1 | // =============================================================== // |
---|
2 | // // |
---|
3 | // File : adfile.cxx // |
---|
4 | // Purpose : various IO related functions // |
---|
5 | // // |
---|
6 | // Institute of Microbiology (Technical University Munich) // |
---|
7 | // http://www.arb-home.de/ // |
---|
8 | // // |
---|
9 | // =============================================================== // |
---|
10 | |
---|
11 | #include <dirent.h> |
---|
12 | #include <unistd.h> |
---|
13 | #include <sys/stat.h> |
---|
14 | |
---|
15 | #include <arb_sort.h> |
---|
16 | #include <arb_str.h> |
---|
17 | #include <arb_strarray.h> |
---|
18 | #include <arb_file.h> |
---|
19 | #include <arb_pathlen.h> |
---|
20 | |
---|
21 | #include "gb_local.h" |
---|
22 | #include "gb_load.h" |
---|
23 | |
---|
24 | GB_CSTR GB_getcwd() { |
---|
25 | // get the current working directory |
---|
26 | // (directory from which application has been started) |
---|
27 | RETURN_ONETIME_ALLOC(getcwd(0, ARB_PATH_MAX)); |
---|
28 | } |
---|
29 | |
---|
30 | GB_ERROR gb_scan_directory(char *basename, gb_scandir *sd) { |
---|
31 | // goes to header: __ATTR__USERESULT_TODO |
---|
32 | // look for quick saves (basename = yyy/xxx no arb ending !!!!) |
---|
33 | char *path = ARB_strdup(basename); |
---|
34 | const char *fulldir = "."; |
---|
35 | char *file = strrchr(path, '/'); |
---|
36 | DIR *dirp; |
---|
37 | int curindex; |
---|
38 | char *suffix; |
---|
39 | dirent *dp; |
---|
40 | struct stat st; |
---|
41 | const char *oldstyle = ".arb.quick"; |
---|
42 | char buffer[ARB_PATH_MAX]; |
---|
43 | int oldstylelen = strlen(oldstyle); |
---|
44 | int filelen; |
---|
45 | |
---|
46 | if (file) { |
---|
47 | *(file++) = 0; |
---|
48 | fulldir = path; |
---|
49 | } |
---|
50 | else { |
---|
51 | file = path; |
---|
52 | } |
---|
53 | |
---|
54 | memset((char*)sd, 0, sizeof(*sd)); |
---|
55 | sd->type = GB_SCAN_NO_QUICK; |
---|
56 | sd->highest_quick_index = -1; |
---|
57 | sd->newest_quick_index = -1; |
---|
58 | sd->date_of_quick_file = 0; |
---|
59 | |
---|
60 | dirp = opendir(fulldir); |
---|
61 | if (!dirp) { |
---|
62 | GB_ERROR error = GBS_global_string("Directory %s of file %s.arb not readable", fulldir, file); |
---|
63 | free(path); |
---|
64 | return error; |
---|
65 | } |
---|
66 | filelen = strlen(file); |
---|
67 | for (dp = readdir(dirp); dp != NULL; dp = readdir(dirp)) { |
---|
68 | if (strncmp(dp->d_name, file, filelen)) continue; |
---|
69 | suffix = dp->d_name + filelen; |
---|
70 | if (suffix[0] != '.') continue; |
---|
71 | if (!strncmp(suffix, oldstyle, oldstylelen)) { |
---|
72 | if (sd->type == GB_SCAN_NEW_QUICK) { |
---|
73 | printf("Warning: Found new and old changes files, using new\n"); |
---|
74 | continue; |
---|
75 | } |
---|
76 | sd->type = GB_SCAN_OLD_QUICK; |
---|
77 | curindex = atoi(suffix+oldstylelen); |
---|
78 | goto check_time_and_date; |
---|
79 | } |
---|
80 | if (strlen(suffix) == 4 && |
---|
81 | suffix[0] == '.' && |
---|
82 | suffix[1] == 'a' && |
---|
83 | isdigit(suffix[2]) && |
---|
84 | isdigit(suffix[3])) { |
---|
85 | if (sd->type == GB_SCAN_OLD_QUICK) { |
---|
86 | printf("Warning: Found new and old changes files, using new\n"); |
---|
87 | } |
---|
88 | sd->type = GB_SCAN_NEW_QUICK; |
---|
89 | curindex = atoi(suffix+2); |
---|
90 | goto check_time_and_date; |
---|
91 | } |
---|
92 | continue; |
---|
93 | check_time_and_date : |
---|
94 | if (curindex > sd->highest_quick_index) sd->highest_quick_index = curindex; |
---|
95 | sprintf(buffer, "%s/%s", fulldir, dp->d_name); |
---|
96 | stat(buffer, &st); |
---|
97 | if ((unsigned long)st.st_mtime > sd->date_of_quick_file) { |
---|
98 | sd->date_of_quick_file = st.st_mtime; |
---|
99 | sd->newest_quick_index = curindex; |
---|
100 | } |
---|
101 | continue; |
---|
102 | } |
---|
103 | closedir(dirp); |
---|
104 | free(path); |
---|
105 | return 0; |
---|
106 | } |
---|
107 | |
---|
108 | |
---|
109 | char *GB_find_all_files(const char *dir, const char *mask, bool filename_only) { |
---|
110 | /* Returns a string containing the filenames of all files matching mask. |
---|
111 | The single filenames are separated by '*'. |
---|
112 | if 'filename_only' is true -> string contains only filenames w/o path |
---|
113 | |
---|
114 | returns 0 if no files found (or directory not found). |
---|
115 | in this case an error may be exported |
---|
116 | |
---|
117 | 'mask' may contain wildcards (*?) or |
---|
118 | it may be a regular expression ('/regexp/') |
---|
119 | */ |
---|
120 | |
---|
121 | DIR *dirp; |
---|
122 | struct stat st; |
---|
123 | char *result = 0; |
---|
124 | |
---|
125 | dirp = opendir(dir); |
---|
126 | if (dirp) { |
---|
127 | GBS_string_matcher *matcher = GBS_compile_matcher(mask, GB_IGNORE_CASE); |
---|
128 | if (matcher) { |
---|
129 | for (dirent *dp = readdir(dirp); dp != NULL; dp = readdir(dirp)) { |
---|
130 | if (GBS_string_matches_regexp(dp->d_name, matcher)) { |
---|
131 | char buffer[ARB_PATH_MAX]; |
---|
132 | sprintf(buffer, "%s/%s", dir, dp->d_name); |
---|
133 | if (stat(buffer, &st) == 0 && S_ISREG(st.st_mode)) { // regular file ? |
---|
134 | if (filename_only) strcpy(buffer, dp->d_name); |
---|
135 | if (result) { |
---|
136 | freeset(result, GBS_global_string_copy("%s*%s", result, buffer)); |
---|
137 | } |
---|
138 | else { |
---|
139 | result = ARB_strdup(buffer); |
---|
140 | } |
---|
141 | } |
---|
142 | } |
---|
143 | } |
---|
144 | GBS_free_matcher(matcher); |
---|
145 | } |
---|
146 | closedir(dirp); |
---|
147 | } |
---|
148 | |
---|
149 | return result; |
---|
150 | } |
---|
151 | |
---|
152 | char *GB_find_latest_file(const char *dir, const char *mask) { |
---|
153 | /* returns the name of the newest file in dir 'dir' matching 'mask' |
---|
154 | * or NULL (in this case an error may be exported) |
---|
155 | * |
---|
156 | * 'mask' may contain wildcards (*?) or |
---|
157 | * it may be a regular expression ('/regexp/') |
---|
158 | */ |
---|
159 | |
---|
160 | DIR *dirp; |
---|
161 | struct stat st; |
---|
162 | char *result = 0; |
---|
163 | |
---|
164 | dirp = opendir(dir); |
---|
165 | if (dirp) { |
---|
166 | GBS_string_matcher *matcher = GBS_compile_matcher(mask, GB_IGNORE_CASE); |
---|
167 | if (matcher) { |
---|
168 | GB_ULONG newest = 0; |
---|
169 | for (dirent *dp = readdir(dirp); dp != NULL; dp = readdir(dirp)) { |
---|
170 | if (GBS_string_matches_regexp(dp->d_name, matcher)) { |
---|
171 | char buffer[ARB_PATH_MAX]; |
---|
172 | sprintf(buffer, "%s/%s", dir, dp->d_name); |
---|
173 | if (stat(buffer, &st) == 0 && |
---|
174 | S_ISREG(st.st_mode) && |
---|
175 | (GB_ULONG)st.st_mtime > newest) { |
---|
176 | newest = st.st_mtime; |
---|
177 | freedup(result, dp->d_name); |
---|
178 | } |
---|
179 | } |
---|
180 | } |
---|
181 | GBS_free_matcher(matcher); |
---|
182 | } |
---|
183 | closedir(dirp); |
---|
184 | } |
---|
185 | return result; |
---|
186 | } |
---|
187 | |
---|
188 | static const char *GB_existing_file(const char *file, bool warn_when_not_found) { |
---|
189 | // return 'file' if it's an existing readable file |
---|
190 | // return NULL otherwise |
---|
191 | |
---|
192 | gb_assert(file); |
---|
193 | if (GB_is_readablefile(file)) return file; |
---|
194 | if (warn_when_not_found) GB_warningf("Could not find '%s'", file); |
---|
195 | return NULL; |
---|
196 | } |
---|
197 | |
---|
198 | char *GB_lib_file(bool warn_when_not_found, const char *libprefix, const char *filename) { |
---|
199 | // Search a file in '$ARBHOME/lib/libprefix' |
---|
200 | // Return NULL if not found |
---|
201 | return nulldup(GB_existing_file(GB_path_in_ARBLIB(libprefix, filename), warn_when_not_found)); |
---|
202 | } |
---|
203 | |
---|
204 | char *GB_property_file(bool warn_when_not_found, const char *filename) { |
---|
205 | // Search a file in '$ARB_PROP' or its default in '$ARBHOME/lib/arb_default' |
---|
206 | // Return NULL if neighter found |
---|
207 | |
---|
208 | char *result = nulldup(GB_existing_file(GB_path_in_arbprop(filename), warn_when_not_found)); |
---|
209 | if (!result) result = GB_lib_file(warn_when_not_found, "arb_default", filename); |
---|
210 | return result; |
---|
211 | } |
---|
212 | |
---|
213 | void GBS_read_dir(StrArray& names, const char *dir, const char *mask) { |
---|
214 | /* Return full pathnames of files in directory 'dir'. |
---|
215 | * |
---|
216 | * Filter through 'mask': |
---|
217 | * - mask == NULL -> return all files |
---|
218 | * - in format '/expr/' -> use regular expression (case sensitive) |
---|
219 | * - else it does a simple string match with wildcards ("?*") |
---|
220 | * |
---|
221 | * Result are inserted into 'names' and 'names' is sorted alphanumerically. |
---|
222 | * Note: 'names' are not cleared, so several calls with the same StrArray get collected. |
---|
223 | * |
---|
224 | * In case of error, 'names' is empty and error is exported. |
---|
225 | * |
---|
226 | * Special case: If 'dir' is the name of a file, return an array with file as only element |
---|
227 | */ |
---|
228 | |
---|
229 | gb_assert(dir); // dir == NULL was allowed before 12/2008, forbidden now! |
---|
230 | |
---|
231 | if (dir[0]) { |
---|
232 | char *fulldir = ARB_strdup(GB_canonical_path(dir)); |
---|
233 | DIR *dirstream = opendir(fulldir); |
---|
234 | |
---|
235 | if (!dirstream) { |
---|
236 | if (GB_is_readablefile(fulldir)) { // fixed: returned true for directories before (4/2012) |
---|
237 | names.put(fulldir); // transfer ownership |
---|
238 | fulldir = NULL; |
---|
239 | } |
---|
240 | else { |
---|
241 | // @@@ does too much voodoo here - fix |
---|
242 | char *lslash = strrchr(fulldir, '/'); |
---|
243 | |
---|
244 | if (lslash) { |
---|
245 | lslash[0] = 0; |
---|
246 | char *name = lslash+1; |
---|
247 | if (GB_is_directory(fulldir)) { |
---|
248 | GBS_read_dir(names, fulldir, name); // @@@ concat mask to name? |
---|
249 | } |
---|
250 | lslash[0] = '/'; |
---|
251 | } |
---|
252 | |
---|
253 | if (names.empty()) GB_export_errorf("can't read directory '%s'", fulldir); // @@@ wrong place for error; maybe return error as result |
---|
254 | } |
---|
255 | } |
---|
256 | else { |
---|
257 | if (mask == NULL) mask = "*"; |
---|
258 | |
---|
259 | GBS_string_matcher *matcher = GBS_compile_matcher(mask, GB_MIND_CASE); |
---|
260 | if (matcher) { |
---|
261 | dirent *entry; |
---|
262 | while ((entry = readdir(dirstream)) != 0) { |
---|
263 | const char *name = entry->d_name; |
---|
264 | |
---|
265 | if (name[0] == '.' && (name[1] == 0 || (name[1] == '.' && name[2] == 0))) { |
---|
266 | ; // skip '.' and '..' |
---|
267 | } |
---|
268 | else { |
---|
269 | if (GBS_string_matches_regexp(name, matcher)) { |
---|
270 | const char *full = GB_concat_path(fulldir, name); |
---|
271 | if (!GB_is_directory(full)) { // skip directories |
---|
272 | names.put(ARB_strdup(full)); |
---|
273 | } |
---|
274 | } |
---|
275 | } |
---|
276 | } |
---|
277 | |
---|
278 | names.sort(GB_string_comparator, 0); |
---|
279 | |
---|
280 | GBS_free_matcher(matcher); |
---|
281 | } |
---|
282 | |
---|
283 | closedir(dirstream); |
---|
284 | } |
---|
285 | |
---|
286 | free(fulldir); |
---|
287 | } |
---|
288 | } |
---|
289 | |
---|
290 | const char *GB_get_arb_revision_tag() { |
---|
291 | /*! retrieve arb revision tag (e.g. "14358@trunk") from ../lib/revision_info.txt |
---|
292 | * File is generated by ../SOURCE_TOOLS/build_info.pl |
---|
293 | */ |
---|
294 | static SmartCharPtr tag; |
---|
295 | |
---|
296 | if (tag.isNull()) { |
---|
297 | char *name = GB_lib_file(true, NULL, "revision_info.txt"); |
---|
298 | tag = GB_read_file(name); |
---|
299 | |
---|
300 | char *nl = strchr(&*tag, '\n'); |
---|
301 | if (nl) *nl = 0; |
---|
302 | |
---|
303 | if (tag.isNull()) tag = GBS_global_string_copy("<failed to read %s>", name); |
---|
304 | free(name); |
---|
305 | } |
---|
306 | return &*tag; |
---|
307 | } |
---|
308 | |
---|
309 | |
---|
310 | // -------------------------------------------------------------------------------- |
---|
311 | |
---|
312 | #ifdef UNIT_TESTS |
---|
313 | #include <test_unit.h> |
---|
314 | |
---|
315 | static char *remove_path(const char *fullname, void *cl_path) { |
---|
316 | const char *path = (const char *)cl_path; |
---|
317 | return ARB_strdup(fullname+(ARB_strBeginsWith(fullname, path) ? strlen(path) : 0)); |
---|
318 | } |
---|
319 | |
---|
320 | static void GBT_transform_names(StrArray& dest, const StrArray& source, char *transform(const char *, void *), void *client_data) { |
---|
321 | for (int i = 0; source[i]; ++i) dest.put(transform(source[i], client_data)); |
---|
322 | } |
---|
323 | |
---|
324 | #define TEST_JOINED_FULLDIR_CONTENT_EQUALS(fulldir,mask,expected) do { \ |
---|
325 | StrArray contents; \ |
---|
326 | GBS_read_dir(contents, fulldir, mask); \ |
---|
327 | StrArray contents_no_path; \ |
---|
328 | GBT_transform_names(contents_no_path, contents, remove_path, (void*)fulldir); \ |
---|
329 | char *joined = GBT_join_strings(contents_no_path, '!'); \ |
---|
330 | TEST_EXPECT_EQUAL(joined, expected); \ |
---|
331 | free(joined); \ |
---|
332 | } while(0) |
---|
333 | |
---|
334 | #define TEST_JOINED_DIR_CONTENT_EQUALS(subdir,mask,expected) do { \ |
---|
335 | char *fulldir = ARB_strdup(GB_path_in_ARBHOME(subdir)); \ |
---|
336 | TEST_JOINED_FULLDIR_CONTENT_EQUALS(fulldir,mask,expected); \ |
---|
337 | free(fulldir); \ |
---|
338 | } while(0) |
---|
339 | |
---|
340 | void TEST_GBS_read_dir() { |
---|
341 | TEST_JOINED_DIR_CONTENT_EQUALS("GDE/CLUSTAL", "*.c", "/amenu.c!/clustalv.c!/gcgcheck.c!/myers.c!/sequence.c!/showpair.c!/trees.c!/upgma.c!/util.c"); |
---|
342 | TEST_JOINED_DIR_CONTENT_EQUALS("GDE/CLUSTAL", "/s.*\\.c/", "/clustalv.c!/myers.c!/sequence.c!/showpair.c!/trees.c"); |
---|
343 | |
---|
344 | // test a dir containing subdirectories |
---|
345 | TEST_JOINED_DIR_CONTENT_EQUALS("SL", NULL, "/Makefile!/README"); |
---|
346 | TEST_JOINED_DIR_CONTENT_EQUALS("SL", "*", "/Makefile!/README"); |
---|
347 | |
---|
348 | TEST_JOINED_FULLDIR_CONTENT_EQUALS("", "", ""); // allow GBS_read_dir to be called with "" -> returns empty filelist |
---|
349 | } |
---|
350 | |
---|
351 | void TEST_find_file() { |
---|
352 | gb_getenv_hook old = GB_install_getenv_hook(arb_test::fakeenv); |
---|
353 | |
---|
354 | TEST_EXPECT_EQUAL(GB_existing_file("min_ascii.arb", false), "min_ascii.arb"); |
---|
355 | TEST_EXPECT_NULL(GB_existing_file("nosuchfile", false)); |
---|
356 | |
---|
357 | char *tcporg = GB_lib_file(false, "", "arb_tcp_org.dat"); |
---|
358 | TEST_EXPECT_EQUAL(tcporg, GB_path_in_ARBHOME("lib/arb_tcp_org.dat")); |
---|
359 | TEST_EXPECT_NULL(GB_lib_file(true, "bla", "blub")); |
---|
360 | free(tcporg); |
---|
361 | |
---|
362 | char *status = GB_property_file(false, "status.arb"); |
---|
363 | TEST_EXPECT_EQUAL(status, GB_path_in_ARBHOME("lib/arb_default/status.arb")); |
---|
364 | TEST_EXPECT_NULL(GB_property_file(true, "undhepp")); |
---|
365 | free(status); |
---|
366 | |
---|
367 | TEST_EXPECT_EQUAL((void*)arb_test::fakeenv, (void*)GB_install_getenv_hook(old)); |
---|
368 | } |
---|
369 | TEST_PUBLISH(TEST_find_file); |
---|
370 | |
---|
371 | #endif // UNIT_TESTS |
---|