Ported fix for #73 to the refactored code.

coldsweat/commands.py

@@ -147,7 +147,7 @@ def command_update(self, options, args):
  else:
  print 'Database is already up-to-date.'
  except OperationalError, ex: 
- logger.error('caught exception updating database schema: (%s)' % ex)
+ logger.error(u'caught exception updating database schema: (%s)' % ex)
  print 'Error while running database update. See log file for more information.'
 
 def run():

coldsweat/controllers.py

@@ -53,10 +53,10 @@ def add_subscription(self, feed, group):
  try:
  subscription = Subscription.create(user=self.user, feed=feed, group=group)
  except IntegrityError:
- logger.debug('user %s has already feed %s in her subscriptions' % (self.user.username, feed.self_link)) 
+ logger.debug(u'user %s has already feed %s in her subscriptions' % (self.user.username, feed.self_link)) 
  return None
 
- logger.debug('added feed %s for user %s' % (feed.self_link, self.user.username)) 
+ logger.debug(u'added feed %s for user %s' % (feed.self_link, self.user.username)) 
  return subscription 
 
  def remove_subscription(self, feed):
@@ -80,26 +80,26 @@ def mark_entry(self, entry, status):
  try:
  Read.create(user=self.user, entry=entry)
  except IntegrityError:
- logger.debug('entry %s already marked as read, ignored' % entry.id)
+ logger.debug(u'entry %s already marked as read, ignored' % entry.id)
  return
  elif status == 'unread':
  count = Read.delete().where((Read.user==self.user) & (Read.entry==entry)).execute()
  if not count:
- logger.debug('entry %s never marked as read, ignored' % entry.id)
+ logger.debug(u'entry %s never marked as read, ignored' % entry.id)
  return
  elif status == 'saved':
  try:
  Saved.create(user=self.user, entry=entry)
  except IntegrityError:
- logger.debug('entry %s already saved, ignored' % entry.id)
+ logger.debug(u'entry %s already saved, ignored' % entry.id)
  return
  elif status == 'unsaved':
  count = Saved.delete().where((Saved.user==self.user) & (Saved.entry==entry)).execute()
  if not count:
- logger.debug('entry %s never saved, ignored' % entry.id)
+ logger.debug(u'entry %s never saved, ignored' % entry.id)
  return
 
- logger.debug('entry %s %s' % (entry.id, status))
+ logger.debug(u'entry %s %s' % (entry.id, status))
 
  def get_unread_entries(self, *select): 
  #@@TODO: include saved information too
@@ -172,7 +172,7 @@ def add_feed(self, feed, fetch_data=False):
 
  try:
  previous_feed = Feed.get(Feed.self_link == feed.self_link)
- logger.debug('feed %s has been already added to database, skipped' % feed.self_link)
+ logger.debug(u'feed %s has been already added to database, skipped' % feed.self_link)
  return previous_feed
  except Feed.DoesNotExist:
  pass
@@ -226,7 +226,7 @@ def add_feeds_from_file(self, filename):
  group = Group.get(Group.title==group.title)
  except Group.DoesNotExist:
  group.save()
- logger.debug('added group %s to database' % group.title)
+ logger.debug(u'added group %s to database' % group.title)
 
  groups.append(group)
 
@@ -262,7 +262,7 @@ def fetch_feeds(self, feeds): #@@TODO add processes param?
 
  load_plugins()
 
- logger.debug("starting fetcher")
+ logger.debug(u"starting fetcher")
  trigger_event('fetch_started')
 
  if config.fetcher.processes:
@@ -276,7 +276,7 @@ def fetch_feeds(self, feeds): #@@TODO add processes param?
 
  trigger_event('fetch_done', feeds)
 
- logger.info("%d feeds checked in %.2fs" % (len(feeds), time.time() - start)) 
+ logger.info(u"%d feeds checked in %.2fs" % (len(feeds), time.time() - start)) 
 
 
  def fetch_all_feeds(self):
@@ -289,7 +289,7 @@ def fetch_all_feeds(self):
 
  feeds = list(q)
  if not feeds:
- logger.debug("no feeds found to refresh, halted")
+ logger.debug(u"no feeds found to refresh, halted")
  return
 
  self.fetch_feeds(feeds)
@@ -299,7 +299,7 @@ def feed_worker(feed):
 
  #@@REMOVEME: just delete feed if there are no subscribers
 # if not feed.subscriptions:
-# logger.debug("feed %s has no subscribers, skipped" % feed.self_link)
+# logger.debug(u"feed %s has no subscribers, skipped" % feed.self_link)
 # return
 
  # Each worker has its own connection

coldsweat/fetcher.py

@@ -54,7 +54,7 @@ def handle_304(self, response):
  '''
  Not modified
  '''
- logger.debug("%s hasn't been modified, skipped" % self.netloc) 
+ logger.debug(u"%s hasn't been modified, skipped" % self.netloc) 
 
  raise HTTPNotModified
 
@@ -65,7 +65,7 @@ def handle_410(self, response):
  self.feed.is_enabled = False
  self.feed.error_count += 1 
 
- logger.warn("%s is gone, disabled" % self.netloc)
+ logger.warn(u"%s is gone, disabled" % self.netloc)
  self._synthesize_entry('Feed has been removed from the origin server.')
 
  raise HTTPGone
@@ -80,14 +80,14 @@ def handle_301(self, response):
  Feed.get(self_link=self_link)
  except Feed.DoesNotExist:
  self.feed.self_link = self_link 
- logger.info("%s has changed its location, updated to %s" % (self.netloc, self_link))
+ logger.info(u"%s has changed its location, updated to %s" % (self.netloc, self_link))
  else:
  self.feed.is_enabled = False
  self.last_status = DuplicatedFeedError.code
  self.feed.error_count += 1 
 
  self._synthesize_entry('Feed has a duplicated web address.')
- logger.warn("new %s location %s is duplicated, disabled" % (self.netloc, self_link)) 
+ logger.warn(u"new %s location %s is duplicated, disabled" % (self.netloc, self_link)) 
 
  raise DuplicatedFeedError #HTTPMovedPermanently
 
@@ -107,7 +107,7 @@ def handle_200(self, response):
 
  def fetch_feed(self):
 
- logger.debug("fetching %s" % self.netloc)
+ logger.debug(u"fetching %s" % self.netloc)
 
  # Check freshness
  for value in [self.feed.last_checked_on, self.feed.last_updated_on]:
@@ -118,7 +118,7 @@ def fetch_feed(self):
  # deal with large seconds values
  delta = datetime_as_epoch(self.instant) - datetime_as_epoch(value) 
  if delta < config.fetcher.min_interval:
- logger.debug("%s is below minimun fetch interval, skipped" % self.netloc)
+ logger.debug(u"%s is below minimun fetch interval, skipped" % self.netloc)
  return 
 
  try:
@@ -130,7 +130,7 @@ def fetch_feed(self):
  # Record any network error as 'Service Unavailable'
  self.last_status = HTTPServiceUnavailable.code
  self.feed.error_count += 1 
- logger.warn("a network error occured while fetching %s, skipped" % self.netloc)
+ logger.warn(u"a network error occured while fetching %s, skipped" % self.netloc)
  return
 
  self.feed.last_checked_on = self.instant
@@ -147,7 +147,7 @@ def fetch_feed(self):
  except (HTTPError, HTTPNotModified, DuplicatedFeedError):
  return # Bail out
  except AttributeError:
- logger.warn("%s replied with status %d, aborted" % (self.netloc, response.status_code))
+ logger.warn(u"%s replied with status %d, aborted" % (self.netloc, response.status_code))
  return
  finally:
  self.feed.save()
@@ -166,7 +166,7 @@ def parse_feed(self, data):
  soup = feedparser.parse(data) 
  # Got parsing error?
  if hasattr(soup, 'bozo') and soup.bozo:
- logger.debug("%s caused a parser error (%s), tried to parse it anyway" % (self.netloc, soup.bozo_exception))
+ logger.debug(u"%s caused a parser error (%s), tried to parse it anyway" % (self.netloc, soup.bozo_exception))
 
  ft = FeedTranslator(soup.feed)
 
@@ -187,21 +187,21 @@ def parse_feed(self, data):
  guid = t.get_guid(default=link)
 
  if not guid:
- logger.warn('could not find GUID for entry from %s, skipped' % self.netloc)
+ logger.warn(u'could not find GUID for entry from %s, skipped' % self.netloc)
  continue
 
  timestamp = t.get_timestamp(self.instant)
  content_type, content = t.get_content(('text/plain', ''))
 
  # Skip ancient entries 
  if (self.instant - timestamp).days > config.fetcher.max_history:
- logger.debug("entry %s from %s is over maximum history, skipped" % (guid, self.netloc))
+ logger.debug(u"entry %s from %s is over maximum history, skipped" % (guid, self.netloc))
  continue
 
  try:
  # If entry is already in database with same GUID, then skip it
  Entry.get(guid=guid)
- logger.debug("duplicated entry %s, skipped" % guid)
+ logger.debug(u"duplicated entry %s, skipped" % guid)
  continue
  except Entry.DoesNotExist:
  pass
@@ -233,7 +233,7 @@ def fetch_icon(self):
  self.feed.icon = self._google_favicon_fetcher(self.feed.alternate_link or self.feed.self_link)
  self.feed.icon_last_updated_on = self.instant
 
- logger.debug("saved favicon %s..." % (self.feed.icon[:70])) 
+ logger.debug(u"saved favicon %s..." % (self.feed.icon[:70])) 
  self.feed.save()
 
 
@@ -246,7 +246,7 @@ def _google_favicon_fetcher(self, url):
  try:
  response = fetch_url(endpoint)
  except RequestException, exc:
- logger.warn("could not fetch favicon for %s (%s)" % (url, exc))
+ logger.warn(u"could not fetch favicon for %s (%s)" % (url, exc))
  return Feed.DEFAULT_ICON
 
  return make_data_uri(response.headers['Content-Type'], response.content)
@@ -278,7 +278,7 @@ def add_synthesized_entry(self, title, content_type, content):
  last_updated_on = self.instant
  )
  entry.save()
- logger.debug("synthesized entry %s" % guid) 
+ logger.debug(u"synthesized entry %s" % guid) 
  return entry
 
 
@@ -301,7 +301,7 @@ def get_timestamp(self, default):
  if value:
  # Fix future dates
  return min(tuple_as_datetime(value), default)
- logger.debug('no feed timestamp found, using default') 
+ logger.debug(u'no feed timestamp found, using default') 
  return default
 
  # Nullable fields
@@ -341,7 +341,7 @@ def get_timestamp(self, default):
  if value:
  # Fix future dates
  return min(tuple_as_datetime(value), default)
- logger.debug('no entry timestamp found, using default') 
+ logger.debug(u'no entry timestamp found, using default') 
  return default
 
  def get_title(self, default):
@@ -364,7 +364,7 @@ def get_content(self, default):
  if candidates:
  return candidates[0].type, candidates[0].value
 
- logger.debug('no entry content found, using default') 
+ logger.debug(u'no entry content found, using default') 
  return default
 
  # Nullable fields
@@ -432,7 +432,7 @@ def fetch_url(url, timeout=10, etag=None, modified_since=None):
  try:
  response = requests.get(url, timeout=timeout, headers=request_headers)
  except RequestException, exc:
- logger.debug("tried to fetch %s but got %s" % (url, exc.__class__.__name__))
+ logger.debug(u"tried to fetch %s but got %s" % (url, exc.__class__.__name__))
  raise exc
 
  return response