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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.7 by root, Sat Oct 26 23:02:55 2013 UTC vs.
Revision 1.57 by root, Fri Nov 25 12:16:12 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
16 108
17#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
18 110
19#define SB do { 111#define SB do {
20#define SE } while (0) 112#define SE } while (0)
31#else 123#else
32# define CBOR_SLOW 0 124# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
34#endif 126#endif
35 127
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 130
39typedef struct { 131typedef struct {
40 U32 flags; 132 U32 flags;
41 U32 max_depth; 133 U32 max_depth;
42 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 141} CBOR;
44 142
45ecb_inline void 143ecb_inline void
46cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
47{ 145{
48 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 147 cbor->max_depth = 512;
148}
149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
50} 155}
51 156
52///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
53// utility functions 158// utility functions
54 159
76 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 182#endif
78 } 183 }
79} 184}
80 185
81///////////////////////////////////////////////////////////////////////////// 186// minimum length of a string to be registered for stringref
82// fp hell 187ecb_inline int
83 188minimum_string_length (UV idx)
84//TODO 189{
190 return idx > 23
191 ? idx > 0xffU
192 ? idx > 0xffffU
193 ? idx > 0xffffffffU
194 ? 11
195 : 7
196 : 5
197 : 4
198 : 3;
199}
85 200
86///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
87// encoder 202// encoder
88 203
89// structure used for encoding CBOR 204// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
94 SV *sv; // result scalar 209 SV *sv; // result scalar
95 CBOR cbor; 210 CBOR cbor;
96 U32 depth; // recursion level 211 U32 depth; // recursion level
212 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
213 UV stringref_idx;
214 HV *shareable; // ptr => index, or 0
215 UV shareable_idx;
97} enc_t; 216} enc_t;
98 217
99ecb_inline void 218ecb_inline void
100need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
101{ 220{
118static void 237static void
119encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
120{ 239{
121 need (enc, 9); 240 need (enc, 9);
122 241
123 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
125 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
126 { 245 {
127 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 247 *enc->cur++ = len;
129 } 248 }
130 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
131 { 250 {
132 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 253 *enc->cur++ = len;
135 } 254 }
136 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
137 { 256 {
138 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 261 *enc->cur++ = len;
143 } 262 }
144 else 263 else
145 { 264 {
146 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 273 *enc->cur++ = len;
155 } 274 }
156} 275}
157 276
158static void 277ecb_inline void
278encode_tag (enc_t *enc, UV tag)
279{
280 encode_uint (enc, MAJOR_TAG, tag);
281}
282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 286{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 317 need (enc, len);
163 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
164 enc->cur += len; 319 enc->cur += len;
165} 320}
166 321
322ecb_inline void
323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
324{
325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
328
329 if (SvOK (*svp))
330 {
331 // already registered, use stringref
332 encode_tag (enc, CBOR_TAG_STRINGREF);
333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
334 return;
335 }
336 else if (len >= minimum_string_length (enc->stringref_idx))
337 {
338 // register only
339 sv_setuv (*svp, enc->stringref_idx);
340 ++enc->stringref_idx;
341 }
342 }
343
344 encode_str (enc, upgrade_utf8, utf8, str, len);
345}
346
167static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
168 348
169static void 349static void
170encode_av (enc_t *enc, AV *av) 350encode_av (enc_t *enc, AV *av)
171{ 351{
174 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
176 356
177 ++enc->depth; 357 ++enc->depth;
178 358
179 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 360
361 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
182 { 363 {
183 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
186 373
187 --enc->depth; 374 --enc->depth;
188} 375}
189 376
190static void 377static void
199 386
200 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
202 389
203 if (mg) 390 if (mg)
204 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 392 else
206 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
207 394
208 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
209 { 396 {
210 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
212 else 399 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 401
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 403 }
217 404
218 if (mg) 405 if (mg)
219 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 407
221 --enc->depth; 408 --enc->depth;
222} 409}
223 410
224// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
225static void 412static void
226encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
227{ 414{
228 svtype svt;
229
230 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
231 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
232 418
233 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
234 { 420 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 422 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 423 : gv_stashpv ("Types::Serialiser::Boolean", 1);
424 HV *error_stash = !CBOR_SLOW || types_error_stash
425 ? types_error_stash
426 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 428 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 430
431 HV *stash = SvSTASH (sv);
432
242 if (SvSTASH (sv) == boolean_stash) 433 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
438 else if (stash == error_stash)
439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
244 else if (SvSTASH (sv) == tagged_stash) 443 else if (stash == tagged_stash)
245 { 444 {
246 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 447
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
251 } 468 }
252 else 469 else
253 { 470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
483 {
484 dSP;
485
486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 490
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 491 PUTBACK;
492 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 493 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 494 SPAGAIN;
268 495
269 // catch this surprisingly common error 496 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 497 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 498 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 499
273 sv = POPs; 500 encode_sv (enc, POPs);
501
274 PUTBACK; 502 PUTBACK;
275 503
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 504 FREETMPS; LEAVE;
279 } 505 }
506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
507 {
508 dSP;
509
510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
513 EXTEND (SP, 2);
514 // we re-bless the reference to get overload and other niceties right
515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
516 PUSHs (sv_cbor);
517
518 PUTBACK;
519 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
520 SPAGAIN;
521
522 // catch this surprisingly common error
523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
525
526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
527 encode_uint (enc, MAJOR_ARRAY, count + 1);
528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
529
530 while (count)
531 encode_sv (enc, SP[1 - count--]);
532
533 PUTBACK;
534
535 FREETMPS; LEAVE;
536 }
280 else 537 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 538 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 540 }
285 else if (svt == SVt_PVHV) 541 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 542 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 543 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 544 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 545 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 546 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 547 encode_tag (enc, CBOR_TAG_INDIRECTION);
548 encode_sv (enc, sv);
549 }
309} 550}
310 551
311static void 552static void
312encode_nv (enc_t *enc, SV *sv) 553encode_nv (enc_t *enc, SV *sv)
313{ 554{
314 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
315 556
316 need (enc, 9); 557 need (enc, 9);
317 558
318 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 561 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
322 { 563 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
324 565
325 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 567
327 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
329 570
330 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
332 } 573 }
333 else 574 else
334 { 575 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
336 577
337 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 579
339 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
341 582
342 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
351 592
352 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
353 { 594 {
354 STRLEN len; 595 STRLEN len;
355 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
357 } 598 }
358 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 600 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
361 { 602 {
362 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 607 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 609 }
369 else if (SvROK (sv)) 610 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 616 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 619}
379 620
380static SV * 621static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
382{ 623{
383 enc_t enc; 624 enc_t enc = { 0 };
384 625
385 enc.cbor = *cbor; 626 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 630
391 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
632
633 if (cbor->flags & F_PACK_STRINGS)
634 {
635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
638 }
639
392 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
393 641
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 644
410 U8 *end; // end of input string 658 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
412 CBOR cbor; 660 CBOR cbor;
413 U32 depth; // recursion depth 661 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
415} dec_t; 666} dec_t;
416 667
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 669
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data")
420 671
421#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
422#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
423 674
424static UV 675static UV
425decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
426{ 677{
427 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
428 { 679 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 680
434 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
435 WANT (2); 685 WANT (1);
436 dec->cur += 2; 686 dec->cur += 1;
437 return dec->cur[-1]; 687 return dec->cur[-1];
438 688 }
439 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
440 WANT (3); 691 WANT (2);
441 dec->cur += 3; 692 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
444 695 }
445 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
446 WANT (5); 698 WANT (4);
447 dec->cur += 5; 699 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
452 704 }
453 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
454 WANT (9); 707 WANT (8);
455 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
456 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
460 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
464 723 }
465 default: 724 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 726
469fail: 727fail:
470 return 0; 728 return 0;
471} 729}
472 730
477{ 735{
478 AV *av = newAV (); 736 AV *av = newAV ();
479 737
480 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
481 739
482 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 741 {
484 ++dec->cur; 742 ++dec->cur;
485 743
486 for (;;) 744 for (;;)
487 { 745 {
488 WANT (1); 746 WANT (1);
489 747
490 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 749 {
492 ++dec->cur; 750 ++dec->cur;
493 break; 751 break;
494 } 752 }
495 753
496 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
497 } 755 }
498 } 756 }
499 else 757 else
500 { 758 {
501 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
502 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 762 av_fill (av, len - 1);
504 763
505 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
507 } 766 }
513 SvREFCNT_dec (av); 772 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 773 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 774 return &PL_sv_undef;
516} 775}
517 776
777static void
778decode_he (dec_t *dec, HV *hv)
779{
780 // for speed reasons, we specialcase single-string
781 // byte or utf-8 strings as keys, but only when !stringref
782
783 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
785 {
786 STRLEN len = decode_uint (dec);
787 char *key = (char *)dec->cur;
788
789 WANT (len);
790 dec->cur += len;
791
792 hv_store (hv, key, len, decode_sv (dec), 0);
793
794 return;
795 }
796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
797 {
798 STRLEN len = decode_uint (dec);
799 char *key = (char *)dec->cur;
800
801 WANT (len);
802 dec->cur += len;
803
804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
807
808 hv_store (hv, key, -len, decode_sv (dec), 0);
809
810 return;
811 }
812
813 SV *k = decode_sv (dec);
814 SV *v = decode_sv (dec);
815
816 hv_store_ent (hv, k, v, 0);
817 SvREFCNT_dec (k);
818
819fail:
820 ;
821}
822
518static SV * 823static SV *
519decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
520{ 825{
521 HV *hv = newHV (); 826 HV *hv = newHV ();
522 827
523 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
524 829
525 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 831 {
527 ++dec->cur; 832 ++dec->cur;
528 833
529 for (;;) 834 for (;;)
530 { 835 {
531 WANT (1); 836 WANT (1);
532 837
533 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 839 {
535 ++dec->cur; 840 ++dec->cur;
536 break; 841 break;
537 } 842 }
538 843
539 SV *k = decode_sv (dec); 844 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 845 }
544 } 846 }
545 else 847 else
546 { 848 {
547 int len = decode_uint (dec); 849 UV pairs = decode_uint (dec);
548 850
851 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
852
549 while (len--) 853 while (pairs--)
550 { 854 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 855 }
557 856
558 DEC_DEC_DEPTH; 857 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 858 return newRV_noinc ((SV *)hv);
560 859
567static SV * 866static SV *
568decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
569{ 868{
570 SV *sv = 0; 869 SV *sv = 0;
571 870
572 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 872 {
873 // indefinite length strings
574 ++dec->cur; 874 ++dec->cur;
575 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
576 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
577 879
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 880 for (;;)
580 { 881 {
581 WANT (1); 882 WANT (1);
582 883
583 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 886 {
585 ++dec->cur; 887 ++dec->cur;
586 break; 888 break;
587 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 892
589 sv_catsv (sv, decode_sv (dec)); 893 STRLEN len = decode_uint (dec);
894
895 WANT (len);
896 sv_catpvn (sv, dec->cur, len);
897 dec->cur += len;
590 } 898 }
591 } 899 }
592 else 900 else
593 { 901 {
594 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
595 903
596 WANT (len); 904 WANT (len);
597 sv = newSVpvn (dec->cur, len); 905 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 906 dec->cur += len;
907
908 if (ecb_expect_false (dec->stringref)
909 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
910 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 911 }
600 912
601 if (utf8) 913 if (utf8)
914 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918
602 SvUTF8_on (sv); 919 SvUTF8_on (sv);
920 }
603 921
604 return sv; 922 return sv;
605 923
606fail: 924fail:
607 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
609} 927}
610 928
611static SV * 929static SV *
612decode_tagged (dec_t *dec) 930decode_tagged (dec_t *dec)
613{ 931{
932 SV *sv = 0;
614 UV tag = decode_uint (dec); 933 UV tag = decode_uint (dec);
934
935 WANT (1);
936
937 switch (tag)
938 {
939 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 940 sv = decode_sv (dec);
941 break;
616 942
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 943 case CBOR_TAG_INDIRECTION:
944 sv = newRV_noinc (decode_sv (dec));
945 break;
946
947 case CBOR_TAG_STRINGREF_NAMESPACE:
948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
951 ENTER;
952
953 SAVESPTR (dec->stringref);
954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
955
956 sv = decode_sv (dec);
957
958 LEAVE;
959 }
960 break;
961
962 case CBOR_TAG_STRINGREF:
963 {
964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
966
967 UV idx = decode_uint (dec);
968
969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
970 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
971
972 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
973 }
974 break;
975
976 case CBOR_TAG_VALUE_SHAREABLE:
977 {
978 if (ecb_expect_false (!dec->shareable))
979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
983 sv = newSV (0);
984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
985
986 SV *osv = decode_sv (dec);
987 sv_setsv (sv, osv);
988 SvREFCNT_dec_NN (osv);
989 }
990 else
991 {
992 av_push (dec->shareable, &PL_sv_undef);
993 int idx = AvFILLp (dec->shareable);
994 sv = decode_sv (dec);
995 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
996 }
997 }
998 break;
999
1000 case CBOR_TAG_VALUE_SHAREDREF:
1001 {
1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1004
1005 UV idx = decode_uint (dec);
1006
1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1009
1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1011
1012 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1014 }
1015 break;
1016
1017 case CBOR_TAG_PERL_OBJECT:
1018 {
1019 sv = decode_sv (dec);
1020
1021 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1022 ERR ("corrupted CBOR data (non-array perl object)");
1023
1024 AV *av = (AV *)SvRV (sv);
1025 int len = av_len (av) + 1;
1026 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1027
1028 if (!stash)
1029 ERR ("cannot decode perl-object (package does not exist)");
1030
1031 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1032
1033 if (!method)
1034 ERR ("cannot decode perl-object (package does not have a THAW method)");
1035
1036 dSP;
1037
1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
1040 EXTEND (SP, len + 1);
1041 // we re-bless the reference to get overload and other niceties right
1042 PUSHs (*av_fetch (av, 0, 1));
1043 PUSHs (sv_cbor);
1044
1045 int i;
1046
1047 for (i = 1; i < len; ++i)
1048 PUSHs (*av_fetch (av, i, 1));
1049
1050 PUTBACK;
1051 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1052 SPAGAIN;
1053
1054 if (SvTRUE (ERRSV))
1055 {
1056 FREETMPS; LEAVE;
1057 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1058 }
1059
1060 SvREFCNT_dec (sv);
1061 sv = SvREFCNT_inc (POPs);
1062
1063 PUTBACK;
1064
1065 FREETMPS; LEAVE;
1066 }
1067 break;
1068
1069 default:
1070 {
1071 sv = decode_sv (dec);
1072
1073 dSP;
1074 ENTER; SAVETMPS;
1075 SAVESTACK_POS ();
1076 PUSHMARK (SP);
1077 EXTEND (SP, 2);
1078 PUSHs (newSVuv (tag));
1079 PUSHs (sv);
1080
1081 PUTBACK;
1082 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1083 SPAGAIN;
1084
1085 if (SvTRUE (ERRSV))
1086 {
1087 FREETMPS; LEAVE;
1088 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1089 }
1090
1091 if (count)
1092 {
1093 SvREFCNT_dec (sv);
1094 sv = SvREFCNT_inc (POPs);
1095 }
1096 else
1097 {
1098 AV *av = newAV ();
1099 av_push (av, newSVuv (tag));
1100 av_push (av, sv);
1101
1102 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1103 ? cbor_tagged_stash
1104 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1105 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1106 }
1107
1108 PUTBACK;
1109
1110 FREETMPS; LEAVE;
1111 }
1112 break;
1113 }
1114
618 return sv; 1115 return sv;
619 1116
620 AV *av = newAV (); 1117fail:
621 av_push (av, newSVuv (tag)); 1118 SvREFCNT_dec (sv);
622 av_push (av, sv); 1119 return &PL_sv_undef;
623
624 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
625 ? cbor_tagged_stash
626 : gv_stashpv ("CBOR::XS::Tagged" , 1);
627
628 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
629} 1120}
630 1121
631static SV * 1122static SV *
632decode_sv (dec_t *dec) 1123decode_sv (dec_t *dec)
633{ 1124{
634 WANT (1); 1125 WANT (1);
635 1126
636 switch (*dec->cur >> 5) 1127 switch (*dec->cur >> MAJOR_SHIFT)
637 { 1128 {
638 case 0: // unsigned int 1129 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
639 return newSVuv (decode_uint (dec)); 1130 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
640 case 1: // negative int 1131 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
641 return newSViv (-1 - (IV)decode_uint (dec)); 1132 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
642 case 2: // octet string 1133 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
643 return decode_str (dec, 0); 1134 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
644 case 3: // utf-8 string 1135 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
645 return decode_str (dec, 1); 1136
646 case 4: // array 1137 case MAJOR_MISC >> MAJOR_SHIFT:
647 return decode_av (dec);
648 case 5: // map
649 return decode_hv (dec);
650 case 6: // tag
651 return decode_tagged (dec);
652 case 7: // misc
653 switch (*dec->cur++ & 31) 1138 switch (*dec->cur++ & MINOR_MASK)
654 { 1139 {
655 case 20: 1140 case SIMPLE_FALSE:
656#if CBOR_SLOW 1141#if CBOR_SLOW
657 cbor_false = get_bool ("CBOR::XS::false"); 1142 types_false = get_bool ("Types::Serialiser::false");
658#endif 1143#endif
659 return newSVsv (cbor_false); 1144 return newSVsv (types_false);
660 case 21: 1145 case SIMPLE_TRUE:
661#if CBOR_SLOW 1146#if CBOR_SLOW
662 cbor_true = get_bool ("CBOR::XS::true"); 1147 types_true = get_bool ("Types::Serialiser::true");
663#endif 1148#endif
664 return newSVsv (cbor_true); 1149 return newSVsv (types_true);
665 case 22: 1150 case SIMPLE_NULL:
666 return newSVsv (&PL_sv_undef); 1151 return newSVsv (&PL_sv_undef);
1152 case SIMPLE_UNDEF:
1153#if CBOR_SLOW
1154 types_error = get_bool ("Types::Serialiser::error");
1155#endif
1156 return newSVsv (types_error);
667 1157
668 case 25: 1158 case MISC_FLOAT16:
669 { 1159 {
670 WANT (2); 1160 WANT (2);
671 1161
672 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1162 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
673 dec->cur += 2; 1163 dec->cur += 2;
674 1164
675 return newSVnv (ecb_binary16_to_float (fp)); 1165 return newSVnv (ecb_binary16_to_float (fp));
676 } 1166 }
677 1167
678 case 26: 1168 case MISC_FLOAT32:
679 { 1169 {
680 uint32_t fp; 1170 uint32_t fp;
681 WANT (4); 1171 WANT (4);
682 memcpy (&fp, dec->cur, 4); 1172 memcpy (&fp, dec->cur, 4);
683 dec->cur += 4; 1173 dec->cur += 4;
686 fp = ecb_bswap32 (fp); 1176 fp = ecb_bswap32 (fp);
687 1177
688 return newSVnv (ecb_binary32_to_float (fp)); 1178 return newSVnv (ecb_binary32_to_float (fp));
689 } 1179 }
690 1180
691 case 27: 1181 case MISC_FLOAT64:
692 { 1182 {
693 uint64_t fp; 1183 uint64_t fp;
694 WANT (8); 1184 WANT (8);
695 memcpy (&fp, dec->cur, 8); 1185 memcpy (&fp, dec->cur, 8);
696 dec->cur += 8; 1186 dec->cur += 8;
699 fp = ecb_bswap64 (fp); 1189 fp = ecb_bswap64 (fp);
700 1190
701 return newSVnv (ecb_binary64_to_double (fp)); 1191 return newSVnv (ecb_binary64_to_double (fp));
702 } 1192 }
703 1193
704 // 0..19 unassigned 1194 // 0..19 unassigned simple
705 // 24 reserved + unassigned (reserved values are not encodable) 1195 // 24 reserved + unassigned simple (reserved values are not encodable)
1196 // 28-30 unassigned misc
1197 // 31 break code
706 default: 1198 default:
707 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1199 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
708 } 1200 }
709 1201
710 break; 1202 break;
711 } 1203 }
712 1204
715} 1207}
716 1208
717static SV * 1209static SV *
718decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1210decode_cbor (SV *string, CBOR *cbor, char **offset_return)
719{ 1211{
720 dec_t dec; 1212 dec_t dec = { 0 };
721 SV *sv; 1213 SV *sv;
1214 STRLEN len;
1215 char *data = SvPVbyte (string, len);
722 1216
723 /* work around bugs in 5.10 where manipulating magic values
724 * makes perl ignore the magic in subsequent accesses.
725 * also make a copy of non-PV values, to get them into a clean
726 * state (SvPV should do that, but it's buggy, see below).
727 */
728 /*SvGETMAGIC (string);*/
729 if (SvMAGICAL (string) || !SvPOK (string))
730 string = sv_2mortal (newSVsv (string));
731
732 SvUPGRADE (string, SVt_PV);
733
734 /* work around a bug in perl 5.10, which causes SvCUR to fail an
735 * assertion with -DDEBUGGING, although SvCUR is documented to
736 * return the xpv_cur field which certainly exists after upgrading.
737 * according to nicholas clark, calling SvPOK fixes this.
738 * But it doesn't fix it, so try another workaround, call SvPV_nolen
739 * and hope for the best.
740 * Damnit, SvPV_nolen still trips over yet another assertion. This
741 * assertion business is seriously broken, try yet another workaround
742 * for the broken -DDEBUGGING.
743 */
744 {
745#ifdef DEBUGGING
746 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
747#else
748 STRLEN offset = SvCUR (string);
749#endif
750
751 if (offset > cbor->max_size && cbor->max_size) 1217 if (len > cbor->max_size && cbor->max_size)
752 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1218 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
753 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1219 (unsigned long)len, (unsigned long)cbor->max_size);
754 }
755
756 sv_utf8_downgrade (string, 0);
757 1220
758 dec.cbor = *cbor; 1221 dec.cbor = *cbor;
759 dec.cur = (U8 *)SvPVX (string); 1222 dec.cur = (U8 *)data;
760 dec.end = (U8 *)SvEND (string); 1223 dec.end = (U8 *)data + len;
761 dec.err = 0;
762 dec.depth = 0;
763 1224
764 sv = decode_sv (&dec); 1225 sv = decode_sv (&dec);
765 1226
766 if (offset_return) 1227 if (offset_return)
767 *offset_return = dec.cur; 1228 *offset_return = dec.cur;
770 if (dec.cur != dec.end && !dec.err) 1231 if (dec.cur != dec.end && !dec.err)
771 dec.err = "garbage after CBOR object"; 1232 dec.err = "garbage after CBOR object";
772 1233
773 if (dec.err) 1234 if (dec.err)
774 { 1235 {
1236 if (dec.shareable)
1237 {
1238 // need to break cyclic links, which whould all be in shareable
1239 int i;
1240 SV **svp;
1241
1242 for (i = av_len (dec.shareable) + 1; i--; )
1243 if ((svp = av_fetch (dec.shareable, i, 0)))
1244 sv_setsv (*svp, &PL_sv_undef);
1245 }
1246
775 SvREFCNT_dec (sv); 1247 SvREFCNT_dec (sv);
776 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1248 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
777 } 1249 }
778 1250
779 sv = sv_2mortal (sv); 1251 sv = sv_2mortal (sv);
780 1252
781 return sv; 1253 return sv;
782} 1254}
783 1255
1256/////////////////////////////////////////////////////////////////////////////
1257// incremental parser
1258
1259#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1260
1261// returns 0 for notyet, 1 for success or error
1262static int
1263incr_parse (CBOR *self, SV *cborstr)
1264{
1265 STRLEN cur;
1266 SvPV (cborstr, cur);
1267
1268 while (ecb_expect_true (self->incr_need <= cur))
1269 {
1270 // table of integer count bytes
1271 static I8 incr_len[MINOR_MASK + 1] = {
1272 0, 0, 0, 0, 0, 0, 0, 0,
1273 0, 0, 0, 0, 0, 0, 0, 0,
1274 0, 0, 0, 0, 0, 0, 0, 0,
1275 1, 2, 4, 8,-1,-1,-1,-2
1276 };
1277
1278 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1279 U8 m = *p & MINOR_MASK;
1280 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1281 I8 ilen = incr_len[m];
1282
1283 self->incr_need = self->incr_pos + 1;
1284
1285 if (ecb_expect_false (ilen < 0))
1286 {
1287 if (m != MINOR_INDEF)
1288 return 1; // error
1289
1290 if (*p == (MAJOR_MISC | MINOR_INDEF))
1291 {
1292 if (count >= 0)
1293 return 1; // error
1294
1295 count = 1;
1296 }
1297 else
1298 {
1299 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1300 count = -1;
1301 }
1302 }
1303 else
1304 {
1305 self->incr_need += ilen;
1306 if (ecb_expect_false (self->incr_need > cur))
1307 return 0;
1308
1309 int major = *p >> MAJOR_SHIFT;
1310
1311 switch (major)
1312 {
1313 case MAJOR_TAG >> MAJOR_SHIFT:
1314 ++count; // tags merely prefix another value
1315 break;
1316
1317 case MAJOR_BYTES >> MAJOR_SHIFT:
1318 case MAJOR_TEXT >> MAJOR_SHIFT:
1319 case MAJOR_ARRAY >> MAJOR_SHIFT:
1320 case MAJOR_MAP >> MAJOR_SHIFT:
1321 {
1322 UV len;
1323
1324 if (ecb_expect_false (ilen))
1325 {
1326 len = 0;
1327
1328 do {
1329 len = (len << 8) | *++p;
1330 } while (--ilen);
1331 }
1332 else
1333 len = m;
1334
1335 switch (major)
1336 {
1337 case MAJOR_BYTES >> MAJOR_SHIFT:
1338 case MAJOR_TEXT >> MAJOR_SHIFT:
1339 self->incr_need += len;
1340 if (ecb_expect_false (self->incr_need > cur))
1341 return 0;
1342
1343 break;
1344
1345 case MAJOR_MAP >> MAJOR_SHIFT:
1346 len <<= 1;
1347 case MAJOR_ARRAY >> MAJOR_SHIFT:
1348 if (len)
1349 {
1350 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1351 count = len + 1;
1352 }
1353 break;
1354 }
1355 }
1356 }
1357 }
1358
1359 self->incr_pos = self->incr_need;
1360
1361 if (count > 0)
1362 {
1363 while (!--count)
1364 {
1365 if (!AvFILLp (self->incr_count))
1366 return 1; // done
1367
1368 SvREFCNT_dec_NN (av_pop (self->incr_count));
1369 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1370 }
1371
1372 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1373 }
1374 }
1375
1376 return 0;
1377}
1378
1379
784///////////////////////////////////////////////////////////////////////////// 1380/////////////////////////////////////////////////////////////////////////////
785// XS interface functions 1381// XS interface functions
786 1382
787MODULE = CBOR::XS PACKAGE = CBOR::XS 1383MODULE = CBOR::XS PACKAGE = CBOR::XS
788 1384
789BOOT: 1385BOOT:
790{ 1386{
791 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1387 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
792 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
793 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1388 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
794 1389
795 cbor_true = get_bool ("CBOR::XS::true"); 1390 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
796 cbor_false = get_bool ("CBOR::XS::false"); 1391 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1392
1393 types_true = get_bool ("Types::Serialiser::true" );
1394 types_false = get_bool ("Types::Serialiser::false");
1395 types_error = get_bool ("Types::Serialiser::error");
1396
1397 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1398
1399 sv_cbor = newSVpv ("CBOR", 0);
1400 SvREADONLY_on (sv_cbor);
1401
1402 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
797} 1403}
798 1404
799PROTOTYPES: DISABLE 1405PROTOTYPES: DISABLE
800 1406
801void CLONE (...) 1407void CLONE (...)
802 CODE: 1408 CODE:
803 cbor_stash = 0; 1409 cbor_stash = 0;
804 cbor_boolean_stash = 0;
805 cbor_tagged_stash = 0; 1410 cbor_tagged_stash = 0;
1411 types_error_stash = 0;
1412 types_boolean_stash = 0;
806 1413
807void new (char *klass) 1414void new (char *klass)
808 PPCODE: 1415 PPCODE:
809{ 1416{
810 SV *pv = NEWSV (0, sizeof (CBOR)); 1417 SV *pv = NEWSV (0, sizeof (CBOR));
818 1425
819void shrink (CBOR *self, int enable = 1) 1426void shrink (CBOR *self, int enable = 1)
820 ALIAS: 1427 ALIAS:
821 shrink = F_SHRINK 1428 shrink = F_SHRINK
822 allow_unknown = F_ALLOW_UNKNOWN 1429 allow_unknown = F_ALLOW_UNKNOWN
1430 allow_sharing = F_ALLOW_SHARING
1431 allow_cycles = F_ALLOW_CYCLES
1432 pack_strings = F_PACK_STRINGS
1433 text_keys = F_TEXT_KEYS
1434 text_strings = F_TEXT_STRINGS
1435 validate_utf8 = F_VALIDATE_UTF8
823 PPCODE: 1436 PPCODE:
824{ 1437{
825 if (enable) 1438 if (enable)
826 self->flags |= ix; 1439 self->flags |= ix;
827 else 1440 else
832 1445
833void get_shrink (CBOR *self) 1446void get_shrink (CBOR *self)
834 ALIAS: 1447 ALIAS:
835 get_shrink = F_SHRINK 1448 get_shrink = F_SHRINK
836 get_allow_unknown = F_ALLOW_UNKNOWN 1449 get_allow_unknown = F_ALLOW_UNKNOWN
1450 get_allow_sharing = F_ALLOW_SHARING
1451 get_allow_cycles = F_ALLOW_CYCLES
1452 get_pack_strings = F_PACK_STRINGS
1453 get_text_keys = F_TEXT_KEYS
1454 get_text_strings = F_TEXT_STRINGS
1455 get_validate_utf8 = F_VALIDATE_UTF8
837 PPCODE: 1456 PPCODE:
838 XPUSHs (boolSV (self->flags & ix)); 1457 XPUSHs (boolSV (self->flags & ix));
839 1458
840void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1459void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
841 PPCODE: 1460 PPCODE:
857 CODE: 1476 CODE:
858 RETVAL = self->max_size; 1477 RETVAL = self->max_size;
859 OUTPUT: 1478 OUTPUT:
860 RETVAL 1479 RETVAL
861 1480
862#if 0 //TODO 1481void filter (CBOR *self, SV *filter = 0)
863
864void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
865 PPCODE: 1482 PPCODE:
866{
867 SvREFCNT_dec (self->cb_object); 1483 SvREFCNT_dec (self->filter);
868 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1484 self->filter = filter ? newSVsv (filter) : filter;
869
870 XPUSHs (ST (0)); 1485 XPUSHs (ST (0));
871}
872 1486
873void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1487SV *get_filter (CBOR *self)
874 PPCODE: 1488 CODE:
875{ 1489 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
876 if (!self->cb_sk_object) 1490 OUTPUT:
877 self->cb_sk_object = newHV (); 1491 RETVAL
878
879 if (SvOK (cb))
880 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
881 else
882 {
883 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
884
885 if (!HvKEYS (self->cb_sk_object))
886 {
887 SvREFCNT_dec (self->cb_sk_object);
888 self->cb_sk_object = 0;
889 }
890 }
891
892 XPUSHs (ST (0));
893}
894
895#endif
896 1492
897void encode (CBOR *self, SV *scalar) 1493void encode (CBOR *self, SV *scalar)
898 PPCODE: 1494 PPCODE:
899 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1495 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
900 XPUSHs (scalar); 1496 XPUSHs (scalar);
913 EXTEND (SP, 2); 1509 EXTEND (SP, 2);
914 PUSHs (sv); 1510 PUSHs (sv);
915 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1511 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
916} 1512}
917 1513
918#if 0 1514void incr_parse (CBOR *self, SV *cborstr)
1515 ALIAS:
1516 incr_parse_multiple = 1
1517 PPCODE:
1518{
1519 if (SvUTF8 (cborstr))
1520 sv_utf8_downgrade (cborstr, 0);
1521
1522 if (!self->incr_count)
1523 {
1524 self->incr_count = newAV ();
1525 self->incr_pos = 0;
1526 self->incr_need = 1;
1527
1528 av_push (self->incr_count, newSViv (1));
1529 }
1530
1531 do
1532 {
1533 if (!incr_parse (self, cborstr))
1534 {
1535 if (self->incr_need > self->max_size && self->max_size)
1536 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1537 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1538
1539 break;
1540 }
1541
1542 SV *sv;
1543 char *offset;
1544
1545 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1546 XPUSHs (sv);
1547
1548 sv_chop (cborstr, offset);
1549
1550 av_clear (self->incr_count);
1551 av_push (self->incr_count, newSViv (1));
1552
1553 self->incr_pos = 0;
1554 self->incr_need = self->incr_pos + 1;
1555 }
1556 while (ix);
1557}
1558
1559void incr_reset (CBOR *self)
1560 CODE:
1561{
1562 SvREFCNT_dec (self->incr_count);
1563 self->incr_count = 0;
1564}
919 1565
920void DESTROY (CBOR *self) 1566void DESTROY (CBOR *self)
921 CODE: 1567 PPCODE:
922 SvREFCNT_dec (self->cb_sk_object); 1568 cbor_free (self);
923 SvREFCNT_dec (self->cb_object);
924
925#endif
926 1569
927PROTOTYPES: ENABLE 1570PROTOTYPES: ENABLE
928 1571
929void encode_cbor (SV *scalar) 1572void encode_cbor (SV *scalar)
1573 ALIAS:
1574 encode_cbor = 0
1575 encode_cbor_sharing = F_ALLOW_SHARING
930 PPCODE: 1576 PPCODE:
931{ 1577{
932 CBOR cbor; 1578 CBOR cbor;
933 cbor_init (&cbor); 1579 cbor_init (&cbor);
1580 cbor.flags |= ix;
934 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1581 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
935 XPUSHs (scalar); 1582 XPUSHs (scalar);
936} 1583}
937 1584
938void decode_cbor (SV *cborstr) 1585void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines