ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.5 by root, Sat Oct 26 21:14:20 2013 UTC vs.
Revision 1.62 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
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_dec_NN
27# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
28#endif
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
67// known tags
68enum cbor_tag
69{
70 // extensions
71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
72 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
73 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
74 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
76 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
77 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
78
79 // rfc7049
80 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
81 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
82 CBOR_TAG_POS_BIGNUM = 2, // byte string
83 CBOR_TAG_NEG_BIGNUM = 3, // byte string
84 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
85 CBOR_TAG_BIGFLOAT = 5, // array
86
87 CBOR_TAG_CONV_B64U = 21, // base64url, any
88 CBOR_TAG_CONV_B64 = 22, // base64, any
89 CBOR_TAG_CONV_HEX = 23, // base16, any
90 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
91
92 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
93 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
94 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
95 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
96 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
97
98 CBOR_TAG_MAGIC = 55799, // self-describe cbor
99};
100
14#define F_SHRINK 0x00000200UL 101#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 102#define F_ALLOW_UNKNOWN 0x00000002UL
103#define F_ALLOW_SHARING 0x00000004UL
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
16 110
17#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
18 112
19#define SB do { 113#define SB do {
20#define SE } while (0) 114#define SE } while (0)
31#else 125#else
32# define CBOR_SLOW 0 126# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 127# define CBOR_STASH cbor_stash
34#endif 128#endif
35 129
36static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 130static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 131static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 132
39typedef struct { 133typedef struct {
40 U32 flags; 134 U32 flags;
41 U32 max_depth; 135 U32 max_depth;
42 STRLEN max_size; 136 STRLEN max_size;
137 SV *filter;
43 138
44 SV *cb_object; 139 // for the incremental parser
45 HV *cb_sk_object; 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.
46} CBOR; 143} CBOR;
47 144
48ecb_inline void 145ecb_inline void
49cbor_init (CBOR *cbor) 146cbor_init (CBOR *cbor)
50{ 147{
51 Zero (cbor, 1, CBOR); 148 Zero (cbor, 1, CBOR);
52 cbor->max_depth = 512; 149 cbor->max_depth = 512;
150}
151
152ecb_inline void
153cbor_free (CBOR *cbor)
154{
155 SvREFCNT_dec (cbor->filter);
156 SvREFCNT_dec (cbor->incr_count);
53} 157}
54 158
55///////////////////////////////////////////////////////////////////////////// 159/////////////////////////////////////////////////////////////////////////////
56// utility functions 160// utility functions
57 161
79 SvPV_renew (sv, SvCUR (sv) + 1); 183 SvPV_renew (sv, SvCUR (sv) + 1);
80#endif 184#endif
81 } 185 }
82} 186}
83 187
84///////////////////////////////////////////////////////////////////////////// 188// minimum length of a string to be registered for stringref
85// fp hell 189ecb_inline int
86 190minimum_string_length (UV idx)
87//TODO 191{
192 return idx <= 23 ? 3
193 : idx <= 0xffU ? 4
194 : idx <= 0xffffU ? 5
195 : idx <= 0xffffffffU ? 7
196 : 11;
197}
88 198
89///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
90// encoder 200// encoder
91 201
92// structure used for encoding CBOR 202// structure used for encoding CBOR
95 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
96 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
97 SV *sv; // result scalar 207 SV *sv; // result scalar
98 CBOR cbor; 208 CBOR cbor;
99 U32 depth; // recursion level 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;
100} enc_t; 214} enc_t;
101 215
102ecb_inline void 216ecb_inline void
103need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
104{ 218{
105 if (ecb_expect_false (enc->cur + len >= enc->end)) 219 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
106 { 220 {
107 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
108 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
109 enc->cur = SvPVX (enc->sv) + cur; 223 enc->cur = SvPVX (enc->sv) + cur;
110 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
121static void 235static void
122encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
123{ 237{
124 need (enc, 9); 238 need (enc, 9);
125 239
126 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
127 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
128 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
129 { 243 {
130 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
131 *enc->cur++ = len; 245 *enc->cur++ = len;
132 } 246 }
133 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
134 { 248 {
135 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
136 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
137 *enc->cur++ = len; 251 *enc->cur++ = len;
138 } 252 }
139 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
140 { 254 {
141 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
142 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
143 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
144 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
145 *enc->cur++ = len; 259 *enc->cur++ = len;
146 } 260 }
147 else 261 else
148 { 262 {
149 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
150 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
151 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
152 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
153 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
154 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
156 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
157 *enc->cur++ = len; 271 *enc->cur++ = len;
158 } 272 }
159} 273}
160 274
161static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
282ecb_noinline static void
162encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
163{ 284{
164 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 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);
165 need (enc, len); 315 need (enc, len);
166 memcpy (enc->cur, str, len); 316 memcpy (enc->cur, str, len);
167 enc->cur += len; 317 enc->cur += len;
168} 318}
169 319
320ecb_inline void
321encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
322{
323 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
324 {
325 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
326
327 if (SvOK (*svp))
328 {
329 // already registered, use stringref
330 encode_tag (enc, CBOR_TAG_STRINGREF);
331 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
332 return;
333 }
334 else if (len >= minimum_string_length (enc->stringref_idx))
335 {
336 // register only
337 sv_setuv (*svp, enc->stringref_idx);
338 ++enc->stringref_idx;
339 }
340 }
341
342 encode_str (enc, upgrade_utf8, utf8, str, len);
343}
344
170static void encode_sv (enc_t *enc, SV *sv); 345static void encode_sv (enc_t *enc, SV *sv);
171 346
172static void 347static void
173encode_av (enc_t *enc, AV *av) 348encode_av (enc_t *enc, AV *av)
174{ 349{
177 if (enc->depth >= enc->cbor.max_depth) 352 if (enc->depth >= enc->cbor.max_depth)
178 croak (ERR_NESTING_EXCEEDED); 353 croak (ERR_NESTING_EXCEEDED);
179 354
180 ++enc->depth; 355 ++enc->depth;
181 356
182 encode_uint (enc, 0x80, len + 1); 357 encode_uint (enc, MAJOR_ARRAY, len + 1);
183 358
359 if (SvMAGICAL (av))
184 for (i = 0; i <= len; ++i) 360 for (i = 0; i <= len; ++i)
185 { 361 {
186 SV **svp = av_fetch (av, i, 0); 362 SV **svp = av_fetch (av, i, 0);
187 encode_sv (enc, svp ? *svp : &PL_sv_undef); 363 encode_sv (enc, svp ? *svp : &PL_sv_undef);
188 } 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 }
189 371
190 --enc->depth; 372 --enc->depth;
191} 373}
192 374
193static void 375static void
202 384
203 int pairs = hv_iterinit (hv); 385 int pairs = hv_iterinit (hv);
204 int mg = SvMAGICAL (hv); 386 int mg = SvMAGICAL (hv);
205 387
206 if (mg) 388 if (mg)
207 encode_ch (enc, 0xa0 | 31); 389 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
208 else 390 else
209 encode_uint (enc, 0xa0, pairs); 391 encode_uint (enc, MAJOR_MAP, pairs);
210 392
211 while ((he = hv_iternext (hv))) 393 while ((he = hv_iternext (hv)))
212 { 394 {
213 if (HeKLEN (he) == HEf_SVKEY) 395 if (HeKLEN (he) == HEf_SVKEY)
214 encode_sv (enc, HeSVKEY (he)); 396 encode_sv (enc, HeSVKEY (he));
215 else 397 else
216 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));
217 399
218 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));
219 } 401 }
220 402
221 if (mg) 403 if (mg)
222 encode_ch (enc, 0xe0 | 31); 404 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
223 405
224 --enc->depth; 406 --enc->depth;
225} 407}
226 408
227// encode objects, arrays and special \0=false and \1=true values. 409// encode objects, arrays and special \0=false and \1=true values.
228static void 410static void
229encode_rv (enc_t *enc, SV *sv) 411encode_rv (enc_t *enc, SV *sv)
230{ 412{
231 svtype svt;
232
233 SvGETMAGIC (sv); 413 SvGETMAGIC (sv);
414
234 svt = SvTYPE (sv); 415 svtype svt = SvTYPE (sv);
235 416
236 if (ecb_expect_false (SvOBJECT (sv))) 417 if (ecb_expect_false (SvOBJECT (sv)))
237 { 418 {
238 HV *stash = !CBOR_SLOW || cbor_boolean_stash 419 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
239 ? cbor_boolean_stash 420 ? types_boolean_stash
421 : gv_stashpv ("Types::Serialiser::Boolean", 1);
422 HV *error_stash = !CBOR_SLOW || types_error_stash
423 ? types_error_stash
424 : gv_stashpv ("Types::Serialiser::Error", 1);
425 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
426 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Boolean", 1); 427 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 428
242 if (SvSTASH (sv) == stash) 429 HV *stash = SvSTASH (sv);
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 430
431 if (stash == boolean_stash)
432 {
433 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
434 return;
435 }
436 else if (stash == error_stash)
437 {
438 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
439 return;
440 }
441 else if (stash == tagged_stash)
442 {
443 if (svt != SVt_PVAV)
444 croak ("encountered CBOR::XS::Tagged object that isn't an array");
445
446 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 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))
462 {
463 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
464 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
465 return;
466 }
244 else 467 else
245 { 468 {
246#if 0 //TODO 469 sv_setuv (*svp, enc->shareable_idx);
247 if (enc->cbor.flags & F_CONV_BLESSED) 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))));
483 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
248 { 484 {
485 dSP;
486
487 ENTER; SAVETMPS;
488 PUSHMARK (SP);
249 // 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
250 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
251
252 if (to_cbor)
253 {
254 dSP;
255
256 ENTER; SAVETMPS; PUSHMARK (SP);
257 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 490 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
258 491
259 // calling with G_SCALAR ensures that we always get a 1 return value
260 PUTBACK; 492 PUTBACK;
493 // G_SCALAR ensures that return value is 1
261 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 494 call_sv ((SV *)GvCV (method), G_SCALAR);
262 SPAGAIN; 495 SPAGAIN;
263 496
264 // catch this surprisingly common error 497 // catch this surprisingly common error
265 if (SvROK (TOPs) && SvRV (TOPs) == sv) 498 if (SvROK (TOPs) && SvRV (TOPs) == sv)
266 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 499 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
267 500
268 sv = POPs;
269 PUTBACK;
270
271 encode_sv (enc, sv); 501 encode_sv (enc, POPs);
272 502
503 PUTBACK;
504
273 FREETMPS; LEAVE; 505 FREETMPS; LEAVE;
274 }
275 else if (enc->cbor.flags & F_ALLOW_BLESSED)
276 encode_str (enc, "null", 4, 0);
277 else
278 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
279 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
280 } 506 }
281 else if (enc->cbor.flags & F_ALLOW_BLESSED) 507 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
282 encode_str (enc, "null", 4, 0); 508 {
509 dSP;
510
511 ENTER; SAVETMPS;
512 SAVESTACK_POS ();
513 PUSHMARK (SP);
514 EXTEND (SP, 2);
515 // we re-bless the reference to get overload and other niceties right
516 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
517 PUSHs (sv_cbor);
518
519 PUTBACK;
520 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
521 SPAGAIN;
522
523 // catch this surprisingly common error
524 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
525 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
526
527 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
528 encode_uint (enc, MAJOR_ARRAY, count + 1);
529 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
530
531 while (count)
532 encode_sv (enc, SP[1 - count--]);
533
534 PUTBACK;
535
536 FREETMPS; LEAVE;
537 }
283 else 538 else
284 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 539 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
285 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 540 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
286#endif
287 }
288 } 541 }
289 else if (svt == SVt_PVHV) 542 else if (svt == SVt_PVHV)
290 encode_hv (enc, (HV *)sv); 543 encode_hv (enc, (HV *)sv);
291 else if (svt == SVt_PVAV) 544 else if (svt == SVt_PVAV)
292 encode_av (enc, (AV *)sv); 545 encode_av (enc, (AV *)sv);
293 else if (svt < SVt_PVAV)
294 {
295 STRLEN len = 0;
296 char *pv = svt ? SvPV (sv, len) : 0;
297
298 if (len == 1 && *pv == '1')
299 encode_ch (enc, 0xe0 | 21);
300 else if (len == 1 && *pv == '0')
301 encode_ch (enc, 0xe0 | 20);
302 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
303 encode_ch (enc, 0xe0 | 23);
304 else
305 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
309 encode_ch (enc, 0xe0 | 23);
310 else 546 else
311 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 547 {
312 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 548 encode_tag (enc, CBOR_TAG_INDIRECTION);
549 encode_sv (enc, sv);
550 }
313} 551}
314 552
315static void 553static void
316encode_nv (enc_t *enc, SV *sv) 554encode_nv (enc_t *enc, SV *sv)
317{ 555{
318 double nv = SvNVX (sv); 556 double nv = SvNVX (sv);
319 557
320 need (enc, 9); 558 need (enc, 9);
321 559
322 if (ecb_expect_false (nv == (U32)nv)) 560 if (ecb_expect_false (nv == (NV)(U32)nv))
323 encode_uint (enc, 0x00, (U32)nv); 561 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
324 //TODO: maybe I32? 562 //TODO: maybe I32?
325 else if (ecb_expect_false (nv == (float)nv)) 563 else if (ecb_expect_false (nv == (float)nv))
326 { 564 {
327 uint32_t fp = ecb_float_to_binary32 (nv); 565 uint32_t fp = ecb_float_to_binary32 (nv);
328 566
329 *enc->cur++ = 0xe0 | 26; 567 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
330 568
331 if (!ecb_big_endian ()) 569 if (!ecb_big_endian ())
332 fp = ecb_bswap32 (fp); 570 fp = ecb_bswap32 (fp);
333 571
334 memcpy (enc->cur, &fp, 4); 572 memcpy (enc->cur, &fp, 4);
336 } 574 }
337 else 575 else
338 { 576 {
339 uint64_t fp = ecb_double_to_binary64 (nv); 577 uint64_t fp = ecb_double_to_binary64 (nv);
340 578
341 *enc->cur++ = 0xe0 | 27; 579 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
342 580
343 if (!ecb_big_endian ()) 581 if (!ecb_big_endian ())
344 fp = ecb_bswap64 (fp); 582 fp = ecb_bswap64 (fp);
345 583
346 memcpy (enc->cur, &fp, 8); 584 memcpy (enc->cur, &fp, 8);
355 593
356 if (SvPOKp (sv)) 594 if (SvPOKp (sv))
357 { 595 {
358 STRLEN len; 596 STRLEN len;
359 char *str = SvPV (sv, len); 597 char *str = SvPV (sv, len);
360 encode_str (enc, SvUTF8 (sv), str, len); 598 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
361 } 599 }
362 else if (SvNOKp (sv)) 600 else if (SvNOKp (sv))
363 encode_nv (enc, sv); 601 encode_nv (enc, sv);
364 else if (SvIOKp (sv)) 602 else if (SvIOKp (sv))
365 { 603 {
366 if (SvIsUV (sv)) 604 if (SvIsUV (sv))
367 encode_uint (enc, 0x00, SvUVX (sv)); 605 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
368 else if (SvIVX (sv) >= 0) 606 else if (SvIVX (sv) >= 0)
369 encode_uint (enc, 0x00, SvIVX (sv)); 607 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
370 else 608 else
371 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 609 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
372 } 610 }
373 else if (SvROK (sv)) 611 else if (SvROK (sv))
374 encode_rv (enc, SvRV (sv)); 612 encode_rv (enc, SvRV (sv));
375 else if (!SvOK (sv)) 613 else if (!SvOK (sv))
376 encode_ch (enc, 0xe0 | 22); 614 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
377 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 615 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
378 encode_ch (enc, 0xe0 | 23); 616 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
379 else 617 else
380 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",
381 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 619 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
382} 620}
383 621
384static SV * 622static SV *
385encode_cbor (SV *scalar, CBOR *cbor) 623encode_cbor (SV *scalar, CBOR *cbor)
386{ 624{
387 enc_t enc; 625 enc_t enc = { 0 };
388 626
389 enc.cbor = *cbor; 627 enc.cbor = *cbor;
390 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 628 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
391 enc.cur = SvPVX (enc.sv); 629 enc.cur = SvPVX (enc.sv);
392 enc.end = SvEND (enc.sv); 630 enc.end = SvEND (enc.sv);
393 enc.depth = 0;
394 631
395 SvPOK_only (enc.sv); 632 SvPOK_only (enc.sv);
633
634 if (cbor->flags & F_PACK_STRINGS)
635 {
636 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
637 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
638 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
639 }
640
396 encode_sv (&enc, scalar); 641 encode_sv (&enc, scalar);
397 642
398 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 643 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
399 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 644 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
400 645
414 U8 *end; // end of input string 659 U8 *end; // end of input string
415 const char *err; // parse error, if != 0 660 const char *err; // parse error, if != 0
416 CBOR cbor; 661 CBOR cbor;
417 U32 depth; // recursion depth 662 U32 depth; // recursion depth
418 U32 maxdepth; // recursion depth limit 663 U32 maxdepth; // recursion depth limit
664 AV *shareable;
665 AV *stringref;
666 SV *decode_tagged;
667 SV *err_sv; // optional sv for error, needs to be freed
419} dec_t; 668} dec_t;
420 669
421#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);
422 677
423#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;
424 681
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
425#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)
426#define DEC_DEC_DEPTH --dec->depth 713#define DEC_DEC_DEPTH --dec->depth
427 714
428static UV 715static UV
429decode_uint (dec_t *dec) 716decode_uint (dec_t *dec)
430{ 717{
431 switch (*dec->cur & 31) 718 U8 m = *dec->cur & MINOR_MASK;
432 { 719 ++dec->cur;
433 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
434 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
435 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
436 return *dec->cur++ & 31;
437 720
438 case 24: 721 if (ecb_expect_true (m < LENGTH_EXT1))
722 return m;
723 else if (ecb_expect_true (m == LENGTH_EXT1))
724 {
439 WANT (2); 725 WANT (1);
440 dec->cur += 2; 726 dec->cur += 1;
441 return dec->cur[-1]; 727 return dec->cur[-1];
442 728 }
443 case 25: 729 else if (ecb_expect_true (m == LENGTH_EXT2))
730 {
444 WANT (3); 731 WANT (2);
445 dec->cur += 3; 732 dec->cur += 2;
446 return (((UV)dec->cur[-2]) << 8) 733 return (((UV)dec->cur[-2]) << 8)
447 | ((UV)dec->cur[-1]); 734 | ((UV)dec->cur[-1]);
448 735 }
449 case 26: 736 else if (ecb_expect_true (m == LENGTH_EXT4))
737 {
450 WANT (5); 738 WANT (4);
451 dec->cur += 5; 739 dec->cur += 4;
452 return (((UV)dec->cur[-4]) << 24) 740 return (((UV)dec->cur[-4]) << 24)
453 | (((UV)dec->cur[-3]) << 16) 741 | (((UV)dec->cur[-3]) << 16)
454 | (((UV)dec->cur[-2]) << 8) 742 | (((UV)dec->cur[-2]) << 8)
455 | ((UV)dec->cur[-1]); 743 | ((UV)dec->cur[-1]);
456 744 }
457 case 27: 745 else if (ecb_expect_true (m == LENGTH_EXT8))
746 {
458 WANT (9); 747 WANT (8);
459 dec->cur += 9; 748 dec->cur += 8;
749
750 return
751#if UVSIZE < 8
752 0
753#else
460 return (((UV)dec->cur[-8]) << 56) 754 (((UV)dec->cur[-8]) << 56)
461 | (((UV)dec->cur[-7]) << 48) 755 | (((UV)dec->cur[-7]) << 48)
462 | (((UV)dec->cur[-6]) << 40) 756 | (((UV)dec->cur[-6]) << 40)
463 | (((UV)dec->cur[-5]) << 32) 757 | (((UV)dec->cur[-5]) << 32)
758#endif
464 | (((UV)dec->cur[-4]) << 24) 759 | (((UV)dec->cur[-4]) << 24)
465 | (((UV)dec->cur[-3]) << 16) 760 | (((UV)dec->cur[-3]) << 16)
466 | (((UV)dec->cur[-2]) << 8) 761 | (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 762 | ((UV)dec->cur[-1]);
468 763 }
469 default: 764 else
470 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 765 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
471 }
472 766
473fail: 767fail:
474 return 0; 768 return 0;
475} 769}
476 770
481{ 775{
482 AV *av = newAV (); 776 AV *av = newAV ();
483 777
484 DEC_INC_DEPTH; 778 DEC_INC_DEPTH;
485 779
486 if ((*dec->cur & 31) == 31) 780 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
487 { 781 {
488 ++dec->cur; 782 ++dec->cur;
489 783
490 for (;;) 784 for (;;)
491 { 785 {
492 WANT (1); 786 WANT (1);
493 787
494 if (*dec->cur == (0xe0 | 31)) 788 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
495 { 789 {
496 ++dec->cur; 790 ++dec->cur;
497 break; 791 break;
498 } 792 }
499 793
500 av_push (av, decode_sv (dec)); 794 av_push (av, decode_sv (dec));
501 } 795 }
502 } 796 }
503 else 797 else
504 { 798 {
505 int i, len = decode_uint (dec); 799 UV i, len = decode_uint (dec);
506 800
801 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
507 av_fill (av, len - 1); 802 av_fill (av, len - 1);
508 803
509 for (i = 0; i < len; ++i) 804 for (i = 0; i < len; ++i)
510 AvARRAY (av)[i] = decode_sv (dec); 805 AvARRAY (av)[i] = decode_sv (dec);
511 } 806 }
517 SvREFCNT_dec (av); 812 SvREFCNT_dec (av);
518 DEC_DEC_DEPTH; 813 DEC_DEC_DEPTH;
519 return &PL_sv_undef; 814 return &PL_sv_undef;
520} 815}
521 816
817static void
818decode_he (dec_t *dec, HV *hv)
819{
820 // for speed reasons, we specialcase single-string
821 // byte or utf-8 strings as keys, but only when !stringref
822
823 if (ecb_expect_true (!dec->stringref))
824 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
825 {
826 STRLEN len = decode_uint (dec);
827 char *key = (char *)dec->cur;
828
829 WANT (len);
830 dec->cur += len;
831
832 hv_store (hv, key, len, decode_sv (dec), 0);
833
834 return;
835 }
836 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
837 {
838 STRLEN len = decode_uint (dec);
839 char *key = (char *)dec->cur;
840
841 WANT (len);
842 dec->cur += len;
843
844 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
845 if (!is_utf8_string (key, len))
846 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
847
848 hv_store (hv, key, -len, decode_sv (dec), 0);
849
850 return;
851 }
852
853 SV *k = decode_sv (dec);
854 SV *v = decode_sv (dec);
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
884 hv_store_ent (hv, k, v, 0);
885 SvREFCNT_dec (k);
886
887fail:
888 ;
889}
890
522static SV * 891static SV *
523decode_hv (dec_t *dec) 892decode_hv (dec_t *dec)
524{ 893{
525 HV *hv = newHV (); 894 HV *hv = newHV ();
526 895
527 DEC_INC_DEPTH; 896 DEC_INC_DEPTH;
528 897
529 if ((*dec->cur & 31) == 31) 898 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
530 { 899 {
531 ++dec->cur; 900 ++dec->cur;
532 901
533 for (;;) 902 for (;;)
534 { 903 {
535 WANT (1); 904 WANT (1);
536 905
537 if (*dec->cur == (0xe0 | 31)) 906 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
538 { 907 {
539 ++dec->cur; 908 ++dec->cur;
540 break; 909 break;
541 } 910 }
542 911
543 SV *k = decode_sv (dec); 912 decode_he (dec, hv);
544 SV *v = decode_sv (dec);
545
546 hv_store_ent (hv, k, v, 0);
547 } 913 }
548 } 914 }
549 else 915 else
550 { 916 {
551 int len = decode_uint (dec); 917 UV pairs = decode_uint (dec);
552 918
919 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
920
553 while (len--) 921 while (pairs--)
554 { 922 decode_he (dec, hv);
555 SV *k = decode_sv (dec);
556 SV *v = decode_sv (dec);
557
558 hv_store_ent (hv, k, v, 0);
559 }
560 } 923 }
561 924
562 DEC_DEC_DEPTH; 925 DEC_DEC_DEPTH;
563 return newRV_noinc ((SV *)hv); 926 return newRV_noinc ((SV *)hv);
564
565#if 0
566 SV *sv;
567 HV *hv = newHV ();
568
569 DEC_INC_DEPTH;
570 decode_ws (dec);
571
572 for (;;)
573 {
574 // heuristic: assume that
575 // a) decode_str + hv_store_ent are abysmally slow.
576 // b) most hash keys are short, simple ascii text.
577 // => try to "fast-match" such strings to avoid
578 // the overhead of decode_str + hv_store_ent.
579 {
580 SV *value;
581 char *p = dec->cur;
582 char *e = p + 24; // only try up to 24 bytes
583
584 for (;;)
585 {
586 // the >= 0x80 is false on most architectures
587 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
588 {
589 // slow path, back up and use decode_str
590 SV *key = decode_str (dec);
591 if (!key)
592 goto fail;
593
594 decode_ws (dec); EXPECT_CH (':');
595
596 decode_ws (dec);
597 value = decode_sv (dec);
598 if (!value)
599 {
600 SvREFCNT_dec (key);
601 goto fail;
602 }
603
604 hv_store_ent (hv, key, value, 0);
605 SvREFCNT_dec (key);
606
607 break;
608 }
609 else if (*p == '"')
610 {
611 // fast path, got a simple key
612 char *key = dec->cur;
613 int len = p - key;
614 dec->cur = p + 1;
615
616 decode_ws (dec); EXPECT_CH (':');
617
618 decode_ws (dec);
619 value = decode_sv (dec);
620 if (!value)
621 goto fail;
622
623 hv_store (hv, key, len, value, 0);
624
625 break;
626 }
627
628 ++p;
629 }
630 }
631
632 decode_ws (dec);
633
634 if (*dec->cur == '}')
635 {
636 ++dec->cur;
637 break;
638 }
639
640 if (*dec->cur != ',')
641 ERR (", or } expected while parsing object/hash");
642
643 ++dec->cur;
644
645 decode_ws (dec);
646
647 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
648 {
649 ++dec->cur;
650 break;
651 }
652 }
653
654 DEC_DEC_DEPTH;
655 sv = newRV_noinc ((SV *)hv);
656
657 // check filter callbacks
658 if (dec->cbor.flags & F_HOOK)
659 {
660 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
661 {
662 HE *cb, *he;
663
664 hv_iterinit (hv);
665 he = hv_iternext (hv);
666 hv_iterinit (hv);
667
668 // the next line creates a mortal sv each time its called.
669 // might want to optimise this for common cases.
670 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
671
672 if (cb)
673 {
674 dSP;
675 int count;
676
677 ENTER; SAVETMPS; PUSHMARK (SP);
678 XPUSHs (HeVAL (he));
679 sv_2mortal (sv);
680
681 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
682
683 if (count == 1)
684 {
685 sv = newSVsv (POPs);
686 FREETMPS; LEAVE;
687 return sv;
688 }
689
690 SvREFCNT_inc (sv);
691 FREETMPS; LEAVE;
692 }
693 }
694
695 if (dec->cbor.cb_object)
696 {
697 dSP;
698 int count;
699
700 ENTER; SAVETMPS; PUSHMARK (SP);
701 XPUSHs (sv_2mortal (sv));
702
703 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
704
705 if (count == 1)
706 {
707 sv = newSVsv (POPs);
708 FREETMPS; LEAVE;
709 return sv;
710 }
711
712 SvREFCNT_inc (sv);
713 FREETMPS; LEAVE;
714 }
715 }
716
717 return sv;
718#endif
719 927
720fail: 928fail:
721 SvREFCNT_dec (hv); 929 SvREFCNT_dec (hv);
722 DEC_DEC_DEPTH; 930 DEC_DEC_DEPTH;
723 return &PL_sv_undef; 931 return &PL_sv_undef;
724} 932}
725 933
726static SV * 934static SV *
727decode_str (dec_t *dec, int utf8) 935decode_str (dec_t *dec, int utf8)
728{ 936{
729 SV *sv; 937 SV *sv = 0;
730 938
731 if ((*dec->cur & 31) == 31) 939 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 940 {
941 // indefinite length strings
733 ++dec->cur; 942 ++dec->cur;
734 943
944 U8 major = *dec->cur & MAJOR_MISC;
945
735 sv = newSVpvn ("", 0); 946 sv = newSVpvn ("", 0);
736 947
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 948 for (;;)
739 { 949 {
740 WANT (1); 950 WANT (1);
741 951
742 if (*dec->cur == (0xe0 | 31)) 952 if ((*dec->cur - major) > LENGTH_EXT8)
953 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 954 {
744 ++dec->cur; 955 ++dec->cur;
745 break; 956 break;
746 } 957 }
958 else
959 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 960
748 SV *sv2 = decode_sv (dec); 961 STRLEN len = decode_uint (dec);
749 sv_catsv (sv, sv2); 962
963 WANT (len);
964 sv_catpvn (sv, dec->cur, len);
965 dec->cur += len;
750 } 966 }
751 } 967 }
752 else 968 else
753 { 969 {
754 STRLEN len = decode_uint (dec); 970 STRLEN len = decode_uint (dec);
755 971
756 WANT (len); 972 WANT (len);
757 sv = newSVpvn (dec->cur, len); 973 sv = newSVpvn (dec->cur, len);
758 dec->cur += len; 974 dec->cur += len;
975
976 if (ecb_expect_false (dec->stringref)
977 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
978 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
759 } 979 }
760 980
761 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
762 SvUTF8_on (sv); 987 SvUTF8_on (sv);
988 }
763 989
764 return sv; 990 return sv;
765 991
766fail: 992fail:
993 SvREFCNT_dec (sv);
767 return &PL_sv_undef; 994 return &PL_sv_undef;
768} 995}
769 996
770static SV * 997static SV *
771decode_tagged (dec_t *dec) 998decode_tagged (dec_t *dec)
772{ 999{
1000 SV *sv = 0;
773 UV tag = decode_uint (dec); 1001 UV tag = decode_uint (dec);
1002
1003 WANT (1);
1004
1005 switch (tag)
1006 {
1007 case CBOR_TAG_MAGIC:
774 SV *sv = decode_sv (dec); 1008 sv = decode_sv (dec);
1009 break;
775 1010
776 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1011 case CBOR_TAG_INDIRECTION:
1012 sv = newRV_noinc (decode_sv (dec));
1013 break;
1014
1015 case CBOR_TAG_STRINGREF_NAMESPACE:
1016 {
1017 // do nmot use SAVETMPS/FREETMPS, as these will
1018 // erase mortalised caches, e.g. "shareable"
1019 ENTER;
1020
1021 SAVESPTR (dec->stringref);
1022 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1023
1024 sv = decode_sv (dec);
1025
1026 LEAVE;
1027 }
1028 break;
1029
1030 case CBOR_TAG_STRINGREF:
1031 {
1032 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1033 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1034
1035 UV idx = decode_uint (dec);
1036
1037 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
1038 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1039
1040 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1041 }
1042 break;
1043
1044 case CBOR_TAG_VALUE_SHAREABLE:
1045 {
1046 if (ecb_expect_false (!dec->shareable))
1047 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1048
1049 if (dec->cbor.flags & F_ALLOW_CYCLES)
1050 {
1051 sv = newSV (0);
1052 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1053
1054 SV *osv = decode_sv (dec);
1055 sv_setsv (sv, 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 }
1065 }
1066 break;
1067
1068 case CBOR_TAG_VALUE_SHAREDREF:
1069 {
1070 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1071 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1072
1073 UV idx = decode_uint (dec);
1074
1075 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1076 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1077
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");
1082 }
1083 break;
1084
1085 case CBOR_TAG_PERL_OBJECT:
1086 {
1087 if (dec->cbor.flags & F_FORBID_OBJECTS)
1088 goto filter;
1089
1090 sv = decode_sv (dec);
1091
1092 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1093 ERR ("corrupted CBOR data (non-array perl object)");
1094
1095 AV *av = (AV *)SvRV (sv);
1096 int len = av_len (av) + 1;
1097 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1098
1099 if (!stash)
1100 ERR ("cannot decode perl-object (package does not exist)");
1101
1102 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1103
1104 if (!method)
1105 ERR ("cannot decode perl-object (package does not have a THAW method)");
1106
1107 dSP;
1108
1109 ENTER; SAVETMPS;
1110 PUSHMARK (SP);
1111 EXTEND (SP, len + 1);
1112 // we re-bless the reference to get overload and other niceties right
1113 PUSHs (*av_fetch (av, 0, 1));
1114 PUSHs (sv_cbor);
1115
1116 int i;
1117
1118 for (i = 1; i < len; ++i)
1119 PUSHs (*av_fetch (av, i, 1));
1120
1121 PUTBACK;
1122 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1123 SPAGAIN;
1124
1125 if (SvTRUE (ERRSV))
1126 {
1127 FREETMPS; LEAVE;
1128 ERR_ERRSV;
1129 }
1130
1131 SvREFCNT_dec (sv);
1132 sv = SvREFCNT_inc (POPs);
1133
1134 PUTBACK;
1135
1136 FREETMPS; LEAVE;
1137 }
1138 break;
1139
1140 default:
1141 filter:
1142 {
1143 SV *tag_sv = newSVuv (tag);
1144
1145 sv = decode_sv (dec);
1146
1147 dSP;
1148 ENTER; SAVETMPS;
1149 SAVESTACK_POS ();
1150 PUSHMARK (SP);
1151 EXTEND (SP, 2);
1152 PUSHs (tag_sv);
1153 PUSHs (sv);
1154
1155 PUTBACK;
1156 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1157 SPAGAIN;
1158
1159 if (SvTRUE (ERRSV))
1160 {
1161 SvREFCNT_dec (tag_sv);
1162 FREETMPS; LEAVE;
1163 ERR_ERRSV;
1164 }
1165
1166 if (count)
1167 {
1168 SvREFCNT_dec (tag_sv);
1169 SvREFCNT_dec (sv);
1170 sv = SvREFCNT_inc (POPs);
1171 }
1172 else
1173 {
1174 AV *av = newAV ();
1175 av_push (av, tag_sv);
1176 av_push (av, sv);
1177
1178 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1179 ? cbor_tagged_stash
1180 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1181 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1182 }
1183
1184 PUTBACK;
1185
1186 FREETMPS; LEAVE;
1187 }
1188 break;
1189 }
1190
777 return sv; 1191 return sv;
778 1192
779 AV *av = newAV (); 1193fail:
780 av_push (av, newSVuv (tag)); 1194 SvREFCNT_dec (sv);
781 av_push (av, sv); 1195 return &PL_sv_undef;
782 return newRV_noinc ((SV *)av);
783} 1196}
784 1197
785static SV * 1198static SV *
786decode_sv (dec_t *dec) 1199decode_sv (dec_t *dec)
787{ 1200{
788 WANT (1); 1201 WANT (1);
789 1202
790 switch (*dec->cur >> 5) 1203 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1204 {
792 case 0: // unsigned int 1205 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1206 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1207 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1208 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1209 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1210 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1211 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1212
800 return decode_str (dec, 1); 1213 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 return decode_tagged (dec);
807 case 7: // misc
808 switch (*dec->cur++ & 31) 1214 switch (*dec->cur++ & MINOR_MASK)
809 { 1215 {
810 case 20: 1216 case SIMPLE_FALSE:
811#if CBOR_SLOW 1217#if CBOR_SLOW
812 cbor_false = get_bool ("CBOR::XS::false"); 1218 types_false = get_bool ("Types::Serialiser::false");
813#endif 1219#endif
814 return newSVsv (cbor_false); 1220 return newSVsv (types_false);
815 case 21: 1221 case SIMPLE_TRUE:
816#if CBOR_SLOW 1222#if CBOR_SLOW
817 cbor_true = get_bool ("CBOR::XS::true"); 1223 types_true = get_bool ("Types::Serialiser::true");
818#endif 1224#endif
819 return newSVsv (cbor_true); 1225 return newSVsv (types_true);
820 case 22: 1226 case SIMPLE_NULL:
821 return newSVsv (&PL_sv_undef); 1227 return newSVsv (&PL_sv_undef);
1228 case SIMPLE_UNDEF:
1229#if CBOR_SLOW
1230 types_error = get_bool ("Types::Serialiser::error");
1231#endif
1232 return newSVsv (types_error);
822 1233
823 case 25: 1234 case MISC_FLOAT16:
824 { 1235 {
825 WANT (2); 1236 WANT (2);
826 1237
827 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1238 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
828 dec->cur += 2; 1239 dec->cur += 2;
829 1240
830 return newSVnv (ecb_binary16_to_float (fp)); 1241 return newSVnv (ecb_binary16_to_float (fp));
831 } 1242 }
832 1243
833 case 26: 1244 case MISC_FLOAT32:
834 { 1245 {
835 uint32_t fp; 1246 uint32_t fp;
836 WANT (4); 1247 WANT (4);
837 memcpy (&fp, dec->cur, 4); 1248 memcpy (&fp, dec->cur, 4);
838 dec->cur += 4; 1249 dec->cur += 4;
841 fp = ecb_bswap32 (fp); 1252 fp = ecb_bswap32 (fp);
842 1253
843 return newSVnv (ecb_binary32_to_float (fp)); 1254 return newSVnv (ecb_binary32_to_float (fp));
844 } 1255 }
845 1256
846 case 27: 1257 case MISC_FLOAT64:
847 { 1258 {
848 uint64_t fp; 1259 uint64_t fp;
849 WANT (8); 1260 WANT (8);
850 memcpy (&fp, dec->cur, 8); 1261 memcpy (&fp, dec->cur, 8);
851 dec->cur += 8; 1262 dec->cur += 8;
854 fp = ecb_bswap64 (fp); 1265 fp = ecb_bswap64 (fp);
855 1266
856 return newSVnv (ecb_binary64_to_double (fp)); 1267 return newSVnv (ecb_binary64_to_double (fp));
857 } 1268 }
858 1269
859 // 0..19 unassigned 1270 // 0..19 unassigned simple
860 // 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
861 default: 1274 default:
862 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1275 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
863 } 1276 }
864 1277
865 break; 1278 break;
866 } 1279 }
867#if 0
868 switch (*dec->cur)
869 {
870 //case '"': ++dec->cur; return decode_str (dec);
871 case '[': ++dec->cur; return decode_av (dec);
872 case '{': ++dec->cur; return decode_hv (dec);
873
874 case '-':
875 case '0': case '1': case '2': case '3': case '4':
876 case '5': case '6': case '7': case '8': case '9':
877 //TODO return decode_num (dec);
878
879 case 't':
880 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
881 {
882 dec->cur += 4;
883#if CBOR_SLOW
884 cbor_true = get_bool ("CBOR::XS::true");
885#endif
886 return newSVsv (cbor_true);
887 }
888 else
889 ERR ("'true' expected");
890
891 break;
892
893 case 'f':
894 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
895 {
896 dec->cur += 5;
897#if CBOR_SLOW
898 cbor_false = get_bool ("CBOR::XS::false");
899#endif
900 return newSVsv (cbor_false);
901 }
902 else
903 ERR ("'false' expected");
904
905 break;
906
907 case 'n':
908 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
909 {
910 dec->cur += 4;
911 return newSVsv (&PL_sv_undef);
912 }
913 else
914 ERR ("'null' expected");
915
916 break;
917
918 default:
919 ERR ("malformed CBOR string, neither array, object, number, string or atom");
920 break;
921 }
922#endif
923 1280
924fail: 1281fail:
925 return &PL_sv_undef; 1282 return &PL_sv_undef;
926} 1283}
927 1284
928static SV * 1285static SV *
929decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1286decode_cbor (SV *string, CBOR *cbor, char **offset_return)
930{ 1287{
931 dec_t dec; 1288 dec_t dec = { 0 };
932 SV *sv; 1289 SV *sv;
1290 STRLEN len;
1291 char *data = SvPVbyte (string, len);
933 1292
934 /* work around bugs in 5.10 where manipulating magic values
935 * makes perl ignore the magic in subsequent accesses.
936 * also make a copy of non-PV values, to get them into a clean
937 * state (SvPV should do that, but it's buggy, see below).
938 */
939 /*SvGETMAGIC (string);*/
940 if (SvMAGICAL (string) || !SvPOK (string))
941 string = sv_2mortal (newSVsv (string));
942
943 SvUPGRADE (string, SVt_PV);
944
945 /* work around a bug in perl 5.10, which causes SvCUR to fail an
946 * assertion with -DDEBUGGING, although SvCUR is documented to
947 * return the xpv_cur field which certainly exists after upgrading.
948 * according to nicholas clark, calling SvPOK fixes this.
949 * But it doesn't fix it, so try another workaround, call SvPV_nolen
950 * and hope for the best.
951 * Damnit, SvPV_nolen still trips over yet another assertion. This
952 * assertion business is seriously broken, try yet another workaround
953 * for the broken -DDEBUGGING.
954 */
955 {
956#ifdef DEBUGGING
957 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
958#else
959 STRLEN offset = SvCUR (string);
960#endif
961
962 if (offset > cbor->max_size && cbor->max_size) 1293 if (len > cbor->max_size && cbor->max_size)
963 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1294 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
964 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1295 (unsigned long)len, (unsigned long)cbor->max_size);
965 }
966
967 sv_utf8_downgrade (string, 0);
968 1296
969 dec.cbor = *cbor; 1297 dec.cbor = *cbor;
970 dec.cur = (U8 *)SvPVX (string); 1298 dec.cur = (U8 *)data;
971 dec.end = (U8 *)SvEND (string); 1299 dec.end = (U8 *)data + len;
972 dec.err = 0;
973 dec.depth = 0;
974
975 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
976 ;//TODO dec.cbor.flags |= F_HOOK;
977 1300
978 sv = decode_sv (&dec); 1301 sv = decode_sv (&dec);
979 1302
980 if (offset_return) 1303 if (offset_return)
981 *offset_return = dec.cur; 1304 *offset_return = dec.cur;
984 if (dec.cur != dec.end && !dec.err) 1307 if (dec.cur != dec.end && !dec.err)
985 dec.err = "garbage after CBOR object"; 1308 dec.err = "garbage after CBOR object";
986 1309
987 if (dec.err) 1310 if (dec.err)
988 { 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
989 SvREFCNT_dec (sv); 1323 SvREFCNT_dec (sv);
1324
1325 if (dec.err_sv)
1326 sv_2mortal (dec.err_sv);
1327
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (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);
991 } 1329 }
992 1330
993 sv = sv_2mortal (sv); 1331 sv = sv_2mortal (sv);
994 1332
995 return sv; 1333 return sv;
996} 1334}
997 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
998///////////////////////////////////////////////////////////////////////////// 1460/////////////////////////////////////////////////////////////////////////////
999// XS interface functions 1461// XS interface functions
1000 1462
1001MODULE = CBOR::XS PACKAGE = CBOR::XS 1463MODULE = CBOR::XS PACKAGE = CBOR::XS
1002 1464
1003BOOT: 1465BOOT:
1004{ 1466{
1005 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1467 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1006 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1468 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1007 1469
1008 cbor_true = get_bool ("CBOR::XS::true"); 1470 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1009 cbor_false = get_bool ("CBOR::XS::false"); 1471 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1472
1473 types_true = get_bool ("Types::Serialiser::true" );
1474 types_false = get_bool ("Types::Serialiser::false");
1475 types_error = get_bool ("Types::Serialiser::error");
1476
1477 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1478
1479 sv_cbor = newSVpv ("CBOR", 0);
1480 SvREADONLY_on (sv_cbor);
1481
1482 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1010} 1483}
1011 1484
1012PROTOTYPES: DISABLE 1485PROTOTYPES: DISABLE
1013 1486
1014void CLONE (...) 1487void CLONE (...)
1015 CODE: 1488 CODE:
1016 cbor_stash = 0; 1489 cbor_stash = 0;
1490 cbor_tagged_stash = 0;
1491 types_error_stash = 0;
1017 cbor_boolean_stash = 0; 1492 types_boolean_stash = 0;
1018 1493
1019void new (char *klass) 1494void new (char *klass)
1020 PPCODE: 1495 PPCODE:
1021{ 1496{
1022 SV *pv = NEWSV (0, sizeof (CBOR)); 1497 SV *pv = NEWSV (0, sizeof (CBOR));
1030 1505
1031void shrink (CBOR *self, int enable = 1) 1506void shrink (CBOR *self, int enable = 1)
1032 ALIAS: 1507 ALIAS:
1033 shrink = F_SHRINK 1508 shrink = F_SHRINK
1034 allow_unknown = F_ALLOW_UNKNOWN 1509 allow_unknown = F_ALLOW_UNKNOWN
1510 allow_sharing = F_ALLOW_SHARING
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
1035 PPCODE: 1517 PPCODE:
1036{ 1518{
1037 if (enable) 1519 if (enable)
1038 self->flags |= ix; 1520 self->flags |= ix;
1039 else 1521 else
1044 1526
1045void get_shrink (CBOR *self) 1527void get_shrink (CBOR *self)
1046 ALIAS: 1528 ALIAS:
1047 get_shrink = F_SHRINK 1529 get_shrink = F_SHRINK
1048 get_allow_unknown = F_ALLOW_UNKNOWN 1530 get_allow_unknown = F_ALLOW_UNKNOWN
1531 get_allow_sharing = F_ALLOW_SHARING
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
1049 PPCODE: 1538 PPCODE:
1050 XPUSHs (boolSV (self->flags & ix)); 1539 XPUSHs (boolSV (self->flags & ix));
1051 1540
1052void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1541void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1053 PPCODE: 1542 PPCODE:
1069 CODE: 1558 CODE:
1070 RETVAL = self->max_size; 1559 RETVAL = self->max_size;
1071 OUTPUT: 1560 OUTPUT:
1072 RETVAL 1561 RETVAL
1073 1562
1074#if 0 //TODO 1563void filter (CBOR *self, SV *filter = 0)
1075
1076void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1564 PPCODE:
1078{
1079 SvREFCNT_dec (self->cb_object); 1565 SvREFCNT_dec (self->filter);
1080 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1566 self->filter = filter ? newSVsv (filter) : filter;
1081
1082 XPUSHs (ST (0)); 1567 XPUSHs (ST (0));
1083}
1084 1568
1085void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1569SV *get_filter (CBOR *self)
1086 PPCODE: 1570 CODE:
1087{ 1571 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1088 if (!self->cb_sk_object) 1572 OUTPUT:
1089 self->cb_sk_object = newHV (); 1573 RETVAL
1090
1091 if (SvOK (cb))
1092 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1093 else
1094 {
1095 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1096
1097 if (!HvKEYS (self->cb_sk_object))
1098 {
1099 SvREFCNT_dec (self->cb_sk_object);
1100 self->cb_sk_object = 0;
1101 }
1102 }
1103
1104 XPUSHs (ST (0));
1105}
1106
1107#endif
1108 1574
1109void encode (CBOR *self, SV *scalar) 1575void encode (CBOR *self, SV *scalar)
1110 PPCODE: 1576 PPCODE:
1111 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1577 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1112 XPUSHs (scalar); 1578 XPUSHs (scalar);
1125 EXTEND (SP, 2); 1591 EXTEND (SP, 2);
1126 PUSHs (sv); 1592 PUSHs (sv);
1127 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1593 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1128} 1594}
1129 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
1130void DESTROY (CBOR *self) 1648void DESTROY (CBOR *self)
1131 CODE: 1649 PPCODE:
1132 SvREFCNT_dec (self->cb_sk_object); 1650 cbor_free (self);
1133 SvREFCNT_dec (self->cb_object);
1134 1651
1135PROTOTYPES: ENABLE 1652PROTOTYPES: ENABLE
1136 1653
1137void encode_cbor (SV *scalar) 1654void encode_cbor (SV *scalar)
1655 ALIAS:
1656 encode_cbor = 0
1657 encode_cbor_sharing = F_ALLOW_SHARING
1138 PPCODE: 1658 PPCODE:
1139{ 1659{
1140 CBOR cbor; 1660 CBOR cbor;
1141 cbor_init (&cbor); 1661 cbor_init (&cbor);
1662 cbor.flags |= ix;
1142 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1663 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1143 XPUSHs (scalar); 1664 XPUSHs (scalar);
1144} 1665}
1145 1666
1146void decode_cbor (SV *cborstr) 1667void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines