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

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.53 by root, Sun Apr 24 19:51:41 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14// known tags, rfc7049 15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
23 85
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 90
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 96
35 CBOR_TAG_MAGIC = 55799 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
36}; 98};
37 99
38#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
39#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_UTF8_STRINGS 0x00000020UL
106#define F_UTF8_KEYS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
40 108
41#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
42 110
43#define SB do { 111#define SB do {
44#define SE } while (0) 112#define SE } while (0)
55#else 123#else
56# define CBOR_SLOW 0 124# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
58#endif 126#endif
59 127
60static 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::
61static SV *cbor_true, *cbor_false; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 130
63typedef struct { 131typedef struct {
64 U32 flags; 132 U32 flags;
65 U32 max_depth; 133 U32 max_depth;
66 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.
67} CBOR; 141} CBOR;
68 142
69ecb_inline void 143ecb_inline void
70cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
71{ 145{
72 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
73 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);
74} 155}
75 156
76///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
77// utility functions 158// utility functions
78 159
100 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 182#endif
102 } 183 }
103} 184}
104 185
105///////////////////////////////////////////////////////////////////////////// 186// minimum length of a string to be registered for stringref
106// fp hell 187ecb_inline int
107 188minimum_string_length (UV idx)
108//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}
109 200
110///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
111// encoder 202// encoder
112 203
113// structure used for encoding CBOR 204// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
118 SV *sv; // result scalar 209 SV *sv; // result scalar
119 CBOR cbor; 210 CBOR cbor;
120 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;
121} enc_t; 216} enc_t;
122 217
123ecb_inline void 218ecb_inline void
124need (enc_t *enc, STRLEN len) 219need (enc_t *enc, STRLEN len)
125{ 220{
142static void 237static void
143encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
144{ 239{
145 need (enc, 9); 240 need (enc, 9);
146 241
147 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
149 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
150 { 245 {
151 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 247 *enc->cur++ = len;
153 } 248 }
154 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
155 { 250 {
156 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 253 *enc->cur++ = len;
159 } 254 }
160 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
161 { 256 {
162 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 261 *enc->cur++ = len;
167 } 262 }
168 else 263 else
169 { 264 {
170 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 273 *enc->cur++ = len;
179 } 274 }
180} 275}
181 276
182static 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
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 286{
185 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);
186 need (enc, len); 317 need (enc, len);
187 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
188 enc->cur += len; 319 enc->cur += len;
189} 320}
190 321
322ecb_inline
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
191static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
192 348
193static void 349static void
194encode_av (enc_t *enc, AV *av) 350encode_av (enc_t *enc, AV *av)
195{ 351{
198 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
200 356
201 ++enc->depth; 357 ++enc->depth;
202 358
203 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 360
361 if (SvMAGICAL (av))
205 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
206 { 363 {
207 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
209 } 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 }
210 373
211 --enc->depth; 374 --enc->depth;
212} 375}
213 376
214static void 377static void
223 386
224 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
226 389
227 if (mg) 390 if (mg)
228 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 392 else
230 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
231 394
232 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
233 { 396 {
234 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
236 else 399 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 401
239 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));
240 } 403 }
241 404
242 if (mg) 405 if (mg)
243 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 407
245 --enc->depth; 408 --enc->depth;
246} 409}
247 410
248// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
249static void 412static void
250encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
251{ 414{
252 svtype svt;
253
254 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
416
255 svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
256 418
257 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
258 { 420 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 421 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 422 ? types_boolean_stash
261 : 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);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 428 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 430
431 HV *stash = SvSTASH (sv);
432
266 if (SvSTASH (sv) == boolean_stash) 433 if (stash == boolean_stash)
267 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 }
268 else if (SvSTASH (sv) == tagged_stash) 443 else if (stash == tagged_stash)
269 { 444 {
270 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 447
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 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;
275 } 468 }
276 else 469 else
277 { 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);
278 // 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
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))); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 490
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 491 PUTBACK;
492 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 493 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 494 SPAGAIN;
292 495
293 // catch this surprisingly common error 496 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 497 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))); 498 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
296 499
297 sv = POPs; 500 encode_sv (enc, POPs);
501
298 PUTBACK; 502 PUTBACK;
299 503
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 504 FREETMPS; LEAVE;
303 } 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 }
304 else 537 else
305 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",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 540 }
309 else if (svt == SVt_PVHV) 541 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 542 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 543 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 544 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 545 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 546 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 547 encode_tag (enc, CBOR_TAG_INDIRECTION);
548 encode_sv (enc, sv);
549 }
333} 550}
334 551
335static void 552static void
336encode_nv (enc_t *enc, SV *sv) 553encode_nv (enc_t *enc, SV *sv)
337{ 554{
338 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
339 556
340 need (enc, 9); 557 need (enc, 9);
341 558
342 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 561 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
346 { 563 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
348 565
349 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 567
351 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
353 570
354 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
356 } 573 }
357 else 574 else
358 { 575 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
360 577
361 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 579
363 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
365 582
366 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
375 592
376 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
377 { 594 {
378 STRLEN len; 595 STRLEN len;
379 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len);
381 } 598 }
382 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 600 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
385 { 602 {
386 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 607 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 609 }
393 else if (SvROK (sv)) 610 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 616 else
400 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",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 619}
403 620
404static SV * 621static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
406{ 623{
407 enc_t enc; 624 enc_t enc = { 0 };
408 625
409 enc.cbor = *cbor; 626 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 630
415 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
416 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
417 641
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *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
420 644
434 U8 *end; // end of input string 658 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
436 CBOR cbor; 660 CBOR cbor;
437 U32 depth; // recursion depth 661 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
439} dec_t; 666} dec_t;
440 667
441#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
442 669
443#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 (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
446#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
447 674
448static UV 675static UV
449decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
450{ 677{
451 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
452 { 679 ++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;
457 680
458 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
459 WANT (2); 685 WANT (1);
460 dec->cur += 2; 686 dec->cur += 1;
461 return dec->cur[-1]; 687 return dec->cur[-1];
462 688 }
463 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
464 WANT (3); 691 WANT (2);
465 dec->cur += 3; 692 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
468 695 }
469 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
470 WANT (5); 698 WANT (4);
471 dec->cur += 5; 699 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
476 704 }
477 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
478 WANT (9); 707 WANT (8);
479 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
480 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
484 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
488 723 }
489 default: 724 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 726
493fail: 727fail:
494 return 0; 728 return 0;
495} 729}
496 730
501{ 735{
502 AV *av = newAV (); 736 AV *av = newAV ();
503 737
504 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
505 739
506 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 741 {
508 ++dec->cur; 742 ++dec->cur;
509 743
510 for (;;) 744 for (;;)
511 { 745 {
512 WANT (1); 746 WANT (1);
513 747
514 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 749 {
516 ++dec->cur; 750 ++dec->cur;
517 break; 751 break;
518 } 752 }
519 753
522 } 756 }
523 else 757 else
524 { 758 {
525 int i, len = decode_uint (dec); 759 int i, len = decode_uint (dec);
526 760
761 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); 762 av_fill (av, len - 1);
528 763
529 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
531 } 766 }
537 SvREFCNT_dec (av); 772 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 773 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 774 return &PL_sv_undef;
540} 775}
541 776
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 I32 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 I32 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
542static SV * 823static SV *
543decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
544{ 825{
545 HV *hv = newHV (); 826 HV *hv = newHV ();
546 827
547 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
548 829
549 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 831 {
551 ++dec->cur; 832 ++dec->cur;
552 833
553 for (;;) 834 for (;;)
554 { 835 {
555 WANT (1); 836 WANT (1);
556 837
557 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 839 {
559 ++dec->cur; 840 ++dec->cur;
560 break; 841 break;
561 } 842 }
562 843
563 SV *k = decode_sv (dec); 844 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 845 }
568 } 846 }
569 else 847 else
570 { 848 {
571 int len = decode_uint (dec); 849 int pairs = decode_uint (dec);
572 850
573 while (len--) 851 while (pairs--)
574 { 852 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 } 853 }
581 854
582 DEC_DEC_DEPTH; 855 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 856 return newRV_noinc ((SV *)hv);
584 857
591static SV * 864static SV *
592decode_str (dec_t *dec, int utf8) 865decode_str (dec_t *dec, int utf8)
593{ 866{
594 SV *sv = 0; 867 SV *sv = 0;
595 868
596 if ((*dec->cur & 31) == 31) 869 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 870 {
871 // indefinite length strings
598 ++dec->cur; 872 ++dec->cur;
599 873
874 U8 major = *dec->cur & MAJOR_MISC;
875
600 sv = newSVpvn ("", 0); 876 sv = newSVpvn ("", 0);
601 877
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 878 for (;;)
604 { 879 {
605 WANT (1); 880 WANT (1);
606 881
607 if (*dec->cur == (0xe0 | 31)) 882 if ((*dec->cur - major) > LENGTH_EXT8)
883 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 884 {
609 ++dec->cur; 885 ++dec->cur;
610 break; 886 break;
611 } 887 }
888 else
889 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 890
613 sv_catsv (sv, decode_sv (dec)); 891 STRLEN len = decode_uint (dec);
892
893 WANT (len);
894 sv_catpvn (sv, dec->cur, len);
895 dec->cur += len;
614 } 896 }
615 } 897 }
616 else 898 else
617 { 899 {
618 STRLEN len = decode_uint (dec); 900 STRLEN len = decode_uint (dec);
619 901
620 WANT (len); 902 WANT (len);
621 sv = newSVpvn (dec->cur, len); 903 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 904 dec->cur += len;
905
906 if (ecb_expect_false (dec->stringref)
907 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
908 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 909 }
624 910
625 if (utf8) 911 if (utf8)
912 {
913 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
914 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
915 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
916
626 SvUTF8_on (sv); 917 SvUTF8_on (sv);
918 }
627 919
628 return sv; 920 return sv;
629 921
630fail: 922fail:
631 SvREFCNT_dec (sv); 923 SvREFCNT_dec (sv);
633} 925}
634 926
635static SV * 927static SV *
636decode_tagged (dec_t *dec) 928decode_tagged (dec_t *dec)
637{ 929{
930 SV *sv = 0;
638 UV tag = decode_uint (dec); 931 UV tag = decode_uint (dec);
932
933 WANT (1);
934
935 switch (tag)
936 {
937 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 938 sv = decode_sv (dec);
939 break;
640 940
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 941 case CBOR_TAG_INDIRECTION:
942 sv = newRV_noinc (decode_sv (dec));
943 break;
944
945 case CBOR_TAG_STRINGREF_NAMESPACE:
946 {
947 ENTER; SAVETMPS;
948
949 SAVESPTR (dec->stringref);
950 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
951
952 sv = decode_sv (dec);
953
954 FREETMPS; LEAVE;
955 }
956 break;
957
958 case CBOR_TAG_STRINGREF:
959 {
960 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
961 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
962
963 UV idx = decode_uint (dec);
964
965 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
966 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
967
968 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
969 }
970 break;
971
972 case CBOR_TAG_VALUE_SHAREABLE:
973 {
974 if (ecb_expect_false (!dec->shareable))
975 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
976
977 if (dec->cbor.flags & F_ALLOW_CYCLES)
978 {
979 sv = newSV (0);
980 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
981
982 SV *osv = decode_sv (dec);
983 sv_setsv (sv, osv);
984 SvREFCNT_dec_NN (osv);
985 }
986 else
987 {
988 av_push (dec->shareable, &PL_sv_undef);
989 int idx = AvFILLp (dec->shareable);
990 sv = decode_sv (dec);
991 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
992 }
993 }
994 break;
995
996 case CBOR_TAG_VALUE_SHAREDREF:
997 {
998 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
999 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1000
1001 UV idx = decode_uint (dec);
1002
1003 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1004 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1005
1006 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1007
1008 if (sv == &PL_sv_undef)
1009 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1010 }
1011 break;
1012
1013 case CBOR_TAG_PERL_OBJECT:
1014 {
1015 sv = decode_sv (dec);
1016
1017 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1018 ERR ("corrupted CBOR data (non-array perl object)");
1019
1020 AV *av = (AV *)SvRV (sv);
1021 int len = av_len (av) + 1;
1022 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1023
1024 if (!stash)
1025 ERR ("cannot decode perl-object (package does not exist)");
1026
1027 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1028
1029 if (!method)
1030 ERR ("cannot decode perl-object (package does not have a THAW method)");
1031
1032 dSP;
1033
1034 ENTER; SAVETMPS;
1035 PUSHMARK (SP);
1036 EXTEND (SP, len + 1);
1037 // we re-bless the reference to get overload and other niceties right
1038 PUSHs (*av_fetch (av, 0, 1));
1039 PUSHs (sv_cbor);
1040
1041 int i;
1042
1043 for (i = 1; i < len; ++i)
1044 PUSHs (*av_fetch (av, i, 1));
1045
1046 PUTBACK;
1047 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1048 SPAGAIN;
1049
1050 if (SvTRUE (ERRSV))
1051 {
1052 FREETMPS; LEAVE;
1053 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1054 }
1055
1056 SvREFCNT_dec (sv);
1057 sv = SvREFCNT_inc (POPs);
1058
1059 PUTBACK;
1060
1061 FREETMPS; LEAVE;
1062 }
1063 break;
1064
1065 default:
1066 {
1067 sv = decode_sv (dec);
1068
1069 dSP;
1070 ENTER; SAVETMPS;
1071 SAVESTACK_POS ();
1072 PUSHMARK (SP);
1073 EXTEND (SP, 2);
1074 PUSHs (newSVuv (tag));
1075 PUSHs (sv);
1076
1077 PUTBACK;
1078 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1079 SPAGAIN;
1080
1081 if (SvTRUE (ERRSV))
1082 {
1083 FREETMPS; LEAVE;
1084 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1085 }
1086
1087 if (count)
1088 {
1089 SvREFCNT_dec (sv);
1090 sv = SvREFCNT_inc (POPs);
1091 }
1092 else
1093 {
1094 AV *av = newAV ();
1095 av_push (av, newSVuv (tag));
1096 av_push (av, sv);
1097
1098 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1099 ? cbor_tagged_stash
1100 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1101 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1102 }
1103
1104 PUTBACK;
1105
1106 FREETMPS; LEAVE;
1107 }
1108 break;
1109 }
1110
642 return sv; 1111 return sv;
643 1112
644 AV *av = newAV (); 1113fail:
645 av_push (av, newSVuv (tag)); 1114 SvREFCNT_dec (sv);
646 av_push (av, sv); 1115 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} 1116}
654 1117
655static SV * 1118static SV *
656decode_sv (dec_t *dec) 1119decode_sv (dec_t *dec)
657{ 1120{
658 WANT (1); 1121 WANT (1);
659 1122
660 switch (*dec->cur >> 5) 1123 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1124 {
662 case 0: // unsigned int 1125 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1126 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1127 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1128 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1129 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1130 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1131 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1132
670 case 4: // array 1133 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) 1134 switch (*dec->cur++ & MINOR_MASK)
678 { 1135 {
679 case 20: 1136 case SIMPLE_FALSE:
680#if CBOR_SLOW 1137#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1138 types_false = get_bool ("Types::Serialiser::false");
682#endif 1139#endif
683 return newSVsv (cbor_false); 1140 return newSVsv (types_false);
684 case 21: 1141 case SIMPLE_TRUE:
685#if CBOR_SLOW 1142#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1143 types_true = get_bool ("Types::Serialiser::true");
687#endif 1144#endif
688 return newSVsv (cbor_true); 1145 return newSVsv (types_true);
689 case 22: 1146 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1147 return newSVsv (&PL_sv_undef);
1148 case SIMPLE_UNDEF:
1149#if CBOR_SLOW
1150 types_error = get_bool ("Types::Serialiser::error");
1151#endif
1152 return newSVsv (types_error);
691 1153
692 case 25: 1154 case MISC_FLOAT16:
693 { 1155 {
694 WANT (2); 1156 WANT (2);
695 1157
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1158 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1159 dec->cur += 2;
698 1160
699 return newSVnv (ecb_binary16_to_float (fp)); 1161 return newSVnv (ecb_binary16_to_float (fp));
700 } 1162 }
701 1163
702 case 26: 1164 case MISC_FLOAT32:
703 { 1165 {
704 uint32_t fp; 1166 uint32_t fp;
705 WANT (4); 1167 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1168 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1169 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1172 fp = ecb_bswap32 (fp);
711 1173
712 return newSVnv (ecb_binary32_to_float (fp)); 1174 return newSVnv (ecb_binary32_to_float (fp));
713 } 1175 }
714 1176
715 case 27: 1177 case MISC_FLOAT64:
716 { 1178 {
717 uint64_t fp; 1179 uint64_t fp;
718 WANT (8); 1180 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1181 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1182 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1185 fp = ecb_bswap64 (fp);
724 1186
725 return newSVnv (ecb_binary64_to_double (fp)); 1187 return newSVnv (ecb_binary64_to_double (fp));
726 } 1188 }
727 1189
728 // 0..19 unassigned 1190 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1191 // 24 reserved + unassigned simple (reserved values are not encodable)
1192 // 28-30 unassigned misc
1193 // 31 break code
730 default: 1194 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1195 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
732 } 1196 }
733 1197
734 break; 1198 break;
735 } 1199 }
736 1200
739} 1203}
740 1204
741static SV * 1205static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1206decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1207{
744 dec_t dec; 1208 dec_t dec = { 0 };
745 SV *sv; 1209 SV *sv;
1210 STRLEN len;
1211 char *data = SvPVbyte (string, len);
746 1212
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) 1213 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", 1214 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); 1215 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1216
782 dec.cbor = *cbor; 1217 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1218 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1219 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1220
788 sv = decode_sv (&dec); 1221 sv = decode_sv (&dec);
789 1222
790 if (offset_return) 1223 if (offset_return)
791 *offset_return = dec.cur; 1224 *offset_return = dec.cur;
794 if (dec.cur != dec.end && !dec.err) 1227 if (dec.cur != dec.end && !dec.err)
795 dec.err = "garbage after CBOR object"; 1228 dec.err = "garbage after CBOR object";
796 1229
797 if (dec.err) 1230 if (dec.err)
798 { 1231 {
1232 if (dec.shareable)
1233 {
1234 // need to break cyclic links, which whould all be in shareable
1235 int i;
1236 SV **svp;
1237
1238 for (i = av_len (dec.shareable) + 1; i--; )
1239 if ((svp = av_fetch (dec.shareable, i, 0)))
1240 sv_setsv (*svp, &PL_sv_undef);
1241 }
1242
799 SvREFCNT_dec (sv); 1243 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1244 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1245 }
802 1246
803 sv = sv_2mortal (sv); 1247 sv = sv_2mortal (sv);
804 1248
805 return sv; 1249 return sv;
806} 1250}
807 1251
1252/////////////////////////////////////////////////////////////////////////////
1253// incremental parser
1254
1255#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1256
1257// returns 0 for notyet, 1 for success or error
1258static int
1259incr_parse (CBOR *self, SV *cborstr)
1260{
1261 STRLEN cur;
1262 SvPV (cborstr, cur);
1263
1264 while (ecb_expect_true (self->incr_need <= cur))
1265 {
1266 // table of integer count bytes
1267 static I8 incr_len[MINOR_MASK + 1] = {
1268 0, 0, 0, 0, 0, 0, 0, 0,
1269 0, 0, 0, 0, 0, 0, 0, 0,
1270 0, 0, 0, 0, 0, 0, 0, 0,
1271 1, 2, 4, 8,-1,-1,-1,-2
1272 };
1273
1274 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1275 U8 m = *p & MINOR_MASK;
1276 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1277 I8 ilen = incr_len[m];
1278
1279 self->incr_need = self->incr_pos + 1;
1280
1281 if (ecb_expect_false (ilen < 0))
1282 {
1283 if (m != MINOR_INDEF)
1284 return 1; // error
1285
1286 if (*p == (MAJOR_MISC | MINOR_INDEF))
1287 {
1288 if (count >= 0)
1289 return 1; // error
1290
1291 count = 1;
1292 }
1293 else
1294 {
1295 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1296 count = -1;
1297 }
1298 }
1299 else
1300 {
1301 self->incr_need += ilen;
1302 if (ecb_expect_false (self->incr_need > cur))
1303 return 0;
1304
1305 int major = *p >> MAJOR_SHIFT;
1306
1307 switch (major)
1308 {
1309 case MAJOR_TAG >> MAJOR_SHIFT:
1310 ++count; // tags merely prefix another value
1311 break;
1312
1313 case MAJOR_BYTES >> MAJOR_SHIFT:
1314 case MAJOR_TEXT >> MAJOR_SHIFT:
1315 case MAJOR_ARRAY >> MAJOR_SHIFT:
1316 case MAJOR_MAP >> MAJOR_SHIFT:
1317 {
1318 UV len;
1319
1320 if (ecb_expect_false (ilen))
1321 {
1322 len = 0;
1323
1324 do {
1325 len = (len << 8) | *++p;
1326 } while (--ilen);
1327 }
1328 else
1329 len = m;
1330
1331 switch (major)
1332 {
1333 case MAJOR_BYTES >> MAJOR_SHIFT:
1334 case MAJOR_TEXT >> MAJOR_SHIFT:
1335 self->incr_need += len;
1336 if (ecb_expect_false (self->incr_need > cur))
1337 return 0;
1338
1339 break;
1340
1341 case MAJOR_MAP >> MAJOR_SHIFT:
1342 len <<= 1;
1343 case MAJOR_ARRAY >> MAJOR_SHIFT:
1344 if (len)
1345 {
1346 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1347 count = len + 1;
1348 }
1349 break;
1350 }
1351 }
1352 }
1353 }
1354
1355 self->incr_pos = self->incr_need;
1356
1357 if (count > 0)
1358 {
1359 while (!--count)
1360 {
1361 if (!AvFILLp (self->incr_count))
1362 return 1; // done
1363
1364 SvREFCNT_dec_NN (av_pop (self->incr_count));
1365 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1366 }
1367
1368 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1369 }
1370 }
1371
1372 return 0;
1373}
1374
1375
808///////////////////////////////////////////////////////////////////////////// 1376/////////////////////////////////////////////////////////////////////////////
809// XS interface functions 1377// XS interface functions
810 1378
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1379MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1380
813BOOT: 1381BOOT:
814{ 1382{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1383 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); 1384 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1385
819 cbor_true = get_bool ("CBOR::XS::true"); 1386 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1387 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1388
1389 types_true = get_bool ("Types::Serialiser::true" );
1390 types_false = get_bool ("Types::Serialiser::false");
1391 types_error = get_bool ("Types::Serialiser::error");
1392
1393 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1394
1395 sv_cbor = newSVpv ("CBOR", 0);
1396 SvREADONLY_on (sv_cbor);
821} 1397}
822 1398
823PROTOTYPES: DISABLE 1399PROTOTYPES: DISABLE
824 1400
825void CLONE (...) 1401void CLONE (...)
826 CODE: 1402 CODE:
827 cbor_stash = 0; 1403 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1404 cbor_tagged_stash = 0;
1405 types_error_stash = 0;
1406 types_boolean_stash = 0;
830 1407
831void new (char *klass) 1408void new (char *klass)
832 PPCODE: 1409 PPCODE:
833{ 1410{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1411 SV *pv = NEWSV (0, sizeof (CBOR));
842 1419
843void shrink (CBOR *self, int enable = 1) 1420void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1421 ALIAS:
845 shrink = F_SHRINK 1422 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1423 allow_unknown = F_ALLOW_UNKNOWN
1424 allow_sharing = F_ALLOW_SHARING
1425 allow_cycles = F_ALLOW_CYCLES
1426 pack_strings = F_PACK_STRINGS
1427 utf8_strings = F_UTF8_STRINGS
1428 validate_utf8 = F_VALIDATE_UTF8
847 PPCODE: 1429 PPCODE:
848{ 1430{
849 if (enable) 1431 if (enable)
850 self->flags |= ix; 1432 self->flags |= ix;
851 else 1433 else
856 1438
857void get_shrink (CBOR *self) 1439void get_shrink (CBOR *self)
858 ALIAS: 1440 ALIAS:
859 get_shrink = F_SHRINK 1441 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1442 get_allow_unknown = F_ALLOW_UNKNOWN
1443 get_allow_sharing = F_ALLOW_SHARING
1444 get_allow_cycles = F_ALLOW_CYCLES
1445 get_pack_strings = F_PACK_STRINGS
1446 get_validate_utf8 = F_VALIDATE_UTF8
861 PPCODE: 1447 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1448 XPUSHs (boolSV (self->flags & ix));
863 1449
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1450void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1451 PPCODE:
881 CODE: 1467 CODE:
882 RETVAL = self->max_size; 1468 RETVAL = self->max_size;
883 OUTPUT: 1469 OUTPUT:
884 RETVAL 1470 RETVAL
885 1471
886#if 0 //TODO 1472void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1473 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1474 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1475 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1476 XPUSHs (ST (0));
895}
896 1477
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1478SV *get_filter (CBOR *self)
898 PPCODE: 1479 CODE:
899{ 1480 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1481 OUTPUT:
901 self->cb_sk_object = newHV (); 1482 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 1483
921void encode (CBOR *self, SV *scalar) 1484void encode (CBOR *self, SV *scalar)
922 PPCODE: 1485 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1486 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1487 XPUSHs (scalar);
937 EXTEND (SP, 2); 1500 EXTEND (SP, 2);
938 PUSHs (sv); 1501 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1502 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1503}
941 1504
942#if 0 1505void incr_parse (CBOR *self, SV *cborstr)
1506 ALIAS:
1507 incr_parse_multiple = 1
1508 PPCODE:
1509{
1510 if (SvUTF8 (cborstr))
1511 sv_utf8_downgrade (cborstr, 0);
1512
1513 if (!self->incr_count)
1514 {
1515 self->incr_count = newAV ();
1516 self->incr_pos = 0;
1517 self->incr_need = 1;
1518
1519 av_push (self->incr_count, newSViv (1));
1520 }
1521
1522 do
1523 {
1524 if (!incr_parse (self, cborstr))
1525 {
1526 if (self->incr_need > self->max_size && self->max_size)
1527 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1528 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1529
1530 break;
1531 }
1532
1533 SV *sv;
1534 char *offset;
1535
1536 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1537 XPUSHs (sv);
1538
1539 sv_chop (cborstr, offset);
1540
1541 av_clear (self->incr_count);
1542 av_push (self->incr_count, newSViv (1));
1543
1544 self->incr_pos = 0;
1545 self->incr_need = self->incr_pos + 1;
1546 }
1547 while (ix);
1548}
1549
1550void incr_reset (CBOR *self)
1551 CODE:
1552{
1553 SvREFCNT_dec (self->incr_count);
1554 self->incr_count = 0;
1555}
943 1556
944void DESTROY (CBOR *self) 1557void DESTROY (CBOR *self)
945 CODE: 1558 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1559 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1560
951PROTOTYPES: ENABLE 1561PROTOTYPES: ENABLE
952 1562
953void encode_cbor (SV *scalar) 1563void encode_cbor (SV *scalar)
1564 ALIAS:
1565 encode_cbor = 0
1566 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1567 PPCODE:
955{ 1568{
956 CBOR cbor; 1569 CBOR cbor;
957 cbor_init (&cbor); 1570 cbor_init (&cbor);
1571 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1572 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1573 XPUSHs (scalar);
960} 1574}
961 1575
962void decode_cbor (SV *cborstr) 1576void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines