From b18273bbb2177de299f31abd6d64886f9d039838 Mon Sep 17 00:00:00 2001
From: low012 <low012@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Tue, 7 Nov 2006 20:42:37 +0000
Subject: [PATCH] *) YaCy will extract URLs from <embed> tags now. *) YaCy will
 extract URLs from <param> tags now if the name of the parameter is "movie".

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@2934 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 .../htmlFilter/htmlFilterContentScraper.java    | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/source/de/anomic/htmlFilter/htmlFilterContentScraper.java b/source/de/anomic/htmlFilter/htmlFilterContentScraper.java
index 6196fabc0..54fd5252a 100644
--- a/source/de/anomic/htmlFilter/htmlFilterContentScraper.java
+++ b/source/de/anomic/htmlFilter/htmlFilterContentScraper.java
@@ -4,6 +4,8 @@
 // first published on http://www.anomic.de
 // Frankfurt, Germany, 2004
 //
+// Contains contributions by Marc Nause [MN]
+//
 // $LastChangedDate$
 // $LastChangedRevision$
 // $LastChangedBy$
@@ -79,6 +81,8 @@ public class htmlFilterContentScraper extends htmlFilterAbstractScraper implemen
         linkTags0.add("meta");
         linkTags0.add("area");
         linkTags0.add("link");
+        linkTags0.add("embed");     //added by [MN]
+        linkTags0.add("param");     //added by [MN]
 
         linkTags1 = new TreeSet(insensitiveCollator);
         linkTags1.add("a");
@@ -203,7 +207,18 @@ public class htmlFilterContentScraper extends htmlFilterAbstractScraper implemen
                 }
             }
         }
-        
+        //start contrib [MN]
+        if (tagname.equalsIgnoreCase("embed")) {
+            anchors.put(absolutePath(tagopts.getProperty("src", "")), tagopts.getProperty("name",""));
+        }
+        if (tagname.equalsIgnoreCase("param")) {
+            String name = tagopts.getProperty("name", "");
+            if (name.equalsIgnoreCase("movie")) {
+                anchors.put(absolutePath(tagopts.getProperty("value", "")),name);
+            }
+        }
+        //end contrib [MN]
+
         // fire event
         fireScrapeTag0(tagname, tagopts);
     }