HtmlEncoder


1   package jodd.servlet;
2   
3   import java.io.UnsupportedEncodingException;
4   import java.net.URLDecoder;
5   import java.net.URLEncoder;
6   
7   /**
8    * Encodes text strings and URLs to be HTML-safe.
9    */
10  public final class HtmlEncoder {
11  
12      private static float newSizeFactor = 1.3f;
13  
14  
15      /**
16       * Lookup table for use in encode() method.
17       *
18       * @see #encode
19       */
20      public static final String[] TABLE_HTML = new String[256];
21  
22      /**
23       * Lookup table for use in encodeTextXxx() methods.
24       *
25       * @see #encodeText
26       * @see #encodeTextSmart
27       * @see #encodeTextStrict
28       */
29      public static final String[] TABLE_HTML_STRICT = new String[256];
30  
31      static {
32          for (int i = 0; i < 10; i++) {
33              TABLE_HTML[i] = "&#00" + i + ";";
34          }
35          for (int i = 10; i < 32; i++) {
36              TABLE_HTML[i] = "&#0" + i + ";";
37          }
38          for (int i = 32; i < 128; i++) {
39              TABLE_HTML[i] = String.valueOf((char)i);
40          }
41          for (int i = 128; i < 256; i++) {
42              TABLE_HTML[i] = "&#" + i + ";";
43          }
44  
45          // special characters
46          TABLE_HTML['\''] = "&#039;";    // apostrophe ('&apos;' doesn't work - it is not by the w3 specs)
47          TABLE_HTML['\"'] = "&quot;";    // double quote
48          TABLE_HTML['&'] = "&amp;";      // ampersand
49          TABLE_HTML['<'] = "&lt;";       // lower than
50          TABLE_HTML['>'] = "&gt;";       // greater than
51  
52          // strict table
53          System.arraycopy(TABLE_HTML, 0, TABLE_HTML_STRICT, 0, 256);
54          TABLE_HTML_STRICT[' '] = "&nbsp;";
55          TABLE_HTML_STRICT['\n'] = "<br>";       // ascii 10
56          TABLE_HTML_STRICT['\r'] = "<br>";       // ascii 13
57      }
58  
59      // ---------------------------------------------------------------- encoding
60  
61      /**
62       * Encode string to HTML-safe text. Extra characters are encoded as decimals,
63       * and five special characters are replaced with their HTML values:
64       * <li>' with &amp;#039;</li>
65       * <li>" with &amp;quot;</li>
66       * <li>&amp; with &amp;amp;</li>
67       * <li>&lt; with &amp;lt;</li>
68       * <li>&gt; with &amp;gt;</li>
69       *
70       * @param string input string
71       *
72       * @return HTML-safe string
73       * @see #encodeText
74       */
75      public static String encode(String string) {
76          if ((string == null) || (string.length() == 0)) {
77              return "";
78          }
79          int n = string.length();
80          StringBuffer buffer = new StringBuffer((int) (n * newSizeFactor));
81          int tableLen = TABLE_HTML.length;
82          char c;
83          for (int i = 0; i < n; i++) {
84              c = string.charAt(i);
85              if (c < tableLen) {
86                  buffer.append(TABLE_HTML[c]);
87              } else {
88                  buffer.append("&#").append((int)c).append(';');
89              }
90          }
91          return buffer.toString();
92      }
93  
94      /**
95       * Encodes text int HTML-safe text and preserves format. Additionaly, the following
96       * characters are replaced:
97       * <li>' ' with &amp;nbsp;</li>
98       * <li>\n with &lt;br&gt;</li>
99       * <li>\r with &lt;br&gt;</li>
100      * <br><br>
101      * Additionaly, this method takes care about CRLF and LF texts and handles
102      * both.
103      * 
104      * Common problem with this method is that spaces are not breakable, so they
105      * may break the outline of the page.
106      *
107      * @param string input string
108      *
109      * @return HTML-safe format
110      */
111     public static String encodeTextStrict(String string) {
112         if ((string == null) || (string.length() == 0)) {
113             return "";
114         }
115         int n = string.length();
116         StringBuffer buffer = new StringBuffer((int) (n * newSizeFactor));
117         int tableLen = TABLE_HTML_STRICT.length;
118         char c = 0, prev = 0;
119         for (int i = 0; i < n; i++, prev = c) {
120             c = string.charAt(i);
121 
122             if ((c == '\n') && (prev == '\r')) {
123                 continue;       // previously '\r' (CR) was encoded, so skip '\n' (LF)
124             }
125             if (c < tableLen) {
126                 buffer.append(TABLE_HTML_STRICT[c]);
127             } else {
128                 buffer.append("&#").append((int)c).append(';');
129             }
130         }
131         return buffer.toString();
132     }
133 
134     /**
135      * Encodes text int HTML-safe text and preserves format except spaces.
136      * Additionaly, the following characters are replaced:
137      *
138      * <li>\n with &lt;br&gt;</li>
139      * <li>\r with &lt;br&gt;</li>
140      * <br><br>
141      * Additionaly, this method takes care about CRLF and LF texts and handles
142      * both.
143      *
144      * @param string input string
145      *
146      * @return HTML-safe format
147      */
148     public static String encodeText(String string) {
149         if ((string == null) || (string.length() == 0)) {
150             return "";
151         }
152         int n = string.length();
153         StringBuffer buffer = new StringBuffer((int) (n * newSizeFactor));
154         int tableLen = TABLE_HTML_STRICT.length;
155         char c = 0, prev = 0;
156         for (int i = 0; i < n; i++, prev = c) {
157             c = string.charAt(i);;
158 
159             if (c == ' ') {
160                 buffer.append(' ');
161                 continue;
162             }
163             if ((c == '\n') && (prev == '\r')) {
164                 continue;       // previously '\r' (CR) was encoded, so skip '\n' (LF)
165             }
166             if (c < tableLen) {
167                 buffer.append(TABLE_HTML_STRICT[c]);
168             } else {
169                 buffer.append("&#").append((int)c).append(';');
170             }
171         }
172         return buffer.toString();
173     }
174 
175     /**
176      * Encodes text int HTML-safe text and preserves format using smart spaces.
177      * Additionaly, the following characters are replaced:
178      *
179      * <li>\n with &lt;br&gt;</li>
180      * <li>\r with &lt;br&gt;</li>
181      * <br><br>
182      * Additionaly, this method takes care about CRLF and LF texts and handles
183      * both.<br>
184      *
185      * This method is special since it preserves format, but with combination of
186      * not-breakable spaces and common spaces, so breaks are availiable.
187      *
188      * @param string input string
189      *
190      * @return HTML-safe format
191      */
192     public static String encodeTextSmart(String string) {
193         if ((string == null) || (string.length() == 0)) {
194             return "";
195         }
196         int n = string.length();
197         StringBuffer buffer = new StringBuffer((int) (n * newSizeFactor));
198         int tableLen = TABLE_HTML_STRICT.length;
199         char c = 0, prev = 0;
200         boolean prevSpace = false;
201         for (int i = 0; i < n; i++, prev = c) {
202             c = string.charAt(i);;
203 
204             if (c == ' ') {
205                 if (prev != ' ') {
206                     prevSpace = false;
207                 }
208                 if (prevSpace == false) {
209                     buffer.append(' ');
210                 } else {
211                     buffer.append("&nbsp;");
212                 }
213                 prevSpace = !prevSpace;
214                 continue;
215             }
216             if ((c == '\n') && (prev == '\r')) {
217                 continue;       // previously '\r' (CR) was encoded, so skip '\n' (LF)
218             }
219             if (c < tableLen) {
220                 buffer.append(TABLE_HTML_STRICT[c]);
221             } else {
222                 buffer.append("&#").append((int)c).append(';');
223             }
224         }
225         return buffer.toString();
226     }
227 
228 
229     // ---------------------------------------------------------------- URL encode/decode
230 
231     /**
232      * Encodes HTML JavaScript for page output using ISO-88591-1 encoding. Null
233      * strings are converted to empty ones. Unfortunatelly, this encoding is not
234      * comatible with the javascripts functions escape/unescape.
235      *
236      * @param string input string
237      *
238      * @return HTML ready string.
239      */
240     public static String encodeUrl(String string) {
241         return encodeUrl(string, "ISO-8859-1");
242     }
243     /**
244      * Encodes HTML JavaScript for page output. Null strings are converted to
245      * empty ones.
246      *
247      * @param string   input string
248      * @param encoding
249      *
250      * @return HTML ready string.
251      */
252     public static String encodeUrl(String string, String encoding) {
253         if (string == null) {
254             return "";
255         }
256         try {
257             return URLEncoder.encode(string, encoding);
258         } catch (UnsupportedEncodingException e) {
259             return null;
260         }
261     }
262 
263     /**
264      * Encodes HTML JavaScript for page output using ISO-88591-1 encoding. Null
265      * strings are converted to empty ones.
266      *
267      * @param string input
268      *
269      * @return HTML ready string.
270      */
271     public static String decodeUrl(String string) {
272         return decodeUrl(string, "ISO-8859-1");
273     }
274 
275     /**
276      * Encodes HTML JavaScript for page output. Null strings are converted to empty ones.
277      *
278      * @param string   input
279      * @param encoding encoding
280      *
281      * @return HTML ready string.
282      */
283     public static String decodeUrl(String string, String encoding) {
284         if (string == null) {
285             return "";
286         }
287         try {
288             return URLDecoder.decode(string, encoding);
289         } catch (UnsupportedEncodingException e) {
290             return null;
291         }
292     }
293 
294 }
295
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Java Books Remove Frame
Popular Tags