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 1.65 by root, Thu Nov 15 19:52:41 2018 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_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
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
14#define F_SHRINK 0x00000200UL 104#define F_SHRINK 0x00000001UL
15#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
16 113
17#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
18 115
19#define SB do { 116#define SB do {
20#define SE } while (0) 117#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)
32 118
33#define IN_RANGE_INC(type,val,beg,end) \ 119#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 120 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 121 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 122
42#else 128#else
43# define CBOR_SLOW 0 129# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 130# define CBOR_STASH cbor_stash
45#endif 131#endif
46 132
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 133static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 134static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 135
50typedef struct { 136typedef struct {
51 U32 flags; 137 U32 flags;
52 U32 max_depth; 138 U32 max_depth;
53 STRLEN max_size; 139 STRLEN max_size;
140 SV *filter;
54 141
55 SV *cb_object; 142 // for the incremental parser
56 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.
57} CBOR; 146} CBOR;
58 147
59INLINE void 148ecb_inline void
60cbor_init (CBOR *cbor) 149cbor_init (CBOR *cbor)
61{ 150{
62 Zero (cbor, 1, CBOR); 151 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 152 cbor->max_depth = 512;
64} 153}
65 154
155ecb_inline void
156cbor_free (CBOR *cbor)
157{
158 SvREFCNT_dec (cbor->filter);
159 SvREFCNT_dec (cbor->incr_count);
160}
161
66///////////////////////////////////////////////////////////////////////////// 162/////////////////////////////////////////////////////////////////////////////
67// utility functions 163// utility functions
68 164
69INLINE SV * 165ecb_inline SV *
70get_bool (const char *name) 166get_bool (const char *name)
71{ 167{
72 SV *sv = get_sv (name, 1); 168 SV *sv = get_sv (name, 1);
73 169
74 SvREADONLY_on (sv); 170 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 171 SvREADONLY_on (SvRV (sv));
76 172
77 return sv; 173 return sv;
78} 174}
79 175
80INLINE void 176ecb_inline void
81shrink (SV *sv) 177shrink (SV *sv)
82{ 178{
83 sv_utf8_downgrade (sv, 1); 179 sv_utf8_downgrade (sv, 1);
84 180
85 if (SvLEN (sv) > SvCUR (sv) + 1) 181 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 186 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 187#endif
92 } 188 }
93} 189}
94 190
95///////////////////////////////////////////////////////////////////////////// 191// minimum length of a string to be registered for stringref
96// fp hell 192ecb_inline int
97 193minimum_string_length (UV idx)
98//TODO 194{
195 return idx <= 23 ? 3
196 : idx <= 0xffU ? 4
197 : idx <= 0xffffU ? 5
198 : idx <= 0xffffffffU ? 7
199 : 11;
200}
99 201
100///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
101// encoder 203// encoder
102 204
103// structure used for encoding CBOR 205// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 208 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 209 char *end; // SvEND (sv)
108 SV *sv; // result scalar 210 SV *sv; // result scalar
109 CBOR cbor; 211 CBOR cbor;
110 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;
111} enc_t; 217} enc_t;
112 218
113INLINE void 219ecb_inline void
114need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
115{ 221{
116 if (expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
117 { 223 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 228 }
123} 229}
124 230
125INLINE void 231ecb_inline void
126encode_ch (enc_t *enc, char ch) 232encode_ch (enc_t *enc, char ch)
127{ 233{
128 need (enc, 1); 234 need (enc, 1);
129 *enc->cur++ = ch; 235 *enc->cur++ = ch;
130} 236}
132static void 238static void
133encode_uint (enc_t *enc, int major, UV len) 239encode_uint (enc_t *enc, int major, UV len)
134{ 240{
135 need (enc, 9); 241 need (enc, 9);
136 242
137 if (len < 24) 243 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 244 *enc->cur++ = major | len;
139 else if (len < 0x100) 245 else if (ecb_expect_true (len <= 0xffU))
140 { 246 {
141 *enc->cur++ = major | 24; 247 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 248 *enc->cur++ = len;
143 } 249 }
144 else if (len < 0x10000) 250 else if (len <= 0xffffU)
145 { 251 {
146 *enc->cur++ = major | 25; 252 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 253 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 254 *enc->cur++ = len;
149 } 255 }
150 else if (len < 0x100000000) 256 else if (len <= 0xffffffffU)
151 { 257 {
152 *enc->cur++ = major | 26; 258 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 259 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 260 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 261 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 262 *enc->cur++ = len;
157 } 263 }
158 else if (len) 264 else
159 { 265 {
160 *enc->cur++ = major | 27; 266 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 267 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 268 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 269 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 270 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 271 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 273 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 274 *enc->cur++ = len;
169 } 275 }
170} 276}
171 277
172static 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
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 287{
175 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);
176 need (enc, len); 318 need (enc, len);
177 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
178 enc->cur += len; 320 enc->cur += len;
179} 321}
180 322
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
181static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
182 349
183static void 350static void
184encode_av (enc_t *enc, AV *av) 351encode_av (enc_t *enc, AV *av)
185{ 352{
188 if (enc->depth >= enc->cbor.max_depth) 355 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 356 croak (ERR_NESTING_EXCEEDED);
190 357
191 ++enc->depth; 358 ++enc->depth;
192 359
193 encode_uint (enc, 0x80, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 361
362 if (ecb_expect_false (SvMAGICAL (av)))
195 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
196 { 364 {
197 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 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 }
200 374
201 --enc->depth; 375 --enc->depth;
202} 376}
203 377
204static void 378static void
212 ++enc->depth; 386 ++enc->depth;
213 387
214 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
216 390
217 if (mg) 391 if (ecb_expect_false (mg))
218 encode_ch (enc, 0xa0 | 31); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 393 else
220 encode_uint (enc, 0xa0, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
221 395
222 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
223 { 397 {
224 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
226 else 400 else
227 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));
228 402
229 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));
230 } 404 }
231 405
232 if (mg) 406 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xe0 | 31); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 408
235 --enc->depth; 409 --enc->depth;
236} 410}
237 411
238// encode objects, arrays and special \0=false and \1=true values. 412// encode objects, arrays and special \0=false and \1=true values.
239static void 413static void
240encode_rv (enc_t *enc, SV *sv) 414encode_rv (enc_t *enc, SV *sv)
241{ 415{
242 svtype svt;
243
244 SvGETMAGIC (sv); 416 SvGETMAGIC (sv);
417
245 svt = SvTYPE (sv); 418 svtype svt = SvTYPE (sv);
246 419
247 if (expect_false (SvOBJECT (sv))) 420 if (ecb_expect_false (SvOBJECT (sv)))
248 { 421 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 422 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? 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
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 430 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 431
253 if (SvSTASH (sv) == stash) 432 HV *stash = SvSTASH (sv);
254 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 }
255 else 470 else
256 { 471 {
257#if 0 //TODO 472 sv_setuv (*svp, enc->shareable_idx);
258 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)))
259 { 487 {
488 dSP;
489
490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
260 // 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
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))); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 494
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 495 PUTBACK;
496 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 497 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 498 SPAGAIN;
274 499
275 // catch this surprisingly common error 500 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 501 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))); 502 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 503
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 504 encode_sv (enc, POPs);
283 505
506 PUTBACK;
507
284 FREETMPS; LEAVE; 508 FREETMPS; LEAVE;
285 } 509 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
287 encode_str (enc, "null", 4, 0); 511 {
288 else 512 dSP;
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 513
290 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 514 ENTER; SAVETMPS;
515 PUSHMARK (SP);
516 EXTEND (SP, 2);
517 // we re-bless the reference to get overload and other niceties right
518 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
519 PUSHs (sv_cbor);
520
521 PUTBACK;
522 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
523 SPAGAIN;
524
525 // catch this surprisingly common error
526 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
527 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
528
529 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
530 encode_uint (enc, MAJOR_ARRAY, count + 1);
531 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
532
533 {
534 int i;
535
536 for (i = 0; i < count; ++i)
537 encode_sv (enc, SP[i + 1 - count]);
538
539 SP -= count;
291 } 540 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 541
293 encode_str (enc, "null", 4, 0); 542 PUTBACK;
543
544 FREETMPS; LEAVE;
545 }
294 else 546 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 547 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 549 }
300 else if (svt == SVt_PVHV) 550 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 551 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 552 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 553 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 554 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 555 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 556 encode_tag (enc, CBOR_TAG_INDIRECTION);
557 encode_sv (enc, sv);
558 }
324} 559}
325 560
326static void 561static void
327encode_nv (enc_t *enc, SV *sv) 562encode_nv (enc_t *enc, SV *sv)
328{ 563{
329 double nv = SvNVX (sv); 564 double nv = SvNVX (sv);
330 565
331 need (enc, 9); 566 need (enc, 9);
332 567
333 if (expect_false (nv == (U32)nv)) 568 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 569 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 570 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 571 else if (ecb_expect_false (nv == (float)nv))
337 { 572 {
573 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
574
338 uint32_t fp = ecb_float_to_binary32 (nv); 575 uint32_t fp = ecb_float_to_binary32 (nv);
339
340 *enc->cur++ = 0xe0 | 26;
341 576
342 if (!ecb_big_endian ()) 577 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 578 fp = ecb_bswap32 (fp);
344 579
345 memcpy (enc->cur, &fp, 4); 580 memcpy (enc->cur, &fp, 4);
346 enc->cur += 4; 581 enc->cur += 4;
347 } 582 }
348 else 583 else
349 { 584 {
585 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
586
350 uint64_t fp = ecb_double_to_binary64 (nv); 587 uint64_t fp = ecb_double_to_binary64 (nv);
351
352 *enc->cur++ = 0xe0 | 27;
353 588
354 if (!ecb_big_endian ()) 589 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 590 fp = ecb_bswap64 (fp);
356 591
357 memcpy (enc->cur, &fp, 8); 592 memcpy (enc->cur, &fp, 8);
366 601
367 if (SvPOKp (sv)) 602 if (SvPOKp (sv))
368 { 603 {
369 STRLEN len; 604 STRLEN len;
370 char *str = SvPV (sv, len); 605 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 606 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372 } 607 }
373 else if (SvNOKp (sv)) 608 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 609 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 610 else if (SvIOKp (sv))
376 { 611 {
377 if (SvIsUV (sv)) 612 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 613 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 614 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 615 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 616 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 617 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 618 }
384 else if (SvROK (sv)) 619 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 620 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 621 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 622 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 623 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 624 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 625 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 626 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 627 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 628}
394 629
395static SV * 630static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 631encode_cbor (SV *scalar, CBOR *cbor)
397{ 632{
398 enc_t enc; 633 enc_t enc = { 0 };
399 634
400 enc.cbor = *cbor; 635 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 636 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 637 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 638 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 639
406 SvPOK_only (enc.sv); 640 SvPOK_only (enc.sv);
641
642 if (cbor->flags & F_PACK_STRINGS)
643 {
644 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
645 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
646 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
647 }
648
407 encode_sv (&enc, scalar); 649 encode_sv (&enc, scalar);
408 650
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 651 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 652 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 653
425 U8 *end; // end of input string 667 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 668 const char *err; // parse error, if != 0
427 CBOR cbor; 669 CBOR cbor;
428 U32 depth; // recursion depth 670 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
672 AV *shareable;
673 AV *stringref;
674 SV *decode_tagged;
675 SV *err_sv; // optional sv for error, needs to be freed
430} dec_t; 676} dec_t;
431 677
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 678// set dec->err to ERRSV
679ecb_cold static void
680err_errsv (dec_t *dec)
681{
682 if (!dec->err)
683 {
684 dec->err_sv = newSVsv (ERRSV);
433 685
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 686 // chop off the trailing \n
687 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
688 *SvEND (dec->err_sv) = 0;
435 689
690 dec->err = SvPVutf8_nolen (dec->err_sv);
691 }
692}
693
694// the following functions are used to reduce code size and help the compiler to optimise
695ecb_cold static void
696err_set (dec_t *dec, const char *reason)
697{
698 if (!dec->err)
699 dec->err = reason;
700}
701
702ecb_cold static void
703err_unexpected_end (dec_t *dec)
704{
705 err_set (dec, "unexpected end of CBOR data");
706}
707
708#define ERR_DO(do) SB do; goto fail; SE
709#define ERR(reason) ERR_DO (err_set (dec, reason))
710#define ERR_ERRSV ERR_DO (err_errsv (dec))
711
712#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
713
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 714#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 715#define DEC_DEC_DEPTH --dec->depth
438 716
439static UV 717static UV
440decode_uint (dec_t *dec) 718decode_uint (dec_t *dec)
441{ 719{
442 switch (*dec->cur & 31) 720 U8 m = *dec->cur & MINOR_MASK;
443 { 721 ++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;
448 722
449 case 24: 723 if (ecb_expect_true (m < LENGTH_EXT1))
724 return m;
725 else if (ecb_expect_true (m == LENGTH_EXT1))
726 {
450 WANT (2); 727 WANT (1);
451 dec->cur += 2; 728 dec->cur += 1;
452 return dec->cur[-1]; 729 return dec->cur[-1];
453 730 }
454 case 25: 731 else if (ecb_expect_true (m == LENGTH_EXT2))
732 {
455 WANT (3); 733 WANT (2);
456 dec->cur += 3; 734 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 735 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 736 | ((UV)dec->cur[-1]);
459 737 }
460 case 26: 738 else if (ecb_expect_true (m == LENGTH_EXT4))
739 {
461 WANT (5); 740 WANT (4);
462 dec->cur += 5; 741 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 742 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 743 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 744 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 745 | ((UV)dec->cur[-1]);
467 746 }
468 case 27: 747 else if (ecb_expect_true (m == LENGTH_EXT8))
748 {
469 WANT (9); 749 WANT (8);
470 dec->cur += 9; 750 dec->cur += 8;
751
752 return
753#if UVSIZE < 8
754 0
755#else
471 return (((UV)dec->cur[-8]) << 56) 756 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 757 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 758 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 759 | (((UV)dec->cur[-5]) << 32)
760#endif
475 | (((UV)dec->cur[-4]) << 24) 761 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 762 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 763 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 764 | ((UV)dec->cur[-1]);
479 765 }
480 default: 766 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 767 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 768
484fail: 769fail:
485 return 0; 770 return 0;
486} 771}
487 772
492{ 777{
493 AV *av = newAV (); 778 AV *av = newAV ();
494 779
495 DEC_INC_DEPTH; 780 DEC_INC_DEPTH;
496 781
497 if ((*dec->cur & 31) == 31) 782 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 783 {
499 ++dec->cur; 784 ++dec->cur;
500 785
501 for (;;) 786 for (;;)
502 { 787 {
503 WANT (1); 788 WANT (1);
504 789
505 if (*dec->cur == (0xe0 | 31)) 790 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
506 { 791 {
507 ++dec->cur; 792 ++dec->cur;
508 break; 793 break;
509 } 794 }
510 795
511 av_push (av, decode_sv (dec)); 796 av_push (av, decode_sv (dec));
512 } 797 }
513 } 798 }
514 else 799 else
515 { 800 {
516 int i, len = decode_uint (dec); 801 UV i, len = decode_uint (dec);
517 802
803 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); 804 av_fill (av, len - 1);
519 805
520 for (i = 0; i < len; ++i) 806 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 807 AvARRAY (av)[i] = decode_sv (dec);
522 } 808 }
523 809
524 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
525 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
526 812
527fail: 813fail:
528 SvREFCNT_dec (av); 814 SvREFCNT_dec_NN (av);
529 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 816 return &PL_sv_undef;
531} 817}
532 818
819static void
820decode_he (dec_t *dec, HV *hv)
821{
822 // for speed reasons, we specialcase single-string
823 // byte or utf-8 strings as keys, but only when !stringref
824
825 if (ecb_expect_true (!dec->stringref))
826 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
827 {
828 STRLEN len = decode_uint (dec);
829 char *key = (char *)dec->cur;
830
831 WANT (len);
832 dec->cur += len;
833
834 hv_store (hv, key, len, decode_sv (dec), 0);
835
836 return;
837 }
838 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
839 {
840 STRLEN len = decode_uint (dec);
841 char *key = (char *)dec->cur;
842
843 WANT (len);
844 dec->cur += len;
845
846 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
847 if (!is_utf8_string (key, len))
848 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
849
850 hv_store (hv, key, -len, decode_sv (dec), 0);
851
852 return;
853 }
854
855 SV *k = decode_sv (dec);
856 SV *v = decode_sv (dec);
857
858 // we leak memory if uncaught exceptions are thrown by random magical
859 // methods, and this is hopefully the only place where it can happen,
860 // so if there is a chance of an exception, take the very slow path.
861 // since catching exceptions is "undocumented/internal/forbidden" by
862 // the new p5p powers, we need to call out to a perl function :/
863 if (ecb_expect_false (SvAMAGIC (k)))
864 {
865 dSP;
866
867 ENTER; SAVETMPS;
868 PUSHMARK (SP);
869 EXTEND (SP, 3);
870 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
871 PUSHs (sv_2mortal (k));
872 PUSHs (sv_2mortal (v));
873
874 PUTBACK;
875 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
876 SPAGAIN;
877
878 FREETMPS; LEAVE;
879
880 if (SvTRUE (ERRSV))
881 ERR_ERRSV;
882
883 return;
884 }
885
886 hv_store_ent (hv, k, v, 0);
887 SvREFCNT_dec_NN (k);
888
889fail:
890 ;
891}
892
533static SV * 893static SV *
534decode_hv (dec_t *dec) 894decode_hv (dec_t *dec)
535{ 895{
536 HV *hv = newHV (); 896 HV *hv = newHV ();
537 897
538 DEC_INC_DEPTH; 898 DEC_INC_DEPTH;
539 899
540 if ((*dec->cur & 31) == 31) 900 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 901 {
542 ++dec->cur; 902 ++dec->cur;
543 903
544 for (;;) 904 for (;;)
545 { 905 {
546 WANT (1); 906 WANT (1);
547 907
548 if (*dec->cur == (0xe0 | 31)) 908 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
549 { 909 {
550 ++dec->cur; 910 ++dec->cur;
551 break; 911 break;
552 } 912 }
553 913
554 SV *k = decode_sv (dec); 914 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 915 }
559 } 916 }
560 else 917 else
561 { 918 {
562 int len = decode_uint (dec); 919 UV pairs = decode_uint (dec);
563 920
921 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
922
564 while (len--) 923 while (pairs--)
565 { 924 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 } 925 }
572 926
573 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 928 return newRV_noinc ((SV *)hv);
575 929
576#if 0 930fail:
577 SV *sv; 931 SvREFCNT_dec_NN (hv);
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH; 932 DEC_DEC_DEPTH;
581 decode_ws (dec); 933 return &PL_sv_undef;
934}
582 935
583 for (;;) 936static SV *
584 { 937decode_str (dec_t *dec, int utf8)
585 // heuristic: assume that 938{
586 // a) decode_str + hv_store_ent are abysmally slow. 939 SV *sv = 0;
587 // b) most hash keys are short, simple ascii text. 940
588 // => try to "fast-match" such strings to avoid 941 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
589 // the overhead of decode_str + hv_store_ent.
590 { 942 {
591 SV *value; 943 // indefinite length strings
592 char *p = dec->cur; 944 ++dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594 945
946 U8 major = *dec->cur & MAJOR_MISC;
947
948 sv = newSVpvn ("", 0);
949
595 for (;;) 950 for (;;)
596 { 951 {
597 // the >= 0x80 is false on most architectures 952 WANT (1);
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 953
954 if ((*dec->cur - major) > LENGTH_EXT8)
955 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 956 {
600 // slow path, back up and use decode_str 957 ++dec->cur;
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; 958 break;
619 } 959 }
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 { 960 else
685 dSP; 961 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
686 int count;
687 962
688 ENTER; SAVETMPS; PUSHMARK (SP); 963 STRLEN len = decode_uint (dec);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691 964
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
731fail:
732 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH;
734 return &PL_sv_undef;
735}
736
737static SV *
738decode_str (dec_t *dec, int utf8)
739{
740 SV *sv;
741
742 if ((*dec->cur & 31) == 31)
743 {
744 ++dec->cur;
745
746 sv = newSVpvn ("", 0);
747
748 // not very fast, and certainly not robust against illegal input
749 for (;;)
750 {
751 WANT (1); 965 WANT (len);
752 966 sv_catpvn (sv, dec->cur, len);
753 if (*dec->cur == (0xe0 | 31)) 967 dec->cur += len;
754 {
755 ++dec->cur;
756 break;
757 }
758
759 SV *sv2 = decode_sv (dec);
760 sv_catsv (sv, sv2);
761 } 968 }
762 } 969 }
763 else 970 else
764 { 971 {
765 STRLEN len = decode_uint (dec); 972 STRLEN len = decode_uint (dec);
766 973
767 WANT (len); 974 WANT (len);
768 sv = newSVpvn (dec->cur, len); 975 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 976 dec->cur += len;
977
978 if (ecb_expect_false (dec->stringref)
979 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
980 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 981 }
771 982
772 if (utf8) 983 if (utf8)
984 {
985 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
986 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
987 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
988
773 SvUTF8_on (sv); 989 SvUTF8_on (sv);
990 }
774 991
775 return sv; 992 return sv;
776 993
777fail: 994fail:
995 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 996 return &PL_sv_undef;
779} 997}
780 998
781static SV * 999static SV *
782decode_tagged (dec_t *dec) 1000decode_tagged (dec_t *dec)
783{ 1001{
1002 SV *sv = 0;
784 UV tag = decode_uint (dec); 1003 UV tag = decode_uint (dec);
1004
1005 WANT (1);
1006
1007 switch (tag)
1008 {
1009 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 1010 sv = decode_sv (dec);
1011 break;
786 1012
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1013 case CBOR_TAG_INDIRECTION:
1014 sv = newRV_noinc (decode_sv (dec));
1015 break;
1016
1017 case CBOR_TAG_STRINGREF_NAMESPACE:
1018 {
1019 // do not use SAVETMPS/FREETMPS, as these will
1020 // erase mortalised caches, e.g. "shareable"
1021 ENTER;
1022
1023 SAVESPTR (dec->stringref);
1024 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1025
1026 sv = decode_sv (dec);
1027
1028 LEAVE;
1029 }
1030 break;
1031
1032 case CBOR_TAG_STRINGREF:
1033 {
1034 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1035 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1036
1037 UV idx = decode_uint (dec);
1038
1039 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1040 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1041
1042 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1043 }
1044 break;
1045
1046 case CBOR_TAG_VALUE_SHAREABLE:
1047 {
1048 if (ecb_expect_false (!dec->shareable))
1049 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1050
1051 if (dec->cbor.flags & F_ALLOW_CYCLES)
1052 {
1053 sv = newSV (0);
1054 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1055
1056 SV *osv = decode_sv (dec);
1057 sv_setsv (sv, osv);
1058 SvREFCNT_dec_NN (osv);
1059 }
1060 else
1061 {
1062 av_push (dec->shareable, &PL_sv_undef);
1063 int idx = AvFILLp (dec->shareable);
1064 sv = decode_sv (dec);
1065 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1066 }
1067 }
1068 break;
1069
1070 case CBOR_TAG_VALUE_SHAREDREF:
1071 {
1072 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1073 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1074
1075 UV idx = decode_uint (dec);
1076
1077 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1078 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1079
1080 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1081
1082 if (sv == &PL_sv_undef)
1083 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1084 }
1085 break;
1086
1087 case CBOR_TAG_PERL_OBJECT:
1088 {
1089 if (dec->cbor.flags & F_FORBID_OBJECTS)
1090 goto filter;
1091
1092 sv = decode_sv (dec);
1093
1094 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1095 ERR ("corrupted CBOR data (non-array perl object)");
1096
1097 AV *av = (AV *)SvRV (sv);
1098 int len = av_len (av) + 1;
1099 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1100
1101 if (!stash)
1102 ERR ("cannot decode perl-object (package does not exist)");
1103
1104 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1105
1106 if (!method)
1107 ERR ("cannot decode perl-object (package does not have a THAW method)");
1108
1109 dSP;
1110
1111 ENTER; SAVETMPS;
1112 PUSHMARK (SP);
1113 EXTEND (SP, len + 1);
1114 // we re-bless the reference to get overload and other niceties right
1115 PUSHs (*av_fetch (av, 0, 1));
1116 PUSHs (sv_cbor);
1117
1118 int i;
1119
1120 for (i = 1; i < len; ++i)
1121 PUSHs (*av_fetch (av, i, 1));
1122
1123 PUTBACK;
1124 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1125 SPAGAIN;
1126
1127 if (SvTRUE (ERRSV))
1128 {
1129 FREETMPS; LEAVE;
1130 ERR_ERRSV;
1131 }
1132
1133 SvREFCNT_dec_NN (sv);
1134 sv = SvREFCNT_inc (POPs);
1135
1136 PUTBACK;
1137
1138 FREETMPS; LEAVE;
1139 }
1140 break;
1141
1142 default:
1143 filter:
1144 {
1145 SV *tag_sv = newSVuv (tag);
1146
1147 sv = decode_sv (dec);
1148
1149 dSP;
1150 ENTER; SAVETMPS;
1151 PUSHMARK (SP);
1152 EXTEND (SP, 2);
1153 PUSHs (tag_sv);
1154 PUSHs (sv);
1155
1156 PUTBACK;
1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1158 SPAGAIN;
1159
1160 if (SvTRUE (ERRSV))
1161 {
1162 SvREFCNT_dec_NN (tag_sv);
1163 FREETMPS; LEAVE;
1164 ERR_ERRSV;
1165 }
1166
1167 if (count)
1168 {
1169 SvREFCNT_dec_NN (tag_sv);
1170 SvREFCNT_dec_NN (sv);
1171 sv = SvREFCNT_inc_NN (TOPs);
1172 SP -= count;
1173 }
1174 else
1175 {
1176 AV *av = newAV ();
1177 av_push (av, tag_sv);
1178 av_push (av, sv);
1179
1180 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1181 ? cbor_tagged_stash
1182 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1183 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1184 }
1185
1186 PUTBACK;
1187
1188 FREETMPS; LEAVE;
1189 }
1190 break;
1191 }
1192
788 return sv; 1193 return sv;
789 1194
790 AV *av = newAV (); 1195fail:
791 av_push (av, newSVuv (tag)); 1196 SvREFCNT_dec (sv);
792 av_push (av, sv); 1197 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1198}
795 1199
796static SV * 1200static SV *
797decode_sv (dec_t *dec) 1201decode_sv (dec_t *dec)
798{ 1202{
799 WANT (1); 1203 WANT (1);
800 1204
801 switch (*dec->cur >> 5) 1205 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1206 {
803 case 0: // unsigned int 1207 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1208 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1209 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1210 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1211 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1212 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1213 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1214
811 return decode_str (dec, 1); 1215 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) 1216 switch (*dec->cur++ & MINOR_MASK)
820 { 1217 {
821 case 20: 1218 case SIMPLE_FALSE:
822#if CBOR_SLOW 1219#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1220 types_false = get_bool ("Types::Serialiser::false");
824#endif 1221#endif
825 return newSVsv (cbor_false); 1222 return newSVsv (types_false);
826 case 21: 1223 case SIMPLE_TRUE:
827#if CBOR_SLOW 1224#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1225 types_true = get_bool ("Types::Serialiser::true");
829#endif 1226#endif
830 return newSVsv (cbor_true); 1227 return newSVsv (types_true);
831 case 22: 1228 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1229 return newSVsv (&PL_sv_undef);
1230 case SIMPLE_UNDEF:
1231#if CBOR_SLOW
1232 types_error = get_bool ("Types::Serialiser::error");
1233#endif
1234 return newSVsv (types_error);
833 1235
834 case 25: 1236 case MISC_FLOAT16:
835 { 1237 {
836 WANT (2); 1238 WANT (2);
837 1239
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1240 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1241 dec->cur += 2;
840 1242
841 return newSVnv (ecb_binary16_to_float (fp)); 1243 return newSVnv (ecb_binary16_to_float (fp));
842 } 1244 }
843 1245
844 case 26: 1246 case MISC_FLOAT32:
845 { 1247 {
846 uint32_t fp; 1248 uint32_t fp;
847 WANT (4); 1249 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1250 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1251 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1254 fp = ecb_bswap32 (fp);
853 1255
854 return newSVnv (ecb_binary32_to_float (fp)); 1256 return newSVnv (ecb_binary32_to_float (fp));
855 } 1257 }
856 1258
857 case 27: 1259 case MISC_FLOAT64:
858 { 1260 {
859 uint64_t fp; 1261 uint64_t fp;
860 WANT (8); 1262 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1263 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1264 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1267 fp = ecb_bswap64 (fp);
866 1268
867 return newSVnv (ecb_binary64_to_double (fp)); 1269 return newSVnv (ecb_binary64_to_double (fp));
868 } 1270 }
869 1271
870 // 0..19 unassigned 1272 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1273 // 24 reserved + unassigned simple (reserved values are not encodable)
1274 // 28-30 unassigned misc
1275 // 31 break code
872 default: 1276 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1277 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1278 }
875 1279
876 break; 1280 break;
877 } 1281 }
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 1282
935fail: 1283fail:
936 return &PL_sv_undef; 1284 return &PL_sv_undef;
937} 1285}
938 1286
939static SV * 1287static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1288decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1289{
942 dec_t dec; 1290 dec_t dec = { 0 };
943 SV *sv; 1291 SV *sv;
1292 STRLEN len;
1293 char *data = SvPVbyte (string, len);
944 1294
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) 1295 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", 1296 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); 1297 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1298
980 dec.cbor = *cbor; 1299 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1300 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1301 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 1302
989 sv = decode_sv (&dec); 1303 sv = decode_sv (&dec);
990 1304
991 if (offset_return) 1305 if (offset_return)
992 *offset_return = dec.cur; 1306 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1309 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1310 dec.err = "garbage after CBOR object";
997 1311
998 if (dec.err) 1312 if (dec.err)
999 { 1313 {
1314 if (dec.shareable)
1315 {
1316 // need to break cyclic links, which would all be in shareable
1317 int i;
1318 SV **svp;
1319
1320 for (i = av_len (dec.shareable) + 1; i--; )
1321 if ((svp = av_fetch (dec.shareable, i, 0)))
1322 sv_setsv (*svp, &PL_sv_undef);
1323 }
1324
1000 SvREFCNT_dec (sv); 1325 SvREFCNT_dec_NN (sv);
1326
1327 if (dec.err_sv)
1328 sv_2mortal (dec.err_sv);
1329
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1330 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1331 }
1003 1332
1004 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1005 1334
1006 return sv; 1335 return sv;
1007} 1336}
1008 1337
1338/////////////////////////////////////////////////////////////////////////////
1339// incremental parser
1340
1341#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1342
1343// returns 0 for notyet, 1 for success or error
1344static int
1345incr_parse (CBOR *self, SV *cborstr)
1346{
1347 STRLEN cur;
1348 SvPV (cborstr, cur);
1349
1350 while (ecb_expect_true (self->incr_need <= cur))
1351 {
1352 // table of integer count bytes
1353 static I8 incr_len[MINOR_MASK + 1] = {
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 0, 0, 0, 0, 0, 0, 0, 0,
1356 0, 0, 0, 0, 0, 0, 0, 0,
1357 1, 2, 4, 8,-1,-1,-1,-2
1358 };
1359
1360 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1361 U8 m = *p & MINOR_MASK;
1362 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1363 I8 ilen = incr_len[m];
1364
1365 self->incr_need = self->incr_pos + 1;
1366
1367 if (ecb_expect_false (ilen < 0))
1368 {
1369 if (m != MINOR_INDEF)
1370 return 1; // error
1371
1372 if (*p == (MAJOR_MISC | MINOR_INDEF))
1373 {
1374 if (count >= 0)
1375 return 1; // error
1376
1377 count = 1;
1378 }
1379 else
1380 {
1381 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1382 count = -1;
1383 }
1384 }
1385 else
1386 {
1387 self->incr_need += ilen;
1388 if (ecb_expect_false (self->incr_need > cur))
1389 return 0;
1390
1391 int major = *p >> MAJOR_SHIFT;
1392
1393 switch (major)
1394 {
1395 case MAJOR_TAG >> MAJOR_SHIFT:
1396 ++count; // tags merely prefix another value
1397 break;
1398
1399 case MAJOR_BYTES >> MAJOR_SHIFT:
1400 case MAJOR_TEXT >> MAJOR_SHIFT:
1401 case MAJOR_ARRAY >> MAJOR_SHIFT:
1402 case MAJOR_MAP >> MAJOR_SHIFT:
1403 {
1404 UV len;
1405
1406 if (ecb_expect_false (ilen))
1407 {
1408 len = 0;
1409
1410 do {
1411 len = (len << 8) | *++p;
1412 } while (--ilen);
1413 }
1414 else
1415 len = m;
1416
1417 switch (major)
1418 {
1419 case MAJOR_BYTES >> MAJOR_SHIFT:
1420 case MAJOR_TEXT >> MAJOR_SHIFT:
1421 self->incr_need += len;
1422 if (ecb_expect_false (self->incr_need > cur))
1423 return 0;
1424
1425 break;
1426
1427 case MAJOR_MAP >> MAJOR_SHIFT:
1428 len <<= 1;
1429 case MAJOR_ARRAY >> MAJOR_SHIFT:
1430 if (len)
1431 {
1432 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1433 count = len + 1;
1434 }
1435 break;
1436 }
1437 }
1438 }
1439 }
1440
1441 self->incr_pos = self->incr_need;
1442
1443 if (count > 0)
1444 {
1445 while (!--count)
1446 {
1447 if (!AvFILLp (self->incr_count))
1448 return 1; // done
1449
1450 SvREFCNT_dec_NN (av_pop (self->incr_count));
1451 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1452 }
1453
1454 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1455 }
1456 }
1457
1458 return 0;
1459}
1460
1461
1009///////////////////////////////////////////////////////////////////////////// 1462/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1463// XS interface functions
1011 1464
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1465MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1466
1014BOOT: 1467BOOT:
1015{ 1468{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1469 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1470 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1471
1019 cbor_true = get_bool ("CBOR::XS::true"); 1472 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1473 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1474
1475 types_true = get_bool ("Types::Serialiser::true" );
1476 types_false = get_bool ("Types::Serialiser::false");
1477 types_error = get_bool ("Types::Serialiser::error");
1478
1479 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1480
1481 sv_cbor = newSVpv ("CBOR", 0);
1482 SvREADONLY_on (sv_cbor);
1483
1484 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1021} 1485}
1022 1486
1023PROTOTYPES: DISABLE 1487PROTOTYPES: DISABLE
1024 1488
1025void CLONE (...) 1489void CLONE (...)
1026 CODE: 1490 CODE:
1027 cbor_stash = 0; 1491 cbor_stash = 0;
1492 cbor_tagged_stash = 0;
1493 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1494 types_boolean_stash = 0;
1029 1495
1030void new (char *klass) 1496void new (char *klass)
1031 PPCODE: 1497 PPCODE:
1032{ 1498{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1499 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1507
1042void shrink (CBOR *self, int enable = 1) 1508void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1509 ALIAS:
1044 shrink = F_SHRINK 1510 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1511 allow_unknown = F_ALLOW_UNKNOWN
1512 allow_sharing = F_ALLOW_SHARING
1513 allow_cycles = F_ALLOW_CYCLES
1514 forbid_objects = F_FORBID_OBJECTS
1515 pack_strings = F_PACK_STRINGS
1516 text_keys = F_TEXT_KEYS
1517 text_strings = F_TEXT_STRINGS
1518 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1519 PPCODE:
1047{ 1520{
1048 if (enable) 1521 if (enable)
1049 self->flags |= ix; 1522 self->flags |= ix;
1050 else 1523 else
1055 1528
1056void get_shrink (CBOR *self) 1529void get_shrink (CBOR *self)
1057 ALIAS: 1530 ALIAS:
1058 get_shrink = F_SHRINK 1531 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1532 get_allow_unknown = F_ALLOW_UNKNOWN
1533 get_allow_sharing = F_ALLOW_SHARING
1534 get_allow_cycles = F_ALLOW_CYCLES
1535 get_forbid_objects = F_FORBID_OBJECTS
1536 get_pack_strings = F_PACK_STRINGS
1537 get_text_keys = F_TEXT_KEYS
1538 get_text_strings = F_TEXT_STRINGS
1539 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1540 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1541 XPUSHs (boolSV (self->flags & ix));
1062 1542
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1543void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1544 PPCODE:
1080 CODE: 1560 CODE:
1081 RETVAL = self->max_size; 1561 RETVAL = self->max_size;
1082 OUTPUT: 1562 OUTPUT:
1083 RETVAL 1563 RETVAL
1084 1564
1085#if 0 //TODO 1565void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1566 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1567 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1568 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1569 XPUSHs (ST (0));
1094}
1095 1570
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1571SV *get_filter (CBOR *self)
1097 PPCODE: 1572 CODE:
1098{ 1573 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1574 OUTPUT:
1100 self->cb_sk_object = newHV (); 1575 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 1576
1120void encode (CBOR *self, SV *scalar) 1577void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1578 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1579 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1580 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1593 EXTEND (SP, 2);
1137 PUSHs (sv); 1594 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1595 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1596}
1140 1597
1598void incr_parse (CBOR *self, SV *cborstr)
1599 ALIAS:
1600 incr_parse_multiple = 1
1601 PPCODE:
1602{
1603 if (SvUTF8 (cborstr))
1604 sv_utf8_downgrade (cborstr, 0);
1605
1606 if (!self->incr_count)
1607 {
1608 self->incr_count = newAV ();
1609 self->incr_pos = 0;
1610 self->incr_need = 1;
1611
1612 av_push (self->incr_count, newSViv (1));
1613 }
1614
1615 do
1616 {
1617 if (!incr_parse (self, cborstr))
1618 {
1619 if (self->incr_need > self->max_size && self->max_size)
1620 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1621 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1622
1623 break;
1624 }
1625
1626 SV *sv;
1627 char *offset;
1628
1629 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1630 XPUSHs (sv);
1631
1632 sv_chop (cborstr, offset);
1633
1634 av_clear (self->incr_count);
1635 av_push (self->incr_count, newSViv (1));
1636
1637 self->incr_pos = 0;
1638 self->incr_need = self->incr_pos + 1;
1639 }
1640 while (ix);
1641}
1642
1643void incr_reset (CBOR *self)
1644 CODE:
1645{
1646 SvREFCNT_dec (self->incr_count);
1647 self->incr_count = 0;
1648}
1649
1141void DESTROY (CBOR *self) 1650void DESTROY (CBOR *self)
1142 CODE: 1651 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1652 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1653
1146PROTOTYPES: ENABLE 1654PROTOTYPES: ENABLE
1147 1655
1148void encode_cbor (SV *scalar) 1656void encode_cbor (SV *scalar)
1657 ALIAS:
1658 encode_cbor = 0
1659 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1660 PPCODE:
1150{ 1661{
1151 CBOR cbor; 1662 CBOR cbor;
1152 cbor_init (&cbor); 1663 cbor_init (&cbor);
1664 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1665 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1666 XPUSHs (scalar);
1155} 1667}
1156 1668
1157void decode_cbor (SV *cborstr) 1669void decode_cbor (SV *cborstr)
1161 cbor_init (&cbor); 1673 cbor_init (&cbor);
1162 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1674 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1163 XPUSHs (cborstr); 1675 XPUSHs (cborstr);
1164} 1676}
1165 1677
1678#ifdef __AFL_COMPILER
1679
1680void
1681afl_init ()
1682 CODE:
1683 __AFL_INIT ();
1684
1685int
1686afl_loop (unsigned int count = 10000)
1687 CODE:
1688 RETVAL = __AFL_LOOP (count);
1689 OUTPUT:
1690 RETVAL
1691
1692#endif
1693

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines