From 32ef0fbf1ff60cff7ddc4fc68af2d6d175e4933a Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Sun, 31 Aug 2014 17:00:14 -0500
Subject: [PATCH] Add a bunch of temporary debugging code.

---
 earwigbot/wiki/copyvios/workers.py | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/earwigbot/wiki/copyvios/workers.py b/earwigbot/wiki/copyvios/workers.py
index 6085793..d521d6f 100644
--- a/earwigbot/wiki/copyvios/workers.py
+++ b/earwigbot/wiki/copyvios/workers.py
@@ -21,6 +21,7 @@
 # SOFTWARE.
 
 from gzip import GzipFile
+from logging import getLogger
 from math import log
 from Queue import Empty, Queue
 from socket import error
@@ -142,6 +143,8 @@ class _CopyvioWorker(object):
         self._queue = None
         self._opener = build_opener()
 
+        self._logger = None
+
     def _open_url(self, source):
         """Open a URL and return its parsed content, or None.
 
@@ -202,23 +205,32 @@ class _CopyvioWorker(object):
             timeout = None
 
         if self._queue:
+            self._logger.debug(u"Popping source from existing queue ({0})".format(self._site))
             source = self._queue.pop()
+            self._logger.debug(u"Got URL: {0}".format(source.url))
             with self._queues.lock:
                 if not self._queue:
+                    self._logger.debug(u"Destroying site {0}".format(self._site))
                     del self._queues.sites[self._site]
                     self._queue = None
         else:
+            self._logger.debug("Waiting for unassigned URL queue")
             site, queue = self._queues.unassigned.get(timeout=timeout)
             if site is StopIteration:
                 return StopIteration
+            self._logger.debug(u"Got queue: {0}".format(site))
             source = queue.pop()
+            self._logger.debug(u"Got URL: {0}".format(source.url))
             with self._queues.lock:
                 if not queue:
+                    self._logger.debug(u"Destroying site {0}".format(site))
                     del self._queues.sites[site]
                 else:
+                    self._logger.debug(u"Saving site {0}".format(site))
                     self._site = site
                     self._queue = queue
         if not source.active():
+            self._logger.debug(u"Inactive source; trying again")
             return self._dequeue()
         return source
 
@@ -246,6 +258,7 @@ class _CopyvioWorker(object):
         thread.name = "cvworker-" + name
         thread.daemon = True
         thread.start()
+        self._logger = getLogger("earwigbot.wiki.cvworker." + name)
 
 
 class CopyvioWorkspace(object):
@@ -358,7 +371,9 @@ class CopyvioWorkspace(object):
         """Wait for the workers to finish handling the sources."""
         self._logger.debug("Waiting on {0} sources".format(len(self.sources)))
         for source in self.sources:
+            self._logger.debug("Waiting on source: {0}".format(source.url))
             source.join(self._until)
+        self._logger.debug("Done waiting")
 
     def compare(self, source, source_chain):
         """Compare a source to the article, and update the best known one."""