Project

General

Profile

1
<?php
2
/**
3
 * @category        ISTeasy
4
 * @package         DatabaseSearchEngine 1
5
 * @author          Werner von der Decken
6
 * @copyright       2011, ISTeasy-project
7
 * @license         http://www.gnu.org/licenses/gpl.html
8
 * @version         $Id: DseTwo.php 2070 2014-01-03 01:21:42Z darkviper $
9
 * @filesource		$HeadURL: svn://isteam.dynxs.de/wb-archiv/branches/2.8.x/wb/framework/DseTwo.php $
10
 * @description     Searchengine to browse whoole database for text.
11
 *                  Black- or whitelist is possible
12
 *                  min requirements: PHP 5.2.2, mySQL 5.1
13
 *                  this is a authorisised GPL-lizensed derivate from the original
14
 *                  ISTeasy class DseOne which is available under a cc-by-sa-3.0 license
15
*/
16
/* -------------------------------------------------------- */
17
// Must include code to stop this file being accessed directly
18
if(!defined('WB_PATH')) {
19
	require_once(dirname(__FILE__).'/globalExceptionHandler.php');
20
	throw new IllegalFileException();
21
}
22
/* -------------------------------------------------------- */
23

    
24
class DseTwo {
25

    
26
	const USE_ALL       = 0;
27
	const USE_BLACKLIST = 1;
28
	const USE_WHITELIST = 2;
29

    
30
	const RETURN_UNUSED = 0;
31
	const RETURN_USED   = 1;
32
	/**
33
	 *
34
	 * @var object database 
35
	 */
36
	private $_db;
37
	/**
38
	 *
39
	 * @var string prefix of tables to search for
40
	 */
41
	private $_TablePrefix;
42
	/**
43
	 *
44
	 * @var string name of the database
45
	 */
46
	private $_db_name;
47
	/**
48
	 *
49
	 * @var array list of unneeded tables.fields
50
	 */
51
	private $_ControllList;
52
	private $_ControllListTyp;
53
	private $_ControllListTypen = array('All','BlackList','WhiteList');
54

    
55
	private $_Queries;
56
	private $_BasePath = '';
57
	private $_CachePath = '';
58
	private $_TCacheFile = '';
59
	private $_DCachePrefix = '';
60
	private $_bUseCache = true;
61
	/**
62
	 *
63
	 * @param object $database global database object
64
	 */
65
	public function __construct()
66
	{
67
		$this->_ControllList = array();
68
		$this->_TCacheFile = 'Ie'.__CLASS__.'CacheTables';
69
		$this->_DCachePrefix = 'Ie'.__CLASS__.'CacheDir';
70
		$this->_Queries = array();
71
	}
72
	/**
73
	 *
74
	 * @param string $name name of the property
75
	 *        (db_handle, db_name, table_prefix, base_dir, cache_dir, use_cache)
76
	 * @param mixed $value value of the property
77
	 */
78
	public function  __set($name, $value) {
79

    
80
		switch(strtolower($name)):
81
			case 'db_handle':
82
				if($value) { $this->_db = $value; }
83
				break;
84
			case 'db_name':
85
				if($value != '') { $this->_db_name = $value; }
86
				break;
87
			case 'table_prefix':
88
				if($value != '') { $this->_TablePrefix = $value; }
89
				break;
90
			case 'base_dir':
91
				if($value != '') {
92
					$this->_BasePath = rtrim(str_replace('\\', '/', $value) , '/');
93
				}
94
				break;
95
			case 'cache_dir':
96
				$value = rtrim(str_replace('\\', '/', $value) , '/');
97
				if(!is_dir($value)) {
98
					if(!mkdir($value, 0777, true)) {
99
						$this->_CachePath = '';
100
						$this->_bUseCache = false;
101
						break;
102
					}
103
				}
104
				if(is_writable($value)) {
105
					$this->_CachePath = $value;
106
					$this->_bUseCache = true;
107
				}else {
108
					$this->_CachePath = '';
109
					$this->_bUseCache = false;
110
				}
111
				break;
112
			default:
113
				throw new InvalidArgumentException( __CLASS__.'::'.$name );
114
				break;
115
		endswitch;
116
	}
117

    
118
	/**
119
	 * delete all table cache files
120
	 */
121
	public function clearCache()
122
	{
123
		foreach($this->_ControllListTypen as $type) {
124
			$cFile = $this->_CachePath.'/'.$this->_TCacheFile.$type;
125
			if(file_exists($cFile)) { @unlink($cFile); }
126
		}
127
	}
128
	/**
129
	 *
130
	 * @param string $blacklist path/filename of the blacklist
131
	 * @param int $type const USE_NO_LIST / USE_BLACKLIST / USE_WHITELIST
132
	 * @return bool false if no or empty list is available
133
	 */
134
	public function addControllList($sControllList, $type = self::USE_BLACKLIST)
135
	{
136
		$this->_ControllList = array();
137
		$this->_ControllListTyp = $type;
138
		if(is_readable($sControllList)) {
139
			if(($list = file($sControllList, FILE_IGNORE_NEW_LINES|FILE_SKIP_EMPTY_LINES)) !== false)
140
			{
141
				$list = preg_grep('/^\s*?[^#;]/', $list);
142
				$this->_ControllList = preg_replace('/^\s*?(.*)\s*?$/', $this->_TablePrefix.'$1', $list);
143
				unset($list);
144
			}
145
		}else {
146
			$this->_ControllListTyp = self::USE_ALL;
147
		}
148
		if(($type == self::USE_BLACKLIST) && (sizeof($this->_ControllList) > 0)) {
149
			$this->_ControllListTyp = self::USE_ALL;
150
		}
151
		return (sizeof($this->_ControllList) > 0);
152
	}
153
	/**
154
	 *
155
	 * @param string $sDirToSearch directory to scan (relative to base_dir)
156
	 * @param integer $bRetunMode select matching or unmatching files
157
	 * @return array list of matching files
158
	 */
159
	public function getMatchesFromDir($sDirToSearch, $bRetunMode = self::RETURN_USED)
160
	{
161
		$aResultFileList = array();
162
		$aNewFileList = array();
163
		$sDirToSearch = trim(str_replace('\\', '/', $sDirToSearch) , '/');
164
		$sPathToSearch = $this->_BasePath.'/'.$sDirToSearch;
165
		$sCacheFile = $this->_DCachePrefix.$bRetunMode.urlencode('/'.$sDirToSearch);
166
		$sCacheFile = $this->_CachePath.'/'.$sCacheFile;
167
		if(sizeof($this->_Queries) <= 0) { $this->_getTableQueries(); }
168
		// read fileList from directory
169
		try{
170
			foreach( new DirectoryIterator($sPathToSearch) as $fileinfo ) {
171
			// at first collect all files from target directory
172
				$fileName = $fileinfo->getFilename();
173
				if(($fileinfo->isFile()) &&
174
				   (!$fileinfo->isDot()) &&
175
				   ($fileinfo->getFilename() != 'index.php')) {
176
				   $aNewFileList[] = $fileinfo->getFilename();
177
				}
178
			}
179
		}catch(UnexpectedValueException $e) {}
180
		// make checksum of current directory
181
		$bCacheValid = false;
182
		if($this->_bUseCache) {
183
			$checkSum = crc32(serialize($aNewFileList));
184
			if(is_readable($sCacheFile)){
185
			// read cachefile if available
186
				$aResultFileList = unserialize(file_get_contents($sCacheFile));
187
				if($checkSum == array_shift($aResultFileList)) {
188
				// compare new checksum against checksum from cachefile
189
					$bCacheValid = true;
190
				}
191
			}
192
		}
193
		if(!$bCacheValid) {
194
		// skip this loop if valid cache is available
195
			$aResultFileList = array();
196
			while (list( , $sFilename) = each($aNewFileList)) {
197
				// iterate all tables and search for filename
198
				if( $this->_getMatch($sDirToSearch.'/'.$sFilename) !== false) {
199
					if($bRetunMode == self::RETURN_USED) { $aResultFileList[] = $sFilename; }
200
				}else {
201
					if($bRetunMode == self::RETURN_UNUSED) { $aResultFileList[] = $sFilename; }
202
				}
203
			}
204
			// calculate new checksum
205
			$newCheckSum = crc32(serialize($aResultFileList));
206
			// add checksum to array
207
			array_unshift($aResultFileList,  $newCheckSum);
208
			// try to write serialized array into new cachefile
209
			if(file_put_contents($sCacheFile, serialize($aResultFileList)) === false) {
210
				throw new RuntimeException();
211
			}
212
			// remove checksum again
213
			array_shift($aResultFileList);
214
		}
215
		unset($aNewFileList);
216
		return $aResultFileList;
217
	}
218
	/**
219
	 *
220
	 * @param <type> $sFilename
221
	 * @return bool true if file found in db
222
	 */
223
	private function _getMatch($sFilename)
224
	{
225
		$result = 0;
226
		$sFilename = str_replace('_', '\_', $sFilename);
227
		$sSearch = '%'.str_replace('/', '_', $sFilename).'%';
228
		while (list( , $sQuery) = each($this->_Queries)) {
229
			$sql = sprintf($sQuery, $sSearch);
230
			if( ($res = mysql_query($sql, $this->_db)) ) {
231
				if( ($result = intval(mysql_fetch_array($res))) > 0 )  { break; }
232
			}
233
		}
234
		return ($result != 0);
235
	}
236
	/**
237
	 *
238
	 */
239
	private function _getTableQueries()
240
	{
241
		if($this->_bUseCache) {
242
		// try to read queries from cace
243
			$sCacheFile = $this->_CachePath.'/'.$this->_TCacheFile.$this->_ControllListTypen[$this->_ControllListTyp];
244
			try {
245
				if(is_readable($sCacheFile)) {
246
					$this->_Queries = unserialize(file_get_contents($sCacheFile));
247
				}
248
			}catch(Exception $e) {
249
				$this->_Queries = array();
250
			}
251
		}
252
		if(sizeof($this->_Queries) > 0) { return; } // queries alreade loaded from cache
253
		$TP = str_replace('_','\_', $this->_TablePrefix);
254
		$sql  = 'SELECT TABLE_NAME `table`, COLUMN_NAME `column` ';
255
		$sql .= 'FROM INFORMATION_SCHEMA.COLUMNS ';
256
		$sql .= 'WHERE `table_schema` = \''.$this->_db_name.'\' AND ';
257
		$sql .=        '`table_name` LIKE \''.$TP.'%\' AND ';
258
		$sql .=        '(`data_type` LIKE \'%text\' OR ';
259
		$sql .=           '(`data_type` = \'varchar\' AND `character_maximum_length` > 20)';
260
		$sql .=        ')' ;
261
		$sql .= 'ORDER BY `table`, `column`';
262
		if(($res = mysql_query($sql, $this->_db))) {
263
			$lastTable = '';
264
			$aOrStatements = array();
265
			$sPrefix = '';
266
			while($rec = mysql_fetch_assoc($res))
267
			{ // loop through all found tables/fields
268
				$sTableColumn = $rec['table'].'.'.$rec['column'];
269
				switch($this->_ControllListTyp):
270
				// test against controll list
271
					case self::USE_BLACKLIST:
272
						$needRecord = true;
273
						if(in_array($rec['table'], $this->_ControllList) ||
274
						   in_array($sTableColumn, $this->_ControllList))
275
						{
276
							$needRecord = false;
277
						}
278
						break;
279
					case self::USE_WHITELIST:
280
						$needRecord = false;
281
						if(in_array($rec['table'], $this->_ControllList) ||
282
						   in_array($sTableColumn, $this->_ControllList))
283
						{
284
							$needRecord = true;
285
						}
286
						break;
287
					default: // self::USE_ALL
288
						$needRecord = true;
289
						break;
290
				endswitch;
291
				if($needRecord) {
292
					if($lastTable != $rec['table']) {
293
						if(sizeof($aOrStatements)!= 0){
294
						// close previous table
295
							$this->_Queries[] = $sPrefix.implode(') OR (', $aOrStatements).')';
296
						}
297
					// start a new table
298
						$sPrefix = 'SELECT COUNT(*) `count` FROM `'.$rec['table'].'` WHERE( ';
299
						$aOrStatements = array();
300
						$lastTable = $rec['table'];
301
					}
302
					// add table.column to query
303
					$aOrStatements[] = '`'.$rec['table'].'`.`'.$rec['column'].'` LIKE \'%1$s\'';
304
				}
305
			}
306
			if(sizeof($aOrStatements)!= 0){
307
			// close last table
308
				$this->_Queries[] = $sPrefix.implode(') OR (', $aOrStatements).')';
309
			}
310
			mysql_free_result($res);
311
		}
312
		if($this->_bUseCache) {
313
		// try to write queries into the cache
314
			if(file_put_contents($sCacheFile, serialize($this->_Queries)) === false) {
315
				throw new RuntimeException('unable to write file ['.$sCacheFile.']');
316
			}
317
		}
318
	}
319

    
320
}
321
?>
(3-3/40)