Matrix-Mirrors/maubot-rss
1
0
Fork 0
mirror of https://github.com/maubot/rss.git synced 2024-10-01 08:25:36 -04:00
Code Issues Packages Projects Releases Wiki Activity

Improve handling of missing keys and invalid feeds

Browse Source
This commit is contained in:
Tulir Asokan 2018-11-27 22:00:50 +02:00
parent 55293815fb
commit 2ee880d9cb
1 changed files with 26 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
rss/bot.py
View File

@ -60,9 +60,8 @@ class RSSBot(Plugin):
await self._poll_feeds() await self._poll_feeds()
except asyncio.CancelledError: except asyncio.CancelledError:
self.log.debug("Polling stopped") self.log.debug("Polling stopped")
pass
except Exception: except Exception:
self.log.exception("Failed to poll feeds") self.log.exception("Fatal error while polling feeds")
async def _broadcast(self, feed: Feed, entry: Entry, subscriptions: List[RoomID]) -> None: async def _broadcast(self, feed: Feed, entry: Entry, subscriptions: List[RoomID]) -> None:
text = f"New post in {feed.title}: {entry.title} ({entry.link})" text = f"New post in {feed.title}: {entry.title} ({entry.link})"
@ -96,7 +95,10 @@ class RSSBot(Plugin):
async def _poll_feeds(self) -> None: async def _poll_feeds(self) -> None:
self.log.debug("Polling started") self.log.debug("Polling started")
while True: while True:
await self._poll_once() try:
await self._poll_once()
except Exception:
self.log.exception("Error while polling feeds")
await asyncio.sleep(self.config["update_interval"] * 60, loop=self.loop) await asyncio.sleep(self.config["update_interval"] * 60, loop=self.loop)
async def read_feed(self, url: str): async def read_feed(self, url: str):
@ -105,11 +107,23 @@ class RSSBot(Plugin):
return feedparser.parse(content) return feedparser.parse(content)
@staticmethod @staticmethod
def find_entries(feed_id: int, entries: List[Any]) -> List[Entry]: def get_date(entry: Any) -> datetime:
try:
return datetime.fromtimestamp(mktime(entry["published_parsed"]))
except (KeyError, TypeError):
pass
try:
return datetime.fromtimestamp(mktime(entry["date_parsed"]))
except (KeyError, TypeError):
pass
return datetime.now()
@classmethod
def find_entries(cls, feed_id: int, entries: List[Any]) -> List[Entry]:
return [Entry( return [Entry(
feed_id=feed_id, feed_id=feed_id,
id=entry.id, id=entry.id,
date=datetime.fromtimestamp(mktime(entry.published_parsed)), date=cls.get_date(entry),
title=entry.title, title=entry.title,
summary=entry.description, summary=entry.description,
link=entry.link, link=entry.link,
@ -129,9 +143,13 @@ class RSSBot(Plugin):
feed = self.db.get_feed_by_url(url) feed = self.db.get_feed_by_url(url)
if not feed: if not feed:
metadata = await self.read_feed(url) metadata = await self.read_feed(url)
feed = self.db.create_feed(url, metadata["channel"]["title"], if metadata.bozo:
metadata["channel"]["description"], await evt.reply("That doesn't look like a valid feed.")
metadata["channel"]["link"]) return
channel = metadata.get("channel", {})
feed = self.db.create_feed(url, channel.get("title", url),
channel.get("description", ""),
channel.get("link", ""))
self.db.add_entries(self.find_entries(feed.id, metadata.entries)) self.db.add_entries(self.find_entries(feed.id, metadata.entries))
self.db.subscribe(feed.id, evt.room_id, evt.sender) self.db.subscribe(feed.id, evt.room_id, evt.sender)
await evt.reply(f"Subscribed to feed ID {feed.id}: [{feed.title}]({feed.url})") await evt.reply(f"Subscribed to feed ID {feed.id}: [{feed.title}]({feed.url})")