Extension:FileIndexer

(Redirected from Manual:Uploaded document searching)

Jump to: navigation, search

This extension requires patches to core MediaWiki code when used with MediaWiki 1.8 and before. Extensions implemented using patches may be disabled by or interfere with upgrades and security patches. If you would like to use this extension, we strongly recommend you upgrade to a version later than 1.8. Alternatively, if a suitable alternative without a patch is available, we recommend you use that extension instead.

<?php
$wgHooks['UploadForm:BeforeProcessing'][] = 'ScanFileForIndex';
 
//Testing if these commands are available
if($wgFileIndexerDebug==true)
{
	isCommandPresent("/usr/bin/pdftotext");
	isCommandPresent("/usr/bin/iconv");
	isCommandPresent("/usr/bin/antiword");
	isCommandPresent("/usr/bin/xls2csv");
	isCommandPresent("/usr/bin/catppt");
	isCommandPresent("/usr/bin/strings");
}
 
function ScanFileForIndex($uploadFormObj) 
{
 
         $NewDesc = '';
         $RemoveTags = false;  //remove HTML-Tags created during conversion?
 
         $extension = substr(strrchr($uploadFormObj->mDesiredDestName, '.'),1); //extract the extension of the destination filename
 
          switch(strtolower($extension)) //methods for text extraction 
          {
                case "pdf": 
                {                 
                        //using XPDF and iconv for conversion purposes
                        $toexec = "/usr/bin/pdftotext  -raw -nopgbrk " . $uploadFormObj->mTempPath . " -";
                        $toexec.="| /usr/bin/iconv -f ISO-8859-1 -t UTF-8";
                        break;
                }
 
                case "dot": {}
                case "doc": 
                {
                        //using antiword
                        $toexec = "/usr/bin/antiword -s ".$uploadFormObj->mTempPath;
                        break;
                }
 
                case "xls":
                {
                        $toexec = "/usr/bin/xls2csv ".$uploadFormObj->mTempPath;
                        break;
                }
 
                case "ppt":
                {
                        $toexec = "/usr/bin/catppt ".$uploadFormObj->mTempPath; 
                        break;
                }
 
                case "rtf": # any file extension with text in it will be okay here
                {
                        $toexec = "/usr/bin/strings ".$uploadFormObj->mTempPath; # string's output isn't neat, but it works.
                        break;
                }
 
                //OpenOffice.org documents
                case "ods": {}
                case "odp": {}
                case "odg": {}
                case "odt":
                {
                        $toexec = "unzip -p " . $uploadFormObj->mTempPath . " content.xml";
                        $RemoveTags = true;
                        break;
                }
        }
        if ($toexec != "")
        {
                exec($toexec, $DocText);
                $NewDesc = $uploadFormObj->mComment . "\r\n" . "<!-- ";
                foreach ($DocText as $DocLine) 
                {
                        if($RemoveTags == false)
                        {
                                $NewDesc .= "\r\n" . str_replace("-->","",$DocLine);
                        }
                        else
                        {
                                $NewDesc .= "\r\n" . strip_tags(str_replace("-->","",$DocLine));
                        }
                }
                $NewDesc .= "\r\n" . " -->";
                $uploadFormObj->mComment = $NewDesc;
        }
        return $uploadFormObj;
}
 
function isCommandPresent($command)
{
	if(file_exists($command)==false)
	{
		//extract the command from the path
		$lastSlash = strrpos($command, '/');
		if($lastSlash!='')
		{
			$commandWithoutSlashes = substr($command, $lastSlash+1);
		}
		else
		{
			$commandWithoutSlashes = $command;
		}
 
		$toexec = "whereis $commandWithoutSlashes";
		//lookup the command
		exec($toexec, $whereis);
		echo "FileIndexer: The file $command is missing ... whereis result: $whereis[0] <br>";
	}
}
 
/**
  * Add extension information to Special:Version
 */
$wgExtensionCredits['other'][] = array(
        'name' => 'FileIndexer',
        'author' => 'MHart and Flominator',
        'description' => 'makes uploaded documents searchable',
        'url' => 'http://www.mediawiki.org/wiki/Extension:FileIndexer'
        );

<?php
$wgHooks['UploadForm:BeforeProcessing'][] = 'ScanFileForIndex';
 
 
//Testing if these commands are available
if($wgFileIndexerDebug==true)
{
	isCommandPresent("/usr/bin/pdftotext");
	isCommandPresent("/usr/bin/iconv");
	isCommandPresent("/usr/bin/antiword");
	isCommandPresent("/usr/bin/xls2csv");
	isCommandPresent("/usr/bin/catppt");
	isCommandPresent("/usr/bin/strings");
}
 
function ScanFileForIndex($uploadFormObj) 
{
	 $NewDesc = '';
	 $RemoveTags = false;  //remove HTML-Tags created during conversion?
 
	 $extension = substr(strrchr($uploadFormObj->mDestFile, '.'),1); //extract the extension of the destination filename
 
	  switch(strtolower($extension)) //methods for text extraction 
	  {
		case "pdf": 
		{
			//using XPDF and iconv for conversion purposes
			$toexec = "/usr/local/bin/pdftotext  -raw -nopgbrk " . $uploadFormObj->mUploadTempName . " -";
			// Alternative: $toexec = "/usr/bin/pdftotext  -raw -nopgbrk " . $uploadFormObj->mUploadTempName . " -";
			$toexec.="| iconv -f ISO-8859-1 -t UTF-8";
			break;
		}
 
		case "dot": {}
		case "doc": 
		{
			//using antiword 
			$toexec = "/usr/bin/antiword -s ".$uploadFormObj->mUploadTempName;
			break;
		}
 
		case "xls":
		{
			$toexec = "/usr/bin/xls2csv ".$uploadFormObj->mUploadTempName;
			break;
		}
 
		case "ppt":
		{
			$toexec = "/usr/bin/catppt ".$uploadFormObj->mUploadTempName; 
			break;
		}
 
		case "rtf": # any file extension with text in it will be okay here
		{
			$toexec = "/usr/bin/strings ".$uploadFormObj->mUploadTempName; # string's output isn't neat, but it works.
			break;
		}
 
		//OpenOffice.org documents
		case "ods": {}
		case "odp": {}
		case "odg": {}
		case "odt":
		{
			$toexec = "unzip -p " . $uploadFormObj->mUploadTempName . " content.xml";
			$RemoveTags = true;
			break;
		}
	}
	if ($toexec != "")
	{
		exec($toexec, $DocText);
		$NewDesc = $uploadFormObj->mUploadDescription . "\r\n" . "<!-- ";
		foreach ($DocText as $DocLine) 
		{
			if($RemoveTags == false)
			{
				$NewDesc .= "\r\n" . str_replace("-->","",$DocLine);
			}
			else
			{
				$NewDesc .= "\r\n" . strip_tags(str_replace("-->","",$DocLine));
			}
		}
		$NewDesc .= "\r\n" . " -->";
		$uploadFormObj->mUploadDescription = $NewDesc;
	}
	return $uploadFormObj;
}
 
function isCommandPresent($command)
{
	if(file_exists($command)==false)
	{
		//extract the command from the path
		$lastSlash = strrpos($command, '/');
		if($lastSlash!='')
		{
			$commandWithoutSlashes = substr($command, $lastSlash+1);
		}
		else
		{
			$commandWithoutSlashes = $command;
		}
 
		$toexec = "whereis $commandWithoutSlashes";
		//lookup the command
		exec($toexec, $whereis);
		echo "FileIndexer: The file $command is missing ... whereis result: $whereis[0] <br>";
	}
}
 
/**
  * Add extension information to Special:Version
 */
$wgExtensionCredits['other'][] = array(
	'name' => 'FileIndexer',
	'author' => 'MHart and Flominator',
	'description' => 'makes uploaded documents searchable',
	'url' => 'http://www.mediawiki.org/wiki/Extension:FileIndexer'
	);

	function processUpload() {
		global $wgUser, $wgOut;
 
		if( !wfRunHooks( 'UploadForm:BeforeProcessing', array( &$this ) ) )
		{
			wfDebug( "Hook 'UploadForm:BeforeProcessing' broke processing the file." );
			return false;
		}
 
		/* Check for PHP error if any, requires php 4.2 or newer */

#Makes uploaded documents searchable
include("extensions/FileIndexer/FileIndexer.php");

apt-get install <toolname>

$wgFileIndexerDebug = true;