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 1.6 by root, Sat Oct 26 22:25:47 2013 UTC vs.
Revision 1.52 by root, Sun Apr 24 19:31:55 2016 UTC

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 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_VALIDATE_UTF8 0x00000040UL
16 107
17#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
18 109
19#define SB do { 110#define SB do {
20#define SE } while (0) 111#define SE } while (0)
31#else 122#else
32# define CBOR_SLOW 0 123# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
34#endif 125#endif
35 126
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 129
39typedef struct { 130typedef struct {
40 U32 flags; 131 U32 flags;
41 U32 max_depth; 132 U32 max_depth;
42 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 140} CBOR;
44 141
45ecb_inline void 142ecb_inline void
46cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
47{ 144{
48 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 146 cbor->max_depth = 512;
147}
148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
50} 154}
51 155
52///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
53// utility functions 157// utility functions
54 158
76 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 181#endif
78 } 182 }
79} 183}
80 184
81///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
82// fp hell 186ecb_inline int
83 187minimum_string_length (UV idx)
84//TODO 188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
85 199
86///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
87// encoder 201// encoder
88 202
89// structure used for encoding CBOR 203// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
94 SV *sv; // result scalar 208 SV *sv; // result scalar
95 CBOR cbor; 209 CBOR cbor;
96 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;
97} enc_t; 215} enc_t;
98 216
99ecb_inline void 217ecb_inline void
100need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
101{ 219{
118static void 236static void
119encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
120{ 238{
121 need (enc, 9); 239 need (enc, 9);
122 240
123 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
125 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
126 { 244 {
127 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 246 *enc->cur++ = len;
129 } 247 }
130 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
131 { 249 {
132 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 252 *enc->cur++ = len;
135 } 253 }
136 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
137 { 255 {
138 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 260 *enc->cur++ = len;
143 } 261 }
144 else 262 else
145 { 263 {
146 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 272 *enc->cur++ = len;
155 } 273 }
156} 274}
157 275
158static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 284{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 // exceptional path for bytze strings that need to be utf8-encoded
289 STRLEN ulen = len;
290 U8 *p, *pend = (U8 *)str + len;
291
292 for (p = (U8 *)str; p < pend; ++p)
293 ulen += *p >> 7; // count set high bits
294
295 encode_uint (enc, MAJOR_TEXT, ulen);
296
297 need (enc, ulen);
298 for (p = (U8 *)str; p < pend; ++p)
299 if (*p < 0x80)
300 *enc->cur++ = *p;
301 else
302 {
303 *enc->cur++ = 0xc0 + (*p >> 6);
304 *enc->cur++ = 0x80 + (*p & 63);
305 }
306
307 return;
308 }
309
310 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 311 need (enc, len);
163 memcpy (enc->cur, str, len); 312 memcpy (enc->cur, str, len);
164 enc->cur += len; 313 enc->cur += len;
165} 314}
166 315
316static void
317encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
318{
319 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
320 {
321 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
322
323 if (SvOK (*svp))
324 {
325 // already registered, use stringref
326 encode_tag (enc, CBOR_TAG_STRINGREF);
327 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
328 return;
329 }
330 else if (len >= minimum_string_length (enc->stringref_idx))
331 {
332 // register only
333 sv_setuv (*svp, enc->stringref_idx);
334 ++enc->stringref_idx;
335 }
336 }
337
338 encode_str (enc, utf8, str, len);
339}
340
167static void encode_sv (enc_t *enc, SV *sv); 341static void encode_sv (enc_t *enc, SV *sv);
168 342
169static void 343static void
170encode_av (enc_t *enc, AV *av) 344encode_av (enc_t *enc, AV *av)
171{ 345{
174 if (enc->depth >= enc->cbor.max_depth) 348 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 349 croak (ERR_NESTING_EXCEEDED);
176 350
177 ++enc->depth; 351 ++enc->depth;
178 352
179 encode_uint (enc, 0x80, len + 1); 353 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 354
355 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 356 for (i = 0; i <= len; ++i)
182 { 357 {
183 SV **svp = av_fetch (av, i, 0); 358 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 359 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 360 }
361 else
362 for (i = 0; i <= len; ++i)
363 {
364 SV *sv = AvARRAY (av)[i];
365 encode_sv (enc, sv ? sv : &PL_sv_undef);
366 }
186 367
187 --enc->depth; 368 --enc->depth;
188} 369}
189 370
190static void 371static void
199 380
200 int pairs = hv_iterinit (hv); 381 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 382 int mg = SvMAGICAL (hv);
202 383
203 if (mg) 384 if (mg)
204 encode_ch (enc, 0xa0 | 31); 385 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 386 else
206 encode_uint (enc, 0xa0, pairs); 387 encode_uint (enc, MAJOR_MAP, pairs);
207 388
208 while ((he = hv_iternext (hv))) 389 while ((he = hv_iternext (hv)))
209 { 390 {
210 if (HeKLEN (he) == HEf_SVKEY) 391 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 392 encode_sv (enc, HeSVKEY (he));
212 else 393 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 394 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 395
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 396 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 397 }
217 398
218 if (mg) 399 if (mg)
219 encode_ch (enc, 0xe0 | 31); 400 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 401
221 --enc->depth; 402 --enc->depth;
222} 403}
223 404
224// encode objects, arrays and special \0=false and \1=true values. 405// encode objects, arrays and special \0=false and \1=true values.
225static void 406static void
226encode_rv (enc_t *enc, SV *sv) 407encode_rv (enc_t *enc, SV *sv)
227{ 408{
228 svtype svt;
229
230 SvGETMAGIC (sv); 409 SvGETMAGIC (sv);
410
231 svt = SvTYPE (sv); 411 svtype svt = SvTYPE (sv);
232 412
233 if (ecb_expect_false (SvOBJECT (sv))) 413 if (ecb_expect_false (SvOBJECT (sv)))
234 { 414 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 415 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 416 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 417 : gv_stashpv ("Types::Serialiser::Boolean", 1);
418 HV *error_stash = !CBOR_SLOW || types_error_stash
419 ? types_error_stash
420 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 421 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 422 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 423 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 424
425 HV *stash = SvSTASH (sv);
426
242 if (SvSTASH (sv) == boolean_stash) 427 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 428 {
429 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
430 return;
431 }
432 else if (stash == error_stash)
433 {
434 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 return;
436 }
244 else if (SvSTASH (sv) == tagged_stash) 437 else if (stash == tagged_stash)
245 { 438 {
246 if (svt != SVt_PVAV) 439 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 440 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 441
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 442 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 443 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
444
445 return;
446 }
447 }
448
449 if (ecb_expect_false (SvREFCNT (sv) > 1)
450 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
451 {
452 if (!enc->shareable)
453 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
454
455 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
456
457 if (SvOK (*svp))
458 {
459 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
460 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
461 return;
251 } 462 }
252 else 463 else
253 { 464 {
465 sv_setuv (*svp, enc->shareable_idx);
466 ++enc->shareable_idx;
467 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
468 }
469 }
470
471 if (ecb_expect_false (SvOBJECT (sv)))
472 {
473 HV *stash = SvSTASH (sv);
474 GV *method;
475
476 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
477 {
478 dSP;
479
480 ENTER; SAVETMPS;
481 PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 482 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 483 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 484
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 485 PUTBACK;
486 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 487 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 488 SPAGAIN;
268 489
269 // catch this surprisingly common error 490 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 491 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 492 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 493
273 sv = POPs; 494 encode_sv (enc, POPs);
495
274 PUTBACK; 496 PUTBACK;
275 497
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 498 FREETMPS; LEAVE;
279 } 499 }
500 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
501 {
502 dSP;
503
504 ENTER; SAVETMPS;
505 SAVESTACK_POS ();
506 PUSHMARK (SP);
507 EXTEND (SP, 2);
508 // we re-bless the reference to get overload and other niceties right
509 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
510 PUSHs (sv_cbor);
511
512 PUTBACK;
513 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
514 SPAGAIN;
515
516 // catch this surprisingly common error
517 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
518 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
519
520 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
521 encode_uint (enc, MAJOR_ARRAY, count + 1);
522 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
523
524 while (count)
525 encode_sv (enc, SP[1 - count--]);
526
527 PUTBACK;
528
529 FREETMPS; LEAVE;
530 }
280 else 531 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 532 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 533 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 534 }
285 else if (svt == SVt_PVHV) 535 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 536 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 537 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 538 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 539 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 540 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 541 encode_tag (enc, CBOR_TAG_INDIRECTION);
542 encode_sv (enc, sv);
543 }
309} 544}
310 545
311static void 546static void
312encode_nv (enc_t *enc, SV *sv) 547encode_nv (enc_t *enc, SV *sv)
313{ 548{
314 double nv = SvNVX (sv); 549 double nv = SvNVX (sv);
315 550
316 need (enc, 9); 551 need (enc, 9);
317 552
318 if (ecb_expect_false (nv == (U32)nv)) 553 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 554 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 555 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 556 else if (ecb_expect_false (nv == (float)nv))
322 { 557 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 558 uint32_t fp = ecb_float_to_binary32 (nv);
324 559
325 *enc->cur++ = 0xe0 | 26; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 561
327 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 563 fp = ecb_bswap32 (fp);
329 564
330 memcpy (enc->cur, &fp, 4); 565 memcpy (enc->cur, &fp, 4);
332 } 567 }
333 else 568 else
334 { 569 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 570 uint64_t fp = ecb_double_to_binary64 (nv);
336 571
337 *enc->cur++ = 0xe0 | 27; 572 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 573
339 if (!ecb_big_endian ()) 574 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 575 fp = ecb_bswap64 (fp);
341 576
342 memcpy (enc->cur, &fp, 8); 577 memcpy (enc->cur, &fp, 8);
351 586
352 if (SvPOKp (sv)) 587 if (SvPOKp (sv))
353 { 588 {
354 STRLEN len; 589 STRLEN len;
355 char *str = SvPV (sv, len); 590 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 591 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 592 }
358 else if (SvNOKp (sv)) 593 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 594 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 595 else if (SvIOKp (sv))
361 { 596 {
362 if (SvIsUV (sv)) 597 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 598 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 599 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 600 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 601 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 602 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 603 }
369 else if (SvROK (sv)) 604 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 605 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 606 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 607 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 608 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 609 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 610 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 611 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 612 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 613}
379 614
380static SV * 615static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 616encode_cbor (SV *scalar, CBOR *cbor)
382{ 617{
383 enc_t enc; 618 enc_t enc = { 0 };
384 619
385 enc.cbor = *cbor; 620 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 621 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 622 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 623 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 624
391 SvPOK_only (enc.sv); 625 SvPOK_only (enc.sv);
626
627 if (cbor->flags & F_PACK_STRINGS)
628 {
629 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
630 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
631 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
632 }
633
392 encode_sv (&enc, scalar); 634 encode_sv (&enc, scalar);
393 635
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 637 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 638
410 U8 *end; // end of input string 652 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
412 CBOR cbor; 654 CBOR cbor;
413 U32 depth; // recursion depth 655 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 656 U32 maxdepth; // recursion depth limit
657 AV *shareable;
658 AV *stringref;
659 SV *decode_tagged;
415} dec_t; 660} dec_t;
416 661
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 662#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 663
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 664#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
422#define DEC_DEC_DEPTH --dec->depth 667#define DEC_DEC_DEPTH --dec->depth
423 668
424static UV 669static UV
425decode_uint (dec_t *dec) 670decode_uint (dec_t *dec)
426{ 671{
427 switch (*dec->cur & 31) 672 U8 m = *dec->cur & MINOR_MASK;
428 { 673 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 674
434 case 24: 675 if (ecb_expect_true (m < LENGTH_EXT1))
676 return m;
677 else if (ecb_expect_true (m == LENGTH_EXT1))
678 {
435 WANT (2); 679 WANT (1);
436 dec->cur += 2; 680 dec->cur += 1;
437 return dec->cur[-1]; 681 return dec->cur[-1];
438 682 }
439 case 25: 683 else if (ecb_expect_true (m == LENGTH_EXT2))
684 {
440 WANT (3); 685 WANT (2);
441 dec->cur += 3; 686 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 687 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 688 | ((UV)dec->cur[-1]);
444 689 }
445 case 26: 690 else if (ecb_expect_true (m == LENGTH_EXT4))
691 {
446 WANT (5); 692 WANT (4);
447 dec->cur += 5; 693 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 694 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 695 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 696 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 697 | ((UV)dec->cur[-1]);
452 698 }
453 case 27: 699 else if (ecb_expect_true (m == LENGTH_EXT8))
700 {
454 WANT (9); 701 WANT (8);
455 dec->cur += 9; 702 dec->cur += 8;
703
704 return
705#if UVSIZE < 8
706 0
707#else
456 return (((UV)dec->cur[-8]) << 56) 708 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 709 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 710 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 711 | (((UV)dec->cur[-5]) << 32)
712#endif
460 | (((UV)dec->cur[-4]) << 24) 713 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 714 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 715 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 716 | ((UV)dec->cur[-1]);
464 717 }
465 default: 718 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 719 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 720
469fail: 721fail:
470 return 0; 722 return 0;
471} 723}
472 724
477{ 729{
478 AV *av = newAV (); 730 AV *av = newAV ();
479 731
480 DEC_INC_DEPTH; 732 DEC_INC_DEPTH;
481 733
482 if ((*dec->cur & 31) == 31) 734 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 735 {
484 ++dec->cur; 736 ++dec->cur;
485 737
486 for (;;) 738 for (;;)
487 { 739 {
488 WANT (1); 740 WANT (1);
489 741
490 if (*dec->cur == (0xe0 | 31)) 742 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 743 {
492 ++dec->cur; 744 ++dec->cur;
493 break; 745 break;
494 } 746 }
495 747
498 } 750 }
499 else 751 else
500 { 752 {
501 int i, len = decode_uint (dec); 753 int i, len = decode_uint (dec);
502 754
755 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 756 av_fill (av, len - 1);
504 757
505 for (i = 0; i < len; ++i) 758 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 759 AvARRAY (av)[i] = decode_sv (dec);
507 } 760 }
513 SvREFCNT_dec (av); 766 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 767 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 768 return &PL_sv_undef;
516} 769}
517 770
771static void
772decode_he (dec_t *dec, HV *hv)
773{
774 // for speed reasons, we specialcase single-string
775 // byte or utf-8 strings as keys, but only when !stringref
776
777 if (ecb_expect_true (!dec->stringref))
778 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
779 {
780 I32 len = decode_uint (dec);
781 char *key = (char *)dec->cur;
782
783 WANT (len);
784 dec->cur += len;
785
786 hv_store (hv, key, len, decode_sv (dec), 0);
787
788 return;
789 }
790 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
791 {
792 I32 len = decode_uint (dec);
793 char *key = (char *)dec->cur;
794
795 WANT (len);
796 dec->cur += len;
797
798 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
799 if (!is_utf8_string (key, len))
800 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
801
802 hv_store (hv, key, -len, decode_sv (dec), 0);
803
804 return;
805 }
806
807 SV *k = decode_sv (dec);
808 SV *v = decode_sv (dec);
809
810 hv_store_ent (hv, k, v, 0);
811 SvREFCNT_dec (k);
812
813fail:
814 ;
815}
816
518static SV * 817static SV *
519decode_hv (dec_t *dec) 818decode_hv (dec_t *dec)
520{ 819{
521 HV *hv = newHV (); 820 HV *hv = newHV ();
522 821
523 DEC_INC_DEPTH; 822 DEC_INC_DEPTH;
524 823
525 if ((*dec->cur & 31) == 31) 824 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 825 {
527 ++dec->cur; 826 ++dec->cur;
528 827
529 for (;;) 828 for (;;)
530 { 829 {
531 WANT (1); 830 WANT (1);
532 831
533 if (*dec->cur == (0xe0 | 31)) 832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 833 {
535 ++dec->cur; 834 ++dec->cur;
536 break; 835 break;
537 } 836 }
538 837
539 SV *k = decode_sv (dec); 838 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 839 }
544 } 840 }
545 else 841 else
546 { 842 {
547 int len = decode_uint (dec); 843 int pairs = decode_uint (dec);
548 844
549 while (len--) 845 while (pairs--)
550 { 846 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 847 }
557 848
558 DEC_DEC_DEPTH; 849 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 850 return newRV_noinc ((SV *)hv);
560 851
567static SV * 858static SV *
568decode_str (dec_t *dec, int utf8) 859decode_str (dec_t *dec, int utf8)
569{ 860{
570 SV *sv = 0; 861 SV *sv = 0;
571 862
572 if ((*dec->cur & 31) == 31) 863 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 864 {
865 // indefinite length strings
574 ++dec->cur; 866 ++dec->cur;
575 867
868 U8 major = *dec->cur & MAJOR_MISC;
869
576 sv = newSVpvn ("", 0); 870 sv = newSVpvn ("", 0);
577 871
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 872 for (;;)
580 { 873 {
581 WANT (1); 874 WANT (1);
582 875
583 if (*dec->cur == (0xe0 | 31)) 876 if ((*dec->cur - major) > LENGTH_EXT8)
877 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 878 {
585 ++dec->cur; 879 ++dec->cur;
586 break; 880 break;
587 } 881 }
882 else
883 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 884
589 sv_catsv (sv, decode_sv (dec)); 885 STRLEN len = decode_uint (dec);
886
887 WANT (len);
888 sv_catpvn (sv, dec->cur, len);
889 dec->cur += len;
590 } 890 }
591 } 891 }
592 else 892 else
593 { 893 {
594 STRLEN len = decode_uint (dec); 894 STRLEN len = decode_uint (dec);
595 895
596 WANT (len); 896 WANT (len);
597 sv = newSVpvn (dec->cur, len); 897 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 898 dec->cur += len;
899
900 if (ecb_expect_false (dec->stringref)
901 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
902 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 903 }
600 904
601 if (utf8) 905 if (utf8)
906 {
907 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
908 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
909 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
910
602 SvUTF8_on (sv); 911 SvUTF8_on (sv);
912 }
603 913
604 return sv; 914 return sv;
605 915
606fail: 916fail:
607 SvREFCNT_dec (sv); 917 SvREFCNT_dec (sv);
609} 919}
610 920
611static SV * 921static SV *
612decode_tagged (dec_t *dec) 922decode_tagged (dec_t *dec)
613{ 923{
924 SV *sv = 0;
614 UV tag = decode_uint (dec); 925 UV tag = decode_uint (dec);
926
927 WANT (1);
928
929 switch (tag)
930 {
931 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 932 sv = decode_sv (dec);
933 break;
616 934
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 935 case CBOR_TAG_INDIRECTION:
936 sv = newRV_noinc (decode_sv (dec));
937 break;
938
939 case CBOR_TAG_STRINGREF_NAMESPACE:
940 {
941 ENTER; SAVETMPS;
942
943 SAVESPTR (dec->stringref);
944 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
945
946 sv = decode_sv (dec);
947
948 FREETMPS; LEAVE;
949 }
950 break;
951
952 case CBOR_TAG_STRINGREF:
953 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
956
957 UV idx = decode_uint (dec);
958
959 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
960 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
961
962 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
963 }
964 break;
965
966 case CBOR_TAG_VALUE_SHAREABLE:
967 {
968 if (ecb_expect_false (!dec->shareable))
969 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
970
971 if (dec->cbor.flags & F_ALLOW_CYCLES)
972 {
973 sv = newSV (0);
974 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
975
976 SV *osv = decode_sv (dec);
977 sv_setsv (sv, osv);
978 SvREFCNT_dec_NN (osv);
979 }
980 else
981 {
982 av_push (dec->shareable, &PL_sv_undef);
983 int idx = AvFILLp (dec->shareable);
984 sv = decode_sv (dec);
985 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
986 }
987 }
988 break;
989
990 case CBOR_TAG_VALUE_SHAREDREF:
991 {
992 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
993 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
994
995 UV idx = decode_uint (dec);
996
997 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
998 ERR ("corrupted CBOR data (sharedref index out of bounds)");
999
1000 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1001
1002 if (sv == &PL_sv_undef)
1003 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1004 }
1005 break;
1006
1007 case CBOR_TAG_PERL_OBJECT:
1008 {
1009 sv = decode_sv (dec);
1010
1011 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1012 ERR ("corrupted CBOR data (non-array perl object)");
1013
1014 AV *av = (AV *)SvRV (sv);
1015 int len = av_len (av) + 1;
1016 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1017
1018 if (!stash)
1019 ERR ("cannot decode perl-object (package does not exist)");
1020
1021 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1022
1023 if (!method)
1024 ERR ("cannot decode perl-object (package does not have a THAW method)");
1025
1026 dSP;
1027
1028 ENTER; SAVETMPS;
1029 PUSHMARK (SP);
1030 EXTEND (SP, len + 1);
1031 // we re-bless the reference to get overload and other niceties right
1032 PUSHs (*av_fetch (av, 0, 1));
1033 PUSHs (sv_cbor);
1034
1035 int i;
1036
1037 for (i = 1; i < len; ++i)
1038 PUSHs (*av_fetch (av, i, 1));
1039
1040 PUTBACK;
1041 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1042 SPAGAIN;
1043
1044 if (SvTRUE (ERRSV))
1045 {
1046 FREETMPS; LEAVE;
1047 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1048 }
1049
1050 SvREFCNT_dec (sv);
1051 sv = SvREFCNT_inc (POPs);
1052
1053 PUTBACK;
1054
1055 FREETMPS; LEAVE;
1056 }
1057 break;
1058
1059 default:
1060 {
1061 sv = decode_sv (dec);
1062
1063 dSP;
1064 ENTER; SAVETMPS;
1065 SAVESTACK_POS ();
1066 PUSHMARK (SP);
1067 EXTEND (SP, 2);
1068 PUSHs (newSVuv (tag));
1069 PUSHs (sv);
1070
1071 PUTBACK;
1072 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1073 SPAGAIN;
1074
1075 if (SvTRUE (ERRSV))
1076 {
1077 FREETMPS; LEAVE;
1078 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1079 }
1080
1081 if (count)
1082 {
1083 SvREFCNT_dec (sv);
1084 sv = SvREFCNT_inc (POPs);
1085 }
1086 else
1087 {
1088 AV *av = newAV ();
1089 av_push (av, newSVuv (tag));
1090 av_push (av, sv);
1091
1092 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1093 ? cbor_tagged_stash
1094 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1095 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1096 }
1097
1098 PUTBACK;
1099
1100 FREETMPS; LEAVE;
1101 }
1102 break;
1103 }
1104
618 return sv; 1105 return sv;
619 1106
620 AV *av = newAV (); 1107fail:
621 av_push (av, newSVuv (tag)); 1108 SvREFCNT_dec (sv);
622 av_push (av, sv); 1109 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 1110}
625 1111
626static SV * 1112static SV *
627decode_sv (dec_t *dec) 1113decode_sv (dec_t *dec)
628{ 1114{
629 WANT (1); 1115 WANT (1);
630 1116
631 switch (*dec->cur >> 5) 1117 switch (*dec->cur >> MAJOR_SHIFT)
632 { 1118 {
633 case 0: // unsigned int 1119 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
634 return newSVuv (decode_uint (dec)); 1120 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
635 case 1: // negative int 1121 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
636 return newSViv (-1 - (IV)decode_uint (dec)); 1122 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
637 case 2: // octet string 1123 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
638 return decode_str (dec, 0); 1124 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
639 case 3: // utf-8 string 1125 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
640 return decode_str (dec, 1); 1126
641 case 4: // array 1127 case MAJOR_MISC >> MAJOR_SHIFT:
642 return decode_av (dec);
643 case 5: // map
644 return decode_hv (dec);
645 case 6: // tag
646 return decode_tagged (dec);
647 case 7: // misc
648 switch (*dec->cur++ & 31) 1128 switch (*dec->cur++ & MINOR_MASK)
649 { 1129 {
650 case 20: 1130 case SIMPLE_FALSE:
651#if CBOR_SLOW 1131#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1132 types_false = get_bool ("Types::Serialiser::false");
653#endif 1133#endif
654 return newSVsv (cbor_false); 1134 return newSVsv (types_false);
655 case 21: 1135 case SIMPLE_TRUE:
656#if CBOR_SLOW 1136#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1137 types_true = get_bool ("Types::Serialiser::true");
658#endif 1138#endif
659 return newSVsv (cbor_true); 1139 return newSVsv (types_true);
660 case 22: 1140 case SIMPLE_NULL:
661 return newSVsv (&PL_sv_undef); 1141 return newSVsv (&PL_sv_undef);
1142 case SIMPLE_UNDEF:
1143#if CBOR_SLOW
1144 types_error = get_bool ("Types::Serialiser::error");
1145#endif
1146 return newSVsv (types_error);
662 1147
663 case 25: 1148 case MISC_FLOAT16:
664 { 1149 {
665 WANT (2); 1150 WANT (2);
666 1151
667 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1152 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
668 dec->cur += 2; 1153 dec->cur += 2;
669 1154
670 return newSVnv (ecb_binary16_to_float (fp)); 1155 return newSVnv (ecb_binary16_to_float (fp));
671 } 1156 }
672 1157
673 case 26: 1158 case MISC_FLOAT32:
674 { 1159 {
675 uint32_t fp; 1160 uint32_t fp;
676 WANT (4); 1161 WANT (4);
677 memcpy (&fp, dec->cur, 4); 1162 memcpy (&fp, dec->cur, 4);
678 dec->cur += 4; 1163 dec->cur += 4;
681 fp = ecb_bswap32 (fp); 1166 fp = ecb_bswap32 (fp);
682 1167
683 return newSVnv (ecb_binary32_to_float (fp)); 1168 return newSVnv (ecb_binary32_to_float (fp));
684 } 1169 }
685 1170
686 case 27: 1171 case MISC_FLOAT64:
687 { 1172 {
688 uint64_t fp; 1173 uint64_t fp;
689 WANT (8); 1174 WANT (8);
690 memcpy (&fp, dec->cur, 8); 1175 memcpy (&fp, dec->cur, 8);
691 dec->cur += 8; 1176 dec->cur += 8;
694 fp = ecb_bswap64 (fp); 1179 fp = ecb_bswap64 (fp);
695 1180
696 return newSVnv (ecb_binary64_to_double (fp)); 1181 return newSVnv (ecb_binary64_to_double (fp));
697 } 1182 }
698 1183
699 // 0..19 unassigned 1184 // 0..19 unassigned simple
700 // 24 reserved + unassigned (reserved values are not encodable) 1185 // 24 reserved + unassigned simple (reserved values are not encodable)
1186 // 28-30 unassigned misc
1187 // 31 break code
701 default: 1188 default:
702 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1189 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
703 } 1190 }
704 1191
705 break; 1192 break;
706 } 1193 }
707 1194
710} 1197}
711 1198
712static SV * 1199static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1200decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1201{
715 dec_t dec; 1202 dec_t dec = { 0 };
716 SV *sv; 1203 SV *sv;
1204 STRLEN len;
1205 char *data = SvPVbyte (string, len);
717 1206
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 1207 if (len > cbor->max_size && cbor->max_size)
747 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1208 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1209 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1210
753 dec.cbor = *cbor; 1211 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1212 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1213 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1214
759 sv = decode_sv (&dec); 1215 sv = decode_sv (&dec);
760 1216
761 if (offset_return) 1217 if (offset_return)
762 *offset_return = dec.cur; 1218 *offset_return = dec.cur;
765 if (dec.cur != dec.end && !dec.err) 1221 if (dec.cur != dec.end && !dec.err)
766 dec.err = "garbage after CBOR object"; 1222 dec.err = "garbage after CBOR object";
767 1223
768 if (dec.err) 1224 if (dec.err)
769 { 1225 {
1226 if (dec.shareable)
1227 {
1228 // need to break cyclic links, which whould all be in shareable
1229 int i;
1230 SV **svp;
1231
1232 for (i = av_len (dec.shareable) + 1; i--; )
1233 if ((svp = av_fetch (dec.shareable, i, 0)))
1234 sv_setsv (*svp, &PL_sv_undef);
1235 }
1236
770 SvREFCNT_dec (sv); 1237 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1238 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1239 }
773 1240
774 sv = sv_2mortal (sv); 1241 sv = sv_2mortal (sv);
775 1242
776 return sv; 1243 return sv;
777} 1244}
778 1245
1246/////////////////////////////////////////////////////////////////////////////
1247// incremental parser
1248
1249#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1250
1251// returns 0 for notyet, 1 for success or error
1252static int
1253incr_parse (CBOR *self, SV *cborstr)
1254{
1255 STRLEN cur;
1256 SvPV (cborstr, cur);
1257
1258 while (ecb_expect_true (self->incr_need <= cur))
1259 {
1260 // table of integer count bytes
1261 static I8 incr_len[MINOR_MASK + 1] = {
1262 0, 0, 0, 0, 0, 0, 0, 0,
1263 0, 0, 0, 0, 0, 0, 0, 0,
1264 0, 0, 0, 0, 0, 0, 0, 0,
1265 1, 2, 4, 8,-1,-1,-1,-2
1266 };
1267
1268 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1269 U8 m = *p & MINOR_MASK;
1270 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1271 I8 ilen = incr_len[m];
1272
1273 self->incr_need = self->incr_pos + 1;
1274
1275 if (ecb_expect_false (ilen < 0))
1276 {
1277 if (m != MINOR_INDEF)
1278 return 1; // error
1279
1280 if (*p == (MAJOR_MISC | MINOR_INDEF))
1281 {
1282 if (count >= 0)
1283 return 1; // error
1284
1285 count = 1;
1286 }
1287 else
1288 {
1289 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1290 count = -1;
1291 }
1292 }
1293 else
1294 {
1295 self->incr_need += ilen;
1296 if (ecb_expect_false (self->incr_need > cur))
1297 return 0;
1298
1299 int major = *p >> MAJOR_SHIFT;
1300
1301 switch (major)
1302 {
1303 case MAJOR_TAG >> MAJOR_SHIFT:
1304 ++count; // tags merely prefix another value
1305 break;
1306
1307 case MAJOR_BYTES >> MAJOR_SHIFT:
1308 case MAJOR_TEXT >> MAJOR_SHIFT:
1309 case MAJOR_ARRAY >> MAJOR_SHIFT:
1310 case MAJOR_MAP >> MAJOR_SHIFT:
1311 {
1312 UV len;
1313
1314 if (ecb_expect_false (ilen))
1315 {
1316 len = 0;
1317
1318 do {
1319 len = (len << 8) | *++p;
1320 } while (--ilen);
1321 }
1322 else
1323 len = m;
1324
1325 switch (major)
1326 {
1327 case MAJOR_BYTES >> MAJOR_SHIFT:
1328 case MAJOR_TEXT >> MAJOR_SHIFT:
1329 self->incr_need += len;
1330 if (ecb_expect_false (self->incr_need > cur))
1331 return 0;
1332
1333 break;
1334
1335 case MAJOR_MAP >> MAJOR_SHIFT:
1336 len <<= 1;
1337 case MAJOR_ARRAY >> MAJOR_SHIFT:
1338 if (len)
1339 {
1340 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1341 count = len + 1;
1342 }
1343 break;
1344 }
1345 }
1346 }
1347 }
1348
1349 self->incr_pos = self->incr_need;
1350
1351 if (count > 0)
1352 {
1353 while (!--count)
1354 {
1355 if (!AvFILLp (self->incr_count))
1356 return 1; // done
1357
1358 SvREFCNT_dec_NN (av_pop (self->incr_count));
1359 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1360 }
1361
1362 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1363 }
1364 }
1365
1366 return 0;
1367}
1368
1369
779///////////////////////////////////////////////////////////////////////////// 1370/////////////////////////////////////////////////////////////////////////////
780// XS interface functions 1371// XS interface functions
781 1372
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1373MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1374
784BOOT: 1375BOOT:
785{ 1376{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1377 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1378 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1379
790 cbor_true = get_bool ("CBOR::XS::true"); 1380 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1381 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1382
1383 types_true = get_bool ("Types::Serialiser::true" );
1384 types_false = get_bool ("Types::Serialiser::false");
1385 types_error = get_bool ("Types::Serialiser::error");
1386
1387 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1388
1389 sv_cbor = newSVpv ("CBOR", 0);
1390 SvREADONLY_on (sv_cbor);
792} 1391}
793 1392
794PROTOTYPES: DISABLE 1393PROTOTYPES: DISABLE
795 1394
796void CLONE (...) 1395void CLONE (...)
797 CODE: 1396 CODE:
798 cbor_stash = 0; 1397 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1398 cbor_tagged_stash = 0;
1399 types_error_stash = 0;
1400 types_boolean_stash = 0;
801 1401
802void new (char *klass) 1402void new (char *klass)
803 PPCODE: 1403 PPCODE:
804{ 1404{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1405 SV *pv = NEWSV (0, sizeof (CBOR));
813 1413
814void shrink (CBOR *self, int enable = 1) 1414void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1415 ALIAS:
816 shrink = F_SHRINK 1416 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1417 allow_unknown = F_ALLOW_UNKNOWN
1418 allow_sharing = F_ALLOW_SHARING
1419 allow_cycles = F_ALLOW_CYCLES
1420 pack_strings = F_PACK_STRINGS
1421 utf8_strings = F_UTF8_STRINGS
1422 validate_utf8 = F_VALIDATE_UTF8
818 PPCODE: 1423 PPCODE:
819{ 1424{
820 if (enable) 1425 if (enable)
821 self->flags |= ix; 1426 self->flags |= ix;
822 else 1427 else
827 1432
828void get_shrink (CBOR *self) 1433void get_shrink (CBOR *self)
829 ALIAS: 1434 ALIAS:
830 get_shrink = F_SHRINK 1435 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1436 get_allow_unknown = F_ALLOW_UNKNOWN
1437 get_allow_sharing = F_ALLOW_SHARING
1438 get_allow_cycles = F_ALLOW_CYCLES
1439 get_pack_strings = F_PACK_STRINGS
1440 get_validate_utf8 = F_VALIDATE_UTF8
832 PPCODE: 1441 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1442 XPUSHs (boolSV (self->flags & ix));
834 1443
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1444void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1445 PPCODE:
852 CODE: 1461 CODE:
853 RETVAL = self->max_size; 1462 RETVAL = self->max_size;
854 OUTPUT: 1463 OUTPUT:
855 RETVAL 1464 RETVAL
856 1465
857#if 0 //TODO 1466void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1467 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1468 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1469 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1470 XPUSHs (ST (0));
866}
867 1471
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1472SV *get_filter (CBOR *self)
869 PPCODE: 1473 CODE:
870{ 1474 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1475 OUTPUT:
872 self->cb_sk_object = newHV (); 1476 RETVAL
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891 1477
892void encode (CBOR *self, SV *scalar) 1478void encode (CBOR *self, SV *scalar)
893 PPCODE: 1479 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1480 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1481 XPUSHs (scalar);
908 EXTEND (SP, 2); 1494 EXTEND (SP, 2);
909 PUSHs (sv); 1495 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1496 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1497}
912 1498
913#if 0 1499void incr_parse (CBOR *self, SV *cborstr)
1500 ALIAS:
1501 incr_parse_multiple = 1
1502 PPCODE:
1503{
1504 if (SvUTF8 (cborstr))
1505 sv_utf8_downgrade (cborstr, 0);
1506
1507 if (!self->incr_count)
1508 {
1509 self->incr_count = newAV ();
1510 self->incr_pos = 0;
1511 self->incr_need = 1;
1512
1513 av_push (self->incr_count, newSViv (1));
1514 }
1515
1516 do
1517 {
1518 if (!incr_parse (self, cborstr))
1519 {
1520 if (self->incr_need > self->max_size && self->max_size)
1521 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1522 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1523
1524 break;
1525 }
1526
1527 SV *sv;
1528 char *offset;
1529
1530 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1531 XPUSHs (sv);
1532
1533 sv_chop (cborstr, offset);
1534
1535 av_clear (self->incr_count);
1536 av_push (self->incr_count, newSViv (1));
1537
1538 self->incr_pos = 0;
1539 self->incr_need = self->incr_pos + 1;
1540 }
1541 while (ix);
1542}
1543
1544void incr_reset (CBOR *self)
1545 CODE:
1546{
1547 SvREFCNT_dec (self->incr_count);
1548 self->incr_count = 0;
1549}
914 1550
915void DESTROY (CBOR *self) 1551void DESTROY (CBOR *self)
916 CODE: 1552 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1553 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1554
922PROTOTYPES: ENABLE 1555PROTOTYPES: ENABLE
923 1556
924void encode_cbor (SV *scalar) 1557void encode_cbor (SV *scalar)
1558 ALIAS:
1559 encode_cbor = 0
1560 encode_cbor_sharing = F_ALLOW_SHARING
925 PPCODE: 1561 PPCODE:
926{ 1562{
927 CBOR cbor; 1563 CBOR cbor;
928 cbor_init (&cbor); 1564 cbor_init (&cbor);
1565 cbor.flags |= ix;
929 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1566 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
930 XPUSHs (scalar); 1567 XPUSHs (scalar);
931} 1568}
932 1569
933void decode_cbor (SV *cborstr) 1570void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines