328 if (SvOK (*svp))
329 {
330 // already registered, use stringref
331 encode_tag (enc, CBOR_TAG_STRINGREF);
332 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
333 return;
334 }
335 else if (len >= minimum_string_length (enc->stringref_idx))
336 {
337 // register only
338 sv_setuv (*svp, enc->stringref_idx);
339 ++enc->stringref_idx;
340 }
341 }
342
343 encode_str (enc, upgrade_utf8, utf8, str, len);
344}
345
170static void encode_sv (enc_t *enc, SV *sv); 346static void encode_sv (enc_t *enc, SV *sv);
171 347
172static void 348static void
173encode_av (enc_t *enc, AV *av) 349encode_av (enc_t *enc, AV *av)
174{ 350{
177 if (enc->depth >= enc->cbor.max_depth) 353 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 354 croak (ERR_NESTING_EXCEEDED);
179 355
180 ++enc->depth; 356 ++enc->depth;
181 357
182 encode_uint (enc, 0x80, len + 1); 358 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 359
360 if (SvMAGICAL (av))
184 for (i = 0; i <= len; ++i) 361 for (i = 0; i <= len; ++i)
185 { 362 {
186 SV **svp = av_fetch (av, i, 0); 363 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 364 encode_sv (enc, svp ? *svp : &PL_sv_undef);
188 } 365 }
366 else
367 for (i = 0; i <= len; ++i)
368 {
369 SV *sv = AvARRAY (av)[i];
370 encode_sv (enc, sv ? sv : &PL_sv_undef);
371 }
189 372
190 --enc->depth; 373 --enc->depth;
191} 374}
192 375
193static void 376static void
202 385
203 int pairs = hv_iterinit (hv); 386 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 387 int mg = SvMAGICAL (hv);
205 388
206 if (mg) 389 if (mg)
207 encode_ch (enc, 0xa0 | 31); 390 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 391 else
209 encode_uint (enc, 0xa0, pairs); 392 encode_uint (enc, MAJOR_MAP, pairs);
210 393
211 while ((he = hv_iternext (hv))) 394 while ((he = hv_iternext (hv)))
212 { 395 {
213 if (HeKLEN (he) == HEf_SVKEY) 396 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 397 encode_sv (enc, HeSVKEY (he));
215 else 398 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 399 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217 400
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 401 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 402 }
220 403
221 if (mg) 404 if (mg)
222 encode_ch (enc, 0xe0 | 31); 405 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 406
224 --enc->depth; 407 --enc->depth;
225} 408}
226 409
227// encode objects, arrays and special \0=false and \1=true values. 410// encode objects, arrays and special \0=false and \1=true values.
228static void 411static void
229encode_rv (enc_t *enc, SV *sv) 412encode_rv (enc_t *enc, SV *sv)
230{ 413{
231 svtype svt;
232
233 SvGETMAGIC (sv); 414 SvGETMAGIC (sv);
415
234 svt = SvTYPE (sv); 416 svtype svt = SvTYPE (sv);
235 417
236 if (ecb_expect_false (SvOBJECT (sv))) 418 if (ecb_expect_false (SvOBJECT (sv)))
237 { 419 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 420 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 421 ? types_boolean_stash
422 : gv_stashpv ("Types::Serialiser::Boolean", 1);
423 HV *error_stash = !CBOR_SLOW || types_error_stash
424 ? types_error_stash
425 : gv_stashpv ("Types::Serialiser::Error", 1);
426 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
427 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 428 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 429
242 if (SvSTASH (sv) == stash) 430 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 431
432 if (stash == boolean_stash)
433 {
434 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
435 return;
436 }
437 else if (stash == error_stash)
438 {
439 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
440 return;
441 }
442 else if (stash == tagged_stash)
443 {
444 if (svt != SVt_PVAV)
445 croak ("encountered CBOR::XS::Tagged object that isn't an array");
446
447 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
448 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
449
450 return;
451 }
452 }
453
454 if (ecb_expect_false (SvREFCNT (sv) > 1)
455 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
456 {
457 if (!enc->shareable)
458 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
459
460 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
461
462 if (SvOK (*svp))
463 {
464 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
465 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
466 return;
467 }
244 else 468 else
245 { 469 {
246#if 0 //TODO 470 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 471 ++enc->shareable_idx;
472 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
473 }
474 }
475
476 if (ecb_expect_false (SvOBJECT (sv)))
477 {
478 HV *stash = SvSTASH (sv);
479 GV *method;
480
481 if (enc->cbor.flags & F_FORBID_OBJECTS)
482 croak ("encountered object '%s', but forbid_objects is enabled",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
484 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 485 {
486 dSP;
487
488 ENTER; SAVETMPS;
489 PUSHMARK (SP);
249 // we re-bless the reference to get overload and other niceties right 490 // 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))); 491 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 492
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 493 PUTBACK;
494 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 495 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 496 SPAGAIN;
263 497
264 // catch this surprisingly common error 498 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 499 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))); 500 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 501
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 502 encode_sv (enc, POPs);
272 503
504 PUTBACK;
505
273 FREETMPS; LEAVE; 506 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",
279 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280 } 507 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 508 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
282 encode_str (enc, "null", 4, 0); 509 {
510 dSP;
511
512 ENTER; SAVETMPS;
513 SAVESTACK_POS ();
514 PUSHMARK (SP);
515 EXTEND (SP, 2);
516 // we re-bless the reference to get overload and other niceties right
517 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
518 PUSHs (sv_cbor);
519
520 PUTBACK;
521 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
522 SPAGAIN;
523
524 // catch this surprisingly common error
525 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
526 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
527
528 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
529 encode_uint (enc, MAJOR_ARRAY, count + 1);
530 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
531
532 while (count)
533 encode_sv (enc, SP[1 - count--]);
534
535 PUTBACK;
536
537 FREETMPS; LEAVE;
538 }
283 else 539 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 540 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 541 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 542 }
289 else if (svt == SVt_PVHV) 543 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 544 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 545 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 546 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 547 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 548 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 549 encode_tag (enc, CBOR_TAG_INDIRECTION);
550 encode_sv (enc, sv);
551 }
313} 552}
314 553
315static void 554static void
316encode_nv (enc_t *enc, SV *sv) 555encode_nv (enc_t *enc, SV *sv)
317{ 556{
318 double nv = SvNVX (sv); 557 double nv = SvNVX (sv);
319 558
320 need (enc, 9); 559 need (enc, 9);
321 560
322 if (ecb_expect_false (nv == (U32)nv)) 561 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 562 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 563 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 564 else if (ecb_expect_false (nv == (float)nv))
326 { 565 {
327 uint32_t fp = ecb_float_to_binary32 (nv); 566 uint32_t fp = ecb_float_to_binary32 (nv);
328 567
329 *enc->cur++ = 0xe0 | 26; 568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
330 569
331 if (!ecb_big_endian ()) 570 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp); 571 fp = ecb_bswap32 (fp);
333 572
334 memcpy (enc->cur, &fp, 4); 573 memcpy (enc->cur, &fp, 4);
336 } 575 }
337 else 576 else
338 { 577 {
339 uint64_t fp = ecb_double_to_binary64 (nv); 578 uint64_t fp = ecb_double_to_binary64 (nv);
340 579
341 *enc->cur++ = 0xe0 | 27; 580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
342 581
343 if (!ecb_big_endian ()) 582 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp); 583 fp = ecb_bswap64 (fp);
345 584
346 memcpy (enc->cur, &fp, 8); 585 memcpy (enc->cur, &fp, 8);
355 594
356 if (SvPOKp (sv)) 595 if (SvPOKp (sv))
357 { 596 {
358 STRLEN len; 597 STRLEN len;
359 char *str = SvPV (sv, len); 598 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 599 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
361 } 600 }
362 else if (SvNOKp (sv)) 601 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 602 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 603 else if (SvIOKp (sv))
365 { 604 {
366 if (SvIsUV (sv)) 605 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
368 else if (SvIVX (sv) >= 0) 607 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
370 else 609 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 610 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
372 } 611 }
373 else if (SvROK (sv)) 612 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 613 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 614 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 616 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 618 else
380 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 619 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 620 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 621}
383 622
384static SV * 623static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 624encode_cbor (SV *scalar, CBOR *cbor)
386{ 625{
387 enc_t enc; 626 enc_t enc = { 0 };
388 627
389 enc.cbor = *cbor; 628 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 629 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 630 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 631 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 632
395 SvPOK_only (enc.sv); 633 SvPOK_only (enc.sv);
634
635 if (cbor->flags & F_PACK_STRINGS)
636 {
637 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
638 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
639 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
640 }
641
396 encode_sv (&enc, scalar); 642 encode_sv (&enc, scalar);
397 643
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 644 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 645 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 646
414 U8 *end; // end of input string 660 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 661 const char *err; // parse error, if != 0
416 CBOR cbor; 662 CBOR cbor;
417 U32 depth; // recursion depth 663 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 664 U32 maxdepth; // recursion depth limit
665 AV *shareable;
666 AV *stringref;
667 SV *decode_tagged;
668 SV *err_sv; // optional sv for error, needs to be freed
419} dec_t; 669} dec_t;
420 670
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 671// set dec->err to ERRSV
672ecb_cold static void
673err_errsv (dec_t *dec)
674{
675 if (!dec->err)
676 {
677 dec->err_sv = newSVsv (ERRSV);
422 678
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 679 // chop off the trailing \n
680 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
681 *SvEND (dec->err_sv) = 0;
424 682
683 dec->err = SvPVutf8_nolen (dec->err_sv);
684 }
685}
686
687// the following functions are used to reduce code size and help the compiler to optimise
688ecb_cold static void
689err_set (dec_t *dec, const char *reason)
690{
691 if (!dec->err)
692 dec->err = reason;
693}
694
695ecb_cold static void
696err_unexpected_end (dec_t *dec)
697{
698 err_set (dec, "unexpected end of CBOR data");
699}
700
701ecb_cold static void
702err_nesting_exceeded (dec_t *dec)
703{
704 err_set (dec, ERR_NESTING_EXCEEDED);
705}
706
707#define ERR_DO(do) SB do; goto fail; SE
708#define ERR(reason) ERR_DO (err_set (dec, reason))
709#define ERR_ERRSV ERR_DO (err_errsv (dec))
710
711#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
712
425#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 713#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
426#define DEC_DEC_DEPTH --dec->depth 714#define DEC_DEC_DEPTH --dec->depth
427 715
428static UV 716static UV
429decode_uint (dec_t *dec) 717decode_uint (dec_t *dec)
430{ 718{
431 switch (*dec->cur & 31) 719 U8 m = *dec->cur & MINOR_MASK;
432 { 720 ++dec->cur;
433 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436 return *dec->cur++ & 31;
437 721
438 case 24: 722 if (ecb_expect_true (m < LENGTH_EXT1))
723 return m;
724 else if (ecb_expect_true (m == LENGTH_EXT1))
725 {
439 WANT (2); 726 WANT (1);
440 dec->cur += 2; 727 dec->cur += 1;
441 return dec->cur[-1]; 728 return dec->cur[-1];
442 729 }
443 case 25: 730 else if (ecb_expect_true (m == LENGTH_EXT2))
731 {
444 WANT (3); 732 WANT (2);
445 dec->cur += 3; 733 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 734 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 735 | ((UV)dec->cur[-1]);
448 736 }
449 case 26: 737 else if (ecb_expect_true (m == LENGTH_EXT4))
738 {
450 WANT (5); 739 WANT (4);
451 dec->cur += 5; 740 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 741 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 742 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 743 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 744 | ((UV)dec->cur[-1]);
456 745 }
457 case 27: 746 else if (ecb_expect_true (m == LENGTH_EXT8))
747 {
458 WANT (9); 748 WANT (8);
459 dec->cur += 9; 749 dec->cur += 8;
750
751 return
752#if UVSIZE < 8
753 0
754#else
460 return (((UV)dec->cur[-8]) << 56) 755 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 756 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 757 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 758 | (((UV)dec->cur[-5]) << 32)
759#endif
464 | (((UV)dec->cur[-4]) << 24) 760 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 761 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 762 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 763 | ((UV)dec->cur[-1]);
468 764 }
469 default: 765 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 766 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 767
473fail: 768fail:
474 return 0; 769 return 0;
475} 770}
476 771
481{ 776{
482 AV *av = newAV (); 777 AV *av = newAV ();
483 778
484 DEC_INC_DEPTH; 779 DEC_INC_DEPTH;
485 780
486 if ((*dec->cur & 31) == 31) 781 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 782 {
488 ++dec->cur; 783 ++dec->cur;
489 784
490 for (;;) 785 for (;;)
491 { 786 {
492 WANT (1); 787 WANT (1);
493 788
494 if (*dec->cur == (0xe0 | 31)) 789 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
495 { 790 {
496 ++dec->cur; 791 ++dec->cur;
497 break; 792 break;
498 } 793 }
499 794
500 av_push (av, decode_sv (dec)); 795 av_push (av, decode_sv (dec));
501 } 796 }
502 } 797 }
503 else 798 else
504 { 799 {
505 int i, len = decode_uint (dec); 800 UV i, len = decode_uint (dec);
506 801
802 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
507 av_fill (av, len - 1); 803 av_fill (av, len - 1);
508 804
509 for (i = 0; i < len; ++i) 805 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 806 AvARRAY (av)[i] = decode_sv (dec);
511 } 807 }
517 SvREFCNT_dec (av); 813 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 814 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 815 return &PL_sv_undef;
520} 816}
521 817
818static void
819decode_he (dec_t *dec, HV *hv)
820{
821 // for speed reasons, we specialcase single-string
822 // byte or utf-8 strings as keys, but only when !stringref
823
824 if (ecb_expect_true (!dec->stringref))
825 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
826 {
827 STRLEN len = decode_uint (dec);
828 char *key = (char *)dec->cur;
829
830 WANT (len);
831 dec->cur += len;
832
833 hv_store (hv, key, len, decode_sv (dec), 0);
834
835 return;
836 }
837 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
838 {
839 STRLEN len = decode_uint (dec);
840 char *key = (char *)dec->cur;
841
842 WANT (len);
843 dec->cur += len;
844
845 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
846 if (!is_utf8_string (key, len))
847 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
848
849 hv_store (hv, key, -len, decode_sv (dec), 0);
850
851 return;
852 }
853
854 SV *k = decode_sv (dec);
855 SV *v = decode_sv (dec);
856
857 // we leak memory if uncaught exceptions are thrown by random magical
858 // methods, and this is hopefully the only place where it can happen,
859 // so if there is a chance of an exception, take the very slow path.
860 // since catching exceptions is "undocumented/internal/forbidden" by
861 // the new p5p powers, we need to call out to a perl function :/
862 if (ecb_expect_false (SvAMAGIC (k)))
863 {
864 dSP;
865
866 ENTER; SAVETMPS;
867 PUSHMARK (SP);
868 EXTEND (SP, 3);
869 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
870 PUSHs (sv_2mortal (k));
871 PUSHs (sv_2mortal (v));
872
873 PUTBACK;
874 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
875 SPAGAIN;
876
877 FREETMPS; LEAVE;
878
879 if (SvTRUE (ERRSV))
880 ERR_ERRSV;
881
882 return;
883 }
884
885 hv_store_ent (hv, k, v, 0);
886 SvREFCNT_dec (k);
887
888fail:
889 ;
890}
891
522static SV * 892static SV *
523decode_hv (dec_t *dec) 893decode_hv (dec_t *dec)
524{ 894{
525 HV *hv = newHV (); 895 HV *hv = newHV ();
526 896
527 DEC_INC_DEPTH; 897 DEC_INC_DEPTH;
528 898
529 if ((*dec->cur & 31) == 31) 899 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 900 {
531 ++dec->cur; 901 ++dec->cur;
532 902
533 for (;;) 903 for (;;)
534 { 904 {
535 WANT (1); 905 WANT (1);
536 906
537 if (*dec->cur == (0xe0 | 31)) 907 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
538 { 908 {
539 ++dec->cur; 909 ++dec->cur;
540 break; 910 break;
541 } 911 }
542 912
543 SV *k = decode_sv (dec); 913 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 914 }
548 } 915 }
549 else 916 else
550 { 917 {
551 int len = decode_uint (dec); 918 UV pairs = decode_uint (dec);
552 919
920 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
921
553 while (len--) 922 while (pairs--)
554 { 923 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 } 924 }
561 925
562 DEC_DEC_DEPTH; 926 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 927 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
719 928
720fail: 929fail:
721 SvREFCNT_dec (hv); 930 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 931 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 932 return &PL_sv_undef;
724} 933}
725 934
726static SV * 935static SV *
727decode_str (dec_t *dec, int utf8) 936decode_str (dec_t *dec, int utf8)
728{ 937{
729 SV *sv; 938 SV *sv = 0;
730 939
731 if ((*dec->cur & 31) == 31) 940 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 941 {
942 // indefinite length strings
733 ++dec->cur; 943 ++dec->cur;
734 944
945 U8 major = *dec->cur & MAJOR_MISC;
946
735 sv = newSVpvn ("", 0); 947 sv = newSVpvn ("", 0);
736 948
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 949 for (;;)
739 { 950 {
740 WANT (1); 951 WANT (1);
741 952
742 if (*dec->cur == (0xe0 | 31)) 953 if ((*dec->cur - major) > LENGTH_EXT8)
954 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 955 {
744 ++dec->cur; 956 ++dec->cur;
745 break; 957 break;
746 } 958 }
959 else
960 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 961
748 SV *sv2 = decode_sv (dec); 962 STRLEN len = decode_uint (dec);
749 sv_catsv (sv, sv2); 963
964 WANT (len);
965 sv_catpvn (sv, dec->cur, len);
966 dec->cur += len;
750 } 967 }
751 } 968 }
752 else 969 else
753 { 970 {
754 STRLEN len = decode_uint (dec); 971 STRLEN len = decode_uint (dec);
755 972
756 WANT (len); 973 WANT (len);
757 sv = newSVpvn (dec->cur, len); 974 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 975 dec->cur += len;
976
977 if (ecb_expect_false (dec->stringref)
978 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
979 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 980 }
760 981
761 if (utf8) 982 if (utf8)
983 {
984 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
985 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
986 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
987
762 SvUTF8_on (sv); 988 SvUTF8_on (sv);
989 }
763 990
764 return sv; 991 return sv;
765 992
766fail: 993fail:
994 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 995 return &PL_sv_undef;
768} 996}
769 997
770static SV * 998static SV *
771decode_tagged (dec_t *dec) 999decode_tagged (dec_t *dec)
772{ 1000{
1001 SV *sv = 0;
773 UV tag = decode_uint (dec); 1002 UV tag = decode_uint (dec);
1003
1004 WANT (1);
1005
1006 switch (tag)
1007 {
1008 case CBOR_TAG_MAGIC:
774 SV *sv = decode_sv (dec); 1009 sv = decode_sv (dec);
1010 break;
775 1011
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1012 case CBOR_TAG_INDIRECTION:
1013 sv = newRV_noinc (decode_sv (dec));
1014 break;
1015
1016 case CBOR_TAG_STRINGREF_NAMESPACE:
1017 {
1018 // do nmot use SAVETMPS/FREETMPS, as these will
1019 // erase mortalised caches, e.g. "shareable"
1020 ENTER;
1021
1022 SAVESPTR (dec->stringref);
1023 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1024
1025 sv = decode_sv (dec);
1026
1027 LEAVE;
1028 }
1029 break;
1030
1031 case CBOR_TAG_STRINGREF:
1032 {
1033 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1034 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1035
1036 UV idx = decode_uint (dec);
1037
1038 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
1039 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1040
1041 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1042 }
1043 break;
1044
1045 case CBOR_TAG_VALUE_SHAREABLE:
1046 {
1047 if (ecb_expect_false (!dec->shareable))
1048 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1049
1050 if (dec->cbor.flags & F_ALLOW_CYCLES)
1051 {
1052 sv = newSV (0);
1053 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1054
1055 SV *osv = decode_sv (dec);
1056 sv_setsv (sv, osv);
1057 SvREFCNT_dec_NN (osv);
1058 }
1059 else
1060 {
1061 av_push (dec->shareable, &PL_sv_undef);
1062 int idx = AvFILLp (dec->shareable);
1063 sv = decode_sv (dec);
1064 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1065 }
1066 }
1067 break;
1068
1069 case CBOR_TAG_VALUE_SHAREDREF:
1070 {
1071 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1072 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1073
1074 UV idx = decode_uint (dec);
1075
1076 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1077 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1078
1079 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1080
1081 if (sv == &PL_sv_undef)
1082 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1083 }
1084 break;
1085
1086 case CBOR_TAG_PERL_OBJECT:
1087 {
1088 if (dec->cbor.flags & F_FORBID_OBJECTS)
1089 goto filter;
1090
1091 sv = decode_sv (dec);
1092
1093 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1094 ERR ("corrupted CBOR data (non-array perl object)");
1095
1096 AV *av = (AV *)SvRV (sv);
1097 int len = av_len (av) + 1;
1098 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1099
1100 if (!stash)
1101 ERR ("cannot decode perl-object (package does not exist)");
1102
1103 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1104
1105 if (!method)
1106 ERR ("cannot decode perl-object (package does not have a THAW method)");
1107
1108 dSP;
1109
1110 ENTER; SAVETMPS;
1111 PUSHMARK (SP);
1112 EXTEND (SP, len + 1);
1113 // we re-bless the reference to get overload and other niceties right
1114 PUSHs (*av_fetch (av, 0, 1));
1115 PUSHs (sv_cbor);
1116
1117 int i;
1118
1119 for (i = 1; i < len; ++i)
1120 PUSHs (*av_fetch (av, i, 1));
1121
1122 PUTBACK;
1123 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1124 SPAGAIN;
1125
1126 if (SvTRUE (ERRSV))
1127 {
1128 FREETMPS; LEAVE;
1129 ERR_ERRSV;
1130 }
1131
1132 SvREFCNT_dec (sv);
1133 sv = SvREFCNT_inc (POPs);
1134
1135 PUTBACK;
1136
1137 FREETMPS; LEAVE;
1138 }
1139 break;
1140
1141 default:
1142 filter:
1143 {
1144 SV *tag_sv = newSVuv (tag);
1145
1146 sv = decode_sv (dec);
1147
1148 dSP;
1149 ENTER; SAVETMPS;
1150 SAVESTACK_POS ();
1151 PUSHMARK (SP);
1152 EXTEND (SP, 2);
1153 PUSHs (tag_sv);
1154 PUSHs (sv);
1155
1156 PUTBACK;
1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1158 SPAGAIN;
1159
1160 if (SvTRUE (ERRSV))
1161 {
1162 SvREFCNT_dec (tag_sv);
1163 FREETMPS; LEAVE;
1164 ERR_ERRSV;
1165 }
1166
1167 if (count)
1168 {
1169 SvREFCNT_dec (tag_sv);
1170 SvREFCNT_dec (sv);
1171 sv = SvREFCNT_inc (POPs);
1172 }
1173 else
1174 {
1175 AV *av = newAV ();
1176 av_push (av, tag_sv);
1177 av_push (av, sv);
1178
1179 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1180 ? cbor_tagged_stash
1181 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1182 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1183 }
1184
1185 PUTBACK;
1186
1187 FREETMPS; LEAVE;
1188 }
1189 break;
1190 }
1191
777 return sv; 1192 return sv;
778 1193
779 AV *av = newAV (); 1194fail:
780 av_push (av, newSVuv (tag)); 1195 SvREFCNT_dec (sv);
781 av_push (av, sv); 1196 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1197}
784 1198
785static SV * 1199static SV *
786decode_sv (dec_t *dec) 1200decode_sv (dec_t *dec)
787{ 1201{
788 WANT (1); 1202 WANT (1);
789 1203
790 switch (*dec->cur >> 5) 1204 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1205 {
792 case 0: // unsigned int 1206 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1207 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1208 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1209 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1210 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1211 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1212 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1213
800 return decode_str (dec, 1); 1214 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 return decode_tagged (dec);
807 case 7: // misc
808 switch (*dec->cur++ & 31) 1215 switch (*dec->cur++ & MINOR_MASK)
809 { 1216 {
810 case 20: 1217 case SIMPLE_FALSE:
811#if CBOR_SLOW 1218#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1219 types_false = get_bool ("Types::Serialiser::false");
813#endif 1220#endif
814 return newSVsv (cbor_false); 1221 return newSVsv (types_false);
815 case 21: 1222 case SIMPLE_TRUE:
816#if CBOR_SLOW 1223#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1224 types_true = get_bool ("Types::Serialiser::true");
818#endif 1225#endif
819 return newSVsv (cbor_true); 1226 return newSVsv (types_true);
820 case 22: 1227 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1228 return newSVsv (&PL_sv_undef);
1229 case SIMPLE_UNDEF:
1230#if CBOR_SLOW
1231 types_error = get_bool ("Types::Serialiser::error");
1232#endif
1233 return newSVsv (types_error);
822 1234
823 case 25: 1235 case MISC_FLOAT16:
824 { 1236 {
825 WANT (2); 1237 WANT (2);
826 1238
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1239 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1240 dec->cur += 2;
829 1241
830 return newSVnv (ecb_binary16_to_float (fp)); 1242 return newSVnv (ecb_binary16_to_float (fp));
831 } 1243 }
832 1244
833 case 26: 1245 case MISC_FLOAT32:
834 { 1246 {
835 uint32_t fp; 1247 uint32_t fp;
836 WANT (4); 1248 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1249 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1250 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1253 fp = ecb_bswap32 (fp);
842 1254
843 return newSVnv (ecb_binary32_to_float (fp)); 1255 return newSVnv (ecb_binary32_to_float (fp));
844 } 1256 }
845 1257
846 case 27: 1258 case MISC_FLOAT64:
847 { 1259 {
848 uint64_t fp; 1260 uint64_t fp;
849 WANT (8); 1261 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1262 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1263 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1266 fp = ecb_bswap64 (fp);
855 1267
856 return newSVnv (ecb_binary64_to_double (fp)); 1268 return newSVnv (ecb_binary64_to_double (fp));
857 } 1269 }
858 1270
859 // 0..19 unassigned 1271 // 0..19 unassigned simple
860 // 24 reserved + unassigned (reserved values are not encodable) 1272 // 24 reserved + unassigned simple (reserved values are not encodable)
1273 // 28-30 unassigned misc
1274 // 31 break code
861 default: 1275 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1276 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
863 } 1277 }
864 1278
865 break; 1279 break;
866 } 1280 }
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 1281
924fail: 1282fail:
925 return &PL_sv_undef; 1283 return &PL_sv_undef;
926} 1284}
927 1285
928static SV * 1286static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1287decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1288{
931 dec_t dec; 1289 dec_t dec = { 0 };
932 SV *sv; 1290 SV *sv;
1291 STRLEN len;
1292 char *data = SvPVbyte (string, len);
933 1293
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) 1294 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", 1295 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); 1296 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1297
969 dec.cbor = *cbor; 1298 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1299 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1300 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 1301
978 sv = decode_sv (&dec); 1302 sv = decode_sv (&dec);
979 1303
980 if (offset_return) 1304 if (offset_return)
981 *offset_return = dec.cur; 1305 *offset_return = dec.cur;
984 if (dec.cur != dec.end && !dec.err) 1308 if (dec.cur != dec.end && !dec.err)
985 dec.err = "garbage after CBOR object"; 1309 dec.err = "garbage after CBOR object";
986 1310
987 if (dec.err) 1311 if (dec.err)
988 { 1312 {
1313 if (dec.shareable)
1314 {
1315 // need to break cyclic links, which would all be in shareable
1316 int i;
1317 SV **svp;
1318
1319 for (i = av_len (dec.shareable) + 1; i--; )
1320 if ((svp = av_fetch (dec.shareable, i, 0)))
1321 sv_setsv (*svp, &PL_sv_undef);
1322 }
1323
989 SvREFCNT_dec (sv); 1324 SvREFCNT_dec (sv);
1325
1326 if (dec.err_sv)
1327 sv_2mortal (dec.err_sv);
1328
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1329 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1330 }
992 1331
993 sv = sv_2mortal (sv); 1332 sv = sv_2mortal (sv);
994 1333
995 return sv; 1334 return sv;
996} 1335}
997 1336
1337/////////////////////////////////////////////////////////////////////////////
1338// incremental parser
1339
1340#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1341
1342// returns 0 for notyet, 1 for success or error
1343static int
1344incr_parse (CBOR *self, SV *cborstr)
1345{
1346 STRLEN cur;
1347 SvPV (cborstr, cur);
1348
1349 while (ecb_expect_true (self->incr_need <= cur))
1350 {
1351 // table of integer count bytes
1352 static I8 incr_len[MINOR_MASK + 1] = {
1353 0, 0, 0, 0, 0, 0, 0, 0,
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 1, 2, 4, 8,-1,-1,-1,-2
1357 };
1358
1359 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1360 U8 m = *p & MINOR_MASK;
1361 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1362 I8 ilen = incr_len[m];
1363
1364 self->incr_need = self->incr_pos + 1;
1365
1366 if (ecb_expect_false (ilen < 0))
1367 {
1368 if (m != MINOR_INDEF)
1369 return 1; // error
1370
1371 if (*p == (MAJOR_MISC | MINOR_INDEF))
1372 {
1373 if (count >= 0)
1374 return 1; // error
1375
1376 count = 1;
1377 }
1378 else
1379 {
1380 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1381 count = -1;
1382 }
1383 }
1384 else
1385 {
1386 self->incr_need += ilen;
1387 if (ecb_expect_false (self->incr_need > cur))
1388 return 0;
1389
1390 int major = *p >> MAJOR_SHIFT;
1391
1392 switch (major)
1393 {
1394 case MAJOR_TAG >> MAJOR_SHIFT:
1395 ++count; // tags merely prefix another value
1396 break;
1397
1398 case MAJOR_BYTES >> MAJOR_SHIFT:
1399 case MAJOR_TEXT >> MAJOR_SHIFT:
1400 case MAJOR_ARRAY >> MAJOR_SHIFT:
1401 case MAJOR_MAP >> MAJOR_SHIFT:
1402 {
1403 UV len;
1404
1405 if (ecb_expect_false (ilen))
1406 {
1407 len = 0;
1408
1409 do {
1410 len = (len << 8) | *++p;
1411 } while (--ilen);
1412 }
1413 else
1414 len = m;
1415
1416 switch (major)
1417 {
1418 case MAJOR_BYTES >> MAJOR_SHIFT:
1419 case MAJOR_TEXT >> MAJOR_SHIFT:
1420 self->incr_need += len;
1421 if (ecb_expect_false (self->incr_need > cur))
1422 return 0;
1423
1424 break;
1425
1426 case MAJOR_MAP >> MAJOR_SHIFT:
1427 len <<= 1;
1428 case MAJOR_ARRAY >> MAJOR_SHIFT:
1429 if (len)
1430 {
1431 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1432 count = len + 1;
1433 }
1434 break;
1435 }
1436 }
1437 }
1438 }
1439
1440 self->incr_pos = self->incr_need;
1441
1442 if (count > 0)
1443 {
1444 while (!--count)
1445 {
1446 if (!AvFILLp (self->incr_count))
1447 return 1; // done
1448
1449 SvREFCNT_dec_NN (av_pop (self->incr_count));
1450 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1451 }
1452
1453 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1454 }
1455 }
1456
1457 return 0;
1458}
1459
1460
998///////////////////////////////////////////////////////////////////////////// 1461/////////////////////////////////////////////////////////////////////////////
999// XS interface functions 1462// XS interface functions
1000 1463
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1464MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1465
1003BOOT: 1466BOOT:
1004{ 1467{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1468 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1469 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1470
1008 cbor_true = get_bool ("CBOR::XS::true"); 1471 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1472 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1473
1474 types_true = get_bool ("Types::Serialiser::true" );
1475 types_false = get_bool ("Types::Serialiser::false");
1476 types_error = get_bool ("Types::Serialiser::error");
1477
1478 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1479
1480 sv_cbor = newSVpv ("CBOR", 0);
1481 SvREADONLY_on (sv_cbor);
1482
1483 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1010} 1484}
1011 1485
1012PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
1013 1487
1014void CLONE (...) 1488void CLONE (...)
1015 CODE: 1489 CODE:
1016 cbor_stash = 0; 1490 cbor_stash = 0;
1491 cbor_tagged_stash = 0;
1492 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1493 types_boolean_stash = 0;
1018 1494
1019void new (char *klass) 1495void new (char *klass)
1020 PPCODE: 1496 PPCODE:
1021{ 1497{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1498 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1506
1031void shrink (CBOR *self, int enable = 1) 1507void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1508 ALIAS:
1033 shrink = F_SHRINK 1509 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1510 allow_unknown = F_ALLOW_UNKNOWN
1511 allow_sharing = F_ALLOW_SHARING
1512 allow_cycles = F_ALLOW_CYCLES
1513 forbid_objects = F_FORBID_OBJECTS
1514 pack_strings = F_PACK_STRINGS
1515 text_keys = F_TEXT_KEYS
1516 text_strings = F_TEXT_STRINGS
1517 validate_utf8 = F_VALIDATE_UTF8
1035 PPCODE: 1518 PPCODE:
1036{ 1519{
1037 if (enable) 1520 if (enable)
1038 self->flags |= ix; 1521 self->flags |= ix;
1039 else 1522 else
1044 1527
1045void get_shrink (CBOR *self) 1528void get_shrink (CBOR *self)
1046 ALIAS: 1529 ALIAS:
1047 get_shrink = F_SHRINK 1530 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1531 get_allow_unknown = F_ALLOW_UNKNOWN
1532 get_allow_sharing = F_ALLOW_SHARING
1533 get_allow_cycles = F_ALLOW_CYCLES
1534 get_forbid_objects = F_FORBID_OBJECTS
1535 get_pack_strings = F_PACK_STRINGS
1536 get_text_keys = F_TEXT_KEYS
1537 get_text_strings = F_TEXT_STRINGS
1538 get_validate_utf8 = F_VALIDATE_UTF8
1049 PPCODE: 1539 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1540 XPUSHs (boolSV (self->flags & ix));
1051 1541
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1542void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1543 PPCODE:
1069 CODE: 1559 CODE:
1070 RETVAL = self->max_size; 1560 RETVAL = self->max_size;
1071 OUTPUT: 1561 OUTPUT:
1072 RETVAL 1562 RETVAL
1073 1563
1074#if 0 //TODO 1564void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1565 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1566 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1567 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1568 XPUSHs (ST (0));
1083}
1084 1569
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1570SV *get_filter (CBOR *self)
1086 PPCODE: 1571 CODE:
1087{ 1572 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1573 OUTPUT:
1089 self->cb_sk_object = newHV (); 1574 RETVAL
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 1575
1109void encode (CBOR *self, SV *scalar) 1576void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1577 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1578 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1579 XPUSHs (scalar);
1125 EXTEND (SP, 2); 1592 EXTEND (SP, 2);
1126 PUSHs (sv); 1593 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1594 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1595}
1129 1596
1597void incr_parse (CBOR *self, SV *cborstr)
1598 ALIAS:
1599 incr_parse_multiple = 1
1600 PPCODE:
1601{
1602 if (SvUTF8 (cborstr))
1603 sv_utf8_downgrade (cborstr, 0);
1604
1605 if (!self->incr_count)
1606 {
1607 self->incr_count = newAV ();
1608 self->incr_pos = 0;
1609 self->incr_need = 1;
1610
1611 av_push (self->incr_count, newSViv (1));
1612 }
1613
1614 do
1615 {
1616 if (!incr_parse (self, cborstr))
1617 {
1618 if (self->incr_need > self->max_size && self->max_size)
1619 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1620 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1621
1622 break;
1623 }
1624
1625 SV *sv;
1626 char *offset;
1627
1628 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1629 XPUSHs (sv);
1630
1631 sv_chop (cborstr, offset);
1632
1633 av_clear (self->incr_count);
1634 av_push (self->incr_count, newSViv (1));
1635
1636 self->incr_pos = 0;
1637 self->incr_need = self->incr_pos + 1;
1638 }
1639 while (ix);
1640}
1641
1642void incr_reset (CBOR *self)
1643 CODE:
1644{
1645 SvREFCNT_dec (self->incr_count);
1646 self->incr_count = 0;
1647}
1648
1130void DESTROY (CBOR *self) 1649void DESTROY (CBOR *self)
1131 CODE: 1650 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1651 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1652
1135PROTOTYPES: ENABLE 1653PROTOTYPES: ENABLE
1136 1654
1137void encode_cbor (SV *scalar) 1655void encode_cbor (SV *scalar)
1656 ALIAS:
1657 encode_cbor = 0
1658 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1659 PPCODE:
1139{ 1660{
1140 CBOR cbor; 1661 CBOR cbor;
1141 cbor_init (&cbor); 1662 cbor_init (&cbor);
1663 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1664 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1665 XPUSHs (scalar);
1144} 1666}
1145 1667
1146void decode_cbor (SV *cborstr) 1668void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines