Project

General

Profile

1 552 thorn
<?php
2
3 554 Ruebenwurz
// $Id$
4 552 thorn
5
/*
6
7
 Website Baker Project <http://www.websitebaker.org/>
8 915 Ruebenwurz
 Copyright (C) 2004-2009, Ryan Djurovich
9 552 thorn
10
 Website Baker is free software; you can redistribute it and/or modify
11
 it under the terms of the GNU General Public License as published by
12
 the Free Software Foundation; either version 2 of the License, or
13
 (at your option) any later version.
14
15
 Website Baker is distributed in the hope that it will be useful,
16
 but WITHOUT ANY WARRANTY; without even the implied warranty of
17
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18
 GNU General Public License for more details.
19
20
 You should have received a copy of the GNU General Public License
21
 along with Website Baker; if not, write to the Free Software
22
 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
23
24
*/
25
26
// make the url-string for highlighting
27
function make_url_searchstring($search_match, $search_url_array) {
28
	$link = "";
29
	if ($search_match != 'exact') {
30
		$str = implode(" ", $search_url_array);
31
		$link = "?searchresult=1&amp;sstring=".urlencode($str);
32
	} else {
33 761 thorn
		$str = str_replace(' ', '_', $search_url_array[0]);
34 552 thorn
		$link = "?searchresult=2&amp;sstring=".urlencode($str);
35
	}
36
	return $link;
37
}
38
39
// make date and time for "last modified by... on ..."-string
40
function get_page_modified($page_modified_when) {
41
	global $TEXT;
42
	if($page_modified_when > 0) {
43
		$date = gmdate(DATE_FORMAT, $page_modified_when+TIMEZONE);
44
		$time = gmdate(TIME_FORMAT, $page_modified_when+TIMEZONE);
45
	} else {
46
		$date = $TEXT['UNKNOWN'].' '.$TEXT['DATE'];
47
		$time = $TEXT['UNKNOWN'].' '.$TEXT['TIME'];
48
	}
49
	return array($date, $time);
50
}
51
52
// make username and displayname for "last modified by... on ..."-string
53
function get_page_modified_by($page_modified_by, $users) {
54
	global $TEXT;
55
	if($page_modified_by > 0) {
56
			$username = $users[$page_modified_by]['username'];
57
			$displayname = $users[$page_modified_by]['display_name'];
58
		} else {
59
			$username = "";
60
			$displayname = $TEXT['UNKNOWN'];
61
		}
62
	return array($username, $displayname);
63
}
64
65
// checks if _all_ searchwords matches
66
function is_all_matched($text, $search_words) {
67
	$all_matched = true;
68
	foreach ($search_words AS $word) {
69 881 thorn
		if(!preg_match('/'.$word.'/ui', $text)) {
70 552 thorn
			$all_matched = false;
71
			break;
72
		}
73
	}
74
	return $all_matched;
75
}
76
77
// checks if _any_ of the searchwords matches
78
function is_any_matched($text, $search_words) {
79
	$any_matched = false;
80 761 thorn
	$word = '('.implode('|', $search_words).')';
81 881 thorn
	if(preg_match('/'.$word.'/ui', $text)) {
82 552 thorn
		$any_matched = true;
83
	}
84
	return $any_matched;
85
}
86
87
// collects the matches from text in excerpt_array
88
function get_excerpts($text, $search_words, $max_excerpt_num) {
89
	$match_array = array();
90
	$excerpt_array = array();
91 761 thorn
	$word = '('.implode('|', $search_words).')';
92 989 aldus
93
	//Filter droplets from the page data
94
	preg_match_all('~\[\[(.*?)\]\]~', $text, $matches);
95
	foreach ($matches[1] as $match) {
96
		$text = str_replace('[['.$match.']]', '', $text);
97
	}
98
99 552 thorn
	// Build the regex-string
100 857 thorn
	if(strpos(strtoupper(PHP_OS), 'WIN')===0) { // windows -> see below
101
		$str1=".!?;";
102
		$str2=".!?;";
103
	} else { // linux & Co.
104
		// start-sign: .!?; + INVERTED EXCLAMATION MARK - INVERTED QUESTION MARK - DOUBLE EXCLAMATION MARK - INTERROBANG - EXCLAMATION QUESTION MARK - QUESTION EXCLAMATION MARK - DOUBLE QUESTION MARK - HALFWIDTH IDEOGRAPHIC FULL STOP - IDEOGRAPHIC FULL STOP - IDEOGRAPHIC COMMA
105
		$str1=".!?;"."\xC2\xA1"."\xC2\xBF"."\xE2\x80\xBC"."\xE2\x80\xBD"."\xE2\x81\x89"."\xE2\x81\x88"."\xE2\x81\x87"."\xEF\xBD\xA1"."\xE3\x80\x82"."\xE3\x80\x81";
106
		// stop-sign: .!?; + DOUBLE EXCLAMATION MARK - INTERROBANG - EXCLAMATION QUESTION MARK - QUESTION EXCLAMATION MARK - DOUBLE QUESTION MARK - HALFWIDTH IDEOGRAPHIC FULL STOP - IDEOGRAPHIC FULL STOP - IDEOGRAPHIC COMMA
107
		$str2=".!?;"."\xE2\x80\xBC"."\xE2\x80\xBD"."\xE2\x81\x89"."\xE2\x81\x88"."\xE2\x81\x87"."\xEF\xBD\xA1"."\xE3\x80\x82"."\xE3\x80\x81";
108
	}
109 881 thorn
	$regex='/(?:^|\b|['.$str1.'])([^'.$str1.']{0,200}?'.$word.'[^'.$str2.']{0,200}(?:['.$str2.']|\b|$))/uis';
110 857 thorn
	if(version_compare(PHP_VERSION, '4.3.3', '>=') &&
111 869 thorn
	   strpos(strtoupper(PHP_OS), 'WIN')!==0
112
	) { // this may crash windows server, so skip if on windows
113 761 thorn
		// jump from match to match, get excerpt, stop if $max_excerpt_num is reached
114
		$last_end = 0; $offset = 0;
115 881 thorn
		while(preg_match('/'.$word.'/uis', $text, $match_array, PREG_OFFSET_CAPTURE, $last_end)) {
116 761 thorn
			$offset = ($match_array[0][1]-206 < $last_end)?$last_end:$match_array[0][1]-206;
117
			if(preg_match($regex, $text, $matches, PREG_OFFSET_CAPTURE, $offset)) {
118
				$last_end = $matches[1][1]+strlen($matches[1][0])-1;
119 857 thorn
				if(!preg_match('/\b[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\./', $matches[1][0])) // skip excerpts with email-addresses
120 761 thorn
					$excerpt_array[] = trim($matches[1][0]);
121
				if(count($excerpt_array)>=$max_excerpt_num) {
122
					$excerpt_array = array_unique($excerpt_array);
123
					if(count($excerpt_array) >= $max_excerpt_num)
124
						break;
125
				}
126 869 thorn
			} else { // problem: preg_match failed - can't find a start- or stop-sign
127 761 thorn
				$last_end += 201; // jump forward and try again
128
			}
129 552 thorn
		}
130 869 thorn
	} else { // compatible, but may be very slow with large pages
131 761 thorn
		if(preg_match_all($regex, $text, $match_array)) {
132
			foreach($match_array[1] AS $string) {
133 989 aldus
				if(!preg_match('/\b[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\./', $string))  // skip excerpts with email-addresses
134 761 thorn
					$excerpt_array[] = trim($string);
135 989 aldus
136 761 thorn
			}
137
		}
138 552 thorn
	}
139
	return $excerpt_array;
140
}
141
142
// makes excerpt_array a string ready to print out
143
function prepare_excerpts($excerpt_array, $search_words, $max_excerpt_num) {
144
	// excerpts: text before and after a single excerpt, html-tag for markup
145
	$EXCERPT_BEFORE =       '...&nbsp;';
146
	$EXCERPT_AFTER =        '&nbsp;...<br />';
147
	$EXCERPT_MARKUP_START = '<b>';
148
	$EXCERPT_MARKUP_END =   '</b>';
149
	// remove duplicate matches from $excerpt_array, if any.
150
	$excerpt_array = array_unique($excerpt_array);
151
	// use the first $max_excerpt_num excerpts only
152
	if(count($excerpt_array) > $max_excerpt_num) {
153
		$excerpt_array = array_slice($excerpt_array, 0, $max_excerpt_num);
154
	}
155
	// prepare search-string
156
	$string = "(".implode("|", $search_words).")";
157
	// we want markup on search-results page,
158
	// but we need some 'magic' to prevent <br />, <b>... from being highlighted
159
	$excerpt = '';
160
	foreach($excerpt_array as $str) {
161
		$excerpt .= '#,,#'.preg_replace("/($string)/iu","#,,,,#$1#,,,,,#",$str).'#,,,#';
162
	}
163 881 thorn
	$excerpt = str_replace(array('&','<','>','"','\'',"\xC2\xA0"), array('&amp;','&lt;','&gt;','&quot;','&#039;','&nbsp;'), $excerpt);
164 761 thorn
	$excerpt = str_replace(array('#,,,,#','#,,,,,#'), array($EXCERPT_MARKUP_START,$EXCERPT_MARKUP_END), $excerpt);
165
	$excerpt = str_replace(array('#,,#','#,,,#'), array($EXCERPT_BEFORE,$EXCERPT_AFTER), $excerpt);
166 552 thorn
	// prepare to write out
167
	if(DEFAULT_CHARSET != 'utf-8') {
168
		$excerpt = umlauts_to_entities($excerpt, 'UTF-8');
169
	}
170
	return $excerpt;
171
}
172
173 761 thorn
// work out what the link-anchor should be
174 552 thorn
function make_url_target($page_link_target, $text, $search_words) {
175
	// 1. e.g. $page_link_target=="&monthno=5&year=2007" - module-dependent target. Do nothing.
176
	// 2. $page_link_target=="#!wb_section_..." - the user wants the section-target, so do nothing.
177
	// 3. $page_link_target=="#wb_section_..." - try to find a better target, use the section-target as fallback.
178
	// 4. $page_link_target=="" - do nothing
179 761 thorn
	if(version_compare(PHP_VERSION, '4.3.3', ">=") && substr($page_link_target,0,12)=='#wb_section_') {
180 552 thorn
		$word = '('.implode('|', $search_words).')';
181 881 thorn
		preg_match('/'.$word.'/ui', $text, $match, PREG_OFFSET_CAPTURE);
182 552 thorn
		if($match && is_array($match[0])) {
183 761 thorn
			$x=$match[0][1]; // position of first match
184 552 thorn
			// is there an anchor nearby?
185 869 thorn
			if(preg_match_all('/<(?:[^>]+id|\s*a[^>]+name)\s*=\s*"(.*)"/iU', substr($text,0,$x), $match, PREG_OFFSET_CAPTURE)) {
186 552 thorn
				$anchor='';
187
				foreach($match[1] AS $array) {
188
					if($array[1] > $x) {
189
						break;
190
					}
191
					$anchor = $array[0];
192
				}
193
				if($anchor != '') {
194
					$page_link_target = '#'.$anchor;
195
				}
196
			}
197
		}
198
	}
199
	elseif(substr($page_link_target,0,13)=='#!wb_section_') {
200
		$page_link_target = '#'.substr($page_link_target, 2);
201
	}
202 872 thorn
203
	// since wb 2.7.1 the section-anchor is configurable - SEC_ANCHOR holds the anchor name
204
	if(substr($page_link_target,0,12)=='#wb_section_') {
205
		if(defined('SEC_ANCHOR') && SEC_ANCHOR!='') {
206
			$sec_id = substr($page_link_target, 12);
207
			$page_link_target = '#'.SEC_ANCHOR.$sec_id;
208
		} else { // section-anchors are disabled
209
			$page_link_target = '';
210
		}
211
	}
212
213 552 thorn
	return $page_link_target;
214
}
215
216
// wrapper for compatibility with old print_excerpt()
217
function print_excerpt($page_link, $page_link_target, $page_title, $page_description, $page_modified_when, $page_modified_by, $text, $max_excerpt_num, $func_vars, $pic_link="") {
218
	$mod_vars = array(
219
		'page_link' => $page_link,
220
		'page_link_target' => $page_link_target,
221
		'page_title' => $page_title,
222
		'page_description' => $page_description,
223
		'page_modified_when' => $page_modified_when,
224
		'page_modified_by' => $page_modified_by,
225
		'text' => $text,
226
		'max_excerpt_num' => $max_excerpt_num,
227
		'pic_link' => $pic_link
228
	);
229
	print_excerpt2($mod_vars, $func_vars);
230
}
231
232
/* These functions can be used in module-supplied search_funcs
233
 * -----------------------------------------------------------
234
 * print_excerpt2() - the main-function to use in all search_funcs
235
 * print_excerpt() - wrapper for compatibility-reason. Use print_excerpt2() instead.
236
 * list_files_dirs() - lists all files and dirs below a given directory
237
 * clear_filelist() - keeps only wanted or removes unwanted entries in file-list.
238
 */
239
240
// prints the excerpts for one section
241
function print_excerpt2($mod_vars, $func_vars) {
242
	extract($func_vars, EXTR_PREFIX_ALL, 'func');
243
	extract($mod_vars, EXTR_PREFIX_ALL, 'mod');
244
	global $TEXT;
245
	// check $mod_...vars
246
	if(!isset($mod_page_link))          $mod_page_link = $func_page_link;
247
	if(!isset($mod_page_link_target))   $mod_page_link_target = "";
248
	if(!isset($mod_page_title))         $mod_page_title = $func_page_title;
249
	if(!isset($mod_page_description))   $mod_page_description = $func_page_description;
250
	if(!isset($mod_page_modified_when)) $mod_page_modified_when = $func_page_modified_when;
251
	if(!isset($mod_page_modified_by))   $mod_page_modified_by = $func_page_modified_by;
252
	if(!isset($mod_text))               $mod_text = "";
253
	if(!isset($mod_max_excerpt_num))    $mod_max_excerpt_num = $func_default_max_excerpt;
254
	if(!isset($mod_pic_link))           $mod_pic_link = "";
255
	if(!isset($mod_no_highlight))       $mod_no_highlight = false;
256 761 thorn
	if(!isset($func_enable_flush))      $func_enable_flush = false; // set this in db: wb_search.cfg_enable_flush [READ THE DOC BEFORE]
257 870 thorn
	if(isset($mod_ext_charset)) $mod_ext_charset = strtolower($mod_ext_charset);
258
	else $mod_ext_charset = '';
259
260 552 thorn
	if($mod_text == "") // nothing to do
261
		{ return false; }
262 870 thorn
263 552 thorn
	if($mod_no_highlight) // no highlighting
264 743 thorn
		{ $mod_page_link_target = "&amp;nohighlight=1".$mod_page_link_target; }
265 761 thorn
	// clean the text:
266 881 thorn
	$mod_text = preg_replace('#<(!--.*--|style.*</style|script.*</script)>#iU', ' ', $mod_text);
267
	$mod_text = preg_replace('#<(br( /)?|dt|/dd|/?(h[1-6]|tr|table|p|li|ul|pre|code|div|hr))[^>]*>#i', '.', $mod_text);
268
	$mod_text = preg_replace('/(\v\s?|\s\s)+/', ' ', $mod_text);
269
	$mod_text = preg_replace('/\s\./', '.', $mod_text);
270 870 thorn
	if($mod_ext_charset!='') { // data from external database may have a different charset
271 869 thorn
		require_once(WB_PATH.'/framework/functions-utf8.php');
272 870 thorn
		switch($mod_ext_charset) {
273
		case 'latin1':
274
		case 'cp1252':
275
			$mod_text = charset_to_utf8($mod_text, 'CP1252');
276
			break;
277
		case 'cp1251':
278
			$mod_text = charset_to_utf8($mod_text, 'CP1251');
279
			break;
280
		case 'latin2':
281
			$mod_text = charset_to_utf8($mod_text, 'ISO-8859-2');
282
			break;
283
		case 'hebrew':
284
			$mod_text = charset_to_utf8($mod_text, 'ISO-8859-8');
285
			break;
286
		case 'greek':
287
			$mod_text = charset_to_utf8($mod_text, 'ISO-8859-7');
288
			break;
289
		case 'latin5':
290
			$mod_text = charset_to_utf8($mod_text, 'ISO-8859-9');
291
			break;
292
		case 'latin7':
293
			$mod_text = charset_to_utf8($mod_text, 'ISO-8859-13');
294
			break;
295
		case 'utf8':
296
		default:
297
			$mod_text = charset_to_utf8($mod_text, 'UTF-8');
298
		}
299 869 thorn
	} else {
300 552 thorn
	$mod_text = entities_to_umlauts($mod_text, 'UTF-8');
301 869 thorn
	}
302 777 thorn
	$anchor_text = $mod_text; // make an copy containing html-tags
303 552 thorn
	$mod_text = strip_tags($mod_text);
304 881 thorn
	$mod_text = str_replace(array('&gt;','&lt;','&amp;','&quot;','&#039;','&apos;','&nbsp;'), array('>','<','&','"','\'','\'',"\xC2\xA0"), $mod_text);
305 869 thorn
	$mod_text = '.'.trim($mod_text).'.';
306 552 thorn
	// Do a fast scan over $mod_text first. This will speedup things a lot.
307
	if($func_search_match == 'all') {
308 761 thorn
		if(!is_all_matched($mod_text, $func_search_words))
309 552 thorn
			return false;
310
	}
311
	elseif(!is_any_matched($mod_text, $func_search_words)) {
312
		return false;
313
	}
314 777 thorn
	// search for an better anchor - this have to be done before strip_tags() (may fail if search-string contains <, &, amp, gt, lt, ...)
315
	$anchor =  make_url_target($mod_page_link_target, $anchor_text, $func_search_words);
316 552 thorn
317 761 thorn
	// make the link from $mod_page_link, add anchor
318
	$link = "";
319
	$link = page_link($mod_page_link);
320 833 thorn
	if(strpos($mod_page_link, 'http:')===FALSE)
321
		$link .= make_url_searchstring($func_search_match, $func_search_url_array);
322 761 thorn
	$link .= $anchor;
323
324 552 thorn
	// now get the excerpt
325
	$excerpt = "";
326
	$excerpt_array = array();
327
	if($mod_max_excerpt_num > 0) {
328
		if(!$excerpt_array = get_excerpts($mod_text, $func_search_words, $mod_max_excerpt_num)) {
329
			return false;
330
		}
331
		$excerpt = prepare_excerpts($excerpt_array, $func_search_words, $mod_max_excerpt_num);
332
	}
333
334 761 thorn
	// handle thumbs - to deactivate this look in the module's search.php: $show_thumb (or maybe in the module's settings-page)
335 552 thorn
	if($mod_pic_link != "") {
336 761 thorn
		$excerpt = '<table class="excerpt_thumb" width="100%" cellspacing="0" cellpadding="0" border="0"><tbody><tr><td width="110" valign="top"><a href="'.$link.'"><img src="'.WB_URL.'/'.MEDIA_DIRECTORY.$mod_pic_link.'" alt="" /></a></td><td>'.$excerpt.'</td></tr></tbody></table>';
337 552 thorn
	}
338
339
	// print-out the excerpt
340
	$vars = array();
341
	$values = array();
342
	list($date, $time) = get_page_modified($mod_page_modified_when);
343
	list($username, $displayname) = get_page_modified_by($mod_page_modified_by, $func_users);
344
	$vars = array('[LINK]', '[TITLE]', '[DESCRIPTION]', '[USERNAME]','[DISPLAY_NAME]','[DATE]','[TIME]','[TEXT_LAST_UPDATED_BY]','[TEXT_ON]','[EXCERPT]');
345
	$values = array(
346
		$link,
347
		$mod_page_title,
348
		$mod_page_description,
349
		$username,
350
		$displayname,
351
		$date,
352
		$time,
353
		$TEXT['LAST_UPDATED_BY'],
354
		$TEXT['ON'],
355
		$excerpt
356
	);
357
	echo str_replace($vars, $values, $func_results_loop_string);
358 761 thorn
	if($func_enable_flush) { // ATTN: this will bypass output-filters and may break template-layout or -filters
359
		ob_flush();flush();
360
	}
361 552 thorn
	return true;
362
}
363
364 771 thorn
// list all files and dirs in $dir (recursive), omits '.', '..', and hidden files/dirs
365 552 thorn
// returns an array of two arrays ($files[] and $dirs[]).
366
// usage: list($files,$dirs) = list_files_dirs($directory);
367
//        $depth: get subdirs (true/false)
368
function list_files_dirs($dir, $depth=true, $files=array(), $dirs=array()) {
369
	$dh=opendir($dir);
370
	while(($file = readdir($dh)) !== false) {
371 771 thorn
		if($file{0} == '.' || $file == '..') {
372 552 thorn
			continue;
373
		}
374
		if(is_dir($dir.'/'.$file)) {
375
			if($depth) {
376
				$dirs[] = $dir.'/'.$file;
377
				list($files, $dirs) = list_files_dirs($dir.'/'.$file, $depth, $files, $dirs);
378
			}
379
		} else {
380
			$files[] = $dir.'/'.$file;
381
		}
382
	}
383
	closedir($dh);
384
	natcasesort($files);
385
	natcasesort($dirs);
386
	return(array($files, $dirs));
387
}
388
389
// keeps only wanted entries in array $files. $str have to be an eregi()-compatible regex
390
function clear_filelist($files, $str, $keep=true) {
391 761 thorn
	// options: $keep = true  : remove all non-matching entries
392
	//          $keep = false : remove all matching entries
393 552 thorn
	$c_filelist = array();
394
	if($str == '')
395
		return $files;
396
	foreach($files as $file) {
397
		if($keep) {
398
			if(eregi($str, $file)) {
399
				$c_filelist[] = $file;
400
			}
401
		} else {
402
			if(!eregi($str, $file)) {
403
				$c_filelist[] = $file;
404
			}
405
		}
406
	}
407
	return($c_filelist);
408
}
409
410
?>