… | |
… | |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
11 | |
11 | |
12 | #include "ecb.h" |
12 | #include "ecb.h" |
13 | |
13 | |
|
|
14 | // compatibility with perl <5.18 |
|
|
15 | #ifndef HvNAMELEN_get |
|
|
16 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
17 | #endif |
|
|
18 | #ifndef HvNAMELEN |
|
|
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
20 | #endif |
|
|
21 | #ifndef HvNAMEUTF8 |
|
|
22 | # define HvNAMEUTF8(hv) 0 |
|
|
23 | #endif |
|
|
24 | |
|
|
25 | // known tags |
|
|
26 | enum cbor_tag |
|
|
27 | { |
|
|
28 | // inofficial extensions (pending iana registration) |
|
|
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
|
|
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
|
|
31 | CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing |
|
|
32 | CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing |
|
|
33 | CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref |
|
|
34 | CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref |
|
|
35 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
|
|
36 | |
|
|
37 | // rfc7049 |
|
|
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
|
|
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
|
|
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
|
|
41 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
|
|
42 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
|
|
43 | CBOR_TAG_BIGFLOAT = 5, // array |
|
|
44 | |
|
|
45 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
|
|
46 | CBOR_TAG_CONV_B64 = 22, // base64, any |
|
|
47 | CBOR_TAG_CONV_HEX = 23, // base16, any |
|
|
48 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
|
|
49 | |
|
|
50 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
|
|
51 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
|
|
52 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
|
|
53 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
|
|
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
|
|
55 | |
|
|
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
|
|
57 | }; |
|
|
58 | |
14 | #define F_SHRINK 0x00000200UL |
59 | #define F_SHRINK 0x00000001UL |
15 | #define F_ALLOW_UNKNOWN 0x00002000UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
|
|
62 | #define F_DEDUP_STRINGS 0x00000008UL //TODO |
16 | |
63 | |
17 | #define INIT_SIZE 32 // initial scalar size to be allocated |
64 | #define INIT_SIZE 32 // initial scalar size to be allocated |
18 | |
65 | |
19 | #define SB do { |
66 | #define SB do { |
20 | #define SE } while (0) |
67 | #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 | |
68 | |
33 | #define IN_RANGE_INC(type,val,beg,end) \ |
69 | #define IN_RANGE_INC(type,val,beg,end) \ |
34 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
70 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
35 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
71 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
36 | |
72 | |
… | |
… | |
42 | #else |
78 | #else |
43 | # define CBOR_SLOW 0 |
79 | # define CBOR_SLOW 0 |
44 | # define CBOR_STASH cbor_stash |
80 | # define CBOR_STASH cbor_stash |
45 | #endif |
81 | #endif |
46 | |
82 | |
47 | static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: |
83 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
48 | static SV *cbor_true, *cbor_false; |
84 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
49 | |
85 | |
50 | typedef struct { |
86 | typedef struct { |
51 | U32 flags; |
87 | U32 flags; |
52 | U32 max_depth; |
88 | U32 max_depth; |
53 | STRLEN max_size; |
89 | STRLEN max_size; |
54 | |
|
|
55 | SV *cb_object; |
|
|
56 | HV *cb_sk_object; |
|
|
57 | } CBOR; |
90 | } CBOR; |
58 | |
91 | |
59 | INLINE void |
92 | ecb_inline void |
60 | cbor_init (CBOR *cbor) |
93 | cbor_init (CBOR *cbor) |
61 | { |
94 | { |
62 | Zero (cbor, 1, CBOR); |
95 | Zero (cbor, 1, CBOR); |
63 | cbor->max_depth = 512; |
96 | cbor->max_depth = 512; |
64 | } |
97 | } |
65 | |
98 | |
66 | ///////////////////////////////////////////////////////////////////////////// |
99 | ///////////////////////////////////////////////////////////////////////////// |
67 | // utility functions |
100 | // utility functions |
68 | |
101 | |
69 | INLINE SV * |
102 | ecb_inline SV * |
70 | get_bool (const char *name) |
103 | get_bool (const char *name) |
71 | { |
104 | { |
72 | SV *sv = get_sv (name, 1); |
105 | SV *sv = get_sv (name, 1); |
73 | |
106 | |
74 | SvREADONLY_on (sv); |
107 | SvREADONLY_on (sv); |
75 | SvREADONLY_on (SvRV (sv)); |
108 | SvREADONLY_on (SvRV (sv)); |
76 | |
109 | |
77 | return sv; |
110 | return sv; |
78 | } |
111 | } |
79 | |
112 | |
80 | INLINE void |
113 | ecb_inline void |
81 | shrink (SV *sv) |
114 | shrink (SV *sv) |
82 | { |
115 | { |
83 | sv_utf8_downgrade (sv, 1); |
116 | sv_utf8_downgrade (sv, 1); |
84 | |
117 | |
85 | if (SvLEN (sv) > SvCUR (sv) + 1) |
118 | if (SvLEN (sv) > SvCUR (sv) + 1) |
… | |
… | |
90 | SvPV_renew (sv, SvCUR (sv) + 1); |
123 | SvPV_renew (sv, SvCUR (sv) + 1); |
91 | #endif |
124 | #endif |
92 | } |
125 | } |
93 | } |
126 | } |
94 | |
127 | |
95 | ///////////////////////////////////////////////////////////////////////////// |
128 | // minimum length of a string to be registered for stringref |
96 | // fp hell |
129 | ecb_inline int |
97 | |
130 | minimum_string_length (UV idx) |
98 | //TODO |
131 | { |
|
|
132 | return idx > 23 |
|
|
133 | ? idx > 0xffU |
|
|
134 | ? idx > 0xffffU |
|
|
135 | ? idx > 0xffffffffU |
|
|
136 | ? 7 |
|
|
137 | : 6 |
|
|
138 | : 5 |
|
|
139 | : 4 |
|
|
140 | : 3; |
|
|
141 | } |
99 | |
142 | |
100 | ///////////////////////////////////////////////////////////////////////////// |
143 | ///////////////////////////////////////////////////////////////////////////// |
101 | // encoder |
144 | // encoder |
102 | |
145 | |
103 | // structure used for encoding CBOR |
146 | // structure used for encoding CBOR |
… | |
… | |
106 | char *cur; // SvPVX (sv) + current output position |
149 | char *cur; // SvPVX (sv) + current output position |
107 | char *end; // SvEND (sv) |
150 | char *end; // SvEND (sv) |
108 | SV *sv; // result scalar |
151 | SV *sv; // result scalar |
109 | CBOR cbor; |
152 | CBOR cbor; |
110 | U32 depth; // recursion level |
153 | U32 depth; // recursion level |
|
|
154 | HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
|
|
155 | UV stringref_idx; |
|
|
156 | HV *shareable; // ptr => index, or 0 |
|
|
157 | UV shareable_idx; |
111 | } enc_t; |
158 | } enc_t; |
112 | |
159 | |
113 | INLINE void |
160 | ecb_inline void |
114 | need (enc_t *enc, STRLEN len) |
161 | need (enc_t *enc, STRLEN len) |
115 | { |
162 | { |
116 | if (expect_false (enc->cur + len >= enc->end)) |
163 | if (ecb_expect_false (enc->cur + len >= enc->end)) |
117 | { |
164 | { |
118 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
165 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
119 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
166 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
120 | enc->cur = SvPVX (enc->sv) + cur; |
167 | enc->cur = SvPVX (enc->sv) + cur; |
121 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
168 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
122 | } |
169 | } |
123 | } |
170 | } |
124 | |
171 | |
125 | INLINE void |
172 | ecb_inline void |
126 | encode_ch (enc_t *enc, char ch) |
173 | encode_ch (enc_t *enc, char ch) |
127 | { |
174 | { |
128 | need (enc, 1); |
175 | need (enc, 1); |
129 | *enc->cur++ = ch; |
176 | *enc->cur++ = ch; |
130 | } |
177 | } |
… | |
… | |
167 | *enc->cur++ = len >> 8; |
214 | *enc->cur++ = len >> 8; |
168 | *enc->cur++ = len; |
215 | *enc->cur++ = len; |
169 | } |
216 | } |
170 | } |
217 | } |
171 | |
218 | |
|
|
219 | ecb_inline void |
|
|
220 | encode_tag (enc_t *enc, UV tag) |
|
|
221 | { |
|
|
222 | encode_uint (enc, 0xc0, tag); |
|
|
223 | } |
|
|
224 | |
172 | static void |
225 | static void |
173 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
226 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
174 | { |
227 | { |
|
|
228 | if (ecb_expect_false (enc->cbor.flags & F_DEDUP_STRINGS)) |
|
|
229 | { |
|
|
230 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
|
|
231 | |
|
|
232 | if (SvOK (*svp)) |
|
|
233 | { |
|
|
234 | // already registered, use stringref |
|
|
235 | encode_tag (enc, CBOR_TAG_STRINGREF); |
|
|
236 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
237 | return; |
|
|
238 | } |
|
|
239 | else if (len >= minimum_string_length (enc->stringref_idx)) |
|
|
240 | { |
|
|
241 | // register only |
|
|
242 | sv_setuv (*svp, enc->stringref_idx); |
|
|
243 | ++enc->stringref_idx; |
|
|
244 | } |
|
|
245 | } |
|
|
246 | |
175 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
247 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
176 | need (enc, len); |
248 | need (enc, len); |
177 | memcpy (enc->cur, str, len); |
249 | memcpy (enc->cur, str, len); |
178 | enc->cur += len; |
250 | enc->cur += len; |
179 | } |
251 | } |
… | |
… | |
199 | } |
271 | } |
200 | |
272 | |
201 | --enc->depth; |
273 | --enc->depth; |
202 | } |
274 | } |
203 | |
275 | |
|
|
276 | ecb_inline void |
|
|
277 | encode_he (enc_t *enc, HE *he) |
|
|
278 | { |
|
|
279 | } |
|
|
280 | |
204 | static void |
281 | static void |
205 | encode_hv (enc_t *enc, HV *hv) |
282 | encode_hv (enc_t *enc, HV *hv) |
206 | { |
283 | { |
207 | HE *he; |
284 | HE *he; |
208 | |
285 | |
… | |
… | |
224 | if (HeKLEN (he) == HEf_SVKEY) |
301 | if (HeKLEN (he) == HEf_SVKEY) |
225 | encode_sv (enc, HeSVKEY (he)); |
302 | encode_sv (enc, HeSVKEY (he)); |
226 | else |
303 | else |
227 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
304 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
228 | |
305 | |
229 | encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
306 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
230 | } |
307 | } |
231 | |
308 | |
232 | if (mg) |
309 | if (mg) |
233 | encode_ch (enc, 0xe0 | 31); |
310 | encode_ch (enc, 0xe0 | 31); |
234 | |
311 | |
… | |
… | |
237 | |
314 | |
238 | // encode objects, arrays and special \0=false and \1=true values. |
315 | // encode objects, arrays and special \0=false and \1=true values. |
239 | static void |
316 | static void |
240 | encode_rv (enc_t *enc, SV *sv) |
317 | encode_rv (enc_t *enc, SV *sv) |
241 | { |
318 | { |
242 | svtype svt; |
|
|
243 | |
|
|
244 | SvGETMAGIC (sv); |
319 | SvGETMAGIC (sv); |
|
|
320 | |
|
|
321 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
322 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
323 | { |
|
|
324 | if (!enc->shareable) |
|
|
325 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
326 | |
|
|
327 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
328 | |
|
|
329 | if (SvOK (*svp)) |
|
|
330 | { |
|
|
331 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
332 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
333 | return; |
|
|
334 | } |
|
|
335 | else |
|
|
336 | { |
|
|
337 | sv_setuv (*svp, enc->shareable_idx); |
|
|
338 | ++enc->shareable_idx; |
|
|
339 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
340 | } |
|
|
341 | } |
|
|
342 | |
245 | svt = SvTYPE (sv); |
343 | svtype svt = SvTYPE (sv); |
246 | |
344 | |
247 | if (expect_false (SvOBJECT (sv))) |
345 | if (ecb_expect_false (SvOBJECT (sv))) |
248 | { |
346 | { |
249 | HV *stash = !CBOR_SLOW || cbor_boolean_stash |
347 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
250 | ? cbor_boolean_stash |
348 | ? types_boolean_stash |
|
|
349 | : gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
350 | HV *error_stash = !CBOR_SLOW || types_error_stash |
|
|
351 | ? types_error_stash |
|
|
352 | : gv_stashpv ("Types::Serialiser::Error", 1); |
|
|
353 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
|
|
354 | ? cbor_tagged_stash |
251 | : gv_stashpv ("CBOR::XS::Boolean", 1); |
355 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
252 | |
356 | |
253 | if (SvSTASH (sv) == stash) |
357 | HV *stash = SvSTASH (sv); |
|
|
358 | GV *method; |
|
|
359 | |
|
|
360 | if (stash == boolean_stash) |
254 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
361 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
|
|
362 | else if (stash == error_stash) |
|
|
363 | encode_ch (enc, 0xe0 | 23); |
|
|
364 | else if (stash == tagged_stash) |
|
|
365 | { |
|
|
366 | if (svt != SVt_PVAV) |
|
|
367 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
|
|
368 | |
|
|
369 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
|
|
370 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
371 | } |
|
|
372 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
|
|
373 | { |
|
|
374 | dSP; |
|
|
375 | |
|
|
376 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
377 | // we re-bless the reference to get overload and other niceties right |
|
|
378 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
379 | |
|
|
380 | PUTBACK; |
|
|
381 | // G_SCALAR ensures that return value is 1 |
|
|
382 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
383 | SPAGAIN; |
|
|
384 | |
|
|
385 | // catch this surprisingly common error |
|
|
386 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
387 | croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
388 | |
|
|
389 | encode_sv (enc, POPs); |
|
|
390 | |
|
|
391 | PUTBACK; |
|
|
392 | |
|
|
393 | FREETMPS; LEAVE; |
|
|
394 | } |
|
|
395 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
|
|
396 | { |
|
|
397 | dSP; |
|
|
398 | |
|
|
399 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
400 | EXTEND (SP, 2); |
|
|
401 | // we re-bless the reference to get overload and other niceties right |
|
|
402 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
403 | PUSHs (sv_cbor); |
|
|
404 | |
|
|
405 | PUTBACK; |
|
|
406 | int count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
407 | SPAGAIN; |
|
|
408 | |
|
|
409 | // catch this surprisingly common error |
|
|
410 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
411 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
412 | |
|
|
413 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
|
|
414 | encode_uint (enc, 0x80, count + 1); |
|
|
415 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
|
|
416 | |
|
|
417 | while (count) |
|
|
418 | encode_sv (enc, SP[1 - count--]); |
|
|
419 | |
|
|
420 | PUTBACK; |
|
|
421 | |
|
|
422 | FREETMPS; LEAVE; |
|
|
423 | } |
255 | else |
424 | else |
256 | { |
|
|
257 | #if 0 //TODO |
|
|
258 | if (enc->cbor.flags & F_CONV_BLESSED) |
|
|
259 | { |
|
|
260 | // 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))); |
|
|
269 | |
|
|
270 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
271 | PUTBACK; |
|
|
272 | call_sv ((SV *)GvCV (to_cbor), G_SCALAR); |
|
|
273 | SPAGAIN; |
|
|
274 | |
|
|
275 | // catch this surprisingly common error |
|
|
276 | 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))); |
|
|
278 | |
|
|
279 | sv = POPs; |
|
|
280 | PUTBACK; |
|
|
281 | |
|
|
282 | encode_sv (enc, sv); |
|
|
283 | |
|
|
284 | FREETMPS; LEAVE; |
|
|
285 | } |
|
|
286 | else if (enc->cbor.flags & F_ALLOW_BLESSED) |
|
|
287 | encode_str (enc, "null", 4, 0); |
|
|
288 | else |
|
|
289 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", |
425 | croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it", |
290 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
291 | } |
|
|
292 | else if (enc->cbor.flags & F_ALLOW_BLESSED) |
|
|
293 | encode_str (enc, "null", 4, 0); |
|
|
294 | else |
|
|
295 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
296 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
426 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
297 | #endif |
|
|
298 | } |
|
|
299 | } |
427 | } |
300 | else if (svt == SVt_PVHV) |
428 | else if (svt == SVt_PVHV) |
301 | encode_hv (enc, (HV *)sv); |
429 | encode_hv (enc, (HV *)sv); |
302 | else if (svt == SVt_PVAV) |
430 | else if (svt == SVt_PVAV) |
303 | encode_av (enc, (AV *)sv); |
431 | 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 |
432 | else |
322 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
433 | { |
323 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
434 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
435 | encode_sv (enc, sv); |
|
|
436 | } |
324 | } |
437 | } |
325 | |
438 | |
326 | static void |
439 | static void |
327 | encode_nv (enc_t *enc, SV *sv) |
440 | encode_nv (enc_t *enc, SV *sv) |
328 | { |
441 | { |
329 | double nv = SvNVX (sv); |
442 | double nv = SvNVX (sv); |
330 | |
443 | |
331 | need (enc, 9); |
444 | need (enc, 9); |
332 | |
445 | |
333 | if (expect_false (nv == (U32)nv)) |
446 | if (ecb_expect_false (nv == (U32)nv)) |
334 | encode_uint (enc, 0x00, (U32)nv); |
447 | encode_uint (enc, 0x00, (U32)nv); |
335 | //TODO: maybe I32? |
448 | //TODO: maybe I32? |
336 | else if (expect_false (nv == (float)nv)) |
449 | else if (ecb_expect_false (nv == (float)nv)) |
337 | { |
450 | { |
338 | uint32_t fp = ecb_float_to_binary32 (nv); |
451 | uint32_t fp = ecb_float_to_binary32 (nv); |
339 | |
452 | |
340 | *enc->cur++ = 0xe0 | 26; |
453 | *enc->cur++ = 0xe0 | 26; |
341 | |
454 | |
… | |
… | |
393 | } |
506 | } |
394 | |
507 | |
395 | static SV * |
508 | static SV * |
396 | encode_cbor (SV *scalar, CBOR *cbor) |
509 | encode_cbor (SV *scalar, CBOR *cbor) |
397 | { |
510 | { |
398 | enc_t enc; |
511 | enc_t enc = { }; |
399 | |
512 | |
400 | enc.cbor = *cbor; |
513 | enc.cbor = *cbor; |
401 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
514 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
402 | enc.cur = SvPVX (enc.sv); |
515 | enc.cur = SvPVX (enc.sv); |
403 | enc.end = SvEND (enc.sv); |
516 | enc.end = SvEND (enc.sv); |
404 | enc.depth = 0; |
|
|
405 | |
517 | |
406 | SvPOK_only (enc.sv); |
518 | SvPOK_only (enc.sv); |
|
|
519 | |
|
|
520 | if (cbor->flags & F_DEDUP_STRINGS) |
|
|
521 | { |
|
|
522 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
|
|
523 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
524 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
525 | } |
|
|
526 | |
407 | encode_sv (&enc, scalar); |
527 | encode_sv (&enc, scalar); |
408 | |
528 | |
409 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
529 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
410 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
530 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
411 | |
531 | |
… | |
… | |
425 | U8 *end; // end of input string |
545 | U8 *end; // end of input string |
426 | const char *err; // parse error, if != 0 |
546 | const char *err; // parse error, if != 0 |
427 | CBOR cbor; |
547 | CBOR cbor; |
428 | U32 depth; // recursion depth |
548 | U32 depth; // recursion depth |
429 | U32 maxdepth; // recursion depth limit |
549 | U32 maxdepth; // recursion depth limit |
|
|
550 | AV *shareable; |
|
|
551 | AV *stringref; |
430 | } dec_t; |
552 | } dec_t; |
431 | |
553 | |
432 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
554 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
433 | |
555 | |
434 | #define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); |
556 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
435 | |
557 | |
436 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
558 | #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
437 | #define DEC_DEC_DEPTH --dec->depth |
559 | #define DEC_DEC_DEPTH --dec->depth |
438 | |
560 | |
439 | static UV |
561 | static UV |
… | |
… | |
528 | SvREFCNT_dec (av); |
650 | SvREFCNT_dec (av); |
529 | DEC_DEC_DEPTH; |
651 | DEC_DEC_DEPTH; |
530 | return &PL_sv_undef; |
652 | return &PL_sv_undef; |
531 | } |
653 | } |
532 | |
654 | |
|
|
655 | static void |
|
|
656 | decode_he (dec_t *dec, HV *hv) |
|
|
657 | { |
|
|
658 | // for speed reasons, we specialcase single-string |
|
|
659 | // byte or utf-8 strings as keys, but only when !stringref |
|
|
660 | |
|
|
661 | if (expect_true (!dec->stringref)) |
|
|
662 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
|
|
663 | { |
|
|
664 | I32 len = decode_uint (dec); |
|
|
665 | char *key = (char *)dec->cur; |
|
|
666 | |
|
|
667 | dec->cur += len; |
|
|
668 | |
|
|
669 | if (ecb_expect_false (dec->stringref)) |
|
|
670 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
671 | |
|
|
672 | hv_store (hv, key, len, decode_sv (dec), 0); |
|
|
673 | |
|
|
674 | return; |
|
|
675 | } |
|
|
676 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
|
|
677 | { |
|
|
678 | I32 len = decode_uint (dec); |
|
|
679 | char *key = (char *)dec->cur; |
|
|
680 | |
|
|
681 | dec->cur += len; |
|
|
682 | |
|
|
683 | if (ecb_expect_false (dec->stringref)) |
|
|
684 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
|
|
685 | |
|
|
686 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
687 | |
|
|
688 | return; |
|
|
689 | } |
|
|
690 | |
|
|
691 | SV *k = decode_sv (dec); |
|
|
692 | SV *v = decode_sv (dec); |
|
|
693 | |
|
|
694 | hv_store_ent (hv, k, v, 0); |
|
|
695 | SvREFCNT_dec (k); |
|
|
696 | } |
|
|
697 | |
533 | static SV * |
698 | static SV * |
534 | decode_hv (dec_t *dec) |
699 | decode_hv (dec_t *dec) |
535 | { |
700 | { |
536 | HV *hv = newHV (); |
701 | HV *hv = newHV (); |
537 | |
702 | |
… | |
… | |
549 | { |
714 | { |
550 | ++dec->cur; |
715 | ++dec->cur; |
551 | break; |
716 | break; |
552 | } |
717 | } |
553 | |
718 | |
554 | SV *k = decode_sv (dec); |
719 | decode_he (dec, hv); |
555 | SV *v = decode_sv (dec); |
|
|
556 | |
|
|
557 | hv_store_ent (hv, k, v, 0); |
|
|
558 | } |
720 | } |
559 | } |
721 | } |
560 | else |
722 | else |
561 | { |
723 | { |
562 | int len = decode_uint (dec); |
724 | int pairs = decode_uint (dec); |
563 | |
725 | |
564 | while (len--) |
726 | while (pairs--) |
565 | { |
727 | 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 | } |
728 | } |
572 | |
729 | |
573 | DEC_DEC_DEPTH; |
730 | DEC_DEC_DEPTH; |
574 | return newRV_noinc ((SV *)hv); |
731 | return newRV_noinc ((SV *)hv); |
575 | |
|
|
576 | #if 0 |
|
|
577 | SV *sv; |
|
|
578 | HV *hv = newHV (); |
|
|
579 | |
|
|
580 | DEC_INC_DEPTH; |
|
|
581 | decode_ws (dec); |
|
|
582 | |
|
|
583 | for (;;) |
|
|
584 | { |
|
|
585 | // heuristic: assume that |
|
|
586 | // a) decode_str + hv_store_ent are abysmally slow. |
|
|
587 | // b) most hash keys are short, simple ascii text. |
|
|
588 | // => try to "fast-match" such strings to avoid |
|
|
589 | // the overhead of decode_str + hv_store_ent. |
|
|
590 | { |
|
|
591 | SV *value; |
|
|
592 | char *p = dec->cur; |
|
|
593 | char *e = p + 24; // only try up to 24 bytes |
|
|
594 | |
|
|
595 | for (;;) |
|
|
596 | { |
|
|
597 | // the >= 0x80 is false on most architectures |
|
|
598 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
599 | { |
|
|
600 | // slow path, back up and use decode_str |
|
|
601 | SV *key = decode_str (dec); |
|
|
602 | if (!key) |
|
|
603 | goto fail; |
|
|
604 | |
|
|
605 | decode_ws (dec); EXPECT_CH (':'); |
|
|
606 | |
|
|
607 | decode_ws (dec); |
|
|
608 | value = decode_sv (dec); |
|
|
609 | if (!value) |
|
|
610 | { |
|
|
611 | SvREFCNT_dec (key); |
|
|
612 | goto fail; |
|
|
613 | } |
|
|
614 | |
|
|
615 | hv_store_ent (hv, key, value, 0); |
|
|
616 | SvREFCNT_dec (key); |
|
|
617 | |
|
|
618 | break; |
|
|
619 | } |
|
|
620 | else if (*p == '"') |
|
|
621 | { |
|
|
622 | // fast path, got a simple key |
|
|
623 | char *key = dec->cur; |
|
|
624 | int len = p - key; |
|
|
625 | dec->cur = p + 1; |
|
|
626 | |
|
|
627 | decode_ws (dec); EXPECT_CH (':'); |
|
|
628 | |
|
|
629 | decode_ws (dec); |
|
|
630 | value = decode_sv (dec); |
|
|
631 | if (!value) |
|
|
632 | goto fail; |
|
|
633 | |
|
|
634 | hv_store (hv, key, len, value, 0); |
|
|
635 | |
|
|
636 | break; |
|
|
637 | } |
|
|
638 | |
|
|
639 | ++p; |
|
|
640 | } |
|
|
641 | } |
|
|
642 | |
|
|
643 | decode_ws (dec); |
|
|
644 | |
|
|
645 | if (*dec->cur == '}') |
|
|
646 | { |
|
|
647 | ++dec->cur; |
|
|
648 | break; |
|
|
649 | } |
|
|
650 | |
|
|
651 | if (*dec->cur != ',') |
|
|
652 | ERR (", or } expected while parsing object/hash"); |
|
|
653 | |
|
|
654 | ++dec->cur; |
|
|
655 | |
|
|
656 | decode_ws (dec); |
|
|
657 | |
|
|
658 | if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED) |
|
|
659 | { |
|
|
660 | ++dec->cur; |
|
|
661 | break; |
|
|
662 | } |
|
|
663 | } |
|
|
664 | |
|
|
665 | DEC_DEC_DEPTH; |
|
|
666 | sv = newRV_noinc ((SV *)hv); |
|
|
667 | |
|
|
668 | // check filter callbacks |
|
|
669 | if (dec->cbor.flags & F_HOOK) |
|
|
670 | { |
|
|
671 | if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1) |
|
|
672 | { |
|
|
673 | HE *cb, *he; |
|
|
674 | |
|
|
675 | hv_iterinit (hv); |
|
|
676 | he = hv_iternext (hv); |
|
|
677 | hv_iterinit (hv); |
|
|
678 | |
|
|
679 | // the next line creates a mortal sv each time its called. |
|
|
680 | // might want to optimise this for common cases. |
|
|
681 | cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
682 | |
|
|
683 | if (cb) |
|
|
684 | { |
|
|
685 | dSP; |
|
|
686 | int count; |
|
|
687 | |
|
|
688 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
689 | XPUSHs (HeVAL (he)); |
|
|
690 | sv_2mortal (sv); |
|
|
691 | |
|
|
692 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
693 | |
|
|
694 | if (count == 1) |
|
|
695 | { |
|
|
696 | sv = newSVsv (POPs); |
|
|
697 | FREETMPS; LEAVE; |
|
|
698 | return sv; |
|
|
699 | } |
|
|
700 | |
|
|
701 | SvREFCNT_inc (sv); |
|
|
702 | FREETMPS; LEAVE; |
|
|
703 | } |
|
|
704 | } |
|
|
705 | |
|
|
706 | if (dec->cbor.cb_object) |
|
|
707 | { |
|
|
708 | dSP; |
|
|
709 | int count; |
|
|
710 | |
|
|
711 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
712 | XPUSHs (sv_2mortal (sv)); |
|
|
713 | |
|
|
714 | PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN; |
|
|
715 | |
|
|
716 | if (count == 1) |
|
|
717 | { |
|
|
718 | sv = newSVsv (POPs); |
|
|
719 | FREETMPS; LEAVE; |
|
|
720 | return sv; |
|
|
721 | } |
|
|
722 | |
|
|
723 | SvREFCNT_inc (sv); |
|
|
724 | FREETMPS; LEAVE; |
|
|
725 | } |
|
|
726 | } |
|
|
727 | |
|
|
728 | return sv; |
|
|
729 | #endif |
|
|
730 | |
732 | |
731 | fail: |
733 | fail: |
732 | SvREFCNT_dec (hv); |
734 | SvREFCNT_dec (hv); |
733 | DEC_DEC_DEPTH; |
735 | DEC_DEC_DEPTH; |
734 | return &PL_sv_undef; |
736 | return &PL_sv_undef; |
735 | } |
737 | } |
736 | |
738 | |
737 | static SV * |
739 | static SV * |
738 | decode_str (dec_t *dec, int utf8) |
740 | decode_str (dec_t *dec, int utf8) |
739 | { |
741 | { |
740 | SV *sv; |
742 | SV *sv = 0; |
741 | |
743 | |
742 | if ((*dec->cur & 31) == 31) |
744 | if ((*dec->cur & 31) == 31) |
743 | { |
745 | { |
744 | ++dec->cur; |
746 | ++dec->cur; |
745 | |
747 | |
… | |
… | |
754 | { |
756 | { |
755 | ++dec->cur; |
757 | ++dec->cur; |
756 | break; |
758 | break; |
757 | } |
759 | } |
758 | |
760 | |
759 | SV *sv2 = decode_sv (dec); |
|
|
760 | sv_catsv (sv, sv2); |
761 | sv_catsv (sv, decode_sv (dec)); |
761 | } |
762 | } |
762 | } |
763 | } |
763 | else |
764 | else |
764 | { |
765 | { |
765 | STRLEN len = decode_uint (dec); |
766 | STRLEN len = decode_uint (dec); |
… | |
… | |
770 | } |
771 | } |
771 | |
772 | |
772 | if (utf8) |
773 | if (utf8) |
773 | SvUTF8_on (sv); |
774 | SvUTF8_on (sv); |
774 | |
775 | |
|
|
776 | if (ecb_expect_false (dec->stringref) |
|
|
777 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
778 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
|
|
779 | |
775 | return sv; |
780 | return sv; |
776 | |
781 | |
777 | fail: |
782 | fail: |
|
|
783 | SvREFCNT_dec (sv); |
778 | return &PL_sv_undef; |
784 | return &PL_sv_undef; |
779 | } |
785 | } |
780 | |
786 | |
781 | static SV * |
787 | static SV * |
782 | decode_tagged (dec_t *dec) |
788 | decode_tagged (dec_t *dec) |
783 | { |
789 | { |
|
|
790 | SV *sv = 0; |
784 | UV tag = decode_uint (dec); |
791 | UV tag = decode_uint (dec); |
|
|
792 | |
|
|
793 | WANT (1); |
|
|
794 | |
|
|
795 | switch (tag) |
|
|
796 | { |
|
|
797 | case CBOR_TAG_MAGIC: |
|
|
798 | sv = decode_sv (dec); |
|
|
799 | break; |
|
|
800 | |
|
|
801 | case CBOR_TAG_INDIRECTION: |
|
|
802 | sv = newRV_noinc (decode_sv (dec)); |
|
|
803 | break; |
|
|
804 | |
|
|
805 | case CBOR_TAG_STRINGREF_NAMESPACE: |
|
|
806 | { |
|
|
807 | ENTER; SAVETMPS; |
|
|
808 | |
|
|
809 | SAVESPTR (dec->stringref); |
|
|
810 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
811 | |
|
|
812 | sv = decode_sv (dec); |
|
|
813 | |
|
|
814 | FREETMPS; LEAVE; |
|
|
815 | } |
|
|
816 | break; |
|
|
817 | |
|
|
818 | case CBOR_TAG_STRINGREF: |
|
|
819 | { |
|
|
820 | if ((*dec->cur >> 5) != 0) |
|
|
821 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
|
|
822 | |
|
|
823 | UV idx = decode_uint (dec); |
|
|
824 | |
|
|
825 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
|
|
826 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
|
|
827 | |
|
|
828 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
|
|
829 | } |
|
|
830 | break; |
|
|
831 | |
|
|
832 | case CBOR_TAG_VALUE_SHAREABLE: |
|
|
833 | { |
|
|
834 | if (ecb_expect_false (!dec->shareable)) |
|
|
835 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
836 | |
|
|
837 | sv = newSV (0); |
|
|
838 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
|
|
839 | |
785 | SV *sv = decode_sv (dec); |
840 | SV *osv = decode_sv (dec); |
|
|
841 | sv_setsv (sv, osv); |
|
|
842 | SvREFCNT_dec_NN (osv); |
|
|
843 | } |
|
|
844 | break; |
786 | |
845 | |
787 | if (tag == 55799) // 2.4.5 Self-Describe CBOR |
846 | case CBOR_TAG_VALUE_SHAREDREF: |
|
|
847 | { |
|
|
848 | if ((*dec->cur >> 5) != 0) |
|
|
849 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
|
|
850 | |
|
|
851 | UV idx = decode_uint (dec); |
|
|
852 | |
|
|
853 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
|
|
854 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
|
|
855 | |
|
|
856 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
857 | } |
|
|
858 | break; |
|
|
859 | |
|
|
860 | case CBOR_TAG_PERL_OBJECT: |
|
|
861 | { |
|
|
862 | sv = decode_sv (dec); |
|
|
863 | |
|
|
864 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
|
|
865 | ERR ("corrupted CBOR data (non-array perl object)"); |
|
|
866 | |
|
|
867 | AV *av = (AV *)SvRV (sv); |
|
|
868 | int len = av_len (av) + 1; |
|
|
869 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
|
|
870 | |
|
|
871 | if (!stash) |
|
|
872 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
873 | |
|
|
874 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
875 | |
|
|
876 | if (!method) |
|
|
877 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
878 | |
|
|
879 | dSP; |
|
|
880 | |
|
|
881 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
882 | EXTEND (SP, len + 1); |
|
|
883 | // we re-bless the reference to get overload and other niceties right |
|
|
884 | PUSHs (*av_fetch (av, 0, 1)); |
|
|
885 | PUSHs (sv_cbor); |
|
|
886 | |
|
|
887 | int i; |
|
|
888 | |
|
|
889 | for (i = 1; i < len; ++i) |
|
|
890 | PUSHs (*av_fetch (av, i, 1)); |
|
|
891 | |
|
|
892 | PUTBACK; |
|
|
893 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
|
|
894 | SPAGAIN; |
|
|
895 | |
|
|
896 | if (SvTRUE (ERRSV)) |
|
|
897 | { |
|
|
898 | FREETMPS; LEAVE; |
|
|
899 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
900 | } |
|
|
901 | |
|
|
902 | SvREFCNT_dec (sv); |
|
|
903 | sv = SvREFCNT_inc (POPs); |
|
|
904 | |
|
|
905 | PUTBACK; |
|
|
906 | |
|
|
907 | FREETMPS; LEAVE; |
|
|
908 | } |
|
|
909 | break; |
|
|
910 | |
|
|
911 | default: |
|
|
912 | { |
|
|
913 | sv = decode_sv (dec); |
|
|
914 | |
|
|
915 | AV *av = newAV (); |
|
|
916 | av_push (av, newSVuv (tag)); |
|
|
917 | av_push (av, sv); |
|
|
918 | |
|
|
919 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
|
|
920 | ? cbor_tagged_stash |
|
|
921 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
|
|
922 | |
|
|
923 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
924 | } |
|
|
925 | break; |
|
|
926 | } |
|
|
927 | |
788 | return sv; |
928 | return sv; |
789 | |
929 | |
790 | AV *av = newAV (); |
930 | fail: |
791 | av_push (av, newSVuv (tag)); |
931 | SvREFCNT_dec (sv); |
792 | av_push (av, sv); |
932 | return &PL_sv_undef; |
793 | return newRV_noinc ((SV *)av); |
|
|
794 | } |
933 | } |
795 | |
934 | |
796 | static SV * |
935 | static SV * |
797 | decode_sv (dec_t *dec) |
936 | decode_sv (dec_t *dec) |
798 | { |
937 | { |
799 | WANT (1); |
938 | WANT (1); |
800 | |
939 | |
801 | switch (*dec->cur >> 5) |
940 | switch (*dec->cur >> 5) |
802 | { |
941 | { |
803 | case 0: // unsigned int |
942 | case 0: // unsigned int |
804 | //TODO: 64 bit values on 3 2bit perls |
|
|
805 | return newSVuv (decode_uint (dec)); |
943 | return newSVuv (decode_uint (dec)); |
806 | case 1: // negative int |
944 | case 1: // negative int |
807 | return newSViv (-1 - (IV)decode_uint (dec)); |
945 | return newSViv (-1 - (IV)decode_uint (dec)); |
808 | case 2: // octet string |
946 | case 2: // octet string |
809 | return decode_str (dec, 0); |
947 | return decode_str (dec, 0); |
… | |
… | |
818 | case 7: // misc |
956 | case 7: // misc |
819 | switch (*dec->cur++ & 31) |
957 | switch (*dec->cur++ & 31) |
820 | { |
958 | { |
821 | case 20: |
959 | case 20: |
822 | #if CBOR_SLOW |
960 | #if CBOR_SLOW |
823 | cbor_false = get_bool ("CBOR::XS::false"); |
961 | types_false = get_bool ("Types::Serialiser::false"); |
824 | #endif |
962 | #endif |
825 | return newSVsv (cbor_false); |
963 | return newSVsv (types_false); |
826 | case 21: |
964 | case 21: |
827 | #if CBOR_SLOW |
965 | #if CBOR_SLOW |
828 | cbor_true = get_bool ("CBOR::XS::true"); |
966 | types_true = get_bool ("Types::Serialiser::true"); |
829 | #endif |
967 | #endif |
830 | return newSVsv (cbor_true); |
968 | return newSVsv (types_true); |
831 | case 22: |
969 | case 22: |
832 | return newSVsv (&PL_sv_undef); |
970 | return newSVsv (&PL_sv_undef); |
|
|
971 | case 23: |
|
|
972 | #if CBOR_SLOW |
|
|
973 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
974 | #endif |
|
|
975 | return newSVsv (types_error); |
833 | |
976 | |
834 | case 25: |
977 | case 25: |
835 | { |
978 | { |
836 | WANT (2); |
979 | WANT (2); |
837 | |
980 | |
… | |
… | |
873 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1016 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
874 | } |
1017 | } |
875 | |
1018 | |
876 | break; |
1019 | break; |
877 | } |
1020 | } |
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 | |
1021 | |
935 | fail: |
1022 | fail: |
936 | return &PL_sv_undef; |
1023 | return &PL_sv_undef; |
937 | } |
1024 | } |
938 | |
1025 | |
939 | static SV * |
1026 | static SV * |
940 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1027 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
941 | { |
1028 | { |
942 | dec_t dec; |
1029 | dec_t dec = { }; |
943 | SV *sv; |
1030 | SV *sv; |
|
|
1031 | STRLEN len; |
|
|
1032 | char *data = SvPVbyte (string, len); |
944 | |
1033 | |
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) |
1034 | 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", |
1035 | 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); |
1036 | (unsigned long)len, (unsigned long)cbor->max_size); |
976 | } |
|
|
977 | |
|
|
978 | sv_utf8_downgrade (string, 0); |
|
|
979 | |
1037 | |
980 | dec.cbor = *cbor; |
1038 | dec.cbor = *cbor; |
981 | dec.cur = (U8 *)SvPVX (string); |
1039 | dec.cur = (U8 *)data; |
982 | dec.end = (U8 *)SvEND (string); |
1040 | 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 | |
1041 | |
989 | sv = decode_sv (&dec); |
1042 | sv = decode_sv (&dec); |
990 | |
1043 | |
991 | if (offset_return) |
1044 | if (offset_return) |
992 | *offset_return = dec.cur; |
1045 | *offset_return = dec.cur; |
… | |
… | |
996 | dec.err = "garbage after CBOR object"; |
1049 | dec.err = "garbage after CBOR object"; |
997 | |
1050 | |
998 | if (dec.err) |
1051 | if (dec.err) |
999 | { |
1052 | { |
1000 | SvREFCNT_dec (sv); |
1053 | SvREFCNT_dec (sv); |
1001 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); |
1054 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1002 | } |
1055 | } |
1003 | |
1056 | |
1004 | sv = sv_2mortal (sv); |
1057 | sv = sv_2mortal (sv); |
1005 | |
1058 | |
1006 | return sv; |
1059 | return sv; |
… | |
… | |
1012 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1065 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1013 | |
1066 | |
1014 | BOOT: |
1067 | BOOT: |
1015 | { |
1068 | { |
1016 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
1069 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
1017 | cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); |
1070 | cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
1018 | |
1071 | |
1019 | cbor_true = get_bool ("CBOR::XS::true"); |
1072 | types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1020 | cbor_false = get_bool ("CBOR::XS::false"); |
1073 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
|
|
1074 | |
|
|
1075 | types_true = get_bool ("Types::Serialiser::true" ); |
|
|
1076 | types_false = get_bool ("Types::Serialiser::false"); |
|
|
1077 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
1078 | |
|
|
1079 | sv_cbor = newSVpv ("CBOR", 0); |
|
|
1080 | SvREADONLY_on (sv_cbor); |
1021 | } |
1081 | } |
1022 | |
1082 | |
1023 | PROTOTYPES: DISABLE |
1083 | PROTOTYPES: DISABLE |
1024 | |
1084 | |
1025 | void CLONE (...) |
1085 | void CLONE (...) |
1026 | CODE: |
1086 | CODE: |
1027 | cbor_stash = 0; |
1087 | cbor_stash = 0; |
|
|
1088 | cbor_tagged_stash = 0; |
|
|
1089 | types_error_stash = 0; |
1028 | cbor_boolean_stash = 0; |
1090 | types_boolean_stash = 0; |
1029 | |
1091 | |
1030 | void new (char *klass) |
1092 | void new (char *klass) |
1031 | PPCODE: |
1093 | PPCODE: |
1032 | { |
1094 | { |
1033 | SV *pv = NEWSV (0, sizeof (CBOR)); |
1095 | SV *pv = NEWSV (0, sizeof (CBOR)); |
… | |
… | |
1041 | |
1103 | |
1042 | void shrink (CBOR *self, int enable = 1) |
1104 | void shrink (CBOR *self, int enable = 1) |
1043 | ALIAS: |
1105 | ALIAS: |
1044 | shrink = F_SHRINK |
1106 | shrink = F_SHRINK |
1045 | allow_unknown = F_ALLOW_UNKNOWN |
1107 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1108 | allow_sharing = F_ALLOW_SHARING |
|
|
1109 | dedup_strings = F_DEDUP_STRINGS |
1046 | PPCODE: |
1110 | PPCODE: |
1047 | { |
1111 | { |
1048 | if (enable) |
1112 | if (enable) |
1049 | self->flags |= ix; |
1113 | self->flags |= ix; |
1050 | else |
1114 | else |
… | |
… | |
1055 | |
1119 | |
1056 | void get_shrink (CBOR *self) |
1120 | void get_shrink (CBOR *self) |
1057 | ALIAS: |
1121 | ALIAS: |
1058 | get_shrink = F_SHRINK |
1122 | get_shrink = F_SHRINK |
1059 | get_allow_unknown = F_ALLOW_UNKNOWN |
1123 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1124 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1125 | get_dedup_strings = F_DEDUP_STRINGS |
1060 | PPCODE: |
1126 | PPCODE: |
1061 | XPUSHs (boolSV (self->flags & ix)); |
1127 | XPUSHs (boolSV (self->flags & ix)); |
1062 | |
1128 | |
1063 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1129 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1064 | PPCODE: |
1130 | PPCODE: |
… | |
… | |
1079 | int get_max_size (CBOR *self) |
1145 | int get_max_size (CBOR *self) |
1080 | CODE: |
1146 | CODE: |
1081 | RETVAL = self->max_size; |
1147 | RETVAL = self->max_size; |
1082 | OUTPUT: |
1148 | OUTPUT: |
1083 | RETVAL |
1149 | RETVAL |
1084 | |
|
|
1085 | #if 0 //TODO |
|
|
1086 | |
|
|
1087 | void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef) |
|
|
1088 | PPCODE: |
|
|
1089 | { |
|
|
1090 | SvREFCNT_dec (self->cb_object); |
|
|
1091 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
1092 | |
|
|
1093 | XPUSHs (ST (0)); |
|
|
1094 | } |
|
|
1095 | |
|
|
1096 | void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
1097 | PPCODE: |
|
|
1098 | { |
|
|
1099 | if (!self->cb_sk_object) |
|
|
1100 | self->cb_sk_object = newHV (); |
|
|
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 | |
1150 | |
1120 | void encode (CBOR *self, SV *scalar) |
1151 | void encode (CBOR *self, SV *scalar) |
1121 | PPCODE: |
1152 | PPCODE: |
1122 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1153 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1123 | XPUSHs (scalar); |
1154 | XPUSHs (scalar); |
… | |
… | |
1136 | EXTEND (SP, 2); |
1167 | EXTEND (SP, 2); |
1137 | PUSHs (sv); |
1168 | PUSHs (sv); |
1138 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1169 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1139 | } |
1170 | } |
1140 | |
1171 | |
1141 | void DESTROY (CBOR *self) |
|
|
1142 | CODE: |
|
|
1143 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1144 | SvREFCNT_dec (self->cb_object); |
|
|
1145 | |
|
|
1146 | PROTOTYPES: ENABLE |
1172 | PROTOTYPES: ENABLE |
1147 | |
1173 | |
1148 | void encode_cbor (SV *scalar) |
1174 | void encode_cbor (SV *scalar) |
1149 | PPCODE: |
1175 | PPCODE: |
1150 | { |
1176 | { |