|
1
|
<?php
|
|
2
|
/**
|
|
3
|
*
|
|
4
|
* @category frontend
|
|
5
|
* @package search
|
|
6
|
* @author WebsiteBaker Project
|
|
7
|
* @copyright 2004-2009, Ryan Djurovich
|
|
8
|
* @copyright 2009-2011, Website Baker Org. e.V.
|
|
9
|
* @link http://www.websitebaker2.org/
|
|
10
|
* @license http://www.gnu.org/licenses/gpl.html
|
|
11
|
* @platform WebsiteBaker 2.8.x
|
|
12
|
* @requirements PHP 5.2.2 and higher
|
|
13
|
* @version $Id: search_convert.php 1420 2011-01-26 17:43:56Z Luisehahne $
|
|
14
|
* @filesource $HeadURL: svn://isteam.dynxs.de/wb-archiv/branches/2.8.x/wb/search/search_convert.php $
|
|
15
|
* @lastmodified $Date: 2011-01-26 18:43:56 +0100 (Wed, 26 Jan 2011) $
|
|
16
|
*
|
|
17
|
*/
|
|
18
|
|
|
19
|
// Must include code to stop this file being access directly
|
|
20
|
if(defined('WB_PATH') == false) { die("Cannot access this file directly"); }
|
|
21
|
|
|
22
|
if(!isset($search_lang)) $search_lang = LANGUAGE;
|
|
23
|
|
|
24
|
// umlaut to '(upper|lower)' for preg_match()
|
|
25
|
// this is UTF-8-encoded
|
|
26
|
// there is no need for a translation-table anymore since we use u-switch (utf-8) for preg-functions
|
|
27
|
// remember that we use the i-switch, too. [No need for (ae|Ae)]
|
|
28
|
|
|
29
|
$string_ul_umlaut = array();
|
|
30
|
$string_ul_regex = array();
|
|
31
|
|
|
32
|
// but add some national stuff
|
|
33
|
if($search_lang=='DE') { // add special handling for german umlauts (ä==ae, ...)
|
|
34
|
$string_ul_umlaut_add = array(
|
|
35
|
"\xc3\x9f", // german SZ-Ligatur
|
|
36
|
"\xc3\xa4", // german ae
|
|
37
|
"\xc3\xb6", // german oe
|
|
38
|
"\xc3\xbc", // german ue
|
|
39
|
"\xc3\x84", // german Ae
|
|
40
|
"\xc3\x96", // german Oe
|
|
41
|
"\xc3\x9c", // german Ue
|
|
42
|
// these are not that usual
|
|
43
|
"\xEF\xAC\x84", // german ffl-ligatur
|
|
44
|
"ffl", // german ffl-ligatur
|
|
45
|
"\xEF\xAC\x83", // german ffi-ligatur
|
|
46
|
"ffi", // german ffi-ligatur
|
|
47
|
"0xEF\xAC\x80", // german ff-Ligatur
|
|
48
|
"ff", // german ff-Ligatur
|
|
49
|
"\xEF\xAC\x81", // german fi-ligatur
|
|
50
|
"fi", // german fi-ligatur
|
|
51
|
"\xEF\xAC\x82", // german fl-ligatur
|
|
52
|
"fl", // german fl-ligatur
|
|
53
|
"\xEF\xAC\x85", // german st-Ligatur (long s)
|
|
54
|
"st", // german st-Ligatur
|
|
55
|
"\xEF\xAC\x86" // german st-ligatur (round-s)
|
|
56
|
);
|
|
57
|
$string_ul_regex_add = array(
|
|
58
|
"(\xc3\x9f|ss)", // german SZ.Ligatur
|
|
59
|
"(\xc3\xa4|ae)", // german ae
|
|
60
|
"(\xc3\xb6|oe)", // german oe
|
|
61
|
"(\xc3\xbc|ue)", // german ue
|
|
62
|
"(\xc3\x84|Ae)", // german Ae
|
|
63
|
"(\xc3\x96|Oe)", // german Oe
|
|
64
|
"(\xc3\x9c|Ue)", // german Ue
|
|
65
|
// these are not that usual
|
|
66
|
"(\xEF\xAC\x84|ffl)", // german ffl-ligatur
|
|
67
|
"(\xEF\xAC\x84|ffl)", // german ffl-ligatur
|
|
68
|
"(\xEF\xAC\x83|ffi)", // german ffi-ligatur
|
|
69
|
"(\xEF\xAC\x83|ffi)", // german ffi-ligatur
|
|
70
|
"(\xEF\xAC\x80|ff)", // german ff-Ligatur
|
|
71
|
"(\xEF\xAC\x80|ff)", // german ff-Ligatur
|
|
72
|
"(\xEF\xAC\x81|fi)", // german fi-Ligatur
|
|
73
|
"(\xEF\xAC\x81|fi)", // german fi-Ligatur
|
|
74
|
"(\xEF\xAC\x82|fl)", // german fl-ligatur
|
|
75
|
"(\xEF\xAC\x82|fl)", // german fl-ligatur
|
|
76
|
"(\xEF\xAC\x85|st)", // german st-Ligatur (long s)
|
|
77
|
"(\xEF\xAC\x85|st|\xEF\xAC\x86)", // german st-Ligaturs
|
|
78
|
"(\xEF\xAC\x86|st)" // german st-ligatur (round-s)
|
|
79
|
);
|
|
80
|
$string_ul_umlaut = array_merge($string_ul_umlaut_add, $string_ul_umlaut);
|
|
81
|
$string_ul_regex = array_merge($string_ul_regex_add, $string_ul_regex);
|
|
82
|
}
|
|
83
|
|
|
84
|
|
|
85
|
?>
|