xref: /dokuwiki/inc/io.php (revision 5c48379643fe73c0682a2770d0abc793c9bd1513) 
1<?php
2
3/**
4 * File IO functions
5 *
6 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
7 * @author     Andreas Gohr <andi@splitbrain.org>
8 */
9
10use dokuwiki\Utf8\PhpString;
11use dokuwiki\HTTP\DokuHTTPClient;
12use dokuwiki\Extension\Event;
13
14/**
15 * Removes empty directories
16 *
17 * Sends IO_NAMESPACE_DELETED events for 'pages' and 'media' namespaces.
18 * Event data:
19 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
20 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
21 *
22 * @param string $id - a pageid, the namespace of that id will be tried to deleted
23 * @param string $basedir - the config name of the type to delete (datadir or mediadir usally)
24 * @return bool - true if at least one namespace was deleted
25 *
26 * @author  Andreas Gohr <andi@splitbrain.org>
27 * @author Ben Coburn <btcoburn@silicodon.net>
28 */
29function io_sweepNS($id, $basedir = 'datadir')
30{
31    global $conf;
32    $types = ['datadir' => 'pages', 'mediadir' => 'media'];
33    $ns_type = ($types[$basedir] ?? false);
34
35    $delone = false;
36
37    //scan all namespaces
38    while (($id = getNS($id)) !== false) {
39        $dir = $conf[$basedir] . '/' . utf8_encodeFN(str_replace(':', '/', $id));
40
41        //try to delete dir else return
42        if (@rmdir($dir)) {
43            if ($ns_type !== false) {
44                $data = [$id, $ns_type];
45                $delone = true; // we deleted at least one dir
46                Event::createAndTrigger('IO_NAMESPACE_DELETED', $data);
47            }
48        } else {
49            return $delone;
50        }
51    }
52    return $delone;
53}
54
55/**
56 * Used to read in a DokuWiki page from file, and send IO_WIKIPAGE_READ events.
57 *
58 * Generates the action event which delegates to io_readFile().
59 * Action plugins are allowed to modify the page content in transit.
60 * The file path should not be changed.
61 *
62 * Event data:
63 * $data[0]    The raw arguments for io_readFile as an array.
64 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
65 * $data[2]    page_name: The wiki page name.
66 * $data[3]    rev: The page revision, false for current wiki pages.
67 *
68 * @param string $file filename
69 * @param string $id page id
70 * @param bool|int|string $rev revision timestamp
71 * @return string
72 *
73 * @author Ben Coburn <btcoburn@silicodon.net>
74 */
75function io_readWikiPage($file, $id, $rev = false)
76{
77    if (empty($rev)) {
78        $rev = false;
79    }
80    $data = [[$file, true], getNS($id), noNS($id), $rev];
81    return Event::createAndTrigger('IO_WIKIPAGE_READ', $data, '_io_readWikiPage_action', false);
82}
83
84/**
85 * Callback adapter for io_readFile().
86 *
87 * @param array $data event data
88 * @return string
89 *
90 * @author Ben Coburn <btcoburn@silicodon.net>
91 */
92function _io_readWikiPage_action($data)
93{
94    if (is_array($data) && is_array($data[0]) && count($data[0]) === 2) {
95        return io_readFile(...$data[0]);
96    } else {
97        return ''; //callback error
98    }
99}
100
101/**
102 * Returns content of $file as cleaned string.
103 *
104 * Uses gzip if extension is .gz
105 *
106 * If you want to use the returned value in unserialize
107 * be sure to set $clean to false!
108 *
109 *
110 * @param string $file filename
111 * @param bool $clean
112 * @return string|bool the file contents or false on error
113 *
114 * @author  Andreas Gohr <andi@splitbrain.org>
115 */
116function io_readFile($file, $clean = true)
117{
118    $ret = '';
119    if (file_exists($file)) {
120        if (substr($file, -3) == '.gz') {
121            if (!DOKU_HAS_GZIP) return false;
122            $ret = gzfile($file);
123            if (is_array($ret)) {
124                $ret = implode('', $ret);
125            }
126        } elseif (substr($file, -4) == '.bz2') {
127            if (!DOKU_HAS_BZIP) return false;
128            $ret = bzfile($file);
129        } else {
130            $ret = file_get_contents($file);
131        }
132    }
133    if ($ret === null) return false;
134    if ($ret !== false && $clean) {
135        return cleanText($ret);
136    } else {
137        return $ret;
138    }
139}
140
141/**
142 * Returns the content of a .bz2 compressed file as string
143 *
144 * @param string $file filename
145 * @param bool $array return array of lines
146 * @return string|array|bool content or false on error
147 *
148 * @author marcel senf <marcel@rucksackreinigung.de>
149 * @author  Andreas Gohr <andi@splitbrain.org>
150 */
151function bzfile($file, $array = false)
152{
153    $bz = bzopen($file, "r");
154    if ($bz === false) return false;
155
156    if ($array) {
157        $lines = [];
158    }
159    $str = '';
160    while (!feof($bz)) {
161        //8192 seems to be the maximum buffersize?
162        $buffer = bzread($bz, 8192);
163        if (($buffer === false) || (bzerrno($bz) !== 0)) {
164            return false;
165        }
166        $str .= $buffer;
167        if ($array) {
168            $pos = strpos($str, "\n");
169            while ($pos !== false) {
170                $lines[] = substr($str, 0, $pos + 1);
171                $str = substr($str, $pos + 1);
172                $pos = strpos($str, "\n");
173            }
174        }
175    }
176    bzclose($bz);
177    if ($array) {
178        if ($str !== '') {
179            $lines[] = $str;
180        }
181        return $lines;
182    }
183    return $str;
184}
185
186/**
187 * Used to write out a DokuWiki page to file, and send IO_WIKIPAGE_WRITE events.
188 *
189 * This generates an action event and delegates to io_saveFile().
190 * Action plugins are allowed to modify the page content in transit.
191 * The file path should not be changed.
192 * (The append parameter is set to false.)
193 *
194 * Event data:
195 * $data[0]    The raw arguments for io_saveFile as an array.
196 * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
197 * $data[2]    page_name: The wiki page name.
198 * $data[3]    rev: The page revision, false for current wiki pages.
199 *
200 * @param string $file filename
201 * @param string $content
202 * @param string $id page id
203 * @param int|bool|string $rev timestamp of revision
204 * @return bool
205 *
206 * @author Ben Coburn <btcoburn@silicodon.net>
207 */
208function io_writeWikiPage($file, $content, $id, $rev = false)
209{
210    if (empty($rev)) {
211        $rev = false;
212    }
213    if ($rev === false) {
214        io_createNamespace($id); // create namespaces as needed
215    }
216    $data = [[$file, $content, false], getNS($id), noNS($id), $rev];
217    return Event::createAndTrigger('IO_WIKIPAGE_WRITE', $data, '_io_writeWikiPage_action', false);
218}
219
220/**
221 * Callback adapter for io_saveFile().
222 *
223 * @param array $data event data
224 * @return bool
225 *
226 * @author Ben Coburn <btcoburn@silicodon.net>
227 */
228function _io_writeWikiPage_action($data)
229{
230    if (is_array($data) && is_array($data[0]) && count($data[0]) === 3) {
231        $ok = io_saveFile(...$data[0]);
232        // for attic files make sure the file has the mtime of the revision
233        if ($ok && is_int($data[3]) && $data[3] > 0) {
234            @touch($data[0][0], $data[3]);
235        }
236        return $ok;
237    } else {
238        return false; //callback error
239    }
240}
241
242/**
243 * Internal function to save contents to a file.
244 *
245 * @param string $file filename path to file
246 * @param string $content
247 * @param bool $append
248 * @return bool true on success, otherwise false
249 *
250 * @author  Andreas Gohr <andi@splitbrain.org>
251 */
252function _io_saveFile($file, $content, $append)
253{
254    global $conf;
255    $mode = ($append) ? 'ab' : 'wb';
256    $fileexists = file_exists($file);
257
258    if (substr($file, -3) == '.gz') {
259        if (!DOKU_HAS_GZIP) return false;
260        $fh = @gzopen($file, $mode . '9');
261        if (!$fh) return false;
262        gzwrite($fh, $content);
263        gzclose($fh);
264    } elseif (substr($file, -4) == '.bz2') {
265        if (!DOKU_HAS_BZIP) return false;
266        if ($append) {
267            $bzcontent = bzfile($file);
268            if ($bzcontent === false) return false;
269            $content = $bzcontent . $content;
270        }
271        $fh = @bzopen($file, 'w');
272        if (!$fh) return false;
273        bzwrite($fh, $content);
274        bzclose($fh);
275    } else {
276        $fh = @fopen($file, $mode);
277        if (!$fh) return false;
278        fwrite($fh, $content);
279        fclose($fh);
280    }
281
282    if (!$fileexists && $conf['fperm']) {
283        chmod($file, $conf['fperm']);
284    }
285    return true;
286}
287
288/**
289 * Saves $content to $file.
290 *
291 * If the third parameter is set to true the given content
292 * will be appended.
293 *
294 * Uses gzip if extension is .gz
295 * and bz2 if extension is .bz2
296 *
297 * @param string $file filename path to file
298 * @param string $content
299 * @param bool $append
300 * @return bool true on success, otherwise false
301 *
302 * @author  Andreas Gohr <andi@splitbrain.org>
303 */
304function io_saveFile($file, $content, $append = false)
305{
306    io_makeFileDir($file);
307    io_lock($file);
308    if (!_io_saveFile($file, $content, $append)) {
309        msg("Writing $file failed", -1);
310        io_unlock($file);
311        return false;
312    }
313    io_unlock($file);
314    return true;
315}
316
317/**
318 * Replace one or more occurrences of a line in a file.
319 *
320 * The default, when $maxlines is 0 is to delete all matching lines then append a single line.
321 * A regex that matches any part of the line will remove the entire line in this mode.
322 * Captures in $newline are not available.
323 *
324 * Otherwise each line is matched and replaced individually, up to the first $maxlines lines
325 * or all lines if $maxlines is -1. If $regex is true then captures can be used in $newline.
326 *
327 * Be sure to include the trailing newline in $oldline when replacing entire lines.
328 *
329 * Uses gzip if extension is .gz
330 * and bz2 if extension is .bz2
331 *
332 * @param string $file filename
333 * @param string $oldline exact linematch to remove
334 * @param string $newline new line to insert
335 * @param bool $regex use regexp?
336 * @param int $maxlines number of occurrences of the line to replace
337 * @return bool true on success
338 *
339 * @author Steven Danz <steven-danz@kc.rr.com>
340 * @author Christopher Smith <chris@jalakai.co.uk>
341 * @author Patrick Brown <ptbrown@whoopdedo.org>
342 */
343function io_replaceInFile($file, $oldline, $newline, $regex = false, $maxlines = 0)
344{
345    if ((string)$oldline === '') {
346        trigger_error('$oldline parameter cannot be empty in io_replaceInFile()', E_USER_WARNING);
347        return false;
348    }
349
350    if (!file_exists($file)) return true;
351
352    io_lock($file);
353
354    // load into array
355    if (substr($file, -3) == '.gz') {
356        if (!DOKU_HAS_GZIP) return false;
357        $lines = gzfile($file);
358    } elseif (substr($file, -4) == '.bz2') {
359        if (!DOKU_HAS_BZIP) return false;
360        $lines = bzfile($file, true);
361    } else {
362        $lines = file($file);
363    }
364
365    // make non-regexes into regexes
366    $pattern = $regex ? $oldline : '/^' . preg_quote($oldline, '/') . '$/';
367    $replace = $regex ? $newline : addcslashes($newline, '\$');
368
369    // remove matching lines
370    if ($maxlines > 0) {
371        $count = 0;
372        $matched = 0;
373        foreach ($lines as $i => $line) {
374            if ($count >= $maxlines) break;
375            // $matched will be set to 0|1 depending on whether pattern is matched and line replaced
376            $lines[$i] = preg_replace($pattern, $replace, $line, -1, $matched);
377            if ($matched) {
378                $count++;
379            }
380        }
381    } elseif ($maxlines == 0) {
382        $lines = preg_grep($pattern, $lines, PREG_GREP_INVERT);
383        if ((string)$newline !== '') {
384            $lines[] = $newline;
385        }
386    } else {
387        $lines = preg_replace($pattern, $replace, $lines);
388    }
389
390    if (count($lines)) {
391        if (!_io_saveFile($file, implode('', $lines), false)) {
392            msg("Removing content from $file failed", -1);
393            io_unlock($file);
394            return false;
395        }
396    } else {
397        @unlink($file);
398    }
399
400    io_unlock($file);
401    return true;
402}
403
404/**
405 * Delete lines that match $badline from $file.
406 *
407 * Be sure to include the trailing newline in $badline
408 *
409 * @param string $file filename
410 * @param string $badline exact linematch to remove
411 * @param bool $regex use regexp?
412 * @return bool true on success
413 *
414 * @author Patrick Brown <ptbrown@whoopdedo.org>
415 */
416function io_deleteFromFile($file, $badline, $regex = false)
417{
418    return io_replaceInFile($file, $badline, null, $regex, 0);
419}
420
421/**
422 * Tries to lock a file
423 *
424 * Locking is only done for io_savefile and uses directories
425 * inside $conf['lockdir']
426 *
427 * It waits maximal 3 seconds for the lock, after this time
428 * the lock is assumed to be stale and the function goes on
429 *
430 * @param string $file filename
431 *
432 * @author Andreas Gohr <andi@splitbrain.org>
433 */
434function io_lock($file)
435{
436    global $conf;
437
438    $lockDir = $conf['lockdir'] . '/' . md5($file);
439    @ignore_user_abort(1);
440
441    $timeStart = time();
442    do {
443        //waited longer than 3 seconds? -> stale lock
444        if ((time() - $timeStart) > 3) break;
445        $locked = @mkdir($lockDir);
446        if ($locked) {
447            if ($conf['dperm']) {
448                chmod($lockDir, $conf['dperm']);
449            }
450            break;
451        }
452        usleep(50);
453    } while ($locked === false);
454}
455
456/**
457 * Unlocks a file
458 *
459 * @param string $file filename
460 *
461 * @author Andreas Gohr <andi@splitbrain.org>
462 */
463function io_unlock($file)
464{
465    global $conf;
466
467    $lockDir = $conf['lockdir'] . '/' . md5($file);
468    @rmdir($lockDir);
469    @ignore_user_abort(0);
470}
471
472/**
473 * Create missing namespace directories and send the IO_NAMESPACE_CREATED events
474 * in the order of directory creation. (Parent directories first.)
475 *
476 * Event data:
477 * $data[0]    ns: The colon separated namespace path minus the trailing page name.
478 * $data[1]    ns_type: 'pages' or 'media' namespace tree.
479 *
480 * @param string $id page id
481 * @param string $ns_type 'pages' or 'media'
482 *
483 * @author Ben Coburn <btcoburn@silicodon.net>
484 */
485function io_createNamespace($id, $ns_type = 'pages')
486{
487    // verify ns_type
488    $types = ['pages' => 'wikiFN', 'media' => 'mediaFN'];
489    if (!isset($types[$ns_type])) {
490        trigger_error('Bad $ns_type parameter for io_createNamespace().');
491        return;
492    }
493    // make event list
494    $missing = [];
495    $ns_stack = explode(':', $id);
496    $ns = $id;
497    $tmp = dirname($file = call_user_func($types[$ns_type], $ns));
498    while (!@is_dir($tmp) && !(file_exists($tmp) && !is_dir($tmp))) {
499        array_pop($ns_stack);
500        $ns = implode(':', $ns_stack);
501        if (strlen($ns) == 0) {
502            break;
503        }
504        $missing[] = $ns;
505        $tmp = dirname(call_user_func($types[$ns_type], $ns));
506    }
507    // make directories
508    io_makeFileDir($file);
509    // send the events
510    $missing = array_reverse($missing); // inside out
511    foreach ($missing as $ns) {
512        $data = [$ns, $ns_type];
513        Event::createAndTrigger('IO_NAMESPACE_CREATED', $data);
514    }
515}
516
517/**
518 * Create the directory needed for the given file
519 *
520 * @param string $file file name
521 *
522 * @author  Andreas Gohr <andi@splitbrain.org>
523 */
524function io_makeFileDir($file)
525{
526    $dir = dirname($file);
527    if (!@is_dir($dir)) {
528        io_mkdir_p($dir) || msg("Creating directory $dir failed", -1);
529    }
530}
531
532/**
533 * Creates a directory hierachy.
534 *
535 * @param string $target filename
536 * @return bool
537 *
538 * @link    http://php.net/manual/en/function.mkdir.php
539 * @author  <saint@corenova.com>
540 * @author  Andreas Gohr <andi@splitbrain.org>
541 */
542function io_mkdir_p($target)
543{
544    global $conf;
545    if (@is_dir($target) || empty($target)) return true; // best case check first
546    if (file_exists($target) && !is_dir($target)) return false;
547    //recursion
548    if (io_mkdir_p(substr($target, 0, strrpos($target, '/')))) {
549        $ret = @mkdir($target); // crawl back up & create dir tree
550        if ($ret && !empty($conf['dperm'])) {
551            chmod($target, $conf['dperm']);
552        }
553        return $ret;
554    }
555    return false;
556}
557
558/**
559 * Recursively delete a directory
560 *
561 * @param string $path
562 * @param bool $removefiles defaults to false which will delete empty directories only
563 * @return bool
564 *
565 * @author Andreas Gohr <andi@splitbrain.org>
566 */
567function io_rmdir($path, $removefiles = false)
568{
569    if (!is_string($path) || $path == "") return false;
570    if (!file_exists($path)) return true; // it's already gone or was never there, count as success
571
572    if (is_dir($path) && !is_link($path)) {
573        $dirs = [];
574        $files = [];
575        if (!$dh = @opendir($path)) return false;
576        while (false !== ($f = readdir($dh))) {
577            if ($f == '..' || $f == '.') continue;
578
579            // collect dirs and files first
580            if (is_dir("$path/$f") && !is_link("$path/$f")) {
581                $dirs[] = "$path/$f";
582            } elseif ($removefiles) {
583                $files[] = "$path/$f";
584            } else {
585                return false; // abort when non empty
586            }
587        }
588        closedir($dh);
589        // now traverse into  directories first
590        foreach ($dirs as $dir) {
591            if (!io_rmdir($dir, $removefiles)) return false; // abort on any error
592        }
593        // now delete files
594        foreach ($files as $file) {
595            if (!@unlink($file)) return false; //abort on any error
596        }
597        // remove self
598        return @rmdir($path);
599    } elseif ($removefiles) {
600        return @unlink($path);
601    }
602    return false;
603}
604
605/**
606 * Creates a unique temporary directory and returns
607 * its path.
608 *
609 * @return false|string path to new directory or false
610 * @throws Exception
611 *
612 * @author Michael Klier <chi@chimeric.de>
613 */
614function io_mktmpdir()
615{
616    global $conf;
617
618    $base = $conf['tmpdir'];
619    $dir = md5(uniqid(random_int(0, mt_getrandmax()), true));
620    $tmpdir = $base . '/' . $dir;
621
622    if (io_mkdir_p($tmpdir)) {
623        return $tmpdir;
624    } else {
625        return false;
626    }
627}
628
629/**
630 * downloads a file from the net and saves it
631 *
632 * if $useAttachment is false,
633 * - $file is the full filename to save the file, incl. path
634 * - if successful will return true, false otherwise
635 *
636 * if $useAttachment is true,
637 * - $file is the directory where the file should be saved
638 * - if successful will return the name used for the saved file, false otherwise
639 *
640 * @param string $url url to download
641 * @param string $file path to file or directory where to save
642 * @param bool $useAttachment true: try to use name of download, uses otherwise $defaultName
643 *                            false: uses $file as path to file
644 * @param string $defaultName fallback for if using $useAttachment
645 * @param int $maxSize maximum file size
646 * @return bool|string          if failed false, otherwise true or the name of the file in the given dir
647 *
648 * @author Andreas Gohr <andi@splitbrain.org>
649 * @author Chris Smith <chris@jalakai.co.uk>
650 */
651function io_download($url, $file, $useAttachment = false, $defaultName = '', $maxSize = 2_097_152)
652{
653    global $conf;
654    $http = new DokuHTTPClient();
655    $http->max_bodysize = $maxSize;
656    $http->timeout = 25; //max. 25 sec
657    $http->keep_alive = false; // we do single ops here, no need for keep-alive
658
659    $data = $http->get($url);
660    if (!$data) return false;
661
662    $name = '';
663    if ($useAttachment) {
664        if (isset($http->resp_headers['content-disposition'])) {
665            $content_disposition = $http->resp_headers['content-disposition'];
666            $match = [];
667            if (
668                is_string($content_disposition) &&
669                preg_match('/attachment;\s*filename\s*=\s*"([^"]*)"/i', $content_disposition, $match)
670            ) {
671                $name = PhpString::basename($match[1]);
672            }
673        }
674
675        if (!$name) {
676            if (!$defaultName) return false;
677            $name = $defaultName;
678        }
679
680        $file .= $name;
681    }
682
683    $fileexists = file_exists($file);
684    $fp = @fopen($file, "w");
685    if (!$fp) return false;
686    fwrite($fp, $data);
687    fclose($fp);
688    if (!$fileexists && $conf['fperm']) {
689        chmod($file, $conf['fperm']);
690    }
691    if ($useAttachment) return $name;
692    return true;
693}
694
695/**
696 * Windows compatible rename
697 *
698 * rename() can not overwrite existing files on Windows
699 * this function will use copy/unlink instead
700 *
701 * @param string $from
702 * @param string $to
703 * @return bool succes or fail
704 */
705function io_rename($from, $to)
706{
707    global $conf;
708    if (!@rename($from, $to)) {
709        if (@copy($from, $to)) {
710            if ($conf['fperm']) {
711                chmod($to, $conf['fperm']);
712            }
713            @unlink($from);
714            return true;
715        }
716        return false;
717    }
718    return true;
719}
720
721/**
722 * Runs an external command with input and output pipes.
723 * Returns the exit code from the process.
724 *
725 * @param string $cmd
726 * @param string $input input pipe
727 * @param string $output output pipe
728 * @return int exit code from process
729 *
730 * @author Tom N Harris <tnharris@whoopdedo.org>
731 */
732function io_exec($cmd, $input, &$output)
733{
734    $descspec = [
735        0 => ["pipe", "r"],
736        1 => ["pipe", "w"],
737        2 => ["pipe", "w"]
738    ];
739    $ph = proc_open($cmd, $descspec, $pipes);
740    if (!$ph) return -1;
741    fclose($pipes[2]); // ignore stderr
742    fwrite($pipes[0], $input);
743    fclose($pipes[0]);
744    $output = stream_get_contents($pipes[1]);
745    fclose($pipes[1]);
746    return proc_close($ph);
747}
748
749/**
750 * Search a file for matching lines
751 *
752 * This is probably not faster than file()+preg_grep() but less
753 * memory intensive because not the whole file needs to be loaded
754 * at once.
755 *
756 * @param string $file The file to search
757 * @param string $pattern PCRE pattern
758 * @param int $max How many lines to return (0 for all)
759 * @param bool $backref When true returns array with backreferences instead of lines
760 * @return array matching lines or backref, false on error
761 *
762 * @author Andreas Gohr <andi@splitbrain.org>
763 */
764function io_grep($file, $pattern, $max = 0, $backref = false)
765{
766    $fh = @fopen($file, 'r');
767    if (!$fh) return false;
768    $matches = [];
769
770    $cnt = 0;
771    $line = '';
772    while (!feof($fh)) {
773        $line .= fgets($fh, 4096);  // read full line
774        if (substr($line, -1) != "\n") continue;
775
776        // check if line matches
777        if (preg_match($pattern, $line, $match)) {
778            if ($backref) {
779                $matches[] = $match;
780            } else {
781                $matches[] = $line;
782            }
783            $cnt++;
784        }
785        if ($max && $max == $cnt) break;
786        $line = '';
787    }
788    fclose($fh);
789    return $matches;
790}
791
792
793/**
794 * Get size of contents of a file, for a compressed file the uncompressed size
795 * Warning: reading uncompressed size of content of bz-files requires uncompressing
796 *
797 * @param string $file filename path to file
798 * @return int size of file
799 *
800 * @author  Gerrit Uitslag <klapinklapin@gmail.com>
801 */
802function io_getSizeFile($file)
803{
804    if (!file_exists($file)) return 0;
805
806    if (substr($file, -3) == '.gz') {
807        $fp = @fopen($file, "rb");
808        if ($fp === false) return 0;
809        fseek($fp, -4, SEEK_END);
810        $buffer = fread($fp, 4);
811        fclose($fp);
812        $array = unpack("V", $buffer);
813        $uncompressedsize = end($array);
814    } elseif (substr($file, -4) == '.bz2') {
815        if (!DOKU_HAS_BZIP) return 0;
816        $bz = bzopen($file, "r");
817        if ($bz === false) return 0;
818        $uncompressedsize = 0;
819        while (!feof($bz)) {
820            //8192 seems to be the maximum buffersize?
821            $buffer = bzread($bz, 8192);
822            if (($buffer === false) || (bzerrno($bz) !== 0)) {
823                return 0;
824            }
825            $uncompressedsize += strlen($buffer);
826        }
827    } else {
828        $uncompressedsize = filesize($file);
829    }
830
831    return $uncompressedsize;
832}
833