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 1.2 by root, Sat Oct 26 10:41:12 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 UTC

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 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
14#if defined(__BORLANDC__) || defined(_MSC_VER) 15// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
17 28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
18#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
19#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
20 106
21#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
22 108
23#define SB do { 109#define SB do {
24#define SE } while (0) 110#define SE } while (0)
25
26#if __GNUC__ >= 3
27# define expect(expr,value) __builtin_expect ((expr), (value))
28# define INLINE static inline
29#else
30# define expect(expr,value) (expr)
31# define INLINE static
32#endif
33
34#define expect_false(expr) expect ((expr) != 0, 0)
35#define expect_true(expr) expect ((expr) != 0, 1)
36 111
37#define IN_RANGE_INC(type,val,beg,end) \ 112#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 113 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 114 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 115
46#else 121#else
47# define CBOR_SLOW 0 122# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
49#endif 124#endif
50 125
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
53 128
54typedef struct { 129typedef struct {
55 U32 flags; 130 U32 flags;
56 U32 max_depth; 131 U32 max_depth;
57 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
58 134
59 SV *cb_object; 135 // for the incremental parser
60 HV *cb_sk_object; 136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
61} CBOR; 139} CBOR;
62 140
63INLINE void 141ecb_inline void
64cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
65{ 143{
66 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 145 cbor->max_depth = 512;
68} 146}
69 147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
153}
154
70///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
71// utility functions 156// utility functions
72 157
73INLINE SV * 158ecb_inline SV *
74get_bool (const char *name) 159get_bool (const char *name)
75{ 160{
76 SV *sv = get_sv (name, 1); 161 SV *sv = get_sv (name, 1);
77 162
78 SvREADONLY_on (sv); 163 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 164 SvREADONLY_on (SvRV (sv));
80 165
81 return sv; 166 return sv;
82} 167}
83 168
84INLINE void 169ecb_inline void
85shrink (SV *sv) 170shrink (SV *sv)
86{ 171{
87 sv_utf8_downgrade (sv, 1); 172 sv_utf8_downgrade (sv, 1);
88 173
89 if (SvLEN (sv) > SvCUR (sv) + 1) 174 if (SvLEN (sv) > SvCUR (sv) + 1)
94 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 180#endif
96 } 181 }
97} 182}
98 183
99///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
100// fp hell 185ecb_inline int
101 186minimum_string_length (UV idx)
102//TODO 187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
103 198
104///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
105// encoder 200// encoder
106 201
107// structure used for encoding CBOR 202// structure used for encoding CBOR
110 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
111 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
112 SV *sv; // result scalar 207 SV *sv; // result scalar
113 CBOR cbor; 208 CBOR cbor;
114 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;
115} enc_t; 214} enc_t;
116 215
117INLINE void 216ecb_inline void
118need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
119{ 218{
120 if (expect_false (enc->cur + len >= enc->end)) 219 if (ecb_expect_false (enc->cur + len >= enc->end))
121 { 220 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 223 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 225 }
127} 226}
128 227
129INLINE void 228ecb_inline void
130encode_ch (enc_t *enc, char ch) 229encode_ch (enc_t *enc, char ch)
131{ 230{
132 need (enc, 1); 231 need (enc, 1);
133 *enc->cur++ = ch; 232 *enc->cur++ = ch;
134} 233}
136static void 235static void
137encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
138{ 237{
139 need (enc, 9); 238 need (enc, 9);
140 239
141 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
142 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
143 else if (len < 0x100) 242 else if (ecb_expect_true (len <= 0xffU))
144 { 243 {
145 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
146 *enc->cur++ = len; 245 *enc->cur++ = len;
147 } 246 }
148 else if (len < 0x10000) 247 else if (len <= 0xffffU)
149 { 248 {
150 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
151 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 251 *enc->cur++ = len;
153 } 252 }
154 else if (len < 0x100000000) 253 else if (len <= 0xffffffffU)
155 { 254 {
156 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
157 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 259 *enc->cur++ = len;
161 } 260 }
162 else if (len) 261 else
163 { 262 {
164 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
165 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
168 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
169 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
171 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
172 *enc->cur++ = len; 271 *enc->cur++ = len;
173 } 272 }
174} 273}
175 274
176static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
177encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
178{ 283{
179 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
180 need (enc, len); 285 need (enc, len);
181 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
182 enc->cur += len; 287 enc->cur += len;
183} 288}
184 289
290static void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
292{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
313}
314
185static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
186 316
187static void 317static void
188encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
189{ 319{
192 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
193 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
194 324
195 ++enc->depth; 325 ++enc->depth;
196 326
197 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
198 328
329 if (SvMAGICAL (av))
199 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
200 { 331 {
201 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
202 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
203 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
204 338
205 --enc->depth; 339 --enc->depth;
206} 340}
207 341
208static void 342static void
217 351
218 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
219 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
220 354
221 if (mg) 355 if (mg)
222 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
223 else 357 else
224 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
225 359
226 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
227 { 361 {
228 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
230 else 364 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 366
233 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
234 } 368 }
235 369
236 if (mg) 370 if (mg)
237 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
238 372
239 --enc->depth; 373 --enc->depth;
240} 374}
241 375
242// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
243static void 377static void
244encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
245{ 379{
246 svtype svt;
247
248 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
249 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
250 383
251 if (expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
252 { 385 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? cbor_boolean_stash 387 ? types_boolean_stash
388 : gv_stashpv ("Types::Serialiser::Boolean", 1);
389 HV *error_stash = !CBOR_SLOW || types_error_stash
390 ? types_error_stash
391 : gv_stashpv ("Types::Serialiser::Error", 1);
392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
393 ? cbor_tagged_stash
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 395
257 if (SvSTASH (sv) == stash) 396 HV *stash = SvSTASH (sv);
258 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 397
398 if (stash == boolean_stash)
399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
403 else if (stash == error_stash)
404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
408 else if (stash == tagged_stash)
409 {
410 if (svt != SVt_PVAV)
411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
412
413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
259 else 434 else
260 { 435 {
261#if 0 //TODO 436 sv_setuv (*svp, enc->shareable_idx);
262 if (enc->cbor.flags & F_CONV_BLESSED) 437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
263 { 448 {
449 dSP;
450
451 ENTER; SAVETMPS; PUSHMARK (SP);
264 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
265 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
266
267 if (to_cbor)
268 {
269 dSP;
270
271 ENTER; SAVETMPS; PUSHMARK (SP);
272 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
273 454
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK; 455 PUTBACK;
456 // G_SCALAR ensures that return value is 1
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 457 call_sv ((SV *)GvCV (method), G_SCALAR);
277 SPAGAIN; 458 SPAGAIN;
278 459
279 // catch this surprisingly common error 460 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv) 461 if (SvROK (TOPs) && SvRV (TOPs) == sv)
281 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 462 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
282 463
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv); 464 encode_sv (enc, POPs);
287 465
466 PUTBACK;
467
288 FREETMPS; LEAVE; 468 FREETMPS; LEAVE;
289 }
290 else if (enc->cbor.flags & F_ALLOW_BLESSED)
291 encode_str (enc, "null", 4, 0);
292 else
293 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
294 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
295 } 469 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED) 470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
297 encode_str (enc, "null", 4, 0); 471 {
472 dSP;
473
474 ENTER; SAVETMPS; PUSHMARK (SP);
475 EXTEND (SP, 2);
476 // we re-bless the reference to get overload and other niceties right
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor);
479
480 PUTBACK;
481 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
482 SPAGAIN;
483
484 // catch this surprisingly common error
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
487
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491
492 while (count)
493 encode_sv (enc, SP[1 - count--]);
494
495 PUTBACK;
496
497 FREETMPS; LEAVE;
498 }
298 else 499 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 500 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 501 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 502 }
304 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 506 encode_av (enc, (AV *)sv);
308 else if (svt < SVt_PVAV)
309 {
310 STRLEN len = 0;
311 char *pv = svt ? SvPV (sv, len) : 0;
312
313 if (len == 1 && *pv == '1')
314 encode_ch (enc, 0xe0 | 21);
315 else if (len == 1 && *pv == '0')
316 encode_ch (enc, 0xe0 | 20);
317 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
318 encode_ch (enc, 0xe0 | 23);
319 else
320 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
321 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
322 }
323 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
324 encode_ch (enc, 0xe0 | 23);
325 else 507 else
326 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
327 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
328} 512}
329 513
330static void 514static void
331encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
332{ 516{
333 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
334 518
335 need (enc, 9); 519 need (enc, 9);
336 520
337 if (expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
339 //TODO: maybe I32? 523 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
341 { 525 {
342 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
343 527
344 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
345 529
346 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
347 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
348 532
349 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
351 } 535 }
352 else 536 else
353 { 537 {
354 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
355 539
356 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
357 541
358 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
359 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
360 544
361 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
370 554
371 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
372 { 556 {
373 STRLEN len; 557 STRLEN len;
374 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
375 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
376 } 560 }
377 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
378 encode_nv (enc, sv); 562 encode_nv (enc, sv);
379 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
380 { 564 {
381 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
382 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
383 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
384 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
385 else 569 else
386 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
387 } 571 }
388 else if (SvROK (sv)) 572 else if (SvROK (sv))
389 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
390 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
391 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
392 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
393 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 else 578 else
395 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
396 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
397} 581}
398 582
399static SV * 583static SV *
400encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
401{ 585{
402 enc_t enc; 586 enc_t enc = { };
403 587
404 enc.cbor = *cbor; 588 enc.cbor = *cbor;
405 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
406 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
407 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
408 enc.depth = 0;
409 592
410 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
594
595 if (cbor->flags & F_PACK_STRINGS)
596 {
597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
600 }
601
411 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
412 603
413 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
414 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 605 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
415 606
429 U8 *end; // end of input string 620 U8 *end; // end of input string
430 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
431 CBOR cbor; 622 CBOR cbor;
432 U32 depth; // recursion depth 623 U32 depth; // recursion depth
433 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
434} dec_t; 628} dec_t;
435 629
436#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
437 631
438#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
439 633
440#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 634#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
441#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
442 636
443static UV 637static UV
444decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
445{ 639{
446 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
447 { 641 ++dec->cur;
448 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
449 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
450 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
451 return *dec->cur++ & 31;
452 642
453 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
454 WANT (2); 647 WANT (1);
455 dec->cur += 2; 648 dec->cur += 1;
456 return dec->cur[-1]; 649 return dec->cur[-1];
457 650 }
458 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
459 WANT (3); 653 WANT (2);
460 dec->cur += 3; 654 dec->cur += 2;
461 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
462 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
463 657 }
464 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
465 WANT (5); 660 WANT (4);
466 dec->cur += 5; 661 dec->cur += 4;
467 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
468 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
469 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
470 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
471 666 }
472 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
473 WANT (9); 669 WANT (8);
474 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
475 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
476 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
477 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
478 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
479 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
480 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
481 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
482 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
483 685 }
484 default: 686 else
485 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
486 }
487 688
488fail: 689fail:
489 return 0; 690 return 0;
490} 691}
491 692
496{ 697{
497 AV *av = newAV (); 698 AV *av = newAV ();
498 699
499 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
500 701
501 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
502 { 703 {
503 ++dec->cur; 704 ++dec->cur;
504 705
505 for (;;) 706 for (;;)
506 { 707 {
507 WANT (1); 708 WANT (1);
508 709
509 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
510 { 711 {
511 ++dec->cur; 712 ++dec->cur;
512 break; 713 break;
513 } 714 }
514 715
517 } 718 }
518 else 719 else
519 { 720 {
520 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
521 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
522 av_fill (av, len - 1); 724 av_fill (av, len - 1);
523 725
524 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
525 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
526 } 728 }
532 SvREFCNT_dec (av); 734 SvREFCNT_dec (av);
533 DEC_DEC_DEPTH; 735 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 736 return &PL_sv_undef;
535} 737}
536 738
739static void
740decode_he (dec_t *dec, HV *hv)
741{
742 // for speed reasons, we specialcase single-string
743 // byte or utf-8 strings as keys, but only when !stringref
744
745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
756 }
757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
758 {
759 I32 len = decode_uint (dec);
760 char *key = (char *)dec->cur;
761
762 dec->cur += len;
763
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
767
768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
771 }
772
773 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec);
775
776 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k);
778
779fail:
780 ;
781}
782
537static SV * 783static SV *
538decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
539{ 785{
540 HV *hv = newHV (); 786 HV *hv = newHV ();
541 787
542 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
543 789
544 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
545 { 791 {
546 ++dec->cur; 792 ++dec->cur;
547 793
548 for (;;) 794 for (;;)
549 { 795 {
550 WANT (1); 796 WANT (1);
551 797
552 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
553 { 799 {
554 ++dec->cur; 800 ++dec->cur;
555 break; 801 break;
556 } 802 }
557 803
558 SV *k = decode_sv (dec); 804 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 805 }
563 } 806 }
564 else 807 else
565 { 808 {
566 int len = decode_uint (dec); 809 int pairs = decode_uint (dec);
567 810
568 while (len--) 811 while (pairs--)
569 { 812 decode_he (dec, hv);
570 SV *k = decode_sv (dec);
571 SV *v = decode_sv (dec);
572
573 hv_store_ent (hv, k, v, 0);
574 }
575 } 813 }
576 814
577 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 816 return newRV_noinc ((SV *)hv);
579
580#if 0
581 SV *sv;
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH;
585 decode_ws (dec);
586
587 for (;;)
588 {
589 // heuristic: assume that
590 // a) decode_str + hv_store_ent are abysmally slow.
591 // b) most hash keys are short, simple ascii text.
592 // => try to "fast-match" such strings to avoid
593 // the overhead of decode_str + hv_store_ent.
594 {
595 SV *value;
596 char *p = dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598
599 for (;;)
600 {
601 // the >= 0x80 is false on most architectures
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
603 {
604 // slow path, back up and use decode_str
605 SV *key = decode_str (dec);
606 if (!key)
607 goto fail;
608
609 decode_ws (dec); EXPECT_CH (':');
610
611 decode_ws (dec);
612 value = decode_sv (dec);
613 if (!value)
614 {
615 SvREFCNT_dec (key);
616 goto fail;
617 }
618
619 hv_store_ent (hv, key, value, 0);
620 SvREFCNT_dec (key);
621
622 break;
623 }
624 else if (*p == '"')
625 {
626 // fast path, got a simple key
627 char *key = dec->cur;
628 int len = p - key;
629 dec->cur = p + 1;
630
631 decode_ws (dec); EXPECT_CH (':');
632
633 decode_ws (dec);
634 value = decode_sv (dec);
635 if (!value)
636 goto fail;
637
638 hv_store (hv, key, len, value, 0);
639
640 break;
641 }
642
643 ++p;
644 }
645 }
646
647 decode_ws (dec);
648
649 if (*dec->cur == '}')
650 {
651 ++dec->cur;
652 break;
653 }
654
655 if (*dec->cur != ',')
656 ERR (", or } expected while parsing object/hash");
657
658 ++dec->cur;
659
660 decode_ws (dec);
661
662 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
663 {
664 ++dec->cur;
665 break;
666 }
667 }
668
669 DEC_DEC_DEPTH;
670 sv = newRV_noinc ((SV *)hv);
671
672 // check filter callbacks
673 if (dec->cbor.flags & F_HOOK)
674 {
675 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
676 {
677 HE *cb, *he;
678
679 hv_iterinit (hv);
680 he = hv_iternext (hv);
681 hv_iterinit (hv);
682
683 // the next line creates a mortal sv each time its called.
684 // might want to optimise this for common cases.
685 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
686
687 if (cb)
688 {
689 dSP;
690 int count;
691
692 ENTER; SAVETMPS; PUSHMARK (SP);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695
696 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
697
698 if (count == 1)
699 {
700 sv = newSVsv (POPs);
701 FREETMPS; LEAVE;
702 return sv;
703 }
704
705 SvREFCNT_inc (sv);
706 FREETMPS; LEAVE;
707 }
708 }
709
710 if (dec->cbor.cb_object)
711 {
712 dSP;
713 int count;
714
715 ENTER; SAVETMPS; PUSHMARK (SP);
716 XPUSHs (sv_2mortal (sv));
717
718 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
719
720 if (count == 1)
721 {
722 sv = newSVsv (POPs);
723 FREETMPS; LEAVE;
724 return sv;
725 }
726
727 SvREFCNT_inc (sv);
728 FREETMPS; LEAVE;
729 }
730 }
731
732 return sv;
733#endif
734 817
735fail: 818fail:
736 SvREFCNT_dec (hv); 819 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH; 820 DEC_DEC_DEPTH;
738 return &PL_sv_undef; 821 return &PL_sv_undef;
739} 822}
740 823
741static SV * 824static SV *
742decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
743{ 826{
744 SV *sv; 827 SV *sv = 0;
745 828
746 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
747 { 830 {
831 // indefinite length strings
748 ++dec->cur; 832 ++dec->cur;
749 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
750 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
751 837
752 // not very fast, and certainly not robust against illegal input
753 for (;;) 838 for (;;)
754 { 839 {
755 WANT (1); 840 WANT (1);
756 841
757 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
758 { 844 {
759 ++dec->cur; 845 ++dec->cur;
760 break; 846 break;
761 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
762 850
763 SV *sv2 = decode_sv (dec); 851 STRLEN len = decode_uint (dec);
764 sv_catsv (sv, sv2); 852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
765 } 856 }
766 } 857 }
767 else 858 else
768 { 859 {
769 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
770 861
771 WANT (len); 862 WANT (len);
772 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
773 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
774 } 869 }
775 870
776 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
777 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
778 879
779 return sv; 880 return sv;
780 881
781fail: 882fail:
883 SvREFCNT_dec (sv);
884 return &PL_sv_undef;
885}
886
887static SV *
888decode_tagged (dec_t *dec)
889{
890 SV *sv = 0;
891 UV tag = decode_uint (dec);
892
893 WANT (1);
894
895 switch (tag)
896 {
897 case CBOR_TAG_MAGIC:
898 sv = decode_sv (dec);
899 break;
900
901 case CBOR_TAG_INDIRECTION:
902 sv = newRV_noinc (decode_sv (dec));
903 break;
904
905 case CBOR_TAG_STRINGREF_NAMESPACE:
906 {
907 ENTER; SAVETMPS;
908
909 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911
912 sv = decode_sv (dec);
913
914 FREETMPS; LEAVE;
915 }
916 break;
917
918 case CBOR_TAG_STRINGREF:
919 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922
923 UV idx = decode_uint (dec);
924
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 }
930 break;
931
932 case CBOR_TAG_VALUE_SHAREABLE:
933 {
934 if (ecb_expect_false (!dec->shareable))
935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
939 sv = newSV (0);
940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
941
942 SV *osv = decode_sv (dec);
943 sv_setsv (sv, osv);
944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
953 }
954 break;
955
956 case CBOR_TAG_VALUE_SHAREDREF:
957 {
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960
961 UV idx = decode_uint (dec);
962
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
970 }
971 break;
972
973 case CBOR_TAG_PERL_OBJECT:
974 {
975 sv = decode_sv (dec);
976
977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
978 ERR ("corrupted CBOR data (non-array perl object)");
979
980 AV *av = (AV *)SvRV (sv);
981 int len = av_len (av) + 1;
982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
983
984 if (!stash)
985 ERR ("cannot decode perl-object (package does not exist)");
986
987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
988
989 if (!method)
990 ERR ("cannot decode perl-object (package does not have a THAW method)");
991
992 dSP;
993
994 ENTER; SAVETMPS; PUSHMARK (SP);
995 EXTEND (SP, len + 1);
996 // we re-bless the reference to get overload and other niceties right
997 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor);
999
1000 int i;
1001
1002 for (i = 1; i < len; ++i)
1003 PUSHs (*av_fetch (av, i, 1));
1004
1005 PUTBACK;
1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1007 SPAGAIN;
1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
1015 SvREFCNT_dec (sv);
1016 sv = SvREFCNT_inc (POPs);
1017
1018 PUTBACK;
1019
1020 FREETMPS; LEAVE;
1021 }
1022 break;
1023
1024 default:
1025 {
1026 sv = decode_sv (dec);
1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
1051 AV *av = newAV ();
1052 av_push (av, newSVuv (tag));
1053 av_push (av, sv);
1054
1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1056 ? cbor_tagged_stash
1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
1066 }
1067
1068 return sv;
1069
1070fail:
1071 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 1072 return &PL_sv_undef;
783} 1073}
784 1074
785static SV * 1075static SV *
786decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
787{ 1077{
788 WANT (1); 1078 WANT (1);
789 1079
790 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1081 {
792 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1089
800 return decode_str (dec, 1); 1090 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 abort ();
807 break;
808 case 7: // misc
809 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
810 { 1092 {
811 case 20: 1093 case SIMPLE_FALSE:
812#if CBOR_SLOW 1094#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
814#endif 1096#endif
815 return newSVsv (cbor_false); 1097 return newSVsv (types_false);
816 case 21: 1098 case SIMPLE_TRUE:
817#if CBOR_SLOW 1099#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
819#endif 1101#endif
820 return newSVsv (cbor_true); 1102 return newSVsv (types_true);
821 case 22: 1103 case SIMPLE_NULL:
822 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
1105 case SIMPLE_UNDEF:
1106#if CBOR_SLOW
1107 types_error = get_bool ("Types::Serialiser::error");
1108#endif
1109 return newSVsv (types_error);
823 1110
824 case 25: 1111 case MISC_FLOAT16:
825 { 1112 {
826 WANT (2); 1113 WANT (2);
827 1114
828 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
829 dec->cur += 2; 1116 dec->cur += 2;
830 1117
831 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
832 } 1119 }
833 1120
834 case 26: 1121 case MISC_FLOAT32:
835 { 1122 {
836 uint32_t fp; 1123 uint32_t fp;
837 WANT (4); 1124 WANT (4);
838 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
839 dec->cur += 4; 1126 dec->cur += 4;
842 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
843 1130
844 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
845 } 1132 }
846 1133
847 case 27: 1134 case MISC_FLOAT64:
848 { 1135 {
849 uint64_t fp; 1136 uint64_t fp;
850 WANT (8); 1137 WANT (8);
851 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
852 dec->cur += 8; 1139 dec->cur += 8;
855 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
856 1143
857 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
858 } 1145 }
859 1146
860 // 0..19 unassigned 1147 // 0..19 unassigned simple
861 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
862 default: 1151 default:
863 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
864 } 1153 }
865 1154
866 break; 1155 break;
867 } 1156 }
868#if 0
869 switch (*dec->cur)
870 {
871 //case '"': ++dec->cur; return decode_str (dec);
872 case '[': ++dec->cur; return decode_av (dec);
873 case '{': ++dec->cur; return decode_hv (dec);
874
875 case '-':
876 case '0': case '1': case '2': case '3': case '4':
877 case '5': case '6': case '7': case '8': case '9':
878 //TODO return decode_num (dec);
879
880 case 't':
881 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
882 {
883 dec->cur += 4;
884#if CBOR_SLOW
885 cbor_true = get_bool ("CBOR::XS::true");
886#endif
887 return newSVsv (cbor_true);
888 }
889 else
890 ERR ("'true' expected");
891
892 break;
893
894 case 'f':
895 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
896 {
897 dec->cur += 5;
898#if CBOR_SLOW
899 cbor_false = get_bool ("CBOR::XS::false");
900#endif
901 return newSVsv (cbor_false);
902 }
903 else
904 ERR ("'false' expected");
905
906 break;
907
908 case 'n':
909 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
910 {
911 dec->cur += 4;
912 return newSVsv (&PL_sv_undef);
913 }
914 else
915 ERR ("'null' expected");
916
917 break;
918
919 default:
920 ERR ("malformed CBOR string, neither array, object, number, string or atom");
921 break;
922 }
923#endif
924 1157
925fail: 1158fail:
926 return &PL_sv_undef; 1159 return &PL_sv_undef;
927} 1160}
928 1161
929static SV * 1162static SV *
930decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
931{ 1164{
932 dec_t dec; 1165 dec_t dec = { };
933 SV *sv; 1166 SV *sv;
1167 STRLEN len;
1168 char *data = SvPVbyte (string, len);
934 1169
935 /* work around bugs in 5.10 where manipulating magic values
936 * makes perl ignore the magic in subsequent accesses.
937 * also make a copy of non-PV values, to get them into a clean
938 * state (SvPV should do that, but it's buggy, see below).
939 */
940 /*SvGETMAGIC (string);*/
941 if (SvMAGICAL (string) || !SvPOK (string))
942 string = sv_2mortal (newSVsv (string));
943
944 SvUPGRADE (string, SVt_PV);
945
946 /* work around a bug in perl 5.10, which causes SvCUR to fail an
947 * assertion with -DDEBUGGING, although SvCUR is documented to
948 * return the xpv_cur field which certainly exists after upgrading.
949 * according to nicholas clark, calling SvPOK fixes this.
950 * But it doesn't fix it, so try another workaround, call SvPV_nolen
951 * and hope for the best.
952 * Damnit, SvPV_nolen still trips over yet another assertion. This
953 * assertion business is seriously broken, try yet another workaround
954 * for the broken -DDEBUGGING.
955 */
956 {
957#ifdef DEBUGGING
958 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
959#else
960 STRLEN offset = SvCUR (string);
961#endif
962
963 if (offset > cbor->max_size && cbor->max_size) 1170 if (len > cbor->max_size && cbor->max_size)
964 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1171 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
965 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
966 }
967
968 sv_utf8_downgrade (string, 0);
969 1173
970 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
971 dec.cur = (U8 *)SvPVX (string); 1175 dec.cur = (U8 *)data;
972 dec.end = (U8 *)SvEND (string); 1176 dec.end = (U8 *)data + len;
973 dec.err = 0;
974 dec.depth = 0;
975
976 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
977 ;//TODO dec.cbor.flags |= F_HOOK;
978 1177
979 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
980 1179
981 if (offset_return) 1180 if (offset_return)
982 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
985 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
986 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
987 1186
988 if (dec.err) 1187 if (dec.err)
989 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
990 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
991 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
992 } 1202 }
993 1203
994 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
995 1205
996 return sv; 1206 return sv;
997} 1207}
998 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
999///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
1000// XS interface functions 1330// XS interface functions
1001 1331
1002MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
1003 1333
1004BOOT: 1334BOOT:
1005{ 1335{
1006 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1336 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1007 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1337 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1008 1338
1009 cbor_true = get_bool ("CBOR::XS::true"); 1339 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1010 cbor_false = get_bool ("CBOR::XS::false"); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1341
1342 types_true = get_bool ("Types::Serialiser::true" );
1343 types_false = get_bool ("Types::Serialiser::false");
1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347
1348 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor);
1011} 1350}
1012 1351
1013PROTOTYPES: DISABLE 1352PROTOTYPES: DISABLE
1014 1353
1015void CLONE (...) 1354void CLONE (...)
1016 CODE: 1355 CODE:
1017 cbor_stash = 0; 1356 cbor_stash = 0;
1357 cbor_tagged_stash = 0;
1358 types_error_stash = 0;
1018 cbor_boolean_stash = 0; 1359 types_boolean_stash = 0;
1019 1360
1020void new (char *klass) 1361void new (char *klass)
1021 PPCODE: 1362 PPCODE:
1022{ 1363{
1023 SV *pv = NEWSV (0, sizeof (CBOR)); 1364 SV *pv = NEWSV (0, sizeof (CBOR));
1031 1372
1032void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
1033 ALIAS: 1374 ALIAS:
1034 shrink = F_SHRINK 1375 shrink = F_SHRINK
1035 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1381 PPCODE:
1037{ 1382{
1038 if (enable) 1383 if (enable)
1039 self->flags |= ix; 1384 self->flags |= ix;
1040 else 1385 else
1045 1390
1046void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
1047 ALIAS: 1392 ALIAS:
1048 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
1049 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
1050 PPCODE: 1399 PPCODE:
1051 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
1052 1401
1053void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1054 PPCODE: 1403 PPCODE:
1070 CODE: 1419 CODE:
1071 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
1072 OUTPUT: 1421 OUTPUT:
1073 RETVAL 1422 RETVAL
1074 1423
1075#if 0 //TODO 1424void filter (CBOR *self, SV *filter = 0)
1076
1077void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1078 PPCODE: 1425 PPCODE:
1079{
1080 SvREFCNT_dec (self->cb_object); 1426 SvREFCNT_dec (self->filter);
1081 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1427 self->filter = filter ? newSVsv (filter) : filter;
1082
1083 XPUSHs (ST (0)); 1428 XPUSHs (ST (0));
1084}
1085 1429
1086void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1430SV *get_filter (CBOR *self)
1087 PPCODE: 1431 CODE:
1088{ 1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1089 if (!self->cb_sk_object) 1433 OUTPUT:
1090 self->cb_sk_object = newHV (); 1434 RETVAL
1091
1092 if (SvOK (cb))
1093 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1094 else
1095 {
1096 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1097
1098 if (!HvKEYS (self->cb_sk_object))
1099 {
1100 SvREFCNT_dec (self->cb_sk_object);
1101 self->cb_sk_object = 0;
1102 }
1103 }
1104
1105 XPUSHs (ST (0));
1106}
1107
1108#endif
1109 1435
1110void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
1111 PPCODE: 1437 PPCODE:
1112 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1438 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1113 XPUSHs (scalar); 1439 XPUSHs (scalar);
1126 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1127 PUSHs (sv); 1453 PUSHs (sv);
1128 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1129} 1455}
1130 1456
1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
1508
1131void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
1132 CODE: 1510 PPCODE:
1133 SvREFCNT_dec (self->cb_sk_object); 1511 cbor_free (self);
1134 SvREFCNT_dec (self->cb_object);
1135 1512
1136PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1137 1514
1138void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1139 PPCODE: 1519 PPCODE:
1140{ 1520{
1141 CBOR cbor; 1521 CBOR cbor;
1142 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1143 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1144 XPUSHs (scalar); 1525 XPUSHs (scalar);
1145} 1526}
1146 1527
1147void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines