better way to catch utf8 problems

opp/models.py

@@ -50,10 +50,7 @@ def update_db(self, **kwargs):
             kwargs['last_checked'] = time.strftime('%Y-%m-%d %H:%M:%S') 
             query = "UPDATE sources SET {},urlhash=MD5(url) WHERE source_id = %s".format(
                 ",".join(k+"=%s" for k in kwargs.keys()))
-            try:
-                cur.execute(query, tuple(kwargs.values()) + (self.source_id,))
-            except Exception as e: # hack to catch 4-byte unicode strings not supported by mysql
-                debug(1, "COULD NOT WRITE DOC TO DATABASE: %s", e)
+            cur.execute(query, tuple(kwargs.values()) + (self.source_id,))
             debug(3, cur._last_executed)
             db.commit()
 

opp/scraper.py

@@ -313,9 +313,13 @@ def process_link(li, force_reprocess=False, redir_url=None, keep_tempfiles=False
             debug(2, "new source page: setting found_date to 1970")
             doc.found_date = datetime(1970, 1, 1)
 
-    doc.update_db()
-    li.update_db(status=1, doc_id=doc.doc_id)
-
+    try:
+        doc.update_db()
+        li.update_db(status=1, doc_id=doc.doc_id)
+    except Exception as e: # hack to catch 4-byte unicode strings not supported by mysql
+        debug(1, "COULD NOT WRITE DOC TO DATABASE: %s", e)
+        return 0
+
     # categorize, but only if doc has more than 1000 words --
     # otherwise categorization is pretty random:
     if doc.numwords > 700: