xref: /dokuwiki/inc/io.php (revision 1c07b9e622d139fa815c955c89569f96342475fb) 
1<?php
2/**
3 * File IO functions
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Andreas Gohr <andi@splitbrain.org>
7 */
8
9if(!defined('DOKU_INC')) die('meh.');
10
11/**
12 * Removes empty directories
13 *
14 * Sends IO_NAMESPACE_DELETED events for 'pages' and 'media' namespaces.
15 * Event data:
16 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
17 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
18 *
19 * @todo use safemode hack
20 * @param string $id      - a pageid, the namespace of that id will be tried to deleted
21 * @param string $basadir - the config name of the type to delete (datadir or mediadir usally)
22 * @returns bool - true if at least one namespace was deleted
23 * @author  Andreas Gohr <andi@splitbrain.org>
24 * @author Ben Coburn <btcoburn@silicodon.net>
25 */
26function io_sweepNS($id,$basedir='datadir'){
27    global $conf;
28    $types = array ('datadir'=>'pages', 'mediadir'=>'media');
29    $ns_type = (isset($types[$basedir])?$types[$basedir]:false);
30
31    $delone = false;
32
33    //scan all namespaces
34    while(($id = getNS($id)) !== false){
35        $dir = $conf[$basedir].'/'.utf8_encodeFN(str_replace(':','/',$id));
36
37        //try to delete dir else return
38        if(@rmdir($dir)) {
39            if ($ns_type!==false) {
40                $data = array($id, $ns_type);
41                $delone = true; // we deleted at least one dir
42                trigger_event('IO_NAMESPACE_DELETED', $data);
43            }
44        } else { return $delone; }
45    }
46    return $delone;
47}
48
49/**
50 * Used to read in a DokuWiki page from file, and send IO_WIKIPAGE_READ events.
51 *
52 * Generates the action event which delegates to io_readFile().
53 * Action plugins are allowed to modify the page content in transit.
54 * The file path should not be changed.
55 *
56 * Event data:
57 * $data[0]    The raw arguments for io_readFile as an array.
58 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
59 * $data[2]    page_name: The wiki page name.
60 * $data[3]    rev: The page revision, false for current wiki pages.
61 *
62 * @author Ben Coburn <btcoburn@silicodon.net>
63 */
64function io_readWikiPage($file, $id, $rev=false) {
65    if (empty($rev)) { $rev = false; }
66    $data = array(array($file, false), getNS($id), noNS($id), $rev);
67    return trigger_event('IO_WIKIPAGE_READ', $data, '_io_readWikiPage_action', false);
68}
69
70/**
71 * Callback adapter for io_readFile().
72 * @author Ben Coburn <btcoburn@silicodon.net>
73 */
74function _io_readWikiPage_action($data) {
75    if (is_array($data) && is_array($data[0]) && count($data[0])===2) {
76        return call_user_func_array('io_readFile', $data[0]);
77    } else {
78        return ''; //callback error
79    }
80}
81
82/**
83 * Returns content of $file as cleaned string.
84 *
85 * Uses gzip if extension is .gz
86 *
87 * If you want to use the returned value in unserialize
88 * be sure to set $clean to false!
89 *
90 * @author  Andreas Gohr <andi@splitbrain.org>
91 */
92function io_readFile($file,$clean=true){
93    $ret = '';
94    if(@file_exists($file)){
95        if(substr($file,-3) == '.gz'){
96            $ret = join('',gzfile($file));
97        }else if(substr($file,-4) == '.bz2'){
98            $ret = bzfile($file);
99        }else{
100            $ret = file_get_contents($file);
101        }
102    }
103    if($clean){
104        return cleanText($ret);
105    }else{
106        return $ret;
107    }
108}
109/**
110 * Returns the content of a .bz2 compressed file as string
111 * @author marcel senf <marcel@rucksackreinigung.de>
112 */
113
114function bzfile($file){
115    $bz = bzopen($file,"r");
116    while (!feof($bz)){
117        //8192 seems to be the maximum buffersize?
118        $str = $str . bzread($bz,8192);
119    }
120    bzclose($bz);
121    return $str;
122}
123
124
125/**
126 * Used to write out a DokuWiki page to file, and send IO_WIKIPAGE_WRITE events.
127 *
128 * This generates an action event and delegates to io_saveFile().
129 * Action plugins are allowed to modify the page content in transit.
130 * The file path should not be changed.
131 * (The append parameter is set to false.)
132 *
133 * Event data:
134 * $data[0]    The raw arguments for io_saveFile as an array.
135 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
136 * $data[2]    page_name: The wiki page name.
137 * $data[3]    rev: The page revision, false for current wiki pages.
138 *
139 * @author Ben Coburn <btcoburn@silicodon.net>
140 */
141function io_writeWikiPage($file, $content, $id, $rev=false) {
142    if (empty($rev)) { $rev = false; }
143    if ($rev===false) { io_createNamespace($id); } // create namespaces as needed
144    $data = array(array($file, $content, false), getNS($id), noNS($id), $rev);
145    return trigger_event('IO_WIKIPAGE_WRITE', $data, '_io_writeWikiPage_action', false);
146}
147
148/**
149 * Callback adapter for io_saveFile().
150 * @author Ben Coburn <btcoburn@silicodon.net>
151 */
152function _io_writeWikiPage_action($data) {
153    if (is_array($data) && is_array($data[0]) && count($data[0])===3) {
154        return call_user_func_array('io_saveFile', $data[0]);
155    } else {
156        return false; //callback error
157    }
158}
159
160/**
161 * Saves $content to $file.
162 *
163 * If the third parameter is set to true the given content
164 * will be appended.
165 *
166 * Uses gzip if extension is .gz
167 * and bz2 if extension is .bz2
168 *
169 * @author  Andreas Gohr <andi@splitbrain.org>
170 * @return bool true on success
171 */
172function io_saveFile($file,$content,$append=false){
173    global $conf;
174    $mode = ($append) ? 'ab' : 'wb';
175
176    $fileexists = @file_exists($file);
177    io_makeFileDir($file);
178    io_lock($file);
179    if(substr($file,-3) == '.gz'){
180        $fh = @gzopen($file,$mode.'9');
181        if(!$fh){
182            msg("Writing $file failed",-1);
183            io_unlock($file);
184            return false;
185        }
186        gzwrite($fh, $content);
187        gzclose($fh);
188    }else if(substr($file,-4) == '.bz2'){
189        $fh = @bzopen($file,$mode{0});
190        if(!$fh){
191            msg("Writing $file failed", -1);
192            io_unlock($file);
193            return false;
194        }
195        bzwrite($fh, $content);
196        bzclose($fh);
197    }else{
198        $fh = @fopen($file,$mode);
199        if(!$fh){
200            msg("Writing $file failed",-1);
201            io_unlock($file);
202            return false;
203        }
204        fwrite($fh, $content);
205        fclose($fh);
206    }
207
208    if(!$fileexists and !empty($conf['fperm'])) chmod($file, $conf['fperm']);
209    io_unlock($file);
210    return true;
211}
212
213/**
214 * Delete exact linematch for $badline from $file.
215 *
216 * Be sure to include the trailing newline in $badline
217 *
218 * Uses gzip if extension is .gz
219 *
220 * 2005-10-14 : added regex option -- Christopher Smith <chris@jalakai.co.uk>
221 *
222 * @author Steven Danz <steven-danz@kc.rr.com>
223 * @return bool true on success
224 */
225function io_deleteFromFile($file,$badline,$regex=false){
226    if (!@file_exists($file)) return true;
227
228    io_lock($file);
229
230    // load into array
231    if(substr($file,-3) == '.gz'){
232        $lines = gzfile($file);
233    }else{
234        $lines = file($file);
235    }
236
237    // remove all matching lines
238    if ($regex) {
239        $lines = preg_grep($badline,$lines,PREG_GREP_INVERT);
240    } else {
241        $pos = array_search($badline,$lines); //return null or false if not found
242        while(is_int($pos)){
243            unset($lines[$pos]);
244            $pos = array_search($badline,$lines);
245        }
246    }
247
248    if(count($lines)){
249        $content = join('',$lines);
250        if(substr($file,-3) == '.gz'){
251            $fh = @gzopen($file,'wb9');
252            if(!$fh){
253                msg("Removing content from $file failed",-1);
254                io_unlock($file);
255                return false;
256            }
257            gzwrite($fh, $content);
258            gzclose($fh);
259        }else{
260            $fh = @fopen($file,'wb');
261            if(!$fh){
262                msg("Removing content from $file failed",-1);
263                io_unlock($file);
264                return false;
265            }
266            fwrite($fh, $content);
267            fclose($fh);
268        }
269    }else{
270        @unlink($file);
271    }
272
273    io_unlock($file);
274    return true;
275}
276
277/**
278 * Tries to lock a file
279 *
280 * Locking is only done for io_savefile and uses directories
281 * inside $conf['lockdir']
282 *
283 * It waits maximal 3 seconds for the lock, after this time
284 * the lock is assumed to be stale and the function goes on
285 *
286 * @author Andreas Gohr <andi@splitbrain.org>
287 */
288function io_lock($file){
289    global $conf;
290    // no locking if safemode hack
291    if($conf['safemodehack']) return;
292
293    $lockDir = $conf['lockdir'].'/'.md5($file);
294    @ignore_user_abort(1);
295
296    $timeStart = time();
297    do {
298        //waited longer than 3 seconds? -> stale lock
299        if ((time() - $timeStart) > 3) break;
300        $locked = @mkdir($lockDir, $conf['dmode']);
301        if($locked){
302            if(!empty($conf['dperm'])) chmod($lockDir, $conf['dperm']);
303            break;
304        }
305        usleep(50);
306    } while ($locked === false);
307}
308
309/**
310 * Unlocks a file
311 *
312 * @author Andreas Gohr <andi@splitbrain.org>
313 */
314function io_unlock($file){
315    global $conf;
316    // no locking if safemode hack
317    if($conf['safemodehack']) return;
318
319    $lockDir = $conf['lockdir'].'/'.md5($file);
320    @rmdir($lockDir);
321    @ignore_user_abort(0);
322}
323
324/**
325 * Create missing namespace directories and send the IO_NAMESPACE_CREATED events
326 * in the order of directory creation. (Parent directories first.)
327 *
328 * Event data:
329 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
330 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
331 *
332 * @author Ben Coburn <btcoburn@silicodon.net>
333 */
334function io_createNamespace($id, $ns_type='pages') {
335    // verify ns_type
336    $types = array('pages'=>'wikiFN', 'media'=>'mediaFN');
337    if (!isset($types[$ns_type])) {
338        trigger_error('Bad $ns_type parameter for io_createNamespace().');
339        return;
340    }
341    // make event list
342    $missing = array();
343    $ns_stack = explode(':', $id);
344    $ns = $id;
345    $tmp = dirname( $file = call_user_func($types[$ns_type], $ns) );
346    while (!@is_dir($tmp) && !(@file_exists($tmp) && !is_dir($tmp))) {
347        array_pop($ns_stack);
348        $ns = implode(':', $ns_stack);
349        if (strlen($ns)==0) { break; }
350        $missing[] = $ns;
351        $tmp = dirname(call_user_func($types[$ns_type], $ns));
352    }
353    // make directories
354    io_makeFileDir($file);
355    // send the events
356    $missing = array_reverse($missing); // inside out
357    foreach ($missing as $ns) {
358        $data = array($ns, $ns_type);
359        trigger_event('IO_NAMESPACE_CREATED', $data);
360    }
361}
362
363/**
364 * Create the directory needed for the given file
365 *
366 * @author  Andreas Gohr <andi@splitbrain.org>
367 */
368function io_makeFileDir($file){
369    global $conf;
370
371    $dir = dirname($file);
372    if(!@is_dir($dir)){
373        io_mkdir_p($dir) || msg("Creating directory $dir failed",-1);
374    }
375}
376
377/**
378 * Creates a directory hierachy.
379 *
380 * @link    http://www.php.net/manual/en/function.mkdir.php
381 * @author  <saint@corenova.com>
382 * @author  Andreas Gohr <andi@splitbrain.org>
383 */
384function io_mkdir_p($target){
385    global $conf;
386    if (@is_dir($target)||empty($target)) return 1; // best case check first
387    if (@file_exists($target) && !is_dir($target)) return 0;
388    //recursion
389    if (io_mkdir_p(substr($target,0,strrpos($target,'/')))){
390        if($conf['safemodehack']){
391            $dir = preg_replace('/^'.preg_quote(fullpath($conf['ftp']['root']),'/').'/','', $target);
392            return io_mkdir_ftp($dir);
393        }else{
394            $ret = @mkdir($target,$conf['dmode']); // crawl back up & create dir tree
395            if($ret && $conf['dperm']) chmod($target, $conf['dperm']);
396            return $ret;
397        }
398    }
399    return 0;
400}
401
402/**
403 * Creates a directory using FTP
404 *
405 * This is used when the safemode workaround is enabled
406 *
407 * @author <andi@splitbrain.org>
408 */
409function io_mkdir_ftp($dir){
410    global $conf;
411
412    if(!function_exists('ftp_connect')){
413        msg("FTP support not found - safemode workaround not usable",-1);
414        return false;
415    }
416
417    $conn = @ftp_connect($conf['ftp']['host'],$conf['ftp']['port'],10);
418    if(!$conn){
419        msg("FTP connection failed",-1);
420        return false;
421    }
422
423    if(!@ftp_login($conn, $conf['ftp']['user'], conf_decodeString($conf['ftp']['pass']))){
424        msg("FTP login failed",-1);
425        return false;
426    }
427
428    //create directory
429    $ok = @ftp_mkdir($conn, $dir);
430    //set permissions
431    @ftp_site($conn,sprintf("CHMOD %04o %s",$conf['dmode'],$dir));
432
433    @ftp_close($conn);
434    return $ok;
435}
436
437/**
438 * Creates a unique temporary directory and returns
439 * its path.
440 *
441 * @author Michael Klier <chi@chimeric.de>
442 */
443function io_mktmpdir() {
444    global $conf;
445
446    $base = $conf['tmpdir'];
447    $dir  = md5(uniqid(mt_rand(), true));
448    $tmpdir = $base.'/'.$dir;
449
450    if(io_mkdir_p($tmpdir)) {
451        return($tmpdir);
452    } else {
453        return false;
454    }
455}
456
457/**
458 * downloads a file from the net and saves it
459 *
460 * if $useAttachment is false,
461 * - $file is the full filename to save the file, incl. path
462 * - if successful will return true, false otherwise
463 *
464 * if $useAttachment is true,
465 * - $file is the directory where the file should be saved
466 * - if successful will return the name used for the saved file, false otherwise
467 *
468 * @author Andreas Gohr <andi@splitbrain.org>
469 * @author Chris Smith <chris@jalakai.co.uk>
470 */
471function io_download($url,$file,$useAttachment=false,$defaultName='',$maxSize=2097152){
472    global $conf;
473    $http = new DokuHTTPClient();
474    $http->max_bodysize = $maxSize;
475    $http->timeout = 25; //max. 25 sec
476
477    $data = $http->get($url);
478    if(!$data) return false;
479
480    if ($useAttachment) {
481        $name = '';
482        if (isset($http->resp_headers['content-disposition'])) {
483            $content_disposition = $http->resp_headers['content-disposition'];
484            $match=array();
485            if (is_string($content_disposition) &&
486                    preg_match('/attachment;\s*filename\s*=\s*"([^"]*)"/i', $content_disposition, $match)) {
487
488                $name = basename($match[1]);
489        }
490
491        }
492
493        if (!$name) {
494            if (!$defaultName) return false;
495            $name = $defaultName;
496        }
497
498        $file = $file.$name;
499    }
500
501    $fileexists = @file_exists($file);
502    $fp = @fopen($file,"w");
503    if(!$fp) return false;
504    fwrite($fp,$data);
505    fclose($fp);
506    if(!$fileexists and $conf['fperm']) chmod($file, $conf['fperm']);
507    if ($useAttachment) return $name;
508    return true;
509}
510
511/**
512 * Windows compatible rename
513 *
514 * rename() can not overwrite existing files on Windows
515 * this function will use copy/unlink instead
516 */
517function io_rename($from,$to){
518    global $conf;
519    if(!@rename($from,$to)){
520        if(@copy($from,$to)){
521            if($conf['fperm']) chmod($to, $conf['fperm']);
522            @unlink($from);
523            return true;
524        }
525        return false;
526    }
527    return true;
528}
529
530
531/**
532 * Runs an external command and returns it's output as string
533 *
534 * @author Harry Brueckner <harry_b@eml.cc>
535 * @author Andreas Gohr <andi@splitbrain.org>
536 */
537function io_runcmd($cmd, $input, &$output){
538    $descspec = array(
539            0=>array("pipe","r"),
540            1=>array("pipe","w"),
541            2=>array("pipe","w"));
542    $ph = proc_open($cmd, $descspec, $pipes);
543    if(!$ph) return -1;
544    fclose($pipes[2]); // ignore stderr
545    fwrite($pipes[0], $input);
546    fclose($pipes[0]);
547    $output = stream_get_contents($pipes[1]);
548    fclose($pipes[1]);
549    return proc_close($ph);
550}
551
552/**
553 * Search a file for matching lines
554 *
555 * This is probably not faster than file()+preg_grep() but less
556 * memory intensive because not the whole file needs to be loaded
557 * at once.
558 *
559 * @author Andreas Gohr <andi@splitbrain.org>
560 * @param  string $file    The file to search
561 * @param  string $pattern PCRE pattern
562 * @param  int    $max     How many lines to return (0 for all)
563 * @param  bool   $baxkref When true returns array with backreferences instead of lines
564 * @return matching lines or backref, false on error
565 */
566function io_grep($file,$pattern,$max=0,$backref=false){
567    $fh = @fopen($file,'r');
568    if(!$fh) return false;
569    $matches = array();
570
571    $cnt  = 0;
572    $line = '';
573    while (!feof($fh)) {
574        $line .= fgets($fh, 4096);  // read full line
575        if(substr($line,-1) != "\n") continue;
576
577        // check if line matches
578        if(preg_match($pattern,$line,$match)){
579            if($backref){
580                $matches[] = $match;
581            }else{
582                $matches[] = $line;
583            }
584            $cnt++;
585        }
586        if($max && $max == $cnt) break;
587        $line = '';
588    }
589    fclose($fh);
590    return $matches;
591}
592
593