Changeset 11419 for trunk/gsdl

Timestamp:

2006-03-21T09:32:05+12:00 (18 years ago)

Author:

shaoqun

Message:

fix code that test the image file extension and external links

File:

• trunk/gsdl/src/java/org/nzdl/gsdl/GsdlCollageApplet/DownloadUrls.java (modified) (17 diffs)

trunk/gsdl/src/java/org/nzdl/gsdl/GsdlCollageApplet/DownloadUrls.java

@@ -30 +30 @@
     String starting_url_  = null;
 
+    /** the root directory of Greenstone*/  
+    String document_root_ = null;
+
+
     /** CHRIS - Holds the contents of the collection's assoc directory */
     //    File[] assocDir_ = null;
@@ -59 +63 @@
     boolean thread_running_ = true;
 
-    int verbosity_;
+    int verbosity_ = 0;
 
     /** Constructor to initialise a download thread from which images are found, 
@@ -74 +78 @@
             DownloadImages download_images, String starting_url, 
             String href_musthave, String image_mustnothave, 
-            String image_ignore, String image_type, int verbosity)
+            String image_ignore, String image_type, String document_root,int verbosity)
     {
     super("DownloadUrls");
@@ -85 +89 @@
     image_ignore_  = image_ignore;
     image_type_    = image_type;
-    
+    document_root_ = document_root; 
     verbosity_ = verbosity;
+     
+
+    System.err.println("starting_url_ "  + starting_url +"\n"+
+                       "href_musthave_ " +  href_musthave +"\n"+
+               "image_mustnothave_" + image_mustnothave+"\n"+
+               "image_ignore_ "+  image_ignore+"\n"+
+               "image_type_ "+ image_type+"\n"+
+               "document root "+ document_root_
+                   );
+
+
+
+
     }
 
@@ -137 +154 @@
         
         tmp = image_type_.substring(0, image_type_.indexOf("%"));
+             
         if (image_type_.length() > image_type_.indexOf("%") + 1)
-        image_type_ = image_type_.substring(image_type_.indexOf("%") + 1, image_type_.length());
+        image_type_ = image_type_.substring(image_type_.indexOf("%") + 1, image_type_.length());                 
         else
         image_type_ = null;
-
-        if (url_lstring.endsWith(tmp)) {
+            
+        if (!tmp.trim().equals("") && url_lstring.endsWith(tmp)) {
         image_type_ = original_image_type_;
         return true;
@@ -164 +182 @@
     public boolean filter_image(String url_string)
     {
+        
     if (image_ignore_==null || !url_string.startsWith(image_ignore_))
     {
@@ -181 +200 @@
             else
                 image_mustnothave_ = null;
-            
-            if (url_string.indexOf(tmp) >= 0) {
+                                    
+                     
+                         
+            if (!tmp.trim().equals("") && url_string.indexOf(tmp) >= 0) {
+               
                 image_mustnothave_ = original_image_mustnothave_;
                 return false;
             }
-            }
-
+                }
+
+            image_mustnothave_ = original_image_mustnothave_;
+                    
             if (image_mustnothave_ != null && url_string.indexOf(image_mustnothave_) >= 0) {
             image_mustnothave_ = original_image_mustnothave_;
@@ -198 +222 @@
             
             image_mustnothave_ = original_image_mustnothave_;
-            return true;
+            
         }
-        }
-    }
-
-    return false;
+    
+        }
+            
+    }
+    
+    return true;
     }
 
@@ -351 +377 @@
     public void rec_add_images(String new_url, int depth)
     {
-    System.err.println("Parsing url = " + new_url);
+
+       
+    System.err.println("Parsing url =  " + new_url);
     
     if (already_visited(new_url)) return;
@@ -384 +412 @@
         URL url = (URL)src_links.elementAt(i);
         String url_string  = url.toString();
+                
+        //System.err.println(" source links " + i + " ["  + url_string +"]");
 
         if (verbosity_ >= 3) {
@@ -394 +424 @@
                 {
                 img_name = url_string.substring(url_string.lastIndexOf("/") + 1, url_string.length());
-
+                                
                 if (verbosity_ >= 2) {
                     System.err.println("    Filtered: src_link[" + i + "] = " + url_string);
@@ -401 +431 @@
                 if ((external_links_ != null) && (!external_links_.isEmpty())) {
                     String ext = (String) external_links_.get(img_name);
-                    
-                    if (ext != null) 
+                          
+                           
+                    if (ext != null){
                     add_image(url, ext, img_name);
-                    else
+                                       
+                    }
+                    else{
+                
                     add_image(url, new_url, img_name);
+                    }
                 }
                 else {
+                    
                     add_image(url, new_url, img_name);
                 }
+                           
+                
                 }
+        
             }
         
@@ -418 +457 @@
         Vector href_links = curl.getHrefLinks();
         
+      
+        if (verbosity_ >= 2) {
+        System.err.println("  Got href links... there are " + href_links.size() + " of them.");
+        }
+        
+
         // process each of the href links according to the parameters given.
         for (int i = 0; i < href_links.size(); i++) 
@@ -423 +468 @@
             URL url = (URL)href_links.elementAt(i);
             String url_string  = url.toString();
-            
+
+            //System.err.println(" href links " + i + "["  + url_string +"]");
+                    
+
             if (image_file_extension(url_string))
             {
+                                      
                 if (filter_image(url_string))
+
                 {
                     img_name = url_string.substring(url_string.lastIndexOf("/") + 1, url_string.length());
@@ -447 +497 @@
                 if (filter_href(url_string,new_url,depth))
                 {
+            
+                                    System.out.println("*************************************");
                     rec_add_images(url_string,depth+1);
                     
@@ -471 +523 @@
     try {
 
-        if (starting_url_.indexOf("gsdl") >= 0) {
+        if (starting_url_.indexOf(document_root_) >= 0 ){ 
         external_links_ = null;
         return;
@@ -519 +571 @@
     System.err.println("Starting download thread.");
     visited_url_ = new Hashtable();
-    rec_add_images(starting_url_,1);
+    
+        rec_add_images(starting_url_,1);
 
     System.err.println("Download thread finished.");