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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.51 by root, Sun Apr 24 13:15:19 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"
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};
13 65
14// known tags 66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
17 // inofficial extensions (pending iana registration) 69 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 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
20 77
21 // rfc7049 78 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
28 85
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 90
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 96
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 98};
42 99
43#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
44#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_VALIDATE_UTF8 0x00000040UL
45 107
46#define INIT_SIZE 32 // initial scalar size to be allocated 108#define INIT_SIZE 32 // initial scalar size to be allocated
47 109
48#define SB do { 110#define SB do {
49#define SE } while (0) 111#define SE } while (0)
61# define CBOR_SLOW 0 123# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 124# define CBOR_STASH cbor_stash
63#endif 125#endif
64 126
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 127static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 128static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 129
68typedef struct { 130typedef struct {
69 U32 flags; 131 U32 flags;
70 U32 max_depth; 132 U32 max_depth;
71 STRLEN max_size; 133 STRLEN max_size;
134 SV *filter;
135
136 // for the incremental parser
137 STRLEN incr_pos; // the current offset into the text
138 STRLEN incr_need; // minimum bytes needed to decode
139 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 140} CBOR;
73 141
74ecb_inline void 142ecb_inline void
75cbor_init (CBOR *cbor) 143cbor_init (CBOR *cbor)
76{ 144{
77 Zero (cbor, 1, CBOR); 145 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 146 cbor->max_depth = 512;
147}
148
149ecb_inline void
150cbor_free (CBOR *cbor)
151{
152 SvREFCNT_dec (cbor->filter);
153 SvREFCNT_dec (cbor->incr_count);
79} 154}
80 155
81///////////////////////////////////////////////////////////////////////////// 156/////////////////////////////////////////////////////////////////////////////
82// utility functions 157// utility functions
83 158
105 SvPV_renew (sv, SvCUR (sv) + 1); 180 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 181#endif
107 } 182 }
108} 183}
109 184
110///////////////////////////////////////////////////////////////////////////// 185// minimum length of a string to be registered for stringref
111// fp hell 186ecb_inline int
112 187minimum_string_length (UV idx)
113//TODO 188{
189 return idx > 23
190 ? idx > 0xffU
191 ? idx > 0xffffU
192 ? idx > 0xffffffffU
193 ? 11
194 : 7
195 : 5
196 : 4
197 : 3;
198}
114 199
115///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
116// encoder 201// encoder
117 202
118// structure used for encoding CBOR 203// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
123 SV *sv; // result scalar 208 SV *sv; // result scalar
124 CBOR cbor; 209 CBOR cbor;
125 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
126} enc_t; 215} enc_t;
127 216
128ecb_inline void 217ecb_inline void
129need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
130{ 219{
147static void 236static void
148encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
149{ 238{
150 need (enc, 9); 239 need (enc, 9);
151 240
152 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
154 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
155 { 244 {
156 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 246 *enc->cur++ = len;
158 } 247 }
159 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
160 { 249 {
161 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 252 *enc->cur++ = len;
164 } 253 }
165 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
166 { 255 {
167 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 260 *enc->cur++ = len;
172 } 261 }
173 else 262 else
174 { 263 {
175 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 272 *enc->cur++ = len;
184 } 273 }
185} 274}
186 275
187static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 284{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS))
286 if (!utf8)
287 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l);
295 return;
296 }
297
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 299 need (enc, len);
192 memcpy (enc->cur, str, len); 300 memcpy (enc->cur, str, len);
193 enc->cur += len; 301 enc->cur += len;
194} 302}
195 303
304static void
305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
306{
307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310
311 if (SvOK (*svp))
312 {
313 // already registered, use stringref
314 encode_tag (enc, CBOR_TAG_STRINGREF);
315 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
316 return;
317 }
318 else if (len >= minimum_string_length (enc->stringref_idx))
319 {
320 // register only
321 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx;
323 }
324 }
325
326 encode_str (enc, utf8, str, len);
327}
328
196static void encode_sv (enc_t *enc, SV *sv); 329static void encode_sv (enc_t *enc, SV *sv);
197 330
198static void 331static void
199encode_av (enc_t *enc, AV *av) 332encode_av (enc_t *enc, AV *av)
200{ 333{
203 if (enc->depth >= enc->cbor.max_depth) 336 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 337 croak (ERR_NESTING_EXCEEDED);
205 338
206 ++enc->depth; 339 ++enc->depth;
207 340
208 encode_uint (enc, 0x80, len + 1); 341 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 342
343 if (SvMAGICAL (av))
210 for (i = 0; i <= len; ++i) 344 for (i = 0; i <= len; ++i)
211 { 345 {
212 SV **svp = av_fetch (av, i, 0); 346 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 347 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 348 }
349 else
350 for (i = 0; i <= len; ++i)
351 {
352 SV *sv = AvARRAY (av)[i];
353 encode_sv (enc, sv ? sv : &PL_sv_undef);
354 }
215 355
216 --enc->depth; 356 --enc->depth;
217} 357}
218 358
219static void 359static void
228 368
229 int pairs = hv_iterinit (hv); 369 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 370 int mg = SvMAGICAL (hv);
231 371
232 if (mg) 372 if (mg)
233 encode_ch (enc, 0xa0 | 31); 373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 374 else
235 encode_uint (enc, 0xa0, pairs); 375 encode_uint (enc, MAJOR_MAP, pairs);
236 376
237 while ((he = hv_iternext (hv))) 377 while ((he = hv_iternext (hv)))
238 { 378 {
239 if (HeKLEN (he) == HEf_SVKEY) 379 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 380 encode_sv (enc, HeSVKEY (he));
241 else 381 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 383
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 385 }
246 386
247 if (mg) 387 if (mg)
248 encode_ch (enc, 0xe0 | 31); 388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 389
250 --enc->depth; 390 --enc->depth;
251} 391}
252 392
253// encode objects, arrays and special \0=false and \1=true values. 393// encode objects, arrays and special \0=false and \1=true values.
254static void 394static void
255encode_rv (enc_t *enc, SV *sv) 395encode_rv (enc_t *enc, SV *sv)
256{ 396{
257 svtype svt;
258
259 SvGETMAGIC (sv); 397 SvGETMAGIC (sv);
398
260 svt = SvTYPE (sv); 399 svtype svt = SvTYPE (sv);
261 400
262 if (ecb_expect_false (SvOBJECT (sv))) 401 if (ecb_expect_false (SvOBJECT (sv)))
263 { 402 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 403 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 404 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 409 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 410 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 411 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 412
274 HV *stash = SvSTASH (sv); 413 HV *stash = SvSTASH (sv);
275 GV *method;
276 414
277 if (stash == boolean_stash) 415 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 416 {
417 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
418 return;
419 }
279 else if (stash == error_stash) 420 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 421 {
422 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
423 return;
424 }
281 else if (stash == tagged_stash) 425 else if (stash == tagged_stash)
282 { 426 {
283 if (svt != SVt_PVAV) 427 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 428 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 429
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 430 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 431 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
432
433 return;
434 }
435 }
436
437 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 {
440 if (!enc->shareable)
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444
445 if (SvOK (*svp))
288 } 446 {
447 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
448 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
449 return;
450 }
451 else
452 {
453 sv_setuv (*svp, enc->shareable_idx);
454 ++enc->shareable_idx;
455 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
456 }
457 }
458
459 if (ecb_expect_false (SvOBJECT (sv)))
460 {
461 HV *stash = SvSTASH (sv);
462 GV *method;
463
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 465 {
291 dSP; 466 dSP;
292 467
293 ENTER; SAVETMPS; PUSHMARK (SP); 468 ENTER; SAVETMPS;
469 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 470 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 471 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 472
297 PUTBACK; 473 PUTBACK;
298 // G_SCALAR ensures that return value is 1 474 // G_SCALAR ensures that return value is 1
311 } 487 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 489 {
314 dSP; 490 dSP;
315 491
316 ENTER; SAVETMPS; PUSHMARK (SP); 492 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP);
317 EXTEND (SP, 2); 495 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 496 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 498 PUSHs (sv_cbor);
321 499
325 503
326 // catch this surprisingly common error 504 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 507
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 508 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 509 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 511
334 while (count) 512 while (count)
335 encode_sv (enc, SP[1 - count--]); 513 encode_sv (enc, SP[1 - count--]);
336 514
337 PUTBACK; 515 PUTBACK;
344 } 522 }
345 else if (svt == SVt_PVHV) 523 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 524 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 525 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 526 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 527 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 528 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 529 encode_tag (enc, CBOR_TAG_INDIRECTION);
530 encode_sv (enc, sv);
531 }
369} 532}
370 533
371static void 534static void
372encode_nv (enc_t *enc, SV *sv) 535encode_nv (enc_t *enc, SV *sv)
373{ 536{
374 double nv = SvNVX (sv); 537 double nv = SvNVX (sv);
375 538
376 need (enc, 9); 539 need (enc, 9);
377 540
378 if (ecb_expect_false (nv == (U32)nv)) 541 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 542 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 543 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 544 else if (ecb_expect_false (nv == (float)nv))
382 { 545 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 546 uint32_t fp = ecb_float_to_binary32 (nv);
384 547
385 *enc->cur++ = 0xe0 | 26; 548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 549
387 if (!ecb_big_endian ()) 550 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 551 fp = ecb_bswap32 (fp);
389 552
390 memcpy (enc->cur, &fp, 4); 553 memcpy (enc->cur, &fp, 4);
392 } 555 }
393 else 556 else
394 { 557 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 558 uint64_t fp = ecb_double_to_binary64 (nv);
396 559
397 *enc->cur++ = 0xe0 | 27; 560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 561
399 if (!ecb_big_endian ()) 562 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 563 fp = ecb_bswap64 (fp);
401 564
402 memcpy (enc->cur, &fp, 8); 565 memcpy (enc->cur, &fp, 8);
411 574
412 if (SvPOKp (sv)) 575 if (SvPOKp (sv))
413 { 576 {
414 STRLEN len; 577 STRLEN len;
415 char *str = SvPV (sv, len); 578 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 579 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 580 }
418 else if (SvNOKp (sv)) 581 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 582 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 583 else if (SvIOKp (sv))
421 { 584 {
422 if (SvIsUV (sv)) 585 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 586 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 587 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 588 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 589 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 590 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 591 }
429 else if (SvROK (sv)) 592 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 593 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 594 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 595 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 596 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 597 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 598 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 599 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 600 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 601}
439 602
440static SV * 603static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 604encode_cbor (SV *scalar, CBOR *cbor)
442{ 605{
443 enc_t enc; 606 enc_t enc = { 0 };
444 607
445 enc.cbor = *cbor; 608 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 610 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 611 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 612
451 SvPOK_only (enc.sv); 613 SvPOK_only (enc.sv);
614
615 if (cbor->flags & F_PACK_STRINGS)
616 {
617 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
618 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
619 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
620 }
621
452 encode_sv (&enc, scalar); 622 encode_sv (&enc, scalar);
453 623
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 624 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 625 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 626
470 U8 *end; // end of input string 640 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 641 const char *err; // parse error, if != 0
472 CBOR cbor; 642 CBOR cbor;
473 U32 depth; // recursion depth 643 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 644 U32 maxdepth; // recursion depth limit
645 AV *shareable;
646 AV *stringref;
647 SV *decode_tagged;
475} dec_t; 648} dec_t;
476 649
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 651
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 655#define DEC_DEC_DEPTH --dec->depth
483 656
484static UV 657static UV
485decode_uint (dec_t *dec) 658decode_uint (dec_t *dec)
486{ 659{
487 switch (*dec->cur & 31) 660 U8 m = *dec->cur & MINOR_MASK;
488 { 661 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 662
494 case 24: 663 if (ecb_expect_true (m < LENGTH_EXT1))
664 return m;
665 else if (ecb_expect_true (m == LENGTH_EXT1))
666 {
495 WANT (2); 667 WANT (1);
496 dec->cur += 2; 668 dec->cur += 1;
497 return dec->cur[-1]; 669 return dec->cur[-1];
498 670 }
499 case 25: 671 else if (ecb_expect_true (m == LENGTH_EXT2))
672 {
500 WANT (3); 673 WANT (2);
501 dec->cur += 3; 674 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 675 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 676 | ((UV)dec->cur[-1]);
504 677 }
505 case 26: 678 else if (ecb_expect_true (m == LENGTH_EXT4))
679 {
506 WANT (5); 680 WANT (4);
507 dec->cur += 5; 681 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 682 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 683 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 684 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 685 | ((UV)dec->cur[-1]);
512 686 }
513 case 27: 687 else if (ecb_expect_true (m == LENGTH_EXT8))
688 {
514 WANT (9); 689 WANT (8);
515 dec->cur += 9; 690 dec->cur += 8;
691
692 return
693#if UVSIZE < 8
694 0
695#else
516 return (((UV)dec->cur[-8]) << 56) 696 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 697 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 698 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 699 | (((UV)dec->cur[-5]) << 32)
700#endif
520 | (((UV)dec->cur[-4]) << 24) 701 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 702 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 703 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 704 | ((UV)dec->cur[-1]);
524 705 }
525 default: 706 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 707 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 708
529fail: 709fail:
530 return 0; 710 return 0;
531} 711}
532 712
537{ 717{
538 AV *av = newAV (); 718 AV *av = newAV ();
539 719
540 DEC_INC_DEPTH; 720 DEC_INC_DEPTH;
541 721
542 if ((*dec->cur & 31) == 31) 722 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 723 {
544 ++dec->cur; 724 ++dec->cur;
545 725
546 for (;;) 726 for (;;)
547 { 727 {
548 WANT (1); 728 WANT (1);
549 729
550 if (*dec->cur == (0xe0 | 31)) 730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 731 {
552 ++dec->cur; 732 ++dec->cur;
553 break; 733 break;
554 } 734 }
555 735
558 } 738 }
559 else 739 else
560 { 740 {
561 int i, len = decode_uint (dec); 741 int i, len = decode_uint (dec);
562 742
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 744 av_fill (av, len - 1);
564 745
565 for (i = 0; i < len; ++i) 746 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 747 AvARRAY (av)[i] = decode_sv (dec);
567 } 748 }
573 SvREFCNT_dec (av); 754 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 755 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 756 return &PL_sv_undef;
576} 757}
577 758
759static void
760decode_he (dec_t *dec, HV *hv)
761{
762 // for speed reasons, we specialcase single-string
763 // byte or utf-8 strings as keys, but only when !stringref
764
765 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
767 {
768 I32 len = decode_uint (dec);
769 char *key = (char *)dec->cur;
770
771 WANT (len);
772 dec->cur += len;
773
774 hv_store (hv, key, len, decode_sv (dec), 0);
775
776 return;
777 }
778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
779 {
780 I32 len = decode_uint (dec);
781 char *key = (char *)dec->cur;
782
783 WANT (len);
784 dec->cur += len;
785
786 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
787 if (!is_utf8_string (key, len))
788 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
789
790 hv_store (hv, key, -len, decode_sv (dec), 0);
791
792 return;
793 }
794
795 SV *k = decode_sv (dec);
796 SV *v = decode_sv (dec);
797
798 hv_store_ent (hv, k, v, 0);
799 SvREFCNT_dec (k);
800
801fail:
802 ;
803}
804
578static SV * 805static SV *
579decode_hv (dec_t *dec) 806decode_hv (dec_t *dec)
580{ 807{
581 HV *hv = newHV (); 808 HV *hv = newHV ();
582 809
583 DEC_INC_DEPTH; 810 DEC_INC_DEPTH;
584 811
585 if ((*dec->cur & 31) == 31) 812 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 813 {
587 ++dec->cur; 814 ++dec->cur;
588 815
589 for (;;) 816 for (;;)
590 { 817 {
591 WANT (1); 818 WANT (1);
592 819
593 if (*dec->cur == (0xe0 | 31)) 820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 821 {
595 ++dec->cur; 822 ++dec->cur;
596 break; 823 break;
597 } 824 }
598 825
599 SV *k = decode_sv (dec); 826 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 827 }
605 } 828 }
606 else 829 else
607 { 830 {
608 int len = decode_uint (dec); 831 int pairs = decode_uint (dec);
609 832
610 while (len--) 833 while (pairs--)
611 { 834 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 835 }
619 836
620 DEC_DEC_DEPTH; 837 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 838 return newRV_noinc ((SV *)hv);
622 839
629static SV * 846static SV *
630decode_str (dec_t *dec, int utf8) 847decode_str (dec_t *dec, int utf8)
631{ 848{
632 SV *sv = 0; 849 SV *sv = 0;
633 850
634 if ((*dec->cur & 31) == 31) 851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 852 {
853 // indefinite length strings
636 ++dec->cur; 854 ++dec->cur;
637 855
856 U8 major = *dec->cur & MAJOR_MISC;
857
638 sv = newSVpvn ("", 0); 858 sv = newSVpvn ("", 0);
639 859
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 860 for (;;)
642 { 861 {
643 WANT (1); 862 WANT (1);
644 863
645 if (*dec->cur == (0xe0 | 31)) 864 if ((*dec->cur - major) > LENGTH_EXT8)
865 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 866 {
647 ++dec->cur; 867 ++dec->cur;
648 break; 868 break;
649 } 869 }
870 else
871 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 872
651 sv_catsv (sv, decode_sv (dec)); 873 STRLEN len = decode_uint (dec);
874
875 WANT (len);
876 sv_catpvn (sv, dec->cur, len);
877 dec->cur += len;
652 } 878 }
653 } 879 }
654 else 880 else
655 { 881 {
656 STRLEN len = decode_uint (dec); 882 STRLEN len = decode_uint (dec);
657 883
658 WANT (len); 884 WANT (len);
659 sv = newSVpvn (dec->cur, len); 885 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 886 dec->cur += len;
887
888 if (ecb_expect_false (dec->stringref)
889 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
890 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 891 }
662 892
663 if (utf8) 893 if (utf8)
894 {
895 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
896 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
897 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
898
664 SvUTF8_on (sv); 899 SvUTF8_on (sv);
900 }
665 901
666 return sv; 902 return sv;
667 903
668fail: 904fail:
669 SvREFCNT_dec (sv); 905 SvREFCNT_dec (sv);
671} 907}
672 908
673static SV * 909static SV *
674decode_tagged (dec_t *dec) 910decode_tagged (dec_t *dec)
675{ 911{
912 SV *sv = 0;
676 UV tag = decode_uint (dec); 913 UV tag = decode_uint (dec);
914
915 WANT (1);
916
917 switch (tag)
918 {
919 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 920 sv = decode_sv (dec);
921 break;
678 922
679 if (tag == CBOR_TAG_MAGIC) 923 case CBOR_TAG_INDIRECTION:
680 return sv; 924 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 925 break;
682 { 926
927 case CBOR_TAG_STRINGREF_NAMESPACE:
928 {
929 ENTER; SAVETMPS;
930
931 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933
934 sv = decode_sv (dec);
935
936 FREETMPS; LEAVE;
937 }
938 break;
939
940 case CBOR_TAG_STRINGREF:
941 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944
945 UV idx = decode_uint (dec);
946
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 }
952 break;
953
954 case CBOR_TAG_VALUE_SHAREABLE:
955 {
956 if (ecb_expect_false (!dec->shareable))
957 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
958
959 if (dec->cbor.flags & F_ALLOW_CYCLES)
960 {
961 sv = newSV (0);
962 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
963
964 SV *osv = decode_sv (dec);
965 sv_setsv (sv, osv);
966 SvREFCNT_dec_NN (osv);
967 }
968 else
969 {
970 av_push (dec->shareable, &PL_sv_undef);
971 int idx = AvFILLp (dec->shareable);
972 sv = decode_sv (dec);
973 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
974 }
975 }
976 break;
977
978 case CBOR_TAG_VALUE_SHAREDREF:
979 {
980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
982
983 UV idx = decode_uint (dec);
984
985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
986 ERR ("corrupted CBOR data (sharedref index out of bounds)");
987
988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989
990 if (sv == &PL_sv_undef)
991 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
992 }
993 break;
994
995 case CBOR_TAG_PERL_OBJECT:
996 {
997 sv = decode_sv (dec);
998
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 1000 ERR ("corrupted CBOR data (non-array perl object)");
685 1001
686 AV *av = (AV *)SvRV (sv); 1002 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 1003 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 1004 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 1005
690 if (!stash) 1006 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 1007 ERR ("cannot decode perl-object (package does not exist)");
692 1008
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 1009 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 1010
695 if (!method) 1011 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1012 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 1013
698 dSP; 1014 dSP;
699 1015
700 ENTER; SAVETMPS; PUSHMARK (SP); 1016 ENTER; SAVETMPS;
1017 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1018 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1019 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1020 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1021 PUSHs (sv_cbor);
705 1022
706 int i; 1023 int i;
707 1024
708 for (i = 1; i < len; ++i) 1025 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1026 PUSHs (*av_fetch (av, i, 1));
710 1027
711 PUTBACK; 1028 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1029 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1030 SPAGAIN;
714 1031
1032 if (SvTRUE (ERRSV))
1033 {
1034 FREETMPS; LEAVE;
1035 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1036 }
1037
1038 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1039 sv = SvREFCNT_inc (POPs);
716 1040
717 PUTBACK; 1041 PUTBACK;
718 1042
719 FREETMPS; LEAVE; 1043 FREETMPS; LEAVE;
1044 }
1045 break;
720 1046
721 return sv; 1047 default:
722 } 1048 {
723 else 1049 sv = decode_sv (dec);
724 { 1050
1051 dSP;
1052 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP);
1055 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag));
1057 PUSHs (sv);
1058
1059 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN;
1062
1063 if (SvTRUE (ERRSV))
1064 {
1065 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1067 }
1068
1069 if (count)
1070 {
1071 SvREFCNT_dec (sv);
1072 sv = SvREFCNT_inc (POPs);
1073 }
1074 else
1075 {
725 AV *av = newAV (); 1076 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1077 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1078 av_push (av, sv);
728 1079
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1081 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1082 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1083 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1084 }
1085
1086 PUTBACK;
1087
1088 FREETMPS; LEAVE;
1089 }
1090 break;
734 } 1091 }
1092
1093 return sv;
735 1094
736fail: 1095fail:
737 SvREFCNT_dec (sv); 1096 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1097 return &PL_sv_undef;
739} 1098}
741static SV * 1100static SV *
742decode_sv (dec_t *dec) 1101decode_sv (dec_t *dec)
743{ 1102{
744 WANT (1); 1103 WANT (1);
745 1104
746 switch (*dec->cur >> 5) 1105 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1106 {
748 case 0: // unsigned int 1107 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1108 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1109 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1110 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1111 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1112 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1113 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1114
756 case 4: // array 1115 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1116 switch (*dec->cur++ & MINOR_MASK)
764 { 1117 {
765 case 20: 1118 case SIMPLE_FALSE:
766#if CBOR_SLOW 1119#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1120 types_false = get_bool ("Types::Serialiser::false");
768#endif 1121#endif
769 return newSVsv (types_false); 1122 return newSVsv (types_false);
770 case 21: 1123 case SIMPLE_TRUE:
771#if CBOR_SLOW 1124#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1125 types_true = get_bool ("Types::Serialiser::true");
773#endif 1126#endif
774 return newSVsv (types_true); 1127 return newSVsv (types_true);
775 case 22: 1128 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1129 return newSVsv (&PL_sv_undef);
777 case 23: 1130 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1131#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1132 types_error = get_bool ("Types::Serialiser::error");
780#endif 1133#endif
781 return newSVsv (types_error); 1134 return newSVsv (types_error);
782 1135
783 case 25: 1136 case MISC_FLOAT16:
784 { 1137 {
785 WANT (2); 1138 WANT (2);
786 1139
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1140 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1141 dec->cur += 2;
789 1142
790 return newSVnv (ecb_binary16_to_float (fp)); 1143 return newSVnv (ecb_binary16_to_float (fp));
791 } 1144 }
792 1145
793 case 26: 1146 case MISC_FLOAT32:
794 { 1147 {
795 uint32_t fp; 1148 uint32_t fp;
796 WANT (4); 1149 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1150 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1151 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1154 fp = ecb_bswap32 (fp);
802 1155
803 return newSVnv (ecb_binary32_to_float (fp)); 1156 return newSVnv (ecb_binary32_to_float (fp));
804 } 1157 }
805 1158
806 case 27: 1159 case MISC_FLOAT64:
807 { 1160 {
808 uint64_t fp; 1161 uint64_t fp;
809 WANT (8); 1162 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1163 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1164 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1167 fp = ecb_bswap64 (fp);
815 1168
816 return newSVnv (ecb_binary64_to_double (fp)); 1169 return newSVnv (ecb_binary64_to_double (fp));
817 } 1170 }
818 1171
819 // 0..19 unassigned 1172 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1173 // 24 reserved + unassigned simple (reserved values are not encodable)
1174 // 28-30 unassigned misc
1175 // 31 break code
821 default: 1176 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1177 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1178 }
824 1179
825 break; 1180 break;
826 } 1181 }
827 1182
830} 1185}
831 1186
832static SV * 1187static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1188decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1189{
835 dec_t dec; 1190 dec_t dec = { 0 };
836 SV *sv; 1191 SV *sv;
1192 STRLEN len;
1193 char *data = SvPVbyte (string, len);
837 1194
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1195 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1196 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1197 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1198
873 dec.cbor = *cbor; 1199 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1200 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1201 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1202
879 sv = decode_sv (&dec); 1203 sv = decode_sv (&dec);
880 1204
881 if (offset_return) 1205 if (offset_return)
882 *offset_return = dec.cur; 1206 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1209 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1210 dec.err = "garbage after CBOR object";
887 1211
888 if (dec.err) 1212 if (dec.err)
889 { 1213 {
1214 if (dec.shareable)
1215 {
1216 // need to break cyclic links, which whould all be in shareable
1217 int i;
1218 SV **svp;
1219
1220 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef);
1223 }
1224
890 SvREFCNT_dec (sv); 1225 SvREFCNT_dec (sv);
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1227 }
893 1228
894 sv = sv_2mortal (sv); 1229 sv = sv_2mortal (sv);
895 1230
896 return sv; 1231 return sv;
897} 1232}
898 1233
1234/////////////////////////////////////////////////////////////////////////////
1235// incremental parser
1236
1237#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1238
1239// returns 0 for notyet, 1 for success or error
1240static int
1241incr_parse (CBOR *self, SV *cborstr)
1242{
1243 STRLEN cur;
1244 SvPV (cborstr, cur);
1245
1246 while (ecb_expect_true (self->incr_need <= cur))
1247 {
1248 // table of integer count bytes
1249 static I8 incr_len[MINOR_MASK + 1] = {
1250 0, 0, 0, 0, 0, 0, 0, 0,
1251 0, 0, 0, 0, 0, 0, 0, 0,
1252 0, 0, 0, 0, 0, 0, 0, 0,
1253 1, 2, 4, 8,-1,-1,-1,-2
1254 };
1255
1256 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1257 U8 m = *p & MINOR_MASK;
1258 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1259 I8 ilen = incr_len[m];
1260
1261 self->incr_need = self->incr_pos + 1;
1262
1263 if (ecb_expect_false (ilen < 0))
1264 {
1265 if (m != MINOR_INDEF)
1266 return 1; // error
1267
1268 if (*p == (MAJOR_MISC | MINOR_INDEF))
1269 {
1270 if (count >= 0)
1271 return 1; // error
1272
1273 count = 1;
1274 }
1275 else
1276 {
1277 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1278 count = -1;
1279 }
1280 }
1281 else
1282 {
1283 self->incr_need += ilen;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 int major = *p >> MAJOR_SHIFT;
1288
1289 switch (major)
1290 {
1291 case MAJOR_TAG >> MAJOR_SHIFT:
1292 ++count; // tags merely prefix another value
1293 break;
1294
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 case MAJOR_ARRAY >> MAJOR_SHIFT:
1298 case MAJOR_MAP >> MAJOR_SHIFT:
1299 {
1300 UV len;
1301
1302 if (ecb_expect_false (ilen))
1303 {
1304 len = 0;
1305
1306 do {
1307 len = (len << 8) | *++p;
1308 } while (--ilen);
1309 }
1310 else
1311 len = m;
1312
1313 switch (major)
1314 {
1315 case MAJOR_BYTES >> MAJOR_SHIFT:
1316 case MAJOR_TEXT >> MAJOR_SHIFT:
1317 self->incr_need += len;
1318 if (ecb_expect_false (self->incr_need > cur))
1319 return 0;
1320
1321 break;
1322
1323 case MAJOR_MAP >> MAJOR_SHIFT:
1324 len <<= 1;
1325 case MAJOR_ARRAY >> MAJOR_SHIFT:
1326 if (len)
1327 {
1328 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1329 count = len + 1;
1330 }
1331 break;
1332 }
1333 }
1334 }
1335 }
1336
1337 self->incr_pos = self->incr_need;
1338
1339 if (count > 0)
1340 {
1341 while (!--count)
1342 {
1343 if (!AvFILLp (self->incr_count))
1344 return 1; // done
1345
1346 SvREFCNT_dec_NN (av_pop (self->incr_count));
1347 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1348 }
1349
1350 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1351 }
1352 }
1353
1354 return 0;
1355}
1356
1357
899///////////////////////////////////////////////////////////////////////////// 1358/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1359// XS interface functions
901 1360
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1361MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1362
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1369 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1370
912 types_true = get_bool ("Types::Serialiser::true" ); 1371 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1372 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1373 types_error = get_bool ("Types::Serialiser::error");
1374
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
915 1376
916 sv_cbor = newSVpv ("CBOR", 0); 1377 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1378 SvREADONLY_on (sv_cbor);
918} 1379}
919 1380
940 1401
941void shrink (CBOR *self, int enable = 1) 1402void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1403 ALIAS:
943 shrink = F_SHRINK 1404 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1405 allow_unknown = F_ALLOW_UNKNOWN
1406 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES
1408 pack_strings = F_PACK_STRINGS
1409 utf8_strings = F_UTF8_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1411 PPCODE:
946{ 1412{
947 if (enable) 1413 if (enable)
948 self->flags |= ix; 1414 self->flags |= ix;
949 else 1415 else
954 1420
955void get_shrink (CBOR *self) 1421void get_shrink (CBOR *self)
956 ALIAS: 1422 ALIAS:
957 get_shrink = F_SHRINK 1423 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1424 get_allow_unknown = F_ALLOW_UNKNOWN
1425 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES
1427 get_pack_strings = F_PACK_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1429 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1430 XPUSHs (boolSV (self->flags & ix));
961 1431
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1433 PPCODE:
979 CODE: 1449 CODE:
980 RETVAL = self->max_size; 1450 RETVAL = self->max_size;
981 OUTPUT: 1451 OUTPUT:
982 RETVAL 1452 RETVAL
983 1453
984#if 0 //TODO 1454void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1455 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1456 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1457 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1458 XPUSHs (ST (0));
993}
994 1459
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1460SV *get_filter (CBOR *self)
996 PPCODE: 1461 CODE:
997{ 1462 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1463 OUTPUT:
999 self->cb_sk_object = newHV (); 1464 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1465
1019void encode (CBOR *self, SV *scalar) 1466void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1467 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1468 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1469 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1482 EXTEND (SP, 2);
1036 PUSHs (sv); 1483 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1484 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1485}
1039 1486
1040#if 0 1487void incr_parse (CBOR *self, SV *cborstr)
1488 ALIAS:
1489 incr_parse_multiple = 1
1490 PPCODE:
1491{
1492 if (SvUTF8 (cborstr))
1493 sv_utf8_downgrade (cborstr, 0);
1494
1495 if (!self->incr_count)
1496 {
1497 self->incr_count = newAV ();
1498 self->incr_pos = 0;
1499 self->incr_need = 1;
1500
1501 av_push (self->incr_count, newSViv (1));
1502 }
1503
1504 do
1505 {
1506 if (!incr_parse (self, cborstr))
1507 {
1508 if (self->incr_need > self->max_size && self->max_size)
1509 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1510 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1511
1512 break;
1513 }
1514
1515 SV *sv;
1516 char *offset;
1517
1518 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1519 XPUSHs (sv);
1520
1521 sv_chop (cborstr, offset);
1522
1523 av_clear (self->incr_count);
1524 av_push (self->incr_count, newSViv (1));
1525
1526 self->incr_pos = 0;
1527 self->incr_need = self->incr_pos + 1;
1528 }
1529 while (ix);
1530}
1531
1532void incr_reset (CBOR *self)
1533 CODE:
1534{
1535 SvREFCNT_dec (self->incr_count);
1536 self->incr_count = 0;
1537}
1041 1538
1042void DESTROY (CBOR *self) 1539void DESTROY (CBOR *self)
1043 CODE: 1540 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1541 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1542
1049PROTOTYPES: ENABLE 1543PROTOTYPES: ENABLE
1050 1544
1051void encode_cbor (SV *scalar) 1545void encode_cbor (SV *scalar)
1546 ALIAS:
1547 encode_cbor = 0
1548 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1549 PPCODE:
1053{ 1550{
1054 CBOR cbor; 1551 CBOR cbor;
1055 cbor_init (&cbor); 1552 cbor_init (&cbor);
1553 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1554 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1555 XPUSHs (scalar);
1058} 1556}
1059 1557
1060void decode_cbor (SV *cborstr) 1558void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines