# HG changeset patch # User Dirk Olmes # Date 1314176026 -7200 # Node ID 6ea813cfac3313d70d1b635d29eef279d2d2ab8e # Parent e927c6910c387e3d45eca17b0b067d5a441a95c5 pull out common code for updating a feed into an abstract class, have the sqlalchemy backend use that class. diff -r e927c6910c38 -r 6ea813cfac33 backend/AbstractFeedUpdater.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/backend/AbstractFeedUpdater.py Wed Aug 24 10:53:46 2011 +0200 @@ -0,0 +1,54 @@ + +from datetime import datetime +import feedparser +import logging + +STATUS_ERROR = 400 +log = logging.getLogger("FeedUpdater") + +class AbstractFeedUpdater(object): + ''' + Abstract base class for FeedUpdater implementations - handles all the parsing of the feed. + Subclasses need to implement creating and storing the new feed entries. + ''' + + def __init__(self, feed): + self.feed = feed + + def update(self): + log.info("updating " + self.feed.rss_url) + result = self._retrieveFeed() + for entry in result.entries: + self._normalize(entry) + self._processEntry(entry) + self.feed.incrementNextUpdateDate() + + def _retrieveFeed(self): + result = feedparser.parse(self.feed.rss_url) + # bozo flags if a feed is well-formed. +# if result["bozo"] > 0: +# raise FeedUpdateException() + status = result["status"] + if status >= STATUS_ERROR: + raise FeedUpdateException("HTTP status " + str(status)) + return result + + def _normalize(self, entry): + if not hasattr(entry, "id"): + entry.id = entry.link + if not hasattr(entry, "updated_parsed"): + entry.updated_parsed = datetime.today() + else: + entry.updated_parsed = datetime(*entry.updated_parsed[:6]) + if not hasattr(entry, "summary"): + if hasattr(entry, "content"): + entry.summary = entry.content[0].value + else: + entry.summary = "" + + def _processEntry(self, entry): + raise Exception("_processEntry is abstract, subclasses must override") + + +class FeedUpdateException(Exception): + pass diff -r e927c6910c38 -r 6ea813cfac33 backend/sqlalchemy/FeedUpdater.py --- a/backend/sqlalchemy/FeedUpdater.py Wed Aug 24 09:39:10 2011 +0200 +++ b/backend/sqlalchemy/FeedUpdater.py Wed Aug 24 10:53:46 2011 +0200 @@ -1,4 +1,5 @@ +from backend.AbstractFeedUpdater import AbstractFeedUpdater, FeedUpdateException from datetime import datetime from Feed import Feed from FeedEntry import FeedEntry @@ -46,39 +47,18 @@ FeedUpdater(session, newFeed).update() -class FeedUpdater(object): +class FeedUpdater(AbstractFeedUpdater): def __init__(self, session, feed): + AbstractFeedUpdater.__init__(self, feed) self.session = session - self.feed = feed - - def update(self): - log.info("updating " + self.feed.rss_url) - result = self.getFeed() - for entry in result.entries: - self.processEntry(entry) - self.feed.incrementNextUpdateDate() - def getFeed(self): - result = feedparser.parse(self.feed.rss_url) - # bozo flags if a feed is well-formed. -# if result["bozo"] > 0: -# raise FeedUpdateException() - status = result["status"] - if status >= STATUS_ERROR: - raise FeedUpdateException("HTTP status " + str(status)) - return result - - def processEntry(self, entry): - normalize(entry) + def _processEntry(self, entry): feedEntry = FeedEntry.findById(entry.id, self.session) if feedEntry is None: - self.createFeedEntry(entry) + self._createFeedEntry(entry) - def createFeedEntry(self, entry): + def _createFeedEntry(self, entry): new = FeedEntry.create(entry) new.feed = self.feed self.session.add(new) log.info("new feed entry: " + entry.title) - -class FeedUpdateException(Exception): - pass