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 1.3 by root, Sat Oct 26 11:08:34 2013 UTC vs.
Revision 1.50 by root, Thu Feb 25 02:29:22 2016 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
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
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
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#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
16 106
17#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
18 108
19#define SB do { 109#define SB do {
20#define SE } while (0) 110#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 111
33#define IN_RANGE_INC(type,val,beg,end) \ 112#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 113 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 114 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 115
42#else 121#else
43# define CBOR_SLOW 0 122# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
45#endif 124#endif
46 125
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 128
50typedef struct { 129typedef struct {
51 U32 flags; 130 U32 flags;
52 U32 max_depth; 131 U32 max_depth;
53 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
54 134
55 SV *cb_object; 135 // for the incremental parser
56 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.
57} CBOR; 139} CBOR;
58 140
59INLINE void 141ecb_inline void
60cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
61{ 143{
62 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 145 cbor->max_depth = 512;
64} 146}
65 147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
153}
154
66///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
67// utility functions 156// utility functions
68 157
69INLINE SV * 158ecb_inline SV *
70get_bool (const char *name) 159get_bool (const char *name)
71{ 160{
72 SV *sv = get_sv (name, 1); 161 SV *sv = get_sv (name, 1);
73 162
74 SvREADONLY_on (sv); 163 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 164 SvREADONLY_on (SvRV (sv));
76 165
77 return sv; 166 return sv;
78} 167}
79 168
80INLINE void 169ecb_inline void
81shrink (SV *sv) 170shrink (SV *sv)
82{ 171{
83 sv_utf8_downgrade (sv, 1); 172 sv_utf8_downgrade (sv, 1);
84 173
85 if (SvLEN (sv) > SvCUR (sv) + 1) 174 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 180#endif
92 } 181 }
93} 182}
94 183
95///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
96// fp hell 185ecb_inline int
97 186minimum_string_length (UV idx)
98//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}
99 198
100///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
101// encoder 200// encoder
102 201
103// structure used for encoding CBOR 202// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
108 SV *sv; // result scalar 207 SV *sv; // result scalar
109 CBOR cbor; 208 CBOR cbor;
110 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;
111} enc_t; 214} enc_t;
112 215
113INLINE void 216ecb_inline void
114need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
115{ 218{
116 if (expect_false (enc->cur + len >= enc->end)) 219 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 220 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 223 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 225 }
123} 226}
124 227
125INLINE void 228ecb_inline void
126encode_ch (enc_t *enc, char ch) 229encode_ch (enc_t *enc, char ch)
127{ 230{
128 need (enc, 1); 231 need (enc, 1);
129 *enc->cur++ = ch; 232 *enc->cur++ = ch;
130} 233}
132static void 235static void
133encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
134{ 237{
135 need (enc, 9); 238 need (enc, 9);
136 239
137 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
139 else if (len < 0x100) 242 else if (ecb_expect_true (len <= 0xffU))
140 { 243 {
141 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 245 *enc->cur++ = len;
143 } 246 }
144 else if (len < 0x10000) 247 else if (len <= 0xffffU)
145 { 248 {
146 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 251 *enc->cur++ = len;
149 } 252 }
150 else if (len < 0x100000000) 253 else if (len <= 0xffffffffU)
151 { 254 {
152 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 259 *enc->cur++ = len;
157 } 260 }
158 else if (len) 261 else
159 { 262 {
160 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 271 *enc->cur++ = len;
169 } 272 }
170} 273}
171 274
172static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 283{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 285 need (enc, len);
177 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
178 enc->cur += len; 287 enc->cur += len;
179} 288}
180 289
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
181static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
182 316
183static void 317static void
184encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
185{ 319{
188 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
190 324
191 ++enc->depth; 325 ++enc->depth;
192 326
193 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 328
329 if (SvMAGICAL (av))
195 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
196 { 331 {
197 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
200 341
201 --enc->depth; 342 --enc->depth;
202} 343}
203 344
204static void 345static void
213 354
214 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
216 357
217 if (mg) 358 if (mg)
218 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 360 else
220 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
221 362
222 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
223 { 364 {
224 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
226 else 367 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 369
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 371 }
231 372
232 if (mg) 373 if (mg)
233 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 375
235 --enc->depth; 376 --enc->depth;
236} 377}
237 378
238// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
239static void 380static void
240encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
241{ 382{
242 svtype svt;
243
244 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
245 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
246 386
247 if (expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
248 { 388 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 390 ? types_boolean_stash
391 : gv_stashpv ("Types::Serialiser::Boolean", 1);
392 HV *error_stash = !CBOR_SLOW || types_error_stash
393 ? types_error_stash
394 : gv_stashpv ("Types::Serialiser::Error", 1);
395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
396 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 398
253 if (SvSTASH (sv) == stash) 399 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 400
401 if (stash == boolean_stash)
402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
406 else if (stash == error_stash)
407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
411 else if (stash == tagged_stash)
412 {
413 if (svt != SVt_PVAV)
414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
415
416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
255 else 437 else
256 { 438 {
257#if 0 //TODO 439 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 451 {
452 dSP;
453
454 ENTER; SAVETMPS;
455 PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 456 // we re-bless the reference to get overload and other niceties right
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 458
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 459 PUTBACK;
460 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 461 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 462 SPAGAIN;
274 463
275 // catch this surprisingly common error 464 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 465 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 466 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 467
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 468 encode_sv (enc, POPs);
283 469
470 PUTBACK;
471
284 FREETMPS; LEAVE; 472 FREETMPS; LEAVE;
285 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED)
287 encode_str (enc, "null", 4, 0);
288 else
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
290 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291 } 473 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 475 {
476 dSP;
477
478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
481 EXTEND (SP, 2);
482 // we re-bless the reference to get overload and other niceties right
483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
484 PUSHs (sv_cbor);
485
486 PUTBACK;
487 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
488 SPAGAIN;
489
490 // catch this surprisingly common error
491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
493
494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
495 encode_uint (enc, MAJOR_ARRAY, count + 1);
496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
497
498 while (count)
499 encode_sv (enc, SP[1 - count--]);
500
501 PUTBACK;
502
503 FREETMPS; LEAVE;
504 }
294 else 505 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 506 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 507 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 508 }
300 else if (svt == SVt_PVHV) 509 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 510 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 511 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 512 encode_av (enc, (AV *)sv);
304 else if (svt < SVt_PVAV)
305 {
306 STRLEN len = 0;
307 char *pv = svt ? SvPV (sv, len) : 0;
308
309 if (len == 1 && *pv == '1')
310 encode_ch (enc, 0xe0 | 21);
311 else if (len == 1 && *pv == '0')
312 encode_ch (enc, 0xe0 | 20);
313 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314 encode_ch (enc, 0xe0 | 23);
315 else
316 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318 }
319 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320 encode_ch (enc, 0xe0 | 23);
321 else 513 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 514 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 515 encode_tag (enc, CBOR_TAG_INDIRECTION);
516 encode_sv (enc, sv);
517 }
324} 518}
325 519
326static void 520static void
327encode_nv (enc_t *enc, SV *sv) 521encode_nv (enc_t *enc, SV *sv)
328{ 522{
329 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
330 524
331 need (enc, 9); 525 need (enc, 9);
332 526
333 if (expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 529 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
337 { 531 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
339 533
340 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341 535
342 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
344 538
345 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
347 } 541 }
348 else 542 else
349 { 543 {
350 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
351 545
352 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353 547
354 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
356 550
357 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
366 560
367 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
368 { 562 {
369 STRLEN len; 563 STRLEN len;
370 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
372 } 566 }
373 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 568 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
376 { 570 {
377 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 575 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 577 }
384 else if (SvROK (sv)) 578 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 584 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 585 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 587}
394 588
395static SV * 589static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
397{ 591{
398 enc_t enc; 592 enc_t enc = { 0 };
399 593
400 enc.cbor = *cbor; 594 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 598
406 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
600
601 if (cbor->flags & F_PACK_STRINGS)
602 {
603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
606 }
607
407 encode_sv (&enc, scalar); 608 encode_sv (&enc, scalar);
408 609
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 610 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 611 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 612
425 U8 *end; // end of input string 626 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 627 const char *err; // parse error, if != 0
427 CBOR cbor; 628 CBOR cbor;
428 U32 depth; // recursion depth 629 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 630 U32 maxdepth; // recursion depth limit
631 AV *shareable;
632 AV *stringref;
633 SV *decode_tagged;
430} dec_t; 634} dec_t;
431 635
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 636#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
433 637
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 638#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
435 639
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 640#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 641#define DEC_DEC_DEPTH --dec->depth
438 642
439static UV 643static UV
440decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
441{ 645{
442 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
443 { 647 ++dec->cur;
444 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447 return *dec->cur++ & 31;
448 648
449 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
450 WANT (2); 653 WANT (1);
451 dec->cur += 2; 654 dec->cur += 1;
452 return dec->cur[-1]; 655 return dec->cur[-1];
453 656 }
454 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
455 WANT (3); 659 WANT (2);
456 dec->cur += 3; 660 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
459 663 }
460 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
461 WANT (5); 666 WANT (4);
462 dec->cur += 5; 667 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
467 672 }
468 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
469 WANT (9); 675 WANT (8);
470 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
471 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
475 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
479 691 }
480 default: 692 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 694
484fail: 695fail:
485 return 0; 696 return 0;
486} 697}
487 698
492{ 703{
493 AV *av = newAV (); 704 AV *av = newAV ();
494 705
495 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
496 707
497 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 709 {
499 ++dec->cur; 710 ++dec->cur;
500 711
501 for (;;) 712 for (;;)
502 { 713 {
503 WANT (1); 714 WANT (1);
504 715
505 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506 { 717 {
507 ++dec->cur; 718 ++dec->cur;
508 break; 719 break;
509 } 720 }
510 721
513 } 724 }
514 else 725 else
515 { 726 {
516 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
517 728
729 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518 av_fill (av, len - 1); 730 av_fill (av, len - 1);
519 731
520 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
522 } 734 }
528 SvREFCNT_dec (av); 740 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 741 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 742 return &PL_sv_undef;
531} 743}
532 744
745static void
746decode_he (dec_t *dec, HV *hv)
747{
748 // for speed reasons, we specialcase single-string
749 // byte or utf-8 strings as keys, but only when !stringref
750
751 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 WANT (len);
758 dec->cur += len;
759
760 hv_store (hv, key, len, decode_sv (dec), 0);
761
762 return;
763 }
764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
765 {
766 I32 len = decode_uint (dec);
767 char *key = (char *)dec->cur;
768
769 WANT (len);
770 dec->cur += len;
771
772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
775
776 hv_store (hv, key, -len, decode_sv (dec), 0);
777
778 return;
779 }
780
781 SV *k = decode_sv (dec);
782 SV *v = decode_sv (dec);
783
784 hv_store_ent (hv, k, v, 0);
785 SvREFCNT_dec (k);
786
787fail:
788 ;
789}
790
533static SV * 791static SV *
534decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
535{ 793{
536 HV *hv = newHV (); 794 HV *hv = newHV ();
537 795
538 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
539 797
540 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 799 {
542 ++dec->cur; 800 ++dec->cur;
543 801
544 for (;;) 802 for (;;)
545 { 803 {
546 WANT (1); 804 WANT (1);
547 805
548 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549 { 807 {
550 ++dec->cur; 808 ++dec->cur;
551 break; 809 break;
552 } 810 }
553 811
554 SV *k = decode_sv (dec); 812 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 813 }
559 } 814 }
560 else 815 else
561 { 816 {
562 int len = decode_uint (dec); 817 int pairs = decode_uint (dec);
563 818
564 while (len--) 819 while (pairs--)
565 { 820 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 821 }
572 822
573 DEC_DEC_DEPTH; 823 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 824 return newRV_noinc ((SV *)hv);
575
576#if 0
577 SV *sv;
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH;
581 decode_ws (dec);
582
583 for (;;)
584 {
585 // heuristic: assume that
586 // a) decode_str + hv_store_ent are abysmally slow.
587 // b) most hash keys are short, simple ascii text.
588 // => try to "fast-match" such strings to avoid
589 // the overhead of decode_str + hv_store_ent.
590 {
591 SV *value;
592 char *p = dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594
595 for (;;)
596 {
597 // the >= 0x80 is false on most architectures
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599 {
600 // slow path, back up and use decode_str
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break;
619 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 {
685 dSP;
686 int count;
687
688 ENTER; SAVETMPS; PUSHMARK (SP);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730 825
731fail: 826fail:
732 SvREFCNT_dec (hv); 827 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 828 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 829 return &PL_sv_undef;
735} 830}
736 831
737static SV * 832static SV *
738decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
739{ 834{
740 SV *sv; 835 SV *sv = 0;
741 836
742 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743 { 838 {
839 // indefinite length strings
744 ++dec->cur; 840 ++dec->cur;
745 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
746 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
747 845
748 // not very fast, and certainly not robust against illegal input
749 for (;;) 846 for (;;)
750 { 847 {
751 WANT (1); 848 WANT (1);
752 849
753 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754 { 852 {
755 ++dec->cur; 853 ++dec->cur;
756 break; 854 break;
757 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758 858
759 SV *sv2 = decode_sv (dec); 859 STRLEN len = decode_uint (dec);
760 sv_catsv (sv, sv2); 860
861 WANT (len);
862 sv_catpvn (sv, dec->cur, len);
863 dec->cur += len;
761 } 864 }
762 } 865 }
763 else 866 else
764 { 867 {
765 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
766 869
767 WANT (len); 870 WANT (len);
768 sv = newSVpvn (dec->cur, len); 871 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 872 dec->cur += len;
873
874 if (ecb_expect_false (dec->stringref)
875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 877 }
771 878
772 if (utf8) 879 if (utf8)
880 {
881 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
882 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
883 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
884
773 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
774 887
775 return sv; 888 return sv;
776 889
777fail: 890fail:
891 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 892 return &PL_sv_undef;
779} 893}
780 894
781static SV * 895static SV *
782decode_tagged (dec_t *dec) 896decode_tagged (dec_t *dec)
783{ 897{
898 SV *sv = 0;
784 UV tag = decode_uint (dec); 899 UV tag = decode_uint (dec);
900
901 WANT (1);
902
903 switch (tag)
904 {
905 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 906 sv = decode_sv (dec);
907 break;
786 908
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 909 case CBOR_TAG_INDIRECTION:
910 sv = newRV_noinc (decode_sv (dec));
911 break;
912
913 case CBOR_TAG_STRINGREF_NAMESPACE:
914 {
915 ENTER; SAVETMPS;
916
917 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919
920 sv = decode_sv (dec);
921
922 FREETMPS; LEAVE;
923 }
924 break;
925
926 case CBOR_TAG_STRINGREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
934 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
935
936 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_VALUE_SHAREABLE:
941 {
942 if (ecb_expect_false (!dec->shareable))
943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
947 sv = newSV (0);
948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
949
950 SV *osv = decode_sv (dec);
951 sv_setsv (sv, osv);
952 SvREFCNT_dec_NN (osv);
953 }
954 else
955 {
956 av_push (dec->shareable, &PL_sv_undef);
957 int idx = AvFILLp (dec->shareable);
958 sv = decode_sv (dec);
959 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
960 }
961 }
962 break;
963
964 case CBOR_TAG_VALUE_SHAREDREF:
965 {
966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
968
969 UV idx = decode_uint (dec);
970
971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
973
974 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
975
976 if (sv == &PL_sv_undef)
977 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
978 }
979 break;
980
981 case CBOR_TAG_PERL_OBJECT:
982 {
983 sv = decode_sv (dec);
984
985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
986 ERR ("corrupted CBOR data (non-array perl object)");
987
988 AV *av = (AV *)SvRV (sv);
989 int len = av_len (av) + 1;
990 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
991
992 if (!stash)
993 ERR ("cannot decode perl-object (package does not exist)");
994
995 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
996
997 if (!method)
998 ERR ("cannot decode perl-object (package does not have a THAW method)");
999
1000 dSP;
1001
1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
1004 EXTEND (SP, len + 1);
1005 // we re-bless the reference to get overload and other niceties right
1006 PUSHs (*av_fetch (av, 0, 1));
1007 PUSHs (sv_cbor);
1008
1009 int i;
1010
1011 for (i = 1; i < len; ++i)
1012 PUSHs (*av_fetch (av, i, 1));
1013
1014 PUTBACK;
1015 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1016 SPAGAIN;
1017
1018 if (SvTRUE (ERRSV))
1019 {
1020 FREETMPS; LEAVE;
1021 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1022 }
1023
1024 SvREFCNT_dec (sv);
1025 sv = SvREFCNT_inc (POPs);
1026
1027 PUTBACK;
1028
1029 FREETMPS; LEAVE;
1030 }
1031 break;
1032
1033 default:
1034 {
1035 sv = decode_sv (dec);
1036
1037 dSP;
1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
1041 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag));
1043 PUSHs (sv);
1044
1045 PUTBACK;
1046 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1047 SPAGAIN;
1048
1049 if (SvTRUE (ERRSV))
1050 {
1051 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1053 }
1054
1055 if (count)
1056 {
1057 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs);
1059 }
1060 else
1061 {
1062 AV *av = newAV ();
1063 av_push (av, newSVuv (tag));
1064 av_push (av, sv);
1065
1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1067 ? cbor_tagged_stash
1068 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1069 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1070 }
1071
1072 PUTBACK;
1073
1074 FREETMPS; LEAVE;
1075 }
1076 break;
1077 }
1078
788 return sv; 1079 return sv;
789 1080
790 AV *av = newAV (); 1081fail:
791 av_push (av, newSVuv (tag)); 1082 SvREFCNT_dec (sv);
792 av_push (av, sv); 1083 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1084}
795 1085
796static SV * 1086static SV *
797decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
798{ 1088{
799 WANT (1); 1089 WANT (1);
800 1090
801 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1092 {
803 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1100
811 return decode_str (dec, 1); 1101 case MAJOR_MISC >> MAJOR_SHIFT:
812 case 4: // array
813 return decode_av (dec);
814 case 5: // map
815 return decode_hv (dec);
816 case 6: // tag
817 return decode_tagged (dec);
818 case 7: // misc
819 switch (*dec->cur++ & 31) 1102 switch (*dec->cur++ & MINOR_MASK)
820 { 1103 {
821 case 20: 1104 case SIMPLE_FALSE:
822#if CBOR_SLOW 1105#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
824#endif 1107#endif
825 return newSVsv (cbor_false); 1108 return newSVsv (types_false);
826 case 21: 1109 case SIMPLE_TRUE:
827#if CBOR_SLOW 1110#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
829#endif 1112#endif
830 return newSVsv (cbor_true); 1113 return newSVsv (types_true);
831 case 22: 1114 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
1116 case SIMPLE_UNDEF:
1117#if CBOR_SLOW
1118 types_error = get_bool ("Types::Serialiser::error");
1119#endif
1120 return newSVsv (types_error);
833 1121
834 case 25: 1122 case MISC_FLOAT16:
835 { 1123 {
836 WANT (2); 1124 WANT (2);
837 1125
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1127 dec->cur += 2;
840 1128
841 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
842 } 1130 }
843 1131
844 case 26: 1132 case MISC_FLOAT32:
845 { 1133 {
846 uint32_t fp; 1134 uint32_t fp;
847 WANT (4); 1135 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1137 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
853 1141
854 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
855 } 1143 }
856 1144
857 case 27: 1145 case MISC_FLOAT64:
858 { 1146 {
859 uint64_t fp; 1147 uint64_t fp;
860 WANT (8); 1148 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1150 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
866 1154
867 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
868 } 1156 }
869 1157
870 // 0..19 unassigned 1158 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1159 // 24 reserved + unassigned simple (reserved values are not encodable)
1160 // 28-30 unassigned misc
1161 // 31 break code
872 default: 1162 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1164 }
875 1165
876 break; 1166 break;
877 } 1167 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 1168
935fail: 1169fail:
936 return &PL_sv_undef; 1170 return &PL_sv_undef;
937} 1171}
938 1172
939static SV * 1173static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1175{
942 dec_t dec; 1176 dec_t dec = { 0 };
943 SV *sv; 1177 SV *sv;
1178 STRLEN len;
1179 char *data = SvPVbyte (string, len);
944 1180
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1182 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1183 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1184
980 dec.cbor = *cbor; 1185 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1186 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1187 dec.end = (U8 *)data + len;
983 dec.err = 0;
984 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 1188
989 sv = decode_sv (&dec); 1189 sv = decode_sv (&dec);
990 1190
991 if (offset_return) 1191 if (offset_return)
992 *offset_return = dec.cur; 1192 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
997 1197
998 if (dec.err) 1198 if (dec.err)
999 { 1199 {
1200 if (dec.shareable)
1201 {
1202 // need to break cyclic links, which whould all be in shareable
1203 int i;
1204 SV **svp;
1205
1206 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef);
1209 }
1210
1000 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1212 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1213 }
1003 1214
1004 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
1005 1216
1006 return sv; 1217 return sv;
1007} 1218}
1008 1219
1220/////////////////////////////////////////////////////////////////////////////
1221// incremental parser
1222
1223#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1224
1225// returns 0 for notyet, 1 for success or error
1226static int
1227incr_parse (CBOR *self, SV *cborstr)
1228{
1229 STRLEN cur;
1230 SvPV (cborstr, cur);
1231
1232 while (ecb_expect_true (self->incr_need <= cur))
1233 {
1234 // table of integer count bytes
1235 static I8 incr_len[MINOR_MASK + 1] = {
1236 0, 0, 0, 0, 0, 0, 0, 0,
1237 0, 0, 0, 0, 0, 0, 0, 0,
1238 0, 0, 0, 0, 0, 0, 0, 0,
1239 1, 2, 4, 8,-1,-1,-1,-2
1240 };
1241
1242 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1243 U8 m = *p & MINOR_MASK;
1244 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1245 I8 ilen = incr_len[m];
1246
1247 self->incr_need = self->incr_pos + 1;
1248
1249 if (ecb_expect_false (ilen < 0))
1250 {
1251 if (m != MINOR_INDEF)
1252 return 1; // error
1253
1254 if (*p == (MAJOR_MISC | MINOR_INDEF))
1255 {
1256 if (count >= 0)
1257 return 1; // error
1258
1259 count = 1;
1260 }
1261 else
1262 {
1263 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1264 count = -1;
1265 }
1266 }
1267 else
1268 {
1269 self->incr_need += ilen;
1270 if (ecb_expect_false (self->incr_need > cur))
1271 return 0;
1272
1273 int major = *p >> MAJOR_SHIFT;
1274
1275 switch (major)
1276 {
1277 case MAJOR_TAG >> MAJOR_SHIFT:
1278 ++count; // tags merely prefix another value
1279 break;
1280
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 case MAJOR_ARRAY >> MAJOR_SHIFT:
1284 case MAJOR_MAP >> MAJOR_SHIFT:
1285 {
1286 UV len;
1287
1288 if (ecb_expect_false (ilen))
1289 {
1290 len = 0;
1291
1292 do {
1293 len = (len << 8) | *++p;
1294 } while (--ilen);
1295 }
1296 else
1297 len = m;
1298
1299 switch (major)
1300 {
1301 case MAJOR_BYTES >> MAJOR_SHIFT:
1302 case MAJOR_TEXT >> MAJOR_SHIFT:
1303 self->incr_need += len;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 break;
1308
1309 case MAJOR_MAP >> MAJOR_SHIFT:
1310 len <<= 1;
1311 case MAJOR_ARRAY >> MAJOR_SHIFT:
1312 if (len)
1313 {
1314 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1315 count = len + 1;
1316 }
1317 break;
1318 }
1319 }
1320 }
1321 }
1322
1323 self->incr_pos = self->incr_need;
1324
1325 if (count > 0)
1326 {
1327 while (!--count)
1328 {
1329 if (!AvFILLp (self->incr_count))
1330 return 1; // done
1331
1332 SvREFCNT_dec_NN (av_pop (self->incr_count));
1333 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1334 }
1335
1336 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1337 }
1338 }
1339
1340 return 0;
1341}
1342
1343
1009///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1345// XS interface functions
1011 1346
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1348
1014BOOT: 1349BOOT:
1015{ 1350{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1351 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1352 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1353
1019 cbor_true = get_bool ("CBOR::XS::true"); 1354 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1355 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1356
1357 types_true = get_bool ("Types::Serialiser::true" );
1358 types_false = get_bool ("Types::Serialiser::false");
1359 types_error = get_bool ("Types::Serialiser::error");
1360
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1362
1363 sv_cbor = newSVpv ("CBOR", 0);
1364 SvREADONLY_on (sv_cbor);
1021} 1365}
1022 1366
1023PROTOTYPES: DISABLE 1367PROTOTYPES: DISABLE
1024 1368
1025void CLONE (...) 1369void CLONE (...)
1026 CODE: 1370 CODE:
1027 cbor_stash = 0; 1371 cbor_stash = 0;
1372 cbor_tagged_stash = 0;
1373 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1374 types_boolean_stash = 0;
1029 1375
1030void new (char *klass) 1376void new (char *klass)
1031 PPCODE: 1377 PPCODE:
1032{ 1378{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1379 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1387
1042void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1389 ALIAS:
1044 shrink = F_SHRINK 1390 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1392 allow_sharing = F_ALLOW_SHARING
1393 allow_cycles = F_ALLOW_CYCLES
1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1396 PPCODE:
1047{ 1397{
1048 if (enable) 1398 if (enable)
1049 self->flags |= ix; 1399 self->flags |= ix;
1050 else 1400 else
1055 1405
1056void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
1057 ALIAS: 1407 ALIAS:
1058 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1410 get_allow_sharing = F_ALLOW_SHARING
1411 get_allow_cycles = F_ALLOW_CYCLES
1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1414 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
1062 1416
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1418 PPCODE:
1080 CODE: 1434 CODE:
1081 RETVAL = self->max_size; 1435 RETVAL = self->max_size;
1082 OUTPUT: 1436 OUTPUT:
1083 RETVAL 1437 RETVAL
1084 1438
1085#if 0 //TODO 1439void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1440 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1441 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1442 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1443 XPUSHs (ST (0));
1094}
1095 1444
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1445SV *get_filter (CBOR *self)
1097 PPCODE: 1446 CODE:
1098{ 1447 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1448 OUTPUT:
1100 self->cb_sk_object = newHV (); 1449 RETVAL
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119 1450
1120void encode (CBOR *self, SV *scalar) 1451void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1452 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1453 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1454 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
1137 PUSHs (sv); 1468 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1470}
1140 1471
1472void incr_parse (CBOR *self, SV *cborstr)
1473 ALIAS:
1474 incr_parse_multiple = 1
1475 PPCODE:
1476{
1477 if (SvUTF8 (cborstr))
1478 sv_utf8_downgrade (cborstr, 0);
1479
1480 if (!self->incr_count)
1481 {
1482 self->incr_count = newAV ();
1483 self->incr_pos = 0;
1484 self->incr_need = 1;
1485
1486 av_push (self->incr_count, newSViv (1));
1487 }
1488
1489 do
1490 {
1491 if (!incr_parse (self, cborstr))
1492 {
1493 if (self->incr_need > self->max_size && self->max_size)
1494 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1495 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1496
1497 break;
1498 }
1499
1500 SV *sv;
1501 char *offset;
1502
1503 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1504 XPUSHs (sv);
1505
1506 sv_chop (cborstr, offset);
1507
1508 av_clear (self->incr_count);
1509 av_push (self->incr_count, newSViv (1));
1510
1511 self->incr_pos = 0;
1512 self->incr_need = self->incr_pos + 1;
1513 }
1514 while (ix);
1515}
1516
1517void incr_reset (CBOR *self)
1518 CODE:
1519{
1520 SvREFCNT_dec (self->incr_count);
1521 self->incr_count = 0;
1522}
1523
1141void DESTROY (CBOR *self) 1524void DESTROY (CBOR *self)
1142 CODE: 1525 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1526 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1527
1146PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
1147 1529
1148void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1534 PPCODE:
1150{ 1535{
1151 CBOR cbor; 1536 CBOR cbor;
1152 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1540 XPUSHs (scalar);
1155} 1541}
1156 1542
1157void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines