bitbot-3.11-fork/modules/rss.py

148 lines
5.2 KiB
Python
Raw Normal View History

2019-07-08 12:46:12 +00:00
import time
2019-06-23 15:03:15 +00:00
from src import ModuleManager, utils
import feedparser
RSS_INTERVAL = 60 # 1 minute
def _format_entry(feed_title, entry):
title = entry["title"]
author = entry.get("author", None)
author = " by %s" % author if author else ""
link = entry.get("link", None)
link = " - %s" % link if link else ""
feed_title_str = "%s: " % feed_title if feed_title else ""
return "%s%s%s%s" % (feed_title_str, title, author, link)
2019-06-23 15:03:15 +00:00
2019-07-04 10:15:01 +00:00
@utils.export("botset", utils.IntSetting("rss-interval",
"Interval (in seconds) between RSS polls", example="120"))
2019-06-23 15:03:15 +00:00
class Module(ModuleManager.BaseModule):
2019-06-24 19:23:36 +00:00
_name = "RSS"
2019-06-23 15:03:15 +00:00
def on_load(self):
2019-07-04 10:15:01 +00:00
self.timers.add("rss", self.bot.get_setting("rss-interval",
RSS_INTERVAL))
2019-06-23 15:03:15 +00:00
@utils.hook("timer.rss")
def timer(self, event):
2019-07-08 12:46:12 +00:00
start_time = time.monotonic()
self.log.trace("Polling RSS feeds")
2019-06-23 15:03:15 +00:00
event["timer"].redo()
hook_settings = self.bot.database.channel_settings.find_by_setting(
"rss-hooks")
hooks = {}
for server_id, channel_name, urls in hook_settings:
server = self.bot.get_server_by_id(server_id)
if server and channel_name in server.channels:
channel = server.channels.get(channel_name)
for url in urls:
if not url in hooks:
hooks[url] = []
hooks[url].append((server, channel))
2019-06-23 15:03:15 +00:00
pages = utils.http.request_many(hooks.keys())
2019-06-23 15:03:15 +00:00
for url, channels in hooks.items():
if not url in pages:
# async url get failed
continue
feed = feedparser.parse(pages[url].data)
feed_title = feed["feed"].get("title", None)
2019-06-23 15:03:15 +00:00
entry_formatted = {}
for server, channel in channels:
seen_ids = channel.get_setting("rss-seen-ids-%s" % url, [])
2019-06-23 15:03:15 +00:00
new_ids = []
2019-06-23 15:22:45 +00:00
valid = 0
for entry in feed["entries"][::-1]:
2019-06-23 15:03:15 +00:00
if entry["id"] in seen_ids:
new_ids.append(entry["id"])
continue
2019-06-23 15:22:45 +00:00
if valid == 3:
2019-06-23 15:03:15 +00:00
continue
valid += 1
if not entry["id"] in entry_formatted:
output = _format_entry(feed_title, entry)
entry_formatted[entry["id"]] = output
else:
output = entry_formatted[entry["id"]]
self.events.on("send.stdout").call(target=channel,
module_name="RSS", server=server, message=output)
new_ids.append(entry["id"])
channel.set_setting("rss-seen-ids-%s" % url, new_ids)
2019-06-23 15:03:15 +00:00
2019-07-08 12:46:12 +00:00
total_milliseconds = (time.monotonic() - start_time) * 1000
self.log.trace("Polled RSS feeds in %fms", [total_milliseconds])
def _check_url(self, url):
try:
data = utils.http.request(url)
feed = feedparser.parse(data.data)
except Exception as e:
self.log.warn("failed to parse RSS %s", [url], exc_info=True)
feed = None
if not feed or not feed["feed"]:
return None
return [entry["id"] for entry in feed["entries"]]
2019-06-23 15:03:15 +00:00
@utils.hook("received.command.rss", min_args=1, channel_only=True)
def rss(self, event):
"""
:help: Modify RSS/Atom configuration for the current channel
:usage: list
:usage: add <url>
:usage: remove <url>
:permission: rss
"""
changed = False
message = None
rss_hooks = event["target"].get_setting("rss-hooks", [])
subcommand = event["args_split"][0].lower()
if subcommand == "list":
event["stdout"].write("RSS hooks: %s" % ", ".join(rss_hooks))
elif subcommand == "add":
if not len(event["args_split"]) > 1:
raise utils.EventError("Please provide a URL")
url = utils.http.url_sanitise(event["args_split"][1])
2019-06-23 15:03:15 +00:00
if url in rss_hooks:
raise utils.EventError("That URL is already being watched")
seen_ids = self._check_url(url)
if seen_ids == None:
raise utils.EventError("Failed to read feed")
event["target"].set_setting("rss-seen-ids-%s" % url, seen_ids)
2019-06-23 15:03:15 +00:00
rss_hooks.append(url)
changed = True
message = "Added RSS feed"
elif subcommand == "remove":
if not len(event["args_split"]) > 1:
raise utils.EventError("Please provide a URL")
url = utils.http.url_sanitise(event["args_split"][1])
2019-06-23 15:03:15 +00:00
if not url in rss_hooks:
raise utils.EventError("I'm not watching that URL")
rss_hooks.remove(url)
changed = True
message = "Removed RSS feed"
else:
raise utils.EventError("Unknown subcommand '%s'" % subcommand)
if changed:
if rss_hooks:
event["target"].set_setting("rss-hooks", rss_hooks)
else:
event["target"].del_setting("rss-hooks")
2019-06-23 15:03:15 +00:00
event["stdout"].write(message)