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 1.37 by root, Sat Nov 30 18:13:53 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
16 104
17#define INIT_SIZE 32 // initial scalar size to be allocated 105#define INIT_SIZE 32 // initial scalar size to be allocated
18 106
19#define SB do { 107#define SB do {
20#define SE } while (0) 108#define SE } while (0)
31#else 119#else
32# define CBOR_SLOW 0 120# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 121# define CBOR_STASH cbor_stash
34#endif 122#endif
35 123
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 124static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 125static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 126
39typedef struct { 127typedef struct {
40 U32 flags; 128 U32 flags;
41 U32 max_depth; 129 U32 max_depth;
42 STRLEN max_size; 130 STRLEN max_size;
43 131 SV *filter;
44 SV *cb_object;
45 HV *cb_sk_object;
46} CBOR; 132} CBOR;
47 133
48ecb_inline void 134ecb_inline void
49cbor_init (CBOR *cbor) 135cbor_init (CBOR *cbor)
50{ 136{
51 Zero (cbor, 1, CBOR); 137 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 138 cbor->max_depth = 512;
139}
140
141ecb_inline void
142cbor_free (CBOR *cbor)
143{
144 SvREFCNT_dec (cbor->filter);
53} 145}
54 146
55///////////////////////////////////////////////////////////////////////////// 147/////////////////////////////////////////////////////////////////////////////
56// utility functions 148// utility functions
57 149
79 SvPV_renew (sv, SvCUR (sv) + 1); 171 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 172#endif
81 } 173 }
82} 174}
83 175
84///////////////////////////////////////////////////////////////////////////// 176// minimum length of a string to be registered for stringref
85// fp hell 177ecb_inline int
86 178minimum_string_length (UV idx)
87//TODO 179{
180 return idx > 23
181 ? idx > 0xffU
182 ? idx > 0xffffU
183 ? idx > 0xffffffffU
184 ? 11
185 : 7
186 : 5
187 : 4
188 : 3;
189}
88 190
89///////////////////////////////////////////////////////////////////////////// 191/////////////////////////////////////////////////////////////////////////////
90// encoder 192// encoder
91 193
92// structure used for encoding CBOR 194// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 197 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 198 char *end; // SvEND (sv)
97 SV *sv; // result scalar 199 SV *sv; // result scalar
98 CBOR cbor; 200 CBOR cbor;
99 U32 depth; // recursion level 201 U32 depth; // recursion level
202 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
203 UV stringref_idx;
204 HV *shareable; // ptr => index, or 0
205 UV shareable_idx;
100} enc_t; 206} enc_t;
101 207
102ecb_inline void 208ecb_inline void
103need (enc_t *enc, STRLEN len) 209need (enc_t *enc, STRLEN len)
104{ 210{
121static void 227static void
122encode_uint (enc_t *enc, int major, UV len) 228encode_uint (enc_t *enc, int major, UV len)
123{ 229{
124 need (enc, 9); 230 need (enc, 9);
125 231
126 if (len < 24) 232 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 233 *enc->cur++ = major | len;
128 else if (len <= 0xff) 234 else if (ecb_expect_true (len <= 0xffU))
129 { 235 {
130 *enc->cur++ = major | 24; 236 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 237 *enc->cur++ = len;
132 } 238 }
133 else if (len <= 0xffff) 239 else if (len <= 0xffffU)
134 { 240 {
135 *enc->cur++ = major | 25; 241 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 242 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 243 *enc->cur++ = len;
138 } 244 }
139 else if (len <= 0xffffffff) 245 else if (len <= 0xffffffffU)
140 { 246 {
141 *enc->cur++ = major | 26; 247 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 248 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 249 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 251 *enc->cur++ = len;
146 } 252 }
147 else 253 else
148 { 254 {
149 *enc->cur++ = major | 27; 255 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 256 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 257 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 258 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 259 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 260 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 262 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 263 *enc->cur++ = len;
158 } 264 }
159} 265}
160 266
161static void 267ecb_inline void
268encode_tag (enc_t *enc, UV tag)
269{
270 encode_uint (enc, MAJOR_TAG, tag);
271}
272
273ecb_inline void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 274encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 275{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 276 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
165 need (enc, len); 277 need (enc, len);
166 memcpy (enc->cur, str, len); 278 memcpy (enc->cur, str, len);
167 enc->cur += len; 279 enc->cur += len;
168} 280}
169 281
282static void
283encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
284{
285 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
286 {
287 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
288
289 if (SvOK (*svp))
290 {
291 // already registered, use stringref
292 encode_tag (enc, CBOR_TAG_STRINGREF);
293 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
294 return;
295 }
296 else if (len >= minimum_string_length (enc->stringref_idx))
297 {
298 // register only
299 sv_setuv (*svp, enc->stringref_idx);
300 ++enc->stringref_idx;
301 }
302 }
303
304 encode_str (enc, utf8, str, len);
305}
306
170static void encode_sv (enc_t *enc, SV *sv); 307static void encode_sv (enc_t *enc, SV *sv);
171 308
172static void 309static void
173encode_av (enc_t *enc, AV *av) 310encode_av (enc_t *enc, AV *av)
174{ 311{
177 if (enc->depth >= enc->cbor.max_depth) 314 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 315 croak (ERR_NESTING_EXCEEDED);
179 316
180 ++enc->depth; 317 ++enc->depth;
181 318
182 encode_uint (enc, 0x80, len + 1); 319 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 320
184 for (i = 0; i <= len; ++i) 321 for (i = 0; i <= len; ++i)
185 { 322 {
186 SV **svp = av_fetch (av, i, 0); 323 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 324 encode_sv (enc, svp ? *svp : &PL_sv_undef);
202 339
203 int pairs = hv_iterinit (hv); 340 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 341 int mg = SvMAGICAL (hv);
205 342
206 if (mg) 343 if (mg)
207 encode_ch (enc, 0xa0 | 31); 344 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 345 else
209 encode_uint (enc, 0xa0, pairs); 346 encode_uint (enc, MAJOR_MAP, pairs);
210 347
211 while ((he = hv_iternext (hv))) 348 while ((he = hv_iternext (hv)))
212 { 349 {
213 if (HeKLEN (he) == HEf_SVKEY) 350 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 351 encode_sv (enc, HeSVKEY (he));
215 else 352 else
216 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 353 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
217 354
218 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 355 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
219 } 356 }
220 357
221 if (mg) 358 if (mg)
222 encode_ch (enc, 0xe0 | 31); 359 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 360
224 --enc->depth; 361 --enc->depth;
225} 362}
226 363
227// encode objects, arrays and special \0=false and \1=true values. 364// encode objects, arrays and special \0=false and \1=true values.
228static void 365static void
229encode_rv (enc_t *enc, SV *sv) 366encode_rv (enc_t *enc, SV *sv)
230{ 367{
231 svtype svt;
232
233 SvGETMAGIC (sv); 368 SvGETMAGIC (sv);
369
234 svt = SvTYPE (sv); 370 svtype svt = SvTYPE (sv);
235 371
236 if (ecb_expect_false (SvOBJECT (sv))) 372 if (ecb_expect_false (SvOBJECT (sv)))
237 { 373 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 374 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 375 ? types_boolean_stash
376 : gv_stashpv ("Types::Serialiser::Boolean", 1);
377 HV *error_stash = !CBOR_SLOW || types_error_stash
378 ? types_error_stash
379 : gv_stashpv ("Types::Serialiser::Error", 1);
380 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
381 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 382 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 383
242 if (SvSTASH (sv) == stash) 384 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 385
386 if (stash == boolean_stash)
387 {
388 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
389 return;
390 }
391 else if (stash == error_stash)
392 {
393 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 return;
395 }
396 else if (stash == tagged_stash)
397 {
398 if (svt != SVt_PVAV)
399 croak ("encountered CBOR::XS::Tagged object that isn't an array");
400
401 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
402 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
403
404 return;
405 }
406 }
407
408 if (ecb_expect_false (SvREFCNT (sv) > 1)
409 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
410 {
411 if (!enc->shareable)
412 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
413
414 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
415
416 if (SvOK (*svp))
417 {
418 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
419 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
420 return;
421 }
244 else 422 else
245 { 423 {
246#if 0 //TODO 424 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 425 ++enc->shareable_idx;
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
427 }
428 }
429
430 if (ecb_expect_false (SvOBJECT (sv)))
431 {
432 HV *stash = SvSTASH (sv);
433 GV *method;
434
435 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 436 {
437 dSP;
438
439 ENTER; SAVETMPS; PUSHMARK (SP);
249 // we re-bless the reference to get overload and other niceties right 440 // 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))); 441 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 442
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 443 PUTBACK;
444 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 445 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 446 SPAGAIN;
263 447
264 // catch this surprisingly common error 448 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 449 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))); 450 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 451
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 452 encode_sv (enc, POPs);
272 453
454 PUTBACK;
455
273 FREETMPS; LEAVE; 456 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 } 457 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 458 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
282 encode_str (enc, "null", 4, 0); 459 {
460 dSP;
461
462 ENTER; SAVETMPS; PUSHMARK (SP);
463 EXTEND (SP, 2);
464 // we re-bless the reference to get overload and other niceties right
465 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
466 PUSHs (sv_cbor);
467
468 PUTBACK;
469 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
470 SPAGAIN;
471
472 // catch this surprisingly common error
473 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
474 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
475
476 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
477 encode_uint (enc, MAJOR_ARRAY, count + 1);
478 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
479
480 while (count)
481 encode_sv (enc, SP[1 - count--]);
482
483 PUTBACK;
484
485 FREETMPS; LEAVE;
486 }
283 else 487 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 488 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 489 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 490 }
289 else if (svt == SVt_PVHV) 491 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 492 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 493 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 494 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 495 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 496 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 497 encode_tag (enc, CBOR_TAG_INDIRECTION);
498 encode_sv (enc, sv);
499 }
313} 500}
314 501
315static void 502static void
316encode_nv (enc_t *enc, SV *sv) 503encode_nv (enc_t *enc, SV *sv)
317{ 504{
318 double nv = SvNVX (sv); 505 double nv = SvNVX (sv);
319 506
320 need (enc, 9); 507 need (enc, 9);
321 508
322 if (ecb_expect_false (nv == (U32)nv)) 509 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 510 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 511 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 512 else if (ecb_expect_false (nv == (float)nv))
326 { 513 {
327 uint32_t fp = ecb_float_to_binary32 (nv); 514 uint32_t fp = ecb_float_to_binary32 (nv);
328 515
329 *enc->cur++ = 0xe0 | 26; 516 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
330 517
331 if (!ecb_big_endian ()) 518 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp); 519 fp = ecb_bswap32 (fp);
333 520
334 memcpy (enc->cur, &fp, 4); 521 memcpy (enc->cur, &fp, 4);
336 } 523 }
337 else 524 else
338 { 525 {
339 uint64_t fp = ecb_double_to_binary64 (nv); 526 uint64_t fp = ecb_double_to_binary64 (nv);
340 527
341 *enc->cur++ = 0xe0 | 27; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
342 529
343 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp); 531 fp = ecb_bswap64 (fp);
345 532
346 memcpy (enc->cur, &fp, 8); 533 memcpy (enc->cur, &fp, 8);
355 542
356 if (SvPOKp (sv)) 543 if (SvPOKp (sv))
357 { 544 {
358 STRLEN len; 545 STRLEN len;
359 char *str = SvPV (sv, len); 546 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 547 encode_strref (enc, SvUTF8 (sv), str, len);
361 } 548 }
362 else if (SvNOKp (sv)) 549 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 550 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 551 else if (SvIOKp (sv))
365 { 552 {
366 if (SvIsUV (sv)) 553 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv)); 554 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
368 else if (SvIVX (sv) >= 0) 555 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv)); 556 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
370 else 557 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 558 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
372 } 559 }
373 else if (SvROK (sv)) 560 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 561 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 562 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 563 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 564 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 565 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 566 else
380 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 567 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 568 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 569}
383 570
384static SV * 571static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 572encode_cbor (SV *scalar, CBOR *cbor)
386{ 573{
387 enc_t enc; 574 enc_t enc = { };
388 575
389 enc.cbor = *cbor; 576 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 577 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 578 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 579 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 580
395 SvPOK_only (enc.sv); 581 SvPOK_only (enc.sv);
582
583 if (cbor->flags & F_PACK_STRINGS)
584 {
585 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
586 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
587 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
588 }
589
396 encode_sv (&enc, scalar); 590 encode_sv (&enc, scalar);
397 591
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 592 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 593 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 594
414 U8 *end; // end of input string 608 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 609 const char *err; // parse error, if != 0
416 CBOR cbor; 610 CBOR cbor;
417 U32 depth; // recursion depth 611 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 612 U32 maxdepth; // recursion depth limit
613 AV *shareable;
614 AV *stringref;
615 SV *decode_tagged;
419} dec_t; 616} dec_t;
420 617
421#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 618#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
422 619
423#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 620#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 623#define DEC_DEC_DEPTH --dec->depth
427 624
428static UV 625static UV
429decode_uint (dec_t *dec) 626decode_uint (dec_t *dec)
430{ 627{
431 switch (*dec->cur & 31) 628 U8 m = *dec->cur & MINOR_MASK;
432 { 629 ++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;
437 630
438 case 24: 631 if (ecb_expect_true (m < LENGTH_EXT1))
632 return m;
633 else if (ecb_expect_true (m == LENGTH_EXT1))
634 {
439 WANT (2); 635 WANT (1);
440 dec->cur += 2; 636 dec->cur += 1;
441 return dec->cur[-1]; 637 return dec->cur[-1];
442 638 }
443 case 25: 639 else if (ecb_expect_true (m == LENGTH_EXT2))
640 {
444 WANT (3); 641 WANT (2);
445 dec->cur += 3; 642 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 643 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 644 | ((UV)dec->cur[-1]);
448 645 }
449 case 26: 646 else if (ecb_expect_true (m == LENGTH_EXT4))
647 {
450 WANT (5); 648 WANT (4);
451 dec->cur += 5; 649 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 650 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 651 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 652 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 653 | ((UV)dec->cur[-1]);
456 654 }
457 case 27: 655 else if (ecb_expect_true (m == LENGTH_EXT8))
656 {
458 WANT (9); 657 WANT (8);
459 dec->cur += 9; 658 dec->cur += 8;
659
660 return
661#if UVSIZE < 8
662 0
663#else
460 return (((UV)dec->cur[-8]) << 56) 664 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 665 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 666 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 667 | (((UV)dec->cur[-5]) << 32)
668#endif
464 | (((UV)dec->cur[-4]) << 24) 669 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 670 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 671 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 672 | ((UV)dec->cur[-1]);
468 673 }
469 default: 674 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 675 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 676
473fail: 677fail:
474 return 0; 678 return 0;
475} 679}
476 680
481{ 685{
482 AV *av = newAV (); 686 AV *av = newAV ();
483 687
484 DEC_INC_DEPTH; 688 DEC_INC_DEPTH;
485 689
486 if ((*dec->cur & 31) == 31) 690 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 691 {
488 ++dec->cur; 692 ++dec->cur;
489 693
490 for (;;) 694 for (;;)
491 { 695 {
492 WANT (1); 696 WANT (1);
493 697
494 if (*dec->cur == (0xe0 | 31)) 698 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
495 { 699 {
496 ++dec->cur; 700 ++dec->cur;
497 break; 701 break;
498 } 702 }
499 703
502 } 706 }
503 else 707 else
504 { 708 {
505 int i, len = decode_uint (dec); 709 int i, len = decode_uint (dec);
506 710
711 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); 712 av_fill (av, len - 1);
508 713
509 for (i = 0; i < len; ++i) 714 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 715 AvARRAY (av)[i] = decode_sv (dec);
511 } 716 }
517 SvREFCNT_dec (av); 722 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 723 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 724 return &PL_sv_undef;
520} 725}
521 726
727static void
728decode_he (dec_t *dec, HV *hv)
729{
730 // for speed reasons, we specialcase single-string
731 // byte or utf-8 strings as keys, but only when !stringref
732
733 if (ecb_expect_true (!dec->stringref))
734 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
735 {
736 I32 len = decode_uint (dec);
737 char *key = (char *)dec->cur;
738
739 dec->cur += len;
740
741 if (ecb_expect_false (dec->stringref))
742 av_push (dec->stringref, newSVpvn (key, len));
743
744 hv_store (hv, key, len, decode_sv (dec), 0);
745
746 return;
747 }
748 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
749 {
750 I32 len = decode_uint (dec);
751 char *key = (char *)dec->cur;
752
753 dec->cur += len;
754
755 if (ecb_expect_false (dec->stringref))
756 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
757
758 hv_store (hv, key, -len, decode_sv (dec), 0);
759
760 return;
761 }
762
763 SV *k = decode_sv (dec);
764 SV *v = decode_sv (dec);
765
766 hv_store_ent (hv, k, v, 0);
767 SvREFCNT_dec (k);
768}
769
522static SV * 770static SV *
523decode_hv (dec_t *dec) 771decode_hv (dec_t *dec)
524{ 772{
525 HV *hv = newHV (); 773 HV *hv = newHV ();
526 774
527 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
528 776
529 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 778 {
531 ++dec->cur; 779 ++dec->cur;
532 780
533 for (;;) 781 for (;;)
534 { 782 {
535 WANT (1); 783 WANT (1);
536 784
537 if (*dec->cur == (0xe0 | 31)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
538 { 786 {
539 ++dec->cur; 787 ++dec->cur;
540 break; 788 break;
541 } 789 }
542 790
543 SV *k = decode_sv (dec); 791 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 792 }
548 } 793 }
549 else 794 else
550 { 795 {
551 int len = decode_uint (dec); 796 int pairs = decode_uint (dec);
552 797
553 while (len--) 798 while (pairs--)
554 { 799 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 } 800 }
561 801
562 DEC_DEC_DEPTH; 802 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 803 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
719 804
720fail: 805fail:
721 SvREFCNT_dec (hv); 806 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 807 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 808 return &PL_sv_undef;
724} 809}
725 810
726static SV * 811static SV *
727decode_str (dec_t *dec, int utf8) 812decode_str (dec_t *dec, int utf8)
728{ 813{
729 SV *sv; 814 SV *sv = 0;
730 815
731 if ((*dec->cur & 31) == 31) 816 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 817 {
818 // indefinite length strings
733 ++dec->cur; 819 ++dec->cur;
734 820
821 U8 major = *dec->cur & MAJOR_MISC;
822
735 sv = newSVpvn ("", 0); 823 sv = newSVpvn ("", 0);
736 824
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 825 for (;;)
739 { 826 {
740 WANT (1); 827 WANT (1);
741 828
742 if (*dec->cur == (0xe0 | 31)) 829 if ((*dec->cur - major) > LENGTH_EXT8)
830 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 831 {
744 ++dec->cur; 832 ++dec->cur;
745 break; 833 break;
746 } 834 }
835 else
836 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 837
748 SV *sv2 = decode_sv (dec); 838 STRLEN len = decode_uint (dec);
749 sv_catsv (sv, sv2); 839
840 WANT (len);
841 sv_catpvn (sv, dec->cur, len);
842 dec->cur += len;
750 } 843 }
751 } 844 }
752 else 845 else
753 { 846 {
754 STRLEN len = decode_uint (dec); 847 STRLEN len = decode_uint (dec);
755 848
756 WANT (len); 849 WANT (len);
757 sv = newSVpvn (dec->cur, len); 850 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 851 dec->cur += len;
852
853 if (ecb_expect_false (dec->stringref)
854 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
855 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 856 }
760 857
761 if (utf8) 858 if (utf8)
762 SvUTF8_on (sv); 859 SvUTF8_on (sv);
763 860
764 return sv; 861 return sv;
765 862
766fail: 863fail:
864 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 865 return &PL_sv_undef;
768} 866}
769 867
770static SV * 868static SV *
771decode_tagged (dec_t *dec) 869decode_tagged (dec_t *dec)
772{ 870{
871 SV *sv = 0;
773 UV tag = decode_uint (dec); 872 UV tag = decode_uint (dec);
873
874 WANT (1);
875
876 switch (tag)
877 {
878 case CBOR_TAG_MAGIC:
774 SV *sv = decode_sv (dec); 879 sv = decode_sv (dec);
880 break;
775 881
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 882 case CBOR_TAG_INDIRECTION:
883 sv = newRV_noinc (decode_sv (dec));
884 break;
885
886 case CBOR_TAG_STRINGREF_NAMESPACE:
887 {
888 ENTER; SAVETMPS;
889
890 SAVESPTR (dec->stringref);
891 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
892
893 sv = decode_sv (dec);
894
895 FREETMPS; LEAVE;
896 }
897 break;
898
899 case CBOR_TAG_STRINGREF:
900 {
901 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
902 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
903
904 UV idx = decode_uint (dec);
905
906 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
907 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
908
909 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
910 }
911 break;
912
913 case CBOR_TAG_VALUE_SHAREABLE:
914 {
915 if (ecb_expect_false (!dec->shareable))
916 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
917
918 if (dec->cbor.flags & F_ALLOW_CYCLES)
919 {
920 sv = newSV (0);
921 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
922
923 SV *osv = decode_sv (dec);
924 sv_setsv (sv, osv);
925 SvREFCNT_dec_NN (osv);
926 }
927 else
928 {
929 av_push (dec->shareable, &PL_sv_undef);
930 int idx = AvFILLp (dec->shareable);
931 sv = decode_sv (dec);
932 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
933 }
934 }
935 break;
936
937 case CBOR_TAG_VALUE_SHAREDREF:
938 {
939 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
940 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
941
942 UV idx = decode_uint (dec);
943
944 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
945 ERR ("corrupted CBOR data (sharedref index out of bounds)");
946
947 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
948
949 if (sv == &PL_sv_undef)
950 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
951 }
952 break;
953
954 case CBOR_TAG_PERL_OBJECT:
955 {
956 sv = decode_sv (dec);
957
958 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
959 ERR ("corrupted CBOR data (non-array perl object)");
960
961 AV *av = (AV *)SvRV (sv);
962 int len = av_len (av) + 1;
963 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
964
965 if (!stash)
966 ERR ("cannot decode perl-object (package does not exist)");
967
968 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
969
970 if (!method)
971 ERR ("cannot decode perl-object (package does not have a THAW method)");
972
973 dSP;
974
975 ENTER; SAVETMPS; PUSHMARK (SP);
976 EXTEND (SP, len + 1);
977 // we re-bless the reference to get overload and other niceties right
978 PUSHs (*av_fetch (av, 0, 1));
979 PUSHs (sv_cbor);
980
981 int i;
982
983 for (i = 1; i < len; ++i)
984 PUSHs (*av_fetch (av, i, 1));
985
986 PUTBACK;
987 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
988 SPAGAIN;
989
990 if (SvTRUE (ERRSV))
991 {
992 FREETMPS; LEAVE;
993 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
994 }
995
996 SvREFCNT_dec (sv);
997 sv = SvREFCNT_inc (POPs);
998
999 PUTBACK;
1000
1001 FREETMPS; LEAVE;
1002 }
1003 break;
1004
1005 default:
1006 {
1007 sv = decode_sv (dec);
1008
1009 dSP;
1010 ENTER; SAVETMPS; PUSHMARK (SP);
1011 EXTEND (SP, 2);
1012 PUSHs (newSVuv (tag));
1013 PUSHs (sv);
1014
1015 PUTBACK;
1016 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1017 SPAGAIN;
1018
1019 if (SvTRUE (ERRSV))
1020 {
1021 FREETMPS; LEAVE;
1022 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1023 }
1024
1025 if (count)
1026 {
1027 SvREFCNT_dec (sv);
1028 sv = SvREFCNT_inc (POPs);
1029 }
1030 else
1031 {
1032 AV *av = newAV ();
1033 av_push (av, newSVuv (tag));
1034 av_push (av, sv);
1035
1036 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1037 ? cbor_tagged_stash
1038 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1039 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1040 }
1041
1042 PUTBACK;
1043
1044 FREETMPS; LEAVE;
1045 }
1046 break;
1047 }
1048
777 return sv; 1049 return sv;
778 1050
779 AV *av = newAV (); 1051fail:
780 av_push (av, newSVuv (tag)); 1052 SvREFCNT_dec (sv);
781 av_push (av, sv); 1053 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1054}
784 1055
785static SV * 1056static SV *
786decode_sv (dec_t *dec) 1057decode_sv (dec_t *dec)
787{ 1058{
788 WANT (1); 1059 WANT (1);
789 1060
790 switch (*dec->cur >> 5) 1061 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1062 {
792 case 0: // unsigned int 1063 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1064 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1065 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1066 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1067 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1068 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1069 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1070
800 return decode_str (dec, 1); 1071 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) 1072 switch (*dec->cur++ & MINOR_MASK)
809 { 1073 {
810 case 20: 1074 case SIMPLE_FALSE:
811#if CBOR_SLOW 1075#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1076 types_false = get_bool ("Types::Serialiser::false");
813#endif 1077#endif
814 return newSVsv (cbor_false); 1078 return newSVsv (types_false);
815 case 21: 1079 case SIMPLE_TRUE:
816#if CBOR_SLOW 1080#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1081 types_true = get_bool ("Types::Serialiser::true");
818#endif 1082#endif
819 return newSVsv (cbor_true); 1083 return newSVsv (types_true);
820 case 22: 1084 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1085 return newSVsv (&PL_sv_undef);
1086 case SIMPLE_UNDEF:
1087#if CBOR_SLOW
1088 types_error = get_bool ("Types::Serialiser::error");
1089#endif
1090 return newSVsv (types_error);
822 1091
823 case 25: 1092 case MISC_FLOAT16:
824 { 1093 {
825 WANT (2); 1094 WANT (2);
826 1095
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1096 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1097 dec->cur += 2;
829 1098
830 return newSVnv (ecb_binary16_to_float (fp)); 1099 return newSVnv (ecb_binary16_to_float (fp));
831 } 1100 }
832 1101
833 case 26: 1102 case MISC_FLOAT32:
834 { 1103 {
835 uint32_t fp; 1104 uint32_t fp;
836 WANT (4); 1105 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1106 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1107 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1110 fp = ecb_bswap32 (fp);
842 1111
843 return newSVnv (ecb_binary32_to_float (fp)); 1112 return newSVnv (ecb_binary32_to_float (fp));
844 } 1113 }
845 1114
846 case 27: 1115 case MISC_FLOAT64:
847 { 1116 {
848 uint64_t fp; 1117 uint64_t fp;
849 WANT (8); 1118 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1119 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1120 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1123 fp = ecb_bswap64 (fp);
855 1124
856 return newSVnv (ecb_binary64_to_double (fp)); 1125 return newSVnv (ecb_binary64_to_double (fp));
857 } 1126 }
858 1127
859 // 0..19 unassigned 1128 // 0..19 unassigned simple
860 // 24 reserved + unassigned (reserved values are not encodable) 1129 // 24 reserved + unassigned (reserved values are not encodable)
861 default: 1130 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1131 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
863 } 1132 }
864 1133
865 break; 1134 break;
866 } 1135 }
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 1136
924fail: 1137fail:
925 return &PL_sv_undef; 1138 return &PL_sv_undef;
926} 1139}
927 1140
928static SV * 1141static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1142decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1143{
931 dec_t dec; 1144 dec_t dec = { };
932 SV *sv; 1145 SV *sv;
1146 STRLEN len;
1147 char *data = SvPVbyte (string, len);
933 1148
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) 1149 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", 1150 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); 1151 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1152
969 dec.cbor = *cbor; 1153 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1154 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1155 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 1156
978 sv = decode_sv (&dec); 1157 sv = decode_sv (&dec);
979 1158
980 if (offset_return) 1159 if (offset_return)
981 *offset_return = dec.cur; 1160 *offset_return = dec.cur;
985 dec.err = "garbage after CBOR object"; 1164 dec.err = "garbage after CBOR object";
986 1165
987 if (dec.err) 1166 if (dec.err)
988 { 1167 {
989 SvREFCNT_dec (sv); 1168 SvREFCNT_dec (sv);
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1169 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
991 } 1170 }
992 1171
993 sv = sv_2mortal (sv); 1172 sv = sv_2mortal (sv);
994 1173
995 return sv; 1174 return sv;
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1180MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1181
1003BOOT: 1182BOOT:
1004{ 1183{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1184 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1185 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1186
1008 cbor_true = get_bool ("CBOR::XS::true"); 1187 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1188 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1189
1190 types_true = get_bool ("Types::Serialiser::true" );
1191 types_false = get_bool ("Types::Serialiser::false");
1192 types_error = get_bool ("Types::Serialiser::error");
1193
1194 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1195
1196 sv_cbor = newSVpv ("CBOR", 0);
1197 SvREADONLY_on (sv_cbor);
1010} 1198}
1011 1199
1012PROTOTYPES: DISABLE 1200PROTOTYPES: DISABLE
1013 1201
1014void CLONE (...) 1202void CLONE (...)
1015 CODE: 1203 CODE:
1016 cbor_stash = 0; 1204 cbor_stash = 0;
1205 cbor_tagged_stash = 0;
1206 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1207 types_boolean_stash = 0;
1018 1208
1019void new (char *klass) 1209void new (char *klass)
1020 PPCODE: 1210 PPCODE:
1021{ 1211{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1212 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1220
1031void shrink (CBOR *self, int enable = 1) 1221void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1222 ALIAS:
1033 shrink = F_SHRINK 1223 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1224 allow_unknown = F_ALLOW_UNKNOWN
1225 allow_sharing = F_ALLOW_SHARING
1226 allow_cycles = F_ALLOW_CYCLES
1227 pack_strings = F_PACK_STRINGS
1035 PPCODE: 1228 PPCODE:
1036{ 1229{
1037 if (enable) 1230 if (enable)
1038 self->flags |= ix; 1231 self->flags |= ix;
1039 else 1232 else
1044 1237
1045void get_shrink (CBOR *self) 1238void get_shrink (CBOR *self)
1046 ALIAS: 1239 ALIAS:
1047 get_shrink = F_SHRINK 1240 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1241 get_allow_unknown = F_ALLOW_UNKNOWN
1242 get_allow_sharing = F_ALLOW_SHARING
1243 get_allow_cycles = F_ALLOW_CYCLES
1244 get_pack_strings = F_PACK_STRINGS
1049 PPCODE: 1245 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1246 XPUSHs (boolSV (self->flags & ix));
1051 1247
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1248void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1249 PPCODE:
1069 CODE: 1265 CODE:
1070 RETVAL = self->max_size; 1266 RETVAL = self->max_size;
1071 OUTPUT: 1267 OUTPUT:
1072 RETVAL 1268 RETVAL
1073 1269
1074#if 0 //TODO 1270void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1271 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1272 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1273 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1274 XPUSHs (ST (0));
1083}
1084 1275
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1276SV *get_filter (CBOR *self)
1086 PPCODE: 1277 CODE:
1087{ 1278 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1279 OUTPUT:
1089 self->cb_sk_object = newHV (); 1280 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 1281
1109void encode (CBOR *self, SV *scalar) 1282void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1283 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1284 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1285 XPUSHs (scalar);
1126 PUSHs (sv); 1299 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1300 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1301}
1129 1302
1130void DESTROY (CBOR *self) 1303void DESTROY (CBOR *self)
1131 CODE: 1304 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1305 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1306
1135PROTOTYPES: ENABLE 1307PROTOTYPES: ENABLE
1136 1308
1137void encode_cbor (SV *scalar) 1309void encode_cbor (SV *scalar)
1310 ALIAS:
1311 encode_cbor = 0
1312 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1313 PPCODE:
1139{ 1314{
1140 CBOR cbor; 1315 CBOR cbor;
1141 cbor_init (&cbor); 1316 cbor_init (&cbor);
1317 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1318 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1319 XPUSHs (scalar);
1144} 1320}
1145 1321
1146void decode_cbor (SV *cborstr) 1322void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines