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 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.50 by root, Thu Feb 25 02:29:22 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
14// known tags, rfc7049 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
15enum cbor_tag 67enum cbor_tag
16{ 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
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
23 85
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 90
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 96
35 CBOR_TAG_MAGIC = 55799 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
36}; 98};
37 99
38#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
39#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
40 106
41#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
42 108
43#define SB do { 109#define SB do {
44#define SE } while (0) 110#define SE } while (0)
55#else 121#else
56# define CBOR_SLOW 0 122# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
58#endif 124#endif
59 125
60static 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::
61static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 128
63typedef struct { 129typedef struct {
64 U32 flags; 130 U32 flags;
65 U32 max_depth; 131 U32 max_depth;
66 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.
67} CBOR; 139} CBOR;
68 140
69ecb_inline void 141ecb_inline void
70cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
71{ 143{
72 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
73 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);
74} 153}
75 154
76///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
77// utility functions 156// utility functions
78 157
100 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 180#endif
102 } 181 }
103} 182}
104 183
105///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
106// fp hell 185ecb_inline int
107 186minimum_string_length (UV idx)
108//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}
109 198
110///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
111// encoder 200// encoder
112 201
113// structure used for encoding CBOR 202// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
118 SV *sv; // result scalar 207 SV *sv; // result scalar
119 CBOR cbor; 208 CBOR cbor;
120 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;
121} enc_t; 214} enc_t;
122 215
123ecb_inline void 216ecb_inline void
124need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
125{ 218{
142static void 235static void
143encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
144{ 237{
145 need (enc, 9); 238 need (enc, 9);
146 239
147 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
149 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
150 { 243 {
151 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 245 *enc->cur++ = len;
153 } 246 }
154 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
155 { 248 {
156 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 251 *enc->cur++ = len;
159 } 252 }
160 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
161 { 254 {
162 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 259 *enc->cur++ = len;
167 } 260 }
168 else 261 else
169 { 262 {
170 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 271 *enc->cur++ = len;
179 } 272 }
180} 273}
181 274
182static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 283{
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
186 need (enc, len); 285 need (enc, len);
187 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
188 enc->cur += len; 287 enc->cur += len;
189} 288}
190 289
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
191static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
192 316
193static void 317static void
194encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
195{ 319{
198 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
200 324
201 ++enc->depth; 325 ++enc->depth;
202 326
203 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 328
329 if (SvMAGICAL (av))
205 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
206 { 331 {
207 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
209 } 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 }
210 341
211 --enc->depth; 342 --enc->depth;
212} 343}
213 344
214static void 345static void
223 354
224 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
226 357
227 if (mg) 358 if (mg)
228 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 360 else
230 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
231 362
232 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
233 { 364 {
234 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
236 else 367 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 369
239 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));
240 } 371 }
241 372
242 if (mg) 373 if (mg)
243 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 375
245 --enc->depth; 376 --enc->depth;
246} 377}
247 378
248// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
249static void 380static void
250encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
251{ 382{
252 svtype svt;
253
254 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
255 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
256 386
257 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
258 { 388 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 390 ? types_boolean_stash
261 : 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);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 396 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 398
399 HV *stash = SvSTASH (sv);
400
266 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
267 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 }
268 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
269 { 412 {
270 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 415
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 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;
275 } 436 }
276 else 437 else
277 { 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;
455 PUSHMARK (SP);
278 // we re-bless the reference to get overload and other niceties right 456 // we re-bless the reference to get overload and other niceties right
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 458
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 459 PUTBACK;
460 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 461 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 462 SPAGAIN;
292 463
293 // catch this surprisingly common error 464 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 465 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 466 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 467
297 sv = POPs; 468 encode_sv (enc, POPs);
469
298 PUTBACK; 470 PUTBACK;
299 471
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 472 FREETMPS; LEAVE;
303 } 473 }
474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
475 {
476 dSP;
477
478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
481 EXTEND (SP, 2);
482 // we re-bless the reference to get overload and other niceties right
483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
484 PUSHs (sv_cbor);
485
486 PUTBACK;
487 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
488 SPAGAIN;
489
490 // catch this surprisingly common error
491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
493
494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
495 encode_uint (enc, MAJOR_ARRAY, count + 1);
496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
497
498 while (count)
499 encode_sv (enc, SP[1 - count--]);
500
501 PUTBACK;
502
503 FREETMPS; LEAVE;
504 }
304 else 505 else
305 croak ("encountered object '%s', but no TO_CBOR method available on it", 506 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 507 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 508 }
309 else if (svt == SVt_PVHV) 509 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 510 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 511 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 512 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 513 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 514 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 515 encode_tag (enc, CBOR_TAG_INDIRECTION);
516 encode_sv (enc, sv);
517 }
333} 518}
334 519
335static void 520static void
336encode_nv (enc_t *enc, SV *sv) 521encode_nv (enc_t *enc, SV *sv)
337{ 522{
338 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
339 524
340 need (enc, 9); 525 need (enc, 9);
341 526
342 if (ecb_expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 529 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
346 { 531 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
348 533
349 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 535
351 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
353 538
354 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
356 } 541 }
357 else 542 else
358 { 543 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
360 545
361 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 547
363 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
365 550
366 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
375 560
376 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
377 { 562 {
378 STRLEN len; 563 STRLEN len;
379 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 566 }
382 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 568 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
385 { 570 {
386 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 575 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 577 }
393 else if (SvROK (sv)) 578 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 584 else
400 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 585 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 587}
403 588
404static SV * 589static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
406{ 591{
407 enc_t enc; 592 enc_t enc = { 0 };
408 593
409 enc.cbor = *cbor; 594 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 598
415 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
600
601 if (cbor->flags & F_PACK_STRINGS)
602 {
603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
606 }
607
416 encode_sv (&enc, scalar); 608 encode_sv (&enc, scalar);
417 609
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 610 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 611 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 612
434 U8 *end; // end of input string 626 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 627 const char *err; // parse error, if != 0
436 CBOR cbor; 628 CBOR cbor;
437 U32 depth; // recursion depth 629 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 630 U32 maxdepth; // recursion depth limit
631 AV *shareable;
632 AV *stringref;
633 SV *decode_tagged;
439} dec_t; 634} dec_t;
440 635
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 636#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 637
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 638#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
446#define DEC_DEC_DEPTH --dec->depth 641#define DEC_DEC_DEPTH --dec->depth
447 642
448static UV 643static UV
449decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
450{ 645{
451 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
452 { 647 ++dec->cur;
453 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
454 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
455 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
456 return *dec->cur++ & 31;
457 648
458 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
459 WANT (2); 653 WANT (1);
460 dec->cur += 2; 654 dec->cur += 1;
461 return dec->cur[-1]; 655 return dec->cur[-1];
462 656 }
463 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
464 WANT (3); 659 WANT (2);
465 dec->cur += 3; 660 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
468 663 }
469 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
470 WANT (5); 666 WANT (4);
471 dec->cur += 5; 667 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
476 672 }
477 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
478 WANT (9); 675 WANT (8);
479 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
480 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
484 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
488 691 }
489 default: 692 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 694
493fail: 695fail:
494 return 0; 696 return 0;
495} 697}
496 698
501{ 703{
502 AV *av = newAV (); 704 AV *av = newAV ();
503 705
504 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
505 707
506 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 709 {
508 ++dec->cur; 710 ++dec->cur;
509 711
510 for (;;) 712 for (;;)
511 { 713 {
512 WANT (1); 714 WANT (1);
513 715
514 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 717 {
516 ++dec->cur; 718 ++dec->cur;
517 break; 719 break;
518 } 720 }
519 721
522 } 724 }
523 else 725 else
524 { 726 {
525 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
526 728
729 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
527 av_fill (av, len - 1); 730 av_fill (av, len - 1);
528 731
529 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
531 } 734 }
537 SvREFCNT_dec (av); 740 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 741 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 742 return &PL_sv_undef;
540} 743}
541 744
745static void
746decode_he (dec_t *dec, HV *hv)
747{
748 // for speed reasons, we specialcase single-string
749 // byte or utf-8 strings as keys, but only when !stringref
750
751 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 WANT (len);
758 dec->cur += len;
759
760 hv_store (hv, key, len, decode_sv (dec), 0);
761
762 return;
763 }
764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
765 {
766 I32 len = decode_uint (dec);
767 char *key = (char *)dec->cur;
768
769 WANT (len);
770 dec->cur += len;
771
772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
775
776 hv_store (hv, key, -len, decode_sv (dec), 0);
777
778 return;
779 }
780
781 SV *k = decode_sv (dec);
782 SV *v = decode_sv (dec);
783
784 hv_store_ent (hv, k, v, 0);
785 SvREFCNT_dec (k);
786
787fail:
788 ;
789}
790
542static SV * 791static SV *
543decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
544{ 793{
545 HV *hv = newHV (); 794 HV *hv = newHV ();
546 795
547 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
548 797
549 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 799 {
551 ++dec->cur; 800 ++dec->cur;
552 801
553 for (;;) 802 for (;;)
554 { 803 {
555 WANT (1); 804 WANT (1);
556 805
557 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 807 {
559 ++dec->cur; 808 ++dec->cur;
560 break; 809 break;
561 } 810 }
562 811
563 SV *k = decode_sv (dec); 812 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 813 }
568 } 814 }
569 else 815 else
570 { 816 {
571 int len = decode_uint (dec); 817 int pairs = decode_uint (dec);
572 818
573 while (len--) 819 while (pairs--)
574 { 820 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 821 }
581 822
582 DEC_DEC_DEPTH; 823 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 824 return newRV_noinc ((SV *)hv);
584 825
591static SV * 832static SV *
592decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
593{ 834{
594 SV *sv = 0; 835 SV *sv = 0;
595 836
596 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 838 {
839 // indefinite length strings
598 ++dec->cur; 840 ++dec->cur;
599 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
600 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
601 845
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 846 for (;;)
604 { 847 {
605 WANT (1); 848 WANT (1);
606 849
607 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 852 {
609 ++dec->cur; 853 ++dec->cur;
610 break; 854 break;
611 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 858
613 sv_catsv (sv, decode_sv (dec)); 859 STRLEN len = decode_uint (dec);
860
861 WANT (len);
862 sv_catpvn (sv, dec->cur, len);
863 dec->cur += len;
614 } 864 }
615 } 865 }
616 else 866 else
617 { 867 {
618 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
619 869
620 WANT (len); 870 WANT (len);
621 sv = newSVpvn (dec->cur, len); 871 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 872 dec->cur += len;
873
874 if (ecb_expect_false (dec->stringref)
875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 877 }
624 878
625 if (utf8) 879 if (utf8)
880 {
881 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
882 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
883 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
884
626 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
627 887
628 return sv; 888 return sv;
629 889
630fail: 890fail:
631 SvREFCNT_dec (sv); 891 SvREFCNT_dec (sv);
633} 893}
634 894
635static SV * 895static SV *
636decode_tagged (dec_t *dec) 896decode_tagged (dec_t *dec)
637{ 897{
898 SV *sv = 0;
638 UV tag = decode_uint (dec); 899 UV tag = decode_uint (dec);
900
901 WANT (1);
902
903 switch (tag)
904 {
905 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 906 sv = decode_sv (dec);
907 break;
640 908
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 909 case CBOR_TAG_INDIRECTION:
910 sv = newRV_noinc (decode_sv (dec));
911 break;
912
913 case CBOR_TAG_STRINGREF_NAMESPACE:
914 {
915 ENTER; SAVETMPS;
916
917 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919
920 sv = decode_sv (dec);
921
922 FREETMPS; LEAVE;
923 }
924 break;
925
926 case CBOR_TAG_STRINGREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
934 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
935
936 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_VALUE_SHAREABLE:
941 {
942 if (ecb_expect_false (!dec->shareable))
943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
947 sv = newSV (0);
948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
949
950 SV *osv = decode_sv (dec);
951 sv_setsv (sv, osv);
952 SvREFCNT_dec_NN (osv);
953 }
954 else
955 {
956 av_push (dec->shareable, &PL_sv_undef);
957 int idx = AvFILLp (dec->shareable);
958 sv = decode_sv (dec);
959 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
960 }
961 }
962 break;
963
964 case CBOR_TAG_VALUE_SHAREDREF:
965 {
966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
968
969 UV idx = decode_uint (dec);
970
971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
973
974 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
975
976 if (sv == &PL_sv_undef)
977 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
978 }
979 break;
980
981 case CBOR_TAG_PERL_OBJECT:
982 {
983 sv = decode_sv (dec);
984
985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
986 ERR ("corrupted CBOR data (non-array perl object)");
987
988 AV *av = (AV *)SvRV (sv);
989 int len = av_len (av) + 1;
990 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
991
992 if (!stash)
993 ERR ("cannot decode perl-object (package does not exist)");
994
995 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
996
997 if (!method)
998 ERR ("cannot decode perl-object (package does not have a THAW method)");
999
1000 dSP;
1001
1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
1004 EXTEND (SP, len + 1);
1005 // we re-bless the reference to get overload and other niceties right
1006 PUSHs (*av_fetch (av, 0, 1));
1007 PUSHs (sv_cbor);
1008
1009 int i;
1010
1011 for (i = 1; i < len; ++i)
1012 PUSHs (*av_fetch (av, i, 1));
1013
1014 PUTBACK;
1015 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1016 SPAGAIN;
1017
1018 if (SvTRUE (ERRSV))
1019 {
1020 FREETMPS; LEAVE;
1021 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1022 }
1023
1024 SvREFCNT_dec (sv);
1025 sv = SvREFCNT_inc (POPs);
1026
1027 PUTBACK;
1028
1029 FREETMPS; LEAVE;
1030 }
1031 break;
1032
1033 default:
1034 {
1035 sv = decode_sv (dec);
1036
1037 dSP;
1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
1041 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag));
1043 PUSHs (sv);
1044
1045 PUTBACK;
1046 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1047 SPAGAIN;
1048
1049 if (SvTRUE (ERRSV))
1050 {
1051 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1053 }
1054
1055 if (count)
1056 {
1057 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs);
1059 }
1060 else
1061 {
1062 AV *av = newAV ();
1063 av_push (av, newSVuv (tag));
1064 av_push (av, sv);
1065
1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1067 ? cbor_tagged_stash
1068 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1069 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1070 }
1071
1072 PUTBACK;
1073
1074 FREETMPS; LEAVE;
1075 }
1076 break;
1077 }
1078
642 return sv; 1079 return sv;
643 1080
644 AV *av = newAV (); 1081fail:
645 av_push (av, newSVuv (tag)); 1082 SvREFCNT_dec (sv);
646 av_push (av, sv); 1083 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 1084}
654 1085
655static SV * 1086static SV *
656decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
657{ 1088{
658 WANT (1); 1089 WANT (1);
659 1090
660 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1092 {
662 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1100
670 case 4: // array 1101 case MAJOR_MISC >> MAJOR_SHIFT:
671 return decode_av (dec);
672 case 5: // map
673 return decode_hv (dec);
674 case 6: // tag
675 return decode_tagged (dec);
676 case 7: // misc
677 switch (*dec->cur++ & 31) 1102 switch (*dec->cur++ & MINOR_MASK)
678 { 1103 {
679 case 20: 1104 case SIMPLE_FALSE:
680#if CBOR_SLOW 1105#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
682#endif 1107#endif
683 return newSVsv (cbor_false); 1108 return newSVsv (types_false);
684 case 21: 1109 case SIMPLE_TRUE:
685#if CBOR_SLOW 1110#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
687#endif 1112#endif
688 return newSVsv (cbor_true); 1113 return newSVsv (types_true);
689 case 22: 1114 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
1116 case SIMPLE_UNDEF:
1117#if CBOR_SLOW
1118 types_error = get_bool ("Types::Serialiser::error");
1119#endif
1120 return newSVsv (types_error);
691 1121
692 case 25: 1122 case MISC_FLOAT16:
693 { 1123 {
694 WANT (2); 1124 WANT (2);
695 1125
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1127 dec->cur += 2;
698 1128
699 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
700 } 1130 }
701 1131
702 case 26: 1132 case MISC_FLOAT32:
703 { 1133 {
704 uint32_t fp; 1134 uint32_t fp;
705 WANT (4); 1135 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1137 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
711 1141
712 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
713 } 1143 }
714 1144
715 case 27: 1145 case MISC_FLOAT64:
716 { 1146 {
717 uint64_t fp; 1147 uint64_t fp;
718 WANT (8); 1148 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1150 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
724 1154
725 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
726 } 1156 }
727 1157
728 // 0..19 unassigned 1158 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1159 // 24 reserved + unassigned simple (reserved values are not encodable)
1160 // 28-30 unassigned misc
1161 // 31 break code
730 default: 1162 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
732 } 1164 }
733 1165
734 break; 1166 break;
735 } 1167 }
736 1168
739} 1171}
740 1172
741static SV * 1173static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1175{
744 dec_t dec; 1176 dec_t dec = { 0 };
745 SV *sv; 1177 SV *sv;
1178 STRLEN len;
1179 char *data = SvPVbyte (string, len);
746 1180
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
776 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1182 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1183 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1184
782 dec.cbor = *cbor; 1185 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1186 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1187 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1188
788 sv = decode_sv (&dec); 1189 sv = decode_sv (&dec);
789 1190
790 if (offset_return) 1191 if (offset_return)
791 *offset_return = dec.cur; 1192 *offset_return = dec.cur;
794 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
795 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
796 1197
797 if (dec.err) 1198 if (dec.err)
798 { 1199 {
1200 if (dec.shareable)
1201 {
1202 // need to break cyclic links, which whould all be in shareable
1203 int i;
1204 SV **svp;
1205
1206 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef);
1209 }
1210
799 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1212 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1213 }
802 1214
803 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
804 1216
805 return sv; 1217 return sv;
806} 1218}
807 1219
1220/////////////////////////////////////////////////////////////////////////////
1221// incremental parser
1222
1223#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1224
1225// returns 0 for notyet, 1 for success or error
1226static int
1227incr_parse (CBOR *self, SV *cborstr)
1228{
1229 STRLEN cur;
1230 SvPV (cborstr, cur);
1231
1232 while (ecb_expect_true (self->incr_need <= cur))
1233 {
1234 // table of integer count bytes
1235 static I8 incr_len[MINOR_MASK + 1] = {
1236 0, 0, 0, 0, 0, 0, 0, 0,
1237 0, 0, 0, 0, 0, 0, 0, 0,
1238 0, 0, 0, 0, 0, 0, 0, 0,
1239 1, 2, 4, 8,-1,-1,-1,-2
1240 };
1241
1242 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1243 U8 m = *p & MINOR_MASK;
1244 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1245 I8 ilen = incr_len[m];
1246
1247 self->incr_need = self->incr_pos + 1;
1248
1249 if (ecb_expect_false (ilen < 0))
1250 {
1251 if (m != MINOR_INDEF)
1252 return 1; // error
1253
1254 if (*p == (MAJOR_MISC | MINOR_INDEF))
1255 {
1256 if (count >= 0)
1257 return 1; // error
1258
1259 count = 1;
1260 }
1261 else
1262 {
1263 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1264 count = -1;
1265 }
1266 }
1267 else
1268 {
1269 self->incr_need += ilen;
1270 if (ecb_expect_false (self->incr_need > cur))
1271 return 0;
1272
1273 int major = *p >> MAJOR_SHIFT;
1274
1275 switch (major)
1276 {
1277 case MAJOR_TAG >> MAJOR_SHIFT:
1278 ++count; // tags merely prefix another value
1279 break;
1280
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 case MAJOR_ARRAY >> MAJOR_SHIFT:
1284 case MAJOR_MAP >> MAJOR_SHIFT:
1285 {
1286 UV len;
1287
1288 if (ecb_expect_false (ilen))
1289 {
1290 len = 0;
1291
1292 do {
1293 len = (len << 8) | *++p;
1294 } while (--ilen);
1295 }
1296 else
1297 len = m;
1298
1299 switch (major)
1300 {
1301 case MAJOR_BYTES >> MAJOR_SHIFT:
1302 case MAJOR_TEXT >> MAJOR_SHIFT:
1303 self->incr_need += len;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 break;
1308
1309 case MAJOR_MAP >> MAJOR_SHIFT:
1310 len <<= 1;
1311 case MAJOR_ARRAY >> MAJOR_SHIFT:
1312 if (len)
1313 {
1314 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1315 count = len + 1;
1316 }
1317 break;
1318 }
1319 }
1320 }
1321 }
1322
1323 self->incr_pos = self->incr_need;
1324
1325 if (count > 0)
1326 {
1327 while (!--count)
1328 {
1329 if (!AvFILLp (self->incr_count))
1330 return 1; // done
1331
1332 SvREFCNT_dec_NN (av_pop (self->incr_count));
1333 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1334 }
1335
1336 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1337 }
1338 }
1339
1340 return 0;
1341}
1342
1343
808///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
809// XS interface functions 1345// XS interface functions
810 1346
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1348
813BOOT: 1349BOOT:
814{ 1350{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1351 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1352 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1353
819 cbor_true = get_bool ("CBOR::XS::true"); 1354 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1355 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1356
1357 types_true = get_bool ("Types::Serialiser::true" );
1358 types_false = get_bool ("Types::Serialiser::false");
1359 types_error = get_bool ("Types::Serialiser::error");
1360
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1362
1363 sv_cbor = newSVpv ("CBOR", 0);
1364 SvREADONLY_on (sv_cbor);
821} 1365}
822 1366
823PROTOTYPES: DISABLE 1367PROTOTYPES: DISABLE
824 1368
825void CLONE (...) 1369void CLONE (...)
826 CODE: 1370 CODE:
827 cbor_stash = 0; 1371 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1372 cbor_tagged_stash = 0;
1373 types_error_stash = 0;
1374 types_boolean_stash = 0;
830 1375
831void new (char *klass) 1376void new (char *klass)
832 PPCODE: 1377 PPCODE:
833{ 1378{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1379 SV *pv = NEWSV (0, sizeof (CBOR));
842 1387
843void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1389 ALIAS:
845 shrink = F_SHRINK 1390 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1392 allow_sharing = F_ALLOW_SHARING
1393 allow_cycles = F_ALLOW_CYCLES
1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
847 PPCODE: 1396 PPCODE:
848{ 1397{
849 if (enable) 1398 if (enable)
850 self->flags |= ix; 1399 self->flags |= ix;
851 else 1400 else
856 1405
857void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
858 ALIAS: 1407 ALIAS:
859 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1410 get_allow_sharing = F_ALLOW_SHARING
1411 get_allow_cycles = F_ALLOW_CYCLES
1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
861 PPCODE: 1414 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
863 1416
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1418 PPCODE:
881 CODE: 1434 CODE:
882 RETVAL = self->max_size; 1435 RETVAL = self->max_size;
883 OUTPUT: 1436 OUTPUT:
884 RETVAL 1437 RETVAL
885 1438
886#if 0 //TODO 1439void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1440 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1441 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1442 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1443 XPUSHs (ST (0));
895}
896 1444
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1445SV *get_filter (CBOR *self)
898 PPCODE: 1446 CODE:
899{ 1447 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1448 OUTPUT:
901 self->cb_sk_object = newHV (); 1449 RETVAL
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1450
921void encode (CBOR *self, SV *scalar) 1451void encode (CBOR *self, SV *scalar)
922 PPCODE: 1452 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1453 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1454 XPUSHs (scalar);
937 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
938 PUSHs (sv); 1468 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1470}
941 1471
942#if 0 1472void incr_parse (CBOR *self, SV *cborstr)
1473 ALIAS:
1474 incr_parse_multiple = 1
1475 PPCODE:
1476{
1477 if (SvUTF8 (cborstr))
1478 sv_utf8_downgrade (cborstr, 0);
1479
1480 if (!self->incr_count)
1481 {
1482 self->incr_count = newAV ();
1483 self->incr_pos = 0;
1484 self->incr_need = 1;
1485
1486 av_push (self->incr_count, newSViv (1));
1487 }
1488
1489 do
1490 {
1491 if (!incr_parse (self, cborstr))
1492 {
1493 if (self->incr_need > self->max_size && self->max_size)
1494 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1495 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1496
1497 break;
1498 }
1499
1500 SV *sv;
1501 char *offset;
1502
1503 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1504 XPUSHs (sv);
1505
1506 sv_chop (cborstr, offset);
1507
1508 av_clear (self->incr_count);
1509 av_push (self->incr_count, newSViv (1));
1510
1511 self->incr_pos = 0;
1512 self->incr_need = self->incr_pos + 1;
1513 }
1514 while (ix);
1515}
1516
1517void incr_reset (CBOR *self)
1518 CODE:
1519{
1520 SvREFCNT_dec (self->incr_count);
1521 self->incr_count = 0;
1522}
943 1523
944void DESTROY (CBOR *self) 1524void DESTROY (CBOR *self)
945 CODE: 1525 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1526 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1527
951PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
952 1529
953void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1534 PPCODE:
955{ 1535{
956 CBOR cbor; 1536 CBOR cbor;
957 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1540 XPUSHs (scalar);
960} 1541}
961 1542
962void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines