diff options
Diffstat (limited to 'src/fs/fs_uri.c')
-rw-r--r-- | src/fs/fs_uri.c | 2055 |
1 files changed, 0 insertions, 2055 deletions
diff --git a/src/fs/fs_uri.c b/src/fs/fs_uri.c deleted file mode 100644 index 73ea5d60d..000000000 --- a/src/fs/fs_uri.c +++ /dev/null | |||
@@ -1,2055 +0,0 @@ | |||
1 | /* | ||
2 | This file is part of GNUnet. | ||
3 | Copyright (C) 2003--2014 GNUnet e.V. | ||
4 | |||
5 | GNUnet is free software: you can redistribute it and/or modify it | ||
6 | under the terms of the GNU Affero General Public License as published | ||
7 | by the Free Software Foundation, either version 3 of the License, | ||
8 | or (at your option) any later version. | ||
9 | |||
10 | GNUnet is distributed in the hope that it will be useful, but | ||
11 | WITHOUT ANY WARRANTY; without even the implied warranty of | ||
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
13 | Affero General Public License for more details. | ||
14 | |||
15 | You should have received a copy of the GNU Affero General Public License | ||
16 | along with this program. If not, see <http://www.gnu.org/licenses/>. | ||
17 | |||
18 | SPDX-License-Identifier: AGPL3.0-or-later | ||
19 | */ | ||
20 | |||
21 | /** | ||
22 | * @file fs/fs_uri.c | ||
23 | * @brief Parses and produces uri strings. | ||
24 | * @author Igor Wronsky, Christian Grothoff | ||
25 | * | ||
26 | * GNUnet URIs are of the general form "gnunet://MODULE/IDENTIFIER". | ||
27 | * The specific structure of "IDENTIFIER" depends on the module and | ||
28 | * maybe differentiated into additional subcategories if applicable. | ||
29 | * This module only deals with fs identifiers (MODULE = "fs"). | ||
30 | * <p> | ||
31 | * | ||
32 | * This module only parses URIs for the AFS module. The FS URIs fall | ||
33 | * into four categories, "chk", "sks", "ksk" and "loc". The first three | ||
34 | * categories were named in analogy (!) to Freenet, but they do NOT | ||
35 | * work in exactly the same way. They are very similar from the user's | ||
36 | * point of view (unique file identifier, subspace, keyword), but the | ||
37 | * implementation is rather different in pretty much every detail. | ||
38 | * The concrete URI formats are: | ||
39 | * | ||
40 | * </p> | ||
41 | * | ||
42 | * <ul><li> | ||
43 | * | ||
44 | * First, there are URIs that identify a file. They have the format | ||
45 | * "gnunet://fs/chk/HEX1.HEX2.SIZE". These URIs can be used to | ||
46 | * download the file. The description, filename, mime-type and other | ||
47 | * meta-data is NOT part of the file-URI since a URI uniquely | ||
48 | * identifies a resource (and the contents of the file would be the | ||
49 | * same even if it had a different description). | ||
50 | * | ||
51 | * </li><li> | ||
52 | * | ||
53 | * The second category identifies entries in a namespace. The format | ||
54 | * is "gnunet://fs/sks/NAMESPACE/IDENTIFIER" where the namespace | ||
55 | * should be given in HEX. Applications may allow using a nickname | ||
56 | * for the namespace if the nickname is not ambiguous. The identifier | ||
57 | * can be either an ASCII sequence or a HEX-encoding. If the | ||
58 | * identifier is in ASCII but the format is ambiguous and could denote | ||
59 | * a HEX-string a "/" is appended to indicate ASCII encoding. | ||
60 | * | ||
61 | * </li> <li> | ||
62 | * | ||
63 | * The third category identifies ordinary searches. The format is | ||
64 | * "gnunet://fs/ksk/KEYWORD[+KEYWORD]*". Using the "+" syntax | ||
65 | * it is possible to encode searches with the boolean "AND" operator. | ||
66 | * "+" is used since it indicates a commutative 'and' operation and | ||
67 | * is unlikely to be used in a keyword by itself. | ||
68 | * | ||
69 | * </li><li> | ||
70 | * | ||
71 | * The last category identifies a datum on a specific machine. The | ||
72 | * format is "gnunet://fs/loc/HEX1.HEX2.SIZE.PEER.SIG.EXPTIME". PEER is | ||
73 | * the BinName of the public key of the peer storing the datum. The | ||
74 | * signature (SIG) certifies that this peer has this content. | ||
75 | * HEX1, HEX2 and SIZE correspond to a 'chk' URI. | ||
76 | * | ||
77 | * </li></ul> | ||
78 | * | ||
79 | * The encoding for hexadecimal values is defined in the hashing.c | ||
80 | * module in the gnunetutil library and discussed there. | ||
81 | * | ||
82 | */ | ||
83 | #include "platform.h" | ||
84 | #include "gnunet_fs_service.h" | ||
85 | #include "gnunet_signatures.h" | ||
86 | #include "fs_api.h" | ||
87 | #include <unitypes.h> | ||
88 | #include <unicase.h> | ||
89 | #include <uniconv.h> | ||
90 | #include <unistr.h> | ||
91 | #include <unistdio.h> | ||
92 | |||
93 | |||
94 | /** | ||
95 | * Get a unique key from a URI. This is for putting URIs | ||
96 | * into HashMaps. The key may change between FS implementations. | ||
97 | * | ||
98 | * @param uri uri to convert to a unique key | ||
99 | * @param key where to store the unique key | ||
100 | * @return #GNUNET_OK on success | ||
101 | */ | ||
102 | int | ||
103 | GNUNET_FS_uri_to_key (const struct GNUNET_FS_Uri *uri, | ||
104 | struct GNUNET_HashCode *key) | ||
105 | { | ||
106 | switch (uri->type) | ||
107 | { | ||
108 | case GNUNET_FS_URI_CHK: | ||
109 | *key = uri->data.chk.chk.query; | ||
110 | return GNUNET_OK; | ||
111 | |||
112 | case GNUNET_FS_URI_SKS: | ||
113 | GNUNET_CRYPTO_hash (uri->data.sks.identifier, | ||
114 | strlen (uri->data.sks.identifier), | ||
115 | key); | ||
116 | return GNUNET_OK; | ||
117 | |||
118 | case GNUNET_FS_URI_KSK: | ||
119 | if (uri->data.ksk.keywordCount > 0) | ||
120 | { | ||
121 | GNUNET_CRYPTO_hash (uri->data.ksk.keywords[0], | ||
122 | strlen (uri->data.ksk.keywords[0]), | ||
123 | key); | ||
124 | return GNUNET_OK; | ||
125 | } | ||
126 | else | ||
127 | { | ||
128 | memset (key, 0, sizeof(struct GNUNET_HashCode)); | ||
129 | return GNUNET_SYSERR; | ||
130 | } | ||
131 | break; | ||
132 | |||
133 | case GNUNET_FS_URI_LOC: | ||
134 | GNUNET_CRYPTO_hash (&uri->data.loc.fi, | ||
135 | sizeof(struct FileIdentifier) | ||
136 | + sizeof(struct GNUNET_PeerIdentity), | ||
137 | key); | ||
138 | return GNUNET_OK; | ||
139 | |||
140 | default: | ||
141 | memset (key, 0, sizeof(struct GNUNET_HashCode)); | ||
142 | return GNUNET_SYSERR; | ||
143 | } | ||
144 | } | ||
145 | |||
146 | |||
147 | /** | ||
148 | * Convert keyword URI to a human readable format | ||
149 | * (i.e. the search query that was used in the first place) | ||
150 | * | ||
151 | * @param uri ksk uri to convert to a string | ||
152 | * @return string with the keywords | ||
153 | */ | ||
154 | char * | ||
155 | GNUNET_FS_uri_ksk_to_string_fancy (const struct GNUNET_FS_Uri *uri) | ||
156 | { | ||
157 | size_t n; | ||
158 | char *ret; | ||
159 | unsigned int i; | ||
160 | const char *keyword; | ||
161 | char **keywords; | ||
162 | unsigned int keywordCount; | ||
163 | |||
164 | if ((NULL == uri) || (GNUNET_FS_URI_KSK != uri->type)) | ||
165 | { | ||
166 | GNUNET_break (0); | ||
167 | return NULL; | ||
168 | } | ||
169 | keywords = uri->data.ksk.keywords; | ||
170 | keywordCount = uri->data.ksk.keywordCount; | ||
171 | n = keywordCount + 1; | ||
172 | for (i = 0; i < keywordCount; i++) | ||
173 | { | ||
174 | keyword = keywords[i]; | ||
175 | n += strlen (keyword) - 1; | ||
176 | if (NULL != strstr (&keyword[1], " ")) | ||
177 | n += 2; | ||
178 | if (keyword[0] == '+') | ||
179 | n++; | ||
180 | } | ||
181 | ret = GNUNET_malloc (n); | ||
182 | strcpy (ret, ""); | ||
183 | for (i = 0; i < keywordCount; i++) | ||
184 | { | ||
185 | keyword = keywords[i]; | ||
186 | if (NULL != strstr (&keyword[1], " ")) | ||
187 | { | ||
188 | strcat (ret, "\""); | ||
189 | if (keyword[0] == '+') | ||
190 | strcat (ret, keyword); | ||
191 | else | ||
192 | strcat (ret, &keyword[1]); | ||
193 | strcat (ret, "\""); | ||
194 | } | ||
195 | else | ||
196 | { | ||
197 | if (keyword[0] == '+') | ||
198 | strcat (ret, keyword); | ||
199 | else | ||
200 | strcat (ret, &keyword[1]); | ||
201 | } | ||
202 | strcat (ret, " "); | ||
203 | } | ||
204 | return ret; | ||
205 | } | ||
206 | |||
207 | |||
208 | /** | ||
209 | * Given a keyword with %-encoding (and possibly quotes to protect | ||
210 | * spaces), return a copy of the keyword without %-encoding and | ||
211 | * without double-quotes (%22). Also, add a space at the beginning | ||
212 | * if there is not a '+'. | ||
213 | * | ||
214 | * @param in string with %-encoding | ||
215 | * @param emsg where to store the parser error message (if any) | ||
216 | * @return decoded string with leading space (or preserved plus) | ||
217 | */ | ||
218 | static char * | ||
219 | percent_decode_keyword (const char *in, char **emsg) | ||
220 | { | ||
221 | char *out; | ||
222 | char *ret; | ||
223 | unsigned int rpos; | ||
224 | unsigned int wpos; | ||
225 | unsigned int hx; | ||
226 | |||
227 | out = GNUNET_strdup (in); | ||
228 | rpos = 0; | ||
229 | wpos = 0; | ||
230 | while (out[rpos] != '\0') | ||
231 | { | ||
232 | if (out[rpos] == '%') | ||
233 | { | ||
234 | if (1 != sscanf (&out[rpos + 1], "%2X", &hx)) | ||
235 | { | ||
236 | GNUNET_free (out); | ||
237 | *emsg = GNUNET_strdup ( | ||
238 | _ ( /* xgettext:no-c-format */ | ||
239 | "Malformed KSK URI (`%' must be followed by HEX number)")); | ||
240 | return NULL; | ||
241 | } | ||
242 | rpos += 3; | ||
243 | if (hx == '"') | ||
244 | continue; /* skip double quote */ | ||
245 | out[wpos++] = (char) hx; | ||
246 | } | ||
247 | else | ||
248 | { | ||
249 | out[wpos++] = out[rpos++]; | ||
250 | } | ||
251 | } | ||
252 | out[wpos] = '\0'; | ||
253 | if (out[0] == '+') | ||
254 | { | ||
255 | ret = GNUNET_strdup (out); | ||
256 | } | ||
257 | else | ||
258 | { | ||
259 | /* need to prefix with space */ | ||
260 | ret = GNUNET_malloc (strlen (out) + 2); | ||
261 | strcpy (ret, " "); | ||
262 | strcat (ret, out); | ||
263 | } | ||
264 | GNUNET_free (out); | ||
265 | return ret; | ||
266 | } | ||
267 | |||
268 | |||
269 | #define GNUNET_FS_URI_KSK_PREFIX GNUNET_FS_URI_PREFIX GNUNET_FS_URI_KSK_INFIX | ||
270 | |||
271 | /** | ||
272 | * Parse a KSK URI. | ||
273 | * | ||
274 | * @param s an uri string | ||
275 | * @param emsg where to store the parser error message (if any) | ||
276 | * @return NULL on error, otherwise the KSK URI | ||
277 | */ | ||
278 | static struct GNUNET_FS_Uri * | ||
279 | uri_ksk_parse (const char *s, char **emsg) | ||
280 | { | ||
281 | struct GNUNET_FS_Uri *ret; | ||
282 | char **keywords; | ||
283 | unsigned int pos; | ||
284 | int max; | ||
285 | int iret; | ||
286 | int i; | ||
287 | size_t slen; | ||
288 | char *dup; | ||
289 | int saw_quote; | ||
290 | |||
291 | slen = strlen (s); | ||
292 | pos = strlen (GNUNET_FS_URI_KSK_PREFIX); | ||
293 | if ((slen <= pos) || (0 != strncmp (s, GNUNET_FS_URI_KSK_PREFIX, pos))) | ||
294 | return NULL; /* not KSK URI */ | ||
295 | if ((s[slen - 1] == '+') || (s[pos] == '+')) | ||
296 | { | ||
297 | *emsg = | ||
298 | GNUNET_strdup (_ ("Malformed KSK URI (must not begin or end with `+')")); | ||
299 | return NULL; | ||
300 | } | ||
301 | max = 1; | ||
302 | saw_quote = 0; | ||
303 | for (i = pos; i < slen; i++) | ||
304 | { | ||
305 | if ((s[i] == '%') && (&s[i] == strstr (&s[i], "%22"))) | ||
306 | { | ||
307 | saw_quote = (saw_quote + 1) % 2; | ||
308 | i += 3; | ||
309 | continue; | ||
310 | } | ||
311 | if ((s[i] == '+') && (saw_quote == 0)) | ||
312 | { | ||
313 | max++; | ||
314 | if (s[i - 1] == '+') | ||
315 | { | ||
316 | *emsg = GNUNET_strdup (_ ("Malformed KSK URI (`++' not allowed)")); | ||
317 | return NULL; | ||
318 | } | ||
319 | } | ||
320 | } | ||
321 | if (saw_quote == 1) | ||
322 | { | ||
323 | *emsg = GNUNET_strdup (_ ("Malformed KSK URI (quotes not balanced)")); | ||
324 | return NULL; | ||
325 | } | ||
326 | iret = max; | ||
327 | dup = GNUNET_strdup (s); | ||
328 | keywords = GNUNET_new_array (max, char *); | ||
329 | for (i = slen - 1; i >= (int) pos; i--) | ||
330 | { | ||
331 | if ((s[i] == '%') && (&s[i] == strstr (&s[i], "%22"))) | ||
332 | { | ||
333 | saw_quote = (saw_quote + 1) % 2; | ||
334 | continue; | ||
335 | } | ||
336 | if ((dup[i] == '+') && (saw_quote == 0)) | ||
337 | { | ||
338 | keywords[--max] = percent_decode_keyword (&dup[i + 1], emsg); | ||
339 | if (NULL == keywords[max]) | ||
340 | goto CLEANUP; | ||
341 | dup[i] = '\0'; | ||
342 | } | ||
343 | } | ||
344 | keywords[--max] = percent_decode_keyword (&dup[pos], emsg); | ||
345 | if (NULL == keywords[max]) | ||
346 | goto CLEANUP; | ||
347 | GNUNET_assert (0 == max); | ||
348 | GNUNET_free (dup); | ||
349 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
350 | ret->type = GNUNET_FS_URI_KSK; | ||
351 | ret->data.ksk.keywordCount = iret; | ||
352 | ret->data.ksk.keywords = keywords; | ||
353 | return ret; | ||
354 | CLEANUP: | ||
355 | for (i = 0; i < max; i++) | ||
356 | GNUNET_free (keywords[i]); | ||
357 | GNUNET_free (keywords); | ||
358 | GNUNET_free (dup); | ||
359 | return NULL; | ||
360 | } | ||
361 | |||
362 | |||
363 | #define GNUNET_FS_URI_SKS_PREFIX GNUNET_FS_URI_PREFIX GNUNET_FS_URI_SKS_INFIX | ||
364 | |||
365 | /** | ||
366 | * Parse an SKS URI. | ||
367 | * | ||
368 | * @param s an uri string | ||
369 | * @param emsg where to store the parser error message (if any) | ||
370 | * @return NULL on error, SKS URI otherwise | ||
371 | */ | ||
372 | static struct GNUNET_FS_Uri * | ||
373 | uri_sks_parse (const char *s, char **emsg) | ||
374 | { | ||
375 | struct GNUNET_FS_Uri *ret; | ||
376 | struct GNUNET_CRYPTO_EcdsaPublicKey ns; | ||
377 | size_t pos; | ||
378 | char *end; | ||
379 | |||
380 | pos = strlen (GNUNET_FS_URI_SKS_PREFIX); | ||
381 | if ((strlen (s) <= pos) || (0 != strncmp (s, GNUNET_FS_URI_SKS_PREFIX, pos))) | ||
382 | return NULL; /* not an SKS URI */ | ||
383 | end = strchr (&s[pos], '/'); | ||
384 | if ((NULL == end) || | ||
385 | (GNUNET_OK != GNUNET_STRINGS_string_to_data (&s[pos], | ||
386 | end - &s[pos], | ||
387 | &ns, | ||
388 | sizeof(ns)))) | ||
389 | { | ||
390 | *emsg = GNUNET_strdup (_ ("Malformed SKS URI (wrong syntax)")); | ||
391 | return NULL; /* malformed */ | ||
392 | } | ||
393 | end++; /* skip over '/' */ | ||
394 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
395 | ret->type = GNUNET_FS_URI_SKS; | ||
396 | ret->data.sks.ns = ns; | ||
397 | ret->data.sks.identifier = GNUNET_strdup (end); | ||
398 | return ret; | ||
399 | } | ||
400 | |||
401 | |||
402 | #define GNUNET_FS_URI_CHK_PREFIX GNUNET_FS_URI_PREFIX GNUNET_FS_URI_CHK_INFIX | ||
403 | |||
404 | |||
405 | /** | ||
406 | * Parse a CHK URI. | ||
407 | * | ||
408 | * @param s an uri string | ||
409 | * @param emsg where to store the parser error message (if any) | ||
410 | * @return NULL on error, CHK URI otherwise | ||
411 | */ | ||
412 | static struct GNUNET_FS_Uri * | ||
413 | uri_chk_parse (const char *s, char **emsg) | ||
414 | { | ||
415 | struct GNUNET_FS_Uri *ret; | ||
416 | struct FileIdentifier fi; | ||
417 | unsigned int pos; | ||
418 | unsigned long long flen; | ||
419 | size_t slen; | ||
420 | char h1[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)]; | ||
421 | char h2[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)]; | ||
422 | |||
423 | slen = strlen (s); | ||
424 | pos = strlen (GNUNET_FS_URI_CHK_PREFIX); | ||
425 | if ((slen < pos + 2 * sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) + 1) || | ||
426 | (0 != strncmp (s, GNUNET_FS_URI_CHK_PREFIX, pos))) | ||
427 | return NULL; /* not a CHK URI */ | ||
428 | if ((s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] != '.') || | ||
429 | (s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) * 2 - 1] != '.')) | ||
430 | { | ||
431 | *emsg = GNUNET_strdup (_ ("Malformed CHK URI (wrong syntax)")); | ||
432 | return NULL; | ||
433 | } | ||
434 | GNUNET_memcpy (h1, &s[pos], sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)); | ||
435 | h1[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] = '\0'; | ||
436 | GNUNET_memcpy (h2, | ||
437 | &s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)], | ||
438 | sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)); | ||
439 | h2[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] = '\0'; | ||
440 | |||
441 | if ((GNUNET_OK != GNUNET_CRYPTO_hash_from_string (h1, &fi.chk.key)) || | ||
442 | (GNUNET_OK != GNUNET_CRYPTO_hash_from_string (h2, &fi.chk.query)) || | ||
443 | (1 != | ||
444 | sscanf (&s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) * 2], | ||
445 | "%llu", | ||
446 | &flen))) | ||
447 | { | ||
448 | *emsg = GNUNET_strdup (_ ("Malformed CHK URI (failed to decode CHK)")); | ||
449 | return NULL; | ||
450 | } | ||
451 | fi.file_length = GNUNET_htonll (flen); | ||
452 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
453 | ret->type = GNUNET_FS_URI_CHK; | ||
454 | ret->data.chk = fi; | ||
455 | return ret; | ||
456 | } | ||
457 | |||
458 | |||
459 | GNUNET_NETWORK_STRUCT_BEGIN | ||
460 | /** | ||
461 | * Structure that defines how the contents of a location URI must be | ||
462 | * assembled in memory to create or verify the signature of a location | ||
463 | * URI. | ||
464 | */ | ||
465 | struct LocUriAssembly | ||
466 | { | ||
467 | /** | ||
468 | * What is being signed (rest of this struct). | ||
469 | */ | ||
470 | struct GNUNET_CRYPTO_EccSignaturePurpose purpose; | ||
471 | |||
472 | /** | ||
473 | * Expiration time of the offer. | ||
474 | */ | ||
475 | struct GNUNET_TIME_AbsoluteNBO exptime; | ||
476 | |||
477 | /** | ||
478 | * File being offered. | ||
479 | */ | ||
480 | struct FileIdentifier fi; | ||
481 | |||
482 | /** | ||
483 | * Peer offering the file. | ||
484 | */ | ||
485 | struct GNUNET_PeerIdentity peer; | ||
486 | }; | ||
487 | GNUNET_NETWORK_STRUCT_END | ||
488 | |||
489 | |||
490 | #define GNUNET_FS_URI_LOC_PREFIX GNUNET_FS_URI_PREFIX GNUNET_FS_URI_LOC_INFIX | ||
491 | |||
492 | #define SIGNATURE_ASCII_LENGTH 103 | ||
493 | |||
494 | /** | ||
495 | * Parse a LOC URI. | ||
496 | * Also verifies validity of the location URI. | ||
497 | * | ||
498 | * @param s an uri string | ||
499 | * @param emsg where to store the parser error message (if any) | ||
500 | * @return NULL on error, valid LOC URI otherwise | ||
501 | */ | ||
502 | static struct GNUNET_FS_Uri * | ||
503 | uri_loc_parse (const char *s, char **emsg) | ||
504 | { | ||
505 | struct GNUNET_FS_Uri *uri; | ||
506 | char h1[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)]; | ||
507 | char h2[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)]; | ||
508 | unsigned int pos; | ||
509 | unsigned int npos; | ||
510 | unsigned long long exptime; | ||
511 | unsigned long long flen; | ||
512 | struct GNUNET_TIME_Absolute et; | ||
513 | struct GNUNET_CRYPTO_EddsaSignature sig; | ||
514 | struct LocUriAssembly ass; | ||
515 | size_t slen; | ||
516 | |||
517 | slen = strlen (s); | ||
518 | pos = strlen (GNUNET_FS_URI_LOC_PREFIX); | ||
519 | if ((slen < pos + 2 * sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) + 1) || | ||
520 | (0 != strncmp (s, GNUNET_FS_URI_LOC_PREFIX, pos))) | ||
521 | return NULL; /* not a LOC URI */ | ||
522 | if ((s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] != '.') || | ||
523 | (s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) * 2 - 1] != '.')) | ||
524 | { | ||
525 | *emsg = GNUNET_strdup (_ ("LOC URI malformed (wrong syntax)")); | ||
526 | return NULL; | ||
527 | } | ||
528 | GNUNET_memcpy (h1, &s[pos], sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)); | ||
529 | h1[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] = '\0'; | ||
530 | GNUNET_memcpy (h2, | ||
531 | &s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)], | ||
532 | sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded)); | ||
533 | h2[sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) - 1] = '\0'; | ||
534 | |||
535 | if ((GNUNET_OK != GNUNET_CRYPTO_hash_from_string (h1, &ass.fi.chk.key)) || | ||
536 | (GNUNET_OK != GNUNET_CRYPTO_hash_from_string (h2, &ass.fi.chk.query)) || | ||
537 | (1 != | ||
538 | sscanf (&s[pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) * 2], | ||
539 | "%llu", | ||
540 | &flen))) | ||
541 | { | ||
542 | *emsg = GNUNET_strdup (_ ("LOC URI malformed (no CHK)")); | ||
543 | return NULL; | ||
544 | } | ||
545 | ass.fi.file_length = GNUNET_htonll (flen); | ||
546 | |||
547 | npos = pos + sizeof(struct GNUNET_CRYPTO_HashAsciiEncoded) * 2; | ||
548 | while ((s[npos] != '\0') && (s[npos] != '.')) | ||
549 | npos++; | ||
550 | if (s[npos] == '\0') | ||
551 | { | ||
552 | *emsg = GNUNET_strdup (_ ("LOC URI malformed (missing LOC)")); | ||
553 | goto ERR; | ||
554 | } | ||
555 | npos++; | ||
556 | if ((strlen (&s[npos]) <= GNUNET_CRYPTO_PKEY_ASCII_LENGTH + 1) || | ||
557 | ('.' != s[npos + GNUNET_CRYPTO_PKEY_ASCII_LENGTH])) | ||
558 | { | ||
559 | *emsg = | ||
560 | GNUNET_strdup (_ ("LOC URI malformed (wrong syntax for public key)")); | ||
561 | } | ||
562 | if ( | ||
563 | GNUNET_OK != | ||
564 | GNUNET_CRYPTO_eddsa_public_key_from_string (&s[npos], | ||
565 | GNUNET_CRYPTO_PKEY_ASCII_LENGTH, | ||
566 | &ass.peer.public_key)) | ||
567 | { | ||
568 | *emsg = | ||
569 | GNUNET_strdup (_ ("LOC URI malformed (could not decode public key)")); | ||
570 | goto ERR; | ||
571 | } | ||
572 | npos += GNUNET_CRYPTO_PKEY_ASCII_LENGTH; | ||
573 | if (s[npos++] != '.') | ||
574 | { | ||
575 | *emsg = GNUNET_strdup (_ ("LOC URI malformed (could not find signature)")); | ||
576 | goto ERR; | ||
577 | } | ||
578 | if ((strlen (&s[npos]) <= SIGNATURE_ASCII_LENGTH + 1) || | ||
579 | ('.' != s[npos + SIGNATURE_ASCII_LENGTH])) | ||
580 | { | ||
581 | *emsg = | ||
582 | GNUNET_strdup (_ ("LOC URI malformed (wrong syntax for signature)")); | ||
583 | goto ERR; | ||
584 | } | ||
585 | if (GNUNET_OK != | ||
586 | GNUNET_STRINGS_string_to_data (&s[npos], | ||
587 | SIGNATURE_ASCII_LENGTH, | ||
588 | &sig, | ||
589 | sizeof( | ||
590 | struct GNUNET_CRYPTO_EddsaSignature))) | ||
591 | { | ||
592 | *emsg = | ||
593 | GNUNET_strdup (_ ("LOC URI malformed (could not decode signature)")); | ||
594 | goto ERR; | ||
595 | } | ||
596 | npos += SIGNATURE_ASCII_LENGTH; | ||
597 | if (s[npos++] != '.') | ||
598 | { | ||
599 | *emsg = GNUNET_strdup ( | ||
600 | _ ("LOC URI malformed (wrong syntax for expiration time)")); | ||
601 | goto ERR; | ||
602 | } | ||
603 | if (1 != sscanf (&s[npos], "%llu", &exptime)) | ||
604 | { | ||
605 | *emsg = | ||
606 | GNUNET_strdup (_ ("LOC URI malformed (could not parse expiration time)")); | ||
607 | goto ERR; | ||
608 | } | ||
609 | ass.purpose.size = htonl (sizeof(struct LocUriAssembly)); | ||
610 | ass.purpose.purpose = htonl (GNUNET_SIGNATURE_PURPOSE_PEER_PLACEMENT); | ||
611 | et.abs_value_us = exptime * 1000LL * 1000LL; | ||
612 | ass.exptime = GNUNET_TIME_absolute_hton (et); | ||
613 | if (GNUNET_OK != | ||
614 | GNUNET_CRYPTO_eddsa_verify (GNUNET_SIGNATURE_PURPOSE_PEER_PLACEMENT, | ||
615 | &ass, | ||
616 | &sig, | ||
617 | &ass.peer.public_key)) | ||
618 | { | ||
619 | *emsg = | ||
620 | GNUNET_strdup (_ ("LOC URI malformed (signature failed validation)")); | ||
621 | goto ERR; | ||
622 | } | ||
623 | uri = GNUNET_new (struct GNUNET_FS_Uri); | ||
624 | uri->type = GNUNET_FS_URI_LOC; | ||
625 | uri->data.loc.fi = ass.fi; | ||
626 | uri->data.loc.peer = ass.peer; | ||
627 | uri->data.loc.expirationTime = et; | ||
628 | uri->data.loc.contentSignature = sig; | ||
629 | |||
630 | return uri; | ||
631 | ERR: | ||
632 | return NULL; | ||
633 | } | ||
634 | |||
635 | |||
636 | /** | ||
637 | * Convert a UTF-8 String to a URI. | ||
638 | * | ||
639 | * @param uri string to parse | ||
640 | * @param emsg where to store the parser error message (if any) | ||
641 | * @return NULL on error | ||
642 | */ | ||
643 | struct GNUNET_FS_Uri * | ||
644 | GNUNET_FS_uri_parse (const char *uri, char **emsg) | ||
645 | { | ||
646 | struct GNUNET_FS_Uri *ret; | ||
647 | char *msg; | ||
648 | |||
649 | if (NULL == uri) | ||
650 | { | ||
651 | GNUNET_break (0); | ||
652 | if (NULL != emsg) | ||
653 | *emsg = GNUNET_strdup (_ ("invalid argument")); | ||
654 | return NULL; | ||
655 | } | ||
656 | if (NULL == emsg) | ||
657 | emsg = &msg; | ||
658 | *emsg = NULL; | ||
659 | if ((NULL != (ret = uri_chk_parse (uri, emsg))) || | ||
660 | (NULL != (ret = uri_ksk_parse (uri, emsg))) || | ||
661 | (NULL != (ret = uri_sks_parse (uri, emsg))) || | ||
662 | (NULL != (ret = uri_loc_parse (uri, emsg)))) | ||
663 | return ret; | ||
664 | if (NULL == *emsg) | ||
665 | *emsg = GNUNET_strdup (_ ("Unrecognized URI type")); | ||
666 | if (emsg == &msg) | ||
667 | GNUNET_free (msg); | ||
668 | return NULL; | ||
669 | } | ||
670 | |||
671 | |||
672 | /** | ||
673 | * Free URI. | ||
674 | * | ||
675 | * @param uri uri to free | ||
676 | */ | ||
677 | void | ||
678 | GNUNET_FS_uri_destroy (struct GNUNET_FS_Uri *uri) | ||
679 | { | ||
680 | unsigned int i; | ||
681 | |||
682 | switch (uri->type) | ||
683 | { | ||
684 | case GNUNET_FS_URI_KSK: | ||
685 | for (i = 0; i < uri->data.ksk.keywordCount; i++) | ||
686 | GNUNET_free (uri->data.ksk.keywords[i]); | ||
687 | GNUNET_array_grow (uri->data.ksk.keywords, uri->data.ksk.keywordCount, 0); | ||
688 | break; | ||
689 | |||
690 | case GNUNET_FS_URI_SKS: | ||
691 | GNUNET_free (uri->data.sks.identifier); | ||
692 | break; | ||
693 | |||
694 | case GNUNET_FS_URI_LOC: | ||
695 | break; | ||
696 | |||
697 | default: | ||
698 | /* do nothing */ | ||
699 | break; | ||
700 | } | ||
701 | GNUNET_free (uri); | ||
702 | } | ||
703 | |||
704 | |||
705 | /** | ||
706 | * How many keywords are ANDed in this keyword URI? | ||
707 | * | ||
708 | * @param uri ksk uri to get the number of keywords from | ||
709 | * @return 0 if this is not a keyword URI | ||
710 | */ | ||
711 | unsigned int | ||
712 | GNUNET_FS_uri_ksk_get_keyword_count (const struct GNUNET_FS_Uri *uri) | ||
713 | { | ||
714 | if (uri->type != GNUNET_FS_URI_KSK) | ||
715 | return 0; | ||
716 | return uri->data.ksk.keywordCount; | ||
717 | } | ||
718 | |||
719 | |||
720 | /** | ||
721 | * Iterate over all keywords in this keyword URI. | ||
722 | * | ||
723 | * @param uri ksk uri to get the keywords from | ||
724 | * @param iterator function to call on each keyword | ||
725 | * @param iterator_cls closure for iterator | ||
726 | * @return -1 if this is not a keyword URI, otherwise number of | ||
727 | * keywords iterated over until iterator aborted | ||
728 | */ | ||
729 | int | ||
730 | GNUNET_FS_uri_ksk_get_keywords (const struct GNUNET_FS_Uri *uri, | ||
731 | GNUNET_FS_KeywordIterator iterator, | ||
732 | void *iterator_cls) | ||
733 | { | ||
734 | unsigned int i; | ||
735 | char *keyword; | ||
736 | |||
737 | if (uri->type != GNUNET_FS_URI_KSK) | ||
738 | return -1; | ||
739 | if (NULL == iterator) | ||
740 | return uri->data.ksk.keywordCount; | ||
741 | for (i = 0; i < uri->data.ksk.keywordCount; i++) | ||
742 | { | ||
743 | keyword = uri->data.ksk.keywords[i]; | ||
744 | /* first character of keyword indicates | ||
745 | * if it is mandatory or not */ | ||
746 | if (GNUNET_OK != iterator (iterator_cls, &keyword[1],(keyword[0] == '+') )) | ||
747 | return i; | ||
748 | } | ||
749 | return i; | ||
750 | } | ||
751 | |||
752 | |||
753 | /** | ||
754 | * Add the given keyword to the set of keywords represented by the URI. | ||
755 | * Does nothing if the keyword is already present. | ||
756 | * | ||
757 | * @param uri ksk uri to modify | ||
758 | * @param keyword keyword to add | ||
759 | * @param is_mandatory is this keyword mandatory? | ||
760 | */ | ||
761 | void | ||
762 | GNUNET_FS_uri_ksk_add_keyword (struct GNUNET_FS_Uri *uri, | ||
763 | const char *keyword, | ||
764 | int is_mandatory) | ||
765 | { | ||
766 | unsigned int i; | ||
767 | const char *old; | ||
768 | char *n; | ||
769 | |||
770 | GNUNET_assert (uri->type == GNUNET_FS_URI_KSK); | ||
771 | for (i = 0; i < uri->data.ksk.keywordCount; i++) | ||
772 | { | ||
773 | old = uri->data.ksk.keywords[i]; | ||
774 | if (0 == strcmp (&old[1], keyword)) | ||
775 | return; | ||
776 | } | ||
777 | GNUNET_asprintf (&n, is_mandatory ? "+%s" : " %s", keyword); | ||
778 | GNUNET_array_append (uri->data.ksk.keywords, uri->data.ksk.keywordCount, n); | ||
779 | } | ||
780 | |||
781 | |||
782 | /** | ||
783 | * Remove the given keyword from the set of keywords represented by the URI. | ||
784 | * Does nothing if the keyword is not present. | ||
785 | * | ||
786 | * @param uri ksk uri to modify | ||
787 | * @param keyword keyword to add | ||
788 | */ | ||
789 | void | ||
790 | GNUNET_FS_uri_ksk_remove_keyword (struct GNUNET_FS_Uri *uri, | ||
791 | const char *keyword) | ||
792 | { | ||
793 | unsigned int i; | ||
794 | char *old; | ||
795 | |||
796 | GNUNET_assert (uri->type == GNUNET_FS_URI_KSK); | ||
797 | for (i = 0; i < uri->data.ksk.keywordCount; i++) | ||
798 | { | ||
799 | old = uri->data.ksk.keywords[i]; | ||
800 | if (0 == strcmp (&old[1], keyword)) | ||
801 | { | ||
802 | uri->data.ksk.keywords[i] = | ||
803 | uri->data.ksk.keywords[uri->data.ksk.keywordCount - 1]; | ||
804 | GNUNET_array_grow (uri->data.ksk.keywords, | ||
805 | uri->data.ksk.keywordCount, | ||
806 | uri->data.ksk.keywordCount - 1); | ||
807 | GNUNET_free (old); | ||
808 | return; | ||
809 | } | ||
810 | } | ||
811 | } | ||
812 | |||
813 | |||
814 | /** | ||
815 | * Obtain the identity of the peer offering the data | ||
816 | * | ||
817 | * @param uri the location URI to inspect | ||
818 | * @param peer where to store the identify of the peer (presumably) offering the content | ||
819 | * @return #GNUNET_SYSERR if this is not a location URI, otherwise #GNUNET_OK | ||
820 | */ | ||
821 | int | ||
822 | GNUNET_FS_uri_loc_get_peer_identity (const struct GNUNET_FS_Uri *uri, | ||
823 | struct GNUNET_PeerIdentity *peer) | ||
824 | { | ||
825 | if (uri->type != GNUNET_FS_URI_LOC) | ||
826 | return GNUNET_SYSERR; | ||
827 | *peer = uri->data.loc.peer; | ||
828 | return GNUNET_OK; | ||
829 | } | ||
830 | |||
831 | |||
832 | /** | ||
833 | * Obtain the expiration of the LOC URI. | ||
834 | * | ||
835 | * @param uri location URI to get the expiration from | ||
836 | * @return expiration time of the URI | ||
837 | */ | ||
838 | struct GNUNET_TIME_Absolute | ||
839 | GNUNET_FS_uri_loc_get_expiration (const struct GNUNET_FS_Uri *uri) | ||
840 | { | ||
841 | GNUNET_assert (uri->type == GNUNET_FS_URI_LOC); | ||
842 | return uri->data.loc.expirationTime; | ||
843 | } | ||
844 | |||
845 | |||
846 | /** | ||
847 | * Obtain the URI of the content itself. | ||
848 | * | ||
849 | * @param uri location URI to get the content URI from | ||
850 | * @return NULL if argument is not a location URI | ||
851 | */ | ||
852 | struct GNUNET_FS_Uri * | ||
853 | GNUNET_FS_uri_loc_get_uri (const struct GNUNET_FS_Uri *uri) | ||
854 | { | ||
855 | struct GNUNET_FS_Uri *ret; | ||
856 | |||
857 | if (uri->type != GNUNET_FS_URI_LOC) | ||
858 | return NULL; | ||
859 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
860 | ret->type = GNUNET_FS_URI_CHK; | ||
861 | ret->data.chk = uri->data.loc.fi; | ||
862 | return ret; | ||
863 | } | ||
864 | |||
865 | |||
866 | /** | ||
867 | * Construct a location URI (this peer will be used for the location). | ||
868 | * This function should only be called from within gnunet-service-fs, | ||
869 | * as it requires the peer's private key which is generally unavailable | ||
870 | * to processes directly under the user's control. However, for | ||
871 | * testing and as it logically fits under URIs, it is in this API. | ||
872 | * | ||
873 | * @param base_uri content offered by the sender | ||
874 | * @param sign_key private key of the peer | ||
875 | * @param expiration_time how long will the content be offered? | ||
876 | * @return the location URI, NULL on error | ||
877 | */ | ||
878 | struct GNUNET_FS_Uri * | ||
879 | GNUNET_FS_uri_loc_create (const struct GNUNET_FS_Uri *base_uri, | ||
880 | const struct GNUNET_CRYPTO_EddsaPrivateKey *sign_key, | ||
881 | struct GNUNET_TIME_Absolute expiration_time) | ||
882 | { | ||
883 | struct GNUNET_FS_Uri *uri; | ||
884 | struct GNUNET_CRYPTO_EddsaPublicKey my_public_key; | ||
885 | struct LocUriAssembly ass; | ||
886 | struct GNUNET_TIME_Absolute et; | ||
887 | |||
888 | if (GNUNET_FS_URI_CHK != base_uri->type) | ||
889 | return NULL; | ||
890 | /* we round expiration time to full seconds for SKS URIs */ | ||
891 | et.abs_value_us = (expiration_time.abs_value_us / 1000000LL) * 1000000LL; | ||
892 | GNUNET_CRYPTO_eddsa_key_get_public (sign_key, &my_public_key); | ||
893 | ass.purpose.size = htonl (sizeof(struct LocUriAssembly)); | ||
894 | ass.purpose.purpose = htonl (GNUNET_SIGNATURE_PURPOSE_PEER_PLACEMENT); | ||
895 | ass.exptime = GNUNET_TIME_absolute_hton (et); | ||
896 | ass.fi = base_uri->data.chk; | ||
897 | ass.peer.public_key = my_public_key; | ||
898 | uri = GNUNET_new (struct GNUNET_FS_Uri); | ||
899 | uri->type = GNUNET_FS_URI_LOC; | ||
900 | uri->data.loc.fi = base_uri->data.chk; | ||
901 | uri->data.loc.expirationTime = et; | ||
902 | uri->data.loc.peer.public_key = my_public_key; | ||
903 | GNUNET_CRYPTO_eddsa_sign (sign_key, | ||
904 | &ass, | ||
905 | &uri->data.loc.contentSignature); | ||
906 | return uri; | ||
907 | } | ||
908 | |||
909 | |||
910 | /** | ||
911 | * Create an SKS URI from a namespace ID and an identifier. | ||
912 | * | ||
913 | * @param ns namespace ID | ||
914 | * @param id identifier | ||
915 | * @return an FS URI for the given namespace and identifier | ||
916 | */ | ||
917 | struct GNUNET_FS_Uri * | ||
918 | GNUNET_FS_uri_sks_create (const struct GNUNET_CRYPTO_EcdsaPublicKey *ns, | ||
919 | const char *id) | ||
920 | { | ||
921 | struct GNUNET_FS_Uri *ns_uri; | ||
922 | |||
923 | ns_uri = GNUNET_new (struct GNUNET_FS_Uri); | ||
924 | ns_uri->type = GNUNET_FS_URI_SKS; | ||
925 | ns_uri->data.sks.ns = *ns; | ||
926 | ns_uri->data.sks.identifier = GNUNET_strdup (id); | ||
927 | return ns_uri; | ||
928 | } | ||
929 | |||
930 | |||
931 | /** | ||
932 | * Merge the sets of keywords from two KSK URIs. | ||
933 | * (useful for merging the canonicalized keywords with | ||
934 | * the original keywords for sharing). | ||
935 | * | ||
936 | * @param u1 first uri | ||
937 | * @param u2 second uri | ||
938 | * @return merged URI, NULL on error | ||
939 | */ | ||
940 | struct GNUNET_FS_Uri * | ||
941 | GNUNET_FS_uri_ksk_merge (const struct GNUNET_FS_Uri *u1, | ||
942 | const struct GNUNET_FS_Uri *u2) | ||
943 | { | ||
944 | struct GNUNET_FS_Uri *ret; | ||
945 | unsigned int kc; | ||
946 | unsigned int i; | ||
947 | unsigned int j; | ||
948 | int found; | ||
949 | const char *kp; | ||
950 | char **kl; | ||
951 | |||
952 | if ((u1 == NULL) && (u2 == NULL)) | ||
953 | return NULL; | ||
954 | if (u1 == NULL) | ||
955 | return GNUNET_FS_uri_dup (u2); | ||
956 | if (u2 == NULL) | ||
957 | return GNUNET_FS_uri_dup (u1); | ||
958 | if ((u1->type != GNUNET_FS_URI_KSK) || (u2->type != GNUNET_FS_URI_KSK)) | ||
959 | { | ||
960 | GNUNET_break (0); | ||
961 | return NULL; | ||
962 | } | ||
963 | kc = u1->data.ksk.keywordCount; | ||
964 | kl = GNUNET_new_array (kc + u2->data.ksk.keywordCount, char *); | ||
965 | for (i = 0; i < u1->data.ksk.keywordCount; i++) | ||
966 | kl[i] = GNUNET_strdup (u1->data.ksk.keywords[i]); | ||
967 | for (i = 0; i < u2->data.ksk.keywordCount; i++) | ||
968 | { | ||
969 | kp = u2->data.ksk.keywords[i]; | ||
970 | found = 0; | ||
971 | for (j = 0; j < u1->data.ksk.keywordCount; j++) | ||
972 | if (0 == strcmp (kp + 1, kl[j] + 1)) | ||
973 | { | ||
974 | found = 1; | ||
975 | if (kp[0] == '+') | ||
976 | kl[j][0] = '+'; | ||
977 | break; | ||
978 | } | ||
979 | if (0 == found) | ||
980 | kl[kc++] = GNUNET_strdup (kp); | ||
981 | } | ||
982 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
983 | ret->type = GNUNET_FS_URI_KSK; | ||
984 | ret->data.ksk.keywordCount = kc; | ||
985 | ret->data.ksk.keywords = kl; | ||
986 | return ret; | ||
987 | } | ||
988 | |||
989 | |||
990 | /** | ||
991 | * Duplicate URI. | ||
992 | * | ||
993 | * @param uri the URI to duplicate | ||
994 | * @return copy of the URI | ||
995 | */ | ||
996 | struct GNUNET_FS_Uri * | ||
997 | GNUNET_FS_uri_dup (const struct GNUNET_FS_Uri *uri) | ||
998 | { | ||
999 | struct GNUNET_FS_Uri *ret; | ||
1000 | unsigned int i; | ||
1001 | |||
1002 | if (uri == NULL) | ||
1003 | return NULL; | ||
1004 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
1005 | GNUNET_memcpy (ret, uri, sizeof(struct GNUNET_FS_Uri)); | ||
1006 | switch (ret->type) | ||
1007 | { | ||
1008 | case GNUNET_FS_URI_KSK: | ||
1009 | if (ret->data.ksk.keywordCount >= | ||
1010 | GNUNET_MAX_MALLOC_CHECKED / sizeof(char *)) | ||
1011 | { | ||
1012 | GNUNET_break (0); | ||
1013 | GNUNET_free (ret); | ||
1014 | return NULL; | ||
1015 | } | ||
1016 | if (ret->data.ksk.keywordCount > 0) | ||
1017 | { | ||
1018 | ret->data.ksk.keywords = | ||
1019 | GNUNET_new_array (ret->data.ksk.keywordCount, char *); | ||
1020 | for (i = 0; i < ret->data.ksk.keywordCount; i++) | ||
1021 | ret->data.ksk.keywords[i] = GNUNET_strdup (uri->data.ksk.keywords[i]); | ||
1022 | } | ||
1023 | else | ||
1024 | ret->data.ksk.keywords = NULL; /* just to be sure */ | ||
1025 | break; | ||
1026 | |||
1027 | case GNUNET_FS_URI_SKS: | ||
1028 | ret->data.sks.identifier = GNUNET_strdup (uri->data.sks.identifier); | ||
1029 | break; | ||
1030 | |||
1031 | case GNUNET_FS_URI_LOC: | ||
1032 | break; | ||
1033 | |||
1034 | default: | ||
1035 | break; | ||
1036 | } | ||
1037 | return ret; | ||
1038 | } | ||
1039 | |||
1040 | |||
1041 | /** | ||
1042 | * Create an FS URI from a single user-supplied string of keywords. | ||
1043 | * The string is broken up at spaces into individual keywords. | ||
1044 | * Keywords that start with "+" are mandatory. Double-quotes can | ||
1045 | * be used to prevent breaking up strings at spaces (and also | ||
1046 | * to specify non-mandatory keywords starting with "+"). | ||
1047 | * | ||
1048 | * Keywords must contain a balanced number of double quotes and | ||
1049 | * double quotes can not be used in the actual keywords (for | ||
1050 | * example, the string '""foo bar""' will be turned into two | ||
1051 | * "OR"ed keywords 'foo' and 'bar', not into '"foo bar"'. | ||
1052 | * | ||
1053 | * @param keywords the keyword string | ||
1054 | * @param emsg where to store an error message | ||
1055 | * @return an FS URI for the given keywords, NULL | ||
1056 | * if keywords is not legal (i.e. empty). | ||
1057 | */ | ||
1058 | struct GNUNET_FS_Uri * | ||
1059 | GNUNET_FS_uri_ksk_create (const char *keywords, char **emsg) | ||
1060 | { | ||
1061 | char **keywordarr; | ||
1062 | unsigned int num_Words; | ||
1063 | int inWord; | ||
1064 | char *pos; | ||
1065 | struct GNUNET_FS_Uri *uri; | ||
1066 | char *searchString; | ||
1067 | int saw_quote; | ||
1068 | |||
1069 | if (keywords == NULL) | ||
1070 | { | ||
1071 | *emsg = GNUNET_strdup (_ ("No keywords specified!\n")); | ||
1072 | GNUNET_break (0); | ||
1073 | return NULL; | ||
1074 | } | ||
1075 | searchString = GNUNET_strdup (keywords); | ||
1076 | num_Words = 0; | ||
1077 | inWord = 0; | ||
1078 | saw_quote = 0; | ||
1079 | pos = searchString; | ||
1080 | while ('\0' != *pos) | ||
1081 | { | ||
1082 | if ((saw_quote == 0) && (isspace ((unsigned char) *pos))) | ||
1083 | { | ||
1084 | inWord = 0; | ||
1085 | } | ||
1086 | else if (0 == inWord) | ||
1087 | { | ||
1088 | inWord = 1; | ||
1089 | ++num_Words; | ||
1090 | } | ||
1091 | if ('"' == *pos) | ||
1092 | saw_quote = (saw_quote + 1) % 2; | ||
1093 | pos++; | ||
1094 | } | ||
1095 | if (num_Words == 0) | ||
1096 | { | ||
1097 | GNUNET_free (searchString); | ||
1098 | *emsg = GNUNET_strdup (_ ("No keywords specified!\n")); | ||
1099 | return NULL; | ||
1100 | } | ||
1101 | if (saw_quote != 0) | ||
1102 | { | ||
1103 | GNUNET_free (searchString); | ||
1104 | *emsg = GNUNET_strdup (_ ("Number of double-quotes not balanced!\n")); | ||
1105 | return NULL; | ||
1106 | } | ||
1107 | keywordarr = GNUNET_new_array (num_Words, char *); | ||
1108 | num_Words = 0; | ||
1109 | inWord = 0; | ||
1110 | pos = searchString; | ||
1111 | while ('\0' != *pos) | ||
1112 | { | ||
1113 | if ((saw_quote == 0) && (isspace ((unsigned char) *pos))) | ||
1114 | { | ||
1115 | inWord = 0; | ||
1116 | *pos = '\0'; | ||
1117 | } | ||
1118 | else if (0 == inWord) | ||
1119 | { | ||
1120 | keywordarr[num_Words] = pos; | ||
1121 | inWord = 1; | ||
1122 | ++num_Words; | ||
1123 | } | ||
1124 | if ('"' == *pos) | ||
1125 | saw_quote = (saw_quote + 1) % 2; | ||
1126 | pos++; | ||
1127 | } | ||
1128 | uri = | ||
1129 | GNUNET_FS_uri_ksk_create_from_args (num_Words, (const char **) keywordarr); | ||
1130 | GNUNET_free (keywordarr); | ||
1131 | GNUNET_free (searchString); | ||
1132 | return uri; | ||
1133 | } | ||
1134 | |||
1135 | |||
1136 | /** | ||
1137 | * Create an FS URI from a user-supplied command line of keywords. | ||
1138 | * Arguments should start with "+" to indicate mandatory | ||
1139 | * keywords. | ||
1140 | * | ||
1141 | * @param argc number of keywords | ||
1142 | * @param argv keywords (double quotes are not required for | ||
1143 | * keywords containing spaces; however, double | ||
1144 | * quotes are required for keywords starting with | ||
1145 | * "+"); there is no mechanism for having double | ||
1146 | * quotes in the actual keywords (if the user | ||
1147 | * did specifically specify double quotes, the | ||
1148 | * caller should convert each double quote | ||
1149 | * into two single quotes). | ||
1150 | * @return an FS URI for the given keywords, NULL | ||
1151 | * if keywords is not legal (i.e. empty). | ||
1152 | */ | ||
1153 | struct GNUNET_FS_Uri * | ||
1154 | GNUNET_FS_uri_ksk_create_from_args (unsigned int argc, const char **argv) | ||
1155 | { | ||
1156 | unsigned int i; | ||
1157 | struct GNUNET_FS_Uri *uri; | ||
1158 | const char *keyword; | ||
1159 | char *val; | ||
1160 | const char *r; | ||
1161 | char *w; | ||
1162 | char *emsg; | ||
1163 | |||
1164 | if (argc == 0) | ||
1165 | return NULL; | ||
1166 | /* allow URI to be given as one and only keyword and | ||
1167 | * handle accordingly */ | ||
1168 | emsg = NULL; | ||
1169 | if ((argc == 1) && (strlen (argv[0]) > strlen (GNUNET_FS_URI_PREFIX)) && | ||
1170 | (0 == strncmp (argv[0], | ||
1171 | GNUNET_FS_URI_PREFIX, | ||
1172 | strlen (GNUNET_FS_URI_PREFIX))) && | ||
1173 | (NULL != (uri = GNUNET_FS_uri_parse (argv[0], &emsg)))) | ||
1174 | return uri; | ||
1175 | GNUNET_free (emsg); | ||
1176 | uri = GNUNET_new (struct GNUNET_FS_Uri); | ||
1177 | uri->type = GNUNET_FS_URI_KSK; | ||
1178 | uri->data.ksk.keywordCount = argc; | ||
1179 | uri->data.ksk.keywords = GNUNET_new_array (argc, char *); | ||
1180 | for (i = 0; i < argc; i++) | ||
1181 | { | ||
1182 | keyword = argv[i]; | ||
1183 | if (keyword[0] == '+') | ||
1184 | val = GNUNET_strdup (keyword); | ||
1185 | else | ||
1186 | GNUNET_asprintf (&val, " %s", keyword); | ||
1187 | r = val; | ||
1188 | w = val; | ||
1189 | while ('\0' != *r) | ||
1190 | { | ||
1191 | if ('"' == *r) | ||
1192 | r++; | ||
1193 | else | ||
1194 | *(w++) = *(r++); | ||
1195 | } | ||
1196 | *w = '\0'; | ||
1197 | uri->data.ksk.keywords[i] = val; | ||
1198 | } | ||
1199 | return uri; | ||
1200 | } | ||
1201 | |||
1202 | |||
1203 | /** | ||
1204 | * Test if two URIs are equal. | ||
1205 | * | ||
1206 | * @param u1 one of the URIs | ||
1207 | * @param u2 the other URI | ||
1208 | * @return #GNUNET_YES if the URIs are equal | ||
1209 | */ | ||
1210 | int | ||
1211 | GNUNET_FS_uri_test_equal (const struct GNUNET_FS_Uri *u1, | ||
1212 | const struct GNUNET_FS_Uri *u2) | ||
1213 | { | ||
1214 | int ret; | ||
1215 | unsigned int i; | ||
1216 | unsigned int j; | ||
1217 | |||
1218 | GNUNET_assert (u1 != NULL); | ||
1219 | GNUNET_assert (u2 != NULL); | ||
1220 | if (u1->type != u2->type) | ||
1221 | return GNUNET_NO; | ||
1222 | switch (u1->type) | ||
1223 | { | ||
1224 | case GNUNET_FS_URI_CHK: | ||
1225 | if (0 == | ||
1226 | memcmp (&u1->data.chk, &u2->data.chk, sizeof(struct FileIdentifier))) | ||
1227 | return GNUNET_YES; | ||
1228 | return GNUNET_NO; | ||
1229 | |||
1230 | case GNUNET_FS_URI_SKS: | ||
1231 | if ((0 == memcmp (&u1->data.sks.ns, | ||
1232 | &u2->data.sks.ns, | ||
1233 | sizeof(struct GNUNET_CRYPTO_EcdsaPublicKey))) && | ||
1234 | (0 == strcmp (u1->data.sks.identifier, u2->data.sks.identifier))) | ||
1235 | |||
1236 | return GNUNET_YES; | ||
1237 | return GNUNET_NO; | ||
1238 | |||
1239 | case GNUNET_FS_URI_KSK: | ||
1240 | if (u1->data.ksk.keywordCount != u2->data.ksk.keywordCount) | ||
1241 | return GNUNET_NO; | ||
1242 | for (i = 0; i < u1->data.ksk.keywordCount; i++) | ||
1243 | { | ||
1244 | ret = GNUNET_NO; | ||
1245 | for (j = 0; j < u2->data.ksk.keywordCount; j++) | ||
1246 | { | ||
1247 | if (0 == strcmp (u1->data.ksk.keywords[i], u2->data.ksk.keywords[j])) | ||
1248 | { | ||
1249 | ret = GNUNET_YES; | ||
1250 | break; | ||
1251 | } | ||
1252 | } | ||
1253 | if (ret == GNUNET_NO) | ||
1254 | return GNUNET_NO; | ||
1255 | } | ||
1256 | return GNUNET_YES; | ||
1257 | |||
1258 | case GNUNET_FS_URI_LOC: | ||
1259 | if (memcmp (&u1->data.loc, | ||
1260 | &u2->data.loc, | ||
1261 | sizeof(struct FileIdentifier) | ||
1262 | + sizeof(struct GNUNET_CRYPTO_EcdsaPublicKey) | ||
1263 | + sizeof(struct GNUNET_TIME_Absolute) | ||
1264 | + sizeof(unsigned short) + sizeof(unsigned short)) != 0) | ||
1265 | return GNUNET_NO; | ||
1266 | return GNUNET_YES; | ||
1267 | |||
1268 | default: | ||
1269 | return GNUNET_NO; | ||
1270 | } | ||
1271 | } | ||
1272 | |||
1273 | |||
1274 | /** | ||
1275 | * Is this a namespace URI? | ||
1276 | * | ||
1277 | * @param uri the uri to check | ||
1278 | * @return #GNUNET_YES if this is an SKS uri | ||
1279 | */ | ||
1280 | int | ||
1281 | GNUNET_FS_uri_test_sks (const struct GNUNET_FS_Uri *uri) | ||
1282 | { | ||
1283 | return uri->type == GNUNET_FS_URI_SKS; | ||
1284 | } | ||
1285 | |||
1286 | |||
1287 | /** | ||
1288 | * Get the ID of a namespace from the given | ||
1289 | * namespace URI. | ||
1290 | * | ||
1291 | * @param uri the uri to get the namespace ID from | ||
1292 | * @param pseudonym where to store the ID of the namespace | ||
1293 | * @return #GNUNET_OK on success | ||
1294 | */ | ||
1295 | int | ||
1296 | GNUNET_FS_uri_sks_get_namespace (const struct GNUNET_FS_Uri *uri, | ||
1297 | struct GNUNET_CRYPTO_EcdsaPublicKey *pseudonym) | ||
1298 | { | ||
1299 | if (! GNUNET_FS_uri_test_sks (uri)) | ||
1300 | { | ||
1301 | GNUNET_break (0); | ||
1302 | return GNUNET_SYSERR; | ||
1303 | } | ||
1304 | *pseudonym = uri->data.sks.ns; | ||
1305 | return GNUNET_OK; | ||
1306 | } | ||
1307 | |||
1308 | |||
1309 | /** | ||
1310 | * Get the content identifier of an SKS URI. | ||
1311 | * | ||
1312 | * @param uri the sks uri | ||
1313 | * @return NULL on error (not a valid SKS URI) | ||
1314 | */ | ||
1315 | char * | ||
1316 | GNUNET_FS_uri_sks_get_content_id (const struct GNUNET_FS_Uri *uri) | ||
1317 | { | ||
1318 | if (! GNUNET_FS_uri_test_sks (uri)) | ||
1319 | { | ||
1320 | GNUNET_break (0); | ||
1321 | return NULL; | ||
1322 | } | ||
1323 | return GNUNET_strdup (uri->data.sks.identifier); | ||
1324 | } | ||
1325 | |||
1326 | |||
1327 | /** | ||
1328 | * Is this a keyword URI? | ||
1329 | * | ||
1330 | * @param uri the uri | ||
1331 | * @return #GNUNET_YES if this is a KSK uri | ||
1332 | */ | ||
1333 | int | ||
1334 | GNUNET_FS_uri_test_ksk (const struct GNUNET_FS_Uri *uri) | ||
1335 | { | ||
1336 | #if EXTRA_CHECKS | ||
1337 | unsigned int i; | ||
1338 | |||
1339 | if (uri->type == GNUNET_FS_URI_KSK) | ||
1340 | { | ||
1341 | for (i = 0; i < uri->data.ksk.keywordCount; i++) | ||
1342 | GNUNET_assert (uri->data.ksk.keywords[i] != NULL); | ||
1343 | } | ||
1344 | #endif | ||
1345 | return uri->type == GNUNET_FS_URI_KSK; | ||
1346 | } | ||
1347 | |||
1348 | |||
1349 | /** | ||
1350 | * Is this a file (or directory) URI? | ||
1351 | * | ||
1352 | * @param uri the uri to check | ||
1353 | * @return #GNUNET_YES if this is a CHK uri | ||
1354 | */ | ||
1355 | int | ||
1356 | GNUNET_FS_uri_test_chk (const struct GNUNET_FS_Uri *uri) | ||
1357 | { | ||
1358 | return uri->type == GNUNET_FS_URI_CHK; | ||
1359 | } | ||
1360 | |||
1361 | |||
1362 | /** | ||
1363 | * What is the size of the file that this URI | ||
1364 | * refers to? | ||
1365 | * | ||
1366 | * @param uri the CHK URI to inspect | ||
1367 | * @return size of the file as specified in the CHK URI | ||
1368 | */ | ||
1369 | uint64_t | ||
1370 | GNUNET_FS_uri_chk_get_file_size (const struct GNUNET_FS_Uri *uri) | ||
1371 | { | ||
1372 | switch (uri->type) | ||
1373 | { | ||
1374 | case GNUNET_FS_URI_CHK: | ||
1375 | return GNUNET_ntohll (uri->data.chk.file_length); | ||
1376 | |||
1377 | case GNUNET_FS_URI_LOC: | ||
1378 | return GNUNET_ntohll (uri->data.loc.fi.file_length); | ||
1379 | |||
1380 | default: | ||
1381 | GNUNET_assert (0); | ||
1382 | } | ||
1383 | return 0; /* unreachable */ | ||
1384 | } | ||
1385 | |||
1386 | |||
1387 | /** | ||
1388 | * Is this a location URI? | ||
1389 | * | ||
1390 | * @param uri the uri to check | ||
1391 | * @return #GNUNET_YES if this is a LOC uri | ||
1392 | */ | ||
1393 | int | ||
1394 | GNUNET_FS_uri_test_loc (const struct GNUNET_FS_Uri *uri) | ||
1395 | { | ||
1396 | return uri->type == GNUNET_FS_URI_LOC; | ||
1397 | } | ||
1398 | |||
1399 | |||
1400 | /** | ||
1401 | * Add a keyword as non-mandatory (with ' '-prefix) to the | ||
1402 | * given keyword list at offset 'index'. The array is | ||
1403 | * guaranteed to be long enough. | ||
1404 | * | ||
1405 | * @param s keyword to add | ||
1406 | * @param array array to add the keyword to | ||
1407 | * @param index offset where to add the keyword | ||
1408 | */ | ||
1409 | static void | ||
1410 | insert_non_mandatory_keyword (const char *s, char **array, int index) | ||
1411 | { | ||
1412 | char *nkword; | ||
1413 | |||
1414 | GNUNET_asprintf (&nkword, | ||
1415 | " %s", /* space to mark as 'non mandatory' */ | ||
1416 | s); | ||
1417 | array[index] = nkword; | ||
1418 | } | ||
1419 | |||
1420 | |||
1421 | /** | ||
1422 | * Test if the given keyword @a s is already present in the | ||
1423 | * given array, ignoring the '+'-mandatory prefix in the array. | ||
1424 | * | ||
1425 | * @param s keyword to test | ||
1426 | * @param array keywords to test against, with ' ' or '+' prefix to ignore | ||
1427 | * @param array_length length of the @a array | ||
1428 | * @return #GNUNET_YES if the keyword exists, #GNUNET_NO if not | ||
1429 | */ | ||
1430 | static int | ||
1431 | find_duplicate (const char *s, const char **array, int array_length) | ||
1432 | { | ||
1433 | int j; | ||
1434 | |||
1435 | for (j = array_length - 1; j >= 0; j--) | ||
1436 | if (0 == strcmp (&array[j][1], s)) | ||
1437 | return GNUNET_YES; | ||
1438 | return GNUNET_NO; | ||
1439 | } | ||
1440 | |||
1441 | |||
1442 | /** | ||
1443 | * FIXME: comment | ||
1444 | */ | ||
1445 | static char * | ||
1446 | normalize_metadata (enum EXTRACTOR_MetaFormat format, | ||
1447 | const char *data, | ||
1448 | size_t data_len) | ||
1449 | { | ||
1450 | uint8_t *free_str = NULL; | ||
1451 | uint8_t *str_to_normalize = (uint8_t *) data; | ||
1452 | uint8_t *normalized; | ||
1453 | size_t r_len; | ||
1454 | |||
1455 | if (str_to_normalize == NULL) | ||
1456 | return NULL; | ||
1457 | /* Don't trust libextractor */ | ||
1458 | if (format == EXTRACTOR_METAFORMAT_UTF8) | ||
1459 | { | ||
1460 | free_str = (uint8_t *) u8_check ((const uint8_t *) data, data_len); | ||
1461 | if (free_str == NULL) | ||
1462 | free_str = NULL; | ||
1463 | else | ||
1464 | format = EXTRACTOR_METAFORMAT_C_STRING; | ||
1465 | } | ||
1466 | if (format == EXTRACTOR_METAFORMAT_C_STRING) | ||
1467 | { | ||
1468 | free_str = u8_strconv_from_encoding (data, | ||
1469 | locale_charset (), | ||
1470 | iconveh_escape_sequence); | ||
1471 | if (free_str == NULL) | ||
1472 | return NULL; | ||
1473 | } | ||
1474 | |||
1475 | normalized = u8_tolower (str_to_normalize, | ||
1476 | strlen ((char *) str_to_normalize), | ||
1477 | NULL, | ||
1478 | UNINORM_NFD, | ||
1479 | NULL, | ||
1480 | &r_len); | ||
1481 | /* free_str is allocated by libunistring internally, use free() */ | ||
1482 | if (free_str != NULL) | ||
1483 | free (free_str); | ||
1484 | if (normalized != NULL) | ||
1485 | { | ||
1486 | /* u8_tolower allocates a non-NULL-terminated string! */ | ||
1487 | free_str = GNUNET_malloc (r_len + 1); | ||
1488 | GNUNET_memcpy (free_str, normalized, r_len); | ||
1489 | free_str[r_len] = '\0'; | ||
1490 | free (normalized); | ||
1491 | normalized = free_str; | ||
1492 | } | ||
1493 | return (char *) normalized; | ||
1494 | } | ||
1495 | |||
1496 | |||
1497 | /** | ||
1498 | * Counts the number of UTF-8 characters (not bytes) in the string, | ||
1499 | * returns that count. | ||
1500 | */ | ||
1501 | static size_t | ||
1502 | u8_strcount (const uint8_t *s) | ||
1503 | { | ||
1504 | size_t count; | ||
1505 | ucs4_t c; | ||
1506 | |||
1507 | GNUNET_assert (s != NULL); | ||
1508 | if (s[0] == 0) | ||
1509 | return 0; | ||
1510 | for (count = 0; s != NULL; count++) | ||
1511 | s = u8_next (&c, s); | ||
1512 | return count - 1; | ||
1513 | } | ||
1514 | |||
1515 | |||
1516 | /** | ||
1517 | * Break the filename up by matching [], () and {} pairs to make | ||
1518 | * keywords. In case of nesting parentheses only the inner pair counts. | ||
1519 | * You can't escape parentheses to scan something like "[blah\{foo]" to | ||
1520 | * make a "blah{foo" keyword, this function is only a heuristic! | ||
1521 | * | ||
1522 | * @param s string to break down. | ||
1523 | * @param array array to fill with enclosed tokens. If NULL, then tokens | ||
1524 | * are only counted. | ||
1525 | * @param index index at which to start filling the array (entries prior | ||
1526 | * to it are used to check for duplicates). ignored if @a array == NULL. | ||
1527 | * @return number of tokens counted (including duplicates), or number of | ||
1528 | * tokens extracted (excluding duplicates). 0 if there are no | ||
1529 | * matching parens in the string (when counting), or when all tokens | ||
1530 | * were duplicates (when extracting). | ||
1531 | */ | ||
1532 | static int | ||
1533 | get_keywords_from_parens (const char *s, char **array, int index) | ||
1534 | { | ||
1535 | int count = 0; | ||
1536 | char *open_paren; | ||
1537 | char *close_paren; | ||
1538 | char *ss; | ||
1539 | char tmp; | ||
1540 | |||
1541 | if (NULL == s) | ||
1542 | return 0; | ||
1543 | ss = GNUNET_strdup (s); | ||
1544 | open_paren = ss - 1; | ||
1545 | while (NULL != (open_paren = strpbrk (open_paren + 1, "[{("))) | ||
1546 | { | ||
1547 | int match = 0; | ||
1548 | |||
1549 | close_paren = strpbrk (open_paren + 1, "]})"); | ||
1550 | if (NULL == close_paren) | ||
1551 | continue; | ||
1552 | switch (open_paren[0]) | ||
1553 | { | ||
1554 | case '[': | ||
1555 | if (']' == close_paren[0]) | ||
1556 | match = 1; | ||
1557 | break; | ||
1558 | |||
1559 | case '{': | ||
1560 | if ('}' == close_paren[0]) | ||
1561 | match = 1; | ||
1562 | break; | ||
1563 | |||
1564 | case '(': | ||
1565 | if (')' == close_paren[0]) | ||
1566 | match = 1; | ||
1567 | break; | ||
1568 | |||
1569 | default: | ||
1570 | break; | ||
1571 | } | ||
1572 | if (match && (close_paren - open_paren > 1)) | ||
1573 | { | ||
1574 | tmp = close_paren[0]; | ||
1575 | close_paren[0] = '\0'; | ||
1576 | /* Keywords must be at least 3 characters long */ | ||
1577 | if (u8_strcount ((const uint8_t *) &open_paren[1]) <= 2) | ||
1578 | { | ||
1579 | close_paren[0] = tmp; | ||
1580 | continue; | ||
1581 | } | ||
1582 | if (NULL != array) | ||
1583 | { | ||
1584 | char *normalized; | ||
1585 | if (GNUNET_NO == find_duplicate ((const char *) &open_paren[1], | ||
1586 | (const char **) array, | ||
1587 | index + count)) | ||
1588 | { | ||
1589 | insert_non_mandatory_keyword ((const char *) &open_paren[1], | ||
1590 | array, | ||
1591 | index + count); | ||
1592 | count++; | ||
1593 | } | ||
1594 | normalized = normalize_metadata (EXTRACTOR_METAFORMAT_UTF8, | ||
1595 | &open_paren[1], | ||
1596 | close_paren - &open_paren[1]); | ||
1597 | if (normalized != NULL) | ||
1598 | { | ||
1599 | if (GNUNET_NO == find_duplicate ((const char *) normalized, | ||
1600 | (const char **) array, | ||
1601 | index + count)) | ||
1602 | { | ||
1603 | insert_non_mandatory_keyword ((const char *) normalized, | ||
1604 | array, | ||
1605 | index + count); | ||
1606 | count++; | ||
1607 | } | ||
1608 | GNUNET_free (normalized); | ||
1609 | } | ||
1610 | } | ||
1611 | else | ||
1612 | count++; | ||
1613 | close_paren[0] = tmp; | ||
1614 | } | ||
1615 | } | ||
1616 | GNUNET_free (ss); | ||
1617 | return count; | ||
1618 | } | ||
1619 | |||
1620 | |||
1621 | /** | ||
1622 | * Where to break up keywords | ||
1623 | */ | ||
1624 | #define TOKENS "_. /-!?#&+@\"\'\\;:,()[]{}$<>|" | ||
1625 | |||
1626 | /** | ||
1627 | * Break the filename up by TOKENS to make | ||
1628 | * keywords. | ||
1629 | * | ||
1630 | * @param s string to break down. | ||
1631 | * @param array array to fill with tokens. If NULL, then tokens are only | ||
1632 | * counted. | ||
1633 | * @param index index at which to start filling the array (entries prior | ||
1634 | * to it are used to check for duplicates). ignored if @a array == NULL. | ||
1635 | * @return number of tokens (>1) counted (including duplicates), or number of | ||
1636 | * tokens extracted (excluding duplicates). 0 if there are no | ||
1637 | * separators in the string (when counting), or when all tokens were | ||
1638 | * duplicates (when extracting). | ||
1639 | */ | ||
1640 | static int | ||
1641 | get_keywords_from_tokens (const char *s, char **array, int index) | ||
1642 | { | ||
1643 | char *p; | ||
1644 | char *ss; | ||
1645 | int seps = 0; | ||
1646 | |||
1647 | ss = GNUNET_strdup (s); | ||
1648 | for (p = strtok (ss, TOKENS); p != NULL; p = strtok (NULL, TOKENS)) | ||
1649 | { | ||
1650 | /* Keywords must be at least 3 characters long */ | ||
1651 | if (u8_strcount ((const uint8_t *) p) <= 2) | ||
1652 | continue; | ||
1653 | if (NULL != array) | ||
1654 | { | ||
1655 | char *normalized; | ||
1656 | if (GNUNET_NO == find_duplicate (p, (const char **) array, index + seps)) | ||
1657 | { | ||
1658 | insert_non_mandatory_keyword (p, array, index + seps); | ||
1659 | seps++; | ||
1660 | } | ||
1661 | normalized = | ||
1662 | normalize_metadata (EXTRACTOR_METAFORMAT_UTF8, p, strlen (p)); | ||
1663 | if (normalized != NULL) | ||
1664 | { | ||
1665 | if (GNUNET_NO == find_duplicate ((const char *) normalized, | ||
1666 | (const char **) array, | ||
1667 | index + seps)) | ||
1668 | { | ||
1669 | insert_non_mandatory_keyword ((const char *) normalized, | ||
1670 | array, | ||
1671 | index + seps); | ||
1672 | seps++; | ||
1673 | } | ||
1674 | GNUNET_free (normalized); | ||
1675 | } | ||
1676 | } | ||
1677 | else | ||
1678 | seps++; | ||
1679 | } | ||
1680 | GNUNET_free (ss); | ||
1681 | return seps; | ||
1682 | } | ||
1683 | |||
1684 | |||
1685 | #undef TOKENS | ||
1686 | |||
1687 | |||
1688 | /** | ||
1689 | * Function called on each value in the meta data. | ||
1690 | * Adds it to the URI. | ||
1691 | * | ||
1692 | * @param cls URI to update | ||
1693 | * @param plugin_name name of the plugin that produced this value; | ||
1694 | * special values can be used (e.g. '<zlib>' for zlib being | ||
1695 | * used in the main libextractor library and yielding | ||
1696 | * meta data). | ||
1697 | * @param type libextractor-type describing the meta data | ||
1698 | * @param format basic format information about data | ||
1699 | * @param data_mime_type mime-type of data (not of the original file); | ||
1700 | * can be NULL (if mime-type is not known) | ||
1701 | * @param data actual meta-data found | ||
1702 | * @param data_len number of bytes in @a data | ||
1703 | * @return 0 (always) | ||
1704 | */ | ||
1705 | static int | ||
1706 | gather_uri_data (void *cls, | ||
1707 | const char *plugin_name, | ||
1708 | enum EXTRACTOR_MetaType type, | ||
1709 | enum EXTRACTOR_MetaFormat format, | ||
1710 | const char *data_mime_type, | ||
1711 | const char *data, | ||
1712 | size_t data_len) | ||
1713 | { | ||
1714 | struct GNUNET_FS_Uri *uri = cls; | ||
1715 | char *normalized_data; | ||
1716 | const char *sep; | ||
1717 | |||
1718 | if ((format != EXTRACTOR_METAFORMAT_UTF8) && | ||
1719 | (format != EXTRACTOR_METAFORMAT_C_STRING)) | ||
1720 | return 0; | ||
1721 | /* Keywords must be at least 3 characters long | ||
1722 | * If given non-utf8 string it will, most likely, find it to be invalid, | ||
1723 | * and will return the length of its valid part, skipping the keyword. | ||
1724 | * If it does - fix the extractor, not this check! | ||
1725 | */if (u8_strcount ((const uint8_t *) data) <= 2) | ||
1726 | return 0; | ||
1727 | if ((EXTRACTOR_METATYPE_MIMETYPE == type) && | ||
1728 | (NULL != (sep = memchr (data, '/', data_len))) && (sep != data)) | ||
1729 | { | ||
1730 | char *xtra; | ||
1731 | |||
1732 | GNUNET_asprintf (&xtra, "mimetype:%.*s", (int) (sep - data), data); | ||
1733 | if (! find_duplicate (xtra, | ||
1734 | (const char **) uri->data.ksk.keywords, | ||
1735 | uri->data.ksk.keywordCount)) | ||
1736 | { | ||
1737 | insert_non_mandatory_keyword (xtra, | ||
1738 | uri->data.ksk.keywords, | ||
1739 | uri->data.ksk.keywordCount); | ||
1740 | uri->data.ksk.keywordCount++; | ||
1741 | } | ||
1742 | GNUNET_free (xtra); | ||
1743 | } | ||
1744 | |||
1745 | normalized_data = normalize_metadata (format, data, data_len); | ||
1746 | if (! find_duplicate (data, | ||
1747 | (const char **) uri->data.ksk.keywords, | ||
1748 | uri->data.ksk.keywordCount)) | ||
1749 | { | ||
1750 | insert_non_mandatory_keyword (data, | ||
1751 | uri->data.ksk.keywords, | ||
1752 | uri->data.ksk.keywordCount); | ||
1753 | uri->data.ksk.keywordCount++; | ||
1754 | } | ||
1755 | if (NULL != normalized_data) | ||
1756 | { | ||
1757 | if (! find_duplicate (normalized_data, | ||
1758 | (const char **) uri->data.ksk.keywords, | ||
1759 | uri->data.ksk.keywordCount)) | ||
1760 | { | ||
1761 | insert_non_mandatory_keyword (normalized_data, | ||
1762 | uri->data.ksk.keywords, | ||
1763 | uri->data.ksk.keywordCount); | ||
1764 | uri->data.ksk.keywordCount++; | ||
1765 | } | ||
1766 | GNUNET_free (normalized_data); | ||
1767 | } | ||
1768 | return 0; | ||
1769 | } | ||
1770 | |||
1771 | |||
1772 | /** | ||
1773 | * Construct a keyword-URI from meta-data (take all entries | ||
1774 | * in the meta-data and construct one large keyword URI | ||
1775 | * that lists all keywords that can be found in the meta-data). | ||
1776 | * | ||
1777 | * @param md metadata to use | ||
1778 | * @return NULL on error, otherwise a KSK URI | ||
1779 | */ | ||
1780 | struct GNUNET_FS_Uri * | ||
1781 | GNUNET_FS_uri_ksk_create_from_meta_data ( | ||
1782 | const struct GNUNET_CONTAINER_MetaData *md) | ||
1783 | { | ||
1784 | struct GNUNET_FS_Uri *ret; | ||
1785 | char *filename; | ||
1786 | char *full_name = NULL; | ||
1787 | char *ss; | ||
1788 | int ent; | ||
1789 | int tok_keywords = 0; | ||
1790 | int paren_keywords = 0; | ||
1791 | |||
1792 | if (NULL == md) | ||
1793 | return NULL; | ||
1794 | ret = GNUNET_new (struct GNUNET_FS_Uri); | ||
1795 | ret->type = GNUNET_FS_URI_KSK; | ||
1796 | ent = GNUNET_CONTAINER_meta_data_iterate (md, NULL, NULL); | ||
1797 | if (ent > 0) | ||
1798 | { | ||
1799 | full_name = GNUNET_CONTAINER_meta_data_get_first_by_types ( | ||
1800 | md, | ||
1801 | EXTRACTOR_METATYPE_GNUNET_ORIGINAL_FILENAME, | ||
1802 | -1); | ||
1803 | if (NULL != full_name) | ||
1804 | { | ||
1805 | filename = full_name; | ||
1806 | while (NULL != (ss = strstr (filename, DIR_SEPARATOR_STR))) | ||
1807 | filename = ss + 1; | ||
1808 | tok_keywords = get_keywords_from_tokens (filename, NULL, 0); | ||
1809 | paren_keywords = get_keywords_from_parens (filename, NULL, 0); | ||
1810 | } | ||
1811 | /* x3 because there might be a normalized variant of every keyword, | ||
1812 | plus theoretically one more for mime... */ | ||
1813 | ret->data.ksk.keywords = | ||
1814 | GNUNET_new_array ((ent + tok_keywords + paren_keywords) * 3, char *); | ||
1815 | GNUNET_CONTAINER_meta_data_iterate (md, &gather_uri_data, ret); | ||
1816 | } | ||
1817 | if (tok_keywords > 0) | ||
1818 | ret->data.ksk.keywordCount += | ||
1819 | get_keywords_from_tokens (filename, | ||
1820 | ret->data.ksk.keywords, | ||
1821 | ret->data.ksk.keywordCount); | ||
1822 | if (paren_keywords > 0) | ||
1823 | ret->data.ksk.keywordCount += | ||
1824 | get_keywords_from_parens (filename, | ||
1825 | ret->data.ksk.keywords, | ||
1826 | ret->data.ksk.keywordCount); | ||
1827 | if (ent > 0) | ||
1828 | GNUNET_free (full_name); | ||
1829 | return ret; | ||
1830 | } | ||
1831 | |||
1832 | |||
1833 | /** | ||
1834 | * In URI-encoding, does the given character | ||
1835 | * need to be encoded using %-encoding? | ||
1836 | */ | ||
1837 | static int | ||
1838 | needs_percent (char c) | ||
1839 | { | ||
1840 | return(! ((isalnum ((unsigned char) c)) || (c == '-') || (c == '_') || | ||
1841 | (c == '.') || (c == '~'))); | ||
1842 | } | ||
1843 | |||
1844 | |||
1845 | /** | ||
1846 | * Convert a KSK URI to a string. | ||
1847 | * | ||
1848 | * @param uri the URI to convert | ||
1849 | * @return NULL on error (i.e. keywordCount == 0) | ||
1850 | */ | ||
1851 | static char * | ||
1852 | uri_ksk_to_string (const struct GNUNET_FS_Uri *uri) | ||
1853 | { | ||
1854 | char **keywords; | ||
1855 | unsigned int keywordCount; | ||
1856 | size_t n; | ||
1857 | char *ret; | ||
1858 | unsigned int i; | ||
1859 | unsigned int j; | ||
1860 | unsigned int wpos; | ||
1861 | size_t slen; | ||
1862 | const char *keyword; | ||
1863 | |||
1864 | if (uri->type != GNUNET_FS_URI_KSK) | ||
1865 | return NULL; | ||
1866 | keywords = uri->data.ksk.keywords; | ||
1867 | keywordCount = uri->data.ksk.keywordCount; | ||
1868 | n = keywordCount + strlen (GNUNET_FS_URI_PREFIX) | ||
1869 | + strlen (GNUNET_FS_URI_KSK_INFIX) + 1; | ||
1870 | for (i = 0; i < keywordCount; i++) | ||
1871 | { | ||
1872 | keyword = keywords[i]; | ||
1873 | slen = strlen (keyword); | ||
1874 | n += slen; | ||
1875 | for (j = 0; j < slen; j++) | ||
1876 | { | ||
1877 | if ((j == 0) && (keyword[j] == ' ')) | ||
1878 | { | ||
1879 | n--; | ||
1880 | continue; /* skip leading space */ | ||
1881 | } | ||
1882 | if (needs_percent (keyword[j])) | ||
1883 | n += 2; /* will use %-encoding */ | ||
1884 | } | ||
1885 | } | ||
1886 | ret = GNUNET_malloc (n); | ||
1887 | strcpy (ret, GNUNET_FS_URI_PREFIX); | ||
1888 | strcat (ret, GNUNET_FS_URI_KSK_INFIX); | ||
1889 | wpos = strlen (ret); | ||
1890 | for (i = 0; i < keywordCount; i++) | ||
1891 | { | ||
1892 | keyword = keywords[i]; | ||
1893 | slen = strlen (keyword); | ||
1894 | for (j = 0; j < slen; j++) | ||
1895 | { | ||
1896 | if ((j == 0) && (keyword[j] == ' ')) | ||
1897 | continue; /* skip leading space */ | ||
1898 | if (needs_percent (keyword[j])) | ||
1899 | { | ||
1900 | sprintf (&ret[wpos], "%%%02X", (unsigned char) keyword[j]); | ||
1901 | wpos += 3; | ||
1902 | } | ||
1903 | else | ||
1904 | { | ||
1905 | ret[wpos++] = keyword[j]; | ||
1906 | } | ||
1907 | } | ||
1908 | if (i != keywordCount - 1) | ||
1909 | ret[wpos++] = '+'; | ||
1910 | } | ||
1911 | return ret; | ||
1912 | } | ||
1913 | |||
1914 | |||
1915 | /** | ||
1916 | * Convert SKS URI to a string. | ||
1917 | * | ||
1918 | * @param uri sks uri to convert | ||
1919 | * @return NULL on error | ||
1920 | */ | ||
1921 | static char * | ||
1922 | uri_sks_to_string (const struct GNUNET_FS_Uri *uri) | ||
1923 | { | ||
1924 | char *ret; | ||
1925 | char buf[1024]; | ||
1926 | |||
1927 | if (GNUNET_FS_URI_SKS != uri->type) | ||
1928 | return NULL; | ||
1929 | ret = | ||
1930 | GNUNET_STRINGS_data_to_string (&uri->data.sks.ns, | ||
1931 | sizeof(struct GNUNET_CRYPTO_EcdsaPublicKey), | ||
1932 | buf, | ||
1933 | sizeof(buf)); | ||
1934 | GNUNET_assert (NULL != ret); | ||
1935 | ret[0] = '\0'; | ||
1936 | GNUNET_asprintf (&ret, | ||
1937 | "%s%s%s/%s", | ||
1938 | GNUNET_FS_URI_PREFIX, | ||
1939 | GNUNET_FS_URI_SKS_INFIX, | ||
1940 | buf, | ||
1941 | uri->data.sks.identifier); | ||
1942 | return ret; | ||
1943 | } | ||
1944 | |||
1945 | |||
1946 | /** | ||
1947 | * Convert a CHK URI to a string. | ||
1948 | * | ||
1949 | * @param uri chk uri to convert | ||
1950 | * @return NULL on error | ||
1951 | */ | ||
1952 | static char * | ||
1953 | uri_chk_to_string (const struct GNUNET_FS_Uri *uri) | ||
1954 | { | ||
1955 | const struct FileIdentifier *fi; | ||
1956 | char *ret; | ||
1957 | struct GNUNET_CRYPTO_HashAsciiEncoded keyhash; | ||
1958 | struct GNUNET_CRYPTO_HashAsciiEncoded queryhash; | ||
1959 | |||
1960 | if (uri->type != GNUNET_FS_URI_CHK) | ||
1961 | return NULL; | ||
1962 | fi = &uri->data.chk; | ||
1963 | GNUNET_CRYPTO_hash_to_enc (&fi->chk.key, &keyhash); | ||
1964 | GNUNET_CRYPTO_hash_to_enc (&fi->chk.query, &queryhash); | ||
1965 | |||
1966 | GNUNET_asprintf (&ret, | ||
1967 | "%s%s%s.%s.%llu", | ||
1968 | GNUNET_FS_URI_PREFIX, | ||
1969 | GNUNET_FS_URI_CHK_INFIX, | ||
1970 | (const char *) &keyhash, | ||
1971 | (const char *) &queryhash, | ||
1972 | (unsigned long long) GNUNET_ntohll (fi->file_length)); | ||
1973 | return ret; | ||
1974 | } | ||
1975 | |||
1976 | |||
1977 | /** | ||
1978 | * Convert a LOC URI to a string. | ||
1979 | * | ||
1980 | * @param uri loc uri to convert | ||
1981 | * @return NULL on error | ||
1982 | */ | ||
1983 | static char * | ||
1984 | uri_loc_to_string (const struct GNUNET_FS_Uri *uri) | ||
1985 | { | ||
1986 | char *ret; | ||
1987 | struct GNUNET_CRYPTO_HashAsciiEncoded keyhash; | ||
1988 | struct GNUNET_CRYPTO_HashAsciiEncoded queryhash; | ||
1989 | char *peer_id; | ||
1990 | char peer_sig[SIGNATURE_ASCII_LENGTH + 1]; | ||
1991 | |||
1992 | GNUNET_CRYPTO_hash_to_enc (&uri->data.loc.fi.chk.key, &keyhash); | ||
1993 | GNUNET_CRYPTO_hash_to_enc (&uri->data.loc.fi.chk.query, &queryhash); | ||
1994 | peer_id = | ||
1995 | GNUNET_CRYPTO_eddsa_public_key_to_string (&uri->data.loc.peer.public_key); | ||
1996 | GNUNET_assert ( | ||
1997 | NULL != | ||
1998 | GNUNET_STRINGS_data_to_string (&uri->data.loc.contentSignature, | ||
1999 | sizeof(struct GNUNET_CRYPTO_EddsaSignature), | ||
2000 | peer_sig, | ||
2001 | sizeof(peer_sig))); | ||
2002 | GNUNET_asprintf (&ret, | ||
2003 | "%s%s%s.%s.%llu.%s.%s.%llu", | ||
2004 | GNUNET_FS_URI_PREFIX, | ||
2005 | GNUNET_FS_URI_LOC_INFIX, | ||
2006 | (const char *) &keyhash, | ||
2007 | (const char *) &queryhash, | ||
2008 | (unsigned long long) GNUNET_ntohll ( | ||
2009 | uri->data.loc.fi.file_length), | ||
2010 | peer_id, | ||
2011 | peer_sig, | ||
2012 | (unsigned long long) | ||
2013 | uri->data.loc.expirationTime.abs_value_us | ||
2014 | / 1000000LL); | ||
2015 | GNUNET_free (peer_id); | ||
2016 | return ret; | ||
2017 | } | ||
2018 | |||
2019 | |||
2020 | /** | ||
2021 | * Convert a URI to a UTF-8 String. | ||
2022 | * | ||
2023 | * @param uri uri to convert to a string | ||
2024 | * @return the UTF-8 string | ||
2025 | */ | ||
2026 | char * | ||
2027 | GNUNET_FS_uri_to_string (const struct GNUNET_FS_Uri *uri) | ||
2028 | { | ||
2029 | if (uri == NULL) | ||
2030 | { | ||
2031 | GNUNET_break (0); | ||
2032 | return NULL; | ||
2033 | } | ||
2034 | switch (uri->type) | ||
2035 | { | ||
2036 | case GNUNET_FS_URI_KSK: | ||
2037 | return uri_ksk_to_string (uri); | ||
2038 | |||
2039 | case GNUNET_FS_URI_SKS: | ||
2040 | return uri_sks_to_string (uri); | ||
2041 | |||
2042 | case GNUNET_FS_URI_CHK: | ||
2043 | return uri_chk_to_string (uri); | ||
2044 | |||
2045 | case GNUNET_FS_URI_LOC: | ||
2046 | return uri_loc_to_string (uri); | ||
2047 | |||
2048 | default: | ||
2049 | GNUNET_break (0); | ||
2050 | return NULL; | ||
2051 | } | ||
2052 | } | ||
2053 | |||
2054 | |||
2055 | /* end of fs_uri.c */ | ||