HtmlCleanerTest


1   /*
2    * Copyright 2004 Outerthought bvba and Schaubroeck nv
3    *
4    * Licensed under the Apache License, Version 2.0 (the "License");
5    * you may not use this file except in compliance with the License.
6    * You may obtain a copy of the License at
7    *
8    *     http://www.apache.org/licenses/LICENSE-2.0
9    *
10   * Unless required by applicable law or agreed to in writing, software
11   * distributed under the License is distributed on an "AS IS" BASIS,
12   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13   * See the License for the specific language governing permissions and
14   * limitations under the License.
15   */
16  package org.outerj.daisy.htmlcleaner;
17  
18  import junit.framework.TestCase;
19  import org.xml.sax.InputSource  ;
20  
21  import java.io.InputStream  ;
22  import java.io.Reader  ;
23  import java.io.InputStreamReader  ;
24  import java.io.BufferedReader  ;
25  
26  public class HtmlCleanerTest extends TestCase {
27      public void testIt() throws Exception   {
28          HtmlCleanerFactory factory = new HtmlCleanerFactory();
29          InputSource   is = new InputSource  (getClass().getClassLoader().getResourceAsStream("org/outerj/daisy/htmlcleaner/cleanerconf.xml"));
30          HtmlCleanerTemplate template = factory.buildTemplate(is);
31  
32          String   result;
33          HtmlCleaner cleaner = template.newHtmlCleaner();
34          // the \u0004 in there is to check that invalid XML characters are removed
35          result = cleaner.cleanToString("<html><body>\u0004abc</body></html>");
36          assertEquals(readResource("output1.txt"), result);
37  
38          cleaner = template.newHtmlCleaner();
39          result = cleaner.cleanToString("<html xmlns='abc'><body>abc<ul> </ul></body></html>");
40          assertEquals(readResource("output1.txt"), result);
41  
42          cleaner = template.newHtmlCleaner();
43          result = cleaner.cleanToString("<x:html xmlns:x='abc'><x:body x:r='z'>abc</x:body></x:html>");
44          assertEquals(readResource("output1.txt"), result);
45  
46          cleaner = template.newHtmlCleaner();
47          result = cleaner.cleanToString("abc");
48          assertEquals(readResource("output1.txt"), result);
49  
50          cleaner = template.newHtmlCleaner();
51          result = cleaner.cleanToString("<html>abc</html>");
52          assertEquals(readResource("output1.txt"), result);
53  
54          cleaner = template.newHtmlCleaner();
55          result = cleaner.cleanToString("<html><body>abc</html>");
56          assertEquals(readResource("output1.txt"), result);
57  
58          // * free text in body should be embedded in <p>
59          // * two br's should be translated to new paragraph
60          cleaner = template.newHtmlCleaner();
61          result = cleaner.cleanToString("<html><body>abc<br/><br/>def</html>");
62          assertEquals(readResource("output2.txt"), result);
63  
64          // * more then two br's should give same result
65          cleaner = template.newHtmlCleaner();
66          result = cleaner.cleanToString("<html><body>abc<br/><br/><br/>def</html>");
67          assertEquals(readResource("output2.txt"), result);
68  
69          // * two br's are translated to new paragraph
70          // * one br remains one br
71          // * one or more br's before </p> closing tag: remove them
72          cleaner = template.newHtmlCleaner();
73          result = cleaner.cleanToString("<html><body>abc<br/><br/>def<p>xyz<br/>xyz</p><p>yes<br/></p><p>yesyes<br/><br/><br/></html>");
74          assertEquals(readResource("output3.txt"), result);
75  
76          // * table inside a <p> tag
77          cleaner = template.newHtmlCleaner();
78          result = cleaner.cleanToString("<html><body><p><table><tr><td>hello!</td></tr></table></p></html>");
79          assertEquals(readResource("output4.txt"), result);
80  
81          // ul inside a p tag
82          cleaner = template.newHtmlCleaner();
83          result = cleaner.cleanToString("<html><body><p><ul><li>hello!</li></ul></p></html>");
84          assertEquals(readResource("output5.txt"), result);
85  
86          // ul inside a p tag with still some text around it
87          cleaner = template.newHtmlCleaner();
88          result = cleaner.cleanToString("<html><body><p>abc<ul><li>hello!</li></ul>def</p></html>");
89          assertEquals(readResource("output6.txt"), result);
90  
91          // test text reflow
92          cleaner = template.newHtmlCleaner();
93          result = cleaner.cleanToString("<html><body><p>Hi, this is a text longer then 80 characters which will hence be split across multiple lines. Isn't this interesting. No it isn't. Anyhow, have I told you about that time when I invented the wheel? Well, it was a long time ago.</p></html>");
94          assertEquals(readResource("output7.txt"), result);
95  
96          // test removal of not-allowed tags
97          cleaner = template.newHtmlCleaner();
98          result = cleaner.cleanToString("<html><body><p><font>abc</font></p></html>");
99          assertEquals(readResource("output1.txt"), result);
100 
101         // test translation of span with styling
102         cleaner = template.newHtmlCleaner();
103         result = cleaner.cleanToString("<html><body><p><span style='color: green; font-weight:bold  '>abc</span><span style='font-style:italic'>abc</span><span style='font-style:italic;font-weight:bold'>abc</span></p></html>");
104         assertEquals(readResource("output8.txt"), result);
105 
106         // test img src conversion
107         cleaner = template.newHtmlCleaner();
108         result = cleaner.cleanToString("<html><body><img SRC='hi' daisysrc='daisy:123'/></body></html>");
109         assertEquals(readResource("output9.txt"), result);
110 
111         cleaner = template.newHtmlCleaner();
112         result = cleaner.cleanToString("<html><body>Hi this is <strong>strong</strong> and <em>emphasized</em></body></html>");
113         assertEquals(readResource("output10.txt"), result);
114 
115         cleaner = template.newHtmlCleaner();
116         result = cleaner.cleanToString("<html><body>Hi this is <strong>strong</strong><em>emphasized</em></body></html>");
117         assertEquals(readResource("output11.txt"), result);
118 
119         cleaner = template.newHtmlCleaner();
120         result = cleaner.cleanToString("<html><body>Hi this is <strong>strong</strong> <em>emphasized</em></body></html>");
121         assertEquals(readResource("output12.txt"), result);
122 
123         cleaner = template.newHtmlCleaner();
124         result = cleaner.cleanToString("<html><body>aaaa bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb ccc</body></html>");
125         assertEquals(readResource("output13.txt"), result);
126 
127         cleaner = template.newHtmlCleaner();
128         result = cleaner.cleanToString("<html><body>aaaa bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb<img SRC='somewhere'/></body></html>");
129         assertEquals(readResource("output14.txt"), result);
130 
131         cleaner = template.newHtmlCleaner();
132         result = cleaner.cleanToString("<html><body>test test test test test test test                      \n\n test test test test test test test<a HREF='http://outerthought.org'>test</a> test</body></html>");
133         assertEquals(readResource("output15.txt"), result);
134 
135         cleaner = template.newHtmlCleaner();
136         result = cleaner.cleanToString("<html><body><strong>x</strong><img SRC='x.gif'/><strong>x</strong><img SRC='x.gif'/><strong>x</strong><img SRC='x.gif'/><strong>x</strong><img SRC='x.gif'/><strong>x</strong><img SRC='x.gif'/></html>");
137         assertEquals(readResource("output16.txt"), result);
138 
139         cleaner = template.newHtmlCleaner();
140         result = cleaner.cleanToString("<html><body><p> a b c </p></body></html>");
141         assertEquals(readResource("output17.txt"), result);
142 
143         cleaner = template.newHtmlCleaner();
144         result = cleaner.cleanToString("<html><body><p> a b c </p>   </body></html>");
145         assertEquals(readResource("output17.txt"), result);
146 
147         cleaner = template.newHtmlCleaner();
148         result = cleaner.cleanToString("<html><body> a b c <br/> </html>");
149         assertEquals(readResource("output17.txt"), result);
150 
151         // test removal of <br> inside <td>
152         cleaner = template.newHtmlCleaner();
153         result = cleaner.cleanToString("<html><body><table><tbody><tr><td><br/></td></tr></tbody></table></html>");
154         assertEquals(readResource("output18.txt"), result);
155 
156         cleaner = template.newHtmlCleaner();
157         result = cleaner.cleanToString("<html><body><table><tbody><tr><td><br/>\n</td></tr></tbody></table></html>");
158         assertEquals(readResource("output18.txt"), result);
159 
160         String   teststring = "<html><head><link rel=\"stylesheet\" type=\"text/css\" HREF=\"/daisy/resources/skins/default/css/htmlarea.css\" /></head>\n" +
161                 " <body>\n" +
162                 " \n" +
163                 " <p><strong>asfasdfa</strong></p>\n" +
164                 " \n" +
165                 " <p><strong>dfsafsa<br /></strong></p><p><strong><br />asfj aflad <span style=\"font-style: italic;\">fafjls fd<br /></span></strong></p><p><strong><span style=\"font-style: italic;\">saj lfsdj </span>lkjlkjweids<br /></strong></p>\n" +
166                 " \n" +
167                 " </body></html>";
168 
169         cleaner = template.newHtmlCleaner();
170         result = cleaner.cleanToString(teststring);
171         assertEquals(readResource("output19.txt"), result);
172 
173         cleaner = template.newHtmlCleaner();
174         result = cleaner.cleanToString("<html><body><p>abc<strong/></p><p><strong><em><em><em/></em></em></strong></p></body></html>");
175         assertEquals(readResource("output1.txt"), result);
176 
177         cleaner = template.newHtmlCleaner();
178         result = cleaner.cleanToString("<html><body><table><tr><td> <br/></td></tr></table></body></html>");
179         assertEquals(readResource("output20.txt"), result);
180 
181         cleaner = template.newHtmlCleaner();
182         result = cleaner.cleanToString("<html><body>hallo<table><tr><td>nog eens hallo</td></tr></table></body></html>");
183         assertEquals(readResource("output21.txt"), result);
184 
185         cleaner = template.newHtmlCleaner();
186         result = cleaner.cleanToString("<html><body><p>hallo<table><tr><td>nog eens hallo</td></tr></table></p></body></html>");
187         assertEquals(readResource("output21.txt"), result);
188 
189         cleaner = template.newHtmlCleaner();
190         result = cleaner.cleanToString("<html><body><p>hallo<table><tr><td>nog eens hallo<br/><br/>jaja<p>jan piet joris</p></td><td><table><tr><td><p>1</p>2</td></tr></table></td></tr></table></p></body></html>");
191         assertEquals(readResource("output22.txt"), result);
192 
193         cleaner = template.newHtmlCleaner();
194         result = cleaner.cleanToString("<html><body><pre>each<br/>word<br/>on a new<br/>line</pre></body></html>");
195         assertEquals(readResource("output23.txt"), result);
196 
197         cleaner = template.newHtmlCleaner();
198         result = cleaner.cleanToString("<html><body><h1>ab<br/></h1><h1><br/>\n</h1><h1><br/><h2><br/>cd</h2>ef</h1></body></html>");
199         assertEquals(readResource("output24.txt"), result);
200 
201         cleaner = template.newHtmlCleaner();
202         result = cleaner.cleanToString("<html><body>klsaflkjdkadjfkajlfksdjakfdsfka&lt;abc&gt;lsjfladjflsafjlsjflkjaskfjlkjflksjafkdjalfsajfkjalfdlsfaj</body></html>");
203         assertEquals(readResource("output25.txt"), result);
204 
205         // test link href conversion
206         cleaner = template.newHtmlCleaner();
207         result = cleaner.cleanToString("<html><body><a HREF='hi' daisyhref='daisy:123'>boe</a></body></html>");
208         assertEquals(readResource("output26.txt"), result);
209     }
210 
211     String   readResource(String   name) throws Exception   {
212         InputStream   is = getClass().getClassLoader().getResourceAsStream("org/outerj/daisy/htmlcleaner/" + name);
213         Reader   reader = new InputStreamReader  (is, "UTF-8");
214         BufferedReader   bufferedReader = new BufferedReader  (reader);
215 
216         StringBuffer   buffer = new StringBuffer  ();
217         int c = bufferedReader.read();
218         while (c != -1) {
219             buffer.append((char)c);
220             c = bufferedReader.read();
221         }
222 
223         return buffer.toString();
224     }
225 }
226
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags