1
|
|
package net.sf.flock.parser;
|
2
|
|
|
3
|
|
/**
|
4
|
|
* Utility class for dealing with HTML entities.
|
5
|
|
*
|
6
|
|
* @version $Revision: 1.1 $
|
7
|
|
* @author $Author: phraktle $
|
8
|
|
*/
|
9
|
|
public class HTMLUtil {
|
10
|
|
|
11
|
|
private final static String[][] ENTITIES =
|
12
|
|
{{ "<" , "<" } ,
|
13
|
|
{ ">" , ">" } ,
|
14
|
|
{ "&" , "&" } ,
|
15
|
|
{ """ , "\"" } ,
|
16
|
|
{ "à" , "à" } ,
|
17
|
|
{ "À" , "À" } ,
|
18
|
|
{ "â" , "â" } ,
|
19
|
|
{ "ä" , "ä" } ,
|
20
|
|
{ "Ä" , "Ä" } ,
|
21
|
|
{ "Â" , "Â" } ,
|
22
|
|
{ "å" , "å" } ,
|
23
|
|
{ "Å" , "Å" } ,
|
24
|
|
{ "æ" , "æ" } ,
|
25
|
|
{ "Æ" , "Æ" } ,
|
26
|
|
{ "ç" , "ç" } ,
|
27
|
|
{ "Ç" , "Ç" } ,
|
28
|
|
{ "é" , "é" } ,
|
29
|
|
{ "É" , "É" } ,
|
30
|
|
{ "è" , "è" } ,
|
31
|
|
{ "È" , "È" } ,
|
32
|
|
{ "ê" , "ê" } ,
|
33
|
|
{ "Ê" , "Ê" } ,
|
34
|
|
{ "ë" , "ë" } ,
|
35
|
|
{ "Ë" , "Ë" } ,
|
36
|
|
{ "ï" , "ï" } ,
|
37
|
|
{ "Ï" , "Ï" } ,
|
38
|
|
{ "ô" , "ô" } ,
|
39
|
|
{ "Ô" , "Ô" } ,
|
40
|
|
{ "ö" , "ö" } ,
|
41
|
|
{ "Ö" , "Ö" } ,
|
42
|
|
{ "ø" , "ø" } ,
|
43
|
|
{ "Ø" , "Ø" } ,
|
44
|
|
{ "ß" , "ß" } ,
|
45
|
|
{ "ù" , "ù" } ,
|
46
|
|
{ "Ù" , "Ù" } ,
|
47
|
|
{ "û" , "û" } ,
|
48
|
|
{ "Û" , "Û" } ,
|
49
|
|
{ "ü" , "ü" } ,
|
50
|
|
{ "Ü" , "Ü" } ,
|
51
|
|
{ " " , " " } ,
|
52
|
|
{ "®" , "\u00a9" } ,
|
53
|
|
{ "©" , "\u00ae" } ,
|
54
|
|
{ "€" , "\u20a0" } };
|
55
|
|
|
56
|
|
|
57
|
4
|
public static String unescape(String source) {
|
58
|
4
|
if (source==null || source.indexOf('&')==-1) {
|
59
|
|
// no entities
|
60
|
1
|
return source;
|
61
|
|
}
|
62
|
3
|
int len = source.length();
|
63
|
3
|
StringBuffer result = new StringBuffer(len);
|
64
|
3
|
for (int i=0; i<len; i++) {
|
65
|
5
|
int startPos = source.indexOf('&', i);
|
66
|
|
|
67
|
5
|
if (startPos!=i) {
|
68
|
|
// we skipped some chars, append them to result
|
69
|
4
|
result.append(source.substring(i, startPos==-1 ? len : startPos));
|
70
|
|
}
|
71
|
|
|
72
|
5
|
if (startPos==-1) {
|
73
|
|
// no more entities
|
74
|
1
|
break;
|
75
|
|
}
|
76
|
|
|
77
|
4
|
int endPos = source.indexOf(';', startPos);
|
78
|
4
|
if (endPos==-1) {
|
79
|
|
// broken entity
|
80
|
1
|
result.append( source.substring(startPos) );
|
81
|
1
|
break;
|
82
|
|
}
|
83
|
|
|
84
|
3
|
String entity = source.substring(startPos, endPos+1);
|
85
|
3
|
int p=0;
|
86
|
3
|
for (; p<ENTITIES.length; p++) {
|
87
|
46
|
if (entity.equals(ENTITIES[p][0])) {
|
88
|
2
|
result.append( ENTITIES[p][1] );
|
89
|
2
|
break;
|
90
|
|
}
|
91
|
|
}
|
92
|
3
|
if (p>=ENTITIES.length) {
|
93
|
|
// no entity replacement found, leave as-is
|
94
|
1
|
result.append(entity);
|
95
|
|
}
|
96
|
|
|
97
|
|
// skip ahead a littlebit
|
98
|
3
|
i=endPos;
|
99
|
|
}
|
100
|
3
|
return result.toString();
|
101
|
|
}
|
102
|
|
|
103
|
1
|
public static String escape(String source) {
|
104
|
1
|
if (source==null)
|
105
|
1
|
return null;
|
106
|
|
|
107
|
0
|
StringBuffer result = new StringBuffer(source.length());
|
108
|
|
|
109
|
0
|
for (int i=0;i<source.length();i++) {
|
110
|
|
|
111
|
0
|
char ch = source.charAt(i);
|
112
|
|
|
113
|
0
|
if ((int)ch>127) {
|
114
|
0
|
result.append("&#").append((int)ch).append(';');
|
115
|
|
} else {
|
116
|
0
|
result.append(ch);
|
117
|
|
}
|
118
|
|
}
|
119
|
0
|
return result.toString();
|
120
|
|
}
|
121
|
|
|
122
|
|
}
|
123
|
|
|