extension point of geturl #118

Author: code4craft

webmagic-core/src/main/java/us/codecraft/webmagic/scheduler/component/BloomFilterDuplicateRemover.java

@@ -46,14 +46,18 @@ protected BloomFilter<CharSequence> rebuildBloomFilter() {
 
     @Override
     public boolean isDuplicate(Request request, Task task) {
-        boolean isDuplicate = bloomFilter.mightContain(request.getUrl());
+        boolean isDuplicate = bloomFilter.mightContain(getUrl(request));
         if (!isDuplicate) {
-            bloomFilter.put(request.getUrl());
+            bloomFilter.put(getUrl(request));
             counter.incrementAndGet();
         }
         return isDuplicate;
     }
 
+    protected String getUrl(Request request) {
+        return request.getUrl();
+    }
+
     @Override
     public void resetDuplicateCheck(Task task) {
         rebuildBloomFilter();

webmagic-core/src/main/java/us/codecraft/webmagic/scheduler/component/HashSetDuplicateRemover.java

@@ -16,7 +16,11 @@ public class HashSetDuplicateRemover implements DuplicateRemover {
 
     @Override
     public boolean isDuplicate(Request request, Task task) {
-        return !urls.add(request.getUrl());
+        return !urls.add(getUrl(request));
+    }
+
+    protected String getUrl(Request request) {
+        return request.getUrl();
     }
 
     @Override