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 1.29 by root, Thu Nov 28 09:13:12 2013 UTC vs.
Revision 1.62 by root, Sat Nov 26 02:11:52 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 // extensions 70 // extensions
32 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref 71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object 72 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object 73 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing 75 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref 76 CBOR_TAG_STRINGREF_NAMESPACE = 256, // 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 return idx <= 23 ? 3
143 ? idx > 0xffU 193 : idx <= 0xffU ? 4
144 ? idx > 0xffffU 194 : idx <= 0xffffU ? 5
145 ? idx > 0xffffffffU 195 : idx <= 0xffffffffU ? 7
146 ? 11 196 : 11;
147 : 7
148 : 5
149 : 4
150 : 3;
151} 197}
152 198
153///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
154// encoder 200// encoder
155 201
168} enc_t; 214} enc_t;
169 215
170ecb_inline void 216ecb_inline void
171need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
172{ 218{
173 if (ecb_expect_false (enc->cur + len >= enc->end)) 219 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
174 { 220 {
175 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177 enc->cur = SvPVX (enc->sv) + cur; 223 enc->cur = SvPVX (enc->sv) + cur;
178 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
189static void 235static void
190encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
191{ 237{
192 need (enc, 9); 238 need (enc, 9);
193 239
194 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
195 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
196 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
197 { 243 {
198 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
199 *enc->cur++ = len; 245 *enc->cur++ = len;
200 } 246 }
201 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
202 { 248 {
203 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
204 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
205 *enc->cur++ = len; 251 *enc->cur++ = len;
206 } 252 }
207 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
208 { 254 {
209 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
210 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
211 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
212 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
213 *enc->cur++ = len; 259 *enc->cur++ = len;
214 } 260 }
215 else 261 else
216 { 262 {
217 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
218 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
219 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
220 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
221 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
222 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
227} 273}
228 274
229ecb_inline void 275ecb_inline void
230encode_tag (enc_t *enc, UV tag) 276encode_tag (enc_t *enc, UV tag)
231{ 277{
232 encode_uint (enc, 0xc0, tag); 278 encode_uint (enc, MAJOR_TAG, tag);
233} 279}
234 280
235static void 281// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
282ecb_noinline static void
236encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
237{ 284{
285 STRLEN ulen = len;
286 U8 *p, *pend = (U8 *)str + len;
287
288 for (p = (U8 *)str; p < pend; ++p)
289 ulen += *p >> 7; // count set high bits
290
291 encode_uint (enc, MAJOR_TEXT, ulen);
292
293 need (enc, ulen);
294 for (p = (U8 *)str; p < pend; ++p)
295 if (*p < 0x80)
296 *enc->cur++ = *p;
297 else
298 {
299 *enc->cur++ = 0xc0 + (*p >> 6);
300 *enc->cur++ = 0x80 + (*p & 63);
301 }
302}
303
304ecb_inline void
305encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
306{
307 if (ecb_expect_false (upgrade_utf8))
308 if (!utf8)
309 {
310 encode_str_utf8 (enc, utf8, str, len);
311 return;
312 }
313
314 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
315 need (enc, len);
316 memcpy (enc->cur, str, len);
317 enc->cur += len;
318}
319
320ecb_inline void
321encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
322{
238 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) 323 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
239 { 324 {
240 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 325 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
241 326
242 if (SvOK (*svp)) 327 if (SvOK (*svp))
243 { 328 {
244 // already registered, use stringref 329 // already registered, use stringref
245 encode_tag (enc, CBOR_TAG_STRINGREF); 330 encode_tag (enc, CBOR_TAG_STRINGREF);
246 encode_uint (enc, 0x00, SvUV (*svp)); 331 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
247 return; 332 return;
248 } 333 }
249 else if (len >= minimum_string_length (enc->stringref_idx)) 334 else if (len >= minimum_string_length (enc->stringref_idx))
250 { 335 {
251 // register only 336 // register only
252 sv_setuv (*svp, enc->stringref_idx); 337 sv_setuv (*svp, enc->stringref_idx);
253 ++enc->stringref_idx; 338 ++enc->stringref_idx;
254 } 339 }
255 } 340 }
256 341
257 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 342 encode_str (enc, upgrade_utf8, utf8, str, len);
258 need (enc, len);
259 memcpy (enc->cur, str, len);
260 enc->cur += len;
261} 343}
262 344
263static void encode_sv (enc_t *enc, SV *sv); 345static void encode_sv (enc_t *enc, SV *sv);
264 346
265static void 347static void
270 if (enc->depth >= enc->cbor.max_depth) 352 if (enc->depth >= enc->cbor.max_depth)
271 croak (ERR_NESTING_EXCEEDED); 353 croak (ERR_NESTING_EXCEEDED);
272 354
273 ++enc->depth; 355 ++enc->depth;
274 356
275 encode_uint (enc, 0x80, len + 1); 357 encode_uint (enc, MAJOR_ARRAY, len + 1);
276 358
359 if (SvMAGICAL (av))
277 for (i = 0; i <= len; ++i) 360 for (i = 0; i <= len; ++i)
278 { 361 {
279 SV **svp = av_fetch (av, i, 0); 362 SV **svp = av_fetch (av, i, 0);
280 encode_sv (enc, svp ? *svp : &PL_sv_undef); 363 encode_sv (enc, svp ? *svp : &PL_sv_undef);
281 } 364 }
365 else
366 for (i = 0; i <= len; ++i)
367 {
368 SV *sv = AvARRAY (av)[i];
369 encode_sv (enc, sv ? sv : &PL_sv_undef);
370 }
282 371
283 --enc->depth; 372 --enc->depth;
284} 373}
285 374
286static void 375static void
295 384
296 int pairs = hv_iterinit (hv); 385 int pairs = hv_iterinit (hv);
297 int mg = SvMAGICAL (hv); 386 int mg = SvMAGICAL (hv);
298 387
299 if (mg) 388 if (mg)
300 encode_ch (enc, 0xa0 | 31); 389 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
301 else 390 else
302 encode_uint (enc, 0xa0, pairs); 391 encode_uint (enc, MAJOR_MAP, pairs);
303 392
304 while ((he = hv_iternext (hv))) 393 while ((he = hv_iternext (hv)))
305 { 394 {
306 if (HeKLEN (he) == HEf_SVKEY) 395 if (HeKLEN (he) == HEf_SVKEY)
307 encode_sv (enc, HeSVKEY (he)); 396 encode_sv (enc, HeSVKEY (he));
308 else 397 else
309 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 398 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
310 399
311 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 400 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
312 } 401 }
313 402
314 if (mg) 403 if (mg)
315 encode_ch (enc, 0xe0 | 31); 404 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
316 405
317 --enc->depth; 406 --enc->depth;
318} 407}
319 408
320// encode objects, arrays and special \0=false and \1=true values. 409// encode objects, arrays and special \0=false and \1=true values.
321static void 410static void
322encode_rv (enc_t *enc, SV *sv) 411encode_rv (enc_t *enc, SV *sv)
323{ 412{
324 SvGETMAGIC (sv); 413 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 }
347 414
348 svtype svt = SvTYPE (sv); 415 svtype svt = SvTYPE (sv);
349 416
350 if (ecb_expect_false (SvOBJECT (sv))) 417 if (ecb_expect_false (SvOBJECT (sv)))
351 { 418 {
358 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 425 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
359 ? cbor_tagged_stash 426 ? cbor_tagged_stash
360 : gv_stashpv ("CBOR::XS::Tagged" , 1); 427 : gv_stashpv ("CBOR::XS::Tagged" , 1);
361 428
362 HV *stash = SvSTASH (sv); 429 HV *stash = SvSTASH (sv);
363 GV *method;
364 430
365 if (stash == boolean_stash) 431 if (stash == boolean_stash)
366 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 432 {
433 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
434 return;
435 }
367 else if (stash == error_stash) 436 else if (stash == error_stash)
368 encode_ch (enc, 0xe0 | 23); 437 {
438 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
439 return;
440 }
369 else if (stash == tagged_stash) 441 else if (stash == tagged_stash)
370 { 442 {
371 if (svt != SVt_PVAV) 443 if (svt != SVt_PVAV)
372 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 444 croak ("encountered CBOR::XS::Tagged object that isn't an array");
373 445
374 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 446 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
375 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 447 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
448
449 return;
450 }
451 }
452
453 if (ecb_expect_false (SvREFCNT (sv) > 1)
454 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
455 {
456 if (!enc->shareable)
457 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
458
459 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
460
461 if (SvOK (*svp))
376 } 462 {
463 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
464 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
465 return;
466 }
467 else
468 {
469 sv_setuv (*svp, enc->shareable_idx);
470 ++enc->shareable_idx;
471 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
472 }
473 }
474
475 if (ecb_expect_false (SvOBJECT (sv)))
476 {
477 HV *stash = SvSTASH (sv);
478 GV *method;
479
480 if (enc->cbor.flags & F_FORBID_OBJECTS)
481 croak ("encountered object '%s', but forbid_objects is enabled",
482 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
377 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 483 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
378 { 484 {
379 dSP; 485 dSP;
380 486
381 ENTER; SAVETMPS; PUSHMARK (SP); 487 ENTER; SAVETMPS;
488 PUSHMARK (SP);
382 // we re-bless the reference to get overload and other niceties right 489 // we re-bless the reference to get overload and other niceties right
383 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 490 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
384 491
385 PUTBACK; 492 PUTBACK;
386 // G_SCALAR ensures that return value is 1 493 // G_SCALAR ensures that return value is 1
399 } 506 }
400 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 507 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
401 { 508 {
402 dSP; 509 dSP;
403 510
404 ENTER; SAVETMPS; PUSHMARK (SP); 511 ENTER; SAVETMPS;
512 SAVESTACK_POS ();
513 PUSHMARK (SP);
405 EXTEND (SP, 2); 514 EXTEND (SP, 2);
406 // we re-bless the reference to get overload and other niceties right 515 // we re-bless the reference to get overload and other niceties right
407 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 516 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
408 PUSHs (sv_cbor); 517 PUSHs (sv_cbor);
409 518
414 // catch this surprisingly common error 523 // catch this surprisingly common error
415 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 524 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)); 525 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
417 526
418 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 527 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
419 encode_uint (enc, 0x80, count + 1); 528 encode_uint (enc, MAJOR_ARRAY, count + 1);
420 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 529 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
421 530
422 while (count) 531 while (count)
423 encode_sv (enc, SP[1 - count--]); 532 encode_sv (enc, SP[1 - count--]);
424 533
425 PUTBACK; 534 PUTBACK;
446{ 555{
447 double nv = SvNVX (sv); 556 double nv = SvNVX (sv);
448 557
449 need (enc, 9); 558 need (enc, 9);
450 559
451 if (ecb_expect_false (nv == (U32)nv)) 560 if (ecb_expect_false (nv == (NV)(U32)nv))
452 encode_uint (enc, 0x00, (U32)nv); 561 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
453 //TODO: maybe I32? 562 //TODO: maybe I32?
454 else if (ecb_expect_false (nv == (float)nv)) 563 else if (ecb_expect_false (nv == (float)nv))
455 { 564 {
456 uint32_t fp = ecb_float_to_binary32 (nv); 565 uint32_t fp = ecb_float_to_binary32 (nv);
457 566
458 *enc->cur++ = 0xe0 | 26; 567 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
459 568
460 if (!ecb_big_endian ()) 569 if (!ecb_big_endian ())
461 fp = ecb_bswap32 (fp); 570 fp = ecb_bswap32 (fp);
462 571
463 memcpy (enc->cur, &fp, 4); 572 memcpy (enc->cur, &fp, 4);
465 } 574 }
466 else 575 else
467 { 576 {
468 uint64_t fp = ecb_double_to_binary64 (nv); 577 uint64_t fp = ecb_double_to_binary64 (nv);
469 578
470 *enc->cur++ = 0xe0 | 27; 579 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
471 580
472 if (!ecb_big_endian ()) 581 if (!ecb_big_endian ())
473 fp = ecb_bswap64 (fp); 582 fp = ecb_bswap64 (fp);
474 583
475 memcpy (enc->cur, &fp, 8); 584 memcpy (enc->cur, &fp, 8);
484 593
485 if (SvPOKp (sv)) 594 if (SvPOKp (sv))
486 { 595 {
487 STRLEN len; 596 STRLEN len;
488 char *str = SvPV (sv, len); 597 char *str = SvPV (sv, len);
489 encode_str (enc, SvUTF8 (sv), str, len); 598 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
490 } 599 }
491 else if (SvNOKp (sv)) 600 else if (SvNOKp (sv))
492 encode_nv (enc, sv); 601 encode_nv (enc, sv);
493 else if (SvIOKp (sv)) 602 else if (SvIOKp (sv))
494 { 603 {
495 if (SvIsUV (sv)) 604 if (SvIsUV (sv))
496 encode_uint (enc, 0x00, SvUVX (sv)); 605 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
497 else if (SvIVX (sv) >= 0) 606 else if (SvIVX (sv) >= 0)
498 encode_uint (enc, 0x00, SvIVX (sv)); 607 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
499 else 608 else
500 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 609 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
501 } 610 }
502 else if (SvROK (sv)) 611 else if (SvROK (sv))
503 encode_rv (enc, SvRV (sv)); 612 encode_rv (enc, SvRV (sv));
504 else if (!SvOK (sv)) 613 else if (!SvOK (sv))
505 encode_ch (enc, 0xe0 | 22); 614 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
506 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 615 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
507 encode_ch (enc, 0xe0 | 23); 616 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
508 else 617 else
509 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 618 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
510 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 619 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
511} 620}
512 621
513static SV * 622static SV *
514encode_cbor (SV *scalar, CBOR *cbor) 623encode_cbor (SV *scalar, CBOR *cbor)
515{ 624{
516 enc_t enc = { }; 625 enc_t enc = { 0 };
517 626
518 enc.cbor = *cbor; 627 enc.cbor = *cbor;
519 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 628 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
520 enc.cur = SvPVX (enc.sv); 629 enc.cur = SvPVX (enc.sv);
521 enc.end = SvEND (enc.sv); 630 enc.end = SvEND (enc.sv);
522 631
523 SvPOK_only (enc.sv); 632 SvPOK_only (enc.sv);
524 633
525 if (cbor->flags & F_ALLOW_STRINGREF) 634 if (cbor->flags & F_PACK_STRINGS)
526 { 635 {
527 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 636 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
528 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
529 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 638 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
530 } 639 }
553 U32 depth; // recursion depth 662 U32 depth; // recursion depth
554 U32 maxdepth; // recursion depth limit 663 U32 maxdepth; // recursion depth limit
555 AV *shareable; 664 AV *shareable;
556 AV *stringref; 665 AV *stringref;
557 SV *decode_tagged; 666 SV *decode_tagged;
667 SV *err_sv; // optional sv for error, needs to be freed
558} dec_t; 668} dec_t;
559 669
560#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 670// set dec->err to ERRSV
671ecb_cold static void
672err_errsv (dec_t *dec)
673{
674 if (!dec->err)
675 {
676 dec->err_sv = newSVsv (ERRSV);
561 677
562#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 678 // chop off the trailing \n
679 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
680 *SvEND (dec->err_sv) = 0;
563 681
682 dec->err = SvPVutf8_nolen (dec->err_sv);
683 }
684}
685
686// the following functions are used to reduce code size and help the compiler to optimise
687ecb_cold static void
688err_set (dec_t *dec, const char *reason)
689{
690 if (!dec->err)
691 dec->err = reason;
692}
693
694ecb_cold static void
695err_unexpected_end (dec_t *dec)
696{
697 err_set (dec, "unexpected end of CBOR data");
698}
699
700ecb_cold static void
701err_nesting_exceeded (dec_t *dec)
702{
703 err_set (dec, ERR_NESTING_EXCEEDED);
704}
705
706#define ERR_DO(do) SB do; goto fail; SE
707#define ERR(reason) ERR_DO (err_set (dec, reason))
708#define ERR_ERRSV ERR_DO (err_errsv (dec))
709
710#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
711
564#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 712#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 713#define DEC_DEC_DEPTH --dec->depth
566 714
567static UV 715static UV
568decode_uint (dec_t *dec) 716decode_uint (dec_t *dec)
569{ 717{
570 switch (*dec->cur & 31) 718 U8 m = *dec->cur & MINOR_MASK;
571 { 719 ++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;
576 720
577 case 24: 721 if (ecb_expect_true (m < LENGTH_EXT1))
722 return m;
723 else if (ecb_expect_true (m == LENGTH_EXT1))
724 {
578 WANT (2); 725 WANT (1);
579 dec->cur += 2; 726 dec->cur += 1;
580 return dec->cur[-1]; 727 return dec->cur[-1];
581 728 }
582 case 25: 729 else if (ecb_expect_true (m == LENGTH_EXT2))
730 {
583 WANT (3); 731 WANT (2);
584 dec->cur += 3; 732 dec->cur += 2;
585 return (((UV)dec->cur[-2]) << 8) 733 return (((UV)dec->cur[-2]) << 8)
586 | ((UV)dec->cur[-1]); 734 | ((UV)dec->cur[-1]);
587 735 }
588 case 26: 736 else if (ecb_expect_true (m == LENGTH_EXT4))
737 {
589 WANT (5); 738 WANT (4);
590 dec->cur += 5; 739 dec->cur += 4;
591 return (((UV)dec->cur[-4]) << 24) 740 return (((UV)dec->cur[-4]) << 24)
592 | (((UV)dec->cur[-3]) << 16) 741 | (((UV)dec->cur[-3]) << 16)
593 | (((UV)dec->cur[-2]) << 8) 742 | (((UV)dec->cur[-2]) << 8)
594 | ((UV)dec->cur[-1]); 743 | ((UV)dec->cur[-1]);
595 744 }
596 case 27: 745 else if (ecb_expect_true (m == LENGTH_EXT8))
746 {
597 WANT (9); 747 WANT (8);
598 dec->cur += 9; 748 dec->cur += 8;
749
750 return
751#if UVSIZE < 8
752 0
753#else
599 return (((UV)dec->cur[-8]) << 56) 754 (((UV)dec->cur[-8]) << 56)
600 | (((UV)dec->cur[-7]) << 48) 755 | (((UV)dec->cur[-7]) << 48)
601 | (((UV)dec->cur[-6]) << 40) 756 | (((UV)dec->cur[-6]) << 40)
602 | (((UV)dec->cur[-5]) << 32) 757 | (((UV)dec->cur[-5]) << 32)
758#endif
603 | (((UV)dec->cur[-4]) << 24) 759 | (((UV)dec->cur[-4]) << 24)
604 | (((UV)dec->cur[-3]) << 16) 760 | (((UV)dec->cur[-3]) << 16)
605 | (((UV)dec->cur[-2]) << 8) 761 | (((UV)dec->cur[-2]) << 8)
606 | ((UV)dec->cur[-1]); 762 | ((UV)dec->cur[-1]);
607 763 }
608 default: 764 else
609 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 765 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
610 }
611 766
612fail: 767fail:
613 return 0; 768 return 0;
614} 769}
615 770
620{ 775{
621 AV *av = newAV (); 776 AV *av = newAV ();
622 777
623 DEC_INC_DEPTH; 778 DEC_INC_DEPTH;
624 779
625 if ((*dec->cur & 31) == 31) 780 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
626 { 781 {
627 ++dec->cur; 782 ++dec->cur;
628 783
629 for (;;) 784 for (;;)
630 { 785 {
631 WANT (1); 786 WANT (1);
632 787
633 if (*dec->cur == (0xe0 | 31)) 788 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
634 { 789 {
635 ++dec->cur; 790 ++dec->cur;
636 break; 791 break;
637 } 792 }
638 793
639 av_push (av, decode_sv (dec)); 794 av_push (av, decode_sv (dec));
640 } 795 }
641 } 796 }
642 else 797 else
643 { 798 {
644 int i, len = decode_uint (dec); 799 UV i, len = decode_uint (dec);
645 800
801 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); 802 av_fill (av, len - 1);
647 803
648 for (i = 0; i < len; ++i) 804 for (i = 0; i < len; ++i)
649 AvARRAY (av)[i] = decode_sv (dec); 805 AvARRAY (av)[i] = decode_sv (dec);
650 } 806 }
663{ 819{
664 // for speed reasons, we specialcase single-string 820 // for speed reasons, we specialcase single-string
665 // byte or utf-8 strings as keys, but only when !stringref 821 // byte or utf-8 strings as keys, but only when !stringref
666 822
667 if (ecb_expect_true (!dec->stringref)) 823 if (ecb_expect_true (!dec->stringref))
668 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 824 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
669 { 825 {
670 I32 len = decode_uint (dec); 826 STRLEN len = decode_uint (dec);
671 char *key = (char *)dec->cur; 827 char *key = (char *)dec->cur;
672 828
829 WANT (len);
673 dec->cur += len; 830 dec->cur += len;
674 831
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); 832 hv_store (hv, key, len, decode_sv (dec), 0);
679 833
680 return; 834 return;
681 } 835 }
682 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 836 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
683 { 837 {
684 I32 len = decode_uint (dec); 838 STRLEN len = decode_uint (dec);
685 char *key = (char *)dec->cur; 839 char *key = (char *)dec->cur;
686 840
841 WANT (len);
687 dec->cur += len; 842 dec->cur += len;
688 843
689 if (ecb_expect_false (dec->stringref)) 844 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
690 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 845 if (!is_utf8_string (key, len))
846 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
691 847
692 hv_store (hv, key, -len, decode_sv (dec), 0); 848 hv_store (hv, key, -len, decode_sv (dec), 0);
693 849
694 return; 850 return;
695 } 851 }
696 852
697 SV *k = decode_sv (dec); 853 SV *k = decode_sv (dec);
698 SV *v = decode_sv (dec); 854 SV *v = decode_sv (dec);
699 855
856 // we leak memory if uncaught exceptions are thrown by random magical
857 // methods, and this is hopefully the only place where it can happen,
858 // so if there is a chance of an exception, take the very slow path.
859 // since catching exceptions is "undocumented/internal/forbidden" by
860 // the new p5p powers, we need to call out to a perl function :/
861 if (ecb_expect_false (SvAMAGIC (k)))
862 {
863 dSP;
864
865 ENTER; SAVETMPS;
866 PUSHMARK (SP);
867 EXTEND (SP, 3);
868 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
869 PUSHs (sv_2mortal (k));
870 PUSHs (sv_2mortal (v));
871
872 PUTBACK;
873 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
874 SPAGAIN;
875
876 FREETMPS; LEAVE;
877
878 if (SvTRUE (ERRSV))
879 ERR_ERRSV;
880
881 return;
882 }
883
700 hv_store_ent (hv, k, v, 0); 884 hv_store_ent (hv, k, v, 0);
701 SvREFCNT_dec (k); 885 SvREFCNT_dec (k);
886
887fail:
888 ;
702} 889}
703 890
704static SV * 891static SV *
705decode_hv (dec_t *dec) 892decode_hv (dec_t *dec)
706{ 893{
707 HV *hv = newHV (); 894 HV *hv = newHV ();
708 895
709 DEC_INC_DEPTH; 896 DEC_INC_DEPTH;
710 897
711 if ((*dec->cur & 31) == 31) 898 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
712 { 899 {
713 ++dec->cur; 900 ++dec->cur;
714 901
715 for (;;) 902 for (;;)
716 { 903 {
717 WANT (1); 904 WANT (1);
718 905
719 if (*dec->cur == (0xe0 | 31)) 906 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
720 { 907 {
721 ++dec->cur; 908 ++dec->cur;
722 break; 909 break;
723 } 910 }
724 911
725 decode_he (dec, hv); 912 decode_he (dec, hv);
726 } 913 }
727 } 914 }
728 else 915 else
729 { 916 {
730 int pairs = decode_uint (dec); 917 UV pairs = decode_uint (dec);
918
919 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
731 920
732 while (pairs--) 921 while (pairs--)
733 decode_he (dec, hv); 922 decode_he (dec, hv);
734 } 923 }
735 924
745static SV * 934static SV *
746decode_str (dec_t *dec, int utf8) 935decode_str (dec_t *dec, int utf8)
747{ 936{
748 SV *sv = 0; 937 SV *sv = 0;
749 938
750 if ((*dec->cur & 31) == 31) 939 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
751 { 940 {
941 // indefinite length strings
752 ++dec->cur; 942 ++dec->cur;
753 943
944 U8 major = *dec->cur & MAJOR_MISC;
945
754 sv = newSVpvn ("", 0); 946 sv = newSVpvn ("", 0);
755 947
756 // not very fast, and certainly not robust against illegal input
757 for (;;) 948 for (;;)
758 { 949 {
759 WANT (1); 950 WANT (1);
760 951
761 if (*dec->cur == (0xe0 | 31)) 952 if ((*dec->cur - major) > LENGTH_EXT8)
953 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
762 { 954 {
763 ++dec->cur; 955 ++dec->cur;
764 break; 956 break;
765 } 957 }
958 else
959 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
766 960
767 sv_catsv (sv, decode_sv (dec)); 961 STRLEN len = decode_uint (dec);
962
963 WANT (len);
964 sv_catpvn (sv, dec->cur, len);
965 dec->cur += len;
768 } 966 }
769 } 967 }
770 else 968 else
771 { 969 {
772 STRLEN len = decode_uint (dec); 970 STRLEN len = decode_uint (dec);
779 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) 977 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
780 av_push (dec->stringref, SvREFCNT_inc_NN (sv)); 978 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
781 } 979 }
782 980
783 if (utf8) 981 if (utf8)
982 {
983 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
984 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
985 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
986
784 SvUTF8_on (sv); 987 SvUTF8_on (sv);
988 }
785 989
786 return sv; 990 return sv;
787 991
788fail: 992fail:
789 SvREFCNT_dec (sv); 993 SvREFCNT_dec (sv);
808 sv = newRV_noinc (decode_sv (dec)); 1012 sv = newRV_noinc (decode_sv (dec));
809 break; 1013 break;
810 1014
811 case CBOR_TAG_STRINGREF_NAMESPACE: 1015 case CBOR_TAG_STRINGREF_NAMESPACE:
812 { 1016 {
1017 // do nmot use SAVETMPS/FREETMPS, as these will
1018 // erase mortalised caches, e.g. "shareable"
813 ENTER; SAVETMPS; 1019 ENTER;
814 1020
815 SAVESPTR (dec->stringref); 1021 SAVESPTR (dec->stringref);
816 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1022 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
817 1023
818 sv = decode_sv (dec); 1024 sv = decode_sv (dec);
819 1025
820 FREETMPS; LEAVE; 1026 LEAVE;
821 } 1027 }
822 break; 1028 break;
823 1029
824 case CBOR_TAG_STRINGREF: 1030 case CBOR_TAG_STRINGREF:
825 { 1031 {
826 if ((*dec->cur >> 5) != 0) 1032 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
827 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1033 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
828 1034
829 UV idx = decode_uint (dec); 1035 UV idx = decode_uint (dec);
830 1036
831 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1037 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
838 case CBOR_TAG_VALUE_SHAREABLE: 1044 case CBOR_TAG_VALUE_SHAREABLE:
839 { 1045 {
840 if (ecb_expect_false (!dec->shareable)) 1046 if (ecb_expect_false (!dec->shareable))
841 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1047 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
842 1048
1049 if (dec->cbor.flags & F_ALLOW_CYCLES)
1050 {
843 sv = newSV (0); 1051 sv = newSV (0);
844 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1052 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
845 1053
846 SV *osv = decode_sv (dec); 1054 SV *osv = decode_sv (dec);
847 sv_setsv (sv, osv); 1055 sv_setsv (sv, osv);
848 SvREFCNT_dec_NN (osv); 1056 SvREFCNT_dec_NN (osv);
1057 }
1058 else
1059 {
1060 av_push (dec->shareable, &PL_sv_undef);
1061 int idx = AvFILLp (dec->shareable);
1062 sv = decode_sv (dec);
1063 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1064 }
849 } 1065 }
850 break; 1066 break;
851 1067
852 case CBOR_TAG_VALUE_SHAREDREF: 1068 case CBOR_TAG_VALUE_SHAREDREF:
853 { 1069 {
854 if ((*dec->cur >> 5) != 0) 1070 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
855 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1071 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
856 1072
857 UV idx = decode_uint (dec); 1073 UV idx = decode_uint (dec);
858 1074
859 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1075 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
860 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1076 ERR ("corrupted CBOR data (sharedref index out of bounds)");
861 1077
862 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1078 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1079
1080 if (sv == &PL_sv_undef)
1081 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
863 } 1082 }
864 break; 1083 break;
865 1084
866 case CBOR_TAG_PERL_OBJECT: 1085 case CBOR_TAG_PERL_OBJECT:
867 { 1086 {
1087 if (dec->cbor.flags & F_FORBID_OBJECTS)
1088 goto filter;
1089
868 sv = decode_sv (dec); 1090 sv = decode_sv (dec);
869 1091
870 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1092 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
871 ERR ("corrupted CBOR data (non-array perl object)"); 1093 ERR ("corrupted CBOR data (non-array perl object)");
872 1094
882 if (!method) 1104 if (!method)
883 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1105 ERR ("cannot decode perl-object (package does not have a THAW method)");
884 1106
885 dSP; 1107 dSP;
886 1108
887 ENTER; SAVETMPS; PUSHMARK (SP); 1109 ENTER; SAVETMPS;
1110 PUSHMARK (SP);
888 EXTEND (SP, len + 1); 1111 EXTEND (SP, len + 1);
889 // we re-bless the reference to get overload and other niceties right 1112 // we re-bless the reference to get overload and other niceties right
890 PUSHs (*av_fetch (av, 0, 1)); 1113 PUSHs (*av_fetch (av, 0, 1));
891 PUSHs (sv_cbor); 1114 PUSHs (sv_cbor);
892 1115
900 SPAGAIN; 1123 SPAGAIN;
901 1124
902 if (SvTRUE (ERRSV)) 1125 if (SvTRUE (ERRSV))
903 { 1126 {
904 FREETMPS; LEAVE; 1127 FREETMPS; LEAVE;
905 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1128 ERR_ERRSV;
906 } 1129 }
907 1130
908 SvREFCNT_dec (sv); 1131 SvREFCNT_dec (sv);
909 sv = SvREFCNT_inc (POPs); 1132 sv = SvREFCNT_inc (POPs);
910 1133
913 FREETMPS; LEAVE; 1136 FREETMPS; LEAVE;
914 } 1137 }
915 break; 1138 break;
916 1139
917 default: 1140 default:
1141 filter:
918 { 1142 {
1143 SV *tag_sv = newSVuv (tag);
1144
919 sv = decode_sv (dec); 1145 sv = decode_sv (dec);
920 1146
921 dSP; 1147 dSP;
922 ENTER; SAVETMPS; PUSHMARK (SP); 1148 ENTER; SAVETMPS;
1149 SAVESTACK_POS ();
1150 PUSHMARK (SP);
923 EXTEND (SP, 2); 1151 EXTEND (SP, 2);
924 PUSHs (newSVuv (tag)); 1152 PUSHs (tag_sv);
925 PUSHs (sv); 1153 PUSHs (sv);
926 1154
927 PUTBACK; 1155 PUTBACK;
928 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1156 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
929 SPAGAIN; 1157 SPAGAIN;
930 1158
931 if (SvTRUE (ERRSV)) 1159 if (SvTRUE (ERRSV))
932 { 1160 {
1161 SvREFCNT_dec (tag_sv);
933 FREETMPS; LEAVE; 1162 FREETMPS; LEAVE;
934 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1163 ERR_ERRSV;
935 } 1164 }
936 1165
937 if (count) 1166 if (count)
938 { 1167 {
1168 SvREFCNT_dec (tag_sv);
939 SvREFCNT_dec (sv); 1169 SvREFCNT_dec (sv);
940 sv = SvREFCNT_inc (POPs); 1170 sv = SvREFCNT_inc (POPs);
941 } 1171 }
942 else 1172 else
943 { 1173 {
944 AV *av = newAV (); 1174 AV *av = newAV ();
945 av_push (av, newSVuv (tag)); 1175 av_push (av, tag_sv);
946 av_push (av, sv); 1176 av_push (av, sv);
947 1177
948 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1178 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
949 ? cbor_tagged_stash 1179 ? cbor_tagged_stash
950 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1180 : gv_stashpv ("CBOR::XS::Tagged" , 1);
968static SV * 1198static SV *
969decode_sv (dec_t *dec) 1199decode_sv (dec_t *dec)
970{ 1200{
971 WANT (1); 1201 WANT (1);
972 1202
973 switch (*dec->cur >> 5) 1203 switch (*dec->cur >> MAJOR_SHIFT)
974 { 1204 {
975 case 0: // unsigned int 1205 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
976 return newSVuv (decode_uint (dec)); 1206 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
977 case 1: // negative int 1207 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
978 return newSViv (-1 - (IV)decode_uint (dec)); 1208 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
979 case 2: // octet string 1209 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
980 return decode_str (dec, 0); 1210 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
981 case 3: // utf-8 string 1211 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
982 return decode_str (dec, 1); 1212
983 case 4: // array 1213 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) 1214 switch (*dec->cur++ & MINOR_MASK)
991 { 1215 {
992 case 20: 1216 case SIMPLE_FALSE:
993#if CBOR_SLOW 1217#if CBOR_SLOW
994 types_false = get_bool ("Types::Serialiser::false"); 1218 types_false = get_bool ("Types::Serialiser::false");
995#endif 1219#endif
996 return newSVsv (types_false); 1220 return newSVsv (types_false);
997 case 21: 1221 case SIMPLE_TRUE:
998#if CBOR_SLOW 1222#if CBOR_SLOW
999 types_true = get_bool ("Types::Serialiser::true"); 1223 types_true = get_bool ("Types::Serialiser::true");
1000#endif 1224#endif
1001 return newSVsv (types_true); 1225 return newSVsv (types_true);
1002 case 22: 1226 case SIMPLE_NULL:
1003 return newSVsv (&PL_sv_undef); 1227 return newSVsv (&PL_sv_undef);
1004 case 23: 1228 case SIMPLE_UNDEF:
1005#if CBOR_SLOW 1229#if CBOR_SLOW
1006 types_error = get_bool ("Types::Serialiser::error"); 1230 types_error = get_bool ("Types::Serialiser::error");
1007#endif 1231#endif
1008 return newSVsv (types_error); 1232 return newSVsv (types_error);
1009 1233
1010 case 25: 1234 case MISC_FLOAT16:
1011 { 1235 {
1012 WANT (2); 1236 WANT (2);
1013 1237
1014 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1238 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1015 dec->cur += 2; 1239 dec->cur += 2;
1016 1240
1017 return newSVnv (ecb_binary16_to_float (fp)); 1241 return newSVnv (ecb_binary16_to_float (fp));
1018 } 1242 }
1019 1243
1020 case 26: 1244 case MISC_FLOAT32:
1021 { 1245 {
1022 uint32_t fp; 1246 uint32_t fp;
1023 WANT (4); 1247 WANT (4);
1024 memcpy (&fp, dec->cur, 4); 1248 memcpy (&fp, dec->cur, 4);
1025 dec->cur += 4; 1249 dec->cur += 4;
1028 fp = ecb_bswap32 (fp); 1252 fp = ecb_bswap32 (fp);
1029 1253
1030 return newSVnv (ecb_binary32_to_float (fp)); 1254 return newSVnv (ecb_binary32_to_float (fp));
1031 } 1255 }
1032 1256
1033 case 27: 1257 case MISC_FLOAT64:
1034 { 1258 {
1035 uint64_t fp; 1259 uint64_t fp;
1036 WANT (8); 1260 WANT (8);
1037 memcpy (&fp, dec->cur, 8); 1261 memcpy (&fp, dec->cur, 8);
1038 dec->cur += 8; 1262 dec->cur += 8;
1041 fp = ecb_bswap64 (fp); 1265 fp = ecb_bswap64 (fp);
1042 1266
1043 return newSVnv (ecb_binary64_to_double (fp)); 1267 return newSVnv (ecb_binary64_to_double (fp));
1044 } 1268 }
1045 1269
1046 // 0..19 unassigned 1270 // 0..19 unassigned simple
1047 // 24 reserved + unassigned (reserved values are not encodable) 1271 // 24 reserved + unassigned simple (reserved values are not encodable)
1272 // 28-30 unassigned misc
1273 // 31 break code
1048 default: 1274 default:
1049 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1275 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1050 } 1276 }
1051 1277
1052 break; 1278 break;
1053 } 1279 }
1054 1280
1057} 1283}
1058 1284
1059static SV * 1285static SV *
1060decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1286decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1061{ 1287{
1062 dec_t dec = { }; 1288 dec_t dec = { 0 };
1063 SV *sv; 1289 SV *sv;
1064 STRLEN len; 1290 STRLEN len;
1065 char *data = SvPVbyte (string, len); 1291 char *data = SvPVbyte (string, len);
1066 1292
1067 if (len > cbor->max_size && cbor->max_size) 1293 if (len > cbor->max_size && cbor->max_size)
1081 if (dec.cur != dec.end && !dec.err) 1307 if (dec.cur != dec.end && !dec.err)
1082 dec.err = "garbage after CBOR object"; 1308 dec.err = "garbage after CBOR object";
1083 1309
1084 if (dec.err) 1310 if (dec.err)
1085 { 1311 {
1312 if (dec.shareable)
1313 {
1314 // need to break cyclic links, which would all be in shareable
1315 int i;
1316 SV **svp;
1317
1318 for (i = av_len (dec.shareable) + 1; i--; )
1319 if ((svp = av_fetch (dec.shareable, i, 0)))
1320 sv_setsv (*svp, &PL_sv_undef);
1321 }
1322
1086 SvREFCNT_dec (sv); 1323 SvREFCNT_dec (sv);
1324
1325 if (dec.err_sv)
1326 sv_2mortal (dec.err_sv);
1327
1087 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1328 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1088 } 1329 }
1089 1330
1090 sv = sv_2mortal (sv); 1331 sv = sv_2mortal (sv);
1091 1332
1092 return sv; 1333 return sv;
1093} 1334}
1094 1335
1336/////////////////////////////////////////////////////////////////////////////
1337// incremental parser
1338
1339#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1340
1341// returns 0 for notyet, 1 for success or error
1342static int
1343incr_parse (CBOR *self, SV *cborstr)
1344{
1345 STRLEN cur;
1346 SvPV (cborstr, cur);
1347
1348 while (ecb_expect_true (self->incr_need <= cur))
1349 {
1350 // table of integer count bytes
1351 static I8 incr_len[MINOR_MASK + 1] = {
1352 0, 0, 0, 0, 0, 0, 0, 0,
1353 0, 0, 0, 0, 0, 0, 0, 0,
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 1, 2, 4, 8,-1,-1,-1,-2
1356 };
1357
1358 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1359 U8 m = *p & MINOR_MASK;
1360 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1361 I8 ilen = incr_len[m];
1362
1363 self->incr_need = self->incr_pos + 1;
1364
1365 if (ecb_expect_false (ilen < 0))
1366 {
1367 if (m != MINOR_INDEF)
1368 return 1; // error
1369
1370 if (*p == (MAJOR_MISC | MINOR_INDEF))
1371 {
1372 if (count >= 0)
1373 return 1; // error
1374
1375 count = 1;
1376 }
1377 else
1378 {
1379 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1380 count = -1;
1381 }
1382 }
1383 else
1384 {
1385 self->incr_need += ilen;
1386 if (ecb_expect_false (self->incr_need > cur))
1387 return 0;
1388
1389 int major = *p >> MAJOR_SHIFT;
1390
1391 switch (major)
1392 {
1393 case MAJOR_TAG >> MAJOR_SHIFT:
1394 ++count; // tags merely prefix another value
1395 break;
1396
1397 case MAJOR_BYTES >> MAJOR_SHIFT:
1398 case MAJOR_TEXT >> MAJOR_SHIFT:
1399 case MAJOR_ARRAY >> MAJOR_SHIFT:
1400 case MAJOR_MAP >> MAJOR_SHIFT:
1401 {
1402 UV len;
1403
1404 if (ecb_expect_false (ilen))
1405 {
1406 len = 0;
1407
1408 do {
1409 len = (len << 8) | *++p;
1410 } while (--ilen);
1411 }
1412 else
1413 len = m;
1414
1415 switch (major)
1416 {
1417 case MAJOR_BYTES >> MAJOR_SHIFT:
1418 case MAJOR_TEXT >> MAJOR_SHIFT:
1419 self->incr_need += len;
1420 if (ecb_expect_false (self->incr_need > cur))
1421 return 0;
1422
1423 break;
1424
1425 case MAJOR_MAP >> MAJOR_SHIFT:
1426 len <<= 1;
1427 case MAJOR_ARRAY >> MAJOR_SHIFT:
1428 if (len)
1429 {
1430 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1431 count = len + 1;
1432 }
1433 break;
1434 }
1435 }
1436 }
1437 }
1438
1439 self->incr_pos = self->incr_need;
1440
1441 if (count > 0)
1442 {
1443 while (!--count)
1444 {
1445 if (!AvFILLp (self->incr_count))
1446 return 1; // done
1447
1448 SvREFCNT_dec_NN (av_pop (self->incr_count));
1449 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1450 }
1451
1452 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1453 }
1454 }
1455
1456 return 0;
1457}
1458
1459
1095///////////////////////////////////////////////////////////////////////////// 1460/////////////////////////////////////////////////////////////////////////////
1096// XS interface functions 1461// XS interface functions
1097 1462
1098MODULE = CBOR::XS PACKAGE = CBOR::XS 1463MODULE = CBOR::XS PACKAGE = CBOR::XS
1099 1464
1111 1476
1112 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1477 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1113 1478
1114 sv_cbor = newSVpv ("CBOR", 0); 1479 sv_cbor = newSVpv ("CBOR", 0);
1115 SvREADONLY_on (sv_cbor); 1480 SvREADONLY_on (sv_cbor);
1481
1482 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1116} 1483}
1117 1484
1118PROTOTYPES: DISABLE 1485PROTOTYPES: DISABLE
1119 1486
1120void CLONE (...) 1487void CLONE (...)
1139void shrink (CBOR *self, int enable = 1) 1506void shrink (CBOR *self, int enable = 1)
1140 ALIAS: 1507 ALIAS:
1141 shrink = F_SHRINK 1508 shrink = F_SHRINK
1142 allow_unknown = F_ALLOW_UNKNOWN 1509 allow_unknown = F_ALLOW_UNKNOWN
1143 allow_sharing = F_ALLOW_SHARING 1510 allow_sharing = F_ALLOW_SHARING
1144 allow_stringref = F_ALLOW_STRINGREF 1511 allow_cycles = F_ALLOW_CYCLES
1512 forbid_objects = F_FORBID_OBJECTS
1513 pack_strings = F_PACK_STRINGS
1514 text_keys = F_TEXT_KEYS
1515 text_strings = F_TEXT_STRINGS
1516 validate_utf8 = F_VALIDATE_UTF8
1145 PPCODE: 1517 PPCODE:
1146{ 1518{
1147 if (enable) 1519 if (enable)
1148 self->flags |= ix; 1520 self->flags |= ix;
1149 else 1521 else
1155void get_shrink (CBOR *self) 1527void get_shrink (CBOR *self)
1156 ALIAS: 1528 ALIAS:
1157 get_shrink = F_SHRINK 1529 get_shrink = F_SHRINK
1158 get_allow_unknown = F_ALLOW_UNKNOWN 1530 get_allow_unknown = F_ALLOW_UNKNOWN
1159 get_allow_sharing = F_ALLOW_SHARING 1531 get_allow_sharing = F_ALLOW_SHARING
1160 get_allow_stringref = F_ALLOW_STRINGREF 1532 get_allow_cycles = F_ALLOW_CYCLES
1533 get_forbid_objects = F_FORBID_OBJECTS
1534 get_pack_strings = F_PACK_STRINGS
1535 get_text_keys = F_TEXT_KEYS
1536 get_text_strings = F_TEXT_STRINGS
1537 get_validate_utf8 = F_VALIDATE_UTF8
1161 PPCODE: 1538 PPCODE:
1162 XPUSHs (boolSV (self->flags & ix)); 1539 XPUSHs (boolSV (self->flags & ix));
1163 1540
1164void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1541void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1165 PPCODE: 1542 PPCODE:
1214 EXTEND (SP, 2); 1591 EXTEND (SP, 2);
1215 PUSHs (sv); 1592 PUSHs (sv);
1216 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1593 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1217} 1594}
1218 1595
1596void incr_parse (CBOR *self, SV *cborstr)
1597 ALIAS:
1598 incr_parse_multiple = 1
1599 PPCODE:
1600{
1601 if (SvUTF8 (cborstr))
1602 sv_utf8_downgrade (cborstr, 0);
1603
1604 if (!self->incr_count)
1605 {
1606 self->incr_count = newAV ();
1607 self->incr_pos = 0;
1608 self->incr_need = 1;
1609
1610 av_push (self->incr_count, newSViv (1));
1611 }
1612
1613 do
1614 {
1615 if (!incr_parse (self, cborstr))
1616 {
1617 if (self->incr_need > self->max_size && self->max_size)
1618 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1619 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1620
1621 break;
1622 }
1623
1624 SV *sv;
1625 char *offset;
1626
1627 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1628 XPUSHs (sv);
1629
1630 sv_chop (cborstr, offset);
1631
1632 av_clear (self->incr_count);
1633 av_push (self->incr_count, newSViv (1));
1634
1635 self->incr_pos = 0;
1636 self->incr_need = self->incr_pos + 1;
1637 }
1638 while (ix);
1639}
1640
1641void incr_reset (CBOR *self)
1642 CODE:
1643{
1644 SvREFCNT_dec (self->incr_count);
1645 self->incr_count = 0;
1646}
1647
1219void DESTROY (CBOR *self) 1648void DESTROY (CBOR *self)
1220 PPCODE: 1649 PPCODE:
1221 cbor_free (self); 1650 cbor_free (self);
1222 1651
1223PROTOTYPES: ENABLE 1652PROTOTYPES: ENABLE
1224 1653
1225void encode_cbor (SV *scalar) 1654void encode_cbor (SV *scalar)
1655 ALIAS:
1656 encode_cbor = 0
1657 encode_cbor_sharing = F_ALLOW_SHARING
1226 PPCODE: 1658 PPCODE:
1227{ 1659{
1228 CBOR cbor; 1660 CBOR cbor;
1229 cbor_init (&cbor); 1661 cbor_init (&cbor);
1662 cbor.flags |= ix;
1230 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1663 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1231 XPUSHs (scalar); 1664 XPUSHs (scalar);
1232} 1665}
1233 1666
1234void decode_cbor (SV *cborstr) 1667void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines