xref: /dokuwiki/inc/io.php (revision e26a58378c0c7f5d468dde469a7f782b45354426) 
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 $basedir - the config name of the type to delete (datadir or mediadir usally)
22 * @return bool - true if at least one namespace was deleted
23 * @author  Andreas Gohr <andi@splitbrain.org>
24 * @author Ben Coburn <btcoburn@silicodon.net>
25 */
26function io_sweepNS($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, true), 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    $str = '';
117    while (!feof($bz)){
118        //8192 seems to be the maximum buffersize?
119        $str = $str . bzread($bz,8192);
120    }
121    bzclose($bz);
122    return $str;
123}
124
125
126/**
127 * Used to write out a DokuWiki page to file, and send IO_WIKIPAGE_WRITE events.
128 *
129 * This generates an action event and delegates to io_saveFile().
130 * Action plugins are allowed to modify the page content in transit.
131 * The file path should not be changed.
132 * (The append parameter is set to false.)
133 *
134 * Event data:
135 * $data[0]    The raw arguments for io_saveFile as an array.
136 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
137 * $data[2]    page_name: The wiki page name.
138 * $data[3]    rev: The page revision, false for current wiki pages.
139 *
140 * @author Ben Coburn <btcoburn@silicodon.net>
141 */
142function io_writeWikiPage($file, $content, $id, $rev=false) {
143    if (empty($rev)) { $rev = false; }
144    if ($rev===false) { io_createNamespace($id); } // create namespaces as needed
145    $data = array(array($file, $content, false), getNS($id), noNS($id), $rev);
146    return trigger_event('IO_WIKIPAGE_WRITE', $data, '_io_writeWikiPage_action', false);
147}
148
149/**
150 * Callback adapter for io_saveFile().
151 * @author Ben Coburn <btcoburn@silicodon.net>
152 */
153function _io_writeWikiPage_action($data) {
154    if (is_array($data) && is_array($data[0]) && count($data[0])===3) {
155        return call_user_func_array('io_saveFile', $data[0]);
156    } else {
157        return false; //callback error
158    }
159}
160
161/**
162 * Saves $content to $file.
163 *
164 * If the third parameter is set to true the given content
165 * will be appended.
166 *
167 * Uses gzip if extension is .gz
168 * and bz2 if extension is .bz2
169 *
170 * @author  Andreas Gohr <andi@splitbrain.org>
171 * @return bool true on success
172 */
173function io_saveFile($file,$content,$append=false){
174    global $conf;
175    $mode = ($append) ? 'ab' : 'wb';
176
177    $fileexists = @file_exists($file);
178    io_makeFileDir($file);
179    io_lock($file);
180    if(substr($file,-3) == '.gz'){
181        $fh = @gzopen($file,$mode.'9');
182        if(!$fh){
183            msg("Writing $file failed",-1);
184            io_unlock($file);
185            return false;
186        }
187        gzwrite($fh, $content);
188        gzclose($fh);
189    }else if(substr($file,-4) == '.bz2'){
190        $fh = @bzopen($file,$mode{0});
191        if(!$fh){
192            msg("Writing $file failed", -1);
193            io_unlock($file);
194            return false;
195        }
196        bzwrite($fh, $content);
197        bzclose($fh);
198    }else{
199        $fh = @fopen($file,$mode);
200        if(!$fh){
201            msg("Writing $file failed",-1);
202            io_unlock($file);
203            return false;
204        }
205        fwrite($fh, $content);
206        fclose($fh);
207    }
208
209    if(!$fileexists and !empty($conf['fperm'])) chmod($file, $conf['fperm']);
210    io_unlock($file);
211    return true;
212}
213
214/**
215 * Delete exact linematch for $badline from $file.
216 *
217 * Be sure to include the trailing newline in $badline
218 *
219 * Uses gzip if extension is .gz
220 *
221 * 2005-10-14 : added regex option -- Christopher Smith <chris@jalakai.co.uk>
222 *
223 * @author Steven Danz <steven-danz@kc.rr.com>
224 * @return bool true on success
225 */
226function io_deleteFromFile($file,$badline,$regex=false){
227    if (!@file_exists($file)) return true;
228
229    io_lock($file);
230
231    // load into array
232    if(substr($file,-3) == '.gz'){
233        $lines = gzfile($file);
234    }else{
235        $lines = file($file);
236    }
237
238    // remove all matching lines
239    if ($regex) {
240        $lines = preg_grep($badline,$lines,PREG_GREP_INVERT);
241    } else {
242        $pos = array_search($badline,$lines); //return null or false if not found
243        while(is_int($pos)){
244            unset($lines[$pos]);
245            $pos = array_search($badline,$lines);
246        }
247    }
248
249    if(count($lines)){
250        $content = join('',$lines);
251        if(substr($file,-3) == '.gz'){
252            $fh = @gzopen($file,'wb9');
253            if(!$fh){
254                msg("Removing content from $file failed",-1);
255                io_unlock($file);
256                return false;
257            }
258            gzwrite($fh, $content);
259            gzclose($fh);
260        }else{
261            $fh = @fopen($file,'wb');
262            if(!$fh){
263                msg("Removing content from $file failed",-1);
264                io_unlock($file);
265                return false;
266            }
267            fwrite($fh, $content);
268            fclose($fh);
269        }
270    }else{
271        @unlink($file);
272    }
273
274    io_unlock($file);
275    return true;
276}
277
278/**
279 * Tries to lock a file
280 *
281 * Locking is only done for io_savefile and uses directories
282 * inside $conf['lockdir']
283 *
284 * It waits maximal 3 seconds for the lock, after this time
285 * the lock is assumed to be stale and the function goes on
286 *
287 * @author Andreas Gohr <andi@splitbrain.org>
288 */
289function io_lock($file){
290    global $conf;
291    // no locking if safemode hack
292    if($conf['safemodehack']) return;
293
294    $lockDir = $conf['lockdir'].'/'.md5($file);
295    @ignore_user_abort(1);
296
297    $timeStart = time();
298    do {
299        //waited longer than 3 seconds? -> stale lock
300        if ((time() - $timeStart) > 3) break;
301        $locked = @mkdir($lockDir, $conf['dmode']);
302        if($locked){
303            if(!empty($conf['dperm'])) chmod($lockDir, $conf['dperm']);
304            break;
305        }
306        usleep(50);
307    } while ($locked === false);
308}
309
310/**
311 * Unlocks a file
312 *
313 * @author Andreas Gohr <andi@splitbrain.org>
314 */
315function io_unlock($file){
316    global $conf;
317    // no locking if safemode hack
318    if($conf['safemodehack']) return;
319
320    $lockDir = $conf['lockdir'].'/'.md5($file);
321    @rmdir($lockDir);
322    @ignore_user_abort(0);
323}
324
325/**
326 * Create missing namespace directories and send the IO_NAMESPACE_CREATED events
327 * in the order of directory creation. (Parent directories first.)
328 *
329 * Event data:
330 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
331 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
332 *
333 * @author Ben Coburn <btcoburn@silicodon.net>
334 */
335function io_createNamespace($id, $ns_type='pages') {
336    // verify ns_type
337    $types = array('pages'=>'wikiFN', 'media'=>'mediaFN');
338    if (!isset($types[$ns_type])) {
339        trigger_error('Bad $ns_type parameter for io_createNamespace().');
340        return;
341    }
342    // make event list
343    $missing = array();
344    $ns_stack = explode(':', $id);
345    $ns = $id;
346    $tmp = dirname( $file = call_user_func($types[$ns_type], $ns) );
347    while (!@is_dir($tmp) && !(@file_exists($tmp) && !is_dir($tmp))) {
348        array_pop($ns_stack);
349        $ns = implode(':', $ns_stack);
350        if (strlen($ns)==0) { break; }
351        $missing[] = $ns;
352        $tmp = dirname(call_user_func($types[$ns_type], $ns));
353    }
354    // make directories
355    io_makeFileDir($file);
356    // send the events
357    $missing = array_reverse($missing); // inside out
358    foreach ($missing as $ns) {
359        $data = array($ns, $ns_type);
360        trigger_event('IO_NAMESPACE_CREATED', $data);
361    }
362}
363
364/**
365 * Create the directory needed for the given file
366 *
367 * @author  Andreas Gohr <andi@splitbrain.org>
368 */
369function io_makeFileDir($file){
370    $dir = dirname($file);
371    if(!@is_dir($dir)){
372        io_mkdir_p($dir) || msg("Creating directory $dir failed",-1);
373    }
374}
375
376/**
377 * Creates a directory hierachy.
378 *
379 * @link    http://www.php.net/manual/en/function.mkdir.php
380 * @author  <saint@corenova.com>
381 * @author  Andreas Gohr <andi@splitbrain.org>
382 */
383function io_mkdir_p($target){
384    global $conf;
385    if (@is_dir($target)||empty($target)) return 1; // best case check first
386    if (@file_exists($target) && !is_dir($target)) return 0;
387    //recursion
388    if (io_mkdir_p(substr($target,0,strrpos($target,'/')))){
389        if($conf['safemodehack']){
390            $dir = preg_replace('/^'.preg_quote(fullpath($conf['ftp']['root']),'/').'/','', $target);
391            return io_mkdir_ftp($dir);
392        }else{
393            $ret = @mkdir($target,$conf['dmode']); // crawl back up & create dir tree
394            if($ret && !empty($conf['dperm'])) chmod($target, $conf['dperm']);
395            return $ret;
396        }
397    }
398    return 0;
399}
400
401/**
402 * Recursively delete a directory
403 *
404 * @author Andreas Gohr <andi@splitbrain.org>
405 * @param string $path
406 * @param bool   $removefiles defaults to false which will delete empty directories only
407 * @return bool
408 */
409function io_rmdir($path, $removefiles = false) {
410    if(!is_string($path) || $path == "") return false;
411    if(!file_exists($path)) return true; // it's already gone or was never there, count as success
412
413    if(is_dir($path) && !is_link($path)) {
414        $dirs  = array();
415        $files = array();
416
417        if(!$dh = @opendir($path)) return false;
418        while(false !== ($f = readdir($dh))) {
419            if($f == '..' || $f == '.') continue;
420
421            // collect dirs and files first
422            if(is_dir("$path/$f") && !is_link("$path/$f")) {
423                $dirs[] = "$path/$f";
424            } else if($removefiles) {
425                $files[] = "$path/$f";
426            } else {
427                return false; // abort when non empty
428            }
429
430        }
431        closedir($dh);
432
433        // now traverse into  directories first
434        foreach($dirs as $dir) {
435            if(!io_rmdir($dir, $removefiles)) return false; // abort on any error
436        }
437
438        // now delete files
439        foreach($files as $file) {
440            if(!@unlink($file)) return false; //abort on any error
441        }
442
443        // remove self
444        return @rmdir($path);
445    } else if($removefiles) {
446        return @unlink($path);
447    }
448    return false;
449}
450
451/**
452 * Creates a directory using FTP
453 *
454 * This is used when the safemode workaround is enabled
455 *
456 * @author <andi@splitbrain.org>
457 */
458function io_mkdir_ftp($dir){
459    global $conf;
460
461    if(!function_exists('ftp_connect')){
462        msg("FTP support not found - safemode workaround not usable",-1);
463        return false;
464    }
465
466    $conn = @ftp_connect($conf['ftp']['host'],$conf['ftp']['port'],10);
467    if(!$conn){
468        msg("FTP connection failed",-1);
469        return false;
470    }
471
472    if(!@ftp_login($conn, $conf['ftp']['user'], conf_decodeString($conf['ftp']['pass']))){
473        msg("FTP login failed",-1);
474        return false;
475    }
476
477    //create directory
478    $ok = @ftp_mkdir($conn, $dir);
479    //set permissions
480    @ftp_site($conn,sprintf("CHMOD %04o %s",$conf['dmode'],$dir));
481
482    @ftp_close($conn);
483    return $ok;
484}
485
486/**
487 * Creates a unique temporary directory and returns
488 * its path.
489 *
490 * @author Michael Klier <chi@chimeric.de>
491 */
492function io_mktmpdir() {
493    global $conf;
494
495    $base = $conf['tmpdir'];
496    $dir  = md5(uniqid(mt_rand(), true));
497    $tmpdir = $base.'/'.$dir;
498
499    if(io_mkdir_p($tmpdir)) {
500        return($tmpdir);
501    } else {
502        return false;
503    }
504}
505
506/**
507 * downloads a file from the net and saves it
508 *
509 * if $useAttachment is false,
510 * - $file is the full filename to save the file, incl. path
511 * - if successful will return true, false otherwise
512 *
513 * if $useAttachment is true,
514 * - $file is the directory where the file should be saved
515 * - if successful will return the name used for the saved file, false otherwise
516 *
517 * @author Andreas Gohr <andi@splitbrain.org>
518 * @author Chris Smith <chris@jalakai.co.uk>
519 */
520function io_download($url,$file,$useAttachment=false,$defaultName='',$maxSize=2097152){
521    global $conf;
522    $http = new DokuHTTPClient();
523    $http->max_bodysize = $maxSize;
524    $http->timeout = 25; //max. 25 sec
525    $http->keep_alive = false; // we do single ops here, no need for keep-alive
526
527    $data = $http->get($url);
528    if(!$data) return false;
529
530    $name = '';
531    if ($useAttachment) {
532        if (isset($http->resp_headers['content-disposition'])) {
533            $content_disposition = $http->resp_headers['content-disposition'];
534            $match=array();
535            if (is_string($content_disposition) &&
536                    preg_match('/attachment;\s*filename\s*=\s*"([^"]*)"/i', $content_disposition, $match)) {
537
538                $name = utf8_basename($match[1]);
539            }
540
541        }
542
543        if (!$name) {
544            if (!$defaultName) return false;
545            $name = $defaultName;
546        }
547
548        $file = $file.$name;
549    }
550
551    $fileexists = @file_exists($file);
552    $fp = @fopen($file,"w");
553    if(!$fp) return false;
554    fwrite($fp,$data);
555    fclose($fp);
556    if(!$fileexists and $conf['fperm']) chmod($file, $conf['fperm']);
557    if ($useAttachment) return $name;
558    return true;
559}
560
561/**
562 * Windows compatible rename
563 *
564 * rename() can not overwrite existing files on Windows
565 * this function will use copy/unlink instead
566 */
567function io_rename($from,$to){
568    global $conf;
569    if(!@rename($from,$to)){
570        if(@copy($from,$to)){
571            if($conf['fperm']) chmod($to, $conf['fperm']);
572            @unlink($from);
573            return true;
574        }
575        return false;
576    }
577    return true;
578}
579
580/**
581 * Runs an external command with input and output pipes.
582 * Returns the exit code from the process.
583 *
584 * @author Tom N Harris <tnharris@whoopdedo.org>
585 */
586function io_exec($cmd, $input, &$output){
587    $descspec = array(
588            0=>array("pipe","r"),
589            1=>array("pipe","w"),
590            2=>array("pipe","w"));
591    $ph = proc_open($cmd, $descspec, $pipes);
592    if(!$ph) return -1;
593    fclose($pipes[2]); // ignore stderr
594    fwrite($pipes[0], $input);
595    fclose($pipes[0]);
596    $output = stream_get_contents($pipes[1]);
597    fclose($pipes[1]);
598    return proc_close($ph);
599}
600
601/**
602 * Search a file for matching lines
603 *
604 * This is probably not faster than file()+preg_grep() but less
605 * memory intensive because not the whole file needs to be loaded
606 * at once.
607 *
608 * @author Andreas Gohr <andi@splitbrain.org>
609 * @param  string $file    The file to search
610 * @param  string $pattern PCRE pattern
611 * @param  int    $max     How many lines to return (0 for all)
612 * @param  bool   $backref When true returns array with backreferences instead of lines
613 * @return array matching lines or backref, false on error
614 */
615function io_grep($file,$pattern,$max=0,$backref=false){
616    $fh = @fopen($file,'r');
617    if(!$fh) return false;
618    $matches = array();
619
620    $cnt  = 0;
621    $line = '';
622    while (!feof($fh)) {
623        $line .= fgets($fh, 4096);  // read full line
624        if(substr($line,-1) != "\n") continue;
625
626        // check if line matches
627        if(preg_match($pattern,$line,$match)){
628            if($backref){
629                $matches[] = $match;
630            }else{
631                $matches[] = $line;
632            }
633            $cnt++;
634        }
635        if($max && $max == $cnt) break;
636        $line = '';
637    }
638    fclose($fh);
639    return $matches;
640}
641
642