|
1
37
38 package org.htmlcleaner;
39
40 import java.io.*;
41 import java.net.URL  ;
42
43
49 public class Utils {
50
51
55 public static String ltrim(String s) {
56 if (s == null) {
57 return null;
58 }
59
60 int index = 0;
61 int len = s.length();
62
63 while ( index < len && Character.isWhitespace(s.charAt(index)) ) {
64 index++;
65 }
66
67 return (index >= len) ? "" : s.substring(index);
68 }
69
70
74 public static String rtrim(String s) {
75 if (s == null) {
76 return null;
77 }
78
79 int len = s.length();
80 int index = len;
81
82 while ( index > 0 && Character.isWhitespace(s.charAt(index-1)) ) {
83 index--;
84 }
85
86 return (index <= 0) ? "" : s.substring(0, index);
87 }
88
89
95 public static StringBuffer readUrl(URL url, String charset) throws IOException {
96 StringBuffer buffer = new StringBuffer (1024);
97
98 Object content = url.getContent();
99 if (content instanceof InputStream) {
100 InputStreamReader reader = new InputStreamReader((InputStream)content, charset);
101 char[] charArray = new char[1024];
102
103 int charsRead = 0;
104 do {
105 charsRead = reader.read(charArray);
106 if (charsRead >= 0) {
107 buffer.append(charArray, 0, charsRead);
108 }
109 } while (charsRead > 0);
110 }
111
112 return buffer;
113 }
114
115 public static boolean isHexadecimalDigit(char ch) {
116 return Character.isDigit(ch) ||
117 ch == 'A' || ch == 'a' || ch == 'B' || ch == 'b' || ch == 'C' || ch == 'c' ||
118 ch == 'D' || ch == 'd' || ch == 'E' || ch == 'e' || ch == 'F' || ch == 'f';
119 }
120
121
124 public static String escapeXml(String s, boolean advanced, boolean recognizeUnicodeChars, boolean translateSpecialEntities) {
125 if (s != null) {
126 int len = s.length();
127 StringBuffer result = new StringBuffer (len);
128
129 for (int i = 0; i < len; i++) {
130 char ch = s.charAt(i);
131
132 if (ch == '&') {
133 if ( recognizeUnicodeChars && (i < len-1) && (s.charAt(i+1) == '#') ) {
134 int charIndex = i + 2;
135 String unicode = "";
136 while ( charIndex < len &&
137 (isHexadecimalDigit(s.charAt(charIndex)) || s.charAt(charIndex) == 'x' || s.charAt(charIndex) == 'X')
138 ) {
139 unicode += s.charAt(charIndex);
140 charIndex++;
141 }
142 if (charIndex == len || !"".equals(unicode)) {
143 try {
144 char unicodeChar = unicode.toLowerCase().startsWith("x") ?
145 (char)Integer.parseInt(unicode.substring(1), 16) :
146 (char)Integer.parseInt(unicode);
147 if ( "&<>\'\"".indexOf(unicodeChar) < 0 ) {
148 int replaceChunkSize = (charIndex < len && s.charAt(charIndex) == ';') ? unicode.length()+1 : unicode.length();
149 result.append( String.valueOf(unicodeChar) );
150 i += replaceChunkSize + 1;
151 } else {
152 i = charIndex;
153 result.append("&#" + unicode + ";");
154 }
155 } catch (NumberFormatException e) {
156 i = charIndex;
157 result.append("&#" + unicode + ";");
158 }
159 } else {
160 result.append("&");
161 }
162 } else {
163 if (translateSpecialEntities) {
164 String seq = s.substring(i, i+Math.min(10, len-i));
166 int semiIndex = seq.indexOf(';');
167 if (semiIndex > 0) {
168 String entity = seq.substring(1, semiIndex);
169 Integer code = (Integer ) SpecialEntities.entities.get(entity);
170 if (code != null) {
171 int entityLen = entity.length();
172 result.append( (char)code.intValue() );
173 i += entityLen + 1;
174 continue;
175 }
176 }
177 }
178
179 if (advanced) {
180 String sub = s.substring(i);
181 if ( sub.startsWith("&") ) {
182 result.append("&");
183 i += 4;
184 } else if ( sub.startsWith("'") ) {
185 result.append("'");
186 i += 5;
187 } else if ( sub.startsWith(">") ) {
188 result.append(">");
189 i += 3;
190 } else if ( sub.startsWith("<") ) {
191 result.append("<");
192 i += 3;
193 } else if ( sub.startsWith(""") ) {
194 result.append(""");
195 i += 5;
196 } else {
197 result.append("&");
198 }
199
200 continue;
201 }
202
203 result.append("&");
204 }
205 } else if (ch == '\'') {
206 result.append("'");
207 } else if (ch == '>') {
208 result.append(">");
209 } else if (ch == '<') {
210 result.append("<");
211 } else if (ch == '\"') {
212 result.append(""");
213 } else {
214 result.append(ch);
215 }
216 }
217
218 return result.toString();
219 }
220
221 return null;
222 }
223
224 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.