Bugfix, cleanup.

earwigbot/wiki/copyvios/__init__.py

@@ -268,8 +268,8 @@ class CopyvioMixIn(object):
         searcher = self._get_search_engine()
         parser = ArticleTextParser(self.get())
         article = MarkovChain(parser.strip())
-        workspace = _CopyvioWorkspace(article, min_confidence,
-                                      until, self._logger, self._addheaders)
+        workspace = _CopyvioWorkspace(article, min_confidence, until,
+                                      self._logger, self._addheaders)
         if self._exclusions_db:
             self._exclusions_db.sync(self.site.name)
             exclude = lambda u: self._exclusions_db.check(self.site.name, u)
@@ -328,8 +328,8 @@ class CopyvioMixIn(object):
         start_time = time()
         until = (start_time + max_time) if max_time > 0 else None
         article = MarkovChain(ArticleTextParser(self.get()).strip())
-        workspace = _CopyvioWorkspace(article, min_confidence,
-                                      until, self._logger, self._addheaders)
+        workspace = _CopyvioWorkspace(article, min_confidence, until,
+                                      self._logger, self._addheaders, max_time)
         workspace.enqueue([url])
         workspace.spawn(1)
         workspace.wait()

earwigbot/wiki/copyvios/parsers.py

@@ -134,7 +134,8 @@ class ArticleTextParser(BaseTextParser):
         """
         schemes = ("http://", "https://")
         links = mwparserfromhell.parse(self.text).ifilter_external_links()
-        return [link.url for link in links if link.url.startswith(schemes)]
+        return [unicode(link.url) for link in links
+                if link.url.startswith(schemes)]
 
 
 class HTMLTextParser(BaseTextParser):

earwigbot/wiki/copyvios/search.py

@@ -77,9 +77,8 @@ class YahooBOSSSearchEngine(BaseSearchEngine):
             "oauth_nonce": oauth.generate_nonce(),
             "oauth_timestamp": oauth.Request.make_timestamp(),
             "oauth_consumer_key": consumer.key,
-            "q": quote_plus(query.encode("utf8")),
-            "type": "html,text",
-            "format": "json",
+            "q": quote_plus(query.encode("utf8")), "count": 5,
+            "type": "html,text", "format": "json",
         }
 
         req = oauth.Request(method="GET", url=url, parameters=params)