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 1.1 by root, Fri Oct 25 23:09:45 2013 UTC vs.
Revision 1.64 by root, Tue Jun 27 02:03:24 2017 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14#if defined(__BORLANDC__) || defined(_MSC_VER) 16// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
17 32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
103
18#define F_SHRINK 0x00000200UL 104#define F_SHRINK 0x00000001UL
19#define F_ALLOW_UNKNOWN 0x00002000UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
106#define F_ALLOW_SHARING 0x00000004UL
107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
111#define F_TEXT_STRINGS 0x00000080UL
112#define F_VALIDATE_UTF8 0x00000100UL
20 113
21#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
22 115
23#define SB do { 116#define SB do {
24#define SE } while (0) 117#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)
36 118
37#define IN_RANGE_INC(type,val,beg,end) \ 119#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 120 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 121 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 122
46#else 128#else
47# define CBOR_SLOW 0 129# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 130# define CBOR_STASH cbor_stash
49#endif 131#endif
50 132
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 133static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 134static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
53 135
54typedef struct { 136typedef struct {
55 U32 flags; 137 U32 flags;
56 U32 max_depth; 138 U32 max_depth;
57 STRLEN max_size; 139 STRLEN max_size;
140 SV *filter;
58 141
59 SV *cb_object; 142 // for the incremental parser
60 HV *cb_sk_object; 143 STRLEN incr_pos; // the current offset into the text
144 STRLEN incr_need; // minimum bytes needed to decode
145 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
61} CBOR; 146} CBOR;
62 147
63INLINE void 148ecb_inline void
64cbor_init (CBOR *cbor) 149cbor_init (CBOR *cbor)
65{ 150{
66 Zero (cbor, 1, CBOR); 151 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 152 cbor->max_depth = 512;
68} 153}
69 154
155ecb_inline void
156cbor_free (CBOR *cbor)
157{
158 SvREFCNT_dec (cbor->filter);
159 SvREFCNT_dec (cbor->incr_count);
160}
161
70///////////////////////////////////////////////////////////////////////////// 162/////////////////////////////////////////////////////////////////////////////
71// utility functions 163// utility functions
72 164
73INLINE SV * 165ecb_inline SV *
74get_bool (const char *name) 166get_bool (const char *name)
75{ 167{
76 SV *sv = get_sv (name, 1); 168 SV *sv = get_sv (name, 1);
77 169
78 SvREADONLY_on (sv); 170 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 171 SvREADONLY_on (SvRV (sv));
80 172
81 return sv; 173 return sv;
82} 174}
83 175
84INLINE void 176ecb_inline void
85shrink (SV *sv) 177shrink (SV *sv)
86{ 178{
87 sv_utf8_downgrade (sv, 1); 179 sv_utf8_downgrade (sv, 1);
88 180
89 if (SvLEN (sv) > SvCUR (sv) + 1) 181 if (SvLEN (sv) > SvCUR (sv) + 1)
94 SvPV_renew (sv, SvCUR (sv) + 1); 186 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 187#endif
96 } 188 }
97} 189}
98 190
99///////////////////////////////////////////////////////////////////////////// 191// minimum length of a string to be registered for stringref
100// fp hell 192ecb_inline int
101 193minimum_string_length (UV idx)
102//TODO 194{
195 return idx <= 23 ? 3
196 : idx <= 0xffU ? 4
197 : idx <= 0xffffU ? 5
198 : idx <= 0xffffffffU ? 7
199 : 11;
200}
103 201
104///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
105// encoder 203// encoder
106 204
107// structure used for encoding CBOR 205// structure used for encoding CBOR
110 char *cur; // SvPVX (sv) + current output position 208 char *cur; // SvPVX (sv) + current output position
111 char *end; // SvEND (sv) 209 char *end; // SvEND (sv)
112 SV *sv; // result scalar 210 SV *sv; // result scalar
113 CBOR cbor; 211 CBOR cbor;
114 U32 depth; // recursion level 212 U32 depth; // recursion level
213 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
214 UV stringref_idx;
215 HV *shareable; // ptr => index, or 0
216 UV shareable_idx;
115} enc_t; 217} enc_t;
116 218
117INLINE void 219ecb_inline void
118need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
119{ 221{
120 if (expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
121 { 223 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 228 }
127} 229}
128 230
129INLINE void 231ecb_inline void
130encode_ch (enc_t *enc, char ch) 232encode_ch (enc_t *enc, char ch)
131{ 233{
132 need (enc, 1); 234 need (enc, 1);
133 *enc->cur++ = ch; 235 *enc->cur++ = ch;
134} 236}
136static void 238static void
137encode_uint (enc_t *enc, int major, UV len) 239encode_uint (enc_t *enc, int major, UV len)
138{ 240{
139 need (enc, 9); 241 need (enc, 9);
140 242
141 if (len < 24) 243 if (ecb_expect_true (len < LENGTH_EXT1))
142 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
143 else if (len < 0x100) 245 else if (ecb_expect_true (len <= 0xffU))
144 { 246 {
145 *enc->cur++ = major | 24; 247 *enc->cur++ = major | LENGTH_EXT1;
146 *enc->cur++ = len; 248 *enc->cur++ = len;
147 } 249 }
148 else if (len < 0x10000) 250 else if (len <= 0xffffU)
149 { 251 {
150 *enc->cur++ = major | 25; 252 *enc->cur++ = major | LENGTH_EXT2;
151 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 254 *enc->cur++ = len;
153 } 255 }
154 else if (len < 0x100000000) 256 else if (len <= 0xffffffffU)
155 { 257 {
156 *enc->cur++ = major | 26; 258 *enc->cur++ = major | LENGTH_EXT4;
157 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 262 *enc->cur++ = len;
161 } 263 }
162 else if (len) 264 else
163 { 265 {
164 *enc->cur++ = major | 27; 266 *enc->cur++ = major | LENGTH_EXT8;
165 *enc->cur++ = len >> 56; 267 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 268 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 269 *enc->cur++ = len >> 40;
168 *enc->cur++ = len >> 32; 270 *enc->cur++ = len >> 32;
169 *enc->cur++ = len >> 24; 271 *enc->cur++ = len >> 24;
171 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
172 *enc->cur++ = len; 274 *enc->cur++ = len;
173 } 275 }
174} 276}
175 277
176static void 278ecb_inline void
279encode_tag (enc_t *enc, UV tag)
280{
281 encode_uint (enc, MAJOR_TAG, tag);
282}
283
284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
177encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
178{ 287{
179 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
307ecb_inline void
308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
180 need (enc, len); 318 need (enc, len);
181 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
182 enc->cur += len; 320 enc->cur += len;
183} 321}
184 322
323ecb_inline void
324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
325{
326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
327 {
328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
329
330 if (SvOK (*svp))
331 {
332 // already registered, use stringref
333 encode_tag (enc, CBOR_TAG_STRINGREF);
334 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
335 return;
336 }
337 else if (len >= minimum_string_length (enc->stringref_idx))
338 {
339 // register only
340 sv_setuv (*svp, enc->stringref_idx);
341 ++enc->stringref_idx;
342 }
343 }
344
345 encode_str (enc, upgrade_utf8, utf8, str, len);
346}
347
185static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
186 349
187static void 350static void
188encode_av (enc_t *enc, AV *av) 351encode_av (enc_t *enc, AV *av)
189{ 352{
192 if (enc->depth >= enc->cbor.max_depth) 355 if (enc->depth >= enc->cbor.max_depth)
193 croak (ERR_NESTING_EXCEEDED); 356 croak (ERR_NESTING_EXCEEDED);
194 357
195 ++enc->depth; 358 ++enc->depth;
196 359
197 encode_uint (enc, 0x80, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
198 361
362 if (ecb_expect_false (SvMAGICAL (av)))
199 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
200 { 364 {
201 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
202 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
203 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
204 374
205 --enc->depth; 375 --enc->depth;
206} 376}
207 377
208static void 378static void
216 ++enc->depth; 386 ++enc->depth;
217 387
218 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
219 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
220 390
221 if (mg) 391 if (ecb_expect_false (mg))
222 encode_ch (enc, 0xa0 | 31); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
223 else 393 else
224 encode_uint (enc, 0xa0, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
225 395
226 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
227 { 397 {
228 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
230 else 400 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 402
233 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
234 } 404 }
235 405
236 if (mg) 406 if (ecb_expect_false (mg))
237 encode_ch (enc, 0xe0 | 31); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
238 408
239 --enc->depth; 409 --enc->depth;
240} 410}
241 411
242// encode objects, arrays and special \0=false and \1=true values. 412// encode objects, arrays and special \0=false and \1=true values.
243static void 413static void
244encode_rv (enc_t *enc, SV *sv) 414encode_rv (enc_t *enc, SV *sv)
245{ 415{
246 svtype svt;
247
248 SvGETMAGIC (sv); 416 SvGETMAGIC (sv);
417
249 svt = SvTYPE (sv); 418 svtype svt = SvTYPE (sv);
250 419
251 if (expect_false (SvOBJECT (sv))) 420 if (ecb_expect_false (SvOBJECT (sv)))
252 { 421 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 422 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? cbor_boolean_stash 423 ? types_boolean_stash
424 : gv_stashpv ("Types::Serialiser::Boolean", 1);
425 HV *error_stash = !CBOR_SLOW || types_error_stash
426 ? types_error_stash
427 : gv_stashpv ("Types::Serialiser::Error", 1);
428 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
429 ? cbor_tagged_stash
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 430 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 431
257 if (SvSTASH (sv) == stash) 432 HV *stash = SvSTASH (sv);
258 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 433
434 if (stash == boolean_stash)
435 {
436 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
437 return;
438 }
439 else if (stash == error_stash)
440 {
441 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
442 return;
443 }
444 else if (stash == tagged_stash)
445 {
446 if (svt != SVt_PVAV)
447 croak ("encountered CBOR::XS::Tagged object that isn't an array");
448
449 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
450 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
451
452 return;
453 }
454 }
455
456 if (ecb_expect_false (SvREFCNT (sv) > 1)
457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
458 {
459 if (ecb_expect_false (!enc->shareable))
460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
461
462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
463
464 if (SvOK (*svp))
465 {
466 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
467 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
468 return;
469 }
259 else 470 else
260 { 471 {
261#if 0 //TODO 472 sv_setuv (*svp, enc->shareable_idx);
262 if (enc->cbor.flags & F_CONV_BLESSED) 473 ++enc->shareable_idx;
474 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
475 }
476 }
477
478 if (ecb_expect_false (SvOBJECT (sv)))
479 {
480 HV *stash = SvSTASH (sv);
481 GV *method;
482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
263 { 487 {
488 dSP;
489
490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
264 // we re-bless the reference to get overload and other niceties right 492 // 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))); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
273 494
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK; 495 PUTBACK;
496 // G_SCALAR ensures that return value is 1
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 497 call_sv ((SV *)GvCV (method), G_SCALAR);
277 SPAGAIN; 498 SPAGAIN;
278 499
279 // catch this surprisingly common error 500 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv) 501 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))); 502 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
282 503
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv); 504 encode_sv (enc, POPs);
287 505
506 PUTBACK;
507
288 FREETMPS; LEAVE; 508 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 } 509 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
297 encode_str (enc, "null", 4, 0); 511 {
512 dSP;
513
514 ENTER; SAVETMPS;
515 SAVESTACK_POS ();
516 PUSHMARK (SP);
517 EXTEND (SP, 2);
518 // we re-bless the reference to get overload and other niceties right
519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
520 PUSHs (sv_cbor);
521
522 PUTBACK;
523 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
524 SPAGAIN;
525
526 // catch this surprisingly common error
527 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
529
530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
531 encode_uint (enc, MAJOR_ARRAY, count + 1);
532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
533
534 while (count)
535 encode_sv (enc, SP[1 - count--]);
536
537 PUTBACK;
538
539 FREETMPS; LEAVE;
540 }
298 else 541 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 542 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 544 }
304 else if (svt == SVt_PVHV) 545 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 546 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 547 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 548 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 549 else
326 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 550 {
327 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 551 encode_tag (enc, CBOR_TAG_INDIRECTION);
552 encode_sv (enc, sv);
553 }
328} 554}
329 555
330static void 556static void
331encode_nv (enc_t *enc, SV *sv) 557encode_nv (enc_t *enc, SV *sv)
332{ 558{
333 double nv = SvNVX (sv); 559 double nv = SvNVX (sv);
334 560
335 need (enc, 9); 561 need (enc, 9);
336 562
337 if (expect_false (nv == (U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
339 //TODO: maybe I32? 565 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
341 { 567 {
568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
569
342 uint32_t fp = ecb_float_to_binary32 (nv); 570 uint32_t fp = ecb_float_to_binary32 (nv);
343
344 *enc->cur++ = 0xe0 | 26;
345 571
346 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
347 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
348 574
349 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
350 enc->cur += 4; 576 enc->cur += 4;
351 } 577 }
352 else 578 else
353 { 579 {
580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
581
354 uint64_t fp = ecb_double_to_binary64 (nv); 582 uint64_t fp = ecb_double_to_binary64 (nv);
355
356 *enc->cur++ = 0xe0 | 27;
357 583
358 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
359 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
360 586
361 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
370 596
371 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
372 { 598 {
373 STRLEN len; 599 STRLEN len;
374 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
375 encode_str (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
376 } 602 }
377 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
378 encode_nv (enc, sv); 604 encode_nv (enc, sv);
379 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
380 { 606 {
381 if (SvIsUV (sv)) 607 if (SvIsUV (sv))
382 encode_uint (enc, 0x00, SvUVX (sv)); 608 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
383 else if (SvIVX (sv) >= 0) 609 else if (SvIVX (sv) >= 0)
384 encode_uint (enc, 0x00, SvIVX (sv)); 610 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
385 else 611 else
386 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 612 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
387 } 613 }
388 else if (SvROK (sv)) 614 else if (SvROK (sv))
389 encode_rv (enc, SvRV (sv)); 615 encode_rv (enc, SvRV (sv));
390 else if (!SvOK (sv)) 616 else if (!SvOK (sv))
391 encode_ch (enc, 0xe0 | 22); 617 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
392 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 618 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
393 encode_ch (enc, 0xe0 | 23); 619 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 else 620 else
395 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 621 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
396 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 622 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
397} 623}
398 624
399static SV * 625static SV *
400encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
401{ 627{
402 enc_t enc; 628 enc_t enc = { 0 };
403 629
404 enc.cbor = *cbor; 630 enc.cbor = *cbor;
405 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
406 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
407 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
408 enc.depth = 0;
409 634
410 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
636
637 if (cbor->flags & F_PACK_STRINGS)
638 {
639 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
640 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
641 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
642 }
643
411 encode_sv (&enc, scalar); 644 encode_sv (&enc, scalar);
412 645
413 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 646 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
414 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 647 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
415 648
429 U8 *end; // end of input string 662 U8 *end; // end of input string
430 const char *err; // parse error, if != 0 663 const char *err; // parse error, if != 0
431 CBOR cbor; 664 CBOR cbor;
432 U32 depth; // recursion depth 665 U32 depth; // recursion depth
433 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
667 AV *shareable;
668 AV *stringref;
669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
434} dec_t; 671} dec_t;
435 672
436#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
437 680
438#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
439 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703#define ERR_DO(do) SB do; goto fail; SE
704#define ERR(reason) ERR_DO (err_set (dec, reason))
705#define ERR_ERRSV ERR_DO (err_errsv (dec))
706
707#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
708
440#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 709#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
441#define DEC_DEC_DEPTH --dec->depth 710#define DEC_DEC_DEPTH --dec->depth
442 711
443static UV 712static UV
444decode_uint (dec_t *dec) 713decode_uint (dec_t *dec)
445{ 714{
446 switch (*dec->cur & 31) 715 U8 m = *dec->cur & MINOR_MASK;
447 { 716 ++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;
452 717
453 case 24: 718 if (ecb_expect_true (m < LENGTH_EXT1))
719 return m;
720 else if (ecb_expect_true (m == LENGTH_EXT1))
721 {
454 WANT (2); 722 WANT (1);
455 dec->cur += 2; 723 dec->cur += 1;
456 return dec->cur[-1]; 724 return dec->cur[-1];
457 725 }
458 case 25: 726 else if (ecb_expect_true (m == LENGTH_EXT2))
727 {
459 WANT (3); 728 WANT (2);
460 dec->cur += 3; 729 dec->cur += 2;
461 return (((UV)dec->cur[-2]) << 8) 730 return (((UV)dec->cur[-2]) << 8)
462 | ((UV)dec->cur[-1]); 731 | ((UV)dec->cur[-1]);
463 732 }
464 case 26: 733 else if (ecb_expect_true (m == LENGTH_EXT4))
734 {
465 WANT (5); 735 WANT (4);
466 dec->cur += 5; 736 dec->cur += 4;
467 return (((UV)dec->cur[-4]) << 24) 737 return (((UV)dec->cur[-4]) << 24)
468 | (((UV)dec->cur[-3]) << 16) 738 | (((UV)dec->cur[-3]) << 16)
469 | (((UV)dec->cur[-2]) << 8) 739 | (((UV)dec->cur[-2]) << 8)
470 | ((UV)dec->cur[-1]); 740 | ((UV)dec->cur[-1]);
471 741 }
472 case 27: 742 else if (ecb_expect_true (m == LENGTH_EXT8))
743 {
473 WANT (9); 744 WANT (8);
474 dec->cur += 9; 745 dec->cur += 8;
746
747 return
748#if UVSIZE < 8
749 0
750#else
475 return (((UV)dec->cur[-8]) << 56) 751 (((UV)dec->cur[-8]) << 56)
476 | (((UV)dec->cur[-7]) << 48) 752 | (((UV)dec->cur[-7]) << 48)
477 | (((UV)dec->cur[-6]) << 40) 753 | (((UV)dec->cur[-6]) << 40)
478 | (((UV)dec->cur[-5]) << 32) 754 | (((UV)dec->cur[-5]) << 32)
755#endif
479 | (((UV)dec->cur[-4]) << 24) 756 | (((UV)dec->cur[-4]) << 24)
480 | (((UV)dec->cur[-3]) << 16) 757 | (((UV)dec->cur[-3]) << 16)
481 | (((UV)dec->cur[-2]) << 8) 758 | (((UV)dec->cur[-2]) << 8)
482 | ((UV)dec->cur[-1]); 759 | ((UV)dec->cur[-1]);
483 760 }
484 default: 761 else
485 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 762 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
486 }
487 763
488fail: 764fail:
489 return 0; 765 return 0;
490} 766}
491 767
496{ 772{
497 AV *av = newAV (); 773 AV *av = newAV ();
498 774
499 DEC_INC_DEPTH; 775 DEC_INC_DEPTH;
500 776
501 if ((*dec->cur & 31) == 31) 777 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
502 { 778 {
503 ++dec->cur; 779 ++dec->cur;
504 780
505 for (;;) 781 for (;;)
506 { 782 {
507 WANT (1); 783 WANT (1);
508 784
509 if (*dec->cur == 0xe0 | 31) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
510 { 786 {
511 ++dec->cur; 787 ++dec->cur;
512 break; 788 break;
513 } 789 }
514 790
515 av_push (av, decode_sv (dec)); 791 av_push (av, decode_sv (dec));
516 } 792 }
517 } 793 }
518 else 794 else
519 { 795 {
520 int i, len = decode_uint (dec); 796 UV i, len = decode_uint (dec);
521 797
798 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); 799 av_fill (av, len - 1);
523 800
524 for (i = 0; i < len; ++i) 801 for (i = 0; i < len; ++i)
525 AvARRAY (av)[i] = decode_sv (dec); 802 AvARRAY (av)[i] = decode_sv (dec);
526 } 803 }
527 804
528 DEC_DEC_DEPTH; 805 DEC_DEC_DEPTH;
529 return newRV_noinc ((SV *)av); 806 return newRV_noinc ((SV *)av);
530 807
531fail: 808fail:
532 SvREFCNT_dec (av); 809 SvREFCNT_dec_NN (av);
533 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 811 return &PL_sv_undef;
535} 812}
536 813
814static void
815decode_he (dec_t *dec, HV *hv)
816{
817 // for speed reasons, we specialcase single-string
818 // byte or utf-8 strings as keys, but only when !stringref
819
820 if (ecb_expect_true (!dec->stringref))
821 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
822 {
823 STRLEN len = decode_uint (dec);
824 char *key = (char *)dec->cur;
825
826 WANT (len);
827 dec->cur += len;
828
829 hv_store (hv, key, len, decode_sv (dec), 0);
830
831 return;
832 }
833 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
834 {
835 STRLEN len = decode_uint (dec);
836 char *key = (char *)dec->cur;
837
838 WANT (len);
839 dec->cur += len;
840
841 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
842 if (!is_utf8_string (key, len))
843 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
844
845 hv_store (hv, key, -len, decode_sv (dec), 0);
846
847 return;
848 }
849
850 SV *k = decode_sv (dec);
851 SV *v = decode_sv (dec);
852
853 // we leak memory if uncaught exceptions are thrown by random magical
854 // methods, and this is hopefully the only place where it can happen,
855 // so if there is a chance of an exception, take the very slow path.
856 // since catching exceptions is "undocumented/internal/forbidden" by
857 // the new p5p powers, we need to call out to a perl function :/
858 if (ecb_expect_false (SvAMAGIC (k)))
859 {
860 dSP;
861
862 ENTER; SAVETMPS;
863 PUSHMARK (SP);
864 EXTEND (SP, 3);
865 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
866 PUSHs (sv_2mortal (k));
867 PUSHs (sv_2mortal (v));
868
869 PUTBACK;
870 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
871 SPAGAIN;
872
873 FREETMPS; LEAVE;
874
875 if (SvTRUE (ERRSV))
876 ERR_ERRSV;
877
878 return;
879 }
880
881 hv_store_ent (hv, k, v, 0);
882 SvREFCNT_dec_NN (k);
883
884fail:
885 ;
886}
887
537static SV * 888static SV *
538decode_hv (dec_t *dec) 889decode_hv (dec_t *dec)
539{ 890{
540 HV *hv = newHV (); 891 HV *hv = newHV ();
541 892
542 DEC_INC_DEPTH; 893 DEC_INC_DEPTH;
543 894
544 if ((*dec->cur & 31) == 31) 895 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
545 { 896 {
546 ++dec->cur; 897 ++dec->cur;
547 898
548 for (;;) 899 for (;;)
549 { 900 {
550 WANT (1); 901 WANT (1);
551 902
552 if (*dec->cur == 0xe0 | 31) 903 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
553 { 904 {
554 ++dec->cur; 905 ++dec->cur;
555 break; 906 break;
556 } 907 }
557 908
558 SV *k = decode_sv (dec); 909 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 910 }
563 } 911 }
564 else 912 else
565 { 913 {
566 int len = decode_uint (dec); 914 UV pairs = decode_uint (dec);
567 915
916 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
917
568 while (len--) 918 while (pairs--)
569 { 919 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 } 920 }
576 921
577 DEC_DEC_DEPTH; 922 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 923 return newRV_noinc ((SV *)hv);
579 924
580#if 0 925fail:
581 SV *sv; 926 SvREFCNT_dec_NN (hv);
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH; 927 DEC_DEC_DEPTH;
585 decode_ws (dec); 928 return &PL_sv_undef;
929}
586 930
587 for (;;) 931static SV *
588 { 932decode_str (dec_t *dec, int utf8)
589 // heuristic: assume that 933{
590 // a) decode_str + hv_store_ent are abysmally slow. 934 SV *sv = 0;
591 // b) most hash keys are short, simple ascii text. 935
592 // => try to "fast-match" such strings to avoid 936 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
593 // the overhead of decode_str + hv_store_ent.
594 { 937 {
595 SV *value; 938 // indefinite length strings
596 char *p = dec->cur; 939 ++dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598 940
941 U8 major = *dec->cur & MAJOR_MISC;
942
943 sv = newSVpvn ("", 0);
944
599 for (;;) 945 for (;;)
600 { 946 {
601 // the >= 0x80 is false on most architectures 947 WANT (1);
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 948
949 if ((*dec->cur - major) > LENGTH_EXT8)
950 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
603 { 951 {
604 // slow path, back up and use decode_str 952 ++dec->cur;
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; 953 break;
623 } 954 }
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 { 955 else
689 dSP; 956 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
690 int count;
691 957
692 ENTER; SAVETMPS; PUSHMARK (SP); 958 STRLEN len = decode_uint (dec);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695 959
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
735fail:
736 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH;
738 return &PL_sv_undef;
739}
740
741static SV *
742decode_str (dec_t *dec, int utf8)
743{
744 SV *sv;
745
746 if ((*dec->cur & 31) == 31)
747 {
748 ++dec->cur;
749
750 sv = newSVpvn ("", 0);
751
752 // not very fast, and certainly not robust against illegal input
753 for (;;)
754 {
755 WANT (1); 960 WANT (len);
756 961 sv_catpvn (sv, dec->cur, len);
757 if (*dec->cur == 0xe0 | 31) 962 dec->cur += len;
758 {
759 ++dec->cur;
760 break;
761 }
762
763 SV *sv2 = decode_sv (dec);
764 sv_catsv (sv, sv2);
765 } 963 }
766 } 964 }
767 else 965 else
768 { 966 {
769 STRLEN len = decode_uint (dec); 967 STRLEN len = decode_uint (dec);
770 968
771 WANT (len); 969 WANT (len);
772 sv = newSVpvn (dec->cur, len); 970 sv = newSVpvn (dec->cur, len);
773 dec->cur += len; 971 dec->cur += len;
972
973 if (ecb_expect_false (dec->stringref)
974 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
975 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
774 } 976 }
775 977
776 if (utf8) 978 if (utf8)
979 {
980 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
981 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
982 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
983
777 SvUTF8_on (sv); 984 SvUTF8_on (sv);
985 }
778 986
779 return sv; 987 return sv;
780 988
781fail: 989fail:
990 SvREFCNT_dec (sv);
991 return &PL_sv_undef;
992}
993
994static SV *
995decode_tagged (dec_t *dec)
996{
997 SV *sv = 0;
998 UV tag = decode_uint (dec);
999
1000 WANT (1);
1001
1002 switch (tag)
1003 {
1004 case CBOR_TAG_MAGIC:
1005 sv = decode_sv (dec);
1006 break;
1007
1008 case CBOR_TAG_INDIRECTION:
1009 sv = newRV_noinc (decode_sv (dec));
1010 break;
1011
1012 case CBOR_TAG_STRINGREF_NAMESPACE:
1013 {
1014 // do not use SAVETMPS/FREETMPS, as these will
1015 // erase mortalised caches, e.g. "shareable"
1016 ENTER;
1017
1018 SAVESPTR (dec->stringref);
1019 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1020
1021 sv = decode_sv (dec);
1022
1023 LEAVE;
1024 }
1025 break;
1026
1027 case CBOR_TAG_STRINGREF:
1028 {
1029 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1030 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1031
1032 UV idx = decode_uint (dec);
1033
1034 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1035 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1036
1037 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1038 }
1039 break;
1040
1041 case CBOR_TAG_VALUE_SHAREABLE:
1042 {
1043 if (ecb_expect_false (!dec->shareable))
1044 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1045
1046 if (dec->cbor.flags & F_ALLOW_CYCLES)
1047 {
1048 sv = newSV (0);
1049 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1050
1051 SV *osv = decode_sv (dec);
1052 sv_setsv (sv, osv);
1053 SvREFCNT_dec_NN (osv);
1054 }
1055 else
1056 {
1057 av_push (dec->shareable, &PL_sv_undef);
1058 int idx = AvFILLp (dec->shareable);
1059 sv = decode_sv (dec);
1060 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1061 }
1062 }
1063 break;
1064
1065 case CBOR_TAG_VALUE_SHAREDREF:
1066 {
1067 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1068 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1069
1070 UV idx = decode_uint (dec);
1071
1072 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1073 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1074
1075 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1076
1077 if (sv == &PL_sv_undef)
1078 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1079 }
1080 break;
1081
1082 case CBOR_TAG_PERL_OBJECT:
1083 {
1084 if (dec->cbor.flags & F_FORBID_OBJECTS)
1085 goto filter;
1086
1087 sv = decode_sv (dec);
1088
1089 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1090 ERR ("corrupted CBOR data (non-array perl object)");
1091
1092 AV *av = (AV *)SvRV (sv);
1093 int len = av_len (av) + 1;
1094 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1095
1096 if (!stash)
1097 ERR ("cannot decode perl-object (package does not exist)");
1098
1099 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1100
1101 if (!method)
1102 ERR ("cannot decode perl-object (package does not have a THAW method)");
1103
1104 dSP;
1105
1106 ENTER; SAVETMPS;
1107 PUSHMARK (SP);
1108 EXTEND (SP, len + 1);
1109 // we re-bless the reference to get overload and other niceties right
1110 PUSHs (*av_fetch (av, 0, 1));
1111 PUSHs (sv_cbor);
1112
1113 int i;
1114
1115 for (i = 1; i < len; ++i)
1116 PUSHs (*av_fetch (av, i, 1));
1117
1118 PUTBACK;
1119 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1120 SPAGAIN;
1121
1122 if (SvTRUE (ERRSV))
1123 {
1124 FREETMPS; LEAVE;
1125 ERR_ERRSV;
1126 }
1127
1128 SvREFCNT_dec_NN (sv);
1129 sv = SvREFCNT_inc (POPs);
1130
1131 PUTBACK;
1132
1133 FREETMPS; LEAVE;
1134 }
1135 break;
1136
1137 default:
1138 filter:
1139 {
1140 SV *tag_sv = newSVuv (tag);
1141
1142 sv = decode_sv (dec);
1143
1144 dSP;
1145 ENTER; SAVETMPS;
1146 SAVESTACK_POS ();
1147 PUSHMARK (SP);
1148 EXTEND (SP, 2);
1149 PUSHs (tag_sv);
1150 PUSHs (sv);
1151
1152 PUTBACK;
1153 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1154 SPAGAIN;
1155
1156 if (SvTRUE (ERRSV))
1157 {
1158 SvREFCNT_dec_NN (tag_sv);
1159 FREETMPS; LEAVE;
1160 ERR_ERRSV;
1161 }
1162
1163 if (count)
1164 {
1165 SvREFCNT_dec_NN (tag_sv);
1166 SvREFCNT_dec_NN (sv);
1167 sv = SvREFCNT_inc_NN (POPs);
1168 }
1169 else
1170 {
1171 AV *av = newAV ();
1172 av_push (av, tag_sv);
1173 av_push (av, sv);
1174
1175 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1176 ? cbor_tagged_stash
1177 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1178 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1179 }
1180
1181 PUTBACK;
1182
1183 FREETMPS; LEAVE;
1184 }
1185 break;
1186 }
1187
1188 return sv;
1189
1190fail:
1191 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 1192 return &PL_sv_undef;
783} 1193}
784 1194
785static SV * 1195static SV *
786decode_sv (dec_t *dec) 1196decode_sv (dec_t *dec)
787{ 1197{
788 WANT (1); 1198 WANT (1);
789 1199
790 switch (*dec->cur >> 5) 1200 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1201 {
792 case 0: // unsigned int 1202 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1203 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1204 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1205 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1206 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1207 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1208 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1209
800 return decode_str (dec, 1); 1210 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) 1211 switch (*dec->cur++ & MINOR_MASK)
810 { 1212 {
811 case 20: 1213 case SIMPLE_FALSE:
812#if CBOR_SLOW 1214#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 1215 types_false = get_bool ("Types::Serialiser::false");
814#endif 1216#endif
815 return newSVsv (cbor_false); 1217 return newSVsv (types_false);
816 case 21: 1218 case SIMPLE_TRUE:
817#if CBOR_SLOW 1219#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 1220 types_true = get_bool ("Types::Serialiser::true");
819#endif 1221#endif
820 return newSVsv (cbor_true); 1222 return newSVsv (types_true);
821 case 22: 1223 case SIMPLE_NULL:
822 return newSVsv (&PL_sv_undef); 1224 return newSVsv (&PL_sv_undef);
1225 case SIMPLE_UNDEF:
1226#if CBOR_SLOW
1227 types_error = get_bool ("Types::Serialiser::error");
1228#endif
1229 return newSVsv (types_error);
823 1230
824 case 25: 1231 case MISC_FLOAT16:
825 // half float
826 abort ();
827 break; 1232 {
1233 WANT (2);
828 1234
1235 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1236 dec->cur += 2;
1237
1238 return newSVnv (ecb_binary16_to_float (fp));
1239 }
1240
829 case 26: 1241 case MISC_FLOAT32:
830 { 1242 {
831 uint32_t fp; 1243 uint32_t fp;
832 WANT (4); 1244 WANT (4);
833 memcpy (&fp, dec->cur, 4); 1245 memcpy (&fp, dec->cur, 4);
834 dec->cur += 4; 1246 dec->cur += 4;
837 fp = ecb_bswap32 (fp); 1249 fp = ecb_bswap32 (fp);
838 1250
839 return newSVnv (ecb_binary32_to_float (fp)); 1251 return newSVnv (ecb_binary32_to_float (fp));
840 } 1252 }
841 1253
842 case 27: 1254 case MISC_FLOAT64:
843 { 1255 {
844 uint64_t fp; 1256 uint64_t fp;
845 WANT (8); 1257 WANT (8);
846 memcpy (&fp, dec->cur, 8); 1258 memcpy (&fp, dec->cur, 8);
847 dec->cur += 8; 1259 dec->cur += 8;
850 fp = ecb_bswap64 (fp); 1262 fp = ecb_bswap64 (fp);
851 1263
852 return newSVnv (ecb_binary64_to_double (fp)); 1264 return newSVnv (ecb_binary64_to_double (fp));
853 } 1265 }
854 1266
855 // 0..19 unassigned 1267 // 0..19 unassigned simple
856 // 24 reserved + unassigned (reserved values are not encodable) 1268 // 24 reserved + unassigned simple (reserved values are not encodable)
1269 // 28-30 unassigned misc
1270 // 31 break code
857 default: 1271 default:
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1272 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
859 } 1273 }
860 1274
861 break; 1275 break;
862 } 1276 }
863#if 0
864 switch (*dec->cur)
865 {
866 //case '"': ++dec->cur; return decode_str (dec);
867 case '[': ++dec->cur; return decode_av (dec);
868 case '{': ++dec->cur; return decode_hv (dec);
869
870 case '-':
871 case '0': case '1': case '2': case '3': case '4':
872 case '5': case '6': case '7': case '8': case '9':
873 //TODO return decode_num (dec);
874
875 case 't':
876 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
877 {
878 dec->cur += 4;
879#if CBOR_SLOW
880 cbor_true = get_bool ("CBOR::XS::true");
881#endif
882 return newSVsv (cbor_true);
883 }
884 else
885 ERR ("'true' expected");
886
887 break;
888
889 case 'f':
890 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
891 {
892 dec->cur += 5;
893#if CBOR_SLOW
894 cbor_false = get_bool ("CBOR::XS::false");
895#endif
896 return newSVsv (cbor_false);
897 }
898 else
899 ERR ("'false' expected");
900
901 break;
902
903 case 'n':
904 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
905 {
906 dec->cur += 4;
907 return newSVsv (&PL_sv_undef);
908 }
909 else
910 ERR ("'null' expected");
911
912 break;
913
914 default:
915 ERR ("malformed CBOR string, neither array, object, number, string or atom");
916 break;
917 }
918#endif
919 1277
920fail: 1278fail:
921 return &PL_sv_undef; 1279 return &PL_sv_undef;
922} 1280}
923 1281
924static SV * 1282static SV *
925decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1283decode_cbor (SV *string, CBOR *cbor, char **offset_return)
926{ 1284{
927 dec_t dec; 1285 dec_t dec = { 0 };
928 SV *sv; 1286 SV *sv;
1287 STRLEN len;
1288 char *data = SvPVbyte (string, len);
929 1289
930 /* work around bugs in 5.10 where manipulating magic values
931 * makes perl ignore the magic in subsequent accesses.
932 * also make a copy of non-PV values, to get them into a clean
933 * state (SvPV should do that, but it's buggy, see below).
934 */
935 /*SvGETMAGIC (string);*/
936 if (SvMAGICAL (string) || !SvPOK (string))
937 string = sv_2mortal (newSVsv (string));
938
939 SvUPGRADE (string, SVt_PV);
940
941 /* work around a bug in perl 5.10, which causes SvCUR to fail an
942 * assertion with -DDEBUGGING, although SvCUR is documented to
943 * return the xpv_cur field which certainly exists after upgrading.
944 * according to nicholas clark, calling SvPOK fixes this.
945 * But it doesn't fix it, so try another workaround, call SvPV_nolen
946 * and hope for the best.
947 * Damnit, SvPV_nolen still trips over yet another assertion. This
948 * assertion business is seriously broken, try yet another workaround
949 * for the broken -DDEBUGGING.
950 */
951 {
952#ifdef DEBUGGING
953 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
954#else
955 STRLEN offset = SvCUR (string);
956#endif
957
958 if (offset > cbor->max_size && cbor->max_size) 1290 if (len > cbor->max_size && cbor->max_size)
959 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1291 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
960 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1292 (unsigned long)len, (unsigned long)cbor->max_size);
961 }
962
963 sv_utf8_downgrade (string, 0);
964 1293
965 dec.cbor = *cbor; 1294 dec.cbor = *cbor;
966 dec.cur = (U8 *)SvPVX (string); 1295 dec.cur = (U8 *)data;
967 dec.end = (U8 *)SvEND (string); 1296 dec.end = (U8 *)data + len;
968 dec.err = 0;
969 dec.depth = 0;
970
971 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
972 ;//TODO dec.cbor.flags |= F_HOOK;
973 1297
974 sv = decode_sv (&dec); 1298 sv = decode_sv (&dec);
975 1299
976 if (offset_return) 1300 if (offset_return)
977 *offset_return = dec.cur; 1301 *offset_return = dec.cur;
978 1302
979 if (!(offset_return || !sv)) 1303 if (!(offset_return || !sv))
980 {
981 if (*dec.cur && !dec.err) 1304 if (dec.cur != dec.end && !dec.err)
982 {
983 dec.err = "garbage after CBOR object"; 1305 dec.err = "garbage after CBOR object";
984 SvREFCNT_dec (sv); 1306
985 sv = 0; 1307 if (dec.err)
1308 {
1309 if (dec.shareable)
986 } 1310 {
987 } 1311 // need to break cyclic links, which would all be in shareable
1312 int i;
1313 SV **svp;
988 1314
989 if (!sv) 1315 for (i = av_len (dec.shareable) + 1; i--; )
1316 if ((svp = av_fetch (dec.shareable, i, 0)))
1317 sv_setsv (*svp, &PL_sv_undef);
1318 }
1319
1320 SvREFCNT_dec_NN (sv);
1321
1322 if (dec.err_sv)
1323 sv_2mortal (dec.err_sv);
1324
990 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1325 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1326 }
991 1327
992 sv = sv_2mortal (sv); 1328 sv = sv_2mortal (sv);
993 1329
994 return sv; 1330 return sv;
995} 1331}
996 1332
1333/////////////////////////////////////////////////////////////////////////////
1334// incremental parser
1335
1336#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1337
1338// returns 0 for notyet, 1 for success or error
1339static int
1340incr_parse (CBOR *self, SV *cborstr)
1341{
1342 STRLEN cur;
1343 SvPV (cborstr, cur);
1344
1345 while (ecb_expect_true (self->incr_need <= cur))
1346 {
1347 // table of integer count bytes
1348 static I8 incr_len[MINOR_MASK + 1] = {
1349 0, 0, 0, 0, 0, 0, 0, 0,
1350 0, 0, 0, 0, 0, 0, 0, 0,
1351 0, 0, 0, 0, 0, 0, 0, 0,
1352 1, 2, 4, 8,-1,-1,-1,-2
1353 };
1354
1355 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1356 U8 m = *p & MINOR_MASK;
1357 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1358 I8 ilen = incr_len[m];
1359
1360 self->incr_need = self->incr_pos + 1;
1361
1362 if (ecb_expect_false (ilen < 0))
1363 {
1364 if (m != MINOR_INDEF)
1365 return 1; // error
1366
1367 if (*p == (MAJOR_MISC | MINOR_INDEF))
1368 {
1369 if (count >= 0)
1370 return 1; // error
1371
1372 count = 1;
1373 }
1374 else
1375 {
1376 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1377 count = -1;
1378 }
1379 }
1380 else
1381 {
1382 self->incr_need += ilen;
1383 if (ecb_expect_false (self->incr_need > cur))
1384 return 0;
1385
1386 int major = *p >> MAJOR_SHIFT;
1387
1388 switch (major)
1389 {
1390 case MAJOR_TAG >> MAJOR_SHIFT:
1391 ++count; // tags merely prefix another value
1392 break;
1393
1394 case MAJOR_BYTES >> MAJOR_SHIFT:
1395 case MAJOR_TEXT >> MAJOR_SHIFT:
1396 case MAJOR_ARRAY >> MAJOR_SHIFT:
1397 case MAJOR_MAP >> MAJOR_SHIFT:
1398 {
1399 UV len;
1400
1401 if (ecb_expect_false (ilen))
1402 {
1403 len = 0;
1404
1405 do {
1406 len = (len << 8) | *++p;
1407 } while (--ilen);
1408 }
1409 else
1410 len = m;
1411
1412 switch (major)
1413 {
1414 case MAJOR_BYTES >> MAJOR_SHIFT:
1415 case MAJOR_TEXT >> MAJOR_SHIFT:
1416 self->incr_need += len;
1417 if (ecb_expect_false (self->incr_need > cur))
1418 return 0;
1419
1420 break;
1421
1422 case MAJOR_MAP >> MAJOR_SHIFT:
1423 len <<= 1;
1424 case MAJOR_ARRAY >> MAJOR_SHIFT:
1425 if (len)
1426 {
1427 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1428 count = len + 1;
1429 }
1430 break;
1431 }
1432 }
1433 }
1434 }
1435
1436 self->incr_pos = self->incr_need;
1437
1438 if (count > 0)
1439 {
1440 while (!--count)
1441 {
1442 if (!AvFILLp (self->incr_count))
1443 return 1; // done
1444
1445 SvREFCNT_dec_NN (av_pop (self->incr_count));
1446 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1447 }
1448
1449 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1450 }
1451 }
1452
1453 return 0;
1454}
1455
1456
997///////////////////////////////////////////////////////////////////////////// 1457/////////////////////////////////////////////////////////////////////////////
998// XS interface functions 1458// XS interface functions
999 1459
1000MODULE = CBOR::XS PACKAGE = CBOR::XS 1460MODULE = CBOR::XS PACKAGE = CBOR::XS
1001 1461
1002BOOT: 1462BOOT:
1003{ 1463{
1004 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1464 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1005 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1465 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1006 1466
1007 cbor_true = get_bool ("CBOR::XS::true"); 1467 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1008 cbor_false = get_bool ("CBOR::XS::false"); 1468 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1469
1470 types_true = get_bool ("Types::Serialiser::true" );
1471 types_false = get_bool ("Types::Serialiser::false");
1472 types_error = get_bool ("Types::Serialiser::error");
1473
1474 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1475
1476 sv_cbor = newSVpv ("CBOR", 0);
1477 SvREADONLY_on (sv_cbor);
1478
1479 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1009} 1480}
1010 1481
1011PROTOTYPES: DISABLE 1482PROTOTYPES: DISABLE
1012 1483
1013void CLONE (...) 1484void CLONE (...)
1014 CODE: 1485 CODE:
1015 cbor_stash = 0; 1486 cbor_stash = 0;
1487 cbor_tagged_stash = 0;
1488 types_error_stash = 0;
1016 cbor_boolean_stash = 0; 1489 types_boolean_stash = 0;
1017 1490
1018void new (char *klass) 1491void new (char *klass)
1019 PPCODE: 1492 PPCODE:
1020{ 1493{
1021 SV *pv = NEWSV (0, sizeof (CBOR)); 1494 SV *pv = NEWSV (0, sizeof (CBOR));
1029 1502
1030void shrink (CBOR *self, int enable = 1) 1503void shrink (CBOR *self, int enable = 1)
1031 ALIAS: 1504 ALIAS:
1032 shrink = F_SHRINK 1505 shrink = F_SHRINK
1033 allow_unknown = F_ALLOW_UNKNOWN 1506 allow_unknown = F_ALLOW_UNKNOWN
1507 allow_sharing = F_ALLOW_SHARING
1508 allow_cycles = F_ALLOW_CYCLES
1509 forbid_objects = F_FORBID_OBJECTS
1510 pack_strings = F_PACK_STRINGS
1511 text_keys = F_TEXT_KEYS
1512 text_strings = F_TEXT_STRINGS
1513 validate_utf8 = F_VALIDATE_UTF8
1034 PPCODE: 1514 PPCODE:
1035{ 1515{
1036 if (enable) 1516 if (enable)
1037 self->flags |= ix; 1517 self->flags |= ix;
1038 else 1518 else
1043 1523
1044void get_shrink (CBOR *self) 1524void get_shrink (CBOR *self)
1045 ALIAS: 1525 ALIAS:
1046 get_shrink = F_SHRINK 1526 get_shrink = F_SHRINK
1047 get_allow_unknown = F_ALLOW_UNKNOWN 1527 get_allow_unknown = F_ALLOW_UNKNOWN
1528 get_allow_sharing = F_ALLOW_SHARING
1529 get_allow_cycles = F_ALLOW_CYCLES
1530 get_forbid_objects = F_FORBID_OBJECTS
1531 get_pack_strings = F_PACK_STRINGS
1532 get_text_keys = F_TEXT_KEYS
1533 get_text_strings = F_TEXT_STRINGS
1534 get_validate_utf8 = F_VALIDATE_UTF8
1048 PPCODE: 1535 PPCODE:
1049 XPUSHs (boolSV (self->flags & ix)); 1536 XPUSHs (boolSV (self->flags & ix));
1050 1537
1051void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1538void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1052 PPCODE: 1539 PPCODE:
1068 CODE: 1555 CODE:
1069 RETVAL = self->max_size; 1556 RETVAL = self->max_size;
1070 OUTPUT: 1557 OUTPUT:
1071 RETVAL 1558 RETVAL
1072 1559
1073#if 0 //TODO 1560void filter (CBOR *self, SV *filter = 0)
1074
1075void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1076 PPCODE: 1561 PPCODE:
1077{
1078 SvREFCNT_dec (self->cb_object); 1562 SvREFCNT_dec (self->filter);
1079 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1563 self->filter = filter ? newSVsv (filter) : filter;
1080
1081 XPUSHs (ST (0)); 1564 XPUSHs (ST (0));
1082}
1083 1565
1084void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1566SV *get_filter (CBOR *self)
1085 PPCODE: 1567 CODE:
1086{ 1568 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1087 if (!self->cb_sk_object) 1569 OUTPUT:
1088 self->cb_sk_object = newHV (); 1570 RETVAL
1089
1090 if (SvOK (cb))
1091 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1092 else
1093 {
1094 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1095
1096 if (!HvKEYS (self->cb_sk_object))
1097 {
1098 SvREFCNT_dec (self->cb_sk_object);
1099 self->cb_sk_object = 0;
1100 }
1101 }
1102
1103 XPUSHs (ST (0));
1104}
1105
1106#endif
1107 1571
1108void encode (CBOR *self, SV *scalar) 1572void encode (CBOR *self, SV *scalar)
1109 PPCODE: 1573 PPCODE:
1110 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1574 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1111 XPUSHs (scalar); 1575 XPUSHs (scalar);
1124 EXTEND (SP, 2); 1588 EXTEND (SP, 2);
1125 PUSHs (sv); 1589 PUSHs (sv);
1126 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1590 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1127} 1591}
1128 1592
1593void incr_parse (CBOR *self, SV *cborstr)
1594 ALIAS:
1595 incr_parse_multiple = 1
1596 PPCODE:
1597{
1598 if (SvUTF8 (cborstr))
1599 sv_utf8_downgrade (cborstr, 0);
1600
1601 if (!self->incr_count)
1602 {
1603 self->incr_count = newAV ();
1604 self->incr_pos = 0;
1605 self->incr_need = 1;
1606
1607 av_push (self->incr_count, newSViv (1));
1608 }
1609
1610 do
1611 {
1612 if (!incr_parse (self, cborstr))
1613 {
1614 if (self->incr_need > self->max_size && self->max_size)
1615 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1616 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1617
1618 break;
1619 }
1620
1621 SV *sv;
1622 char *offset;
1623
1624 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1625 XPUSHs (sv);
1626
1627 sv_chop (cborstr, offset);
1628
1629 av_clear (self->incr_count);
1630 av_push (self->incr_count, newSViv (1));
1631
1632 self->incr_pos = 0;
1633 self->incr_need = self->incr_pos + 1;
1634 }
1635 while (ix);
1636}
1637
1638void incr_reset (CBOR *self)
1639 CODE:
1640{
1641 SvREFCNT_dec (self->incr_count);
1642 self->incr_count = 0;
1643}
1644
1129void DESTROY (CBOR *self) 1645void DESTROY (CBOR *self)
1130 CODE: 1646 PPCODE:
1131 SvREFCNT_dec (self->cb_sk_object); 1647 cbor_free (self);
1132 SvREFCNT_dec (self->cb_object);
1133 1648
1134PROTOTYPES: ENABLE 1649PROTOTYPES: ENABLE
1135 1650
1136void encode_cbor (SV *scalar) 1651void encode_cbor (SV *scalar)
1652 ALIAS:
1653 encode_cbor = 0
1654 encode_cbor_sharing = F_ALLOW_SHARING
1137 PPCODE: 1655 PPCODE:
1138{ 1656{
1139 CBOR cbor; 1657 CBOR cbor;
1140 cbor_init (&cbor); 1658 cbor_init (&cbor);
1659 cbor.flags |= ix;
1141 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1660 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1142 XPUSHs (scalar); 1661 XPUSHs (scalar);
1143} 1662}
1144 1663
1145void decode_cbor (SV *cborstr) 1664void decode_cbor (SV *cborstr)
1149 cbor_init (&cbor); 1668 cbor_init (&cbor);
1150 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1669 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1151 XPUSHs (cborstr); 1670 XPUSHs (cborstr);
1152} 1671}
1153 1672
1673#ifdef __AFL_COMPILER
1674
1675void
1676afl_init ()
1677 CODE:
1678 __AFL_INIT ();
1679
1680int
1681afl_loop (unsigned int count = 10000)
1682 CODE:
1683 RETVAL = __AFL_LOOP (count);
1684 OUTPUT:
1685 RETVAL
1686
1687#endif
1688

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines