Changeset 11507 for trunk

Timestamp:

2006-03-28T11:35:50+12:00 (18 years ago)

Author:

shaoqun

Message:

add new parameter for wget that specifies which file extenstions should be accepted

File:

• trunk/gli/src/org/greenstone/gatherer/collection/DownloadJob.java (modified) (5 diffs)

trunk/gli/src/org/greenstone/gatherer/collection/DownloadJob.java

@@ -185 +185 @@
      */
     public boolean addLink(String raw_url, int type) {
-    ///ystem.out.println("addLink("+url+", "+type+")");
+    ///System.out.println("addLink("+url+", "+type+")");
     if(!encountered_urls.contains(raw_url)) {
         // Add it to the urls we've seen.
@@ -256 +256 @@
         }
     }
+
+
+      
         
     // The user can choose to mirror all of the page requisites...
@@ -261 +264 @@
         command = command + "-p ";
     }
+    else{ //or only HTML page (ignore associated files e.g images, stylesheets)
+        command = command + "-A " + ".html,.htm,.shm,.shtml,.asp,.php,.cgi,*?*=* ";
+    }
 
     // Download files from other hosts
@@ -282 +288 @@
     }
     // Run it
+     
+        //System.out.println("***"+command);
     try {
         //DebugStream.println("Cmd: " + command); // don't print it out cos it may have the password in it
@@ -543 +551 @@
      */
     public boolean hasSignalledStop() {
-    if(state == DownloadJob.STOPPED || state == DownloadJob.PAUSED || 
+        if(state == DownloadJob.STOPPED || state == DownloadJob.PAUSED || 
        state == DownloadJob.COMPLETE) {
         return true;