Intended to run as a daemon on the host running Discourse, the bot polls Discourse's latest topics (latest.json) then tweets new topics matching your criteria to your registered Twitter feed.*
Tweets may include a topic thumbnail image, if one exists, and Twitter hashtags & mentions, if those exist. (see Twitter's automation rules at https://help.twitter.com/en/rules-and-policies/twitter-automation#replies-mentions).
The file settings.ini, or environment variables, must be populated with
keys from both Discourse and Twitter. Instructions for creating Twitter keys
can be found at
https://realpython.com/twitter-bot-python-tweepy/#creating-twitter-api-authentication-credentials.
Create a Discourse single-user key at ☰ > Settings > API > New API Key.
Valid setup & desired customizations can be tested before installation.
At the command line, running python3 discourse-new-topic-tweetbot.py should
output the tweet-formatted text of the current most newly created topic, along
with the system path of its thumbnail image, if one exists.
The recommended installation method is sudo pip3 install -r requirements.txt
followed by sudo install discourse_new_topic_tweetbot.py /usr/local/bin.
Of course running from a user's local python packages or venv works as well.
It is left to the user to manage running discourse_new_topic_tweetbot.py
as a daemon (e.g. using Docker, systemd, etc.)
If installed on a different host from the Discourse instance, the bot must be customized to download topic thumbnail images, if those are desired.
To use the included Dockerfile, all required settings from settings.ini
must be passed to docker run using environment variables, e.g.
sudo docker build . -t bot
followed by
sudo docker run -it --env-file settings.ini --entrypoint sh bot
to test settings, and
sudo docker run -d --env-file settings.ini bot
to run as a daemon. To access the Discourse server's images, add a bind mount,
e.g.
-v /var/discourse/shared/standalone:/var/discourse/shared/standalone
The file settings.ini contains authentication credentials and settings.
The default POLLING_INTERVAL is 10 minutes, and TOPIC_REFRESH_INTERVAL
interval is 8 minutes, allowing time for Discourse to generate thumbnails.
TWEET_PREPEND TWEET_HASHTAGS TWEET_MENTIONS will be added to the
beginning of each tweet, before the URL of the Discourse topic.
Inclusion of thumbnail media in tweets may be disabled by setting
TWEET_THUMBNAILS to None or 0.
Further customizations may be made in discourse-new-topic-tweetbot.py. Some
of the best candidates are the build_tweet_string function and the
HTMLMentionsParser class.
Thanks to the developers of Discourse https://www.discourse.org/ and the outstanding Discourse community.
Thanks to the developers of the Discourse & Twitter libraries used, and others:
https://pypi.org/project/discourse/
https://pypi.org/project/tweepy/
https://pypi.org/project/python-decouple/
Thanks also to Evolution Host https://evolution-host.com for provision of a Developer VPS.
* At startup, the bot will poll & wait for new topics to appear in
/latest.json (that is, newer by creation date than the newest topic by
creation date found on the first polling), then tweet those newer topics.
If none among the topics appearing in /latest.json (a 30-item-long list)
on the first polling are actually NEW topics (just recently active topics),
the bot will tweet the NEWEST among them and continue on; technically this
is a design flaw, not a bug.