250 |
kaklik |
1 |
<?php |
|
|
2 |
/* $Id: kanji-encoding.lib.php,v 2.2 2003/11/26 22:52:23 rabus Exp $ */ |
|
|
3 |
// vim: expandtab sw=4 ts=4 sts=4: |
|
|
4 |
|
|
|
5 |
|
|
|
6 |
/** |
|
|
7 |
* Set of functions for kanji-encoding convert (available only with japanese |
|
|
8 |
* language) |
|
|
9 |
* |
|
|
10 |
* PHP4 configure requirements: |
|
|
11 |
* --enable-mbstring --enable-mbstr-enc-trans --enable-mbregex |
|
|
12 |
* |
|
|
13 |
* 2002/2/22 - by Yukihiro Kawada <kawada@den.fujifilm.co.jp> |
|
|
14 |
*/ |
|
|
15 |
|
|
|
16 |
/** |
|
|
17 |
* Gets the php internal encoding codes and sets the available encoding |
|
|
18 |
* codes list |
|
|
19 |
* 2002/1/4 by Y.Kawada |
|
|
20 |
* |
|
|
21 |
* @global string the current encoding code |
|
|
22 |
* @global string the available encoding codes list |
|
|
23 |
* |
|
|
24 |
* @return boolean always true |
|
|
25 |
*/ |
|
|
26 |
function PMA_internal_enc_check() { |
|
|
27 |
global $internal_enc, $enc_list; |
|
|
28 |
|
|
|
29 |
$internal_enc = mb_internal_encoding(); |
|
|
30 |
if ($internal_enc == 'EUC-JP') { |
|
|
31 |
$enc_list = 'ASCII,EUC-JP,SJIS,JIS'; |
|
|
32 |
} else { |
|
|
33 |
$enc_list = 'ASCII,SJIS,EUC-JP,JIS'; |
|
|
34 |
} |
|
|
35 |
|
|
|
36 |
return TRUE; |
|
|
37 |
} // end of the 'PMA_internal_enc_check' function |
|
|
38 |
|
|
|
39 |
|
|
|
40 |
/** |
|
|
41 |
* Reverses SJIS & EUC-JP position in the encoding codes list |
|
|
42 |
* 2002/1/4 by Y.Kawada |
|
|
43 |
* |
|
|
44 |
* @global string the available encoding codes list |
|
|
45 |
* |
|
|
46 |
* @return boolean always true |
|
|
47 |
*/ |
|
|
48 |
function PMA_change_enc_order() { |
|
|
49 |
global $enc_list; |
|
|
50 |
|
|
|
51 |
$p = explode(',', $enc_list); |
|
|
52 |
if ($p[1] == 'EUC-JP') { |
|
|
53 |
$enc_list = 'ASCII,SJIS,EUC-JP,JIS'; |
|
|
54 |
} else { |
|
|
55 |
$enc_list = 'ASCII,EUC-JP,SJIS,JIS'; |
|
|
56 |
} |
|
|
57 |
|
|
|
58 |
return TRUE; |
|
|
59 |
} // end of the 'PMA_change_enc_order' function |
|
|
60 |
|
|
|
61 |
|
|
|
62 |
/** |
|
|
63 |
* Kanji string encoding convert |
|
|
64 |
* 2002/1/4 by Y.Kawada |
|
|
65 |
* |
|
|
66 |
* @param string the string to convert |
|
|
67 |
* @param string the destinasion encoding code |
|
|
68 |
* @param string set 'kana' convert to JIS-X208-kana |
|
|
69 |
* |
|
|
70 |
* @global string the available encoding codes list |
|
|
71 |
* |
|
|
72 |
* @return string the converted string |
|
|
73 |
*/ |
|
|
74 |
function PMA_kanji_str_conv($str, $enc, $kana) { |
|
|
75 |
global $enc_list; |
|
|
76 |
|
|
|
77 |
if ($enc == '' && $kana == '') { |
|
|
78 |
return $str; |
|
|
79 |
} |
|
|
80 |
$nw = mb_detect_encoding($str, $enc_list); |
|
|
81 |
|
|
|
82 |
if ($kana == 'kana') { |
|
|
83 |
$dist = mb_convert_kana($str, 'KV', $nw); |
|
|
84 |
$str = $dist; |
|
|
85 |
} |
|
|
86 |
if ($nw != $enc && $enc != '') { |
|
|
87 |
$dist = mb_convert_encoding($str, $enc, $nw); |
|
|
88 |
} else { |
|
|
89 |
$dist = $str; |
|
|
90 |
} |
|
|
91 |
return $dist; |
|
|
92 |
} // end of the 'PMA_kanji_str_conv' function |
|
|
93 |
|
|
|
94 |
|
|
|
95 |
/** |
|
|
96 |
* Kanji file encoding convert |
|
|
97 |
* 2002/1/4 by Y.Kawada |
|
|
98 |
* |
|
|
99 |
* @param string the name of the file to convert |
|
|
100 |
* @param string the destinasion encoding code |
|
|
101 |
* @param string set 'kana' convert to JIS-X208-kana |
|
|
102 |
* |
|
|
103 |
* @return string the name of the converted file |
|
|
104 |
*/ |
|
|
105 |
function PMA_kanji_file_conv($file, $enc, $kana) { |
|
|
106 |
if ($enc == '' && $kana == '') { |
|
|
107 |
return $file; |
|
|
108 |
} |
|
|
109 |
|
|
|
110 |
$tmpfname = tempnam('', $enc); |
|
|
111 |
$fpd = fopen($tmpfname, 'wb'); |
|
|
112 |
$fps = fopen($file, 'r'); |
|
|
113 |
PMA_change_enc_order(); |
|
|
114 |
while (!feof($fps)) { |
|
|
115 |
$line = fgets($fps, 4096); |
|
|
116 |
$dist = PMA_kanji_str_conv($line, $enc, $kana); |
|
|
117 |
fputs($fpd, $dist); |
|
|
118 |
} // end while |
|
|
119 |
PMA_change_enc_order(); |
|
|
120 |
fclose($fps); |
|
|
121 |
fclose($fpd); |
|
|
122 |
unlink($file); |
|
|
123 |
|
|
|
124 |
return $tmpfname; |
|
|
125 |
} // end of the 'PMA_kanji_file_conv' function |
|
|
126 |
|
|
|
127 |
|
|
|
128 |
/** |
|
|
129 |
* Defines radio form fields to switch between encoding modes |
|
|
130 |
* 2002/1/4 by Y.Kawada |
|
|
131 |
* |
|
|
132 |
* @param string spaces character to prepend the output with |
|
|
133 |
* |
|
|
134 |
* @return string xhtml code for the radio controls |
|
|
135 |
*/ |
|
|
136 |
function PMA_set_enc_form($spaces) { |
|
|
137 |
return "\n" |
|
|
138 |
. $spaces . '<input type="radio" name="knjenc" value="" checked="checked" />non' . "\n" |
|
|
139 |
. $spaces . '<input type="radio" name="knjenc" value="EUC-JP" />EUC' . "\n" |
|
|
140 |
. $spaces . '<input type="radio" name="knjenc" value="SJIS" />SJIS' . "\n" |
|
|
141 |
. $spaces . ' ' . $GLOBALS['strEncto'] . '<br />' . "\n" |
|
|
142 |
. $spaces . '<input type="checkbox" name="xkana" value="kana" />' . "\n" |
|
|
143 |
. $spaces . ' ' . $GLOBALS['strXkana'] . '<br />' . "\n"; |
|
|
144 |
} // end of the 'PMA_set_enc_form' function |
|
|
145 |
|
|
|
146 |
|
|
|
147 |
PMA_internal_enc_check(); |
|
|
148 |
|
|
|
149 |
?> |