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 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.38 by root, Sun Dec 1 14:30:52 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
65// known tags
66enum cbor_tag
67{
68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
76
77 // rfc7049
78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
80 CBOR_TAG_POS_BIGNUM = 2, // byte string
81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
83 CBOR_TAG_BIGFLOAT = 5, // array
84
85 CBOR_TAG_CONV_B64U = 21, // base64url, any
86 CBOR_TAG_CONV_B64 = 22, // base64, any
87 CBOR_TAG_CONV_HEX = 23, // base16, any
88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
89
90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97};
98
14#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
16 105
17#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
18 107
19#define SB do { 108#define SB do {
20#define SE } while (0) 109#define SE } while (0)
31#else 120#else
32# define CBOR_SLOW 0 121# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
34#endif 123#endif
35 124
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 127
39typedef struct { 128typedef struct {
40 U32 flags; 129 U32 flags;
41 U32 max_depth; 130 U32 max_depth;
42 STRLEN max_size; 131 STRLEN max_size;
43 132 SV *filter;
44 SV *cb_object;
45 HV *cb_sk_object;
46} CBOR; 133} CBOR;
47 134
48ecb_inline void 135ecb_inline void
49cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
50{ 137{
51 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 139 cbor->max_depth = 512;
140}
141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
53} 146}
54 147
55///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
56// utility functions 149// utility functions
57 150
79 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 173#endif
81 } 174 }
82} 175}
83 176
84///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
85// fp hell 178ecb_inline int
86 179minimum_string_length (UV idx)
87//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
88 191
89///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
90// encoder 193// encoder
91 194
92// structure used for encoding CBOR 195// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
97 SV *sv; // result scalar 200 SV *sv; // result scalar
98 CBOR cbor; 201 CBOR cbor;
99 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
100} enc_t; 207} enc_t;
101 208
102ecb_inline void 209ecb_inline void
103need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
104{ 211{
121static void 228static void
122encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
123{ 230{
124 need (enc, 9); 231 need (enc, 9);
125 232
126 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
128 else if (len <= 0xff) 235 else if (ecb_expect_true (len <= 0xffU))
129 { 236 {
130 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 238 *enc->cur++ = len;
132 } 239 }
133 else if (len <= 0xffff) 240 else if (len <= 0xffffU)
134 { 241 {
135 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 244 *enc->cur++ = len;
138 } 245 }
139 else if (len <= 0xffffffff) 246 else if (len <= 0xffffffffU)
140 { 247 {
141 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 252 *enc->cur++ = len;
146 } 253 }
147 else 254 else
148 { 255 {
149 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 264 *enc->cur++ = len;
158 } 265 }
159} 266}
160 267
161static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 276{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
165 need (enc, len); 278 need (enc, len);
166 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
167 enc->cur += len; 280 enc->cur += len;
168} 281}
169 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
170static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
171 309
172static void 310static void
173encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
174{ 312{
177 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
179 317
180 ++enc->depth; 318 ++enc->depth;
181 319
182 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 321
184 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
185 { 323 {
186 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
202 340
203 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
205 343
206 if (mg) 344 if (mg)
207 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 346 else
209 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
210 348
211 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
212 { 350 {
213 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
215 else 353 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217 355
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 357 }
220 358
221 if (mg) 359 if (mg)
222 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 361
224 --enc->depth; 362 --enc->depth;
225} 363}
226 364
227// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
228static void 366static void
229encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
230{ 368{
231 svtype svt;
232
233 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
234 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
235 372
236 if (ecb_expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
237 { 374 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 376 ? types_boolean_stash
377 : gv_stashpv ("Types::Serialiser::Boolean", 1);
378 HV *error_stash = !CBOR_SLOW || types_error_stash
379 ? types_error_stash
380 : gv_stashpv ("Types::Serialiser::Error", 1);
381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
382 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 384
242 if (SvSTASH (sv) == stash) 385 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386
387 if (stash == boolean_stash)
388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
392 else if (stash == error_stash)
393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
397 else if (stash == tagged_stash)
398 {
399 if (svt != SVt_PVAV)
400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
401
402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
244 else 423 else
245 { 424 {
246#if 0 //TODO 425 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 437 {
438 dSP;
439
440 ENTER; SAVETMPS; PUSHMARK (SP);
249 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
250 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252 if (to_cbor)
253 {
254 dSP;
255
256 ENTER; SAVETMPS; PUSHMARK (SP);
257 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 443
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 444 PUTBACK;
445 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 446 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 447 SPAGAIN;
263 448
264 // catch this surprisingly common error 449 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 450 if (SvROK (TOPs) && SvRV (TOPs) == sv)
266 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 451 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 452
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 453 encode_sv (enc, POPs);
272 454
455 PUTBACK;
456
273 FREETMPS; LEAVE; 457 FREETMPS; LEAVE;
274 }
275 else if (enc->cbor.flags & F_ALLOW_BLESSED)
276 encode_str (enc, "null", 4, 0);
277 else
278 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
279 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280 } 458 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
282 encode_str (enc, "null", 4, 0); 460 {
461 dSP;
462
463 ENTER; SAVETMPS; PUSHMARK (SP);
464 EXTEND (SP, 2);
465 // we re-bless the reference to get overload and other niceties right
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor);
468
469 PUTBACK;
470 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
471 SPAGAIN;
472
473 // catch this surprisingly common error
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480
481 while (count)
482 encode_sv (enc, SP[1 - count--]);
483
484 PUTBACK;
485
486 FREETMPS; LEAVE;
487 }
283 else 488 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 489 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 490 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 491 }
289 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 495 encode_av (enc, (AV *)sv);
293 else if (svt < SVt_PVAV)
294 {
295 STRLEN len = 0;
296 char *pv = svt ? SvPV (sv, len) : 0;
297
298 if (len == 1 && *pv == '1')
299 encode_ch (enc, 0xe0 | 21);
300 else if (len == 1 && *pv == '0')
301 encode_ch (enc, 0xe0 | 20);
302 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303 encode_ch (enc, 0xe0 | 23);
304 else
305 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309 encode_ch (enc, 0xe0 | 23);
310 else 496 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
313} 501}
314 502
315static void 503static void
316encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
317{ 505{
318 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
319 507
320 need (enc, 9); 508 need (enc, 9);
321 509
322 if (ecb_expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 512 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
326 { 514 {
327 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
328 516
329 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
330 518
331 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
333 521
334 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
336 } 524 }
337 else 525 else
338 { 526 {
339 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
340 528
341 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
342 530
343 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
345 533
346 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
355 543
356 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
357 { 545 {
358 STRLEN len; 546 STRLEN len;
359 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
361 } 549 }
362 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 551 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
365 { 553 {
366 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
368 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
370 else 558 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
372 } 560 }
373 else if (SvROK (sv)) 561 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 567 else
380 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 570}
383 571
384static SV * 572static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
386{ 574{
387 enc_t enc; 575 enc_t enc = { };
388 576
389 enc.cbor = *cbor; 577 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 581
395 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
396 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
397 592
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 595
414 U8 *end; // end of input string 609 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
416 CBOR cbor; 611 CBOR cbor;
417 U32 depth; // recursion depth 612 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
419} dec_t; 617} dec_t;
420 618
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422 620
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
426#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
427 625
428static UV 626static UV
429decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
430{ 628{
431 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
432 { 630 ++dec->cur;
433 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436 return *dec->cur++ & 31;
437 631
438 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
439 WANT (2); 636 WANT (1);
440 dec->cur += 2; 637 dec->cur += 1;
441 return dec->cur[-1]; 638 return dec->cur[-1];
442 639 }
443 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
444 WANT (3); 642 WANT (2);
445 dec->cur += 3; 643 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
448 646 }
449 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
450 WANT (5); 649 WANT (4);
451 dec->cur += 5; 650 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
456 655 }
457 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
458 WANT (9); 658 WANT (8);
459 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
460 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
464 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
468 674 }
469 default: 675 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 677
473fail: 678fail:
474 return 0; 679 return 0;
475} 680}
476 681
481{ 686{
482 AV *av = newAV (); 687 AV *av = newAV ();
483 688
484 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
485 690
486 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 692 {
488 ++dec->cur; 693 ++dec->cur;
489 694
490 for (;;) 695 for (;;)
491 { 696 {
492 WANT (1); 697 WANT (1);
493 698
494 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
495 { 700 {
496 ++dec->cur; 701 ++dec->cur;
497 break; 702 break;
498 } 703 }
499 704
502 } 707 }
503 else 708 else
504 { 709 {
505 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
506 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
507 av_fill (av, len - 1); 713 av_fill (av, len - 1);
508 714
509 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
511 } 717 }
517 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 725 return &PL_sv_undef;
520} 726}
521 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
522static SV * 772static SV *
523decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
524{ 774{
525 HV *hv = newHV (); 775 HV *hv = newHV ();
526 776
527 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
528 778
529 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 780 {
531 ++dec->cur; 781 ++dec->cur;
532 782
533 for (;;) 783 for (;;)
534 { 784 {
535 WANT (1); 785 WANT (1);
536 786
537 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
538 { 788 {
539 ++dec->cur; 789 ++dec->cur;
540 break; 790 break;
541 } 791 }
542 792
543 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 794 }
548 } 795 }
549 else 796 else
550 { 797 {
551 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
552 799
553 while (len--) 800 while (pairs--)
554 { 801 decode_he (dec, hv);
555 SV *k = decode_sv (dec);
556 SV *v = decode_sv (dec);
557
558 hv_store_ent (hv, k, v, 0);
559 }
560 } 802 }
561 803
562 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
564
565#if 0
566 SV *sv;
567 HV *hv = newHV ();
568
569 DEC_INC_DEPTH;
570 decode_ws (dec);
571
572 for (;;)
573 {
574 // heuristic: assume that
575 // a) decode_str + hv_store_ent are abysmally slow.
576 // b) most hash keys are short, simple ascii text.
577 // => try to "fast-match" such strings to avoid
578 // the overhead of decode_str + hv_store_ent.
579 {
580 SV *value;
581 char *p = dec->cur;
582 char *e = p + 24; // only try up to 24 bytes
583
584 for (;;)
585 {
586 // the >= 0x80 is false on most architectures
587 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
588 {
589 // slow path, back up and use decode_str
590 SV *key = decode_str (dec);
591 if (!key)
592 goto fail;
593
594 decode_ws (dec); EXPECT_CH (':');
595
596 decode_ws (dec);
597 value = decode_sv (dec);
598 if (!value)
599 {
600 SvREFCNT_dec (key);
601 goto fail;
602 }
603
604 hv_store_ent (hv, key, value, 0);
605 SvREFCNT_dec (key);
606
607 break;
608 }
609 else if (*p == '"')
610 {
611 // fast path, got a simple key
612 char *key = dec->cur;
613 int len = p - key;
614 dec->cur = p + 1;
615
616 decode_ws (dec); EXPECT_CH (':');
617
618 decode_ws (dec);
619 value = decode_sv (dec);
620 if (!value)
621 goto fail;
622
623 hv_store (hv, key, len, value, 0);
624
625 break;
626 }
627
628 ++p;
629 }
630 }
631
632 decode_ws (dec);
633
634 if (*dec->cur == '}')
635 {
636 ++dec->cur;
637 break;
638 }
639
640 if (*dec->cur != ',')
641 ERR (", or } expected while parsing object/hash");
642
643 ++dec->cur;
644
645 decode_ws (dec);
646
647 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648 {
649 ++dec->cur;
650 break;
651 }
652 }
653
654 DEC_DEC_DEPTH;
655 sv = newRV_noinc ((SV *)hv);
656
657 // check filter callbacks
658 if (dec->cbor.flags & F_HOOK)
659 {
660 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661 {
662 HE *cb, *he;
663
664 hv_iterinit (hv);
665 he = hv_iternext (hv);
666 hv_iterinit (hv);
667
668 // the next line creates a mortal sv each time its called.
669 // might want to optimise this for common cases.
670 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672 if (cb)
673 {
674 dSP;
675 int count;
676
677 ENTER; SAVETMPS; PUSHMARK (SP);
678 XPUSHs (HeVAL (he));
679 sv_2mortal (sv);
680
681 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683 if (count == 1)
684 {
685 sv = newSVsv (POPs);
686 FREETMPS; LEAVE;
687 return sv;
688 }
689
690 SvREFCNT_inc (sv);
691 FREETMPS; LEAVE;
692 }
693 }
694
695 if (dec->cbor.cb_object)
696 {
697 dSP;
698 int count;
699
700 ENTER; SAVETMPS; PUSHMARK (SP);
701 XPUSHs (sv_2mortal (sv));
702
703 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705 if (count == 1)
706 {
707 sv = newSVsv (POPs);
708 FREETMPS; LEAVE;
709 return sv;
710 }
711
712 SvREFCNT_inc (sv);
713 FREETMPS; LEAVE;
714 }
715 }
716
717 return sv;
718#endif
719 806
720fail: 807fail:
721 SvREFCNT_dec (hv); 808 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 809 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 810 return &PL_sv_undef;
724} 811}
725 812
726static SV * 813static SV *
727decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
728{ 815{
729 SV *sv; 816 SV *sv = 0;
730 817
731 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 819 {
820 // indefinite length strings
733 ++dec->cur; 821 ++dec->cur;
734 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
735 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
736 826
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 827 for (;;)
739 { 828 {
740 WANT (1); 829 WANT (1);
741 830
742 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 833 {
744 ++dec->cur; 834 ++dec->cur;
745 break; 835 break;
746 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 839
748 SV *sv2 = decode_sv (dec); 840 STRLEN len = decode_uint (dec);
749 sv_catsv (sv, sv2); 841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
750 } 845 }
751 } 846 }
752 else 847 else
753 { 848 {
754 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
755 850
756 WANT (len); 851 WANT (len);
757 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 858 }
760 859
761 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
762 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
763 868
764 return sv; 869 return sv;
765 870
766fail: 871fail:
872 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 873 return &PL_sv_undef;
768} 874}
769 875
770static SV * 876static SV *
771decode_tagged (dec_t *dec) 877decode_tagged (dec_t *dec)
772{ 878{
879 SV *sv = 0;
773 UV tag = decode_uint (dec); 880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
774 SV *sv = decode_sv (dec); 887 sv = decode_sv (dec);
888 break;
775 889
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 890 case CBOR_TAG_INDIRECTION:
891 sv = newRV_noinc (decode_sv (dec));
892 break;
893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)");
968
969 AV *av = (AV *)SvRV (sv);
970 int len = av_len (av) + 1;
971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
972
973 if (!stash)
974 ERR ("cannot decode perl-object (package does not exist)");
975
976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
977
978 if (!method)
979 ERR ("cannot decode perl-object (package does not have a THAW method)");
980
981 dSP;
982
983 ENTER; SAVETMPS; PUSHMARK (SP);
984 EXTEND (SP, len + 1);
985 // we re-bless the reference to get overload and other niceties right
986 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor);
988
989 int i;
990
991 for (i = 1; i < len; ++i)
992 PUSHs (*av_fetch (av, i, 1));
993
994 PUTBACK;
995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
996 SPAGAIN;
997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
1004 SvREFCNT_dec (sv);
1005 sv = SvREFCNT_inc (POPs);
1006
1007 PUTBACK;
1008
1009 FREETMPS; LEAVE;
1010 }
1011 break;
1012
1013 default:
1014 {
1015 sv = decode_sv (dec);
1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
1040 AV *av = newAV ();
1041 av_push (av, newSVuv (tag));
1042 av_push (av, sv);
1043
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
1055 }
1056
777 return sv; 1057 return sv;
778 1058
779 AV *av = newAV (); 1059fail:
780 av_push (av, newSVuv (tag)); 1060 SvREFCNT_dec (sv);
781 av_push (av, sv); 1061 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1062}
784 1063
785static SV * 1064static SV *
786decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
787{ 1066{
788 WANT (1); 1067 WANT (1);
789 1068
790 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1070 {
792 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1078
800 return decode_str (dec, 1); 1079 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 return decode_tagged (dec);
807 case 7: // misc
808 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
809 { 1081 {
810 case 20: 1082 case SIMPLE_FALSE:
811#if CBOR_SLOW 1083#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
813#endif 1085#endif
814 return newSVsv (cbor_false); 1086 return newSVsv (types_false);
815 case 21: 1087 case SIMPLE_TRUE:
816#if CBOR_SLOW 1088#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
818#endif 1090#endif
819 return newSVsv (cbor_true); 1091 return newSVsv (types_true);
820 case 22: 1092 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1094 case SIMPLE_UNDEF:
1095#if CBOR_SLOW
1096 types_error = get_bool ("Types::Serialiser::error");
1097#endif
1098 return newSVsv (types_error);
822 1099
823 case 25: 1100 case MISC_FLOAT16:
824 { 1101 {
825 WANT (2); 1102 WANT (2);
826 1103
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1105 dec->cur += 2;
829 1106
830 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
831 } 1108 }
832 1109
833 case 26: 1110 case MISC_FLOAT32:
834 { 1111 {
835 uint32_t fp; 1112 uint32_t fp;
836 WANT (4); 1113 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1115 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
842 1119
843 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
844 } 1121 }
845 1122
846 case 27: 1123 case MISC_FLOAT64:
847 { 1124 {
848 uint64_t fp; 1125 uint64_t fp;
849 WANT (8); 1126 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1128 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
855 1132
856 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
857 } 1134 }
858 1135
859 // 0..19 unassigned 1136 // 0..19 unassigned simple
860 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
861 default: 1138 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863 } 1140 }
864 1141
865 break; 1142 break;
866 } 1143 }
867#if 0
868 switch (*dec->cur)
869 {
870 //case '"': ++dec->cur; return decode_str (dec);
871 case '[': ++dec->cur; return decode_av (dec);
872 case '{': ++dec->cur; return decode_hv (dec);
873
874 case '-':
875 case '0': case '1': case '2': case '3': case '4':
876 case '5': case '6': case '7': case '8': case '9':
877 //TODO return decode_num (dec);
878
879 case 't':
880 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881 {
882 dec->cur += 4;
883#if CBOR_SLOW
884 cbor_true = get_bool ("CBOR::XS::true");
885#endif
886 return newSVsv (cbor_true);
887 }
888 else
889 ERR ("'true' expected");
890
891 break;
892
893 case 'f':
894 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895 {
896 dec->cur += 5;
897#if CBOR_SLOW
898 cbor_false = get_bool ("CBOR::XS::false");
899#endif
900 return newSVsv (cbor_false);
901 }
902 else
903 ERR ("'false' expected");
904
905 break;
906
907 case 'n':
908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909 {
910 dec->cur += 4;
911 return newSVsv (&PL_sv_undef);
912 }
913 else
914 ERR ("'null' expected");
915
916 break;
917
918 default:
919 ERR ("malformed CBOR string, neither array, object, number, string or atom");
920 break;
921 }
922#endif
923 1144
924fail: 1145fail:
925 return &PL_sv_undef; 1146 return &PL_sv_undef;
926} 1147}
927 1148
928static SV * 1149static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1151{
931 dec_t dec; 1152 dec_t dec = { };
932 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
933 1156
934 /* work around bugs in 5.10 where manipulating magic values
935 * makes perl ignore the magic in subsequent accesses.
936 * also make a copy of non-PV values, to get them into a clean
937 * state (SvPV should do that, but it's buggy, see below).
938 */
939 /*SvGETMAGIC (string);*/
940 if (SvMAGICAL (string) || !SvPOK (string))
941 string = sv_2mortal (newSVsv (string));
942
943 SvUPGRADE (string, SVt_PV);
944
945 /* work around a bug in perl 5.10, which causes SvCUR to fail an
946 * assertion with -DDEBUGGING, although SvCUR is documented to
947 * return the xpv_cur field which certainly exists after upgrading.
948 * according to nicholas clark, calling SvPOK fixes this.
949 * But it doesn't fix it, so try another workaround, call SvPV_nolen
950 * and hope for the best.
951 * Damnit, SvPV_nolen still trips over yet another assertion. This
952 * assertion business is seriously broken, try yet another workaround
953 * for the broken -DDEBUGGING.
954 */
955 {
956#ifdef DEBUGGING
957 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958#else
959 STRLEN offset = SvCUR (string);
960#endif
961
962 if (offset > cbor->max_size && cbor->max_size) 1157 if (len > cbor->max_size && cbor->max_size)
963 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1158 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1160
969 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
972 dec.err = 0;
973 dec.depth = 0;
974
975 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976 ;//TODO dec.cbor.flags |= F_HOOK;
977 1164
978 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
979 1166
980 if (offset_return) 1167 if (offset_return)
981 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
985 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
986 1173
987 if (dec.err) 1174 if (dec.err)
988 { 1175 {
989 SvREFCNT_dec (sv); 1176 SvREFCNT_dec (sv);
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1177 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1178 }
992 1179
993 sv = sv_2mortal (sv); 1180 sv = sv_2mortal (sv);
994 1181
995 return sv; 1182 return sv;
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1188MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1189
1003BOOT: 1190BOOT:
1004{ 1191{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1192 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1193 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1194
1008 cbor_true = get_bool ("CBOR::XS::true"); 1195 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1196 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1197
1198 types_true = get_bool ("Types::Serialiser::true" );
1199 types_false = get_bool ("Types::Serialiser::false");
1200 types_error = get_bool ("Types::Serialiser::error");
1201
1202 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1203
1204 sv_cbor = newSVpv ("CBOR", 0);
1205 SvREADONLY_on (sv_cbor);
1010} 1206}
1011 1207
1012PROTOTYPES: DISABLE 1208PROTOTYPES: DISABLE
1013 1209
1014void CLONE (...) 1210void CLONE (...)
1015 CODE: 1211 CODE:
1016 cbor_stash = 0; 1212 cbor_stash = 0;
1213 cbor_tagged_stash = 0;
1214 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1215 types_boolean_stash = 0;
1018 1216
1019void new (char *klass) 1217void new (char *klass)
1020 PPCODE: 1218 PPCODE:
1021{ 1219{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1220 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1228
1031void shrink (CBOR *self, int enable = 1) 1229void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1230 ALIAS:
1033 shrink = F_SHRINK 1231 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1232 allow_unknown = F_ALLOW_UNKNOWN
1233 allow_sharing = F_ALLOW_SHARING
1234 allow_cycles = F_ALLOW_CYCLES
1235 pack_strings = F_PACK_STRINGS
1236 validate_utf8 = F_VALIDATE_UTF8
1035 PPCODE: 1237 PPCODE:
1036{ 1238{
1037 if (enable) 1239 if (enable)
1038 self->flags |= ix; 1240 self->flags |= ix;
1039 else 1241 else
1044 1246
1045void get_shrink (CBOR *self) 1247void get_shrink (CBOR *self)
1046 ALIAS: 1248 ALIAS:
1047 get_shrink = F_SHRINK 1249 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1250 get_allow_unknown = F_ALLOW_UNKNOWN
1251 get_allow_sharing = F_ALLOW_SHARING
1252 get_allow_cycles = F_ALLOW_CYCLES
1253 get_pack_strings = F_PACK_STRINGS
1254 get_validate_utf8 = F_VALIDATE_UTF8
1049 PPCODE: 1255 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1256 XPUSHs (boolSV (self->flags & ix));
1051 1257
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1258void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1259 PPCODE:
1069 CODE: 1275 CODE:
1070 RETVAL = self->max_size; 1276 RETVAL = self->max_size;
1071 OUTPUT: 1277 OUTPUT:
1072 RETVAL 1278 RETVAL
1073 1279
1074#if 0 //TODO 1280void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1281 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1282 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1283 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1284 XPUSHs (ST (0));
1083}
1084 1285
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1286SV *get_filter (CBOR *self)
1086 PPCODE: 1287 CODE:
1087{ 1288 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1289 OUTPUT:
1089 self->cb_sk_object = newHV (); 1290 RETVAL
1090
1091 if (SvOK (cb))
1092 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093 else
1094 {
1095 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097 if (!HvKEYS (self->cb_sk_object))
1098 {
1099 SvREFCNT_dec (self->cb_sk_object);
1100 self->cb_sk_object = 0;
1101 }
1102 }
1103
1104 XPUSHs (ST (0));
1105}
1106
1107#endif
1108 1291
1109void encode (CBOR *self, SV *scalar) 1292void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1293 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1294 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1295 XPUSHs (scalar);
1126 PUSHs (sv); 1309 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1310 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1311}
1129 1312
1130void DESTROY (CBOR *self) 1313void DESTROY (CBOR *self)
1131 CODE: 1314 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1315 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1316
1135PROTOTYPES: ENABLE 1317PROTOTYPES: ENABLE
1136 1318
1137void encode_cbor (SV *scalar) 1319void encode_cbor (SV *scalar)
1320 ALIAS:
1321 encode_cbor = 0
1322 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1323 PPCODE:
1139{ 1324{
1140 CBOR cbor; 1325 CBOR cbor;
1141 cbor_init (&cbor); 1326 cbor_init (&cbor);
1327 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1328 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1329 XPUSHs (scalar);
1144} 1330}
1145 1331
1146void decode_cbor (SV *cborstr) 1332void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines