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 1.9 by root, Sun Oct 27 10:17:12 2013 UTC vs.
Revision 1.61 by root, Sat Nov 26 02:10:19 2016 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
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};
13 66
14// known tags 67// known tags
15enum cbor_tag 68enum cbor_tag
16{ 69{
17 // inofficial extensions (pending iana registration) 70 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 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
20 78
21 // rfc7049 79 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 80 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 81 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 82 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 83 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 84 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 85 CBOR_TAG_BIGFLOAT = 5, // array
28 86
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 87 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 88 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 89 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 90 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 91
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 92 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 93 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 94 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 95 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 96 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 97
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 98 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 99};
42 100
43#define F_SHRINK 0x00000200UL 101#define F_SHRINK 0x00000001UL
44#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
45 110
46#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
47 112
48#define SB do { 113#define SB do {
49#define SE } while (0) 114#define SE } while (0)
60#else 125#else
61# define CBOR_SLOW 0 126# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 127# define CBOR_STASH cbor_stash
63#endif 128#endif
64 129
65static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 130static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *cbor_true, *cbor_false; 131static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 132
68typedef struct { 133typedef struct {
69 U32 flags; 134 U32 flags;
70 U32 max_depth; 135 U32 max_depth;
71 STRLEN max_size; 136 STRLEN max_size;
137 SV *filter;
138
139 // for the incremental parser
140 STRLEN incr_pos; // the current offset into the text
141 STRLEN incr_need; // minimum bytes needed to decode
142 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 143} CBOR;
73 144
74ecb_inline void 145ecb_inline void
75cbor_init (CBOR *cbor) 146cbor_init (CBOR *cbor)
76{ 147{
77 Zero (cbor, 1, CBOR); 148 Zero (cbor, 1, CBOR);
78 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);
79} 157}
80 158
81///////////////////////////////////////////////////////////////////////////// 159/////////////////////////////////////////////////////////////////////////////
82// utility functions 160// utility functions
83 161
105 SvPV_renew (sv, SvCUR (sv) + 1); 183 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 184#endif
107 } 185 }
108} 186}
109 187
110///////////////////////////////////////////////////////////////////////////// 188// minimum length of a string to be registered for stringref
111// fp hell 189ecb_inline int
190minimum_string_length (UV idx)
191{
112 192
113//TODO 193 return idx <= 23 ? 3
194 : idx <= 0xffU ? 4
195 : idx <= 0xffffU ? 5
196 : idx <= 0xffffffffU ? 7
197 : 11;
198}
114 199
115///////////////////////////////////////////////////////////////////////////// 200/////////////////////////////////////////////////////////////////////////////
116// encoder 201// encoder
117 202
118// structure used for encoding CBOR 203// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 206 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 207 char *end; // SvEND (sv)
123 SV *sv; // result scalar 208 SV *sv; // result scalar
124 CBOR cbor; 209 CBOR cbor;
125 U32 depth; // recursion level 210 U32 depth; // recursion level
211 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
212 UV stringref_idx;
213 HV *shareable; // ptr => index, or 0
214 UV shareable_idx;
126} enc_t; 215} enc_t;
127 216
128ecb_inline void 217ecb_inline void
129need (enc_t *enc, STRLEN len) 218need (enc_t *enc, STRLEN len)
130{ 219{
131 if (ecb_expect_false (enc->cur + len >= enc->end)) 220 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
132 { 221 {
133 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
134 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
135 enc->cur = SvPVX (enc->sv) + cur; 224 enc->cur = SvPVX (enc->sv) + cur;
136 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
147static void 236static void
148encode_uint (enc_t *enc, int major, UV len) 237encode_uint (enc_t *enc, int major, UV len)
149{ 238{
150 need (enc, 9); 239 need (enc, 9);
151 240
152 if (len < 24) 241 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 242 *enc->cur++ = major | len;
154 else if (len <= 0xff) 243 else if (ecb_expect_true (len <= 0xffU))
155 { 244 {
156 *enc->cur++ = major | 24; 245 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 246 *enc->cur++ = len;
158 } 247 }
159 else if (len <= 0xffff) 248 else if (len <= 0xffffU)
160 { 249 {
161 *enc->cur++ = major | 25; 250 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 252 *enc->cur++ = len;
164 } 253 }
165 else if (len <= 0xffffffff) 254 else if (len <= 0xffffffffU)
166 { 255 {
167 *enc->cur++ = major | 26; 256 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 257 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 258 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 259 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 260 *enc->cur++ = len;
172 } 261 }
173 else 262 else
174 { 263 {
175 *enc->cur++ = major | 27; 264 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 265 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 266 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 267 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 268 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 269 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 271 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 272 *enc->cur++ = len;
184 } 273 }
185} 274}
186 275
187static void 276ecb_inline void
277encode_tag (enc_t *enc, UV tag)
278{
279 encode_uint (enc, MAJOR_TAG, tag);
280}
281
282// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
283ecb_noinline static void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 284encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 285{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 286 STRLEN ulen = len;
287 U8 *p, *pend = (U8 *)str + len;
288
289 for (p = (U8 *)str; p < pend; ++p)
290 ulen += *p >> 7; // count set high bits
291
292 encode_uint (enc, MAJOR_TEXT, ulen);
293
294 need (enc, ulen);
295 for (p = (U8 *)str; p < pend; ++p)
296 if (*p < 0x80)
297 *enc->cur++ = *p;
298 else
299 {
300 *enc->cur++ = 0xc0 + (*p >> 6);
301 *enc->cur++ = 0x80 + (*p & 63);
302 }
303}
304
305ecb_inline void
306encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
307{
308 if (ecb_expect_false (upgrade_utf8))
309 if (!utf8)
310 {
311 encode_str_utf8 (enc, utf8, str, len);
312 return;
313 }
314
315 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 316 need (enc, len);
192 memcpy (enc->cur, str, len); 317 memcpy (enc->cur, str, len);
193 enc->cur += len; 318 enc->cur += len;
194} 319}
195 320
321ecb_inline void
322encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
323{
324 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
325 {
326 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
327
328 if (SvOK (*svp))
329 {
330 // already registered, use stringref
331 encode_tag (enc, CBOR_TAG_STRINGREF);
332 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
333 return;
334 }
335 else if (len >= minimum_string_length (enc->stringref_idx))
336 {
337 // register only
338 sv_setuv (*svp, enc->stringref_idx);
339 ++enc->stringref_idx;
340 }
341 }
342
343 encode_str (enc, upgrade_utf8, utf8, str, len);
344}
345
196static void encode_sv (enc_t *enc, SV *sv); 346static void encode_sv (enc_t *enc, SV *sv);
197 347
198static void 348static void
199encode_av (enc_t *enc, AV *av) 349encode_av (enc_t *enc, AV *av)
200{ 350{
203 if (enc->depth >= enc->cbor.max_depth) 353 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 354 croak (ERR_NESTING_EXCEEDED);
205 355
206 ++enc->depth; 356 ++enc->depth;
207 357
208 encode_uint (enc, 0x80, len + 1); 358 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 359
360 if (SvMAGICAL (av))
210 for (i = 0; i <= len; ++i) 361 for (i = 0; i <= len; ++i)
211 { 362 {
212 SV **svp = av_fetch (av, i, 0); 363 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 364 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 365 }
366 else
367 for (i = 0; i <= len; ++i)
368 {
369 SV *sv = AvARRAY (av)[i];
370 encode_sv (enc, sv ? sv : &PL_sv_undef);
371 }
215 372
216 --enc->depth; 373 --enc->depth;
217} 374}
218 375
219static void 376static void
228 385
229 int pairs = hv_iterinit (hv); 386 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 387 int mg = SvMAGICAL (hv);
231 388
232 if (mg) 389 if (mg)
233 encode_ch (enc, 0xa0 | 31); 390 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 391 else
235 encode_uint (enc, 0xa0, pairs); 392 encode_uint (enc, MAJOR_MAP, pairs);
236 393
237 while ((he = hv_iternext (hv))) 394 while ((he = hv_iternext (hv)))
238 { 395 {
239 if (HeKLEN (he) == HEf_SVKEY) 396 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 397 encode_sv (enc, HeSVKEY (he));
241 else 398 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 399 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 400
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 401 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 402 }
246 403
247 if (mg) 404 if (mg)
248 encode_ch (enc, 0xe0 | 31); 405 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 406
250 --enc->depth; 407 --enc->depth;
251} 408}
252 409
253// encode objects, arrays and special \0=false and \1=true values. 410// encode objects, arrays and special \0=false and \1=true values.
254static void 411static void
255encode_rv (enc_t *enc, SV *sv) 412encode_rv (enc_t *enc, SV *sv)
256{ 413{
257 svtype svt;
258
259 SvGETMAGIC (sv); 414 SvGETMAGIC (sv);
415
260 svt = SvTYPE (sv); 416 svtype svt = SvTYPE (sv);
261 417
262 if (ecb_expect_false (SvOBJECT (sv))) 418 if (ecb_expect_false (SvOBJECT (sv)))
263 { 419 {
264 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 420 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 421 ? types_boolean_stash
266 : gv_stashpv ("CBOR::XS::Boolean", 1); 422 : gv_stashpv ("Types::Serialiser::Boolean", 1);
423 HV *error_stash = !CBOR_SLOW || types_error_stash
424 ? types_error_stash
425 : gv_stashpv ("Types::Serialiser::Error", 1);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 426 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 427 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 428 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 429
430 HV *stash = SvSTASH (sv);
431
271 if (SvSTASH (sv) == boolean_stash) 432 if (stash == boolean_stash)
272 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 433 {
434 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
435 return;
436 }
437 else if (stash == error_stash)
438 {
439 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
440 return;
441 }
273 else if (SvSTASH (sv) == tagged_stash) 442 else if (stash == tagged_stash)
274 { 443 {
275 if (svt != SVt_PVAV) 444 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 445 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 446
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 447 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 448 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
449
450 return;
451 }
452 }
453
454 if (ecb_expect_false (SvREFCNT (sv) > 1)
455 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
456 {
457 if (!enc->shareable)
458 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
459
460 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
461
462 if (SvOK (*svp))
463 {
464 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
465 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
466 return;
280 } 467 }
281 else 468 else
282 { 469 {
470 sv_setuv (*svp, enc->shareable_idx);
471 ++enc->shareable_idx;
472 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
473 }
474 }
475
476 if (ecb_expect_false (SvOBJECT (sv)))
477 {
478 HV *stash = SvSTASH (sv);
479 GV *method;
480
481 if (enc->cbor.flags & F_FORBID_OBJECTS)
482 croak ("encountered object '%s', but forbid_objects is enabled",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
484 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
485 {
486 dSP;
487
488 ENTER; SAVETMPS;
489 PUSHMARK (SP);
283 // we re-bless the reference to get overload and other niceties right 490 // we re-bless the reference to get overload and other niceties right
284 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
285
286 if (to_cbor)
287 {
288 dSP;
289
290 ENTER; SAVETMPS; PUSHMARK (SP);
291 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 491 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
292 492
293 // calling with G_SCALAR ensures that we always get a 1 return value
294 PUTBACK; 493 PUTBACK;
494 // G_SCALAR ensures that return value is 1
295 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 495 call_sv ((SV *)GvCV (method), G_SCALAR);
296 SPAGAIN; 496 SPAGAIN;
297 497
298 // catch this surprisingly common error 498 // catch this surprisingly common error
299 if (SvROK (TOPs) && SvRV (TOPs) == sv) 499 if (SvROK (TOPs) && SvRV (TOPs) == sv)
300 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 500 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
301 501
302 sv = POPs; 502 encode_sv (enc, POPs);
503
303 PUTBACK; 504 PUTBACK;
304 505
305 encode_sv (enc, sv);
306
307 FREETMPS; LEAVE; 506 FREETMPS; LEAVE;
308 } 507 }
508 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
509 {
510 dSP;
511
512 ENTER; SAVETMPS;
513 SAVESTACK_POS ();
514 PUSHMARK (SP);
515 EXTEND (SP, 2);
516 // we re-bless the reference to get overload and other niceties right
517 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
518 PUSHs (sv_cbor);
519
520 PUTBACK;
521 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
522 SPAGAIN;
523
524 // catch this surprisingly common error
525 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
526 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
527
528 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
529 encode_uint (enc, MAJOR_ARRAY, count + 1);
530 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
531
532 while (count)
533 encode_sv (enc, SP[1 - count--]);
534
535 PUTBACK;
536
537 FREETMPS; LEAVE;
538 }
309 else 539 else
310 croak ("encountered object '%s', but no TO_CBOR method available on it", 540 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 541 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 542 }
314 else if (svt == SVt_PVHV) 543 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 544 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 545 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 546 encode_av (enc, (AV *)sv);
318 else if (svt < SVt_PVAV)
319 {
320 STRLEN len = 0;
321 char *pv = svt ? SvPV (sv, len) : 0;
322
323 if (len == 1 && *pv == '1')
324 encode_ch (enc, 0xe0 | 21);
325 else if (len == 1 && *pv == '0')
326 encode_ch (enc, 0xe0 | 20);
327 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
328 encode_ch (enc, 0xe0 | 23);
329 else
330 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
331 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
332 }
333 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
334 encode_ch (enc, 0xe0 | 23);
335 else 547 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 548 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 549 encode_tag (enc, CBOR_TAG_INDIRECTION);
550 encode_sv (enc, sv);
551 }
338} 552}
339 553
340static void 554static void
341encode_nv (enc_t *enc, SV *sv) 555encode_nv (enc_t *enc, SV *sv)
342{ 556{
343 double nv = SvNVX (sv); 557 double nv = SvNVX (sv);
344 558
345 need (enc, 9); 559 need (enc, 9);
346 560
347 if (ecb_expect_false (nv == (U32)nv)) 561 if (ecb_expect_false (nv == (NV)(U32)nv))
348 encode_uint (enc, 0x00, (U32)nv); 562 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
349 //TODO: maybe I32? 563 //TODO: maybe I32?
350 else if (ecb_expect_false (nv == (float)nv)) 564 else if (ecb_expect_false (nv == (float)nv))
351 { 565 {
352 uint32_t fp = ecb_float_to_binary32 (nv); 566 uint32_t fp = ecb_float_to_binary32 (nv);
353 567
354 *enc->cur++ = 0xe0 | 26; 568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
355 569
356 if (!ecb_big_endian ()) 570 if (!ecb_big_endian ())
357 fp = ecb_bswap32 (fp); 571 fp = ecb_bswap32 (fp);
358 572
359 memcpy (enc->cur, &fp, 4); 573 memcpy (enc->cur, &fp, 4);
361 } 575 }
362 else 576 else
363 { 577 {
364 uint64_t fp = ecb_double_to_binary64 (nv); 578 uint64_t fp = ecb_double_to_binary64 (nv);
365 579
366 *enc->cur++ = 0xe0 | 27; 580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
367 581
368 if (!ecb_big_endian ()) 582 if (!ecb_big_endian ())
369 fp = ecb_bswap64 (fp); 583 fp = ecb_bswap64 (fp);
370 584
371 memcpy (enc->cur, &fp, 8); 585 memcpy (enc->cur, &fp, 8);
380 594
381 if (SvPOKp (sv)) 595 if (SvPOKp (sv))
382 { 596 {
383 STRLEN len; 597 STRLEN len;
384 char *str = SvPV (sv, len); 598 char *str = SvPV (sv, len);
385 encode_str (enc, SvUTF8 (sv), str, len); 599 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
386 } 600 }
387 else if (SvNOKp (sv)) 601 else if (SvNOKp (sv))
388 encode_nv (enc, sv); 602 encode_nv (enc, sv);
389 else if (SvIOKp (sv)) 603 else if (SvIOKp (sv))
390 { 604 {
391 if (SvIsUV (sv)) 605 if (SvIsUV (sv))
392 encode_uint (enc, 0x00, SvUVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
393 else if (SvIVX (sv) >= 0) 607 else if (SvIVX (sv) >= 0)
394 encode_uint (enc, 0x00, SvIVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
395 else 609 else
396 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 610 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
397 } 611 }
398 else if (SvROK (sv)) 612 else if (SvROK (sv))
399 encode_rv (enc, SvRV (sv)); 613 encode_rv (enc, SvRV (sv));
400 else if (!SvOK (sv)) 614 else if (!SvOK (sv))
401 encode_ch (enc, 0xe0 | 22); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
402 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 616 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
403 encode_ch (enc, 0xe0 | 23); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
404 else 618 else
405 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 619 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
406 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 620 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
407} 621}
408 622
409static SV * 623static SV *
410encode_cbor (SV *scalar, CBOR *cbor) 624encode_cbor (SV *scalar, CBOR *cbor)
411{ 625{
412 enc_t enc; 626 enc_t enc = { 0 };
413 627
414 enc.cbor = *cbor; 628 enc.cbor = *cbor;
415 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 629 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
416 enc.cur = SvPVX (enc.sv); 630 enc.cur = SvPVX (enc.sv);
417 enc.end = SvEND (enc.sv); 631 enc.end = SvEND (enc.sv);
418 enc.depth = 0;
419 632
420 SvPOK_only (enc.sv); 633 SvPOK_only (enc.sv);
634
635 if (cbor->flags & F_PACK_STRINGS)
636 {
637 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
638 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
639 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
640 }
641
421 encode_sv (&enc, scalar); 642 encode_sv (&enc, scalar);
422 643
423 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 644 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
424 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 645 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
425 646
439 U8 *end; // end of input string 660 U8 *end; // end of input string
440 const char *err; // parse error, if != 0 661 const char *err; // parse error, if != 0
441 CBOR cbor; 662 CBOR cbor;
442 U32 depth; // recursion depth 663 U32 depth; // recursion depth
443 U32 maxdepth; // recursion depth limit 664 U32 maxdepth; // recursion depth limit
665 AV *shareable;
666 AV *stringref;
667 SV *decode_tagged;
668 SV *err_sv; // optional sv for error, needs to be freed
444} dec_t; 669} dec_t;
445 670
446#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 671// set dec->err to ERRSV
672ecb_cold static void
673err_errsv (dec_t *dec)
674{
675 if (!dec->err)
676 {
677 dec->err_sv = newSVsv (ERRSV);
447 678
448#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 679 // chop off the trailing \n
680 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
681 *SvEND (dec->err_sv) = 0;
449 682
683 dec->err = SvPVutf8_nolen (dec->err_sv);
684 }
685}
686
687// the following functions are used to reduce code size and help the compiler to optimise
688ecb_cold static void
689err_set (dec_t *dec, const char *reason)
690{
691 if (!dec->err)
692 dec->err = reason;
693}
694
695ecb_cold static void
696err_unexpected_end (dec_t *dec)
697{
698 err_set (dec, "unexpected end of CBOR data");
699}
700
701ecb_cold static void
702err_nesting_exceeded (dec_t *dec)
703{
704 err_set (dec, ERR_NESTING_EXCEEDED);
705}
706
707#define ERR_DO(do) SB do; goto fail; SE
708#define ERR(reason) ERR_DO (err_set (dec, reason))
709#define ERR_ERRSV ERR_DO (err_errsv (dec))
710
711#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
712
450#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 713#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
451#define DEC_DEC_DEPTH --dec->depth 714#define DEC_DEC_DEPTH --dec->depth
452 715
453static UV 716static UV
454decode_uint (dec_t *dec) 717decode_uint (dec_t *dec)
455{ 718{
456 switch (*dec->cur & 31) 719 U8 m = *dec->cur & MINOR_MASK;
457 { 720 ++dec->cur;
458 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
459 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
460 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
461 return *dec->cur++ & 31;
462 721
463 case 24: 722 if (ecb_expect_true (m < LENGTH_EXT1))
723 return m;
724 else if (ecb_expect_true (m == LENGTH_EXT1))
725 {
464 WANT (2); 726 WANT (1);
465 dec->cur += 2; 727 dec->cur += 1;
466 return dec->cur[-1]; 728 return dec->cur[-1];
467 729 }
468 case 25: 730 else if (ecb_expect_true (m == LENGTH_EXT2))
731 {
469 WANT (3); 732 WANT (2);
470 dec->cur += 3; 733 dec->cur += 2;
471 return (((UV)dec->cur[-2]) << 8) 734 return (((UV)dec->cur[-2]) << 8)
472 | ((UV)dec->cur[-1]); 735 | ((UV)dec->cur[-1]);
473 736 }
474 case 26: 737 else if (ecb_expect_true (m == LENGTH_EXT4))
738 {
475 WANT (5); 739 WANT (4);
476 dec->cur += 5; 740 dec->cur += 4;
477 return (((UV)dec->cur[-4]) << 24) 741 return (((UV)dec->cur[-4]) << 24)
478 | (((UV)dec->cur[-3]) << 16) 742 | (((UV)dec->cur[-3]) << 16)
479 | (((UV)dec->cur[-2]) << 8) 743 | (((UV)dec->cur[-2]) << 8)
480 | ((UV)dec->cur[-1]); 744 | ((UV)dec->cur[-1]);
481 745 }
482 case 27: 746 else if (ecb_expect_true (m == LENGTH_EXT8))
747 {
483 WANT (9); 748 WANT (8);
484 dec->cur += 9; 749 dec->cur += 8;
750
751 return
752#if UVSIZE < 8
753 0
754#else
485 return (((UV)dec->cur[-8]) << 56) 755 (((UV)dec->cur[-8]) << 56)
486 | (((UV)dec->cur[-7]) << 48) 756 | (((UV)dec->cur[-7]) << 48)
487 | (((UV)dec->cur[-6]) << 40) 757 | (((UV)dec->cur[-6]) << 40)
488 | (((UV)dec->cur[-5]) << 32) 758 | (((UV)dec->cur[-5]) << 32)
759#endif
489 | (((UV)dec->cur[-4]) << 24) 760 | (((UV)dec->cur[-4]) << 24)
490 | (((UV)dec->cur[-3]) << 16) 761 | (((UV)dec->cur[-3]) << 16)
491 | (((UV)dec->cur[-2]) << 8) 762 | (((UV)dec->cur[-2]) << 8)
492 | ((UV)dec->cur[-1]); 763 | ((UV)dec->cur[-1]);
493 764 }
494 default: 765 else
495 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 766 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
496 }
497 767
498fail: 768fail:
499 return 0; 769 return 0;
500} 770}
501 771
506{ 776{
507 AV *av = newAV (); 777 AV *av = newAV ();
508 778
509 DEC_INC_DEPTH; 779 DEC_INC_DEPTH;
510 780
511 if ((*dec->cur & 31) == 31) 781 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
512 { 782 {
513 ++dec->cur; 783 ++dec->cur;
514 784
515 for (;;) 785 for (;;)
516 { 786 {
517 WANT (1); 787 WANT (1);
518 788
519 if (*dec->cur == (0xe0 | 31)) 789 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
520 { 790 {
521 ++dec->cur; 791 ++dec->cur;
522 break; 792 break;
523 } 793 }
524 794
525 av_push (av, decode_sv (dec)); 795 av_push (av, decode_sv (dec));
526 } 796 }
527 } 797 }
528 else 798 else
529 { 799 {
530 int i, len = decode_uint (dec); 800 UV i, len = decode_uint (dec);
531 801
802 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
532 av_fill (av, len - 1); 803 av_fill (av, len - 1);
533 804
534 for (i = 0; i < len; ++i) 805 for (i = 0; i < len; ++i)
535 AvARRAY (av)[i] = decode_sv (dec); 806 AvARRAY (av)[i] = decode_sv (dec);
536 } 807 }
542 SvREFCNT_dec (av); 813 SvREFCNT_dec (av);
543 DEC_DEC_DEPTH; 814 DEC_DEC_DEPTH;
544 return &PL_sv_undef; 815 return &PL_sv_undef;
545} 816}
546 817
818static void
819decode_he (dec_t *dec, HV *hv)
820{
821 // for speed reasons, we specialcase single-string
822 // byte or utf-8 strings as keys, but only when !stringref
823
824 if (ecb_expect_true (!dec->stringref))
825 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
826 {
827 STRLEN len = decode_uint (dec);
828 char *key = (char *)dec->cur;
829
830 WANT (len);
831 dec->cur += len;
832
833 hv_store (hv, key, len, decode_sv (dec), 0);
834
835 return;
836 }
837 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
838 {
839 STRLEN len = decode_uint (dec);
840 char *key = (char *)dec->cur;
841
842 WANT (len);
843 dec->cur += len;
844
845 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
846 if (!is_utf8_string (key, len))
847 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
848
849 hv_store (hv, key, -len, decode_sv (dec), 0);
850
851 return;
852 }
853
854 SV *k = decode_sv (dec);
855 SV *v = decode_sv (dec);
856
857 // we leak memory if uncaught exceptions are thrown by random magical
858 // methods, and this is hopefully the only place where it can happen,
859 // so if there is a chance of an exception, take the very slow path.
860 // since catching exceptions is "undocumented/internal/forbidden" by
861 // the new p5p powers, we need to call out to a perl function :/
862 if (ecb_expect_false (SvAMAGIC (k)))
863 {
864 dSP;
865
866 ENTER; SAVETMPS;
867 PUSHMARK (SP);
868 EXTEND (SP, 3);
869 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
870 PUSHs (sv_2mortal (k));
871 PUSHs (sv_2mortal (v));
872
873 PUTBACK;
874 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
875 SPAGAIN;
876
877 FREETMPS; LEAVE;
878
879 if (SvTRUE (ERRSV))
880 ERR_ERRSV;
881
882 return;
883 }
884
885 hv_store_ent (hv, k, v, 0);
886 SvREFCNT_dec (k);
887
888fail:
889 ;
890}
891
547static SV * 892static SV *
548decode_hv (dec_t *dec) 893decode_hv (dec_t *dec)
549{ 894{
550 HV *hv = newHV (); 895 HV *hv = newHV ();
551 896
552 DEC_INC_DEPTH; 897 DEC_INC_DEPTH;
553 898
554 if ((*dec->cur & 31) == 31) 899 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
555 { 900 {
556 ++dec->cur; 901 ++dec->cur;
557 902
558 for (;;) 903 for (;;)
559 { 904 {
560 WANT (1); 905 WANT (1);
561 906
562 if (*dec->cur == (0xe0 | 31)) 907 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
563 { 908 {
564 ++dec->cur; 909 ++dec->cur;
565 break; 910 break;
566 } 911 }
567 912
568 SV *k = decode_sv (dec); 913 decode_he (dec, hv);
569 SV *v = decode_sv (dec);
570
571 hv_store_ent (hv, k, v, 0);
572 } 914 }
573 } 915 }
574 else 916 else
575 { 917 {
576 int len = decode_uint (dec); 918 UV pairs = decode_uint (dec);
577 919
920 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
921
578 while (len--) 922 while (pairs--)
579 { 923 decode_he (dec, hv);
580 SV *k = decode_sv (dec);
581 SV *v = decode_sv (dec);
582
583 hv_store_ent (hv, k, v, 0);
584 }
585 } 924 }
586 925
587 DEC_DEC_DEPTH; 926 DEC_DEC_DEPTH;
588 return newRV_noinc ((SV *)hv); 927 return newRV_noinc ((SV *)hv);
589 928
596static SV * 935static SV *
597decode_str (dec_t *dec, int utf8) 936decode_str (dec_t *dec, int utf8)
598{ 937{
599 SV *sv = 0; 938 SV *sv = 0;
600 939
601 if ((*dec->cur & 31) == 31) 940 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
602 { 941 {
942 // indefinite length strings
603 ++dec->cur; 943 ++dec->cur;
604 944
945 U8 major = *dec->cur & MAJOR_MISC;
946
605 sv = newSVpvn ("", 0); 947 sv = newSVpvn ("", 0);
606 948
607 // not very fast, and certainly not robust against illegal input
608 for (;;) 949 for (;;)
609 { 950 {
610 WANT (1); 951 WANT (1);
611 952
612 if (*dec->cur == (0xe0 | 31)) 953 if ((*dec->cur - major) > LENGTH_EXT8)
954 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
613 { 955 {
614 ++dec->cur; 956 ++dec->cur;
615 break; 957 break;
616 } 958 }
959 else
960 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
617 961
618 sv_catsv (sv, decode_sv (dec)); 962 STRLEN len = decode_uint (dec);
963
964 WANT (len);
965 sv_catpvn (sv, dec->cur, len);
966 dec->cur += len;
619 } 967 }
620 } 968 }
621 else 969 else
622 { 970 {
623 STRLEN len = decode_uint (dec); 971 STRLEN len = decode_uint (dec);
624 972
625 WANT (len); 973 WANT (len);
626 sv = newSVpvn (dec->cur, len); 974 sv = newSVpvn (dec->cur, len);
627 dec->cur += len; 975 dec->cur += len;
976
977 if (ecb_expect_false (dec->stringref)
978 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
979 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
628 } 980 }
629 981
630 if (utf8) 982 if (utf8)
983 {
984 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
985 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
986 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
987
631 SvUTF8_on (sv); 988 SvUTF8_on (sv);
989 }
632 990
633 return sv; 991 return sv;
634 992
635fail: 993fail:
636 SvREFCNT_dec (sv); 994 SvREFCNT_dec (sv);
638} 996}
639 997
640static SV * 998static SV *
641decode_tagged (dec_t *dec) 999decode_tagged (dec_t *dec)
642{ 1000{
1001 SV *sv = 0;
643 UV tag = decode_uint (dec); 1002 UV tag = decode_uint (dec);
1003
1004 WANT (1);
1005
1006 switch (tag)
1007 {
1008 case CBOR_TAG_MAGIC:
644 SV *sv = decode_sv (dec); 1009 sv = decode_sv (dec);
1010 break;
645 1011
646 if (tag == CBOR_TAG_MAGIC) 1012 case CBOR_TAG_INDIRECTION:
647 return sv; 1013 sv = newRV_noinc (decode_sv (dec));
1014 break;
648 1015
649 if (tag == CBOR_TAG_PERL_OBJECT) 1016 case CBOR_TAG_STRINGREF_NAMESPACE:
650 { 1017 {
1018 // do nmot use SAVETMPS/FREETMPS, as these will
1019 // erase mortalised caches, e.g. "shareable"
1020 ENTER;
1021
1022 SAVESPTR (dec->stringref);
1023 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1024
1025 sv = decode_sv (dec);
1026
1027 LEAVE;
1028 }
1029 break;
1030
1031 case CBOR_TAG_STRINGREF:
1032 {
1033 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1034 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1035
1036 UV idx = decode_uint (dec);
1037
1038 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
1039 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1040
1041 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1042 }
1043 break;
1044
1045 case CBOR_TAG_VALUE_SHAREABLE:
1046 {
1047 if (ecb_expect_false (!dec->shareable))
1048 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1049
1050 if (dec->cbor.flags & F_ALLOW_CYCLES)
1051 {
1052 sv = newSV (0);
1053 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1054
1055 SV *osv = decode_sv (dec);
1056 sv_setsv (sv, osv);
1057 SvREFCNT_dec_NN (osv);
1058 }
1059 else
1060 {
1061 av_push (dec->shareable, &PL_sv_undef);
1062 int idx = AvFILLp (dec->shareable);
1063 sv = decode_sv (dec);
1064 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1065 }
1066 }
1067 break;
1068
1069 case CBOR_TAG_VALUE_SHAREDREF:
1070 {
1071 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1072 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1073
1074 UV idx = decode_uint (dec);
1075
1076 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
1077 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1078
1079 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1080
1081 if (sv == &PL_sv_undef)
1082 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1083 }
1084 break;
1085
1086 case CBOR_TAG_PERL_OBJECT:
1087 {
1088 if (dec->cbor.flags & F_FORBID_OBJECTS)
1089 goto filter;
1090
1091 sv = decode_sv (dec);
1092
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1093 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 1094 ERR ("corrupted CBOR data (non-array perl object)");
1095
1096 AV *av = (AV *)SvRV (sv);
1097 int len = av_len (av) + 1;
1098 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1099
1100 if (!stash)
1101 ERR ("cannot decode perl-object (package does not exist)");
1102
1103 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1104
1105 if (!method)
1106 ERR ("cannot decode perl-object (package does not have a THAW method)");
1107
1108 dSP;
1109
1110 ENTER; SAVETMPS;
1111 PUSHMARK (SP);
1112 EXTEND (SP, len + 1);
1113 // we re-bless the reference to get overload and other niceties right
1114 PUSHs (*av_fetch (av, 0, 1));
1115 PUSHs (sv_cbor);
1116
1117 int i;
1118
1119 for (i = 1; i < len; ++i)
1120 PUSHs (*av_fetch (av, i, 1));
1121
1122 PUTBACK;
1123 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1124 SPAGAIN;
1125
1126 if (SvTRUE (ERRSV))
1127 {
1128 FREETMPS; LEAVE;
1129 ERR_ERRSV;
1130 }
1131
1132 SvREFCNT_dec (sv);
1133 sv = SvREFCNT_inc (POPs);
1134
1135 PUTBACK;
1136
1137 FREETMPS; LEAVE;
653 1138 }
654 // TODO 1139 break;
655 }
656 1140
1141 default:
1142 filter:
1143 {
1144 SV *tag_sv = newSVuv (tag);
1145
1146 sv = decode_sv (dec);
1147
1148 dSP;
1149 ENTER; SAVETMPS;
1150 SAVESTACK_POS ();
1151 PUSHMARK (SP);
1152 EXTEND (SP, 2);
1153 PUSHs (tag_sv);
1154 PUSHs (sv);
1155
1156 PUTBACK;
1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1158 SPAGAIN;
1159
1160 if (SvTRUE (ERRSV))
1161 {
1162 SvREFCNT_dec (tag_sv);
1163 FREETMPS; LEAVE;
1164 ERR_ERRSV;
1165 }
1166
1167 if (count)
1168 {
1169 SvREFCNT_dec (tag_sv);
1170 SvREFCNT_dec (sv);
1171 sv = SvREFCNT_inc (POPs);
1172 }
1173 else
1174 {
657 AV *av = newAV (); 1175 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 1176 av_push (av, tag_sv);
659 av_push (av, sv); 1177 av_push (av, sv);
660 1178
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1179 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 1180 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1181 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1182 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1183 }
1184
1185 PUTBACK;
1186
1187 FREETMPS; LEAVE;
1188 }
1189 break;
1190 }
1191
1192 return sv;
666 1193
667fail: 1194fail:
668 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 1196 return &PL_sv_undef;
670} 1197}
672static SV * 1199static SV *
673decode_sv (dec_t *dec) 1200decode_sv (dec_t *dec)
674{ 1201{
675 WANT (1); 1202 WANT (1);
676 1203
677 switch (*dec->cur >> 5) 1204 switch (*dec->cur >> MAJOR_SHIFT)
678 { 1205 {
679 case 0: // unsigned int 1206 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
680 return newSVuv (decode_uint (dec)); 1207 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
681 case 1: // negative int 1208 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
682 return newSViv (-1 - (IV)decode_uint (dec)); 1209 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
683 case 2: // octet string 1210 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
684 return decode_str (dec, 0); 1211 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
685 case 3: // utf-8 string 1212 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
686 return decode_str (dec, 1); 1213
687 case 4: // array 1214 case MAJOR_MISC >> MAJOR_SHIFT:
688 return decode_av (dec);
689 case 5: // map
690 return decode_hv (dec);
691 case 6: // tag
692 return decode_tagged (dec);
693 case 7: // misc
694 switch (*dec->cur++ & 31) 1215 switch (*dec->cur++ & MINOR_MASK)
695 { 1216 {
696 case 20: 1217 case SIMPLE_FALSE:
697#if CBOR_SLOW 1218#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1219 types_false = get_bool ("Types::Serialiser::false");
699#endif 1220#endif
700 return newSVsv (cbor_false); 1221 return newSVsv (types_false);
701 case 21: 1222 case SIMPLE_TRUE:
702#if CBOR_SLOW 1223#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1224 types_true = get_bool ("Types::Serialiser::true");
704#endif 1225#endif
705 return newSVsv (cbor_true); 1226 return newSVsv (types_true);
706 case 22: 1227 case SIMPLE_NULL:
707 return newSVsv (&PL_sv_undef); 1228 return newSVsv (&PL_sv_undef);
1229 case SIMPLE_UNDEF:
1230#if CBOR_SLOW
1231 types_error = get_bool ("Types::Serialiser::error");
1232#endif
1233 return newSVsv (types_error);
708 1234
709 case 25: 1235 case MISC_FLOAT16:
710 { 1236 {
711 WANT (2); 1237 WANT (2);
712 1238
713 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1239 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
714 dec->cur += 2; 1240 dec->cur += 2;
715 1241
716 return newSVnv (ecb_binary16_to_float (fp)); 1242 return newSVnv (ecb_binary16_to_float (fp));
717 } 1243 }
718 1244
719 case 26: 1245 case MISC_FLOAT32:
720 { 1246 {
721 uint32_t fp; 1247 uint32_t fp;
722 WANT (4); 1248 WANT (4);
723 memcpy (&fp, dec->cur, 4); 1249 memcpy (&fp, dec->cur, 4);
724 dec->cur += 4; 1250 dec->cur += 4;
727 fp = ecb_bswap32 (fp); 1253 fp = ecb_bswap32 (fp);
728 1254
729 return newSVnv (ecb_binary32_to_float (fp)); 1255 return newSVnv (ecb_binary32_to_float (fp));
730 } 1256 }
731 1257
732 case 27: 1258 case MISC_FLOAT64:
733 { 1259 {
734 uint64_t fp; 1260 uint64_t fp;
735 WANT (8); 1261 WANT (8);
736 memcpy (&fp, dec->cur, 8); 1262 memcpy (&fp, dec->cur, 8);
737 dec->cur += 8; 1263 dec->cur += 8;
740 fp = ecb_bswap64 (fp); 1266 fp = ecb_bswap64 (fp);
741 1267
742 return newSVnv (ecb_binary64_to_double (fp)); 1268 return newSVnv (ecb_binary64_to_double (fp));
743 } 1269 }
744 1270
745 // 0..19 unassigned 1271 // 0..19 unassigned simple
746 // 24 reserved + unassigned (reserved values are not encodable) 1272 // 24 reserved + unassigned simple (reserved values are not encodable)
1273 // 28-30 unassigned misc
1274 // 31 break code
747 default: 1275 default:
748 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1276 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
749 } 1277 }
750 1278
751 break; 1279 break;
752 } 1280 }
753 1281
756} 1284}
757 1285
758static SV * 1286static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1287decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1288{
761 dec_t dec; 1289 dec_t dec = { 0 };
762 SV *sv; 1290 SV *sv;
1291 STRLEN len;
1292 char *data = SvPVbyte (string, len);
763 1293
764 /* work around bugs in 5.10 where manipulating magic values
765 * makes perl ignore the magic in subsequent accesses.
766 * also make a copy of non-PV values, to get them into a clean
767 * state (SvPV should do that, but it's buggy, see below).
768 */
769 /*SvGETMAGIC (string);*/
770 if (SvMAGICAL (string) || !SvPOK (string))
771 string = sv_2mortal (newSVsv (string));
772
773 SvUPGRADE (string, SVt_PV);
774
775 /* work around a bug in perl 5.10, which causes SvCUR to fail an
776 * assertion with -DDEBUGGING, although SvCUR is documented to
777 * return the xpv_cur field which certainly exists after upgrading.
778 * according to nicholas clark, calling SvPOK fixes this.
779 * But it doesn't fix it, so try another workaround, call SvPV_nolen
780 * and hope for the best.
781 * Damnit, SvPV_nolen still trips over yet another assertion. This
782 * assertion business is seriously broken, try yet another workaround
783 * for the broken -DDEBUGGING.
784 */
785 {
786#ifdef DEBUGGING
787 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
788#else
789 STRLEN offset = SvCUR (string);
790#endif
791
792 if (offset > cbor->max_size && cbor->max_size) 1294 if (len > cbor->max_size && cbor->max_size)
793 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1295 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
794 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1296 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1297
799 dec.cbor = *cbor; 1298 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1299 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1300 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1301
805 sv = decode_sv (&dec); 1302 sv = decode_sv (&dec);
806 1303
807 if (offset_return) 1304 if (offset_return)
808 *offset_return = dec.cur; 1305 *offset_return = dec.cur;
811 if (dec.cur != dec.end && !dec.err) 1308 if (dec.cur != dec.end && !dec.err)
812 dec.err = "garbage after CBOR object"; 1309 dec.err = "garbage after CBOR object";
813 1310
814 if (dec.err) 1311 if (dec.err)
815 { 1312 {
1313 if (dec.shareable)
1314 {
1315 // need to break cyclic links, which would all be in shareable
1316 int i;
1317 SV **svp;
1318
1319 for (i = av_len (dec.shareable) + 1; i--; )
1320 if ((svp = av_fetch (dec.shareable, i, 0)))
1321 sv_setsv (*svp, &PL_sv_undef);
1322 }
1323
816 SvREFCNT_dec (sv); 1324 SvREFCNT_dec (sv);
1325
1326 if (dec.err_sv)
1327 sv_2mortal (dec.err_sv);
1328
817 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1329 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
818 } 1330 }
819 1331
820 sv = sv_2mortal (sv); 1332 sv = sv_2mortal (sv);
821 1333
822 return sv; 1334 return sv;
823} 1335}
824 1336
1337/////////////////////////////////////////////////////////////////////////////
1338// incremental parser
1339
1340#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1341
1342// returns 0 for notyet, 1 for success or error
1343static int
1344incr_parse (CBOR *self, SV *cborstr)
1345{
1346 STRLEN cur;
1347 SvPV (cborstr, cur);
1348
1349 while (ecb_expect_true (self->incr_need <= cur))
1350 {
1351 // table of integer count bytes
1352 static I8 incr_len[MINOR_MASK + 1] = {
1353 0, 0, 0, 0, 0, 0, 0, 0,
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 1, 2, 4, 8,-1,-1,-1,-2
1357 };
1358
1359 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1360 U8 m = *p & MINOR_MASK;
1361 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1362 I8 ilen = incr_len[m];
1363
1364 self->incr_need = self->incr_pos + 1;
1365
1366 if (ecb_expect_false (ilen < 0))
1367 {
1368 if (m != MINOR_INDEF)
1369 return 1; // error
1370
1371 if (*p == (MAJOR_MISC | MINOR_INDEF))
1372 {
1373 if (count >= 0)
1374 return 1; // error
1375
1376 count = 1;
1377 }
1378 else
1379 {
1380 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1381 count = -1;
1382 }
1383 }
1384 else
1385 {
1386 self->incr_need += ilen;
1387 if (ecb_expect_false (self->incr_need > cur))
1388 return 0;
1389
1390 int major = *p >> MAJOR_SHIFT;
1391
1392 switch (major)
1393 {
1394 case MAJOR_TAG >> MAJOR_SHIFT:
1395 ++count; // tags merely prefix another value
1396 break;
1397
1398 case MAJOR_BYTES >> MAJOR_SHIFT:
1399 case MAJOR_TEXT >> MAJOR_SHIFT:
1400 case MAJOR_ARRAY >> MAJOR_SHIFT:
1401 case MAJOR_MAP >> MAJOR_SHIFT:
1402 {
1403 UV len;
1404
1405 if (ecb_expect_false (ilen))
1406 {
1407 len = 0;
1408
1409 do {
1410 len = (len << 8) | *++p;
1411 } while (--ilen);
1412 }
1413 else
1414 len = m;
1415
1416 switch (major)
1417 {
1418 case MAJOR_BYTES >> MAJOR_SHIFT:
1419 case MAJOR_TEXT >> MAJOR_SHIFT:
1420 self->incr_need += len;
1421 if (ecb_expect_false (self->incr_need > cur))
1422 return 0;
1423
1424 break;
1425
1426 case MAJOR_MAP >> MAJOR_SHIFT:
1427 len <<= 1;
1428 case MAJOR_ARRAY >> MAJOR_SHIFT:
1429 if (len)
1430 {
1431 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1432 count = len + 1;
1433 }
1434 break;
1435 }
1436 }
1437 }
1438 }
1439
1440 self->incr_pos = self->incr_need;
1441
1442 if (count > 0)
1443 {
1444 while (!--count)
1445 {
1446 if (!AvFILLp (self->incr_count))
1447 return 1; // done
1448
1449 SvREFCNT_dec_NN (av_pop (self->incr_count));
1450 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1451 }
1452
1453 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1454 }
1455 }
1456
1457 return 0;
1458}
1459
1460
825///////////////////////////////////////////////////////////////////////////// 1461/////////////////////////////////////////////////////////////////////////////
826// XS interface functions 1462// XS interface functions
827 1463
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1464MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1465
830BOOT: 1466BOOT:
831{ 1467{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1468 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
833 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
834 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1469 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1470
836 cbor_true = get_bool ("CBOR::XS::true"); 1471 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1472 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1473
1474 types_true = get_bool ("Types::Serialiser::true" );
1475 types_false = get_bool ("Types::Serialiser::false");
1476 types_error = get_bool ("Types::Serialiser::error");
1477
1478 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1479
1480 sv_cbor = newSVpv ("CBOR", 0);
1481 SvREADONLY_on (sv_cbor);
1482
1483 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
838} 1484}
839 1485
840PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
841 1487
842void CLONE (...) 1488void CLONE (...)
843 CODE: 1489 CODE:
844 cbor_stash = 0; 1490 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1491 cbor_tagged_stash = 0;
1492 types_error_stash = 0;
1493 types_boolean_stash = 0;
847 1494
848void new (char *klass) 1495void new (char *klass)
849 PPCODE: 1496 PPCODE:
850{ 1497{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1498 SV *pv = NEWSV (0, sizeof (CBOR));
859 1506
860void shrink (CBOR *self, int enable = 1) 1507void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1508 ALIAS:
862 shrink = F_SHRINK 1509 shrink = F_SHRINK
863 allow_unknown = F_ALLOW_UNKNOWN 1510 allow_unknown = F_ALLOW_UNKNOWN
1511 allow_sharing = F_ALLOW_SHARING
1512 allow_cycles = F_ALLOW_CYCLES
1513 forbid_objects = F_FORBID_OBJECTS
1514 pack_strings = F_PACK_STRINGS
1515 text_keys = F_TEXT_KEYS
1516 text_strings = F_TEXT_STRINGS
1517 validate_utf8 = F_VALIDATE_UTF8
864 PPCODE: 1518 PPCODE:
865{ 1519{
866 if (enable) 1520 if (enable)
867 self->flags |= ix; 1521 self->flags |= ix;
868 else 1522 else
873 1527
874void get_shrink (CBOR *self) 1528void get_shrink (CBOR *self)
875 ALIAS: 1529 ALIAS:
876 get_shrink = F_SHRINK 1530 get_shrink = F_SHRINK
877 get_allow_unknown = F_ALLOW_UNKNOWN 1531 get_allow_unknown = F_ALLOW_UNKNOWN
1532 get_allow_sharing = F_ALLOW_SHARING
1533 get_allow_cycles = F_ALLOW_CYCLES
1534 get_forbid_objects = F_FORBID_OBJECTS
1535 get_pack_strings = F_PACK_STRINGS
1536 get_text_keys = F_TEXT_KEYS
1537 get_text_strings = F_TEXT_STRINGS
1538 get_validate_utf8 = F_VALIDATE_UTF8
878 PPCODE: 1539 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1540 XPUSHs (boolSV (self->flags & ix));
880 1541
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1542void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1543 PPCODE:
898 CODE: 1559 CODE:
899 RETVAL = self->max_size; 1560 RETVAL = self->max_size;
900 OUTPUT: 1561 OUTPUT:
901 RETVAL 1562 RETVAL
902 1563
903#if 0 //TODO 1564void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1565 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1566 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1567 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1568 XPUSHs (ST (0));
912}
913 1569
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1570SV *get_filter (CBOR *self)
915 PPCODE: 1571 CODE:
916{ 1572 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1573 OUTPUT:
918 self->cb_sk_object = newHV (); 1574 RETVAL
919
920 if (SvOK (cb))
921 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
922 else
923 {
924 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
925
926 if (!HvKEYS (self->cb_sk_object))
927 {
928 SvREFCNT_dec (self->cb_sk_object);
929 self->cb_sk_object = 0;
930 }
931 }
932
933 XPUSHs (ST (0));
934}
935
936#endif
937 1575
938void encode (CBOR *self, SV *scalar) 1576void encode (CBOR *self, SV *scalar)
939 PPCODE: 1577 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1578 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1579 XPUSHs (scalar);
954 EXTEND (SP, 2); 1592 EXTEND (SP, 2);
955 PUSHs (sv); 1593 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1594 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1595}
958 1596
959#if 0 1597void incr_parse (CBOR *self, SV *cborstr)
1598 ALIAS:
1599 incr_parse_multiple = 1
1600 PPCODE:
1601{
1602 if (SvUTF8 (cborstr))
1603 sv_utf8_downgrade (cborstr, 0);
1604
1605 if (!self->incr_count)
1606 {
1607 self->incr_count = newAV ();
1608 self->incr_pos = 0;
1609 self->incr_need = 1;
1610
1611 av_push (self->incr_count, newSViv (1));
1612 }
1613
1614 do
1615 {
1616 if (!incr_parse (self, cborstr))
1617 {
1618 if (self->incr_need > self->max_size && self->max_size)
1619 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1620 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1621
1622 break;
1623 }
1624
1625 SV *sv;
1626 char *offset;
1627
1628 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1629 XPUSHs (sv);
1630
1631 sv_chop (cborstr, offset);
1632
1633 av_clear (self->incr_count);
1634 av_push (self->incr_count, newSViv (1));
1635
1636 self->incr_pos = 0;
1637 self->incr_need = self->incr_pos + 1;
1638 }
1639 while (ix);
1640}
1641
1642void incr_reset (CBOR *self)
1643 CODE:
1644{
1645 SvREFCNT_dec (self->incr_count);
1646 self->incr_count = 0;
1647}
960 1648
961void DESTROY (CBOR *self) 1649void DESTROY (CBOR *self)
962 CODE: 1650 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1651 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1652
968PROTOTYPES: ENABLE 1653PROTOTYPES: ENABLE
969 1654
970void encode_cbor (SV *scalar) 1655void encode_cbor (SV *scalar)
1656 ALIAS:
1657 encode_cbor = 0
1658 encode_cbor_sharing = F_ALLOW_SHARING
971 PPCODE: 1659 PPCODE:
972{ 1660{
973 CBOR cbor; 1661 CBOR cbor;
974 cbor_init (&cbor); 1662 cbor_init (&cbor);
1663 cbor.flags |= ix;
975 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1664 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
976 XPUSHs (scalar); 1665 XPUSHs (scalar);
977} 1666}
978 1667
979void decode_cbor (SV *cborstr) 1668void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines