HTMLParser


1   //
2   //                                   ____.
3   //                       __/\ ______|    |__/\.     _______
4   //            __   .____|    |       \   |    +----+       \
5   //    _______|  /--|    |    |    -   \  _    |    :    -   \_________
6   //   \\______: :---|    :    :           |    :    |         \________>
7   //           |__\---\_____________:______:    :____|____:_____\
8   //                                      /_____|
9   //
10  //                 . . . i n   j a h i a   w e   t r u s t . . .
11  //
12  //--------------------------
13  // HTMLParser
14  //--------------------------
15  // J�r�me B�dat 14.12.2000
16  //--------------------------
17  
18  package org.jahia.layout;
19  
20  import org.apache.regexp.RE;
21  import org.apache.regexp.RESyntaxException;
22  import org.jahia.utils.JahiaConsole;
23  
24  
25  public class HTMLParser
26  {
27  
28      private static HTMLParser theObject = null;
29      private String   newHTML;
30      private String   newString;
31      private String   theScripts = "";
32      private boolean matched;
33      private String   theResult;
34      
35      
36      /**
37       * HTMLParser
38       *
39       * @author J�r�me B�dat
40       *
41       */
42      private HTMLParser() 
43      {
44          JahiaConsole.println( "HTML Parser", "Parsing..." );
45      } // end Constructor
46  
47  
48      /**
49       * getInstance
50       *
51       * @author J�r�me B�dat
52       *
53       */
54      public static synchronized HTMLParser getInstance()
55      {
56          if (theObject == null)
57          {
58              theObject = new HTMLParser();
59          }
60          return theObject;
61      } // end getInstance
62  
63  
64      /**
65       * getDHTML  (replace ('")  with \' or \", delete new lines and set the content of script tags to a variable)
66       *
67       * @author J�r�me B�dat
68       *
69       */   
70      public String   getDHTML(String   oldHTML)
71      {
72          try 
73          {
74              newHTML = oldHTML;
75              
76              RE r = new RE("\\n");
77              newHTML = r.subst(newHTML,"");
78              
79              r = new RE("\\r");
80              newHTML = r.subst(newHTML,"");
81   
82              r = new RE("<script(.*?)</script>", RE.MATCH_CASEINDEPENDENT);
83              matched = r.match(newHTML);      
84              theResult = r.getParen(0);
85              if (theResult != null) { setScripts(theResult); }
86              newHTML = r.subst(newHTML,"");
87              
88              r = new RE("<style(.*?)</style>", RE.MATCH_CASEINDEPENDENT);
89              matched = r.match(newHTML);                 
90              theResult = r.getParen(0);
91              if (theResult != null) { setScripts(theResult); }
92              newHTML = r.subst(newHTML,"");
93  
94              r = new RE("<link(.*?)>", RE.MATCH_CASEINDEPENDENT);
95              matched = r.match(newHTML);                 
96              theResult = r.getParen(0);
97              if (theResult != null) { setScripts(theResult); }
98              newHTML = r.subst(newHTML,"");
99  
100             r = new RE("<!(.*?)>");
101             newHTML = r.subst(newHTML,"");
102             
103             r = new RE("\'");
104             newHTML = r.subst(newHTML,"\\'");
105             
106             // r = new RE("\"");
107             // newHTML = r.subst(newHTML,"\\\"");
108             
109         }
110         catch (RESyntaxException e)
111         {
112             JahiaConsole.println("HTML Parser",e.toString());
113         }
114         return newHTML;
115     } // end getDHTML
116 
117 
118     /**
119      * getQuoteString  (replace ('")  with \' or \"
120      *
121      * @author J�r�me B�dat
122      *
123      */   
124     public String   getQuoteString(String   oldHTML) 
125     {
126         try 
127         {
128             newHTML = oldHTML;
129             
130             RE r = new RE("\'");
131             newHTML = r.subst(newHTML,"\\'");
132             
133             // r = new RE("\"");
134             // newHTML = r.subst(newHTML,"\\\"");
135             
136         }
137         catch (RESyntaxException e)
138         {
139             JahiaConsole.println("HTML Parser",e.toString());
140         }
141         return newHTML;
142     } // end getQuoteString
143 
144  
145     /**
146      * getHTMLString  (replace (" ")  with "%nbsp;")
147      *
148      * @author J�r�me B�dat
149      *
150      */
151     public String   getHTMLString(String   oldString) 
152     {
153         try 
154         {
155             newString = oldString;
156             
157             RE r = new RE(" ");
158             newString = r.subst(newString,"&nbsp;");
159         
160         }
161         catch (RESyntaxException e)
162         {
163             JahiaConsole.println("HTML Parser",e.toString());
164         }
165         return newString;
166     } // end getHTMLString
167     
168     
169     /**
170      * setScripts  (Set content of script tags to a single variable)
171      *
172      * @author J�r�me B�dat
173      *
174      */
175     private void setScripts(String   theScript) {
176         theScripts += theScript;
177     } // end setScripts
178 
179 
180     /**
181      * getScripts  (Get content of script tags)
182      *
183      * @author J�r�me B�dat
184      *
185      */    
186     public String   getScripts() {
187         theResult = theScripts;
188         theScripts = ""; 
189         return theResult;
190     } // end getScripts
191 
192 
193 }
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags