1 |
#include "EXTERN.h" |
2 |
#include "perl.h" |
3 |
#include "XSUB.h" |
4 |
|
5 |
#include <assert.h> |
6 |
#include <string.h> |
7 |
#include <stdlib.h> |
8 |
#include <stdio.h> |
9 |
#include <limits.h> |
10 |
#include <float.h> |
11 |
|
12 |
#define ECB_NO_THREADS 1 |
13 |
#include "ecb.h" |
14 |
|
15 |
// compatibility with perl <5.18 |
16 |
#ifndef HvNAMELEN_get |
17 |
# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
18 |
#endif |
19 |
#ifndef HvNAMELEN |
20 |
# define HvNAMELEN(hv) HvNAMELEN_get (hv) |
21 |
#endif |
22 |
#ifndef HvNAMEUTF8 |
23 |
# define HvNAMEUTF8(hv) 0 |
24 |
#endif |
25 |
#ifndef SvREFCNT_dec_NN |
26 |
# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
27 |
#endif |
28 |
|
29 |
// known major and minor types |
30 |
enum cbor_type |
31 |
{ |
32 |
MAJOR_SHIFT = 5, |
33 |
MINOR_MASK = 0x1f, |
34 |
|
35 |
MAJOR_POS_INT = 0 << MAJOR_SHIFT, |
36 |
MAJOR_NEG_INT = 1 << MAJOR_SHIFT, |
37 |
MAJOR_BYTES = 2 << MAJOR_SHIFT, |
38 |
MAJOR_TEXT = 3 << MAJOR_SHIFT, |
39 |
MAJOR_ARRAY = 4 << MAJOR_SHIFT, |
40 |
MAJOR_MAP = 5 << MAJOR_SHIFT, |
41 |
MAJOR_TAG = 6 << MAJOR_SHIFT, |
42 |
MAJOR_MISC = 7 << MAJOR_SHIFT, |
43 |
|
44 |
// INT/STRING/ARRAY/MAP subtypes |
45 |
LENGTH_EXT1 = 24, |
46 |
LENGTH_EXT2 = 25, |
47 |
LENGTH_EXT4 = 26, |
48 |
LENGTH_EXT8 = 27, |
49 |
|
50 |
// SIMPLE types (effectively MISC subtypes) |
51 |
SIMPLE_FALSE = 20, |
52 |
SIMPLE_TRUE = 21, |
53 |
SIMPLE_NULL = 22, |
54 |
SIMPLE_UNDEF = 23, |
55 |
|
56 |
// MISC subtype (unused) |
57 |
MISC_EXT1 = 24, |
58 |
MISC_FLOAT16 = 25, |
59 |
MISC_FLOAT32 = 26, |
60 |
MISC_FLOAT64 = 27, |
61 |
|
62 |
// BYTES/TEXT/ARRAY/MAP |
63 |
MINOR_INDEF = 31, |
64 |
}; |
65 |
|
66 |
// known tags |
67 |
enum cbor_tag |
68 |
{ |
69 |
// extensions |
70 |
CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
71 |
CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
72 |
CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
73 |
CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
74 |
CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
75 |
CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
76 |
CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
77 |
|
78 |
// rfc7049 |
79 |
CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
80 |
CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
81 |
CBOR_TAG_POS_BIGNUM = 2, // byte string |
82 |
CBOR_TAG_NEG_BIGNUM = 3, // byte string |
83 |
CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
84 |
CBOR_TAG_BIGFLOAT = 5, // array |
85 |
|
86 |
CBOR_TAG_CONV_B64U = 21, // base64url, any |
87 |
CBOR_TAG_CONV_B64 = 22, // base64, any |
88 |
CBOR_TAG_CONV_HEX = 23, // base16, any |
89 |
CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
90 |
|
91 |
CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
92 |
CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
93 |
CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
94 |
CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
95 |
CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
96 |
|
97 |
CBOR_TAG_MAGIC = 55799, // self-describe cbor |
98 |
}; |
99 |
|
100 |
#define F_SHRINK 0x00000001UL |
101 |
#define F_ALLOW_UNKNOWN 0x00000002UL |
102 |
#define F_ALLOW_SHARING 0x00000004UL |
103 |
#define F_ALLOW_CYCLES 0x00000008UL |
104 |
#define F_PACK_STRINGS 0x00000010UL |
105 |
#define F_VALIDATE_UTF8 0x00000020UL |
106 |
|
107 |
#define INIT_SIZE 32 // initial scalar size to be allocated |
108 |
|
109 |
#define SB do { |
110 |
#define SE } while (0) |
111 |
|
112 |
#define IN_RANGE_INC(type,val,beg,end) \ |
113 |
((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
114 |
<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
115 |
|
116 |
#define ERR_NESTING_EXCEEDED "cbor text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
117 |
|
118 |
#ifdef USE_ITHREADS |
119 |
# define CBOR_SLOW 1 |
120 |
# define CBOR_STASH (cbor_stash ? cbor_stash : gv_stashpv ("CBOR::XS", 1)) |
121 |
#else |
122 |
# define CBOR_SLOW 0 |
123 |
# define CBOR_STASH cbor_stash |
124 |
#endif |
125 |
|
126 |
static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
127 |
static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
128 |
|
129 |
typedef struct { |
130 |
U32 flags; |
131 |
U32 max_depth; |
132 |
STRLEN max_size; |
133 |
SV *filter; |
134 |
|
135 |
// for the incremental parser |
136 |
STRLEN incr_pos; // the current offset into the text |
137 |
STRLEN incr_need; // minimum bytes needed to decode |
138 |
AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
139 |
} CBOR; |
140 |
|
141 |
ecb_inline void |
142 |
cbor_init (CBOR *cbor) |
143 |
{ |
144 |
Zero (cbor, 1, CBOR); |
145 |
cbor->max_depth = 512; |
146 |
} |
147 |
|
148 |
ecb_inline void |
149 |
cbor_free (CBOR *cbor) |
150 |
{ |
151 |
SvREFCNT_dec (cbor->filter); |
152 |
SvREFCNT_dec (cbor->incr_count); |
153 |
} |
154 |
|
155 |
///////////////////////////////////////////////////////////////////////////// |
156 |
// utility functions |
157 |
|
158 |
ecb_inline SV * |
159 |
get_bool (const char *name) |
160 |
{ |
161 |
SV *sv = get_sv (name, 1); |
162 |
|
163 |
SvREADONLY_on (sv); |
164 |
SvREADONLY_on (SvRV (sv)); |
165 |
|
166 |
return sv; |
167 |
} |
168 |
|
169 |
ecb_inline void |
170 |
shrink (SV *sv) |
171 |
{ |
172 |
sv_utf8_downgrade (sv, 1); |
173 |
|
174 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
175 |
{ |
176 |
#ifdef SvPV_shrink_to_cur |
177 |
SvPV_shrink_to_cur (sv); |
178 |
#elif defined (SvPV_renew) |
179 |
SvPV_renew (sv, SvCUR (sv) + 1); |
180 |
#endif |
181 |
} |
182 |
} |
183 |
|
184 |
// minimum length of a string to be registered for stringref |
185 |
ecb_inline int |
186 |
minimum_string_length (UV idx) |
187 |
{ |
188 |
return idx > 23 |
189 |
? idx > 0xffU |
190 |
? idx > 0xffffU |
191 |
? idx > 0xffffffffU |
192 |
? 11 |
193 |
: 7 |
194 |
: 5 |
195 |
: 4 |
196 |
: 3; |
197 |
} |
198 |
|
199 |
///////////////////////////////////////////////////////////////////////////// |
200 |
// encoder |
201 |
|
202 |
// structure used for encoding CBOR |
203 |
typedef struct |
204 |
{ |
205 |
char *cur; // SvPVX (sv) + current output position |
206 |
char *end; // SvEND (sv) |
207 |
SV *sv; // result scalar |
208 |
CBOR cbor; |
209 |
U32 depth; // recursion level |
210 |
HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
211 |
UV stringref_idx; |
212 |
HV *shareable; // ptr => index, or 0 |
213 |
UV shareable_idx; |
214 |
} enc_t; |
215 |
|
216 |
ecb_inline void |
217 |
need (enc_t *enc, STRLEN len) |
218 |
{ |
219 |
if (ecb_expect_false (enc->cur + len >= enc->end)) |
220 |
{ |
221 |
STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
222 |
SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
223 |
enc->cur = SvPVX (enc->sv) + cur; |
224 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
225 |
} |
226 |
} |
227 |
|
228 |
ecb_inline void |
229 |
encode_ch (enc_t *enc, char ch) |
230 |
{ |
231 |
need (enc, 1); |
232 |
*enc->cur++ = ch; |
233 |
} |
234 |
|
235 |
static void |
236 |
encode_uint (enc_t *enc, int major, UV len) |
237 |
{ |
238 |
need (enc, 9); |
239 |
|
240 |
if (ecb_expect_true (len < LENGTH_EXT1)) |
241 |
*enc->cur++ = major | len; |
242 |
else if (ecb_expect_true (len <= 0xffU)) |
243 |
{ |
244 |
*enc->cur++ = major | LENGTH_EXT1; |
245 |
*enc->cur++ = len; |
246 |
} |
247 |
else if (len <= 0xffffU) |
248 |
{ |
249 |
*enc->cur++ = major | LENGTH_EXT2; |
250 |
*enc->cur++ = len >> 8; |
251 |
*enc->cur++ = len; |
252 |
} |
253 |
else if (len <= 0xffffffffU) |
254 |
{ |
255 |
*enc->cur++ = major | LENGTH_EXT4; |
256 |
*enc->cur++ = len >> 24; |
257 |
*enc->cur++ = len >> 16; |
258 |
*enc->cur++ = len >> 8; |
259 |
*enc->cur++ = len; |
260 |
} |
261 |
else |
262 |
{ |
263 |
*enc->cur++ = major | LENGTH_EXT8; |
264 |
*enc->cur++ = len >> 56; |
265 |
*enc->cur++ = len >> 48; |
266 |
*enc->cur++ = len >> 40; |
267 |
*enc->cur++ = len >> 32; |
268 |
*enc->cur++ = len >> 24; |
269 |
*enc->cur++ = len >> 16; |
270 |
*enc->cur++ = len >> 8; |
271 |
*enc->cur++ = len; |
272 |
} |
273 |
} |
274 |
|
275 |
ecb_inline void |
276 |
encode_tag (enc_t *enc, UV tag) |
277 |
{ |
278 |
encode_uint (enc, MAJOR_TAG, tag); |
279 |
} |
280 |
|
281 |
ecb_inline void |
282 |
encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
283 |
{ |
284 |
encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
285 |
need (enc, len); |
286 |
memcpy (enc->cur, str, len); |
287 |
enc->cur += len; |
288 |
} |
289 |
|
290 |
static void |
291 |
encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
292 |
{ |
293 |
if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
294 |
{ |
295 |
SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
296 |
|
297 |
if (SvOK (*svp)) |
298 |
{ |
299 |
// already registered, use stringref |
300 |
encode_tag (enc, CBOR_TAG_STRINGREF); |
301 |
encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
302 |
return; |
303 |
} |
304 |
else if (len >= minimum_string_length (enc->stringref_idx)) |
305 |
{ |
306 |
// register only |
307 |
sv_setuv (*svp, enc->stringref_idx); |
308 |
++enc->stringref_idx; |
309 |
} |
310 |
} |
311 |
|
312 |
encode_str (enc, utf8, str, len); |
313 |
} |
314 |
|
315 |
static void encode_sv (enc_t *enc, SV *sv); |
316 |
|
317 |
static void |
318 |
encode_av (enc_t *enc, AV *av) |
319 |
{ |
320 |
int i, len = av_len (av); |
321 |
|
322 |
if (enc->depth >= enc->cbor.max_depth) |
323 |
croak (ERR_NESTING_EXCEEDED); |
324 |
|
325 |
++enc->depth; |
326 |
|
327 |
encode_uint (enc, MAJOR_ARRAY, len + 1); |
328 |
|
329 |
if (SvMAGICAL (av)) |
330 |
for (i = 0; i <= len; ++i) |
331 |
{ |
332 |
SV **svp = av_fetch (av, i, 0); |
333 |
encode_sv (enc, svp ? *svp : &PL_sv_undef); |
334 |
} |
335 |
else |
336 |
for (i = 0; i <= len; ++i) |
337 |
{ |
338 |
SV *sv = AvARRAY (av)[i]; |
339 |
encode_sv (enc, sv ? sv : &PL_sv_undef); |
340 |
} |
341 |
|
342 |
--enc->depth; |
343 |
} |
344 |
|
345 |
static void |
346 |
encode_hv (enc_t *enc, HV *hv) |
347 |
{ |
348 |
HE *he; |
349 |
|
350 |
if (enc->depth >= enc->cbor.max_depth) |
351 |
croak (ERR_NESTING_EXCEEDED); |
352 |
|
353 |
++enc->depth; |
354 |
|
355 |
int pairs = hv_iterinit (hv); |
356 |
int mg = SvMAGICAL (hv); |
357 |
|
358 |
if (mg) |
359 |
encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
360 |
else |
361 |
encode_uint (enc, MAJOR_MAP, pairs); |
362 |
|
363 |
while ((he = hv_iternext (hv))) |
364 |
{ |
365 |
if (HeKLEN (he) == HEf_SVKEY) |
366 |
encode_sv (enc, HeSVKEY (he)); |
367 |
else |
368 |
encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
369 |
|
370 |
encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
371 |
} |
372 |
|
373 |
if (mg) |
374 |
encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
375 |
|
376 |
--enc->depth; |
377 |
} |
378 |
|
379 |
// encode objects, arrays and special \0=false and \1=true values. |
380 |
static void |
381 |
encode_rv (enc_t *enc, SV *sv) |
382 |
{ |
383 |
SvGETMAGIC (sv); |
384 |
|
385 |
svtype svt = SvTYPE (sv); |
386 |
|
387 |
if (ecb_expect_false (SvOBJECT (sv))) |
388 |
{ |
389 |
HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
390 |
? types_boolean_stash |
391 |
: gv_stashpv ("Types::Serialiser::Boolean", 1); |
392 |
HV *error_stash = !CBOR_SLOW || types_error_stash |
393 |
? types_error_stash |
394 |
: gv_stashpv ("Types::Serialiser::Error", 1); |
395 |
HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
396 |
? cbor_tagged_stash |
397 |
: gv_stashpv ("CBOR::XS::Tagged" , 1); |
398 |
|
399 |
HV *stash = SvSTASH (sv); |
400 |
|
401 |
if (stash == boolean_stash) |
402 |
{ |
403 |
encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
404 |
return; |
405 |
} |
406 |
else if (stash == error_stash) |
407 |
{ |
408 |
encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
409 |
return; |
410 |
} |
411 |
else if (stash == tagged_stash) |
412 |
{ |
413 |
if (svt != SVt_PVAV) |
414 |
croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
415 |
|
416 |
encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
417 |
encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
418 |
|
419 |
return; |
420 |
} |
421 |
} |
422 |
|
423 |
if (ecb_expect_false (SvREFCNT (sv) > 1) |
424 |
&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
425 |
{ |
426 |
if (!enc->shareable) |
427 |
enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
428 |
|
429 |
SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
430 |
|
431 |
if (SvOK (*svp)) |
432 |
{ |
433 |
encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
434 |
encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
435 |
return; |
436 |
} |
437 |
else |
438 |
{ |
439 |
sv_setuv (*svp, enc->shareable_idx); |
440 |
++enc->shareable_idx; |
441 |
encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
442 |
} |
443 |
} |
444 |
|
445 |
if (ecb_expect_false (SvOBJECT (sv))) |
446 |
{ |
447 |
HV *stash = SvSTASH (sv); |
448 |
GV *method; |
449 |
|
450 |
if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
451 |
{ |
452 |
dSP; |
453 |
|
454 |
ENTER; SAVETMPS; |
455 |
PUSHMARK (SP); |
456 |
// we re-bless the reference to get overload and other niceties right |
457 |
XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
458 |
|
459 |
PUTBACK; |
460 |
// G_SCALAR ensures that return value is 1 |
461 |
call_sv ((SV *)GvCV (method), G_SCALAR); |
462 |
SPAGAIN; |
463 |
|
464 |
// catch this surprisingly common error |
465 |
if (SvROK (TOPs) && SvRV (TOPs) == sv) |
466 |
croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash)); |
467 |
|
468 |
encode_sv (enc, POPs); |
469 |
|
470 |
PUTBACK; |
471 |
|
472 |
FREETMPS; LEAVE; |
473 |
} |
474 |
else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
475 |
{ |
476 |
dSP; |
477 |
|
478 |
ENTER; SAVETMPS; |
479 |
SAVESTACK_POS (); |
480 |
PUSHMARK (SP); |
481 |
EXTEND (SP, 2); |
482 |
// we re-bless the reference to get overload and other niceties right |
483 |
PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
484 |
PUSHs (sv_cbor); |
485 |
|
486 |
PUTBACK; |
487 |
int count = call_sv ((SV *)GvCV (method), G_ARRAY); |
488 |
SPAGAIN; |
489 |
|
490 |
// catch this surprisingly common error |
491 |
if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
492 |
croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
493 |
|
494 |
encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
495 |
encode_uint (enc, MAJOR_ARRAY, count + 1); |
496 |
encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
497 |
|
498 |
while (count) |
499 |
encode_sv (enc, SP[1 - count--]); |
500 |
|
501 |
PUTBACK; |
502 |
|
503 |
FREETMPS; LEAVE; |
504 |
} |
505 |
else |
506 |
croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it", |
507 |
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
508 |
} |
509 |
else if (svt == SVt_PVHV) |
510 |
encode_hv (enc, (HV *)sv); |
511 |
else if (svt == SVt_PVAV) |
512 |
encode_av (enc, (AV *)sv); |
513 |
else |
514 |
{ |
515 |
encode_tag (enc, CBOR_TAG_INDIRECTION); |
516 |
encode_sv (enc, sv); |
517 |
} |
518 |
} |
519 |
|
520 |
static void |
521 |
encode_nv (enc_t *enc, SV *sv) |
522 |
{ |
523 |
double nv = SvNVX (sv); |
524 |
|
525 |
need (enc, 9); |
526 |
|
527 |
if (ecb_expect_false (nv == (NV)(U32)nv)) |
528 |
encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
529 |
//TODO: maybe I32? |
530 |
else if (ecb_expect_false (nv == (float)nv)) |
531 |
{ |
532 |
uint32_t fp = ecb_float_to_binary32 (nv); |
533 |
|
534 |
*enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
535 |
|
536 |
if (!ecb_big_endian ()) |
537 |
fp = ecb_bswap32 (fp); |
538 |
|
539 |
memcpy (enc->cur, &fp, 4); |
540 |
enc->cur += 4; |
541 |
} |
542 |
else |
543 |
{ |
544 |
uint64_t fp = ecb_double_to_binary64 (nv); |
545 |
|
546 |
*enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
547 |
|
548 |
if (!ecb_big_endian ()) |
549 |
fp = ecb_bswap64 (fp); |
550 |
|
551 |
memcpy (enc->cur, &fp, 8); |
552 |
enc->cur += 8; |
553 |
} |
554 |
} |
555 |
|
556 |
static void |
557 |
encode_sv (enc_t *enc, SV *sv) |
558 |
{ |
559 |
SvGETMAGIC (sv); |
560 |
|
561 |
if (SvPOKp (sv)) |
562 |
{ |
563 |
STRLEN len; |
564 |
char *str = SvPV (sv, len); |
565 |
encode_strref (enc, SvUTF8 (sv), str, len); |
566 |
} |
567 |
else if (SvNOKp (sv)) |
568 |
encode_nv (enc, sv); |
569 |
else if (SvIOKp (sv)) |
570 |
{ |
571 |
if (SvIsUV (sv)) |
572 |
encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
573 |
else if (SvIVX (sv) >= 0) |
574 |
encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
575 |
else |
576 |
encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
577 |
} |
578 |
else if (SvROK (sv)) |
579 |
encode_rv (enc, SvRV (sv)); |
580 |
else if (!SvOK (sv)) |
581 |
encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
582 |
else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
583 |
encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
584 |
else |
585 |
croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
586 |
SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
587 |
} |
588 |
|
589 |
static SV * |
590 |
encode_cbor (SV *scalar, CBOR *cbor) |
591 |
{ |
592 |
enc_t enc = { 0 }; |
593 |
|
594 |
enc.cbor = *cbor; |
595 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
596 |
enc.cur = SvPVX (enc.sv); |
597 |
enc.end = SvEND (enc.sv); |
598 |
|
599 |
SvPOK_only (enc.sv); |
600 |
|
601 |
if (cbor->flags & F_PACK_STRINGS) |
602 |
{ |
603 |
encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
604 |
enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
605 |
enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
606 |
} |
607 |
|
608 |
encode_sv (&enc, scalar); |
609 |
|
610 |
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
611 |
*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
612 |
|
613 |
if (enc.cbor.flags & F_SHRINK) |
614 |
shrink (enc.sv); |
615 |
|
616 |
return enc.sv; |
617 |
} |
618 |
|
619 |
///////////////////////////////////////////////////////////////////////////// |
620 |
// decoder |
621 |
|
622 |
// structure used for decoding CBOR |
623 |
typedef struct |
624 |
{ |
625 |
U8 *cur; // current parser pointer |
626 |
U8 *end; // end of input string |
627 |
const char *err; // parse error, if != 0 |
628 |
CBOR cbor; |
629 |
U32 depth; // recursion depth |
630 |
U32 maxdepth; // recursion depth limit |
631 |
AV *shareable; |
632 |
AV *stringref; |
633 |
SV *decode_tagged; |
634 |
} dec_t; |
635 |
|
636 |
#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
637 |
|
638 |
#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
639 |
|
640 |
#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
641 |
#define DEC_DEC_DEPTH --dec->depth |
642 |
|
643 |
static UV |
644 |
decode_uint (dec_t *dec) |
645 |
{ |
646 |
U8 m = *dec->cur & MINOR_MASK; |
647 |
++dec->cur; |
648 |
|
649 |
if (ecb_expect_true (m < LENGTH_EXT1)) |
650 |
return m; |
651 |
else if (ecb_expect_true (m == LENGTH_EXT1)) |
652 |
{ |
653 |
WANT (1); |
654 |
dec->cur += 1; |
655 |
return dec->cur[-1]; |
656 |
} |
657 |
else if (ecb_expect_true (m == LENGTH_EXT2)) |
658 |
{ |
659 |
WANT (2); |
660 |
dec->cur += 2; |
661 |
return (((UV)dec->cur[-2]) << 8) |
662 |
| ((UV)dec->cur[-1]); |
663 |
} |
664 |
else if (ecb_expect_true (m == LENGTH_EXT4)) |
665 |
{ |
666 |
WANT (4); |
667 |
dec->cur += 4; |
668 |
return (((UV)dec->cur[-4]) << 24) |
669 |
| (((UV)dec->cur[-3]) << 16) |
670 |
| (((UV)dec->cur[-2]) << 8) |
671 |
| ((UV)dec->cur[-1]); |
672 |
} |
673 |
else if (ecb_expect_true (m == LENGTH_EXT8)) |
674 |
{ |
675 |
WANT (8); |
676 |
dec->cur += 8; |
677 |
|
678 |
return |
679 |
#if UVSIZE < 8 |
680 |
0 |
681 |
#else |
682 |
(((UV)dec->cur[-8]) << 56) |
683 |
| (((UV)dec->cur[-7]) << 48) |
684 |
| (((UV)dec->cur[-6]) << 40) |
685 |
| (((UV)dec->cur[-5]) << 32) |
686 |
#endif |
687 |
| (((UV)dec->cur[-4]) << 24) |
688 |
| (((UV)dec->cur[-3]) << 16) |
689 |
| (((UV)dec->cur[-2]) << 8) |
690 |
| ((UV)dec->cur[-1]); |
691 |
} |
692 |
else |
693 |
ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
694 |
|
695 |
fail: |
696 |
return 0; |
697 |
} |
698 |
|
699 |
static SV *decode_sv (dec_t *dec); |
700 |
|
701 |
static SV * |
702 |
decode_av (dec_t *dec) |
703 |
{ |
704 |
AV *av = newAV (); |
705 |
|
706 |
DEC_INC_DEPTH; |
707 |
|
708 |
if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
709 |
{ |
710 |
++dec->cur; |
711 |
|
712 |
for (;;) |
713 |
{ |
714 |
WANT (1); |
715 |
|
716 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
717 |
{ |
718 |
++dec->cur; |
719 |
break; |
720 |
} |
721 |
|
722 |
av_push (av, decode_sv (dec)); |
723 |
} |
724 |
} |
725 |
else |
726 |
{ |
727 |
int i, len = decode_uint (dec); |
728 |
|
729 |
WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
730 |
av_fill (av, len - 1); |
731 |
|
732 |
for (i = 0; i < len; ++i) |
733 |
AvARRAY (av)[i] = decode_sv (dec); |
734 |
} |
735 |
|
736 |
DEC_DEC_DEPTH; |
737 |
return newRV_noinc ((SV *)av); |
738 |
|
739 |
fail: |
740 |
SvREFCNT_dec (av); |
741 |
DEC_DEC_DEPTH; |
742 |
return &PL_sv_undef; |
743 |
} |
744 |
|
745 |
static void |
746 |
decode_he (dec_t *dec, HV *hv) |
747 |
{ |
748 |
// for speed reasons, we specialcase single-string |
749 |
// byte or utf-8 strings as keys, but only when !stringref |
750 |
|
751 |
if (ecb_expect_true (!dec->stringref)) |
752 |
if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
753 |
{ |
754 |
I32 len = decode_uint (dec); |
755 |
char *key = (char *)dec->cur; |
756 |
|
757 |
WANT (len); |
758 |
dec->cur += len; |
759 |
|
760 |
hv_store (hv, key, len, decode_sv (dec), 0); |
761 |
|
762 |
return; |
763 |
} |
764 |
else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
765 |
{ |
766 |
I32 len = decode_uint (dec); |
767 |
char *key = (char *)dec->cur; |
768 |
|
769 |
WANT (len); |
770 |
dec->cur += len; |
771 |
|
772 |
if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
773 |
if (!is_utf8_string (key, len)) |
774 |
ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
775 |
|
776 |
hv_store (hv, key, -len, decode_sv (dec), 0); |
777 |
|
778 |
return; |
779 |
} |
780 |
|
781 |
SV *k = decode_sv (dec); |
782 |
SV *v = decode_sv (dec); |
783 |
|
784 |
hv_store_ent (hv, k, v, 0); |
785 |
SvREFCNT_dec (k); |
786 |
|
787 |
fail: |
788 |
; |
789 |
} |
790 |
|
791 |
static SV * |
792 |
decode_hv (dec_t *dec) |
793 |
{ |
794 |
HV *hv = newHV (); |
795 |
|
796 |
DEC_INC_DEPTH; |
797 |
|
798 |
if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
799 |
{ |
800 |
++dec->cur; |
801 |
|
802 |
for (;;) |
803 |
{ |
804 |
WANT (1); |
805 |
|
806 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
807 |
{ |
808 |
++dec->cur; |
809 |
break; |
810 |
} |
811 |
|
812 |
decode_he (dec, hv); |
813 |
} |
814 |
} |
815 |
else |
816 |
{ |
817 |
int pairs = decode_uint (dec); |
818 |
|
819 |
while (pairs--) |
820 |
decode_he (dec, hv); |
821 |
} |
822 |
|
823 |
DEC_DEC_DEPTH; |
824 |
return newRV_noinc ((SV *)hv); |
825 |
|
826 |
fail: |
827 |
SvREFCNT_dec (hv); |
828 |
DEC_DEC_DEPTH; |
829 |
return &PL_sv_undef; |
830 |
} |
831 |
|
832 |
static SV * |
833 |
decode_str (dec_t *dec, int utf8) |
834 |
{ |
835 |
SV *sv = 0; |
836 |
|
837 |
if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
838 |
{ |
839 |
// indefinite length strings |
840 |
++dec->cur; |
841 |
|
842 |
U8 major = *dec->cur & MAJOR_MISC; |
843 |
|
844 |
sv = newSVpvn ("", 0); |
845 |
|
846 |
for (;;) |
847 |
{ |
848 |
WANT (1); |
849 |
|
850 |
if ((*dec->cur - major) > LENGTH_EXT8) |
851 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
852 |
{ |
853 |
++dec->cur; |
854 |
break; |
855 |
} |
856 |
else |
857 |
ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
858 |
|
859 |
STRLEN len = decode_uint (dec); |
860 |
|
861 |
WANT (len); |
862 |
sv_catpvn (sv, dec->cur, len); |
863 |
dec->cur += len; |
864 |
} |
865 |
} |
866 |
else |
867 |
{ |
868 |
STRLEN len = decode_uint (dec); |
869 |
|
870 |
WANT (len); |
871 |
sv = newSVpvn (dec->cur, len); |
872 |
dec->cur += len; |
873 |
|
874 |
if (ecb_expect_false (dec->stringref) |
875 |
&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
876 |
av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
877 |
} |
878 |
|
879 |
if (utf8) |
880 |
{ |
881 |
if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
882 |
if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
883 |
ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
884 |
|
885 |
SvUTF8_on (sv); |
886 |
} |
887 |
|
888 |
return sv; |
889 |
|
890 |
fail: |
891 |
SvREFCNT_dec (sv); |
892 |
return &PL_sv_undef; |
893 |
} |
894 |
|
895 |
static SV * |
896 |
decode_tagged (dec_t *dec) |
897 |
{ |
898 |
SV *sv = 0; |
899 |
UV tag = decode_uint (dec); |
900 |
|
901 |
WANT (1); |
902 |
|
903 |
switch (tag) |
904 |
{ |
905 |
case CBOR_TAG_MAGIC: |
906 |
sv = decode_sv (dec); |
907 |
break; |
908 |
|
909 |
case CBOR_TAG_INDIRECTION: |
910 |
sv = newRV_noinc (decode_sv (dec)); |
911 |
break; |
912 |
|
913 |
case CBOR_TAG_STRINGREF_NAMESPACE: |
914 |
{ |
915 |
ENTER; SAVETMPS; |
916 |
|
917 |
SAVESPTR (dec->stringref); |
918 |
dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
919 |
|
920 |
sv = decode_sv (dec); |
921 |
|
922 |
FREETMPS; LEAVE; |
923 |
} |
924 |
break; |
925 |
|
926 |
case CBOR_TAG_STRINGREF: |
927 |
{ |
928 |
if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
929 |
ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
930 |
|
931 |
UV idx = decode_uint (dec); |
932 |
|
933 |
if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
934 |
ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
935 |
|
936 |
sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
937 |
} |
938 |
break; |
939 |
|
940 |
case CBOR_TAG_VALUE_SHAREABLE: |
941 |
{ |
942 |
if (ecb_expect_false (!dec->shareable)) |
943 |
dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
944 |
|
945 |
if (dec->cbor.flags & F_ALLOW_CYCLES) |
946 |
{ |
947 |
sv = newSV (0); |
948 |
av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
949 |
|
950 |
SV *osv = decode_sv (dec); |
951 |
sv_setsv (sv, osv); |
952 |
SvREFCNT_dec_NN (osv); |
953 |
} |
954 |
else |
955 |
{ |
956 |
av_push (dec->shareable, &PL_sv_undef); |
957 |
int idx = AvFILLp (dec->shareable); |
958 |
sv = decode_sv (dec); |
959 |
av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
960 |
} |
961 |
} |
962 |
break; |
963 |
|
964 |
case CBOR_TAG_VALUE_SHAREDREF: |
965 |
{ |
966 |
if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
967 |
ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
968 |
|
969 |
UV idx = decode_uint (dec); |
970 |
|
971 |
if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
972 |
ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
973 |
|
974 |
sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
975 |
|
976 |
if (sv == &PL_sv_undef) |
977 |
ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
978 |
} |
979 |
break; |
980 |
|
981 |
case CBOR_TAG_PERL_OBJECT: |
982 |
{ |
983 |
sv = decode_sv (dec); |
984 |
|
985 |
if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
986 |
ERR ("corrupted CBOR data (non-array perl object)"); |
987 |
|
988 |
AV *av = (AV *)SvRV (sv); |
989 |
int len = av_len (av) + 1; |
990 |
HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
991 |
|
992 |
if (!stash) |
993 |
ERR ("cannot decode perl-object (package does not exist)"); |
994 |
|
995 |
GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
996 |
|
997 |
if (!method) |
998 |
ERR ("cannot decode perl-object (package does not have a THAW method)"); |
999 |
|
1000 |
dSP; |
1001 |
|
1002 |
ENTER; SAVETMPS; |
1003 |
PUSHMARK (SP); |
1004 |
EXTEND (SP, len + 1); |
1005 |
// we re-bless the reference to get overload and other niceties right |
1006 |
PUSHs (*av_fetch (av, 0, 1)); |
1007 |
PUSHs (sv_cbor); |
1008 |
|
1009 |
int i; |
1010 |
|
1011 |
for (i = 1; i < len; ++i) |
1012 |
PUSHs (*av_fetch (av, i, 1)); |
1013 |
|
1014 |
PUTBACK; |
1015 |
call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
1016 |
SPAGAIN; |
1017 |
|
1018 |
if (SvTRUE (ERRSV)) |
1019 |
{ |
1020 |
FREETMPS; LEAVE; |
1021 |
ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1022 |
} |
1023 |
|
1024 |
SvREFCNT_dec (sv); |
1025 |
sv = SvREFCNT_inc (POPs); |
1026 |
|
1027 |
PUTBACK; |
1028 |
|
1029 |
FREETMPS; LEAVE; |
1030 |
} |
1031 |
break; |
1032 |
|
1033 |
default: |
1034 |
{ |
1035 |
sv = decode_sv (dec); |
1036 |
|
1037 |
dSP; |
1038 |
ENTER; SAVETMPS; |
1039 |
SAVESTACK_POS (); |
1040 |
PUSHMARK (SP); |
1041 |
EXTEND (SP, 2); |
1042 |
PUSHs (newSVuv (tag)); |
1043 |
PUSHs (sv); |
1044 |
|
1045 |
PUTBACK; |
1046 |
int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
1047 |
SPAGAIN; |
1048 |
|
1049 |
if (SvTRUE (ERRSV)) |
1050 |
{ |
1051 |
FREETMPS; LEAVE; |
1052 |
ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
1053 |
} |
1054 |
|
1055 |
if (count) |
1056 |
{ |
1057 |
SvREFCNT_dec (sv); |
1058 |
sv = SvREFCNT_inc (POPs); |
1059 |
} |
1060 |
else |
1061 |
{ |
1062 |
AV *av = newAV (); |
1063 |
av_push (av, newSVuv (tag)); |
1064 |
av_push (av, sv); |
1065 |
|
1066 |
HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1067 |
? cbor_tagged_stash |
1068 |
: gv_stashpv ("CBOR::XS::Tagged" , 1); |
1069 |
sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
1070 |
} |
1071 |
|
1072 |
PUTBACK; |
1073 |
|
1074 |
FREETMPS; LEAVE; |
1075 |
} |
1076 |
break; |
1077 |
} |
1078 |
|
1079 |
return sv; |
1080 |
|
1081 |
fail: |
1082 |
SvREFCNT_dec (sv); |
1083 |
return &PL_sv_undef; |
1084 |
} |
1085 |
|
1086 |
static SV * |
1087 |
decode_sv (dec_t *dec) |
1088 |
{ |
1089 |
WANT (1); |
1090 |
|
1091 |
switch (*dec->cur >> MAJOR_SHIFT) |
1092 |
{ |
1093 |
case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
1094 |
case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
1095 |
case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
1096 |
case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
1097 |
case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
1098 |
case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
1099 |
case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
1100 |
|
1101 |
case MAJOR_MISC >> MAJOR_SHIFT: |
1102 |
switch (*dec->cur++ & MINOR_MASK) |
1103 |
{ |
1104 |
case SIMPLE_FALSE: |
1105 |
#if CBOR_SLOW |
1106 |
types_false = get_bool ("Types::Serialiser::false"); |
1107 |
#endif |
1108 |
return newSVsv (types_false); |
1109 |
case SIMPLE_TRUE: |
1110 |
#if CBOR_SLOW |
1111 |
types_true = get_bool ("Types::Serialiser::true"); |
1112 |
#endif |
1113 |
return newSVsv (types_true); |
1114 |
case SIMPLE_NULL: |
1115 |
return newSVsv (&PL_sv_undef); |
1116 |
case SIMPLE_UNDEF: |
1117 |
#if CBOR_SLOW |
1118 |
types_error = get_bool ("Types::Serialiser::error"); |
1119 |
#endif |
1120 |
return newSVsv (types_error); |
1121 |
|
1122 |
case MISC_FLOAT16: |
1123 |
{ |
1124 |
WANT (2); |
1125 |
|
1126 |
uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1127 |
dec->cur += 2; |
1128 |
|
1129 |
return newSVnv (ecb_binary16_to_float (fp)); |
1130 |
} |
1131 |
|
1132 |
case MISC_FLOAT32: |
1133 |
{ |
1134 |
uint32_t fp; |
1135 |
WANT (4); |
1136 |
memcpy (&fp, dec->cur, 4); |
1137 |
dec->cur += 4; |
1138 |
|
1139 |
if (!ecb_big_endian ()) |
1140 |
fp = ecb_bswap32 (fp); |
1141 |
|
1142 |
return newSVnv (ecb_binary32_to_float (fp)); |
1143 |
} |
1144 |
|
1145 |
case MISC_FLOAT64: |
1146 |
{ |
1147 |
uint64_t fp; |
1148 |
WANT (8); |
1149 |
memcpy (&fp, dec->cur, 8); |
1150 |
dec->cur += 8; |
1151 |
|
1152 |
if (!ecb_big_endian ()) |
1153 |
fp = ecb_bswap64 (fp); |
1154 |
|
1155 |
return newSVnv (ecb_binary64_to_double (fp)); |
1156 |
} |
1157 |
|
1158 |
// 0..19 unassigned simple |
1159 |
// 24 reserved + unassigned simple (reserved values are not encodable) |
1160 |
// 28-30 unassigned misc |
1161 |
// 31 break code |
1162 |
default: |
1163 |
ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
1164 |
} |
1165 |
|
1166 |
break; |
1167 |
} |
1168 |
|
1169 |
fail: |
1170 |
return &PL_sv_undef; |
1171 |
} |
1172 |
|
1173 |
static SV * |
1174 |
decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1175 |
{ |
1176 |
dec_t dec = { 0 }; |
1177 |
SV *sv; |
1178 |
STRLEN len; |
1179 |
char *data = SvPVbyte (string, len); |
1180 |
|
1181 |
if (len > cbor->max_size && cbor->max_size) |
1182 |
croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
1183 |
(unsigned long)len, (unsigned long)cbor->max_size); |
1184 |
|
1185 |
dec.cbor = *cbor; |
1186 |
dec.cur = (U8 *)data; |
1187 |
dec.end = (U8 *)data + len; |
1188 |
|
1189 |
sv = decode_sv (&dec); |
1190 |
|
1191 |
if (offset_return) |
1192 |
*offset_return = dec.cur; |
1193 |
|
1194 |
if (!(offset_return || !sv)) |
1195 |
if (dec.cur != dec.end && !dec.err) |
1196 |
dec.err = "garbage after CBOR object"; |
1197 |
|
1198 |
if (dec.err) |
1199 |
{ |
1200 |
if (dec.shareable) |
1201 |
{ |
1202 |
// need to break cyclic links, which whould all be in shareable |
1203 |
int i; |
1204 |
SV **svp; |
1205 |
|
1206 |
for (i = av_len (dec.shareable) + 1; i--; ) |
1207 |
if ((svp = av_fetch (dec.shareable, i, 0))) |
1208 |
sv_setsv (*svp, &PL_sv_undef); |
1209 |
} |
1210 |
|
1211 |
SvREFCNT_dec (sv); |
1212 |
croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1213 |
} |
1214 |
|
1215 |
sv = sv_2mortal (sv); |
1216 |
|
1217 |
return sv; |
1218 |
} |
1219 |
|
1220 |
///////////////////////////////////////////////////////////////////////////// |
1221 |
// incremental parser |
1222 |
|
1223 |
#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
1224 |
|
1225 |
// returns 0 for notyet, 1 for success or error |
1226 |
static int |
1227 |
incr_parse (CBOR *self, SV *cborstr) |
1228 |
{ |
1229 |
STRLEN cur; |
1230 |
SvPV (cborstr, cur); |
1231 |
|
1232 |
while (ecb_expect_true (self->incr_need <= cur)) |
1233 |
{ |
1234 |
// table of integer count bytes |
1235 |
static I8 incr_len[MINOR_MASK + 1] = { |
1236 |
0, 0, 0, 0, 0, 0, 0, 0, |
1237 |
0, 0, 0, 0, 0, 0, 0, 0, |
1238 |
0, 0, 0, 0, 0, 0, 0, 0, |
1239 |
1, 2, 4, 8,-1,-1,-1,-2 |
1240 |
}; |
1241 |
|
1242 |
const U8 *p = SvPVX (cborstr) + self->incr_pos; |
1243 |
U8 m = *p & MINOR_MASK; |
1244 |
IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
1245 |
I8 ilen = incr_len[m]; |
1246 |
|
1247 |
self->incr_need = self->incr_pos + 1; |
1248 |
|
1249 |
if (ecb_expect_false (ilen < 0)) |
1250 |
{ |
1251 |
if (m != MINOR_INDEF) |
1252 |
return 1; // error |
1253 |
|
1254 |
if (*p == (MAJOR_MISC | MINOR_INDEF)) |
1255 |
{ |
1256 |
if (count >= 0) |
1257 |
return 1; // error |
1258 |
|
1259 |
count = 1; |
1260 |
} |
1261 |
else |
1262 |
{ |
1263 |
av_push (self->incr_count, newSViv (-1)); //TODO: nest |
1264 |
count = -1; |
1265 |
} |
1266 |
} |
1267 |
else |
1268 |
{ |
1269 |
self->incr_need += ilen; |
1270 |
if (ecb_expect_false (self->incr_need > cur)) |
1271 |
return 0; |
1272 |
|
1273 |
int major = *p >> MAJOR_SHIFT; |
1274 |
|
1275 |
switch (major) |
1276 |
{ |
1277 |
case MAJOR_TAG >> MAJOR_SHIFT: |
1278 |
++count; // tags merely prefix another value |
1279 |
break; |
1280 |
|
1281 |
case MAJOR_BYTES >> MAJOR_SHIFT: |
1282 |
case MAJOR_TEXT >> MAJOR_SHIFT: |
1283 |
case MAJOR_ARRAY >> MAJOR_SHIFT: |
1284 |
case MAJOR_MAP >> MAJOR_SHIFT: |
1285 |
{ |
1286 |
UV len; |
1287 |
|
1288 |
if (ecb_expect_false (ilen)) |
1289 |
{ |
1290 |
len = 0; |
1291 |
|
1292 |
do { |
1293 |
len = (len << 8) | *++p; |
1294 |
} while (--ilen); |
1295 |
} |
1296 |
else |
1297 |
len = m; |
1298 |
|
1299 |
switch (major) |
1300 |
{ |
1301 |
case MAJOR_BYTES >> MAJOR_SHIFT: |
1302 |
case MAJOR_TEXT >> MAJOR_SHIFT: |
1303 |
self->incr_need += len; |
1304 |
if (ecb_expect_false (self->incr_need > cur)) |
1305 |
return 0; |
1306 |
|
1307 |
break; |
1308 |
|
1309 |
case MAJOR_MAP >> MAJOR_SHIFT: |
1310 |
len <<= 1; |
1311 |
case MAJOR_ARRAY >> MAJOR_SHIFT: |
1312 |
if (len) |
1313 |
{ |
1314 |
av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
1315 |
count = len + 1; |
1316 |
} |
1317 |
break; |
1318 |
} |
1319 |
} |
1320 |
} |
1321 |
} |
1322 |
|
1323 |
self->incr_pos = self->incr_need; |
1324 |
|
1325 |
if (count > 0) |
1326 |
{ |
1327 |
while (!--count) |
1328 |
{ |
1329 |
if (!AvFILLp (self->incr_count)) |
1330 |
return 1; // done |
1331 |
|
1332 |
SvREFCNT_dec_NN (av_pop (self->incr_count)); |
1333 |
count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
1334 |
} |
1335 |
|
1336 |
SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
1337 |
} |
1338 |
} |
1339 |
|
1340 |
return 0; |
1341 |
} |
1342 |
|
1343 |
|
1344 |
///////////////////////////////////////////////////////////////////////////// |
1345 |
// XS interface functions |
1346 |
|
1347 |
MODULE = CBOR::XS PACKAGE = CBOR::XS |
1348 |
|
1349 |
BOOT: |
1350 |
{ |
1351 |
cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
1352 |
cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
1353 |
|
1354 |
types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1355 |
types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
1356 |
|
1357 |
types_true = get_bool ("Types::Serialiser::true" ); |
1358 |
types_false = get_bool ("Types::Serialiser::false"); |
1359 |
types_error = get_bool ("Types::Serialiser::error"); |
1360 |
|
1361 |
default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1362 |
|
1363 |
sv_cbor = newSVpv ("CBOR", 0); |
1364 |
SvREADONLY_on (sv_cbor); |
1365 |
} |
1366 |
|
1367 |
PROTOTYPES: DISABLE |
1368 |
|
1369 |
void CLONE (...) |
1370 |
CODE: |
1371 |
cbor_stash = 0; |
1372 |
cbor_tagged_stash = 0; |
1373 |
types_error_stash = 0; |
1374 |
types_boolean_stash = 0; |
1375 |
|
1376 |
void new (char *klass) |
1377 |
PPCODE: |
1378 |
{ |
1379 |
SV *pv = NEWSV (0, sizeof (CBOR)); |
1380 |
SvPOK_only (pv); |
1381 |
cbor_init ((CBOR *)SvPVX (pv)); |
1382 |
XPUSHs (sv_2mortal (sv_bless ( |
1383 |
newRV_noinc (pv), |
1384 |
strEQ (klass, "CBOR::XS") ? CBOR_STASH : gv_stashpv (klass, 1) |
1385 |
))); |
1386 |
} |
1387 |
|
1388 |
void shrink (CBOR *self, int enable = 1) |
1389 |
ALIAS: |
1390 |
shrink = F_SHRINK |
1391 |
allow_unknown = F_ALLOW_UNKNOWN |
1392 |
allow_sharing = F_ALLOW_SHARING |
1393 |
allow_cycles = F_ALLOW_CYCLES |
1394 |
pack_strings = F_PACK_STRINGS |
1395 |
validate_utf8 = F_VALIDATE_UTF8 |
1396 |
PPCODE: |
1397 |
{ |
1398 |
if (enable) |
1399 |
self->flags |= ix; |
1400 |
else |
1401 |
self->flags &= ~ix; |
1402 |
|
1403 |
XPUSHs (ST (0)); |
1404 |
} |
1405 |
|
1406 |
void get_shrink (CBOR *self) |
1407 |
ALIAS: |
1408 |
get_shrink = F_SHRINK |
1409 |
get_allow_unknown = F_ALLOW_UNKNOWN |
1410 |
get_allow_sharing = F_ALLOW_SHARING |
1411 |
get_allow_cycles = F_ALLOW_CYCLES |
1412 |
get_pack_strings = F_PACK_STRINGS |
1413 |
get_validate_utf8 = F_VALIDATE_UTF8 |
1414 |
PPCODE: |
1415 |
XPUSHs (boolSV (self->flags & ix)); |
1416 |
|
1417 |
void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1418 |
PPCODE: |
1419 |
self->max_depth = max_depth; |
1420 |
XPUSHs (ST (0)); |
1421 |
|
1422 |
U32 get_max_depth (CBOR *self) |
1423 |
CODE: |
1424 |
RETVAL = self->max_depth; |
1425 |
OUTPUT: |
1426 |
RETVAL |
1427 |
|
1428 |
void max_size (CBOR *self, U32 max_size = 0) |
1429 |
PPCODE: |
1430 |
self->max_size = max_size; |
1431 |
XPUSHs (ST (0)); |
1432 |
|
1433 |
int get_max_size (CBOR *self) |
1434 |
CODE: |
1435 |
RETVAL = self->max_size; |
1436 |
OUTPUT: |
1437 |
RETVAL |
1438 |
|
1439 |
void filter (CBOR *self, SV *filter = 0) |
1440 |
PPCODE: |
1441 |
SvREFCNT_dec (self->filter); |
1442 |
self->filter = filter ? newSVsv (filter) : filter; |
1443 |
XPUSHs (ST (0)); |
1444 |
|
1445 |
SV *get_filter (CBOR *self) |
1446 |
CODE: |
1447 |
RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
1448 |
OUTPUT: |
1449 |
RETVAL |
1450 |
|
1451 |
void encode (CBOR *self, SV *scalar) |
1452 |
PPCODE: |
1453 |
PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1454 |
XPUSHs (scalar); |
1455 |
|
1456 |
void decode (CBOR *self, SV *cborstr) |
1457 |
PPCODE: |
1458 |
PUTBACK; cborstr = decode_cbor (cborstr, self, 0); SPAGAIN; |
1459 |
XPUSHs (cborstr); |
1460 |
|
1461 |
void decode_prefix (CBOR *self, SV *cborstr) |
1462 |
PPCODE: |
1463 |
{ |
1464 |
SV *sv; |
1465 |
char *offset; |
1466 |
PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
1467 |
EXTEND (SP, 2); |
1468 |
PUSHs (sv); |
1469 |
PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1470 |
} |
1471 |
|
1472 |
void incr_parse (CBOR *self, SV *cborstr) |
1473 |
ALIAS: |
1474 |
incr_parse_multiple = 1 |
1475 |
PPCODE: |
1476 |
{ |
1477 |
if (SvUTF8 (cborstr)) |
1478 |
sv_utf8_downgrade (cborstr, 0); |
1479 |
|
1480 |
if (!self->incr_count) |
1481 |
{ |
1482 |
self->incr_count = newAV (); |
1483 |
self->incr_pos = 0; |
1484 |
self->incr_need = 1; |
1485 |
|
1486 |
av_push (self->incr_count, newSViv (1)); |
1487 |
} |
1488 |
|
1489 |
do |
1490 |
{ |
1491 |
if (!incr_parse (self, cborstr)) |
1492 |
{ |
1493 |
if (self->incr_need > self->max_size && self->max_size) |
1494 |
croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
1495 |
(unsigned long)self->incr_need, (unsigned long)self->max_size); |
1496 |
|
1497 |
break; |
1498 |
} |
1499 |
|
1500 |
SV *sv; |
1501 |
char *offset; |
1502 |
|
1503 |
PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
1504 |
XPUSHs (sv); |
1505 |
|
1506 |
sv_chop (cborstr, offset); |
1507 |
|
1508 |
av_clear (self->incr_count); |
1509 |
av_push (self->incr_count, newSViv (1)); |
1510 |
|
1511 |
self->incr_pos = 0; |
1512 |
self->incr_need = self->incr_pos + 1; |
1513 |
} |
1514 |
while (ix); |
1515 |
} |
1516 |
|
1517 |
void incr_reset (CBOR *self) |
1518 |
CODE: |
1519 |
{ |
1520 |
SvREFCNT_dec (self->incr_count); |
1521 |
self->incr_count = 0; |
1522 |
} |
1523 |
|
1524 |
void DESTROY (CBOR *self) |
1525 |
PPCODE: |
1526 |
cbor_free (self); |
1527 |
|
1528 |
PROTOTYPES: ENABLE |
1529 |
|
1530 |
void encode_cbor (SV *scalar) |
1531 |
ALIAS: |
1532 |
encode_cbor = 0 |
1533 |
encode_cbor_sharing = F_ALLOW_SHARING |
1534 |
PPCODE: |
1535 |
{ |
1536 |
CBOR cbor; |
1537 |
cbor_init (&cbor); |
1538 |
cbor.flags |= ix; |
1539 |
PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1540 |
XPUSHs (scalar); |
1541 |
} |
1542 |
|
1543 |
void decode_cbor (SV *cborstr) |
1544 |
PPCODE: |
1545 |
{ |
1546 |
CBOR cbor; |
1547 |
cbor_init (&cbor); |
1548 |
PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
1549 |
XPUSHs (cborstr); |
1550 |
} |
1551 |
|