PyCon Open Spaces Bot

Below you'll find a brief description of how the bot will function and information about the Slack channels and Django Admin that we will be using to control the bot.

The bot will be listening for the word or hasthtag pyconopenspaces to occur in a tweet. If this word is found in a tweet the bot then tries to extract a time and room number from the text. If exactly one room and one time mention are found the bot will then schedule a retweet to be sent out approx. 15 minutes before the time mentioned in the tweet.

Ways to monitor the bot on Slack

There is a slack team setup that will help us monitor the bot during the conference. In this team there are three channels that provide information about the tweets the bot receives and what the bot is planning on doing.

Slack Channels

• outgoing_tweets - This channel will receive a message from the bot anytime that it has scheduled a tweet to be sent. Each message in the slack channel will contain the tweet, user's name, and user's Twitter id.

• need_review - This channel will receive a message anytime a tweet is found with some relevant information but not exactly the one room and one time mention needed to add the tweet to the bot's outgoing tweets. With the information contained in these slack messages we can manually create the retweet through the Django Admin if needed.

• event_conflict - This channel will receive a message anytime a tweet is found where the time and room number exactly match that of an event the bot has already added to its outgoing tweets. This conflict check is done to have the bot only send tweets about an event one time and not every time a specific Open Spaces event is mentioned on Twitter. This helps us weed out retweets an original sender might recieve when promoting their Open Space.

Managing bot with Django Admin

We will be using the Django Admin as an easy way to have control over what the bot is planning on sending. Below is a description of each model in the Open Spaces Admin and the things it controls.

• outgoing config - This model controls configuration information about the bot and how it functions. Changing information in the most current object in this model lets us change how the bot functions without it needing a restart.

  • The auto_send field is a boolean that gives control over whether the bot will automatically stage tweets to be sent without needing additional approval.
  • The default_send_interval field is the number of minutes before an event that the bot will tweet its reminder.
  • The ignore_users field is an array field that holds a list of twitter ids of ignored users. You can add users to this list by updating the user's record in the User model.

• User - This model holds information about the users who have had a tweet picked up by the bot.

  • should ignore - If you set this field to True and then save the object in the Admin the user's twitter id will be automatically added to the ignore users array field in the outgoing config model and the bot will ignore all future tweets from the user.

• Outgoing Tweets - This model contains a record of all the tweets that the bot is planning on sending. Below is a brief description of some of the fields that control what tweets the bot will send. Important note: If a tweet is picked up for an event that is going to occur within the next 30 minutes it will need to be manually approved. This ensures that we have the opportunity to see and act on all tweets before they're sent.

  • approved - This is a choice field that determines whether or not a tweet will be picked up and sent out. The three choice available are Approved, Denied, and Needs_action. Only those tweets with this field set to Approved will be sent.
  • scheduled time - This datetime field is used to decided when a tweet will be sent. If you decide to change the sending time manually make sure the outgoing message in the tweet field reflects the change in time.
  • tweet - This field is the actual tweet that the bot will send out.
  • original tweet - This field contains the original tweet but it is not the one the bot will send as a reminder. By default 15 minutes before an event the bot will retweet and embedded tweet that contains the user's original tweet and a message about the event coming up in 15 minutes.

• Openspaces Events - This model controls which events the bot has already been scheduled to tweet about. When a new event is found that the bot is planning on tweeting about a record is created in this model that has information about the room, time, creator, and tweet. This model is then used to check that the bot only tweets about an event once. Without this check the bot would tweet out any retweets of the original message about the Open Space.

• Streamed Tweets - This model contains all of the tweets that the bot has received. This table is used as a record of the tweets the bot has seen but doesn't control what the bot is planning on sending.

Instructions on how to manually stage a tweet

Sorry these steps are a little clunky, we didn't quite get around to adding a custom admin action to add tweets from the streamed tweets model to the outgoing tweets model.

If you see a tweet in the need_review Slack channel that you would like to add please follow the steps below.

In order to follow the format of the bot's other tweets you'll need three pieces of information.

• tweet's id
• sender's user name
• time the tweet needs to be sent

Both the tweet id and user name can be found in the need_review slack channel or in the streamed tweets model.

To format a tweet in the same way that the bot currently sends tweets see the example below:

• https://twitter.com/{user_name}/status/{tweet_id} - this is how a url must be formatted in a tweet to send an embedded copy of the users original tweet

• Example tweet: Coming up in 15 minutes! https://twitter.com/FakeUserName/status/865246820123469952

In the example above you can see that you're able to add a message about the embedded tweet, please make sure not to exceed the character limit for twitter when doing so. When manually adding a tweet it should be saved to the tweet field in the outgoing tweets model. The original tweet field is normally a copy of the user's original tweet without an embedded url and does not effect what the bot sends. In this case feel free to leave a note about why you needed to manually add the tweet in the original tweet field.

The time that the tweet needs to be sent is left up to us to figure out. All of the times on the server are in UTC which is 7 hours ahead of Portland time. Please take this difference into account when manually saving the time that a tweet should be sent. Later on in the day this could mean saving a tweet to be sent out tomorrow in UTC time. After the tweet is saved, the list view of the admin will convert the tweets outgoing UTC time and display it in local Portland time. This should allow us to visually check tweets in the outgoing tweets model quickly and offer an easy way to confirm that we entered the correct UTC time.