inc/io.php

   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
   9 use dokuwiki\HTTP\DokuHTTPClient;
  10 use dokuwiki\Extension\Event;
  11
  12 /**
  13  * Removes empty directories
  14  *
  15  * Sends IO_NAMESPACE_DELETED events for 'pages' and 'media' namespaces.
  16  * Event data:
  17  * $data[0]    ns: The colon separated namespace path minus the trailing page name.
  18  * $data[1]    ns_type: 'pages' or 'media' namespace tree.
  19  *
  20  * @param string $id      - a pageid, the namespace of that id will be tried to deleted
  21  * @param string $basedir - the config name of the type to delete (datadir or mediadir usally)
  22  * @return bool - true if at least one namespace was deleted
  23  *
  24  * @author  Andreas Gohr <andi@splitbrain.org>
  25  * @author Ben Coburn <btcoburn@silicodon.net>
  26  */
  27 function io_sweepNS($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                 Event::createAndTrigger('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 $rev revision timestamp
  68  * @return string
  69  */
  70 function io_readWikiPage($file, $id, $rev=false) {
  71     if (empty($rev)) { $rev = false; }
  72     $data = array(array($file, true), getNS($id), noNS($id), $rev);
  73     return Event::createAndTrigger('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 $data event data
  82  * @return string
  83  */
  84 function _io_readWikiPage_action($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   $clean
 104  * @return string|bool the file contents or false on error
 105  */
 106 function io_readFile($file,$clean=true){
 107     $ret = '';
 108     if(file_exists($file)){
 109         if(substr($file,-3) == '.gz'){
 110             if(!DOKU_HAS_GZIP) return false;
 111             $ret = gzfile($file);
 112             if(is_array($ret)) $ret = join('', $ret);
 113         }else if(substr($file,-4) == '.bz2'){
 114             if(!DOKU_HAS_BZIP) return false;
 115             $ret = bzfile($file);
 116         }else{
 117             $ret = file_get_contents($file);
 118         }
 119     }
 120     if($ret === null) return false;
 121     if($ret !== false && $clean){
 122         return cleanText($ret);
 123     }else{
 124         return $ret;
 125     }
 126 }
 127 /**
 128  * Returns the content of a .bz2 compressed file as string
 129  *
 130  * @author marcel senf <marcel@rucksackreinigung.de>
 131  * @author  Andreas Gohr <andi@splitbrain.org>
 132  *
 133  * @param string $file filename
 134  * @param bool   $array return array of lines
 135  * @return string|array|bool content or false on error
 136  */
 137 function bzfile($file, $array=false) {
 138     $bz = bzopen($file,"r");
 139     if($bz === false) return false;
 140
 141     if($array) $lines = array();
 142     $str = '';
 143     while (!feof($bz)) {
 144         //8192 seems to be the maximum buffersize?
 145         $buffer = bzread($bz,8192);
 146         if(($buffer === false) || (bzerrno($bz) !== 0)) {
 147             return false;
 148         }
 149         $str = $str . $buffer;
 150         if($array) {
 151             $pos = strpos($str, "\n");
 152             while($pos !== false) {
 153                 $lines[] = substr($str, 0, $pos+1);
 154                 $str = substr($str, $pos+1);
 155                 $pos = strpos($str, "\n");
 156             }
 157         }
 158     }
 159     bzclose($bz);
 160     if($array) {
 161         if($str !== '') $lines[] = $str;
 162         return $lines;
 163     }
 164     return $str;
 165 }
 166
 167 /**
 168  * Used to write out a DokuWiki page to file, and send IO_WIKIPAGE_WRITE events.
 169  *
 170  * This generates an action event and delegates to io_saveFile().
 171  * Action plugins are allowed to modify the page content in transit.
 172  * The file path should not be changed.
 173  * (The append parameter is set to false.)
 174  *
 175  * Event data:
 176  * $data[0]    The raw arguments for io_saveFile as an array.
 177  * $data[1]    ns: The colon separated namespace path minus the trailing page name. (false if root ns)
 178  * $data[2]    page_name: The wiki page name.
 179  * $data[3]    rev: The page revision, false for current wiki pages.
 180  *
 181  * @author Ben Coburn <btcoburn@silicodon.net>
 182  *
 183  * @param string $file      filename
 184  * @param string $content
 185  * @param string $id        page id
 186  * @param int|bool $rev timestamp of revision
 187  * @return bool
 188  */
 189 function io_writeWikiPage($file, $content, $id, $rev=false) {
 190     if (empty($rev)) { $rev = false; }
 191     if ($rev===false) { io_createNamespace($id); } // create namespaces as needed
 192     $data = array(array($file, $content, false), getNS($id), noNS($id), $rev);
 193     return Event::createAndTrigger('IO_WIKIPAGE_WRITE', $data, '_io_writeWikiPage_action', false);
 194 }
 195
 196 /**
 197  * Callback adapter for io_saveFile().
 198  * @author Ben Coburn <btcoburn@silicodon.net>
 199  *
 200  * @param array $data event data
 201  * @return bool
 202  */
 203 function _io_writeWikiPage_action($data) {
 204     if (is_array($data) && is_array($data[0]) && count($data[0])===3) {
 205         $ok = call_user_func_array('io_saveFile', $data[0]);
 206         // for attic files make sure the file has the mtime of the revision
 207         if($ok && is_int($data[3]) && $data[3] > 0) {
 208             @touch($data[0][0], $data[3]);
 209         }
 210         return $ok;
 211     } else {
 212         return false; //callback error
 213     }
 214 }
 215
 216 /**
 217  * Internal function to save contents to a file.
 218  *
 219  * @author  Andreas Gohr <andi@splitbrain.org>
 220  *
 221  * @param string $file filename path to file
 222  * @param string $content
 223  * @param bool   $append
 224  * @return bool true on success, otherwise false
 225  */
 226 function _io_saveFile($file, $content, $append) {
 227     global $conf;
 228     $mode = ($append) ? 'ab' : 'wb';
 229     $fileexists = file_exists($file);
 230
 231     if(substr($file,-3) == '.gz'){
 232         if(!DOKU_HAS_GZIP) return false;
 233         $fh = @gzopen($file,$mode.'9');
 234         if(!$fh) return false;
 235         gzwrite($fh, $content);
 236         gzclose($fh);
 237     }else if(substr($file,-4) == '.bz2'){
 238         if(!DOKU_HAS_BZIP) return false;
 239         if($append) {
 240             $bzcontent = bzfile($file);
 241             if($bzcontent === false) return false;
 242             $content = $bzcontent.$content;
 243         }
 244         $fh = @bzopen($file,'w');
 245         if(!$fh) return false;
 246         bzwrite($fh, $content);
 247         bzclose($fh);
 248     }else{
 249         $fh = @fopen($file,$mode);
 250         if(!$fh) return false;
 251         fwrite($fh, $content);
 252         fclose($fh);
 253     }
 254
 255     if(!$fileexists and $conf['fperm']) chmod($file, $conf['fperm']);
 256     return true;
 257 }
 258
 259 /**
 260  * Saves $content to $file.
 261  *
 262  * If the third parameter is set to true the given content
 263  * will be appended.
 264  *
 265  * Uses gzip if extension is .gz
 266  * and bz2 if extension is .bz2
 267  *
 268  * @author  Andreas Gohr <andi@splitbrain.org>
 269  *
 270  * @param string $file filename path to file
 271  * @param string $content
 272  * @param bool   $append
 273  * @return bool true on success, otherwise false
 274  */
 275 function io_saveFile($file, $content, $append=false) {
 276     io_makeFileDir($file);
 277     io_lock($file);
 278     if(!_io_saveFile($file, $content, $append)) {
 279         msg("Writing $file failed",-1);
 280         io_unlock($file);
 281         return false;
 282     }
 283     io_unlock($file);
 284     return true;
 285 }
 286
 287 /**
 288  * Replace one or more occurrences of a line in a file.
 289  *
 290  * The default, when $maxlines is 0 is to delete all matching lines then append a single line.
 291  * A regex that matches any part of the line will remove the entire line in this mode.
 292  * Captures in $newline are not available.
 293  *
 294  * Otherwise each line is matched and replaced individually, up to the first $maxlines lines
 295  * or all lines if $maxlines is -1. If $regex is true then captures can be used in $newline.
 296  *
 297  * Be sure to include the trailing newline in $oldline when replacing entire lines.
 298  *
 299  * Uses gzip if extension is .gz
 300  * and bz2 if extension is .bz2
 301  *
 302  * @author Steven Danz <steven-danz@kc.rr.com>
 303  * @author Christopher Smith <chris@jalakai.co.uk>
 304  * @author Patrick Brown <ptbrown@whoopdedo.org>
 305  *
 306  * @param string $file     filename
 307  * @param string $oldline  exact linematch to remove
 308  * @param string $newline  new line to insert
 309  * @param bool   $regex    use regexp?
 310  * @param int    $maxlines number of occurrences of the line to replace
 311  * @return bool true on success
 312  */
 313 function io_replaceInFile($file, $oldline, $newline, $regex=false, $maxlines=0) {
 314     if ((string)$oldline === '') {
 315         trigger_error('$oldline parameter cannot be empty in io_replaceInFile()', E_USER_WARNING);
 316         return false;
 317     }
 318
 319     if (!file_exists($file)) return true;
 320
 321     io_lock($file);
 322
 323     // load into array
 324     if(substr($file,-3) == '.gz'){
 325         if(!DOKU_HAS_GZIP) return false;
 326         $lines = gzfile($file);
 327     }else if(substr($file,-4) == '.bz2'){
 328         if(!DOKU_HAS_BZIP) return false;
 329         $lines = bzfile($file, true);
 330     }else{
 331         $lines = file($file);
 332     }
 333
 334     // make non-regexes into regexes
 335     $pattern = $regex ? $oldline : '/^'.preg_quote($oldline,'/').'$/';
 336     $replace = $regex ? $newline : addcslashes($newline, '\$');
 337
 338     // remove matching lines
 339     if ($maxlines > 0) {
 340         $count = 0;
 341         $matched = 0;
 342         foreach($lines as $i => $line) {
 343             if($count >= $maxlines) break;
 344             // $matched will be set to 0|1 depending on whether pattern is matched and line replaced
 345             $lines[$i] = preg_replace($pattern, $replace, $line, -1, $matched);
 346             if ($matched) $count++;
 347         }
 348     } else if ($maxlines == 0) {
 349         $lines = preg_grep($pattern, $lines, PREG_GREP_INVERT);
 350
 351         if ((string)$newline !== ''){
 352             $lines[] = $newline;
 353         }
 354     } else {
 355         $lines = preg_replace($pattern, $replace, $lines);
 356     }
 357
 358     if(count($lines)){
 359         if(!_io_saveFile($file, join('',$lines), false)) {
 360             msg("Removing content from $file failed",-1);
 361             io_unlock($file);
 362             return false;
 363         }
 364     }else{
 365         @unlink($file);
 366     }
 367
 368     io_unlock($file);
 369     return true;
 370 }
 371
 372 /**
 373  * Delete lines that match $badline from $file.
 374  *
 375  * Be sure to include the trailing newline in $badline
 376  *
 377  * @author Patrick Brown <ptbrown@whoopdedo.org>
 378  *
 379  * @param string $file    filename
 380  * @param string $badline exact linematch to remove
 381  * @param bool   $regex   use regexp?
 382  * @return bool true on success
 383  */
 384 function io_deleteFromFile($file,$badline,$regex=false){
 385     return io_replaceInFile($file,$badline,null,$regex,0);
 386 }
 387
 388 /**
 389  * Tries to lock a file
 390  *
 391  * Locking is only done for io_savefile and uses directories
 392  * inside $conf['lockdir']
 393  *
 394  * It waits maximal 3 seconds for the lock, after this time
 395  * the lock is assumed to be stale and the function goes on
 396  *
 397  * @author Andreas Gohr <andi@splitbrain.org>
 398  *
 399  * @param string $file filename
 400  */
 401 function io_lock($file){
 402     global $conf;
 403
 404     $lockDir = $conf['lockdir'].'/'.md5($file);
 405     @ignore_user_abort(1);
 406
 407     $timeStart = time();
 408     do {
 409         //waited longer than 3 seconds? -> stale lock
 410         if ((time() - $timeStart) > 3) break;
 411         $locked = @mkdir($lockDir);
 412         if($locked){
 413             if($conf['dperm']) chmod($lockDir, $conf['dperm']);
 414             break;
 415         }
 416         usleep(50);
 417     } while ($locked === false);
 418 }
 419
 420 /**
 421  * Unlocks a file
 422  *
 423  * @author Andreas Gohr <andi@splitbrain.org>
 424  *
 425  * @param string $file filename
 426  */
 427 function io_unlock($file){
 428     global $conf;
 429
 430     $lockDir = $conf['lockdir'].'/'.md5($file);
 431     @rmdir($lockDir);
 432     @ignore_user_abort(0);
 433 }
 434
 435 /**
 436  * Create missing namespace directories and send the IO_NAMESPACE_CREATED events
 437  * in the order of directory creation. (Parent directories first.)
 438  *
 439  * Event data:
 440  * $data[0]    ns: The colon separated namespace path minus the trailing page name.
 441  * $data[1]    ns_type: 'pages' or 'media' namespace tree.
 442  *
 443  * @author Ben Coburn <btcoburn@silicodon.net>
 444  *
 445  * @param string $id page id
 446  * @param string $ns_type 'pages' or 'media'
 447  */
 448 function io_createNamespace($id, $ns_type='pages') {
 449     // verify ns_type
 450     $types = array('pages'=>'wikiFN', 'media'=>'mediaFN');
 451     if (!isset($types[$ns_type])) {
 452         trigger_error('Bad $ns_type parameter for io_createNamespace().');
 453         return;
 454     }
 455     // make event list
 456     $missing = array();
 457     $ns_stack = explode(':', $id);
 458     $ns = $id;
 459     $tmp = dirname( $file = call_user_func($types[$ns_type], $ns) );
 460     while (!@is_dir($tmp) && !(file_exists($tmp) && !is_dir($tmp))) {
 461         array_pop($ns_stack);
 462         $ns = implode(':', $ns_stack);
 463         if (strlen($ns)==0) { break; }
 464         $missing[] = $ns;
 465         $tmp = dirname(call_user_func($types[$ns_type], $ns));
 466     }
 467     // make directories
 468     io_makeFileDir($file);
 469     // send the events
 470     $missing = array_reverse($missing); // inside out
 471     foreach ($missing as $ns) {
 472         $data = array($ns, $ns_type);
 473         Event::createAndTrigger('IO_NAMESPACE_CREATED', $data);
 474     }
 475 }
 476
 477 /**
 478  * Create the directory needed for the given file
 479  *
 480  * @author  Andreas Gohr <andi@splitbrain.org>
 481  *
 482  * @param string $file file name
 483  */
 484 function io_makeFileDir($file){
 485     $dir = dirname($file);
 486     if(!@is_dir($dir)){
 487         io_mkdir_p($dir) || msg("Creating directory $dir failed",-1);
 488     }
 489 }
 490
 491 /**
 492  * Creates a directory hierachy.
 493  *
 494  * @link    http://php.net/manual/en/function.mkdir.php
 495  * @author  <saint@corenova.com>
 496  * @author  Andreas Gohr <andi@splitbrain.org>
 497  *
 498  * @param string $target filename
 499  * @return bool|int|string
 500  */
 501 function io_mkdir_p($target){
 502     global $conf;
 503     if (@is_dir($target)||empty($target)) return 1; // best case check first
 504     if (file_exists($target) && !is_dir($target)) return 0;
 505     //recursion
 506     if (io_mkdir_p(substr($target,0,strrpos($target,'/')))){
 507         $ret = @mkdir($target); // crawl back up & create dir tree
 508         if($ret && !empty($conf['dperm'])) chmod($target, $conf['dperm']);
 509         return $ret;
 510     }
 511     return 0;
 512 }
 513
 514 /**
 515  * Recursively delete a directory
 516  *
 517  * @author Andreas Gohr <andi@splitbrain.org>
 518  * @param string $path
 519  * @param bool   $removefiles defaults to false which will delete empty directories only
 520  * @return bool
 521  */
 522 function io_rmdir($path, $removefiles = false) {
 523     if(!is_string($path) || $path == "") return false;
 524     if(!file_exists($path)) return true; // it's already gone or was never there, count as success
 525
 526     if(is_dir($path) && !is_link($path)) {
 527         $dirs  = array();
 528         $files = array();
 529
 530         if(!$dh = @opendir($path)) return false;
 531         while(false !== ($f = readdir($dh))) {
 532             if($f == '..' || $f == '.') continue;
 533
 534             // collect dirs and files first
 535             if(is_dir("$path/$f") && !is_link("$path/$f")) {
 536                 $dirs[] = "$path/$f";
 537             } else if($removefiles) {
 538                 $files[] = "$path/$f";
 539             } else {
 540                 return false; // abort when non empty
 541             }
 542
 543         }
 544         closedir($dh);
 545
 546         // now traverse into  directories first
 547         foreach($dirs as $dir) {
 548             if(!io_rmdir($dir, $removefiles)) return false; // abort on any error
 549         }
 550
 551         // now delete files
 552         foreach($files as $file) {
 553             if(!@unlink($file)) return false; //abort on any error
 554         }
 555
 556         // remove self
 557         return @rmdir($path);
 558     } else if($removefiles) {
 559         return @unlink($path);
 560     }
 561     return false;
 562 }
 563
 564 /**
 565  * Creates a unique temporary directory and returns
 566  * its path.
 567  *
 568  * @author Michael Klier <chi@chimeric.de>
 569  *
 570  * @return false|string path to new directory or false
 571  */
 572 function io_mktmpdir() {
 573     global $conf;
 574
 575     $base = $conf['tmpdir'];
 576     $dir  = md5(uniqid(mt_rand(), true));
 577     $tmpdir = $base.'/'.$dir;
 578
 579     if(io_mkdir_p($tmpdir)) {
 580         return($tmpdir);
 581     } else {
 582         return false;
 583     }
 584 }
 585
 586 /**
 587  * downloads a file from the net and saves it
 588  *
 589  * if $useAttachment is false,
 590  * - $file is the full filename to save the file, incl. path
 591  * - if successful will return true, false otherwise
 592  *
 593  * if $useAttachment is true,
 594  * - $file is the directory where the file should be saved
 595  * - if successful will return the name used for the saved file, false otherwise
 596  *
 597  * @author Andreas Gohr <andi@splitbrain.org>
 598  * @author Chris Smith <chris@jalakai.co.uk>
 599  *
 600  * @param string $url           url to download
 601  * @param string $file          path to file or directory where to save
 602  * @param bool   $useAttachment true: try to use name of download, uses otherwise $defaultName
 603  *                              false: uses $file as path to file
 604  * @param string $defaultName   fallback for if using $useAttachment
 605  * @param int    $maxSize       maximum file size
 606  * @return bool|string          if failed false, otherwise true or the name of the file in the given dir
 607  */
 608 function io_download($url,$file,$useAttachment=false,$defaultName='',$maxSize=2097152){
 609     global $conf;
 610     $http = new DokuHTTPClient();
 611     $http->max_bodysize = $maxSize;
 612     $http->timeout = 25; //max. 25 sec
 613     $http->keep_alive = false; // we do single ops here, no need for keep-alive
 614
 615     $data = $http->get($url);
 616     if(!$data) return false;
 617
 618     $name = '';
 619     if ($useAttachment) {
 620         if (isset($http->resp_headers['content-disposition'])) {
 621             $content_disposition = $http->resp_headers['content-disposition'];
 622             $match=array();
 623             if (is_string($content_disposition) &&
 624                     preg_match('/attachment;\s*filename\s*=\s*"([^"]*)"/i', $content_disposition, $match)) {
 625
 626                 $name = \dokuwiki\Utf8\PhpString::basename($match[1]);
 627             }
 628
 629         }
 630
 631         if (!$name) {
 632             if (!$defaultName) return false;
 633             $name = $defaultName;
 634         }
 635
 636         $file = $file.$name;
 637     }
 638
 639     $fileexists = file_exists($file);
 640     $fp = @fopen($file,"w");
 641     if(!$fp) return false;
 642     fwrite($fp,$data);
 643     fclose($fp);
 644     if(!$fileexists and $conf['fperm']) chmod($file, $conf['fperm']);
 645     if ($useAttachment) return $name;
 646     return true;
 647 }
 648
 649 /**
 650  * Windows compatible rename
 651  *
 652  * rename() can not overwrite existing files on Windows
 653  * this function will use copy/unlink instead
 654  *
 655  * @param string $from
 656  * @param string $to
 657  * @return bool succes or fail
 658  */
 659 function io_rename($from,$to){
 660     global $conf;
 661     if(!@rename($from,$to)){
 662         if(@copy($from,$to)){
 663             if($conf['fperm']) chmod($to, $conf['fperm']);
 664             @unlink($from);
 665             return true;
 666         }
 667         return false;
 668     }
 669     return true;
 670 }
 671
 672 /**
 673  * Runs an external command with input and output pipes.
 674  * Returns the exit code from the process.
 675  *
 676  * @author Tom N Harris <tnharris@whoopdedo.org>
 677  *
 678  * @param string $cmd
 679  * @param string $input  input pipe
 680  * @param string $output output pipe
 681  * @return int exit code from process
 682  */
 683 function io_exec($cmd, $input, &$output){
 684     $descspec = array(
 685             0=>array("pipe","r"),
 686             1=>array("pipe","w"),
 687             2=>array("pipe","w"));
 688     $ph = proc_open($cmd, $descspec, $pipes);
 689     if(!$ph) return -1;
 690     fclose($pipes[2]); // ignore stderr
 691     fwrite($pipes[0], $input);
 692     fclose($pipes[0]);
 693     $output = stream_get_contents($pipes[1]);
 694     fclose($pipes[1]);
 695     return proc_close($ph);
 696 }
 697
 698 /**
 699  * Search a file for matching lines
 700  *
 701  * This is probably not faster than file()+preg_grep() but less
 702  * memory intensive because not the whole file needs to be loaded
 703  * at once.
 704  *
 705  * @author Andreas Gohr <andi@splitbrain.org>
 706  * @param  string $file    The file to search
 707  * @param  string $pattern PCRE pattern
 708  * @param  int    $max     How many lines to return (0 for all)
 709  * @param  bool   $backref When true returns array with backreferences instead of lines
 710  * @return array matching lines or backref, false on error
 711  */
 712 function io_grep($file,$pattern,$max=0,$backref=false){
 713     $fh = @fopen($file,'r');
 714     if(!$fh) return false;
 715     $matches = array();
 716
 717     $cnt  = 0;
 718     $line = '';
 719     while (!feof($fh)) {
 720         $line .= fgets($fh, 4096);  // read full line
 721         if(substr($line,-1) != "\n") continue;
 722
 723         // check if line matches
 724         if(preg_match($pattern,$line,$match)){
 725             if($backref){
 726                 $matches[] = $match;
 727             }else{
 728                 $matches[] = $line;
 729             }
 730             $cnt++;
 731         }
 732         if($max && $max == $cnt) break;
 733         $line = '';
 734     }
 735     fclose($fh);
 736     return $matches;
 737 }
 738
 739
 740 /**
 741  * Get size of contents of a file, for a compressed file the uncompressed size
 742  * Warning: reading uncompressed size of content of bz-files requires uncompressing
 743  *
 744  * @author  Gerrit Uitslag <klapinklapin@gmail.com>
 745  *
 746  * @param string $file filename path to file
 747  * @return int size of file
 748  */
 749 function io_getSizeFile($file) {
 750     if (!file_exists($file)) return 0;
 751
 752     if(substr($file,-3) == '.gz'){
 753         $fp = @fopen($file, "rb");
 754         if($fp === false) return 0;
 755
 756         fseek($fp, -4, SEEK_END);
 757         $buffer = fread($fp, 4);
 758         fclose($fp);
 759         $array = unpack("V", $buffer);
 760         $uncompressedsize = end($array);
 761     }else if(substr($file,-4) == '.bz2'){
 762         if(!DOKU_HAS_BZIP) return 0;
 763
 764         $bz = bzopen($file,"r");
 765         if($bz === false) return 0;
 766
 767         $uncompressedsize = 0;
 768         while (!feof($bz)) {
 769             //8192 seems to be the maximum buffersize?
 770             $buffer = bzread($bz,8192);
 771             if(($buffer === false) || (bzerrno($bz) !== 0)) {
 772                 return 0;
 773             }
 774             $uncompressedsize += strlen($buffer);
 775         }
 776     }else{
 777         $uncompressedsize = filesize($file);
 778     }
 779
 780     return $uncompressedsize;
 781  }