Changeset 5944 for trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/HTMLRecogniser.java

Timestamp:

2003-11-24T14:26:35+13:00 (20 years ago)

Author:

cs025

Message:

Index document type, metadata extensions

File:

• trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/HTMLRecogniser.java (modified) (1 diff)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/HTMLRecogniser.java

@@ -9 +9 @@
 public class HTMLRecogniser implements RecogniserInterface
 {
-    DocumentList listRepository;
+  DocumentList listRepository;
 
-    public HTMLRecogniser(DocumentList listRepository)
-    {   this.listRepository = listRepository;
-    }
+  public HTMLRecogniser(DocumentList listRepository)
+  { this.listRepository = listRepository;
+  }
+  
+  public boolean parseDocument(METSFile file)
+  {
+    String MIMEType = file.getMIMEType();
+    if (MIMEType == null || 
+    MIMEType.equals("text/html")) {
+      URL location = file.getLocation();
+      return this.parseDocument(location);
+    }
+    return false;
+  }
 
-    public boolean parseDocument(METSFile file)
-    {
-        String MIMEType = file.getMIMEType();
-        if (MIMEType == null || 
-              MIMEType.equals("text/html")) {
-            URL location = file.getLocation();
-            return this.parseDocument(location);
-        }
-        return false;
-    }
+  public boolean parseDocument(URL url)
+  { String fileName = null;
 
-    public boolean parseDocument(URL url)
-    {   if (url.toString().startsWith("file://")) {
-            String fileName = url.toString().substring(7);
-            if (fileName.endsWith(".htm") ||
-                  fileName.endsWith(".html"))
-          { System.out.println("Posting HTML Document " + fileName);
+    if (url.toString().startsWith("file://")) {
+      fileName = url.toString().substring(7);
+    }
+    else if (url.toString().startsWith("file:/")) {
+      fileName = url.toString().substring(5);
+    }
 
-            HTMLDocument doc = new HTMLDocument(url);
-            this.listRepository.addDocument(doc);
-              return true;
-            }
-        }
-        else {
-            // Get Mime type remotely, and then proceed if required
-            String mimeType = HTTPTools.getMIMEType(url);
+    if (fileName != null) {
+      if (fileName.endsWith(".htm") ||
+      fileName.endsWith(".html"))
+      { System.out.println("Posting HTML Document " + fileName);
 
-            if (mimeType == "text/html")
-            {   System.out.println("Posting HTML Document " + url.toString());
+        HTMLDocument doc = new HTMLDocument(url);
+    this.listRepository.addDocument(doc);
+    return true;
+      }
+    }
+    else {
+      // Get Mime type remotely, and then proceed if required
+      String mimeType = HTTPTools.getMIMEType(url);
+
+      if (mimeType == "text/html")
+      { System.out.println("Posting HTML Document " + url.toString());
             
-                HTMLDocument doc = new HTMLDocument(url);
-                this.listRepository.addDocument(doc);
-                return true;
-            }
-        }
-        return false;
-    }
+        HTMLDocument doc = new HTMLDocument(url);
+    this.listRepository.addDocument(doc);
+    return true;
+      }
+    }
+    return false;
+  }
 }