[Httpunit-commit] CVS: httpunit/src/com/meterware/httpunit HTMLParserFactory.java,NONE,1.1 HTMLPage.java,1.18,1.19 HTMLParser.java,1.2,1.3 HttpUnitOptions.java,1.34,1.35 JTidyHTMLParser.java,1.3,1.4 NekoHTMLParser.java,1.3,1.4 NodeUtils.java,1.13,1.14

[Russell G. <rus...@us...>]

Update of /cvsroot/httpunit/httpunit/src/com/meterware/httpunit
In directory sc8-pr-cvs1:/tmp/cvs-serv24372/src/com/meterware/httpunit

Modified Files:
	HTMLPage.java HTMLParser.java HttpUnitOptions.java 
	JTidyHTMLParser.java NekoHTMLParser.java NodeUtils.java 
Added Files:
	HTMLParserFactory.java 
Log Message:
Add support for parser customization

***** Error reading new file[Errno 2] No such file or directory: 'HTMLParserFactory.java'
Index: HTMLPage.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/HTMLPage.java,v
retrieving revision 1.18
retrieving revision 1.19
diff -u -r1.18 -r1.19
--- HTMLPage.java	13 Dec 2002 12:57:11 -0000	1.18
+++ HTMLPage.java	20 Dec 2002 01:59:47 -0000	1.19
@@ -260,7 +260,7 @@
 
 
     public void parse( String text, URL pageURL ) throws SAXException, IOException {
-        HttpUnitOptions.getHTMLParser().parse( this, pageURL, text );
+        HTMLParserFactory.getHTMLParser().parse( this, pageURL, text );
     }
 
 

Index: HTMLParser.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/HTMLParser.java,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- HTMLParser.java	24 Oct 2002 13:28:53 -0000	1.2
+++ HTMLParser.java	20 Dec 2002 01:59:47 -0000	1.3
@@ -19,15 +19,16 @@
  * DEALINGS IN THE SOFTWARE.
  *
  *******************************************************************************************************************/
-import org.w3c.dom.Node;
 import org.xml.sax.SAXException;
 
 import java.net.URL;
 import java.io.IOException;
 
 /**
+ * A front end to a DOM parser that can handle HTML.
  *
  * @author <a href="mailto:rus...@ht...">Russell Gold</a>
+ * @author <a href="mailto:bw...@xm...">Bernhard Wagner</a>
  **/
 interface HTMLParser {
 
@@ -43,4 +44,16 @@
      * an HTML entity as a special character. This method should convert that character to normal text.
      */
     public String getCleanedText( String string );
+
+
+    /**
+     * Returns true if this parser supports preservation of the case of tag and attribute names.
+     */
+    public boolean supportsPreserveTagCase();
+
+
+    /**
+     * Returns true if this parser can return an HTMLDocument object.
+     */
+    public boolean supportsReturnHTMLDocument();
 }

Index: HttpUnitOptions.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/HttpUnitOptions.java,v
retrieving revision 1.34
retrieving revision 1.35
diff -u -r1.34 -r1.35
--- HttpUnitOptions.java	26 Nov 2002 11:55:57 -0000	1.34
+++ HttpUnitOptions.java	20 Dec 2002 01:59:47 -0000	1.35
@@ -61,17 +61,6 @@
     }
 
 
-    public static HTMLParser getHTMLParser() {
-        if (_nekoParser != null) {
-            return _nekoParser;
-        } else if (_jtidyParser != null) {
-            return _jtidyParser;
-        } else {
-            throw new RuntimeException( "No HTML parser found. Make sure that either nekoHTML.jar or Tidy.jar is in the in classpath" );
-        }
-    }
-
-
     /**
      * Returns true if HttpUnit is accepting and saving cookies. The default is to accept them.
      */
@@ -517,28 +506,10 @@
 
     private static boolean _exceptionsThrownOnScriptError = true;
 
-    private static HTMLParser _jtidyParser;
-
-    private static HTMLParser _nekoParser;
 
     static {
         _listeners = new Vector();
         reset();
 
-        try {
-            Class.forName( "org.w3c.tidy.Parser" );
-            _jtidyParser = (HTMLParser) Class.forName( "com.meterware.httpunit.JTidyHTMLParser" ).newInstance();
-        } catch (InstantiationException e) {
-        } catch (IllegalAccessException e) {
-        } catch (ClassNotFoundException e) {
-        }
-
-        try {
-            Class.forName( "org.cyberneko.html.HTMLConfiguration" );
-            _nekoParser = (HTMLParser) Class.forName( "com.meterware.httpunit.NekoHTMLParser" ).newInstance();
-        } catch (InstantiationException e) {
-        } catch (IllegalAccessException e) {
-        } catch (ClassNotFoundException e) {
-        }
     }
 }

Index: JTidyHTMLParser.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/JTidyHTMLParser.java,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4
--- JTidyHTMLParser.java	25 Nov 2002 22:41:07 -0000	1.3
+++ JTidyHTMLParser.java	20 Dec 2002 01:59:47 -0000	1.4
@@ -19,7 +19,6 @@
  * DEALINGS IN THE SOFTWARE.
  *
  *******************************************************************************************************************/
-import org.w3c.dom.Node;
 import org.w3c.tidy.Tidy;
 import org.xml.sax.SAXException;
 
@@ -48,6 +47,17 @@
     public String getCleanedText( String string ) {
         return (string == null) ? "" : string.replace( NBSP, ' ' );
     }
+
+
+    public boolean supportsPreserveTagCase() {
+        return false;
+    }
+
+
+    public boolean supportsReturnHTMLDocument() {
+        return false;
+    }
+
 
     final private static char NBSP = (char) 160;   // non-breaking space, defined by JTidy
 

Index: NekoHTMLParser.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/NekoHTMLParser.java,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4
--- NekoHTMLParser.java	25 Nov 2002 22:41:07 -0000	1.3
+++ NekoHTMLParser.java	20 Dec 2002 01:59:47 -0000	1.4
@@ -39,13 +39,14 @@
 /**
  *
  * @author <a href="mailto:rus...@ht...">Russell Gold</a>
+ * @author <a href="mailto:bw...@xm...">Bernhard Wagner</a>
  **/
 class NekoHTMLParser implements HTMLParser {
 
 
     public void parse( HTMLPage page, URL pageURL, String pageText ) throws IOException, SAXException {
         try {
-            DOMParser parser = DOMParser.newParser(page);
+            DOMParser parser = DOMParser.newParser( page );
             parser.parse( new InputSource( new StringReader( pageText ) ) );
             page.setRootNode( parser.getDocument() );
         } catch (DOMParser.ScriptException e) {
@@ -58,6 +59,17 @@
         return (string == null) ? "" : string.replace( NBSP, ' ' );
     }
 
+
+    public boolean supportsPreserveTagCase() {
+        return true;
+    }
+
+
+    public boolean supportsReturnHTMLDocument() {
+        return true;
+    }
+
+
     final private static char NBSP = (char) 160;   // non-breaking space, defined by nekoHTML
 }
 
@@ -72,6 +84,12 @@
     /** Filters property identifier. */
     protected static final String FILTERS = "http://cyberneko.org/html/properties/filters";
 
+    /** Element case settings. possible values: "upper", "lower", "match" */
+    protected static final String ELEMS = "http://cyberneko.org/html/properties/names/elems";
+
+    /** Attribute case settings. possible values: "upper", "lower", "no-change" */
+    protected static final String ATTRS = "http://cyberneko.org/html/properties/names/attrs";
+
 
     private HTMLPage _htmlPage;
     private ScriptableDelegate _scriptableObject;
@@ -82,9 +100,23 @@
         configuration.setFeature( AUGMENTATIONS, true );
         final ScriptFilter javaScriptFilter = new ScriptFilter( configuration );
         configuration.setProperty( FILTERS, new XMLDocumentFilter[] { javaScriptFilter } );
-        final DOMParser domParser = new DOMParser( configuration, page );
-        javaScriptFilter.setParser( domParser );
-        return domParser;
+        if (HTMLParserFactory.isPreserveTagCase()) {
+            configuration.setProperty( ELEMS, "match" );
+            configuration.setProperty( ATTRS, "no-change" );
+        }
+
+        try {
+            final DOMParser domParser = new DOMParser( configuration, page );
+            domParser.setFeature( DEFER_NODE_EXPANSION, false );
+            if (HTMLParserFactory.isReturnHTMLDocument()) domParser.setProperty( DOCUMENT_CLASS_NAME, HTML_DOCUMENT_CLASS_NAME );
+            javaScriptFilter.setParser( domParser );
+            return domParser;
+        } catch (SAXNotRecognizedException e) {
+            throw new RuntimeException( e.toString() );
+        } catch (SAXNotSupportedException e) {
+            throw new RuntimeException( e.toString() );
+        }
+
     }
 
 
@@ -124,15 +156,6 @@
     DOMParser( HTMLConfiguration configuration, HTMLPage page ) {
         super( configuration );
         _htmlPage = page;
-
-        try {
-            setFeature( DEFER_NODE_EXPANSION, false );
-            setProperty( DOCUMENT_CLASS_NAME, HTML_DOCUMENT_CLASS_NAME );
-        } catch (SAXNotRecognizedException e) {
-            throw new RuntimeException( e.toString() );
-        } catch (SAXNotSupportedException e) {
-            throw new RuntimeException( e.toString() );
-        }
     }
 
 

Index: NodeUtils.java
===================================================================
RCS file: /cvsroot/httpunit/httpunit/src/com/meterware/httpunit/NodeUtils.java,v
retrieving revision 1.13
retrieving revision 1.14
diff -u -r1.13 -r1.14
--- NodeUtils.java	14 Oct 2002 04:19:06 -0000	1.13
+++ NodeUtils.java	20 Dec 2002 01:59:47 -0000	1.14
@@ -109,7 +109,7 @@
                 return true;
             }
             public void processTextNodeValue( String value ) {
-                sb.append( HttpUnitOptions.getHTMLParser().getCleanedText( value ) );
+                sb.append( HTMLParserFactory.getHTMLParser().getCleanedText( value ) );
             }
         } );
         return sb.toString();