class.external_parser.php

Typo3, 开源里边最强大的

<?php
/***************************************************************
*  Copyright notice
*
*  (c) 2001-2006 Kasper Skaarhoj ([e-mail])
*  All rights reserved
*
*  This script is part of the TYPO3 project. The TYPO3 project is
*  free software; you can redistribute it and/or modify
*  it under the terms of the GNU General Public License as published by
*  the Free Software Foundation; either version 2 of the License, or
*  (at your option) any later version.
*
*  The GNU General Public License can be found at
*  http://www.gnu.org/copyleft/gpl.html.
*  A copy is found in the textfile GPL.txt and important notices to the license
*  from the author is found in LICENSE.txt distributed with these scripts.
*
*
*  This script is distributed in the hope that it will be useful,
*  but WITHOUT ANY WARRANTY; without even the implied warranty of
*  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
*  GNU General Public License for more details.
*
*  This copyright notice MUST APPEAR in all copies of the script!
***************************************************************/
/**
 * External standard parsers for indexed_search
 *
 * @author	Kasper Sk錼h鴍 <[e-mail]>
 * @coauthor	Olivier Simah <[e-mail]>
 */
/**
 * [CLASS/FUNCTION INDEX of SCRIPT]
 *
 *
 *
 *   75: class tx_indexed_search_extparse
 *   94:     function initParser($extension)
 *  214:     function softInit($extension)
 *  247:     function searchTypeMediaTitle($extension)
 *  323:     function isMultiplePageExtension($extension)
 *
 *              SECTION: Reading documents (for parsing)
 *  354:     function readFileContent($ext,$absFile,$cPKey)
 *  521:     function fileContentParts($ext,$absFile)
 *  560:     function splitPdfInfo($pdfInfoArray)
 *  579:     function removeEndJunk($string)
 *
 *              SECTION: Backend analyzer
 *  606:     function getIcon($extension)
 *
 * TOTAL FUNCTIONS: 9
 * (This index is automatically created/updated by the extension "extdeveval")
 *
 */










/**
 * External standard parsers for indexed_search
 * MUST RETURN utf-8 content!
 *
 * @author	Kasper Skaarhoj <[e-mail]>
 * @package TYPO3
 * @subpackage tx_indexedsearch
 */
class tx_indexed_search_extparse {

		// This value is also overridden from config.
	var $pdf_mode = -20;	// zero: whole PDF file is indexed in one. positive value: Indicates number of pages at a time, eg. "5" would means 1-5,6-10,.... Negative integer would indicate (abs value) number of groups. Eg "3" groups of 10 pages would be 1-4,5-8,9-10

		// This array is configured in initialization:
	var $app = array();
	var $ext2itemtype_map = array();
	var $supportedExtensions = array();

	var $pObj;		// Reference to parent object (indexer class)


	/**
	 * Initialize external parser for parsing content.
	 *
	 * @param	string		File extension
	 * @return	boolean		Returns true if extension is supported/enabled, otherwise false.
	 */
	function initParser($extension)	{

			// Then read indexer-config and set if appropriate:
		$indexerConfig = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search']);

			// If windows, apply extension to tool name:
		$exe = (TYPO3_OS == 'WIN') ? '.exe' : ''; // lg
		$extOK = FALSE;
		$mainExtension = '';

			// Ignore extensions
		$ignoreExtensions = t3lib_div::trimExplode(',', strtolower($indexerConfig['ignoreExtensions']),1);
		if (in_array($extension, $ignoreExtensions))	{
			$this->pObj->log_setTSlogMessage('Extension "'.$extension.'" was set to be ignored.',1);
			return FALSE;
		}

			// Switch on file extension:
		switch($extension)	{
			case 'pdf':
					// PDF
				if ($indexerConfig['pdftools'])	{
					$pdfPath = ereg_replace("\/$",'',$indexerConfig['pdftools']).'/';
					if (ini_get('safe_mode') || (@is_file($pdfPath.'pdftotext'.$exe) && @is_file($pdfPath.'pdfinfo'.$exe)))	{
						$this->app['pdfinfo'] = $pdfPath.'pdfinfo'.$exe;
						$this->app['pdftotext'] = $pdfPath.'pdftotext'.$exe;
							// PDF mode:
						$this->pdf_mode = t3lib_div::intInRange($indexerConfig['pdf_mode'],-100,100);
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("PDF tools was not found in paths '".$pdfPath."pdftotext' and/or '".$pdfPath."pdfinfo'",3);
				} else $this->pObj->log_setTSlogMessage('PDF tools disabled',1);
			break;
			case 'doc':
					// Catdoc
				if ($indexerConfig['catdoc'])	{
					$catdocPath = ereg_replace("\/$",'',$indexerConfig['catdoc']).'/';
					if (ini_get('safe_mode') || @is_file($catdocPath.'catdoc'.$exe))	{
						$this->app['catdoc'] = $catdocPath.'catdoc'.$exe;
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("'catdoc' tool for reading Word-files was not found in path '".$catdocPath."catdoc'",3);
				} else $this->pObj->log_setTSlogMessage('catdoc tools (Word-files) disabled',1);
			break;
			case 'pps':		// MS PowerPoint(?)
			case 'ppt':		// MS PowerPoint
					// ppthtml
				if ($indexerConfig['ppthtml'])	{
					$ppthtmlPath = ereg_replace('\/$','',$indexerConfig['ppthtml']).'/';
					if (ini_get('safe_mode') || @is_file($ppthtmlPath.'ppthtml'.$exe)){
						$this->app['ppthtml'] = $ppthtmlPath.'ppthtml'.$exe;
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("'ppthtml' tool for reading Powerpoint-files was not found in path '".$ppthtmlPath."ppthtml'",3);
				} else $this->pObj->log_setTSlogMessage('ppthtml tools (Powerpoint-files) disabled',1);
			break;
			case 'xls':		// MS Excel
					// Xlhtml
				if ($indexerConfig['xlhtml'])	{
					$xlhtmlPath = ereg_replace('\/$','',$indexerConfig['xlhtml']).'/';
					if (ini_get('safe_mode') || @is_file($xlhtmlPath.'xlhtml'.$exe)){
						$this->app['xlhtml'] = $xlhtmlPath.'xlhtml'.$exe;
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("'xlhtml' tool for reading Excel-files was not found in path '".$xlhtmlPath."xlhtml'",3);
				} else $this->pObj->log_setTSlogMessage('xlhtml tools (Excel-files) disabled',1);
			break;
			case 'sxc':		// Open Office Calc.
			case 'sxi':		// Open Office Impress
			case 'sxw':		// Open Office Writer
			case 'ods':		// Oasis OpenDocument Spreadsheet
			case 'odp':		// Oasis OpenDocument Presentation
			case 'odt':		// Oasis OpenDocument Text
				if ($indexerConfig['unzip'])	{
					$unzipPath = preg_replace('/\/$/','',$indexerConfig['unzip']).'/';
					if (ini_get('safe_mode') || @is_file($unzipPath.'unzip'.$exe))	{
						$this->app['unzip'] = $unzipPath.'unzip'.$exe;
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("'unzip' tool for reading OpenOffice.org-files was not found in path '".$unzipPath."unzip'",3);
				} else $this->pObj->log_setTSlogMessage('unzip tool (OpenOffice.org-files) disabled',1);
			break;
			case 'rtf':
					// Catdoc
				if ($indexerConfig['unrtf'])	{
					$unrtfPath = ereg_replace("\/$",'',$indexerConfig['unrtf']).'/';
					if (ini_get('safe_mode') || @is_file($unrtfPath.'unrtf'.$exe))	{
						$this->app['unrtf'] = $unrtfPath.'unrtf'.$exe;
						$extOK = TRUE;
					} else $this->pObj->log_setTSlogMessage("'unrtf' tool for reading RTF-files was not found in path '".$unrtfPath."unrtf'",3);
				} else $this->pObj->log_setTSlogMessage('unrtf tool (RTF-files) disabled',1);
			break;
			case 'txt':		// Raw text
			case 'csv':		// Raw text
			case 'xml':		// PHP strip-tags()
			case 'tif':		// PHP EXIF
				$extOK = TRUE;
			break;
			case 'html':	// PHP strip-tags()
			case 'htm':		// PHP strip-tags()
				$extOK = TRUE;
				$mainExtension = 'html';	// making "html" the common "item_type"
			break;
			case 'jpg':		// PHP EXIF
			case 'jpeg':	// PHP EXIF
				$extOK = TRUE;
				$mainExtension = 'jpeg';	// making "jpeg" the common item_type
			break;
		}

			// If extension was OK:
		if ($extOK)	{
			$this->supportedExtensions[$extension] = TRUE;
			$this->ext2itemtype_map[$extension] = $mainExtension ? $mainExtension : $extension;
			return TRUE;
		}
	}

	/**
	 * Initialize external parser for backend modules
	 * Doesn't evaluate if parser is configured right - more like returning POSSIBLE supported extensions (for showing icons etc) in backend and frontend plugin
	 *
	 * @param	string		File extension to initialize for.
	 * @return	boolean		Returns true if the extension is supported and enabled, otherwise false.
	 */
	function softInit($extension)	{
		switch($extension)	{
			case 'pdf':		// PDF
			case 'doc':		// MS Word files
			case 'pps':		// MS PowerPoint
			case 'ppt':		// MS PowerPoint
			case 'xls':		// MS Excel
			case 'sxc':		// Open Office Calc.
			case 'sxi':		// Open Office Impress
			case 'sxw':		// Open Office Writer
			case 'ods':		// Oasis OpenDocument Spreadsheet
			case 'odp':		// Oasis OpenDocument Presentation
			case 'odt':		// Oasis OpenDocument Text
			case 'rtf':		// RTF documents
			case 'txt':		// ASCII Text documents
			case 'html':	// HTML
			case 'htm':		// HTML
			case 'csv':		// Comma Separated Values
			case 'xml':		// Generic XML
			case 'jpg':		// Jpeg images (EXIF comment)
			case 'jpeg':	// Jpeg images (EXIF comment)
			case 'tif':		// TIF images (EXIF comment)
				return TRUE;
			break;
		}
	}

	/**
	 * Return title of entry in media type selector box.
	 *
	 * @param	string		File extension
	 * @return	string		String with label value of entry in media type search selector box (frontend plugin).
	 */
	function searchTypeMediaTitle($extension)	{

			// Read indexer-config
		$indexerConfig = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search']);

			// Ignore extensions
		$ignoreExtensions = t3lib_div::trimExplode(',', strtolower($indexerConfig['ignoreExtensions']),1);
		if (in_array($extension, $ignoreExtensions))	{
			return FALSE;
		}

			// Switch on file extension:
		switch($extension)	{
			case 'pdf':
					// PDF
				if ($indexerConfig['pdftools'])	{
					return 'PDF';
				}
			break;
			case 'doc':
					// Catdoc
				if ($indexerConfig['catdoc'])	{
					return 'MS Word';
				}
			break;
			case 'pps':		// MS PowerPoint(?)
			case 'ppt':		// MS PowerPoint
					// ppthtml
				if ($indexerConfig['ppthtml'])	{
					return 'MS Powerpoint';
				}
			break;
			case 'xls':		// MS Excel
					// Xlhtml
				if ($indexerConfig['xlhtml'])	{
					return 'MS Excel';
				}
			break;
			case 'sxc':		// Open Office Calc.
			case 'sxi':		// Open Office Impress
			case 'sxw':		// Open Office Writer
			case 'ods':		// Oasis OpenDocument Spreadsheet
			case 'odp':		// Oasis OpenDocument Presentation
			case 'odt':		// Oasis OpenDocument Text
				if ($indexerConfig['unzip'])	{
					return 'Open Office';
				}
			break;
			case 'rtf':
					// Catdoc
				if ($indexerConfig['unrtf'])	{
					return 'RTF';
				}
			break;
			case 'html':	// PHP strip-tags()
			case 'jpeg':	// PHP EXIF
			case 'txt':		// Raw text
			case 'csv':		// Raw text
			case 'xml':		// PHP strip-tags()
			case 'tif':		// PHP EXIF
				return strtoupper($extension);
			break;
				// NO entry (duplicates or blank):
			case 'htm':		// PHP strip-tags()
			case 'jpg':		// PHP EXIF
			default:
			break;
		}
	}