xref: /dokuwiki/inc/io.php (revision 4a690352a4dc4dff668c4189c41d982f838bc8b9) 
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 $basedir - the config name of the type to delete (datadir or mediadir usally)
22 * @return bool - true if at least one namespace was deleted
23 *
24 * @author  Andreas Gohr <andi@splitbrain.org>
25 * @author Ben Coburn <btcoburn@silicodon.net>
26 */
27function io_sweepNS($id,$basedir='datadir'){
28    global $conf;
29    $types = array ('datadir'=>'pages', 'mediadir'=>'media');
30    $ns_type = (isset($types[$basedir])?$types[$basedir]:false);
31
32    $delone = false;
33
34    //scan all namespaces
35    while(($id = getNS($id)) !== false){
36        $dir = $conf[$basedir].'/'.utf8_encodeFN(str_replace(':','/',$id));
37
38        //try to delete dir else return
39        if(@rmdir($dir)) {
40            if ($ns_type!==false) {
41                $data = array($id, $ns_type);
42                $delone = true; // we deleted at least one dir
43                trigger_event('IO_NAMESPACE_DELETED', $data);
44            }
45        } else { return $delone; }
46    }
47    return $delone;
48}
49
50/**
51 * Used to read in a DokuWiki page from file, and send IO_WIKIPAGE_READ events.
52 *
53 * Generates the action event which delegates to io_readFile().
54 * Action plugins are allowed to modify the page content in transit.
55 * The file path should not be changed.
56 *
57 * Event data:
58 * $data[0]    The raw arguments for io_readFile as an array.
59 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
60 * $data[2]    page_name: The wiki page name.
61 * $data[3]    rev: The page revision, false for current wiki pages.
62 *
63 * @author Ben Coburn <btcoburn@silicodon.net>
64 *
65 * @param string   $file filename
66 * @param string   $id page id
67 * @param bool|int $rev revision timestamp
68 * @return string
69 */
70function io_readWikiPage($file, $id, $rev=false) {
71    if (empty($rev)) { $rev = false; }
72    $data = array(array($file, true), getNS($id), noNS($id), $rev);
73    return trigger_event('IO_WIKIPAGE_READ', $data, '_io_readWikiPage_action', false);
74}
75
76/**
77 * Callback adapter for io_readFile().
78 *
79 * @author Ben Coburn <btcoburn@silicodon.net>
80 *
81 * @param array $data event data
82 * @return string
83 */
84function _io_readWikiPage_action($data) {
85    if (is_array($data) && is_array($data[0]) && count($data[0])===2) {
86        return call_user_func_array('io_readFile', $data[0]);
87    } else {
88        return ''; //callback error
89    }
90}
91
92/**
93 * Returns content of $file as cleaned string.
94 *
95 * Uses gzip if extension is .gz
96 *
97 * If you want to use the returned value in unserialize
98 * be sure to set $clean to false!
99 *
100 * @author  Andreas Gohr <andi@splitbrain.org>
101 *
102 * @param string $file  filename
103 * @param bool   $clean
104 * @return string|bool the file contents or false on error
105 */
106function io_readFile($file,$clean=true){
107    $ret = '';
108    if(file_exists($file)){
109        if(substr($file,-3) == '.gz'){
110            $ret = join('',gzfile($file));
111        }else if(substr($file,-4) == '.bz2'){
112            $ret = bzfile($file);
113        }else{
114            $ret = file_get_contents($file);
115        }
116    }
117    if($ret !== false && $clean){
118        return cleanText($ret);
119    }else{
120        return $ret;
121    }
122}
123/**
124 * Returns the content of a .bz2 compressed file as string
125 *
126 * @author marcel senf <marcel@rucksackreinigung.de>
127 * @author  Andreas Gohr <andi@splitbrain.org>
128 *
129 * @param string $file filename
130 * @param bool   $array return array of lines
131 * @return string|array|bool content or false on error
132 */
133function bzfile($file, $array=false) {
134    $bz = bzopen($file,"r");
135    if($bz === false) return false;
136
137    if($array) $lines = array();
138    $str = '';
139    while (!feof($bz)) {
140        //8192 seems to be the maximum buffersize?
141        $buffer = bzread($bz,8192);
142        if(($buffer === false) || (bzerrno($bz) !== 0)) {
143            return false;
144        }
145        $str = $str . $buffer;
146        if($array) {
147            $pos = strpos($str, "\n");
148            while($pos !== false) {
149                $lines[] = substr($str, 0, $pos+1);
150                $str = substr($str, $pos+1);
151                $pos = strpos($str, "\n");
152            }
153        }
154    }
155    bzclose($bz);
156    if($array) {
157        if($str !== '') $lines[] = $str;
158        return $lines;
159    }
160    return $str;
161}
162
163/**
164 * Used to write out a DokuWiki page to file, and send IO_WIKIPAGE_WRITE events.
165 *
166 * This generates an action event and delegates to io_saveFile().
167 * Action plugins are allowed to modify the page content in transit.
168 * The file path should not be changed.
169 * (The append parameter is set to false.)
170 *
171 * Event data:
172 * $data[0]    The raw arguments for io_saveFile as an array.
173 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
174 * $data[2]    page_name: The wiki page name.
175 * $data[3]    rev: The page revision, false for current wiki pages.
176 *
177 * @author Ben Coburn <btcoburn@silicodon.net>
178 *
179 * @param string $file      filename
180 * @param string $content
181 * @param string $id        page id
182 * @param int|bool $rev timestamp of revision
183 * @return bool
184 */
185function io_writeWikiPage($file, $content, $id, $rev=false) {
186    if (empty($rev)) { $rev = false; }
187    if ($rev===false) { io_createNamespace($id); } // create namespaces as needed
188    $data = array(array($file, $content, false), getNS($id), noNS($id), $rev);
189    return trigger_event('IO_WIKIPAGE_WRITE', $data, '_io_writeWikiPage_action', false);
190}
191
192/**
193 * Callback adapter for io_saveFile().
194 * @author Ben Coburn <btcoburn@silicodon.net>
195 *
196 * @param array $data event data
197 * @return bool
198 */
199function _io_writeWikiPage_action($data) {
200    if (is_array($data) && is_array($data[0]) && count($data[0])===3) {
201        return call_user_func_array('io_saveFile', $data[0]);
202    } else {
203        return false; //callback error
204    }
205}
206
207/**
208 * Internal function to save contents to a file.
209 *
210 * @author  Andreas Gohr <andi@splitbrain.org>
211 *
212 * @param string $file filename path to file
213 * @param string $content
214 * @param bool   $append
215 * @return bool true on success, otherwise false
216 */
217function _io_saveFile($file, $content, $append) {
218    global $conf;
219    $mode = ($append) ? 'ab' : 'wb';
220    $fileexists = file_exists($file);
221
222    if(substr($file,-3) == '.gz'){
223        $fh = @gzopen($file,$mode.'9');
224        if(!$fh) return false;
225        gzwrite($fh, $content);
226        gzclose($fh);
227    }else if(substr($file,-4) == '.bz2'){
228        if($append) {
229            $bzcontent = bzfile($file);
230            if($bzcontent === false) return false;
231            $content = $bzcontent.$content;
232        }
233        $fh = @bzopen($file,'w');
234        if(!$fh) return false;
235        bzwrite($fh, $content);
236        bzclose($fh);
237    }else{
238        $fh = @fopen($file,$mode);
239        if(!$fh) return false;
240        fwrite($fh, $content);
241        fclose($fh);
242    }
243
244    if(!$fileexists and !empty($conf['fperm'])) chmod($file, $conf['fperm']);
245    return true;
246}
247
248/**
249 * Saves $content to $file.
250 *
251 * If the third parameter is set to true the given content
252 * will be appended.
253 *
254 * Uses gzip if extension is .gz
255 * and bz2 if extension is .bz2
256 *
257 * @author  Andreas Gohr <andi@splitbrain.org>
258 *
259 * @param string $file filename path to file
260 * @param string $content
261 * @param bool   $append
262 * @return bool true on success, otherwise false
263 */
264function io_saveFile($file, $content, $append=false) {
265    io_makeFileDir($file);
266    io_lock($file);
267    if(!_io_saveFile($file, $content, $append)) {
268        msg("Writing $file failed",-1);
269        io_unlock($file);
270        return false;
271    }
272    io_unlock($file);
273    return true;
274}
275
276/**
277 * Replace one or more occurrences of a line in a file.
278 *
279 * The default, when $maxlines is 0 is to delete all matching lines then append a single line.
280 * A regex that matches any part of the line will remove the entire line in this mode.
281 * Captures in $newline are not available.
282 *
283 * Otherwise each line is matched and replaced individually, up to the first $maxlines lines
284 * or all lines if $maxlines is -1. If $regex is true then captures can be used in $newline.
285 *
286 * Be sure to include the trailing newline in $oldline when replacing entire lines.
287 *
288 * Uses gzip if extension is .gz
289 * and bz2 if extension is .bz2
290 *
291 * @author Steven Danz <steven-danz@kc.rr.com>
292 * @author Christopher Smith <chris@jalakai.co.uk>
293 * @author Patrick Brown <ptbrown@whoopdedo.org>
294 *
295 * @param string $file     filename
296 * @param string $oldline  exact linematch to remove
297 * @param string $newline  new line to insert
298 * @param bool   $regex    use regexp?
299 * @param int    $maxlines number of occurrences of the line to replace
300 * @return bool true on success
301 */
302function io_replaceInFile($file, $oldline, $newline, $regex=false, $maxlines=0) {
303    if ((string)$oldline === '') {
304        trigger_error('$oldline parameter cannot be empty in io_replaceInFile()', E_USER_WARNING);
305        return false;
306    }
307
308    if (!file_exists($file)) return true;
309
310    io_lock($file);
311
312    // load into array
313    if(substr($file,-3) == '.gz'){
314        $lines = gzfile($file);
315    }else if(substr($file,-4) == '.bz2'){
316        $lines = bzfile($file, true);
317    }else{
318        $lines = file($file);
319    }
320
321    // make non-regexes into regexes
322    $pattern = $regex ? $oldline : '/^'.preg_quote($oldline,'/').'$/';
323    $replace = $regex ? $newline : addcslashes($newline, '\$');
324
325    // remove matching lines
326    if ($maxlines > 0) {
327        $count = 0;
328        $matched = 0;
329        while (($count < $maxlines) && (list($i,$line) = each($lines))) {
330            // $matched will be set to 0|1 depending on whether pattern is matched and line replaced
331            $lines[$i] = preg_replace($pattern, $replace, $line, -1, $matched);
332            if ($matched) $count++;
333        }
334    } else if ($maxlines == 0) {
335        $lines = preg_grep($pattern, $lines, PREG_GREP_INVERT);
336
337        if ((string)$newline !== ''){
338            $lines[] = $newline;
339        }
340    } else {
341        $lines = preg_replace($pattern, $replace, $lines);
342    }
343
344    if(count($lines)){
345        if(!_io_saveFile($file, join('',$lines), false)) {
346            msg("Removing content from $file failed",-1);
347            io_unlock($file);
348            return false;
349        }
350    }else{
351        @unlink($file);
352    }
353
354    io_unlock($file);
355    return true;
356}
357
358/**
359 * Delete lines that match $badline from $file.
360 *
361 * Be sure to include the trailing newline in $badline
362 *
363 * @author Patrick Brown <ptbrown@whoopdedo.org>
364 *
365 * @param string $file    filename
366 * @param string $badline exact linematch to remove
367 * @param bool   $regex   use regexp?
368 * @return bool true on success
369 */
370function io_deleteFromFile($file,$badline,$regex=false){
371    return io_replaceInFile($file,$badline,null,$regex,0);
372}
373
374/**
375 * Tries to lock a file
376 *
377 * Locking is only done for io_savefile and uses directories
378 * inside $conf['lockdir']
379 *
380 * It waits maximal 3 seconds for the lock, after this time
381 * the lock is assumed to be stale and the function goes on
382 *
383 * @author Andreas Gohr <andi@splitbrain.org>
384 *
385 * @param string $file filename
386 */
387function io_lock($file){
388    global $conf;
389    // no locking if safemode hack
390    if($conf['safemodehack']) return;
391
392    $lockDir = $conf['lockdir'].'/'.md5($file);
393    @ignore_user_abort(1);
394
395    $timeStart = time();
396    do {
397        //waited longer than 3 seconds? -> stale lock
398        if ((time() - $timeStart) > 3) break;
399        $locked = @mkdir($lockDir, $conf['dmode']);
400        if($locked){
401            if(!empty($conf['dperm'])) chmod($lockDir, $conf['dperm']);
402            break;
403        }
404        usleep(50);
405    } while ($locked === false);
406}
407
408/**
409 * Unlocks a file
410 *
411 * @author Andreas Gohr <andi@splitbrain.org>
412 *
413 * @param string $file filename
414 */
415function io_unlock($file){
416    global $conf;
417    // no locking if safemode hack
418    if($conf['safemodehack']) return;
419
420    $lockDir = $conf['lockdir'].'/'.md5($file);
421    @rmdir($lockDir);
422    @ignore_user_abort(0);
423}
424
425/**
426 * Create missing namespace directories and send the IO_NAMESPACE_CREATED events
427 * in the order of directory creation. (Parent directories first.)
428 *
429 * Event data:
430 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
431 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
432 *
433 * @author Ben Coburn <btcoburn@silicodon.net>
434 *
435 * @param string $id page id
436 * @param string $ns_type 'pages' or 'media'
437 */
438function io_createNamespace($id, $ns_type='pages') {
439    // verify ns_type
440    $types = array('pages'=>'wikiFN', 'media'=>'mediaFN');
441    if (!isset($types[$ns_type])) {
442        trigger_error('Bad $ns_type parameter for io_createNamespace().');
443        return;
444    }
445    // make event list
446    $missing = array();
447    $ns_stack = explode(':', $id);
448    $ns = $id;
449    $tmp = dirname( $file = call_user_func($types[$ns_type], $ns) );
450    while (!@is_dir($tmp) && !(file_exists($tmp) && !is_dir($tmp))) {
451        array_pop($ns_stack);
452        $ns = implode(':', $ns_stack);
453        if (strlen($ns)==0) { break; }
454        $missing[] = $ns;
455        $tmp = dirname(call_user_func($types[$ns_type], $ns));
456    }
457    // make directories
458    io_makeFileDir($file);
459    // send the events
460    $missing = array_reverse($missing); // inside out
461    foreach ($missing as $ns) {
462        $data = array($ns, $ns_type);
463        trigger_event('IO_NAMESPACE_CREATED', $data);
464    }
465}
466
467/**
468 * Create the directory needed for the given file
469 *
470 * @author  Andreas Gohr <andi@splitbrain.org>
471 *
472 * @param string $file file name
473 */
474function io_makeFileDir($file){
475    $dir = dirname($file);
476    if(!@is_dir($dir)){
477        io_mkdir_p($dir) || msg("Creating directory $dir failed",-1);
478    }
479}
480
481/**
482 * Creates a directory hierachy.
483 *
484 * @link    http://www.php.net/manual/en/function.mkdir.php
485 * @author  <saint@corenova.com>
486 * @author  Andreas Gohr <andi@splitbrain.org>
487 *
488 * @param string $target filename
489 * @return bool|int|string
490 */
491function io_mkdir_p($target){
492    global $conf;
493    if (@is_dir($target)||empty($target)) return 1; // best case check first
494    if (file_exists($target) && !is_dir($target)) return 0;
495    //recursion
496    if (io_mkdir_p(substr($target,0,strrpos($target,'/')))){
497        if($conf['safemodehack']){
498            $dir = preg_replace('/^'.preg_quote(fullpath($conf['ftp']['root']),'/').'/','', $target);
499            return io_mkdir_ftp($dir);
500        }else{
501            $ret = @mkdir($target,$conf['dmode']); // crawl back up & create dir tree
502            if($ret && !empty($conf['dperm'])) chmod($target, $conf['dperm']);
503            return $ret;
504        }
505    }
506    return 0;
507}
508
509/**
510 * Recursively delete a directory
511 *
512 * @author Andreas Gohr <andi@splitbrain.org>
513 * @param string $path
514 * @param bool   $removefiles defaults to false which will delete empty directories only
515 * @return bool
516 */
517function io_rmdir($path, $removefiles = false) {
518    if(!is_string($path) || $path == "") return false;
519    if(!file_exists($path)) return true; // it's already gone or was never there, count as success
520
521    if(is_dir($path) && !is_link($path)) {
522        $dirs  = array();
523        $files = array();
524
525        if(!$dh = @opendir($path)) return false;
526        while(false !== ($f = readdir($dh))) {
527            if($f == '..' || $f == '.') continue;
528
529            // collect dirs and files first
530            if(is_dir("$path/$f") && !is_link("$path/$f")) {
531                $dirs[] = "$path/$f";
532            } else if($removefiles) {
533                $files[] = "$path/$f";
534            } else {
535                return false; // abort when non empty
536            }
537
538        }
539        closedir($dh);
540
541        // now traverse into  directories first
542        foreach($dirs as $dir) {
543            if(!io_rmdir($dir, $removefiles)) return false; // abort on any error
544        }
545
546        // now delete files
547        foreach($files as $file) {
548            if(!@unlink($file)) return false; //abort on any error
549        }
550
551        // remove self
552        return @rmdir($path);
553    } else if($removefiles) {
554        return @unlink($path);
555    }
556    return false;
557}
558
559/**
560 * Creates a directory using FTP
561 *
562 * This is used when the safemode workaround is enabled
563 *
564 * @author <andi@splitbrain.org>
565 *
566 * @param string $dir name of the new directory
567 * @return false|string
568 */
569function io_mkdir_ftp($dir){
570    global $conf;
571
572    if(!function_exists('ftp_connect')){
573        msg("FTP support not found - safemode workaround not usable",-1);
574        return false;
575    }
576
577    $conn = @ftp_connect($conf['ftp']['host'],$conf['ftp']['port'],10);
578    if(!$conn){
579        msg("FTP connection failed",-1);
580        return false;
581    }
582
583    if(!@ftp_login($conn, $conf['ftp']['user'], conf_decodeString($conf['ftp']['pass']))){
584        msg("FTP login failed",-1);
585        return false;
586    }
587
588    //create directory
589    $ok = @ftp_mkdir($conn, $dir);
590    //set permissions
591    @ftp_site($conn,sprintf("CHMOD %04o %s",$conf['dmode'],$dir));
592
593    @ftp_close($conn);
594    return $ok;
595}
596
597/**
598 * Creates a unique temporary directory and returns
599 * its path.
600 *
601 * @author Michael Klier <chi@chimeric.de>
602 *
603 * @return false|string path to new directory or false
604 */
605function io_mktmpdir() {
606    global $conf;
607
608    $base = $conf['tmpdir'];
609    $dir  = md5(uniqid(mt_rand(), true));
610    $tmpdir = $base.'/'.$dir;
611
612    if(io_mkdir_p($tmpdir)) {
613        return($tmpdir);
614    } else {
615        return false;
616    }
617}
618
619/**
620 * downloads a file from the net and saves it
621 *
622 * if $useAttachment is false,
623 * - $file is the full filename to save the file, incl. path
624 * - if successful will return true, false otherwise
625 *
626 * if $useAttachment is true,
627 * - $file is the directory where the file should be saved
628 * - if successful will return the name used for the saved file, false otherwise
629 *
630 * @author Andreas Gohr <andi@splitbrain.org>
631 * @author Chris Smith <chris@jalakai.co.uk>
632 *
633 * @param string $url           url to download
634 * @param string $file          path to file or directory where to save
635 * @param bool   $useAttachment if true: try to use name of download, uses otherwise $defaultName, false: uses $file as path to file
636 * @param string $defaultName   fallback for if using $useAttachment
637 * @param int    $maxSize       maximum file size
638 * @return bool|string          if failed false, otherwise true or the name of the file in the given dir
639 */
640function io_download($url,$file,$useAttachment=false,$defaultName='',$maxSize=2097152){
641    global $conf;
642    $http = new DokuHTTPClient();
643    $http->max_bodysize = $maxSize;
644    $http->timeout = 25; //max. 25 sec
645    $http->keep_alive = false; // we do single ops here, no need for keep-alive
646
647    $data = $http->get($url);
648    if(!$data) return false;
649
650    $name = '';
651    if ($useAttachment) {
652        if (isset($http->resp_headers['content-disposition'])) {
653            $content_disposition = $http->resp_headers['content-disposition'];
654            $match=array();
655            if (is_string($content_disposition) &&
656                    preg_match('/attachment;\s*filename\s*=\s*"([^"]*)"/i', $content_disposition, $match)) {
657
658                $name = utf8_basename($match[1]);
659            }
660
661        }
662
663        if (!$name) {
664            if (!$defaultName) return false;
665            $name = $defaultName;
666        }
667
668        $file = $file.$name;
669    }
670
671    $fileexists = file_exists($file);
672    $fp = @fopen($file,"w");
673    if(!$fp) return false;
674    fwrite($fp,$data);
675    fclose($fp);
676    if(!$fileexists and $conf['fperm']) chmod($file, $conf['fperm']);
677    if ($useAttachment) return $name;
678    return true;
679}
680
681/**
682 * Windows compatible rename
683 *
684 * rename() can not overwrite existing files on Windows
685 * this function will use copy/unlink instead
686 *
687 * @param string $from
688 * @param string $to
689 * @return bool succes or fail
690 */
691function io_rename($from,$to){
692    global $conf;
693    if(!@rename($from,$to)){
694        if(@copy($from,$to)){
695            if($conf['fperm']) chmod($to, $conf['fperm']);
696            @unlink($from);
697            return true;
698        }
699        return false;
700    }
701    return true;
702}
703
704/**
705 * Runs an external command with input and output pipes.
706 * Returns the exit code from the process.
707 *
708 * @author Tom N Harris <tnharris@whoopdedo.org>
709 *
710 * @param string $cmd
711 * @param string $input  input pipe
712 * @param string $output output pipe
713 * @return int exit code from process
714 */
715function io_exec($cmd, $input, &$output){
716    $descspec = array(
717            0=>array("pipe","r"),
718            1=>array("pipe","w"),
719            2=>array("pipe","w"));
720    $ph = proc_open($cmd, $descspec, $pipes);
721    if(!$ph) return -1;
722    fclose($pipes[2]); // ignore stderr
723    fwrite($pipes[0], $input);
724    fclose($pipes[0]);
725    $output = stream_get_contents($pipes[1]);
726    fclose($pipes[1]);
727    return proc_close($ph);
728}
729
730/**
731 * Search a file for matching lines
732 *
733 * This is probably not faster than file()+preg_grep() but less
734 * memory intensive because not the whole file needs to be loaded
735 * at once.
736 *
737 * @author Andreas Gohr <andi@splitbrain.org>
738 * @param  string $file    The file to search
739 * @param  string $pattern PCRE pattern
740 * @param  int    $max     How many lines to return (0 for all)
741 * @param  bool   $backref When true returns array with backreferences instead of lines
742 * @return array matching lines or backref, false on error
743 */
744function io_grep($file,$pattern,$max=0,$backref=false){
745    $fh = @fopen($file,'r');
746    if(!$fh) return false;
747    $matches = array();
748
749    $cnt  = 0;
750    $line = '';
751    while (!feof($fh)) {
752        $line .= fgets($fh, 4096);  // read full line
753        if(substr($line,-1) != "\n") continue;
754
755        // check if line matches
756        if(preg_match($pattern,$line,$match)){
757            if($backref){
758                $matches[] = $match;
759            }else{
760                $matches[] = $line;
761            }
762            $cnt++;
763        }
764        if($max && $max == $cnt) break;
765        $line = '';
766    }
767    fclose($fh);
768    return $matches;
769}
770
771