diff --git a/websucker/dump.py b/websucker/dump.py
index b62bd4a..d9be4da 100644
--- a/websucker/dump.py
+++ b/websucker/dump.py
@@ -4,6 +4,7 @@ import cassandra.query
 import json
 import datetime
 import sys
+from urllib.parse import urlparse
 
 cassandra_host = sys.argv[1]
 cassandra_port = sys.argv[2]
@@ -16,9 +17,30 @@ select_documents = "select json * from content"
 with cluster.connect(keyspace) as session:
     #session.row_factory = cassandra.query.dict_factory
     select_html = session.prepare("select json * from html where day=? and domain_name=? LIMIT 1")
+    select_link = session.prepare("select link_status from links where domain_name=? and url_path=? and url_query=? LIMIT 1")
     rows = session.execute(select_documents)
     for row in rows:
         doc = json.loads(row["[json]"])
+        target_link = doc["target_link"]
+        parsed_link = urlparse(target_link)
+        netloc = parsed_link[1].strip().lower()
+        path = parsed_link[2].strip()
+        # strip leading /
+        if len(path) > 1 and path[0] == "/":
+            path = path[1:]
+        query = parsed_link[4]
+        lrows = session.execute(select_link,(netloc,path,query))
+        status = None
+        for l in lrows:
+            status = str(l["link_status"])
+            break
+        #assert status is not None
+        if status is None:
+            continue
+        #print(status)
+        # skip bad links
+        if not status == "good":
+            continue
         dt = doc["update_time"]
         d = dt.split()[0]
         hrows = session.execute(select_html,(d,doc["domain_name"]))
@@ -27,4 +49,6 @@ with cluster.connect(keyspace) as session:
             html = json.loads(h["[json]"])
             break
         doc["html_data"] = html
+        del doc["links"]
+        #print(parsed_link)
         print(json.dumps(doc))