ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/cvsroot/CBOR-XS/XS.xs
(Generate patch)

Comparing cvsroot/CBOR-XS/XS.xs (file contents):
Revision 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 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 encode_sv (enc, AvARRAY (av)[i]);
210 338
211 --enc->depth; 339 --enc->depth;
212} 340}
213 341
214static void 342static void
223 351
224 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
226 354
227 if (mg) 355 if (mg)
228 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 357 else
230 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
231 359
232 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
233 { 361 {
234 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
236 else 364 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 366
239 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
240 } 368 }
241 369
242 if (mg) 370 if (mg)
243 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 372
245 --enc->depth; 373 --enc->depth;
246} 374}
247 375
248// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
249static void 377static void
250encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
251{ 379{
252 svtype svt;
253
254 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
255 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
256 383
257 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
258 { 385 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 387 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 388 : gv_stashpv ("Types::Serialiser::Boolean", 1);
389 HV *error_stash = !CBOR_SLOW || types_error_stash
390 ? types_error_stash
391 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 393 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 395
396 HV *stash = SvSTASH (sv);
397
266 if (SvSTASH (sv) == boolean_stash) 398 if (stash == boolean_stash)
267 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
403 else if (stash == error_stash)
404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
268 else if (SvSTASH (sv) == tagged_stash) 408 else if (stash == tagged_stash)
269 { 409 {
270 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 412
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
275 } 433 }
276 else 434 else
277 { 435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
448 {
449 dSP;
450
451 ENTER; SAVETMPS; PUSHMARK (SP);
278 // we re-bless the reference to get overload and other niceties right 452 // 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))); 453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 454
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 455 PUTBACK;
456 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 457 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 458 SPAGAIN;
292 459
293 // catch this surprisingly common error 460 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 461 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))); 462 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 463
297 sv = POPs; 464 encode_sv (enc, POPs);
465
298 PUTBACK; 466 PUTBACK;
299 467
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 468 FREETMPS; LEAVE;
303 } 469 }
470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
471 {
472 dSP;
473
474 ENTER; SAVETMPS; PUSHMARK (SP);
475 EXTEND (SP, 2);
476 // we re-bless the reference to get overload and other niceties right
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor);
479
480 PUTBACK;
481 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
482 SPAGAIN;
483
484 // catch this surprisingly common error
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
487
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491
492 while (count)
493 encode_sv (enc, SP[1 - count--]);
494
495 PUTBACK;
496
497 FREETMPS; LEAVE;
498 }
304 else 499 else
305 croak ("encountered object '%s', but no TO_CBOR method available on it", 500 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 501 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 502 }
309 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 506 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 507 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
333} 512}
334 513
335static void 514static void
336encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
337{ 516{
338 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
339 518
340 need (enc, 9); 519 need (enc, 9);
341 520
342 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 523 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
346 { 525 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
348 527
349 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 529
351 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
353 532
354 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
356 } 535 }
357 else 536 else
358 { 537 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
360 539
361 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 541
363 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
365 544
366 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
375 554
376 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
377 { 556 {
378 STRLEN len; 557 STRLEN len;
379 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 560 }
382 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 562 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
385 { 564 {
386 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 569 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 571 }
393 else if (SvROK (sv)) 572 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 578 else
400 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 581}
403 582
404static SV * 583static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
406{ 585{
407 enc_t enc; 586 enc_t enc = { };
408 587
409 enc.cbor = *cbor; 588 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 592
415 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
594
595 if (cbor->flags & F_PACK_STRINGS)
596 {
597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
600 }
601
416 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
417 603
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 605 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
420 606
434 U8 *end; // end of input string 620 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
436 CBOR cbor; 622 CBOR cbor;
437 U32 depth; // recursion depth 623 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
439} dec_t; 628} dec_t;
440 629
441#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
442 631
443#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 632#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 635#define DEC_DEC_DEPTH --dec->depth
447 636
448static UV 637static UV
449decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
450{ 639{
451 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
452 { 641 ++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;
457 642
458 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
459 WANT (2); 647 WANT (1);
460 dec->cur += 2; 648 dec->cur += 1;
461 return dec->cur[-1]; 649 return dec->cur[-1];
462 650 }
463 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
464 WANT (3); 653 WANT (2);
465 dec->cur += 3; 654 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
468 657 }
469 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
470 WANT (5); 660 WANT (4);
471 dec->cur += 5; 661 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
476 666 }
477 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
478 WANT (9); 669 WANT (8);
479 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
480 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
484 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
488 685 }
489 default: 686 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 688
493fail: 689fail:
494 return 0; 690 return 0;
495} 691}
496 692
501{ 697{
502 AV *av = newAV (); 698 AV *av = newAV ();
503 699
504 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
505 701
506 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 703 {
508 ++dec->cur; 704 ++dec->cur;
509 705
510 for (;;) 706 for (;;)
511 { 707 {
512 WANT (1); 708 WANT (1);
513 709
514 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 711 {
516 ++dec->cur; 712 ++dec->cur;
517 break; 713 break;
518 } 714 }
519 715
522 } 718 }
523 else 719 else
524 { 720 {
525 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
526 722
723 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); 724 av_fill (av, len - 1);
528 725
529 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
531 } 728 }
537 SvREFCNT_dec (av); 734 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 735 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 736 return &PL_sv_undef;
540} 737}
541 738
739static void
740decode_he (dec_t *dec, HV *hv)
741{
742 // for speed reasons, we specialcase single-string
743 // byte or utf-8 strings as keys, but only when !stringref
744
745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
756 }
757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
758 {
759 I32 len = decode_uint (dec);
760 char *key = (char *)dec->cur;
761
762 dec->cur += len;
763
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
767
768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
771 }
772
773 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec);
775
776 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k);
778
779fail:
780 ;
781}
782
542static SV * 783static SV *
543decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
544{ 785{
545 HV *hv = newHV (); 786 HV *hv = newHV ();
546 787
547 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
548 789
549 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 791 {
551 ++dec->cur; 792 ++dec->cur;
552 793
553 for (;;) 794 for (;;)
554 { 795 {
555 WANT (1); 796 WANT (1);
556 797
557 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 799 {
559 ++dec->cur; 800 ++dec->cur;
560 break; 801 break;
561 } 802 }
562 803
563 SV *k = decode_sv (dec); 804 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 805 }
568 } 806 }
569 else 807 else
570 { 808 {
571 int len = decode_uint (dec); 809 int pairs = decode_uint (dec);
572 810
573 while (len--) 811 while (pairs--)
574 { 812 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 } 813 }
581 814
582 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 816 return newRV_noinc ((SV *)hv);
584 817
591static SV * 824static SV *
592decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
593{ 826{
594 SV *sv = 0; 827 SV *sv = 0;
595 828
596 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 830 {
831 // indefinite length strings
598 ++dec->cur; 832 ++dec->cur;
599 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
600 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
601 837
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 838 for (;;)
604 { 839 {
605 WANT (1); 840 WANT (1);
606 841
607 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 844 {
609 ++dec->cur; 845 ++dec->cur;
610 break; 846 break;
611 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 850
613 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
614 } 856 }
615 } 857 }
616 else 858 else
617 { 859 {
618 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
619 861
620 WANT (len); 862 WANT (len);
621 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 869 }
624 870
625 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
626 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
627 879
628 return sv; 880 return sv;
629 881
630fail: 882fail:
631 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
633} 885}
634 886
635static SV * 887static SV *
636decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
637{ 889{
890 SV *sv = 0;
638 UV tag = decode_uint (dec); 891 UV tag = decode_uint (dec);
892
893 WANT (1);
894
895 switch (tag)
896 {
897 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 898 sv = decode_sv (dec);
899 break;
640 900
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 901 case CBOR_TAG_INDIRECTION:
902 sv = newRV_noinc (decode_sv (dec));
903 break;
904
905 case CBOR_TAG_STRINGREF_NAMESPACE:
906 {
907 ENTER; SAVETMPS;
908
909 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911
912 sv = decode_sv (dec);
913
914 FREETMPS; LEAVE;
915 }
916 break;
917
918 case CBOR_TAG_STRINGREF:
919 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922
923 UV idx = decode_uint (dec);
924
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 }
930 break;
931
932 case CBOR_TAG_VALUE_SHAREABLE:
933 {
934 if (ecb_expect_false (!dec->shareable))
935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
939 sv = newSV (0);
940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
941
942 SV *osv = decode_sv (dec);
943 sv_setsv (sv, osv);
944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
953 }
954 break;
955
956 case CBOR_TAG_VALUE_SHAREDREF:
957 {
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960
961 UV idx = decode_uint (dec);
962
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
970 }
971 break;
972
973 case CBOR_TAG_PERL_OBJECT:
974 {
975 sv = decode_sv (dec);
976
977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
978 ERR ("corrupted CBOR data (non-array perl object)");
979
980 AV *av = (AV *)SvRV (sv);
981 int len = av_len (av) + 1;
982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
983
984 if (!stash)
985 ERR ("cannot decode perl-object (package does not exist)");
986
987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
988
989 if (!method)
990 ERR ("cannot decode perl-object (package does not have a THAW method)");
991
992 dSP;
993
994 ENTER; SAVETMPS; PUSHMARK (SP);
995 EXTEND (SP, len + 1);
996 // we re-bless the reference to get overload and other niceties right
997 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor);
999
1000 int i;
1001
1002 for (i = 1; i < len; ++i)
1003 PUSHs (*av_fetch (av, i, 1));
1004
1005 PUTBACK;
1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1007 SPAGAIN;
1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
1015 SvREFCNT_dec (sv);
1016 sv = SvREFCNT_inc (POPs);
1017
1018 PUTBACK;
1019
1020 FREETMPS; LEAVE;
1021 }
1022 break;
1023
1024 default:
1025 {
1026 sv = decode_sv (dec);
1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
1051 AV *av = newAV ();
1052 av_push (av, newSVuv (tag));
1053 av_push (av, sv);
1054
1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1056 ? cbor_tagged_stash
1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
1066 }
1067
642 return sv; 1068 return sv;
643 1069
644 AV *av = newAV (); 1070fail:
645 av_push (av, newSVuv (tag)); 1071 SvREFCNT_dec (sv);
646 av_push (av, sv); 1072 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} 1073}
654 1074
655static SV * 1075static SV *
656decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
657{ 1077{
658 WANT (1); 1078 WANT (1);
659 1079
660 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1081 {
662 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1089
670 case 4: // array 1090 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) 1091 switch (*dec->cur++ & MINOR_MASK)
678 { 1092 {
679 case 20: 1093 case SIMPLE_FALSE:
680#if CBOR_SLOW 1094#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
682#endif 1096#endif
683 return newSVsv (cbor_false); 1097 return newSVsv (types_false);
684 case 21: 1098 case SIMPLE_TRUE:
685#if CBOR_SLOW 1099#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
687#endif 1101#endif
688 return newSVsv (cbor_true); 1102 return newSVsv (types_true);
689 case 22: 1103 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
1105 case SIMPLE_UNDEF:
1106#if CBOR_SLOW
1107 types_error = get_bool ("Types::Serialiser::error");
1108#endif
1109 return newSVsv (types_error);
691 1110
692 case 25: 1111 case MISC_FLOAT16:
693 { 1112 {
694 WANT (2); 1113 WANT (2);
695 1114
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1116 dec->cur += 2;
698 1117
699 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
700 } 1119 }
701 1120
702 case 26: 1121 case MISC_FLOAT32:
703 { 1122 {
704 uint32_t fp; 1123 uint32_t fp;
705 WANT (4); 1124 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1126 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
711 1130
712 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
713 } 1132 }
714 1133
715 case 27: 1134 case MISC_FLOAT64:
716 { 1135 {
717 uint64_t fp; 1136 uint64_t fp;
718 WANT (8); 1137 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1139 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
724 1143
725 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
726 } 1145 }
727 1146
728 // 0..19 unassigned 1147 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
730 default: 1151 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
732 } 1153 }
733 1154
734 break; 1155 break;
735 } 1156 }
736 1157
739} 1160}
740 1161
741static SV * 1162static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1164{
744 dec_t dec; 1165 dec_t dec = { };
745 SV *sv; 1166 SV *sv;
1167 STRLEN len;
1168 char *data = SvPVbyte (string, len);
746 1169
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) 1170 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", 1171 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); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1173
782 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1175 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1176 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1177
788 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
789 1179
790 if (offset_return) 1180 if (offset_return)
791 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
794 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
795 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
796 1186
797 if (dec.err) 1187 if (dec.err)
798 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
799 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1202 }
802 1203
803 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
804 1205
805 return sv; 1206 return sv;
806} 1207}
807 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
808///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
809// XS interface functions 1330// XS interface functions
810 1331
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1333
813BOOT: 1334BOOT:
814{ 1335{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1336 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); 1337 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1338
819 cbor_true = get_bool ("CBOR::XS::true"); 1339 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1341
1342 types_true = get_bool ("Types::Serialiser::true" );
1343 types_false = get_bool ("Types::Serialiser::false");
1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347
1348 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor);
821} 1350}
822 1351
823PROTOTYPES: DISABLE 1352PROTOTYPES: DISABLE
824 1353
825void CLONE (...) 1354void CLONE (...)
826 CODE: 1355 CODE:
827 cbor_stash = 0; 1356 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1357 cbor_tagged_stash = 0;
1358 types_error_stash = 0;
1359 types_boolean_stash = 0;
830 1360
831void new (char *klass) 1361void new (char *klass)
832 PPCODE: 1362 PPCODE:
833{ 1363{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1364 SV *pv = NEWSV (0, sizeof (CBOR));
842 1372
843void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1374 ALIAS:
845 shrink = F_SHRINK 1375 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
847 PPCODE: 1381 PPCODE:
848{ 1382{
849 if (enable) 1383 if (enable)
850 self->flags |= ix; 1384 self->flags |= ix;
851 else 1385 else
856 1390
857void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
858 ALIAS: 1392 ALIAS:
859 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
861 PPCODE: 1399 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
863 1401
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1403 PPCODE:
881 CODE: 1419 CODE:
882 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
883 OUTPUT: 1421 OUTPUT:
884 RETVAL 1422 RETVAL
885 1423
886#if 0 //TODO 1424void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1425 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1426 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1427 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1428 XPUSHs (ST (0));
895}
896 1429
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1430SV *get_filter (CBOR *self)
898 PPCODE: 1431 CODE:
899{ 1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1433 OUTPUT:
901 self->cb_sk_object = newHV (); 1434 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 1435
921void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
922 PPCODE: 1437 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1438 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1439 XPUSHs (scalar);
937 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
938 PUSHs (sv); 1453 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1455}
941 1456
942#if 0 1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
943 1508
944void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
945 CODE: 1510 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1511 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1512
951PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
952 1514
953void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1519 PPCODE:
955{ 1520{
956 CBOR cbor; 1521 CBOR cbor;
957 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1525 XPUSHs (scalar);
960} 1526}
961 1527
962void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines