|
a |
|
b/src/pathut.cpp |
|
|
1 |
/* Copyright (C) 2004 J.F.Dockes
|
|
|
2 |
* This program is free software; you can redistribute it and/or modify
|
|
|
3 |
* it under the terms of the GNU General Public License as published by
|
|
|
4 |
* the Free Software Foundation; either version 2 of the License, or
|
|
|
5 |
* (at your option) any later version.
|
|
|
6 |
*
|
|
|
7 |
* This program is distributed in the hope that it will be useful,
|
|
|
8 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
9 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
10 |
* GNU General Public License for more details.
|
|
|
11 |
*
|
|
|
12 |
* You should have received a copy of the GNU General Public License
|
|
|
13 |
* along with this program; if not, write to the
|
|
|
14 |
* Free Software Foundation, Inc.,
|
|
|
15 |
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
|
16 |
*/
|
|
|
17 |
|
|
|
18 |
#ifndef TEST_PATHUT
|
|
|
19 |
#ifdef BUILDING_RECOLL
|
|
|
20 |
#include "autoconfig.h"
|
|
|
21 |
#else
|
|
|
22 |
#include "config.h"
|
|
|
23 |
#endif
|
|
|
24 |
|
|
|
25 |
#include <stdio.h>
|
|
|
26 |
#ifdef _WIN32
|
|
|
27 |
#include "dirent.h"
|
|
|
28 |
#include "safefcntl.h"
|
|
|
29 |
#include "safeunistd.h"
|
|
|
30 |
#include "safewindows.h"
|
|
|
31 |
#include "safesysstat.h"
|
|
|
32 |
#else
|
|
|
33 |
#include <fcntl.h>
|
|
|
34 |
#include <unistd.h>
|
|
|
35 |
#include <sys/param.h>
|
|
|
36 |
#include <pwd.h>
|
|
|
37 |
#include <sys/file.h>
|
|
|
38 |
#include <sys/stat.h>
|
|
|
39 |
#include <dirent.h>
|
|
|
40 |
#endif
|
|
|
41 |
#include <math.h>
|
|
|
42 |
#include <errno.h>
|
|
|
43 |
#include <sys/types.h>
|
|
|
44 |
|
|
|
45 |
// Let's include all files where statfs can be defined and hope for no
|
|
|
46 |
// conflict...
|
|
|
47 |
#ifdef HAVE_SYS_MOUNT_H
|
|
|
48 |
#include <sys/mount.h>
|
|
|
49 |
#endif
|
|
|
50 |
#ifdef HAVE_SYS_STATFS_H
|
|
|
51 |
#include <sys/statfs.h>
|
|
|
52 |
#endif
|
|
|
53 |
#ifdef HAVE_SYS_STATVFS_H
|
|
|
54 |
#include <sys/statvfs.h>
|
|
|
55 |
#endif
|
|
|
56 |
#ifdef HAVE_SYS_VFS_H
|
|
|
57 |
#include <sys/vfs.h>
|
|
|
58 |
#endif
|
|
|
59 |
|
|
|
60 |
#include <cstdlib>
|
|
|
61 |
#include <cstring>
|
|
|
62 |
#include <iostream>
|
|
|
63 |
#include <sstream>
|
|
|
64 |
#include <stack>
|
|
|
65 |
#include <set>
|
|
|
66 |
#include <vector>
|
|
|
67 |
|
|
|
68 |
#include "pathut.h"
|
|
|
69 |
#include "smallut.h"
|
|
|
70 |
|
|
|
71 |
using namespace std;
|
|
|
72 |
|
|
|
73 |
#ifdef _WIN32
|
|
|
74 |
/// Convert \ separators to /
|
|
|
75 |
void path_slashize(string& s)
|
|
|
76 |
{
|
|
|
77 |
for (string::size_type i = 0; i < s.size(); i++) {
|
|
|
78 |
if (s[i] == '\\') {
|
|
|
79 |
s[i] = '/';
|
|
|
80 |
}
|
|
|
81 |
}
|
|
|
82 |
}
|
|
|
83 |
static bool path_strlookslikedrive(const string& s)
|
|
|
84 |
{
|
|
|
85 |
return s.size() == 2 && isalpha(s[0]) && s[1] == ':';
|
|
|
86 |
}
|
|
|
87 |
|
|
|
88 |
static bool path_hasdrive(const string& s)
|
|
|
89 |
{
|
|
|
90 |
if (s.size() >= 2 && isalpha(s[0]) && s[1] == ':') {
|
|
|
91 |
return true;
|
|
|
92 |
}
|
|
|
93 |
return false;
|
|
|
94 |
}
|
|
|
95 |
static bool path_isdriveabs(const string& s)
|
|
|
96 |
{
|
|
|
97 |
if (s.size() >= 3 && isalpha(s[0]) && s[1] == ':' && s[2] == '/') {
|
|
|
98 |
return true;
|
|
|
99 |
}
|
|
|
100 |
return false;
|
|
|
101 |
}
|
|
|
102 |
#endif
|
|
|
103 |
|
|
|
104 |
#if defined(HAVE_SYS_MOUNT_H) || defined(HAVE_SYS_STATFS_H) || \
|
|
|
105 |
defined(HAVE_SYS_STATVFS_H) || defined(HAVE_SYS_VFS_H) || defined(_WIN32)
|
|
|
106 |
bool fsocc(const string& path, int *pc, long long *avmbs)
|
|
|
107 |
{
|
|
|
108 |
static const int FSOCC_MB = 1024 * 1024;
|
|
|
109 |
#ifdef _WIN32
|
|
|
110 |
ULARGE_INTEGER freebytesavail;
|
|
|
111 |
ULARGE_INTEGER totalbytes;
|
|
|
112 |
if (!GetDiskFreeSpaceEx(path.c_str(), &freebytesavail,
|
|
|
113 |
&totalbytes, NULL)) {
|
|
|
114 |
return false;
|
|
|
115 |
}
|
|
|
116 |
if (pc) {
|
|
|
117 |
*pc = int((100 * freebytesavail.QuadPart) / totalbytes.QuadPart);
|
|
|
118 |
}
|
|
|
119 |
if (avmbs) {
|
|
|
120 |
*avmbs = int(totalbytes.QuadPart / FSOCC_MB);
|
|
|
121 |
}
|
|
|
122 |
return true;
|
|
|
123 |
#else
|
|
|
124 |
#ifdef sun
|
|
|
125 |
struct statvfs buf;
|
|
|
126 |
if (statvfs(path.c_str(), &buf) != 0) {
|
|
|
127 |
return false;
|
|
|
128 |
}
|
|
|
129 |
#else
|
|
|
130 |
struct statfs buf;
|
|
|
131 |
if (statfs(path.c_str(), &buf) != 0) {
|
|
|
132 |
return false;
|
|
|
133 |
}
|
|
|
134 |
#endif
|
|
|
135 |
|
|
|
136 |
// used blocks
|
|
|
137 |
double fpc = 0.0;
|
|
|
138 |
#define FSOCC_USED (double(buf.f_blocks - buf.f_bfree))
|
|
|
139 |
#define FSOCC_TOTAVAIL (FSOCC_USED + double(buf.f_bavail))
|
|
|
140 |
if (FSOCC_TOTAVAIL > 0) {
|
|
|
141 |
fpc = 100.0 * FSOCC_USED / FSOCC_TOTAVAIL;
|
|
|
142 |
}
|
|
|
143 |
if (pc) {
|
|
|
144 |
*pc = int(fpc);
|
|
|
145 |
}
|
|
|
146 |
if (avmbs) {
|
|
|
147 |
*avmbs = 0;
|
|
|
148 |
if (buf.f_bsize > 0) {
|
|
|
149 |
int ratio = buf.f_bsize > FSOCC_MB ? buf.f_bsize / FSOCC_MB :
|
|
|
150 |
FSOCC_MB / buf.f_bsize;
|
|
|
151 |
|
|
|
152 |
*avmbs = buf.f_bsize > FSOCC_MB ?
|
|
|
153 |
((long long)buf.f_bavail) * ratio :
|
|
|
154 |
((long long)buf.f_bavail) / ratio;
|
|
|
155 |
}
|
|
|
156 |
}
|
|
|
157 |
return true;
|
|
|
158 |
#endif
|
|
|
159 |
}
|
|
|
160 |
#endif // we have found an appropriate include file
|
|
|
161 |
|
|
|
162 |
string path_PATHsep()
|
|
|
163 |
{
|
|
|
164 |
static const string w(";");
|
|
|
165 |
static const string u(":");
|
|
|
166 |
#ifdef _WIN32
|
|
|
167 |
return w;
|
|
|
168 |
#else
|
|
|
169 |
return u;
|
|
|
170 |
#endif
|
|
|
171 |
}
|
|
|
172 |
|
|
|
173 |
void path_catslash(string& s)
|
|
|
174 |
{
|
|
|
175 |
#ifdef _WIN32
|
|
|
176 |
path_slashize(s);
|
|
|
177 |
#endif
|
|
|
178 |
if (s.empty() || s[s.length() - 1] != '/') {
|
|
|
179 |
s += '/';
|
|
|
180 |
}
|
|
|
181 |
}
|
|
|
182 |
|
|
|
183 |
string path_cat(const string& s1, const string& s2)
|
|
|
184 |
{
|
|
|
185 |
string res = s1;
|
|
|
186 |
path_catslash(res);
|
|
|
187 |
res += s2;
|
|
|
188 |
return res;
|
|
|
189 |
}
|
|
|
190 |
|
|
|
191 |
string path_getfather(const string& s)
|
|
|
192 |
{
|
|
|
193 |
string father = s;
|
|
|
194 |
#ifdef _WIN32
|
|
|
195 |
path_slashize(father);
|
|
|
196 |
#endif
|
|
|
197 |
|
|
|
198 |
// ??
|
|
|
199 |
if (father.empty()) {
|
|
|
200 |
return "./";
|
|
|
201 |
}
|
|
|
202 |
|
|
|
203 |
if (path_isroot(father)) {
|
|
|
204 |
return father;
|
|
|
205 |
}
|
|
|
206 |
|
|
|
207 |
if (father[father.length() - 1] == '/') {
|
|
|
208 |
// Input ends with /. Strip it, root special case was tested above
|
|
|
209 |
father.erase(father.length() - 1);
|
|
|
210 |
}
|
|
|
211 |
|
|
|
212 |
string::size_type slp = father.rfind('/');
|
|
|
213 |
if (slp == string::npos) {
|
|
|
214 |
return "./";
|
|
|
215 |
}
|
|
|
216 |
|
|
|
217 |
father.erase(slp);
|
|
|
218 |
path_catslash(father);
|
|
|
219 |
return father;
|
|
|
220 |
}
|
|
|
221 |
|
|
|
222 |
string path_getsimple(const string& s)
|
|
|
223 |
{
|
|
|
224 |
string simple = s;
|
|
|
225 |
#ifdef _WIN32
|
|
|
226 |
path_slashize(simple);
|
|
|
227 |
#endif
|
|
|
228 |
|
|
|
229 |
if (simple.empty()) {
|
|
|
230 |
return simple;
|
|
|
231 |
}
|
|
|
232 |
|
|
|
233 |
string::size_type slp = simple.rfind('/');
|
|
|
234 |
if (slp == string::npos) {
|
|
|
235 |
return simple;
|
|
|
236 |
}
|
|
|
237 |
|
|
|
238 |
simple.erase(0, slp + 1);
|
|
|
239 |
return simple;
|
|
|
240 |
}
|
|
|
241 |
|
|
|
242 |
string path_basename(const string& s, const string& suff)
|
|
|
243 |
{
|
|
|
244 |
string simple = path_getsimple(s);
|
|
|
245 |
string::size_type pos = string::npos;
|
|
|
246 |
if (suff.length() && simple.length() > suff.length()) {
|
|
|
247 |
pos = simple.rfind(suff);
|
|
|
248 |
if (pos != string::npos && pos + suff.length() == simple.length()) {
|
|
|
249 |
return simple.substr(0, pos);
|
|
|
250 |
}
|
|
|
251 |
}
|
|
|
252 |
return simple;
|
|
|
253 |
}
|
|
|
254 |
|
|
|
255 |
string path_suffix(const string& s)
|
|
|
256 |
{
|
|
|
257 |
string::size_type dotp = s.rfind('.');
|
|
|
258 |
if (dotp == string::npos) {
|
|
|
259 |
return string();
|
|
|
260 |
}
|
|
|
261 |
return s.substr(dotp + 1);
|
|
|
262 |
}
|
|
|
263 |
|
|
|
264 |
string path_home()
|
|
|
265 |
{
|
|
|
266 |
#ifdef _WIN32
|
|
|
267 |
string dir;
|
|
|
268 |
const char *cp = getenv("USERPROFILE");
|
|
|
269 |
if (cp != 0) {
|
|
|
270 |
dir = cp;
|
|
|
271 |
}
|
|
|
272 |
if (dir.empty()) {
|
|
|
273 |
cp = getenv("HOMEDRIVE");
|
|
|
274 |
if (cp != 0) {
|
|
|
275 |
const char *cp1 = getenv("HOMEPATH");
|
|
|
276 |
if (cp1 != 0) {
|
|
|
277 |
dir = string(cp) + string(cp1);
|
|
|
278 |
}
|
|
|
279 |
}
|
|
|
280 |
}
|
|
|
281 |
if (dir.empty()) {
|
|
|
282 |
dir = "C:\\";
|
|
|
283 |
}
|
|
|
284 |
dir = path_canon(dir);
|
|
|
285 |
path_catslash(dir);
|
|
|
286 |
return dir;
|
|
|
287 |
#else
|
|
|
288 |
uid_t uid = getuid();
|
|
|
289 |
|
|
|
290 |
struct passwd *entry = getpwuid(uid);
|
|
|
291 |
if (entry == 0) {
|
|
|
292 |
const char *cp = getenv("HOME");
|
|
|
293 |
if (cp) {
|
|
|
294 |
return cp;
|
|
|
295 |
} else {
|
|
|
296 |
return "/";
|
|
|
297 |
}
|
|
|
298 |
}
|
|
|
299 |
|
|
|
300 |
string homedir = entry->pw_dir;
|
|
|
301 |
path_catslash(homedir);
|
|
|
302 |
return homedir;
|
|
|
303 |
#endif
|
|
|
304 |
}
|
|
|
305 |
|
|
|
306 |
// The default place to store the default config and other stuff (e.g webqueue)
|
|
|
307 |
string path_homedata()
|
|
|
308 |
{
|
|
|
309 |
#ifdef _WIN32
|
|
|
310 |
const char *cp = getenv("LOCALAPPDATA");
|
|
|
311 |
string dir;
|
|
|
312 |
if (cp != 0) {
|
|
|
313 |
dir = path_canon(cp);
|
|
|
314 |
}
|
|
|
315 |
if (dir.empty()) {
|
|
|
316 |
dir = path_cat(path_home(), "AppData/Local/");
|
|
|
317 |
}
|
|
|
318 |
return dir;
|
|
|
319 |
#else
|
|
|
320 |
// We should use an xdg-conforming location, but, history...
|
|
|
321 |
return path_home();
|
|
|
322 |
#endif
|
|
|
323 |
}
|
|
|
324 |
|
|
|
325 |
string path_tildexpand(const string& s)
|
|
|
326 |
{
|
|
|
327 |
if (s.empty() || s[0] != '~') {
|
|
|
328 |
return s;
|
|
|
329 |
}
|
|
|
330 |
string o = s;
|
|
|
331 |
#ifdef _WIN32
|
|
|
332 |
path_slashize(o);
|
|
|
333 |
#endif
|
|
|
334 |
|
|
|
335 |
if (s.length() == 1) {
|
|
|
336 |
o.replace(0, 1, path_home());
|
|
|
337 |
} else if (s[1] == '/') {
|
|
|
338 |
o.replace(0, 2, path_home());
|
|
|
339 |
} else {
|
|
|
340 |
string::size_type pos = s.find('/');
|
|
|
341 |
string::size_type l = (pos == string::npos) ? s.length() - 1 : pos - 1;
|
|
|
342 |
#ifdef _WIN32
|
|
|
343 |
// Dont know what this means. Just replace with HOME
|
|
|
344 |
o.replace(0, l + 1, path_home());
|
|
|
345 |
#else
|
|
|
346 |
struct passwd *entry = getpwnam(s.substr(1, l).c_str());
|
|
|
347 |
if (entry) {
|
|
|
348 |
o.replace(0, l + 1, entry->pw_dir);
|
|
|
349 |
}
|
|
|
350 |
#endif
|
|
|
351 |
}
|
|
|
352 |
return o;
|
|
|
353 |
}
|
|
|
354 |
|
|
|
355 |
bool path_isroot(const string& path)
|
|
|
356 |
{
|
|
|
357 |
if (path.size() == 1 && path[0] == '/') {
|
|
|
358 |
return true;
|
|
|
359 |
}
|
|
|
360 |
#ifdef _WIN32
|
|
|
361 |
if (path.size() == 3 && isalpha(path[0]) && path[1] == ':' &&
|
|
|
362 |
(path[2] == '/' || path[2] == '\\')) {
|
|
|
363 |
return true;
|
|
|
364 |
}
|
|
|
365 |
#endif
|
|
|
366 |
return false;
|
|
|
367 |
}
|
|
|
368 |
|
|
|
369 |
bool path_isabsolute(const string& path)
|
|
|
370 |
{
|
|
|
371 |
if (!path.empty() && (path[0] == '/'
|
|
|
372 |
#ifdef _WIN32
|
|
|
373 |
|| path_isdriveabs(path)
|
|
|
374 |
#endif
|
|
|
375 |
)) {
|
|
|
376 |
return true;
|
|
|
377 |
}
|
|
|
378 |
return false;
|
|
|
379 |
}
|
|
|
380 |
|
|
|
381 |
string path_absolute(const string& is)
|
|
|
382 |
{
|
|
|
383 |
if (is.length() == 0) {
|
|
|
384 |
return is;
|
|
|
385 |
}
|
|
|
386 |
string s = is;
|
|
|
387 |
if (!path_isabsolute(s)) {
|
|
|
388 |
char buf[MAXPATHLEN];
|
|
|
389 |
if (!getcwd(buf, MAXPATHLEN)) {
|
|
|
390 |
return string();
|
|
|
391 |
}
|
|
|
392 |
s = path_cat(string(buf), s);
|
|
|
393 |
#ifdef _WIN32
|
|
|
394 |
path_slashize(s);
|
|
|
395 |
#endif
|
|
|
396 |
}
|
|
|
397 |
return s;
|
|
|
398 |
}
|
|
|
399 |
|
|
|
400 |
string path_canon(const string& is, const string* cwd)
|
|
|
401 |
{
|
|
|
402 |
if (is.length() == 0) {
|
|
|
403 |
return is;
|
|
|
404 |
}
|
|
|
405 |
string s = is;
|
|
|
406 |
#ifdef _WIN32
|
|
|
407 |
path_slashize(s);
|
|
|
408 |
// fix possible path from file: absolute url
|
|
|
409 |
if (s.size() && s[0] == '/' && path_hasdrive(s.substr(1))) {
|
|
|
410 |
s = s.substr(1);
|
|
|
411 |
}
|
|
|
412 |
#endif
|
|
|
413 |
|
|
|
414 |
if (!path_isabsolute(s)) {
|
|
|
415 |
char buf[MAXPATHLEN];
|
|
|
416 |
const char *cwdp = buf;
|
|
|
417 |
if (cwd) {
|
|
|
418 |
cwdp = cwd->c_str();
|
|
|
419 |
} else {
|
|
|
420 |
if (!getcwd(buf, MAXPATHLEN)) {
|
|
|
421 |
return string();
|
|
|
422 |
}
|
|
|
423 |
}
|
|
|
424 |
s = path_cat(string(cwdp), s);
|
|
|
425 |
}
|
|
|
426 |
vector<string> elems;
|
|
|
427 |
stringToTokens(s, elems, "/");
|
|
|
428 |
vector<string> cleaned;
|
|
|
429 |
for (vector<string>::const_iterator it = elems.begin();
|
|
|
430 |
it != elems.end(); it++) {
|
|
|
431 |
if (*it == "..") {
|
|
|
432 |
if (!cleaned.empty()) {
|
|
|
433 |
cleaned.pop_back();
|
|
|
434 |
}
|
|
|
435 |
} else if (it->empty() || *it == ".") {
|
|
|
436 |
} else {
|
|
|
437 |
cleaned.push_back(*it);
|
|
|
438 |
}
|
|
|
439 |
}
|
|
|
440 |
string ret;
|
|
|
441 |
if (!cleaned.empty()) {
|
|
|
442 |
for (vector<string>::const_iterator it = cleaned.begin();
|
|
|
443 |
it != cleaned.end(); it++) {
|
|
|
444 |
ret += "/";
|
|
|
445 |
#ifdef _WIN32
|
|
|
446 |
if (it == cleaned.begin() && path_strlookslikedrive(*it)) {
|
|
|
447 |
// Get rid of just added initial "/"
|
|
|
448 |
ret.clear();
|
|
|
449 |
}
|
|
|
450 |
#endif
|
|
|
451 |
ret += *it;
|
|
|
452 |
}
|
|
|
453 |
} else {
|
|
|
454 |
ret = "/";
|
|
|
455 |
}
|
|
|
456 |
return ret;
|
|
|
457 |
}
|
|
|
458 |
|
|
|
459 |
bool path_makepath(const string& ipath, int mode)
|
|
|
460 |
{
|
|
|
461 |
string path = path_canon(ipath);
|
|
|
462 |
vector<string> elems;
|
|
|
463 |
stringToTokens(path, elems, "/");
|
|
|
464 |
path = "/";
|
|
|
465 |
for (vector<string>::const_iterator it = elems.begin();
|
|
|
466 |
it != elems.end(); it++) {
|
|
|
467 |
#ifdef _WIN32
|
|
|
468 |
if (it == elems.begin() && path_strlookslikedrive(*it)) {
|
|
|
469 |
path = "";
|
|
|
470 |
}
|
|
|
471 |
#endif
|
|
|
472 |
path += *it;
|
|
|
473 |
// Not using path_isdir() here, because this cant grok symlinks
|
|
|
474 |
// If we hit an existing file, no worry, mkdir will just fail.
|
|
|
475 |
if (access(path.c_str(), 0) != 0) {
|
|
|
476 |
if (mkdir(path.c_str(), mode) != 0) {
|
|
|
477 |
return false;
|
|
|
478 |
}
|
|
|
479 |
}
|
|
|
480 |
path += "/";
|
|
|
481 |
}
|
|
|
482 |
return true;
|
|
|
483 |
}
|
|
|
484 |
|
|
|
485 |
bool path_isdir(const string& path)
|
|
|
486 |
{
|
|
|
487 |
struct stat st;
|
|
|
488 |
if (lstat(path.c_str(), &st) < 0) {
|
|
|
489 |
return false;
|
|
|
490 |
}
|
|
|
491 |
if (S_ISDIR(st.st_mode)) {
|
|
|
492 |
return true;
|
|
|
493 |
}
|
|
|
494 |
return false;
|
|
|
495 |
}
|
|
|
496 |
|
|
|
497 |
long long path_filesize(const string& path)
|
|
|
498 |
{
|
|
|
499 |
struct stat st;
|
|
|
500 |
if (stat(path.c_str(), &st) < 0) {
|
|
|
501 |
return -1;
|
|
|
502 |
}
|
|
|
503 |
return (long long)st.st_size;
|
|
|
504 |
}
|
|
|
505 |
|
|
|
506 |
int path_fileprops(const std::string path, struct stat *stp, bool follow)
|
|
|
507 |
{
|
|
|
508 |
if (!stp) {
|
|
|
509 |
return -1;
|
|
|
510 |
}
|
|
|
511 |
memset(stp, 0, sizeof(struct stat));
|
|
|
512 |
struct stat mst;
|
|
|
513 |
int ret = follow ? stat(path.c_str(), &mst) : lstat(path.c_str(), &mst);
|
|
|
514 |
if (ret != 0) {
|
|
|
515 |
return ret;
|
|
|
516 |
}
|
|
|
517 |
stp->st_size = mst.st_size;
|
|
|
518 |
stp->st_mode = mst.st_mode;
|
|
|
519 |
stp->st_mtime = mst.st_mtime;
|
|
|
520 |
#ifdef _WIN32
|
|
|
521 |
stp->st_ctime = mst.st_mtime;
|
|
|
522 |
#else
|
|
|
523 |
stp->st_ino = mst.st_ino;
|
|
|
524 |
stp->st_dev = mst.st_dev;
|
|
|
525 |
stp->st_ctime = mst.st_ctime;
|
|
|
526 |
#endif
|
|
|
527 |
return 0;
|
|
|
528 |
}
|
|
|
529 |
|
|
|
530 |
bool path_exists(const string& path)
|
|
|
531 |
{
|
|
|
532 |
return access(path.c_str(), 0) == 0;
|
|
|
533 |
}
|
|
|
534 |
|
|
|
535 |
// Allowed punctuation in the path part of an URI according to RFC2396
|
|
|
536 |
// -_.!~*'():@&=+$,
|
|
|
537 |
/*
|
|
|
538 |
21 !
|
|
|
539 |
22 "
|
|
|
540 |
23 #
|
|
|
541 |
24 $
|
|
|
542 |
25 %
|
|
|
543 |
26 &
|
|
|
544 |
27 '
|
|
|
545 |
28 (
|
|
|
546 |
29 )
|
|
|
547 |
2A *
|
|
|
548 |
2B +
|
|
|
549 |
2C ,
|
|
|
550 |
2D -
|
|
|
551 |
2E .
|
|
|
552 |
2F /
|
|
|
553 |
30 0
|
|
|
554 |
...
|
|
|
555 |
39 9
|
|
|
556 |
3A :
|
|
|
557 |
3B ;
|
|
|
558 |
3C <
|
|
|
559 |
3D =
|
|
|
560 |
3E >
|
|
|
561 |
3F ?
|
|
|
562 |
40 @
|
|
|
563 |
41 A
|
|
|
564 |
...
|
|
|
565 |
5A Z
|
|
|
566 |
5B [
|
|
|
567 |
5C \
|
|
|
568 |
5D ]
|
|
|
569 |
5E ^
|
|
|
570 |
5F _
|
|
|
571 |
60 `
|
|
|
572 |
61 a
|
|
|
573 |
...
|
|
|
574 |
7A z
|
|
|
575 |
7B {
|
|
|
576 |
7C |
|
|
|
577 |
7D }
|
|
|
578 |
7E ~
|
|
|
579 |
7F DEL
|
|
|
580 |
*/
|
|
|
581 |
string url_encode(const string& url, string::size_type offs)
|
|
|
582 |
{
|
|
|
583 |
string out = url.substr(0, offs);
|
|
|
584 |
const char *cp = url.c_str();
|
|
|
585 |
for (string::size_type i = offs; i < url.size(); i++) {
|
|
|
586 |
unsigned int c;
|
|
|
587 |
const char *h = "0123456789ABCDEF";
|
|
|
588 |
c = cp[i];
|
|
|
589 |
if (c <= 0x20 ||
|
|
|
590 |
c >= 0x7f ||
|
|
|
591 |
c == '"' ||
|
|
|
592 |
c == '#' ||
|
|
|
593 |
c == '%' ||
|
|
|
594 |
c == ';' ||
|
|
|
595 |
c == '<' ||
|
|
|
596 |
c == '>' ||
|
|
|
597 |
c == '?' ||
|
|
|
598 |
c == '[' ||
|
|
|
599 |
c == '\\' ||
|
|
|
600 |
c == ']' ||
|
|
|
601 |
c == '^' ||
|
|
|
602 |
c == '`' ||
|
|
|
603 |
c == '{' ||
|
|
|
604 |
c == '|' ||
|
|
|
605 |
c == '}') {
|
|
|
606 |
out += '%';
|
|
|
607 |
out += h[(c >> 4) & 0xf];
|
|
|
608 |
out += h[c & 0xf];
|
|
|
609 |
} else {
|
|
|
610 |
out += char(c);
|
|
|
611 |
}
|
|
|
612 |
}
|
|
|
613 |
return out;
|
|
|
614 |
}
|
|
|
615 |
|
|
|
616 |
string url_gpath(const string& url)
|
|
|
617 |
{
|
|
|
618 |
// Remove the access schema part (or whatever it's called)
|
|
|
619 |
string::size_type colon = url.find_first_of(":");
|
|
|
620 |
if (colon == string::npos || colon == url.size() - 1) {
|
|
|
621 |
return url;
|
|
|
622 |
}
|
|
|
623 |
// If there are non-alphanum chars before the ':', then there
|
|
|
624 |
// probably is no scheme. Whatever...
|
|
|
625 |
for (string::size_type i = 0; i < colon; i++) {
|
|
|
626 |
if (!isalnum(url.at(i))) {
|
|
|
627 |
return url;
|
|
|
628 |
}
|
|
|
629 |
}
|
|
|
630 |
|
|
|
631 |
// In addition we canonize the path to remove empty host parts
|
|
|
632 |
// (for compatibility with older versions of recoll where file://
|
|
|
633 |
// was hardcoded, but the local path was used for doc
|
|
|
634 |
// identification.
|
|
|
635 |
return path_canon(url.substr(colon + 1));
|
|
|
636 |
}
|
|
|
637 |
|
|
|
638 |
string url_parentfolder(const string& url)
|
|
|
639 |
{
|
|
|
640 |
// In general, the parent is the directory above the full path
|
|
|
641 |
string parenturl = path_getfather(url_gpath(url));
|
|
|
642 |
// But if this is http, make sure to keep the host part. Recoll
|
|
|
643 |
// only has file or http urls for now.
|
|
|
644 |
bool isfileurl = urlisfileurl(url);
|
|
|
645 |
if (!isfileurl && parenturl == "/") {
|
|
|
646 |
parenturl = url_gpath(url);
|
|
|
647 |
}
|
|
|
648 |
return isfileurl ? string("file://") + parenturl :
|
|
|
649 |
string("http://") + parenturl;
|
|
|
650 |
}
|
|
|
651 |
|
|
|
652 |
|
|
|
653 |
// Convert to file path if url is like file:
|
|
|
654 |
// Note: this only works with our internal pseudo-urls which are not
|
|
|
655 |
// encoded/escaped
|
|
|
656 |
string fileurltolocalpath(string url)
|
|
|
657 |
{
|
|
|
658 |
if (url.find("file://") == 0) {
|
|
|
659 |
url = url.substr(7, string::npos);
|
|
|
660 |
} else {
|
|
|
661 |
return string();
|
|
|
662 |
}
|
|
|
663 |
|
|
|
664 |
#ifdef _WIN32
|
|
|
665 |
// Absolute file urls are like: file:///c:/mydir/...
|
|
|
666 |
// Get rid of the initial '/'
|
|
|
667 |
if (url.size() >= 3 && url[0] == '/' && isalpha(url[1]) && url[2] == ':') {
|
|
|
668 |
url = url.substr(1);
|
|
|
669 |
}
|
|
|
670 |
#endif
|
|
|
671 |
|
|
|
672 |
// Removing the fragment part. This is exclusively used when
|
|
|
673 |
// executing a viewer for the recoll manual, and we only strip the
|
|
|
674 |
// part after # if it is preceded by .html
|
|
|
675 |
string::size_type pos;
|
|
|
676 |
if ((pos = url.rfind(".html#")) != string::npos) {
|
|
|
677 |
url.erase(pos + 5);
|
|
|
678 |
} else if ((pos = url.rfind(".htm#")) != string::npos) {
|
|
|
679 |
url.erase(pos + 4);
|
|
|
680 |
}
|
|
|
681 |
|
|
|
682 |
return url;
|
|
|
683 |
}
|
|
|
684 |
|
|
|
685 |
static const string cstr_fileu("file://");
|
|
|
686 |
|
|
|
687 |
string path_pathtofileurl(const string& path)
|
|
|
688 |
{
|
|
|
689 |
// We're supposed to receive a canonic absolute path, but on windows we
|
|
|
690 |
// may need to add a '/' in front of the drive spec
|
|
|
691 |
string url(cstr_fileu);
|
|
|
692 |
if (path.empty() || path[0] != '/') {
|
|
|
693 |
url.push_back('/');
|
|
|
694 |
}
|
|
|
695 |
url += path;
|
|
|
696 |
return url;
|
|
|
697 |
}
|
|
|
698 |
|
|
|
699 |
bool urlisfileurl(const string& url)
|
|
|
700 |
{
|
|
|
701 |
return url.find("file://") == 0;
|
|
|
702 |
}
|
|
|
703 |
|
|
|
704 |
bool readdir(const string& dir, string& reason, set<string>& entries)
|
|
|
705 |
{
|
|
|
706 |
struct stat st;
|
|
|
707 |
int statret;
|
|
|
708 |
ostringstream msg;
|
|
|
709 |
DIR *d = 0;
|
|
|
710 |
statret = lstat(dir.c_str(), &st);
|
|
|
711 |
if (statret == -1) {
|
|
|
712 |
msg << "readdir: cant stat " << dir << " errno " << errno;
|
|
|
713 |
goto out;
|
|
|
714 |
}
|
|
|
715 |
if (!S_ISDIR(st.st_mode)) {
|
|
|
716 |
msg << "readdir: " << dir << " not a directory";
|
|
|
717 |
goto out;
|
|
|
718 |
}
|
|
|
719 |
if (access(dir.c_str(), R_OK) < 0) {
|
|
|
720 |
msg << "readdir: no read access to " << dir;
|
|
|
721 |
goto out;
|
|
|
722 |
}
|
|
|
723 |
|
|
|
724 |
d = opendir(dir.c_str());
|
|
|
725 |
if (d == 0) {
|
|
|
726 |
msg << "readdir: cant opendir " << dir << ", errno " << errno;
|
|
|
727 |
goto out;
|
|
|
728 |
}
|
|
|
729 |
|
|
|
730 |
struct dirent *ent;
|
|
|
731 |
while ((ent = readdir(d)) != 0) {
|
|
|
732 |
if (!strcmp(ent->d_name, ".") || !strcmp(ent->d_name, "..")) {
|
|
|
733 |
continue;
|
|
|
734 |
}
|
|
|
735 |
entries.insert(ent->d_name);
|
|
|
736 |
}
|
|
|
737 |
|
|
|
738 |
out:
|
|
|
739 |
if (d) {
|
|
|
740 |
closedir(d);
|
|
|
741 |
}
|
|
|
742 |
reason = msg.str();
|
|
|
743 |
if (reason.empty()) {
|
|
|
744 |
return true;
|
|
|
745 |
}
|
|
|
746 |
return false;
|
|
|
747 |
}
|
|
|
748 |
|
|
|
749 |
// We do not want to mess with the pidfile content in the destructor:
|
|
|
750 |
// the lock might still be in use in a child process. In fact as much
|
|
|
751 |
// as we'd like to reset the pid inside the file when we're done, it
|
|
|
752 |
// would be very difficult to do it right and it's probably best left
|
|
|
753 |
// alone.
|
|
|
754 |
Pidfile::~Pidfile()
|
|
|
755 |
{
|
|
|
756 |
if (m_fd >= 0) {
|
|
|
757 |
::close(m_fd);
|
|
|
758 |
}
|
|
|
759 |
m_fd = -1;
|
|
|
760 |
}
|
|
|
761 |
|
|
|
762 |
pid_t Pidfile::read_pid()
|
|
|
763 |
{
|
|
|
764 |
int fd = ::open(m_path.c_str(), O_RDONLY);
|
|
|
765 |
if (fd == -1) {
|
|
|
766 |
return (pid_t) - 1;
|
|
|
767 |
}
|
|
|
768 |
|
|
|
769 |
char buf[16];
|
|
|
770 |
int i = read(fd, buf, sizeof(buf) - 1);
|
|
|
771 |
::close(fd);
|
|
|
772 |
if (i <= 0) {
|
|
|
773 |
return (pid_t) - 1;
|
|
|
774 |
}
|
|
|
775 |
buf[i] = '\0';
|
|
|
776 |
char *endptr;
|
|
|
777 |
pid_t pid = strtol(buf, &endptr, 10);
|
|
|
778 |
if (endptr != &buf[i]) {
|
|
|
779 |
return (pid_t) - 1;
|
|
|
780 |
}
|
|
|
781 |
return pid;
|
|
|
782 |
}
|
|
|
783 |
|
|
|
784 |
int Pidfile::flopen()
|
|
|
785 |
{
|
|
|
786 |
const char *path = m_path.c_str();
|
|
|
787 |
if ((m_fd = ::open(path, O_RDWR | O_CREAT, 0644)) == -1) {
|
|
|
788 |
m_reason = "Open failed: [" + m_path + "]: " + strerror(errno);
|
|
|
789 |
return -1;
|
|
|
790 |
}
|
|
|
791 |
|
|
|
792 |
#ifdef sun
|
|
|
793 |
struct flock lockdata;
|
|
|
794 |
lockdata.l_start = 0;
|
|
|
795 |
lockdata.l_len = 0;
|
|
|
796 |
lockdata.l_type = F_WRLCK;
|
|
|
797 |
lockdata.l_whence = SEEK_SET;
|
|
|
798 |
if (fcntl(m_fd, F_SETLK, &lockdata) != 0) {
|
|
|
799 |
int serrno = errno;
|
|
|
800 |
(void)::close(m_fd);
|
|
|
801 |
errno = serrno;
|
|
|
802 |
m_reason = "fcntl lock failed";
|
|
|
803 |
return -1;
|
|
|
804 |
}
|
|
|
805 |
#else
|
|
|
806 |
#ifdef _WIN32
|
|
|
807 |
return 0;
|
|
|
808 |
#else
|
|
|
809 |
int operation = LOCK_EX | LOCK_NB;
|
|
|
810 |
if (flock(m_fd, operation) == -1) {
|
|
|
811 |
int serrno = errno;
|
|
|
812 |
(void)::close(m_fd);
|
|
|
813 |
errno = serrno;
|
|
|
814 |
m_reason = "flock failed";
|
|
|
815 |
return -1;
|
|
|
816 |
}
|
|
|
817 |
#endif // ! win32
|
|
|
818 |
#endif // ! sun
|
|
|
819 |
|
|
|
820 |
if (ftruncate(m_fd, 0) != 0) {
|
|
|
821 |
/* can't happen [tm] */
|
|
|
822 |
int serrno = errno;
|
|
|
823 |
(void)::close(m_fd);
|
|
|
824 |
errno = serrno;
|
|
|
825 |
m_reason = "ftruncate failed";
|
|
|
826 |
return -1;
|
|
|
827 |
}
|
|
|
828 |
return 0;
|
|
|
829 |
}
|
|
|
830 |
|
|
|
831 |
pid_t Pidfile::open()
|
|
|
832 |
{
|
|
|
833 |
if (flopen() < 0) {
|
|
|
834 |
return read_pid();
|
|
|
835 |
}
|
|
|
836 |
return (pid_t)0;
|
|
|
837 |
}
|
|
|
838 |
|
|
|
839 |
int Pidfile::write_pid()
|
|
|
840 |
{
|
|
|
841 |
/* truncate to allow multiple calls */
|
|
|
842 |
if (ftruncate(m_fd, 0) == -1) {
|
|
|
843 |
m_reason = "ftruncate failed";
|
|
|
844 |
return -1;
|
|
|
845 |
}
|
|
|
846 |
char pidstr[20];
|
|
|
847 |
sprintf(pidstr, "%u", int(getpid()));
|
|
|
848 |
lseek(m_fd, 0, 0);
|
|
|
849 |
if (::write(m_fd, pidstr, strlen(pidstr)) != (ssize_t)strlen(pidstr)) {
|
|
|
850 |
m_reason = "write failed";
|
|
|
851 |
return -1;
|
|
|
852 |
}
|
|
|
853 |
return 0;
|
|
|
854 |
}
|
|
|
855 |
|
|
|
856 |
int Pidfile::close()
|
|
|
857 |
{
|
|
|
858 |
return ::close(m_fd);
|
|
|
859 |
}
|
|
|
860 |
|
|
|
861 |
int Pidfile::remove()
|
|
|
862 |
{
|
|
|
863 |
return unlink(m_path.c_str());
|
|
|
864 |
}
|
|
|
865 |
|
|
|
866 |
// Call funcs that need static init (not initially reentrant)
|
|
|
867 |
void pathut_init_mt()
|
|
|
868 |
{
|
|
|
869 |
path_home();
|
|
|
870 |
}
|
|
|
871 |
|
|
|
872 |
|
|
|
873 |
#else // TEST_PATHUT
|
|
|
874 |
#include <stdlib.h>
|
|
|
875 |
#include <iostream>
|
|
|
876 |
using namespace std;
|
|
|
877 |
|
|
|
878 |
#include "pathut.h"
|
|
|
879 |
|
|
|
880 |
void path_to_thumb(const string& _input)
|
|
|
881 |
{
|
|
|
882 |
string input(_input);
|
|
|
883 |
// Make absolute path if needed
|
|
|
884 |
if (input[0] != '/') {
|
|
|
885 |
input = path_absolute(input);
|
|
|
886 |
}
|
|
|
887 |
|
|
|
888 |
input = string("file://") + path_canon(input);
|
|
|
889 |
|
|
|
890 |
string path;
|
|
|
891 |
//path = url_encode(input, 7);
|
|
|
892 |
thumbPathForUrl(input, 7, path);
|
|
|
893 |
cout << path << endl;
|
|
|
894 |
}
|
|
|
895 |
|
|
|
896 |
const char *tstvec[] = {"", "/", "/dir", "/dir/", "/dir1/dir2",
|
|
|
897 |
"/dir1/dir2",
|
|
|
898 |
"./dir", "./dir1/", "dir", "../dir", "/dir/toto.c",
|
|
|
899 |
"/dir/.c", "/dir/toto.txt", "toto.txt1"
|
|
|
900 |
};
|
|
|
901 |
|
|
|
902 |
const string ttvec[] = {"/dir", "", "~", "~/sub", "~root", "~root/sub",
|
|
|
903 |
"~nosuch", "~nosuch/sub"
|
|
|
904 |
};
|
|
|
905 |
int nttvec = sizeof(ttvec) / sizeof(string);
|
|
|
906 |
|
|
|
907 |
const char *thisprog;
|
|
|
908 |
|
|
|
909 |
int main(int argc, const char **argv)
|
|
|
910 |
{
|
|
|
911 |
thisprog = *argv++;
|
|
|
912 |
argc--;
|
|
|
913 |
|
|
|
914 |
string s;
|
|
|
915 |
vector<string>::const_iterator it;
|
|
|
916 |
#if 0
|
|
|
917 |
for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
|
|
|
918 |
cout << tstvec[i] << " Father " << path_getfather(tstvec[i]) << endl;
|
|
|
919 |
}
|
|
|
920 |
for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
|
|
|
921 |
cout << tstvec[i] << " Simple " << path_getsimple(tstvec[i]) << endl;
|
|
|
922 |
}
|
|
|
923 |
for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
|
|
|
924 |
cout << tstvec[i] << " Basename " <<
|
|
|
925 |
path_basename(tstvec[i], ".txt") << endl;
|
|
|
926 |
}
|
|
|
927 |
#endif
|
|
|
928 |
|
|
|
929 |
#if 0
|
|
|
930 |
for (int i = 0; i < nttvec; i++) {
|
|
|
931 |
cout << "tildexp: '" << ttvec[i] << "' -> '" <<
|
|
|
932 |
path_tildexpand(ttvec[i]) << "'" << endl;
|
|
|
933 |
}
|
|
|
934 |
#endif
|
|
|
935 |
|
|
|
936 |
#if 0
|
|
|
937 |
const string canontst[] = {"/dir1/../../..", "/////", "",
|
|
|
938 |
"/dir1/../../.././/////dir2///////",
|
|
|
939 |
"../../",
|
|
|
940 |
"../../../../../../../../../../"
|
|
|
941 |
};
|
|
|
942 |
unsigned int nttvec = sizeof(canontst) / sizeof(string);
|
|
|
943 |
for (unsigned int i = 0; i < nttvec; i++) {
|
|
|
944 |
cout << "canon: '" << canontst[i] << "' -> '" <<
|
|
|
945 |
path_canon(canontst[i]) << "'" << endl;
|
|
|
946 |
}
|
|
|
947 |
#endif
|
|
|
948 |
#if 0
|
|
|
949 |
if (argc != 2) {
|
|
|
950 |
cerr << "Usage: trpathut <dir> <pattern>" << endl;
|
|
|
951 |
exit(1);
|
|
|
952 |
}
|
|
|
953 |
string dir = *argv++;
|
|
|
954 |
argc--;
|
|
|
955 |
string pattern = *argv++;
|
|
|
956 |
argc--;
|
|
|
957 |
vector<string> matched = path_dirglob(dir, pattern);
|
|
|
958 |
for (it = matched.begin(); it != matched.end(); it++) {
|
|
|
959 |
cout << *it << endl;
|
|
|
960 |
}
|
|
|
961 |
#endif
|
|
|
962 |
|
|
|
963 |
#if 0
|
|
|
964 |
if (argc != 1) {
|
|
|
965 |
fprintf(stderr, "Usage: fsocc: trpathut <path>\n");
|
|
|
966 |
exit(1);
|
|
|
967 |
}
|
|
|
968 |
string path = *argv++;
|
|
|
969 |
argc--;
|
|
|
970 |
|
|
|
971 |
int pc;
|
|
|
972 |
long long blocks;
|
|
|
973 |
if (!fsocc(path, &pc, &blocks)) {
|
|
|
974 |
fprintf(stderr, "fsocc failed\n");
|
|
|
975 |
return 1;
|
|
|
976 |
}
|
|
|
977 |
printf("pc %d, megabytes %ld\n", pc, blocks);
|
|
|
978 |
#endif
|
|
|
979 |
|
|
|
980 |
#if 0
|
|
|
981 |
Pidfile pidfile("/tmp/pathutpidfile");
|
|
|
982 |
pid_t pid;
|
|
|
983 |
if ((pid = pidfile.open()) != 0) {
|
|
|
984 |
cerr << "open failed. reason: " << pidfile.getreason() <<
|
|
|
985 |
" return " << pid << endl;
|
|
|
986 |
exit(1);
|
|
|
987 |
}
|
|
|
988 |
pidfile.write_pid();
|
|
|
989 |
sleep(10);
|
|
|
990 |
pidfile.close();
|
|
|
991 |
pidfile.remove();
|
|
|
992 |
#endif
|
|
|
993 |
|
|
|
994 |
#if 0
|
|
|
995 |
if (argc > 1) {
|
|
|
996 |
cerr << "Usage: thumbpath <filepath>" << endl;
|
|
|
997 |
exit(1);
|
|
|
998 |
}
|
|
|
999 |
string input;
|
|
|
1000 |
if (argc == 1) {
|
|
|
1001 |
input = *argv++;
|
|
|
1002 |
if (input.empty()) {
|
|
|
1003 |
cerr << "Usage: thumbpath <filepath>" << endl;
|
|
|
1004 |
exit(1);
|
|
|
1005 |
}
|
|
|
1006 |
path_to_thumb(input);
|
|
|
1007 |
} else {
|
|
|
1008 |
while (getline(cin, input)) {
|
|
|
1009 |
path_to_thumb(input);
|
|
|
1010 |
}
|
|
|
1011 |
}
|
|
|
1012 |
|
|
|
1013 |
|
|
|
1014 |
exit(0);
|
|
|
1015 |
#endif
|
|
|
1016 |
|
|
|
1017 |
#if 0
|
|
|
1018 |
if (argc != 1) {
|
|
|
1019 |
cerr << "Usage: trpathut <filename>" << endl;
|
|
|
1020 |
exit(1);
|
|
|
1021 |
}
|
|
|
1022 |
string fn = *argv++;
|
|
|
1023 |
argc--;
|
|
|
1024 |
string ext = path_suffix(fn);
|
|
|
1025 |
cout << "Suffix: [" << ext << "]" << endl;
|
|
|
1026 |
return 0;
|
|
|
1027 |
#endif
|
|
|
1028 |
|
|
|
1029 |
#if 1
|
|
|
1030 |
if (argc != 1) {
|
|
|
1031 |
cerr << "Usage: trpathut url" << endl;
|
|
|
1032 |
exit(1);
|
|
|
1033 |
}
|
|
|
1034 |
string url = *argv++;
|
|
|
1035 |
argc--;
|
|
|
1036 |
|
|
|
1037 |
cout << "File: [" << fileurltolocalpath(url) << "]\n";
|
|
|
1038 |
return 0;
|
|
|
1039 |
#endif
|
|
|
1040 |
|
|
|
1041 |
|
|
|
1042 |
}
|
|
|
1043 |
|
|
|
1044 |
#endif // TEST_PATHUT
|
|
|
1045 |
|