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 1.28 by root, Sat Nov 23 18:30:59 2013 UTC vs.
Revision 1.61 by root, Sat Nov 26 02:10:19 2016 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
23#endif 25#endif
24#ifndef SvREFCNT_dec_NN 26#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 27# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 28#endif
27 29
30// known major and minor types
31enum cbor_type
32{
33 MAJOR_SHIFT = 5,
34 MINOR_MASK = 0x1f,
35
36 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
37 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
38 MAJOR_BYTES = 2 << MAJOR_SHIFT,
39 MAJOR_TEXT = 3 << MAJOR_SHIFT,
40 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
41 MAJOR_MAP = 5 << MAJOR_SHIFT,
42 MAJOR_TAG = 6 << MAJOR_SHIFT,
43 MAJOR_MISC = 7 << MAJOR_SHIFT,
44
45 // INT/STRING/ARRAY/MAP subtypes
46 LENGTH_EXT1 = 24,
47 LENGTH_EXT2 = 25,
48 LENGTH_EXT4 = 26,
49 LENGTH_EXT8 = 27,
50
51 // SIMPLE types (effectively MISC subtypes)
52 SIMPLE_FALSE = 20,
53 SIMPLE_TRUE = 21,
54 SIMPLE_NULL = 22,
55 SIMPLE_UNDEF = 23,
56
57 // MISC subtype (unused)
58 MISC_EXT1 = 24,
59 MISC_FLOAT16 = 25,
60 MISC_FLOAT32 = 26,
61 MISC_FLOAT64 = 27,
62
63 // BYTES/TEXT/ARRAY/MAP
64 MINOR_INDEF = 31,
65};
66
28// known tags 67// known tags
29enum cbor_tag 68enum cbor_tag
30{ 69{
31 // inofficial extensions (pending iana registration) 70 // extensions
71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
32 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 72 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
33 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 73 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
34 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
35 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 75 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 76 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
37 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
38 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 77 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 78
40 // rfc7049 79 // rfc7049
41 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 80 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 81 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43 CBOR_TAG_POS_BIGNUM = 2, // byte string 82 CBOR_TAG_POS_BIGNUM = 2, // byte string
44 CBOR_TAG_NEG_BIGNUM = 3, // byte string 83 CBOR_TAG_NEG_BIGNUM = 3, // byte string
45 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 84 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46 CBOR_TAG_BIGFLOAT = 5, // array 85 CBOR_TAG_BIGFLOAT = 5, // array
47 86
48 CBOR_TAG_CONV_B64U = 21, // base64url, any 87 CBOR_TAG_CONV_B64U = 21, // base64url, any
49 CBOR_TAG_CONV_B64 = 22, // base64, any 88 CBOR_TAG_CONV_B64 = 22, // base64, any
50 CBOR_TAG_CONV_HEX = 23, // base16, any 89 CBOR_TAG_CONV_HEX = 23, // base16, any
51 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 90 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52 91
53 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 92 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 93 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 94 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 95 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 96 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 97
59 CBOR_TAG_MAGIC = 55799 // self-describe cbor 98 CBOR_TAG_MAGIC = 55799, // self-describe cbor
60}; 99};
61 100
62#define F_SHRINK 0x00000001UL 101#define F_SHRINK 0x00000001UL
63#define F_ALLOW_UNKNOWN 0x00000002UL 102#define F_ALLOW_UNKNOWN 0x00000002UL
64#define F_ALLOW_SHARING 0x00000004UL //TODO 103#define F_ALLOW_SHARING 0x00000004UL
65#define F_ALLOW_STRINGREF 0x00000008UL //TODO 104#define F_ALLOW_CYCLES 0x00000008UL
105#define F_FORBID_OBJECTS 0x00000010UL
106#define F_PACK_STRINGS 0x00000020UL
107#define F_TEXT_KEYS 0x00000040UL
108#define F_TEXT_STRINGS 0x00000080UL
109#define F_VALIDATE_UTF8 0x00000100UL
66 110
67#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
68 112
69#define SB do { 113#define SB do {
70#define SE } while (0) 114#define SE } while (0)
89typedef struct { 133typedef struct {
90 U32 flags; 134 U32 flags;
91 U32 max_depth; 135 U32 max_depth;
92 STRLEN max_size; 136 STRLEN max_size;
93 SV *filter; 137 SV *filter;
138
139 // for the incremental parser
140 STRLEN incr_pos; // the current offset into the text
141 STRLEN incr_need; // minimum bytes needed to decode
142 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
94} CBOR; 143} CBOR;
95 144
96ecb_inline void 145ecb_inline void
97cbor_init (CBOR *cbor) 146cbor_init (CBOR *cbor)
98{ 147{
102 151
103ecb_inline void 152ecb_inline void
104cbor_free (CBOR *cbor) 153cbor_free (CBOR *cbor)
105{ 154{
106 SvREFCNT_dec (cbor->filter); 155 SvREFCNT_dec (cbor->filter);
156 SvREFCNT_dec (cbor->incr_count);
107} 157}
108 158
109///////////////////////////////////////////////////////////////////////////// 159/////////////////////////////////////////////////////////////////////////////
110// utility functions 160// utility functions
111 161
137 187
138// minimum length of a string to be registered for stringref 188// minimum length of a string to be registered for stringref
139ecb_inline int 189ecb_inline int
140minimum_string_length (UV idx) 190minimum_string_length (UV idx)
141{ 191{
142 return idx > 23 192
143 ? idx > 0xffU 193 return idx <= 23 ? 3
144 ? idx > 0xffffU 194 : idx <= 0xffU ? 4
195 : idx <= 0xffffU ? 5
145 ? idx > 0xffffffffU 196 : idx <= 0xffffffffU ? 7
146 ? 7 197 : 11;
147 : 6
148 : 5
149 : 4
150 : 3;
151} 198}
152 199
153///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
154// encoder 201// encoder
155 202
168} enc_t; 215} enc_t;
169 216
170ecb_inline void 217ecb_inline void
171need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
172{ 219{
173 if (ecb_expect_false (enc->cur + len >= enc->end)) 220 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
174 { 221 {
175 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177 enc->cur = SvPVX (enc->sv) + cur; 224 enc->cur = SvPVX (enc->sv) + cur;
178 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
189static void 236static void
190encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
191{ 238{
192 need (enc, 9); 239 need (enc, 9);
193 240
194 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
196 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
197 { 244 {
198 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 246 *enc->cur++ = len;
200 } 247 }
201 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
202 { 249 {
203 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 252 *enc->cur++ = len;
206 } 253 }
207 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
208 { 255 {
209 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 260 *enc->cur++ = len;
214 } 261 }
215 else 262 else
216 { 263 {
217 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
227} 274}
228 275
229ecb_inline void 276ecb_inline void
230encode_tag (enc_t *enc, UV tag) 277encode_tag (enc_t *enc, UV tag)
231{ 278{
232 encode_uint (enc, 0xc0, tag); 279 encode_uint (enc, MAJOR_TAG, tag);
233} 280}
234 281
235static void 282// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
283ecb_noinline static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 284encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 285{
286 STRLEN ulen = len;
287 U8 *p, *pend = (U8 *)str + len;
288
289 for (p = (U8 *)str; p < pend; ++p)
290 ulen += *p >> 7; // count set high bits
291
292 encode_uint (enc, MAJOR_TEXT, ulen);
293
294 need (enc, ulen);
295 for (p = (U8 *)str; p < pend; ++p)
296 if (*p < 0x80)
297 *enc->cur++ = *p;
298 else
299 {
300 *enc->cur++ = 0xc0 + (*p >> 6);
301 *enc->cur++ = 0x80 + (*p & 63);
302 }
303}
304
305ecb_inline void
306encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
307{
308 if (ecb_expect_false (upgrade_utf8))
309 if (!utf8)
310 {
311 encode_str_utf8 (enc, utf8, str, len);
312 return;
313 }
314
315 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
316 need (enc, len);
317 memcpy (enc->cur, str, len);
318 enc->cur += len;
319}
320
321ecb_inline void
322encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
323{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 324 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 325 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 326 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 327
242 if (SvOK (*svp)) 328 if (SvOK (*svp))
243 { 329 {
244 // already registered, use stringref 330 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 331 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 332 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 333 return;
248 } 334 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 335 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 336 {
251 // register only 337 // register only
252 sv_setuv (*svp, enc->stringref_idx); 338 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 339 ++enc->stringref_idx;
254 } 340 }
255 } 341 }
256 342
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 343 encode_str (enc, upgrade_utf8, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 344}
262 345
263static void encode_sv (enc_t *enc, SV *sv); 346static void encode_sv (enc_t *enc, SV *sv);
264 347
265static void 348static void
270 if (enc->depth >= enc->cbor.max_depth) 353 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 354 croak (ERR_NESTING_EXCEEDED);
272 355
273 ++enc->depth; 356 ++enc->depth;
274 357
275 encode_uint (enc, 0x80, len + 1); 358 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 359
360 if (SvMAGICAL (av))
277 for (i = 0; i <= len; ++i) 361 for (i = 0; i <= len; ++i)
278 { 362 {
279 SV **svp = av_fetch (av, i, 0); 363 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 364 encode_sv (enc, svp ? *svp : &PL_sv_undef);
281 } 365 }
366 else
367 for (i = 0; i <= len; ++i)
368 {
369 SV *sv = AvARRAY (av)[i];
370 encode_sv (enc, sv ? sv : &PL_sv_undef);
371 }
282 372
283 --enc->depth; 373 --enc->depth;
284} 374}
285 375
286static void 376static void
295 385
296 int pairs = hv_iterinit (hv); 386 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 387 int mg = SvMAGICAL (hv);
298 388
299 if (mg) 389 if (mg)
300 encode_ch (enc, 0xa0 | 31); 390 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 391 else
302 encode_uint (enc, 0xa0, pairs); 392 encode_uint (enc, MAJOR_MAP, pairs);
303 393
304 while ((he = hv_iternext (hv))) 394 while ((he = hv_iternext (hv)))
305 { 395 {
306 if (HeKLEN (he) == HEf_SVKEY) 396 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 397 encode_sv (enc, HeSVKEY (he));
308 else 398 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 399 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 400
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 401 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 402 }
313 403
314 if (mg) 404 if (mg)
315 encode_ch (enc, 0xe0 | 31); 405 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 406
317 --enc->depth; 407 --enc->depth;
318} 408}
319 409
320// encode objects, arrays and special \0=false and \1=true values. 410// encode objects, arrays and special \0=false and \1=true values.
321static void 411static void
322encode_rv (enc_t *enc, SV *sv) 412encode_rv (enc_t *enc, SV *sv)
323{ 413{
324 SvGETMAGIC (sv); 414 SvGETMAGIC (sv);
325
326 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
327 && ecb_expect_false (SvREFCNT (sv) > 1))
328 {
329 if (!enc->shareable)
330 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
331
332 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
333
334 if (SvOK (*svp))
335 {
336 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
337 encode_uint (enc, 0x00, SvUV (*svp));
338 return;
339 }
340 else
341 {
342 sv_setuv (*svp, enc->shareable_idx);
343 ++enc->shareable_idx;
344 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
345 }
346 }
347 415
348 svtype svt = SvTYPE (sv); 416 svtype svt = SvTYPE (sv);
349 417
350 if (ecb_expect_false (SvOBJECT (sv))) 418 if (ecb_expect_false (SvOBJECT (sv)))
351 { 419 {
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 426 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 427 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 428 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 429
362 HV *stash = SvSTASH (sv); 430 HV *stash = SvSTASH (sv);
363 GV *method;
364 431
365 if (stash == boolean_stash) 432 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 433 {
434 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
435 return;
436 }
367 else if (stash == error_stash) 437 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 438 {
439 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
440 return;
441 }
369 else if (stash == tagged_stash) 442 else if (stash == tagged_stash)
370 { 443 {
371 if (svt != SVt_PVAV) 444 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 445 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 446
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 447 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 448 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
449
450 return;
451 }
452 }
453
454 if (ecb_expect_false (SvREFCNT (sv) > 1)
455 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
456 {
457 if (!enc->shareable)
458 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
459
460 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
461
462 if (SvOK (*svp))
376 } 463 {
464 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
465 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
466 return;
467 }
468 else
469 {
470 sv_setuv (*svp, enc->shareable_idx);
471 ++enc->shareable_idx;
472 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
473 }
474 }
475
476 if (ecb_expect_false (SvOBJECT (sv)))
477 {
478 HV *stash = SvSTASH (sv);
479 GV *method;
480
481 if (enc->cbor.flags & F_FORBID_OBJECTS)
482 croak ("encountered object '%s', but forbid_objects is enabled",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 484 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 485 {
379 dSP; 486 dSP;
380 487
381 ENTER; SAVETMPS; PUSHMARK (SP); 488 ENTER; SAVETMPS;
489 PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 490 // we re-bless the reference to get overload and other niceties right
383 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 491 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384 492
385 PUTBACK; 493 PUTBACK;
386 // G_SCALAR ensures that return value is 1 494 // G_SCALAR ensures that return value is 1
399 } 507 }
400 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 508 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401 { 509 {
402 dSP; 510 dSP;
403 511
404 ENTER; SAVETMPS; PUSHMARK (SP); 512 ENTER; SAVETMPS;
513 SAVESTACK_POS ();
514 PUSHMARK (SP);
405 EXTEND (SP, 2); 515 EXTEND (SP, 2);
406 // we re-bless the reference to get overload and other niceties right 516 // we re-bless the reference to get overload and other niceties right
407 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 517 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408 PUSHs (sv_cbor); 518 PUSHs (sv_cbor);
409 519
414 // catch this surprisingly common error 524 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 525 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
416 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 526 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 527
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 528 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 529 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 530 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 531
422 while (count) 532 while (count)
423 encode_sv (enc, SP[1 - count--]); 533 encode_sv (enc, SP[1 - count--]);
424 534
425 PUTBACK; 535 PUTBACK;
446{ 556{
447 double nv = SvNVX (sv); 557 double nv = SvNVX (sv);
448 558
449 need (enc, 9); 559 need (enc, 9);
450 560
451 if (ecb_expect_false (nv == (U32)nv)) 561 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 562 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 563 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 564 else if (ecb_expect_false (nv == (float)nv))
455 { 565 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 566 uint32_t fp = ecb_float_to_binary32 (nv);
457 567
458 *enc->cur++ = 0xe0 | 26; 568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 569
460 if (!ecb_big_endian ()) 570 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 571 fp = ecb_bswap32 (fp);
462 572
463 memcpy (enc->cur, &fp, 4); 573 memcpy (enc->cur, &fp, 4);
465 } 575 }
466 else 576 else
467 { 577 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 578 uint64_t fp = ecb_double_to_binary64 (nv);
469 579
470 *enc->cur++ = 0xe0 | 27; 580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 581
472 if (!ecb_big_endian ()) 582 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 583 fp = ecb_bswap64 (fp);
474 584
475 memcpy (enc->cur, &fp, 8); 585 memcpy (enc->cur, &fp, 8);
484 594
485 if (SvPOKp (sv)) 595 if (SvPOKp (sv))
486 { 596 {
487 STRLEN len; 597 STRLEN len;
488 char *str = SvPV (sv, len); 598 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 599 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490 } 600 }
491 else if (SvNOKp (sv)) 601 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 602 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 603 else if (SvIOKp (sv))
494 { 604 {
495 if (SvIsUV (sv)) 605 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 607 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 609 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 610 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 611 }
502 else if (SvROK (sv)) 612 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 613 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 614 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 616 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 618 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 619 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 620 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 621}
512 622
513static SV * 623static SV *
514encode_cbor (SV *scalar, CBOR *cbor) 624encode_cbor (SV *scalar, CBOR *cbor)
515{ 625{
516 enc_t enc = { }; 626 enc_t enc = { 0 };
517 627
518 enc.cbor = *cbor; 628 enc.cbor = *cbor;
519 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 629 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520 enc.cur = SvPVX (enc.sv); 630 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 631 enc.end = SvEND (enc.sv);
522 632
523 SvPOK_only (enc.sv); 633 SvPOK_only (enc.sv);
524 634
525 if (cbor->flags & F_ALLOW_STRINGREF) 635 if (cbor->flags & F_PACK_STRINGS)
526 { 636 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 637 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 638 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 639 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 640 }
553 U32 depth; // recursion depth 663 U32 depth; // recursion depth
554 U32 maxdepth; // recursion depth limit 664 U32 maxdepth; // recursion depth limit
555 AV *shareable; 665 AV *shareable;
556 AV *stringref; 666 AV *stringref;
557 SV *decode_tagged; 667 SV *decode_tagged;
668 SV *err_sv; // optional sv for error, needs to be freed
558} dec_t; 669} dec_t;
559 670
560#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 671// set dec->err to ERRSV
672ecb_cold static void
673err_errsv (dec_t *dec)
674{
675 if (!dec->err)
676 {
677 dec->err_sv = newSVsv (ERRSV);
561 678
562#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 679 // chop off the trailing \n
680 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
681 *SvEND (dec->err_sv) = 0;
563 682
683 dec->err = SvPVutf8_nolen (dec->err_sv);
684 }
685}
686
687// the following functions are used to reduce code size and help the compiler to optimise
688ecb_cold static void
689err_set (dec_t *dec, const char *reason)
690{
691 if (!dec->err)
692 dec->err = reason;
693}
694
695ecb_cold static void
696err_unexpected_end (dec_t *dec)
697{
698 err_set (dec, "unexpected end of CBOR data");
699}
700
701ecb_cold static void
702err_nesting_exceeded (dec_t *dec)
703{
704 err_set (dec, ERR_NESTING_EXCEEDED);
705}
706
707#define ERR_DO(do) SB do; goto fail; SE
708#define ERR(reason) ERR_DO (err_set (dec, reason))
709#define ERR_ERRSV ERR_DO (err_errsv (dec))
710
711#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
712
564#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 713#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
565#define DEC_DEC_DEPTH --dec->depth 714#define DEC_DEC_DEPTH --dec->depth
566 715
567static UV 716static UV
568decode_uint (dec_t *dec) 717decode_uint (dec_t *dec)
569{ 718{
570 switch (*dec->cur & 31) 719 U8 m = *dec->cur & MINOR_MASK;
571 { 720 ++dec->cur;
572 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
573 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
574 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
575 return *dec->cur++ & 31;
576 721
577 case 24: 722 if (ecb_expect_true (m < LENGTH_EXT1))
723 return m;
724 else if (ecb_expect_true (m == LENGTH_EXT1))
725 {
578 WANT (2); 726 WANT (1);
579 dec->cur += 2; 727 dec->cur += 1;
580 return dec->cur[-1]; 728 return dec->cur[-1];
581 729 }
582 case 25: 730 else if (ecb_expect_true (m == LENGTH_EXT2))
731 {
583 WANT (3); 732 WANT (2);
584 dec->cur += 3; 733 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 734 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 735 | ((UV)dec->cur[-1]);
587 736 }
588 case 26: 737 else if (ecb_expect_true (m == LENGTH_EXT4))
738 {
589 WANT (5); 739 WANT (4);
590 dec->cur += 5; 740 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 741 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 742 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 743 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 744 | ((UV)dec->cur[-1]);
595 745 }
596 case 27: 746 else if (ecb_expect_true (m == LENGTH_EXT8))
747 {
597 WANT (9); 748 WANT (8);
598 dec->cur += 9; 749 dec->cur += 8;
750
751 return
752#if UVSIZE < 8
753 0
754#else
599 return (((UV)dec->cur[-8]) << 56) 755 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 756 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 757 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 758 | (((UV)dec->cur[-5]) << 32)
759#endif
603 | (((UV)dec->cur[-4]) << 24) 760 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 761 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 762 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 763 | ((UV)dec->cur[-1]);
607 764 }
608 default: 765 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 766 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 767
612fail: 768fail:
613 return 0; 769 return 0;
614} 770}
615 771
620{ 776{
621 AV *av = newAV (); 777 AV *av = newAV ();
622 778
623 DEC_INC_DEPTH; 779 DEC_INC_DEPTH;
624 780
625 if ((*dec->cur & 31) == 31) 781 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 782 {
627 ++dec->cur; 783 ++dec->cur;
628 784
629 for (;;) 785 for (;;)
630 { 786 {
631 WANT (1); 787 WANT (1);
632 788
633 if (*dec->cur == (0xe0 | 31)) 789 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 790 {
635 ++dec->cur; 791 ++dec->cur;
636 break; 792 break;
637 } 793 }
638 794
639 av_push (av, decode_sv (dec)); 795 av_push (av, decode_sv (dec));
640 } 796 }
641 } 797 }
642 else 798 else
643 { 799 {
644 int i, len = decode_uint (dec); 800 UV i, len = decode_uint (dec);
645 801
802 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
646 av_fill (av, len - 1); 803 av_fill (av, len - 1);
647 804
648 for (i = 0; i < len; ++i) 805 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 806 AvARRAY (av)[i] = decode_sv (dec);
650 } 807 }
663{ 820{
664 // for speed reasons, we specialcase single-string 821 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 822 // byte or utf-8 strings as keys, but only when !stringref
666 823
667 if (ecb_expect_true (!dec->stringref)) 824 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 825 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 826 {
670 I32 len = decode_uint (dec); 827 STRLEN len = decode_uint (dec);
671 char *key = (char *)dec->cur; 828 char *key = (char *)dec->cur;
672 829
830 WANT (len);
673 dec->cur += len; 831 dec->cur += len;
674 832
675 if (ecb_expect_false (dec->stringref))
676 av_push (dec->stringref, newSVpvn (key, len));
677
678 hv_store (hv, key, len, decode_sv (dec), 0); 833 hv_store (hv, key, len, decode_sv (dec), 0);
679 834
680 return; 835 return;
681 } 836 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 837 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 838 {
684 I32 len = decode_uint (dec); 839 STRLEN len = decode_uint (dec);
685 char *key = (char *)dec->cur; 840 char *key = (char *)dec->cur;
686 841
842 WANT (len);
687 dec->cur += len; 843 dec->cur += len;
688 844
689 if (ecb_expect_false (dec->stringref)) 845 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 846 if (!is_utf8_string (key, len))
847 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 848
692 hv_store (hv, key, -len, decode_sv (dec), 0); 849 hv_store (hv, key, -len, decode_sv (dec), 0);
693 850
694 return; 851 return;
695 } 852 }
696 853
697 SV *k = decode_sv (dec); 854 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 855 SV *v = decode_sv (dec);
699 856
857 // we leak memory if uncaught exceptions are thrown by random magical
858 // methods, and this is hopefully the only place where it can happen,
859 // so if there is a chance of an exception, take the very slow path.
860 // since catching exceptions is "undocumented/internal/forbidden" by
861 // the new p5p powers, we need to call out to a perl function :/
862 if (ecb_expect_false (SvAMAGIC (k)))
863 {
864 dSP;
865
866 ENTER; SAVETMPS;
867 PUSHMARK (SP);
868 EXTEND (SP, 3);
869 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
870 PUSHs (sv_2mortal (k));
871 PUSHs (sv_2mortal (v));
872
873 PUTBACK;
874 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
875 SPAGAIN;
876
877 FREETMPS; LEAVE;
878
879 if (SvTRUE (ERRSV))
880 ERR_ERRSV;
881
882 return;
883 }
884
700 hv_store_ent (hv, k, v, 0); 885 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 886 SvREFCNT_dec (k);
887
888fail:
889 ;
702} 890}
703 891
704static SV * 892static SV *
705decode_hv (dec_t *dec) 893decode_hv (dec_t *dec)
706{ 894{
707 HV *hv = newHV (); 895 HV *hv = newHV ();
708 896
709 DEC_INC_DEPTH; 897 DEC_INC_DEPTH;
710 898
711 if ((*dec->cur & 31) == 31) 899 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 900 {
713 ++dec->cur; 901 ++dec->cur;
714 902
715 for (;;) 903 for (;;)
716 { 904 {
717 WANT (1); 905 WANT (1);
718 906
719 if (*dec->cur == (0xe0 | 31)) 907 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 908 {
721 ++dec->cur; 909 ++dec->cur;
722 break; 910 break;
723 } 911 }
724 912
725 decode_he (dec, hv); 913 decode_he (dec, hv);
726 } 914 }
727 } 915 }
728 else 916 else
729 { 917 {
730 int pairs = decode_uint (dec); 918 UV pairs = decode_uint (dec);
919
920 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731 921
732 while (pairs--) 922 while (pairs--)
733 decode_he (dec, hv); 923 decode_he (dec, hv);
734 } 924 }
735 925
745static SV * 935static SV *
746decode_str (dec_t *dec, int utf8) 936decode_str (dec_t *dec, int utf8)
747{ 937{
748 SV *sv = 0; 938 SV *sv = 0;
749 939
750 if ((*dec->cur & 31) == 31) 940 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 941 {
942 // indefinite length strings
752 ++dec->cur; 943 ++dec->cur;
753 944
945 U8 major = *dec->cur & MAJOR_MISC;
946
754 sv = newSVpvn ("", 0); 947 sv = newSVpvn ("", 0);
755 948
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 949 for (;;)
758 { 950 {
759 WANT (1); 951 WANT (1);
760 952
761 if (*dec->cur == (0xe0 | 31)) 953 if ((*dec->cur - major) > LENGTH_EXT8)
954 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 955 {
763 ++dec->cur; 956 ++dec->cur;
764 break; 957 break;
765 } 958 }
959 else
960 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 961
767 sv_catsv (sv, decode_sv (dec)); 962 STRLEN len = decode_uint (dec);
963
964 WANT (len);
965 sv_catpvn (sv, dec->cur, len);
966 dec->cur += len;
768 } 967 }
769 } 968 }
770 else 969 else
771 { 970 {
772 STRLEN len = decode_uint (dec); 971 STRLEN len = decode_uint (dec);
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 978 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 979 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 980 }
782 981
783 if (utf8) 982 if (utf8)
983 {
984 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
985 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
986 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
987
784 SvUTF8_on (sv); 988 SvUTF8_on (sv);
989 }
785 990
786 return sv; 991 return sv;
787 992
788fail: 993fail:
789 SvREFCNT_dec (sv); 994 SvREFCNT_dec (sv);
808 sv = newRV_noinc (decode_sv (dec)); 1013 sv = newRV_noinc (decode_sv (dec));
809 break; 1014 break;
810 1015
811 case CBOR_TAG_STRINGREF_NAMESPACE: 1016 case CBOR_TAG_STRINGREF_NAMESPACE:
812 { 1017 {
1018 // do nmot use SAVETMPS/FREETMPS, as these will
1019 // erase mortalised caches, e.g. "shareable"
813 ENTER; SAVETMPS; 1020 ENTER;
814 1021
815 SAVESPTR (dec->stringref); 1022 SAVESPTR (dec->stringref);
816 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1023 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817 1024
818 sv = decode_sv (dec); 1025 sv = decode_sv (dec);
819 1026
820 FREETMPS; LEAVE; 1027 LEAVE;
821 } 1028 }
822 break; 1029 break;
823 1030
824 case CBOR_TAG_STRINGREF: 1031 case CBOR_TAG_STRINGREF:
825 { 1032 {
826 if ((*dec->cur >> 5) != 0) 1033 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1034 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 1035
829 UV idx = decode_uint (dec); 1036 UV idx = decode_uint (dec);
830 1037
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1038 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 case CBOR_TAG_VALUE_SHAREABLE: 1045 case CBOR_TAG_VALUE_SHAREABLE:
839 { 1046 {
840 if (ecb_expect_false (!dec->shareable)) 1047 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1048 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 1049
1050 if (dec->cbor.flags & F_ALLOW_CYCLES)
1051 {
843 sv = newSV (0); 1052 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1053 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 1054
846 SV *osv = decode_sv (dec); 1055 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 1056 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 1057 SvREFCNT_dec_NN (osv);
1058 }
1059 else
1060 {
1061 av_push (dec->shareable, &PL_sv_undef);
1062 int idx = AvFILLp (dec->shareable);
1063 sv = decode_sv (dec);
1064 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1065 }
849 } 1066 }
850 break; 1067 break;
851 1068
852 case CBOR_TAG_VALUE_SHAREDREF: 1069 case CBOR_TAG_VALUE_SHAREDREF:
853 { 1070 {
854 if ((*dec->cur >> 5) != 0) 1071 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1072 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 1073
857 UV idx = decode_uint (dec); 1074 UV idx = decode_uint (dec);
858 1075
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1076 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1077 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 1078
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1079 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1080
1081 if (sv == &PL_sv_undef)
1082 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 1083 }
864 break; 1084 break;
865 1085
866 case CBOR_TAG_PERL_OBJECT: 1086 case CBOR_TAG_PERL_OBJECT:
867 { 1087 {
1088 if (dec->cbor.flags & F_FORBID_OBJECTS)
1089 goto filter;
1090
868 sv = decode_sv (dec); 1091 sv = decode_sv (dec);
869 1092
870 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1093 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871 ERR ("corrupted CBOR data (non-array perl object)"); 1094 ERR ("corrupted CBOR data (non-array perl object)");
872 1095
882 if (!method) 1105 if (!method)
883 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1106 ERR ("cannot decode perl-object (package does not have a THAW method)");
884 1107
885 dSP; 1108 dSP;
886 1109
887 ENTER; SAVETMPS; PUSHMARK (SP); 1110 ENTER; SAVETMPS;
1111 PUSHMARK (SP);
888 EXTEND (SP, len + 1); 1112 EXTEND (SP, len + 1);
889 // we re-bless the reference to get overload and other niceties right 1113 // we re-bless the reference to get overload and other niceties right
890 PUSHs (*av_fetch (av, 0, 1)); 1114 PUSHs (*av_fetch (av, 0, 1));
891 PUSHs (sv_cbor); 1115 PUSHs (sv_cbor);
892 1116
900 SPAGAIN; 1124 SPAGAIN;
901 1125
902 if (SvTRUE (ERRSV)) 1126 if (SvTRUE (ERRSV))
903 { 1127 {
904 FREETMPS; LEAVE; 1128 FREETMPS; LEAVE;
905 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1129 ERR_ERRSV;
906 } 1130 }
907 1131
908 SvREFCNT_dec (sv); 1132 SvREFCNT_dec (sv);
909 sv = SvREFCNT_inc (POPs); 1133 sv = SvREFCNT_inc (POPs);
910 1134
913 FREETMPS; LEAVE; 1137 FREETMPS; LEAVE;
914 } 1138 }
915 break; 1139 break;
916 1140
917 default: 1141 default:
1142 filter:
918 { 1143 {
1144 SV *tag_sv = newSVuv (tag);
1145
919 sv = decode_sv (dec); 1146 sv = decode_sv (dec);
920 1147
921 dSP; 1148 dSP;
922 ENTER; SAVETMPS; PUSHMARK (SP); 1149 ENTER; SAVETMPS;
1150 SAVESTACK_POS ();
1151 PUSHMARK (SP);
923 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
924 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
925 PUSHs (sv); 1154 PUSHs (sv);
926 1155
927 PUTBACK; 1156 PUTBACK;
928 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929 SPAGAIN; 1158 SPAGAIN;
930 1159
931 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
932 { 1161 {
1162 SvREFCNT_dec (tag_sv);
933 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
934 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
935 } 1165 }
936 1166
937 if (count) 1167 if (count)
938 { 1168 {
1169 SvREFCNT_dec (tag_sv);
939 SvREFCNT_dec (sv); 1170 SvREFCNT_dec (sv);
940 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc (POPs);
941 } 1172 }
942 else 1173 else
943 { 1174 {
944 AV *av = newAV (); 1175 AV *av = newAV ();
945 av_push (av, newSVuv (tag)); 1176 av_push (av, tag_sv);
946 av_push (av, sv); 1177 av_push (av, sv);
947 1178
948 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1179 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949 ? cbor_tagged_stash 1180 ? cbor_tagged_stash
950 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1181 : gv_stashpv ("CBOR::XS::Tagged" , 1);
968static SV * 1199static SV *
969decode_sv (dec_t *dec) 1200decode_sv (dec_t *dec)
970{ 1201{
971 WANT (1); 1202 WANT (1);
972 1203
973 switch (*dec->cur >> 5) 1204 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1205 {
975 case 0: // unsigned int 1206 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1207 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1208 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1209 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1210 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1211 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1212 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1213
983 case 4: // array 1214 case MAJOR_MISC >> MAJOR_SHIFT:
984 return decode_av (dec);
985 case 5: // map
986 return decode_hv (dec);
987 case 6: // tag
988 return decode_tagged (dec);
989 case 7: // misc
990 switch (*dec->cur++ & 31) 1215 switch (*dec->cur++ & MINOR_MASK)
991 { 1216 {
992 case 20: 1217 case SIMPLE_FALSE:
993#if CBOR_SLOW 1218#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1219 types_false = get_bool ("Types::Serialiser::false");
995#endif 1220#endif
996 return newSVsv (types_false); 1221 return newSVsv (types_false);
997 case 21: 1222 case SIMPLE_TRUE:
998#if CBOR_SLOW 1223#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1224 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1225#endif
1001 return newSVsv (types_true); 1226 return newSVsv (types_true);
1002 case 22: 1227 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1228 return newSVsv (&PL_sv_undef);
1004 case 23: 1229 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1230#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1231 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1232#endif
1008 return newSVsv (types_error); 1233 return newSVsv (types_error);
1009 1234
1010 case 25: 1235 case MISC_FLOAT16:
1011 { 1236 {
1012 WANT (2); 1237 WANT (2);
1013 1238
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1239 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1240 dec->cur += 2;
1016 1241
1017 return newSVnv (ecb_binary16_to_float (fp)); 1242 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1243 }
1019 1244
1020 case 26: 1245 case MISC_FLOAT32:
1021 { 1246 {
1022 uint32_t fp; 1247 uint32_t fp;
1023 WANT (4); 1248 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1249 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1250 dec->cur += 4;
1028 fp = ecb_bswap32 (fp); 1253 fp = ecb_bswap32 (fp);
1029 1254
1030 return newSVnv (ecb_binary32_to_float (fp)); 1255 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1256 }
1032 1257
1033 case 27: 1258 case MISC_FLOAT64:
1034 { 1259 {
1035 uint64_t fp; 1260 uint64_t fp;
1036 WANT (8); 1261 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1262 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1263 dec->cur += 8;
1041 fp = ecb_bswap64 (fp); 1266 fp = ecb_bswap64 (fp);
1042 1267
1043 return newSVnv (ecb_binary64_to_double (fp)); 1268 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1269 }
1045 1270
1046 // 0..19 unassigned 1271 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1272 // 24 reserved + unassigned simple (reserved values are not encodable)
1273 // 28-30 unassigned misc
1274 // 31 break code
1048 default: 1275 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1276 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1277 }
1051 1278
1052 break; 1279 break;
1053 } 1280 }
1054 1281
1057} 1284}
1058 1285
1059static SV * 1286static SV *
1060decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1287decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061{ 1288{
1062 dec_t dec = { }; 1289 dec_t dec = { 0 };
1063 SV *sv; 1290 SV *sv;
1064 STRLEN len; 1291 STRLEN len;
1065 char *data = SvPVbyte (string, len); 1292 char *data = SvPVbyte (string, len);
1066 1293
1067 if (len > cbor->max_size && cbor->max_size) 1294 if (len > cbor->max_size && cbor->max_size)
1081 if (dec.cur != dec.end && !dec.err) 1308 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1309 dec.err = "garbage after CBOR object";
1083 1310
1084 if (dec.err) 1311 if (dec.err)
1085 { 1312 {
1313 if (dec.shareable)
1314 {
1315 // need to break cyclic links, which would all be in shareable
1316 int i;
1317 SV **svp;
1318
1319 for (i = av_len (dec.shareable) + 1; i--; )
1320 if ((svp = av_fetch (dec.shareable, i, 0)))
1321 sv_setsv (*svp, &PL_sv_undef);
1322 }
1323
1086 SvREFCNT_dec (sv); 1324 SvREFCNT_dec (sv);
1325
1326 if (dec.err_sv)
1327 sv_2mortal (dec.err_sv);
1328
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1329 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1330 }
1089 1331
1090 sv = sv_2mortal (sv); 1332 sv = sv_2mortal (sv);
1091 1333
1092 return sv; 1334 return sv;
1093} 1335}
1094 1336
1337/////////////////////////////////////////////////////////////////////////////
1338// incremental parser
1339
1340#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1341
1342// returns 0 for notyet, 1 for success or error
1343static int
1344incr_parse (CBOR *self, SV *cborstr)
1345{
1346 STRLEN cur;
1347 SvPV (cborstr, cur);
1348
1349 while (ecb_expect_true (self->incr_need <= cur))
1350 {
1351 // table of integer count bytes
1352 static I8 incr_len[MINOR_MASK + 1] = {
1353 0, 0, 0, 0, 0, 0, 0, 0,
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 1, 2, 4, 8,-1,-1,-1,-2
1357 };
1358
1359 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1360 U8 m = *p & MINOR_MASK;
1361 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1362 I8 ilen = incr_len[m];
1363
1364 self->incr_need = self->incr_pos + 1;
1365
1366 if (ecb_expect_false (ilen < 0))
1367 {
1368 if (m != MINOR_INDEF)
1369 return 1; // error
1370
1371 if (*p == (MAJOR_MISC | MINOR_INDEF))
1372 {
1373 if (count >= 0)
1374 return 1; // error
1375
1376 count = 1;
1377 }
1378 else
1379 {
1380 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1381 count = -1;
1382 }
1383 }
1384 else
1385 {
1386 self->incr_need += ilen;
1387 if (ecb_expect_false (self->incr_need > cur))
1388 return 0;
1389
1390 int major = *p >> MAJOR_SHIFT;
1391
1392 switch (major)
1393 {
1394 case MAJOR_TAG >> MAJOR_SHIFT:
1395 ++count; // tags merely prefix another value
1396 break;
1397
1398 case MAJOR_BYTES >> MAJOR_SHIFT:
1399 case MAJOR_TEXT >> MAJOR_SHIFT:
1400 case MAJOR_ARRAY >> MAJOR_SHIFT:
1401 case MAJOR_MAP >> MAJOR_SHIFT:
1402 {
1403 UV len;
1404
1405 if (ecb_expect_false (ilen))
1406 {
1407 len = 0;
1408
1409 do {
1410 len = (len << 8) | *++p;
1411 } while (--ilen);
1412 }
1413 else
1414 len = m;
1415
1416 switch (major)
1417 {
1418 case MAJOR_BYTES >> MAJOR_SHIFT:
1419 case MAJOR_TEXT >> MAJOR_SHIFT:
1420 self->incr_need += len;
1421 if (ecb_expect_false (self->incr_need > cur))
1422 return 0;
1423
1424 break;
1425
1426 case MAJOR_MAP >> MAJOR_SHIFT:
1427 len <<= 1;
1428 case MAJOR_ARRAY >> MAJOR_SHIFT:
1429 if (len)
1430 {
1431 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1432 count = len + 1;
1433 }
1434 break;
1435 }
1436 }
1437 }
1438 }
1439
1440 self->incr_pos = self->incr_need;
1441
1442 if (count > 0)
1443 {
1444 while (!--count)
1445 {
1446 if (!AvFILLp (self->incr_count))
1447 return 1; // done
1448
1449 SvREFCNT_dec_NN (av_pop (self->incr_count));
1450 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1451 }
1452
1453 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1454 }
1455 }
1456
1457 return 0;
1458}
1459
1460
1095///////////////////////////////////////////////////////////////////////////// 1461/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1462// XS interface functions
1097 1463
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1464MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1465
1111 1477
1112 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1478 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113 1479
1114 sv_cbor = newSVpv ("CBOR", 0); 1480 sv_cbor = newSVpv ("CBOR", 0);
1115 SvREADONLY_on (sv_cbor); 1481 SvREADONLY_on (sv_cbor);
1482
1483 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116} 1484}
1117 1485
1118PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
1119 1487
1120void CLONE (...) 1488void CLONE (...)
1139void shrink (CBOR *self, int enable = 1) 1507void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1508 ALIAS:
1141 shrink = F_SHRINK 1509 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1510 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1511 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1512 allow_cycles = F_ALLOW_CYCLES
1513 forbid_objects = F_FORBID_OBJECTS
1514 pack_strings = F_PACK_STRINGS
1515 text_keys = F_TEXT_KEYS
1516 text_strings = F_TEXT_STRINGS
1517 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1518 PPCODE:
1146{ 1519{
1147 if (enable) 1520 if (enable)
1148 self->flags |= ix; 1521 self->flags |= ix;
1149 else 1522 else
1155void get_shrink (CBOR *self) 1528void get_shrink (CBOR *self)
1156 ALIAS: 1529 ALIAS:
1157 get_shrink = F_SHRINK 1530 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1531 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1532 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1533 get_allow_cycles = F_ALLOW_CYCLES
1534 get_forbid_objects = F_FORBID_OBJECTS
1535 get_pack_strings = F_PACK_STRINGS
1536 get_text_keys = F_TEXT_KEYS
1537 get_text_strings = F_TEXT_STRINGS
1538 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1539 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1540 XPUSHs (boolSV (self->flags & ix));
1163 1541
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1542void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1543 PPCODE:
1214 EXTEND (SP, 2); 1592 EXTEND (SP, 2);
1215 PUSHs (sv); 1593 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1594 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1595}
1218 1596
1597void incr_parse (CBOR *self, SV *cborstr)
1598 ALIAS:
1599 incr_parse_multiple = 1
1600 PPCODE:
1601{
1602 if (SvUTF8 (cborstr))
1603 sv_utf8_downgrade (cborstr, 0);
1604
1605 if (!self->incr_count)
1606 {
1607 self->incr_count = newAV ();
1608 self->incr_pos = 0;
1609 self->incr_need = 1;
1610
1611 av_push (self->incr_count, newSViv (1));
1612 }
1613
1614 do
1615 {
1616 if (!incr_parse (self, cborstr))
1617 {
1618 if (self->incr_need > self->max_size && self->max_size)
1619 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1620 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1621
1622 break;
1623 }
1624
1625 SV *sv;
1626 char *offset;
1627
1628 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1629 XPUSHs (sv);
1630
1631 sv_chop (cborstr, offset);
1632
1633 av_clear (self->incr_count);
1634 av_push (self->incr_count, newSViv (1));
1635
1636 self->incr_pos = 0;
1637 self->incr_need = self->incr_pos + 1;
1638 }
1639 while (ix);
1640}
1641
1642void incr_reset (CBOR *self)
1643 CODE:
1644{
1645 SvREFCNT_dec (self->incr_count);
1646 self->incr_count = 0;
1647}
1648
1219void DESTROY (CBOR *self) 1649void DESTROY (CBOR *self)
1220 PPCODE: 1650 PPCODE:
1221 cbor_free (self); 1651 cbor_free (self);
1222 1652
1223PROTOTYPES: ENABLE 1653PROTOTYPES: ENABLE
1224 1654
1225void encode_cbor (SV *scalar) 1655void encode_cbor (SV *scalar)
1656 ALIAS:
1657 encode_cbor = 0
1658 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1659 PPCODE:
1227{ 1660{
1228 CBOR cbor; 1661 CBOR cbor;
1229 cbor_init (&cbor); 1662 cbor_init (&cbor);
1663 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1664 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1665 XPUSHs (scalar);
1232} 1666}
1233 1667
1234void decode_cbor (SV *cborstr) 1668void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines