Remove event_id field.

crawldb/hadoop.py

@@ -135,11 +135,11 @@ def mapper(self, line):
 
         # Yield this line if there seems there is no key collision with the previous line:
         if self.last_c and c.ssurt == self.last_c.ssurt and c.timestamp == self.last_c.timestamp:
-            logger.warning("Skipping line %i because the last line appears to collide with this one.")
+            logger.warning("Skipping line %i because the last line appears to collide with this one." % self.line_counter)
             logger.warning("Prev line %s" % self.last_c.line)
             logger.warning("Curr line %s" % c.line)
         else:
-            yield c.upsert_values(self.line_counter)
+            yield c.upsert_values()
 
         # Remember this line as the last line:
         self.last_c = c

crawldb/heritrix.py

@@ -92,7 +92,7 @@ def stats(self):
         """
         return self.stats
 
-    upsert_sql = """UPSERT INTO crawl_log (ssurt, timestamp, event_id, url, host, domain, content_type, content_length, content_digest, via, hop_path, status_code, ip ) VALUES %s"""
+    upsert_sql = """UPSERT INTO crawl_log (ssurt, timestamp, url, host, domain, content_type, content_length, content_digest, via, hop_path, status_code, ip ) VALUES %s"""
 
-    def upsert_values(self, event_id):
-        return (self.ssurt, self.timestamp, event_id, self.url, self.host, self.domain, self.mime, self.content_length, self.hash, self.via, self.hop_path, self.status_code, self.ip)
+    def upsert_values(self):
+        return (self.ssurt, self.timestamp, self.url, self.host, self.domain, self.mime, self.content_length, self.hash, self.via, self.hop_path, self.status_code, self.ip)