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 1.7 by root, Sat Oct 26 23:02:55 2013 UTC vs.
Revision 1.47 by root, Mon Apr 27 20:21:53 2015 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_VALIDATE_UTF8 0x00000020UL
16 106
17#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
18 108
19#define SB do { 109#define SB do {
20#define SE } while (0) 110#define SE } while (0)
31#else 121#else
32# define CBOR_SLOW 0 122# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
34#endif 124#endif
35 125
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 128
39typedef struct { 129typedef struct {
40 U32 flags; 130 U32 flags;
41 U32 max_depth; 131 U32 max_depth;
42 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 139} CBOR;
44 140
45ecb_inline void 141ecb_inline void
46cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
47{ 143{
48 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
50} 153}
51 154
52///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
53// utility functions 156// utility functions
54 157
76 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 180#endif
78 } 181 }
79} 182}
80 183
81///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
82// fp hell 185ecb_inline int
83 186minimum_string_length (UV idx)
84//TODO 187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
85 198
86///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
87// encoder 200// encoder
88 201
89// structure used for encoding CBOR 202// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
94 SV *sv; // result scalar 207 SV *sv; // result scalar
95 CBOR cbor; 208 CBOR cbor;
96 U32 depth; // recursion level 209 U32 depth; // recursion level
210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
212 HV *shareable; // ptr => index, or 0
213 UV shareable_idx;
97} enc_t; 214} enc_t;
98 215
99ecb_inline void 216ecb_inline void
100need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
101{ 218{
118static void 235static void
119encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
120{ 237{
121 need (enc, 9); 238 need (enc, 9);
122 239
123 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
125 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
126 { 243 {
127 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 245 *enc->cur++ = len;
129 } 246 }
130 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
131 { 248 {
132 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 251 *enc->cur++ = len;
135 } 252 }
136 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
137 { 254 {
138 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 259 *enc->cur++ = len;
143 } 260 }
144 else 261 else
145 { 262 {
146 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 271 *enc->cur++ = len;
155 } 272 }
156} 273}
157 274
158static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 283{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 285 need (enc, len);
163 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
164 enc->cur += len; 287 enc->cur += len;
165} 288}
166 289
290static void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
292{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
313}
314
167static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
168 316
169static void 317static void
170encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
171{ 319{
174 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
176 324
177 ++enc->depth; 325 ++enc->depth;
178 326
179 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 328
329 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
182 { 331 {
183 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
186 341
187 --enc->depth; 342 --enc->depth;
188} 343}
189 344
190static void 345static void
199 354
200 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
202 357
203 if (mg) 358 if (mg)
204 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 360 else
206 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
207 362
208 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
209 { 364 {
210 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
212 else 367 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 369
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 371 }
217 372
218 if (mg) 373 if (mg)
219 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 375
221 --enc->depth; 376 --enc->depth;
222} 377}
223 378
224// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
225static void 380static void
226encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
227{ 382{
228 svtype svt;
229
230 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
231 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
232 386
233 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
234 { 388 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 390 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 391 : gv_stashpv ("Types::Serialiser::Boolean", 1);
392 HV *error_stash = !CBOR_SLOW || types_error_stash
393 ? types_error_stash
394 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 396 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 398
399 HV *stash = SvSTASH (sv);
400
242 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
406 else if (stash == error_stash)
407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
244 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
245 { 412 {
246 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 415
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
251 } 436 }
252 else 437 else
253 { 438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 {
452 dSP;
453
454 ENTER; SAVETMPS; PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 455 // 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))); 456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 457
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 458 PUTBACK;
459 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 460 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 461 SPAGAIN;
268 462
269 // catch this surprisingly common error 463 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 464 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))); 465 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 466
273 sv = POPs; 467 encode_sv (enc, POPs);
468
274 PUTBACK; 469 PUTBACK;
275 470
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 471 FREETMPS; LEAVE;
279 } 472 }
473 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
474 {
475 dSP;
476
477 ENTER; SAVETMPS; PUSHMARK (SP);
478 EXTEND (SP, 2);
479 // we re-bless the reference to get overload and other niceties right
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor);
482
483 PUTBACK;
484 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
485 SPAGAIN;
486
487 // catch this surprisingly common error
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
490
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494
495 while (count)
496 encode_sv (enc, SP[1 - count--]);
497
498 PUTBACK;
499
500 FREETMPS; LEAVE;
501 }
280 else 502 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 503 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 505 }
285 else if (svt == SVt_PVHV) 506 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 507 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 508 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 509 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 510 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 511 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 512 encode_tag (enc, CBOR_TAG_INDIRECTION);
513 encode_sv (enc, sv);
514 }
309} 515}
310 516
311static void 517static void
312encode_nv (enc_t *enc, SV *sv) 518encode_nv (enc_t *enc, SV *sv)
313{ 519{
314 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
315 521
316 need (enc, 9); 522 need (enc, 9);
317 523
318 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 526 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
322 { 528 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
324 530
325 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 532
327 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
329 535
330 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
332 } 538 }
333 else 539 else
334 { 540 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
336 542
337 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 544
339 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
341 547
342 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
351 557
352 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
353 { 559 {
354 STRLEN len; 560 STRLEN len;
355 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 563 }
358 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 565 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
361 { 567 {
362 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 572 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 574 }
369 else if (SvROK (sv)) 575 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 581 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 584}
379 585
380static SV * 586static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
382{ 588{
383 enc_t enc; 589 enc_t enc = { };
384 590
385 enc.cbor = *cbor; 591 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 595
391 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
597
598 if (cbor->flags & F_PACK_STRINGS)
599 {
600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
603 }
604
392 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
393 606
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 608 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 609
410 U8 *end; // end of input string 623 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
412 CBOR cbor; 625 CBOR cbor;
413 U32 depth; // recursion depth 626 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
415} dec_t; 631} dec_t;
416 632
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 634
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 635#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 638#define DEC_DEC_DEPTH --dec->depth
423 639
424static UV 640static UV
425decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
426{ 642{
427 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
428 { 644 ++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;
433 645
434 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
435 WANT (2); 650 WANT (1);
436 dec->cur += 2; 651 dec->cur += 1;
437 return dec->cur[-1]; 652 return dec->cur[-1];
438 653 }
439 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
440 WANT (3); 656 WANT (2);
441 dec->cur += 3; 657 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
444 660 }
445 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
446 WANT (5); 663 WANT (4);
447 dec->cur += 5; 664 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
452 669 }
453 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
454 WANT (9); 672 WANT (8);
455 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
456 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
460 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
464 688 }
465 default: 689 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 691
469fail: 692fail:
470 return 0; 693 return 0;
471} 694}
472 695
477{ 700{
478 AV *av = newAV (); 701 AV *av = newAV ();
479 702
480 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
481 704
482 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 706 {
484 ++dec->cur; 707 ++dec->cur;
485 708
486 for (;;) 709 for (;;)
487 { 710 {
488 WANT (1); 711 WANT (1);
489 712
490 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 714 {
492 ++dec->cur; 715 ++dec->cur;
493 break; 716 break;
494 } 717 }
495 718
498 } 721 }
499 else 722 else
500 { 723 {
501 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
502 725
726 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); 727 av_fill (av, len - 1);
504 728
505 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
507 } 731 }
513 SvREFCNT_dec (av); 737 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 738 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 739 return &PL_sv_undef;
516} 740}
517 741
742static void
743decode_he (dec_t *dec, HV *hv)
744{
745 // for speed reasons, we specialcase single-string
746 // byte or utf-8 strings as keys, but only when !stringref
747
748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 {
751 I32 len = decode_uint (dec);
752 char *key = (char *)dec->cur;
753
754 dec->cur += len;
755
756 hv_store (hv, key, len, decode_sv (dec), 0);
757
758 return;
759 }
760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
761 {
762 I32 len = decode_uint (dec);
763 char *key = (char *)dec->cur;
764
765 dec->cur += len;
766
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
770
771 hv_store (hv, key, -len, decode_sv (dec), 0);
772
773 return;
774 }
775
776 SV *k = decode_sv (dec);
777 SV *v = decode_sv (dec);
778
779 hv_store_ent (hv, k, v, 0);
780 SvREFCNT_dec (k);
781
782fail:
783 ;
784}
785
518static SV * 786static SV *
519decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
520{ 788{
521 HV *hv = newHV (); 789 HV *hv = newHV ();
522 790
523 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
524 792
525 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 794 {
527 ++dec->cur; 795 ++dec->cur;
528 796
529 for (;;) 797 for (;;)
530 { 798 {
531 WANT (1); 799 WANT (1);
532 800
533 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 802 {
535 ++dec->cur; 803 ++dec->cur;
536 break; 804 break;
537 } 805 }
538 806
539 SV *k = decode_sv (dec); 807 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 808 }
544 } 809 }
545 else 810 else
546 { 811 {
547 int len = decode_uint (dec); 812 int pairs = decode_uint (dec);
548 813
549 while (len--) 814 while (pairs--)
550 { 815 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 } 816 }
557 817
558 DEC_DEC_DEPTH; 818 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 819 return newRV_noinc ((SV *)hv);
560 820
567static SV * 827static SV *
568decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
569{ 829{
570 SV *sv = 0; 830 SV *sv = 0;
571 831
572 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 833 {
834 // indefinite length strings
574 ++dec->cur; 835 ++dec->cur;
575 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
576 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
577 840
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 841 for (;;)
580 { 842 {
581 WANT (1); 843 WANT (1);
582 844
583 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 847 {
585 ++dec->cur; 848 ++dec->cur;
586 break; 849 break;
587 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 853
589 sv_catsv (sv, decode_sv (dec)); 854 STRLEN len = decode_uint (dec);
855
856 WANT (len);
857 sv_catpvn (sv, dec->cur, len);
858 dec->cur += len;
590 } 859 }
591 } 860 }
592 else 861 else
593 { 862 {
594 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
595 864
596 WANT (len); 865 WANT (len);
597 sv = newSVpvn (dec->cur, len); 866 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 867 dec->cur += len;
868
869 if (ecb_expect_false (dec->stringref)
870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 872 }
600 873
601 if (utf8) 874 if (utf8)
875 {
876 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
877 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
878 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
879
602 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
603 882
604 return sv; 883 return sv;
605 884
606fail: 885fail:
607 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
609} 888}
610 889
611static SV * 890static SV *
612decode_tagged (dec_t *dec) 891decode_tagged (dec_t *dec)
613{ 892{
893 SV *sv = 0;
614 UV tag = decode_uint (dec); 894 UV tag = decode_uint (dec);
895
896 WANT (1);
897
898 switch (tag)
899 {
900 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 901 sv = decode_sv (dec);
902 break;
616 903
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 904 case CBOR_TAG_INDIRECTION:
905 sv = newRV_noinc (decode_sv (dec));
906 break;
907
908 case CBOR_TAG_STRINGREF_NAMESPACE:
909 {
910 ENTER; SAVETMPS;
911
912 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914
915 sv = decode_sv (dec);
916
917 FREETMPS; LEAVE;
918 }
919 break;
920
921 case CBOR_TAG_STRINGREF:
922 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925
926 UV idx = decode_uint (dec);
927
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
929 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
930
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 }
933 break;
934
935 case CBOR_TAG_VALUE_SHAREABLE:
936 {
937 if (ecb_expect_false (!dec->shareable))
938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
942 sv = newSV (0);
943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
944
945 SV *osv = decode_sv (dec);
946 sv_setsv (sv, osv);
947 SvREFCNT_dec_NN (osv);
948 }
949 else
950 {
951 av_push (dec->shareable, &PL_sv_undef);
952 int idx = AvFILLp (dec->shareable);
953 sv = decode_sv (dec);
954 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
955 }
956 }
957 break;
958
959 case CBOR_TAG_VALUE_SHAREDREF:
960 {
961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
963
964 UV idx = decode_uint (dec);
965
966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
968
969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970
971 if (sv == &PL_sv_undef)
972 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
973 }
974 break;
975
976 case CBOR_TAG_PERL_OBJECT:
977 {
978 sv = decode_sv (dec);
979
980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
981 ERR ("corrupted CBOR data (non-array perl object)");
982
983 AV *av = (AV *)SvRV (sv);
984 int len = av_len (av) + 1;
985 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
986
987 if (!stash)
988 ERR ("cannot decode perl-object (package does not exist)");
989
990 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
991
992 if (!method)
993 ERR ("cannot decode perl-object (package does not have a THAW method)");
994
995 dSP;
996
997 ENTER; SAVETMPS; PUSHMARK (SP);
998 EXTEND (SP, len + 1);
999 // we re-bless the reference to get overload and other niceties right
1000 PUSHs (*av_fetch (av, 0, 1));
1001 PUSHs (sv_cbor);
1002
1003 int i;
1004
1005 for (i = 1; i < len; ++i)
1006 PUSHs (*av_fetch (av, i, 1));
1007
1008 PUTBACK;
1009 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1010 SPAGAIN;
1011
1012 if (SvTRUE (ERRSV))
1013 {
1014 FREETMPS; LEAVE;
1015 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1016 }
1017
1018 SvREFCNT_dec (sv);
1019 sv = SvREFCNT_inc (POPs);
1020
1021 PUTBACK;
1022
1023 FREETMPS; LEAVE;
1024 }
1025 break;
1026
1027 default:
1028 {
1029 sv = decode_sv (dec);
1030
1031 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP);
1033 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag));
1035 PUSHs (sv);
1036
1037 PUTBACK;
1038 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1039 SPAGAIN;
1040
1041 if (SvTRUE (ERRSV))
1042 {
1043 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1045 }
1046
1047 if (count)
1048 {
1049 SvREFCNT_dec (sv);
1050 sv = SvREFCNT_inc (POPs);
1051 }
1052 else
1053 {
1054 AV *av = newAV ();
1055 av_push (av, newSVuv (tag));
1056 av_push (av, sv);
1057
1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1059 ? cbor_tagged_stash
1060 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1061 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1062 }
1063
1064 PUTBACK;
1065
1066 FREETMPS; LEAVE;
1067 }
1068 break;
1069 }
1070
618 return sv; 1071 return sv;
619 1072
620 AV *av = newAV (); 1073fail:
621 av_push (av, newSVuv (tag)); 1074 SvREFCNT_dec (sv);
622 av_push (av, sv); 1075 return &PL_sv_undef;
623
624 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
625 ? cbor_tagged_stash
626 : gv_stashpv ("CBOR::XS::Tagged" , 1);
627
628 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
629} 1076}
630 1077
631static SV * 1078static SV *
632decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
633{ 1080{
634 WANT (1); 1081 WANT (1);
635 1082
636 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
637 { 1084 {
638 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
639 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
640 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
641 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
642 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
643 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
644 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
645 return decode_str (dec, 1); 1092
646 case 4: // array 1093 case MAJOR_MISC >> MAJOR_SHIFT:
647 return decode_av (dec);
648 case 5: // map
649 return decode_hv (dec);
650 case 6: // tag
651 return decode_tagged (dec);
652 case 7: // misc
653 switch (*dec->cur++ & 31) 1094 switch (*dec->cur++ & MINOR_MASK)
654 { 1095 {
655 case 20: 1096 case SIMPLE_FALSE:
656#if CBOR_SLOW 1097#if CBOR_SLOW
657 cbor_false = get_bool ("CBOR::XS::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
658#endif 1099#endif
659 return newSVsv (cbor_false); 1100 return newSVsv (types_false);
660 case 21: 1101 case SIMPLE_TRUE:
661#if CBOR_SLOW 1102#if CBOR_SLOW
662 cbor_true = get_bool ("CBOR::XS::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
663#endif 1104#endif
664 return newSVsv (cbor_true); 1105 return newSVsv (types_true);
665 case 22: 1106 case SIMPLE_NULL:
666 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
1108 case SIMPLE_UNDEF:
1109#if CBOR_SLOW
1110 types_error = get_bool ("Types::Serialiser::error");
1111#endif
1112 return newSVsv (types_error);
667 1113
668 case 25: 1114 case MISC_FLOAT16:
669 { 1115 {
670 WANT (2); 1116 WANT (2);
671 1117
672 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
673 dec->cur += 2; 1119 dec->cur += 2;
674 1120
675 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
676 } 1122 }
677 1123
678 case 26: 1124 case MISC_FLOAT32:
679 { 1125 {
680 uint32_t fp; 1126 uint32_t fp;
681 WANT (4); 1127 WANT (4);
682 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
683 dec->cur += 4; 1129 dec->cur += 4;
686 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
687 1133
688 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
689 } 1135 }
690 1136
691 case 27: 1137 case MISC_FLOAT64:
692 { 1138 {
693 uint64_t fp; 1139 uint64_t fp;
694 WANT (8); 1140 WANT (8);
695 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
696 dec->cur += 8; 1142 dec->cur += 8;
699 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
700 1146
701 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
702 } 1148 }
703 1149
704 // 0..19 unassigned 1150 // 0..19 unassigned simple
705 // 24 reserved + unassigned (reserved values are not encodable) 1151 // 24 reserved + unassigned simple (reserved values are not encodable)
1152 // 28-30 unassigned misc
1153 // 31 break code
706 default: 1154 default:
707 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
708 } 1156 }
709 1157
710 break; 1158 break;
711 } 1159 }
712 1160
715} 1163}
716 1164
717static SV * 1165static SV *
718decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1166decode_cbor (SV *string, CBOR *cbor, char **offset_return)
719{ 1167{
720 dec_t dec; 1168 dec_t dec = { };
721 SV *sv; 1169 SV *sv;
1170 STRLEN len;
1171 char *data = SvPVbyte (string, len);
722 1172
723 /* work around bugs in 5.10 where manipulating magic values
724 * makes perl ignore the magic in subsequent accesses.
725 * also make a copy of non-PV values, to get them into a clean
726 * state (SvPV should do that, but it's buggy, see below).
727 */
728 /*SvGETMAGIC (string);*/
729 if (SvMAGICAL (string) || !SvPOK (string))
730 string = sv_2mortal (newSVsv (string));
731
732 SvUPGRADE (string, SVt_PV);
733
734 /* work around a bug in perl 5.10, which causes SvCUR to fail an
735 * assertion with -DDEBUGGING, although SvCUR is documented to
736 * return the xpv_cur field which certainly exists after upgrading.
737 * according to nicholas clark, calling SvPOK fixes this.
738 * But it doesn't fix it, so try another workaround, call SvPV_nolen
739 * and hope for the best.
740 * Damnit, SvPV_nolen still trips over yet another assertion. This
741 * assertion business is seriously broken, try yet another workaround
742 * for the broken -DDEBUGGING.
743 */
744 {
745#ifdef DEBUGGING
746 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
747#else
748 STRLEN offset = SvCUR (string);
749#endif
750
751 if (offset > cbor->max_size && cbor->max_size) 1173 if (len > cbor->max_size && cbor->max_size)
752 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1174 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
753 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1175 (unsigned long)len, (unsigned long)cbor->max_size);
754 }
755
756 sv_utf8_downgrade (string, 0);
757 1176
758 dec.cbor = *cbor; 1177 dec.cbor = *cbor;
759 dec.cur = (U8 *)SvPVX (string); 1178 dec.cur = (U8 *)data;
760 dec.end = (U8 *)SvEND (string); 1179 dec.end = (U8 *)data + len;
761 dec.err = 0;
762 dec.depth = 0;
763 1180
764 sv = decode_sv (&dec); 1181 sv = decode_sv (&dec);
765 1182
766 if (offset_return) 1183 if (offset_return)
767 *offset_return = dec.cur; 1184 *offset_return = dec.cur;
770 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
771 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
772 1189
773 if (dec.err) 1190 if (dec.err)
774 { 1191 {
1192 if (dec.shareable)
1193 {
1194 // need to break cyclic links, which whould all be in shareable
1195 int i;
1196 SV **svp;
1197
1198 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef);
1201 }
1202
775 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
776 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
777 } 1205 }
778 1206
779 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
780 1208
781 return sv; 1209 return sv;
782} 1210}
783 1211
1212/////////////////////////////////////////////////////////////////////////////
1213// incremental parser
1214
1215#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1216
1217// returns 0 for notyet, 1 for success or error
1218static int
1219incr_parse (CBOR *self, SV *cborstr)
1220{
1221 STRLEN cur;
1222 SvPV (cborstr, cur);
1223
1224 while (ecb_expect_true (self->incr_need <= cur))
1225 {
1226 // table of integer count bytes
1227 static I8 incr_len[MINOR_MASK + 1] = {
1228 0, 0, 0, 0, 0, 0, 0, 0,
1229 0, 0, 0, 0, 0, 0, 0, 0,
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 1, 2, 4, 8,-1,-1,-1,-2
1232 };
1233
1234 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1235 U8 m = *p & MINOR_MASK;
1236 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1237 I8 ilen = incr_len[m];
1238
1239 self->incr_need = self->incr_pos + 1;
1240
1241 if (ecb_expect_false (ilen < 0))
1242 {
1243 if (m != MINOR_INDEF)
1244 return 1; // error
1245
1246 if (*p == (MAJOR_MISC | MINOR_INDEF))
1247 {
1248 if (count >= 0)
1249 return 1; // error
1250
1251 count = 1;
1252 }
1253 else
1254 {
1255 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1256 count = -1;
1257 }
1258 }
1259 else
1260 {
1261 self->incr_need += ilen;
1262 if (ecb_expect_false (self->incr_need > cur))
1263 return 0;
1264
1265 int major = *p >> MAJOR_SHIFT;
1266
1267 switch (major)
1268 {
1269 case MAJOR_TAG >> MAJOR_SHIFT:
1270 ++count; // tags merely prefix another value
1271 break;
1272
1273 case MAJOR_BYTES >> MAJOR_SHIFT:
1274 case MAJOR_TEXT >> MAJOR_SHIFT:
1275 case MAJOR_ARRAY >> MAJOR_SHIFT:
1276 case MAJOR_MAP >> MAJOR_SHIFT:
1277 {
1278 UV len;
1279
1280 if (ecb_expect_false (ilen))
1281 {
1282 len = 0;
1283
1284 do {
1285 len = (len << 8) | *++p;
1286 } while (--ilen);
1287 }
1288 else
1289 len = m;
1290
1291 switch (major)
1292 {
1293 case MAJOR_BYTES >> MAJOR_SHIFT:
1294 case MAJOR_TEXT >> MAJOR_SHIFT:
1295 self->incr_need += len;
1296 if (ecb_expect_false (self->incr_need > cur))
1297 return 0;
1298
1299 break;
1300
1301 case MAJOR_MAP >> MAJOR_SHIFT:
1302 len <<= 1;
1303 case MAJOR_ARRAY >> MAJOR_SHIFT:
1304 if (len)
1305 {
1306 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1307 count = len + 1;
1308 }
1309 break;
1310 }
1311 }
1312 }
1313 }
1314
1315 self->incr_pos = self->incr_need;
1316
1317 if (count > 0)
1318 {
1319 while (!--count)
1320 {
1321 if (!AvFILLp (self->incr_count))
1322 return 1; // done
1323
1324 SvREFCNT_dec_NN (av_pop (self->incr_count));
1325 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1326 }
1327
1328 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1329 }
1330 }
1331
1332 return 0;
1333}
1334
1335
784///////////////////////////////////////////////////////////////////////////// 1336/////////////////////////////////////////////////////////////////////////////
785// XS interface functions 1337// XS interface functions
786 1338
787MODULE = CBOR::XS PACKAGE = CBOR::XS 1339MODULE = CBOR::XS PACKAGE = CBOR::XS
788 1340
789BOOT: 1341BOOT:
790{ 1342{
791 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1343 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
792 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
793 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1344 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
794 1345
795 cbor_true = get_bool ("CBOR::XS::true"); 1346 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
796 cbor_false = get_bool ("CBOR::XS::false"); 1347 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1348
1349 types_true = get_bool ("Types::Serialiser::true" );
1350 types_false = get_bool ("Types::Serialiser::false");
1351 types_error = get_bool ("Types::Serialiser::error");
1352
1353 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1354
1355 sv_cbor = newSVpv ("CBOR", 0);
1356 SvREADONLY_on (sv_cbor);
797} 1357}
798 1358
799PROTOTYPES: DISABLE 1359PROTOTYPES: DISABLE
800 1360
801void CLONE (...) 1361void CLONE (...)
802 CODE: 1362 CODE:
803 cbor_stash = 0; 1363 cbor_stash = 0;
804 cbor_boolean_stash = 0;
805 cbor_tagged_stash = 0; 1364 cbor_tagged_stash = 0;
1365 types_error_stash = 0;
1366 types_boolean_stash = 0;
806 1367
807void new (char *klass) 1368void new (char *klass)
808 PPCODE: 1369 PPCODE:
809{ 1370{
810 SV *pv = NEWSV (0, sizeof (CBOR)); 1371 SV *pv = NEWSV (0, sizeof (CBOR));
818 1379
819void shrink (CBOR *self, int enable = 1) 1380void shrink (CBOR *self, int enable = 1)
820 ALIAS: 1381 ALIAS:
821 shrink = F_SHRINK 1382 shrink = F_SHRINK
822 allow_unknown = F_ALLOW_UNKNOWN 1383 allow_unknown = F_ALLOW_UNKNOWN
1384 allow_sharing = F_ALLOW_SHARING
1385 allow_cycles = F_ALLOW_CYCLES
1386 pack_strings = F_PACK_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8
823 PPCODE: 1388 PPCODE:
824{ 1389{
825 if (enable) 1390 if (enable)
826 self->flags |= ix; 1391 self->flags |= ix;
827 else 1392 else
832 1397
833void get_shrink (CBOR *self) 1398void get_shrink (CBOR *self)
834 ALIAS: 1399 ALIAS:
835 get_shrink = F_SHRINK 1400 get_shrink = F_SHRINK
836 get_allow_unknown = F_ALLOW_UNKNOWN 1401 get_allow_unknown = F_ALLOW_UNKNOWN
1402 get_allow_sharing = F_ALLOW_SHARING
1403 get_allow_cycles = F_ALLOW_CYCLES
1404 get_pack_strings = F_PACK_STRINGS
1405 get_validate_utf8 = F_VALIDATE_UTF8
837 PPCODE: 1406 PPCODE:
838 XPUSHs (boolSV (self->flags & ix)); 1407 XPUSHs (boolSV (self->flags & ix));
839 1408
840void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
841 PPCODE: 1410 PPCODE:
857 CODE: 1426 CODE:
858 RETVAL = self->max_size; 1427 RETVAL = self->max_size;
859 OUTPUT: 1428 OUTPUT:
860 RETVAL 1429 RETVAL
861 1430
862#if 0 //TODO 1431void filter (CBOR *self, SV *filter = 0)
863
864void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
865 PPCODE: 1432 PPCODE:
866{
867 SvREFCNT_dec (self->cb_object); 1433 SvREFCNT_dec (self->filter);
868 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1434 self->filter = filter ? newSVsv (filter) : filter;
869
870 XPUSHs (ST (0)); 1435 XPUSHs (ST (0));
871}
872 1436
873void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1437SV *get_filter (CBOR *self)
874 PPCODE: 1438 CODE:
875{ 1439 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
876 if (!self->cb_sk_object) 1440 OUTPUT:
877 self->cb_sk_object = newHV (); 1441 RETVAL
878
879 if (SvOK (cb))
880 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
881 else
882 {
883 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
884
885 if (!HvKEYS (self->cb_sk_object))
886 {
887 SvREFCNT_dec (self->cb_sk_object);
888 self->cb_sk_object = 0;
889 }
890 }
891
892 XPUSHs (ST (0));
893}
894
895#endif
896 1442
897void encode (CBOR *self, SV *scalar) 1443void encode (CBOR *self, SV *scalar)
898 PPCODE: 1444 PPCODE:
899 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1445 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
900 XPUSHs (scalar); 1446 XPUSHs (scalar);
913 EXTEND (SP, 2); 1459 EXTEND (SP, 2);
914 PUSHs (sv); 1460 PUSHs (sv);
915 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1461 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
916} 1462}
917 1463
918#if 0 1464void incr_parse (CBOR *self, SV *cborstr)
1465 ALIAS:
1466 incr_parse_multiple = 1
1467 PPCODE:
1468{
1469 if (SvUTF8 (cborstr))
1470 sv_utf8_downgrade (cborstr, 0);
1471
1472 if (!self->incr_count)
1473 {
1474 self->incr_count = newAV ();
1475 self->incr_pos = 0;
1476 self->incr_need = 1;
1477
1478 av_push (self->incr_count, newSViv (1));
1479 }
1480
1481 do
1482 {
1483 if (!incr_parse (self, cborstr))
1484 {
1485 if (self->incr_need > self->max_size && self->max_size)
1486 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1487 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1488
1489 break;
1490 }
1491
1492 SV *sv;
1493 char *offset;
1494
1495 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1496 XPUSHs (sv);
1497
1498 sv_chop (cborstr, offset);
1499
1500 av_clear (self->incr_count);
1501 av_push (self->incr_count, newSViv (1));
1502
1503 self->incr_pos = 0;
1504 self->incr_need = self->incr_pos + 1;
1505 }
1506 while (ix);
1507}
1508
1509void incr_reset (CBOR *self)
1510 CODE:
1511{
1512 SvREFCNT_dec (self->incr_count);
1513 self->incr_count = 0;
1514}
919 1515
920void DESTROY (CBOR *self) 1516void DESTROY (CBOR *self)
921 CODE: 1517 PPCODE:
922 SvREFCNT_dec (self->cb_sk_object); 1518 cbor_free (self);
923 SvREFCNT_dec (self->cb_object);
924
925#endif
926 1519
927PROTOTYPES: ENABLE 1520PROTOTYPES: ENABLE
928 1521
929void encode_cbor (SV *scalar) 1522void encode_cbor (SV *scalar)
1523 ALIAS:
1524 encode_cbor = 0
1525 encode_cbor_sharing = F_ALLOW_SHARING
930 PPCODE: 1526 PPCODE:
931{ 1527{
932 CBOR cbor; 1528 CBOR cbor;
933 cbor_init (&cbor); 1529 cbor_init (&cbor);
1530 cbor.flags |= ix;
934 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1531 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
935 XPUSHs (scalar); 1532 XPUSHs (scalar);
936} 1533}
937 1534
938void decode_cbor (SV *cborstr) 1535void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines