/Falcon

Email productivity.

Primary LanguagePythonMIT LicenseMIT

Rule based email processing.

How to set it up
  • Run pip install -r requirements.txt
  • Build the desktop_credentials.json file for GMail API.
    • Follow these steps to build the credentials.
    • Move it inside config/.
  • Add emails.json inside data/, it should look like data/emails_templates.json
  • Copy data/rules_sample.csv to data/rules.csv to begin with.
  • Run python/manage.py --update_rules to push rules from data/rules.csv to SQL db.
  • Run python/cleanup.py
Adding more rules to rules.csv

  • The rules csv needs below fields

    • type - If given query applies to true, this command type will be applied.
      • blacklist - Move to trash.
      • whitelist - Falcon will skip this email.
      • label:-<label-name> - Specified label will be removed from email.
      • label:+<label-name> - Specified label will be added to the email. A new label is created if one does not exist already.
    • query - A python expression that needs to evaluate to true for this rule to be applied. Some variables available for us to work with
      • sender - Email ID.
      • labels - Label names associated with given email.
      • tags - This list of tags is populated by Falcon.
        • unsubscribe - Added when Falcon finds an option to unsubscribe from this email, in header or content.
      • subject - Email subject.
      • snippet - Email snippet.
      • text - Email body text.
      • subject_snippet - Email subject plus snippet.
      • content - Entire email content in text.
      • timediff - How old is the email, in seconds.
      • day - Number of seconds in a day.
      • week - Number of seconds in a week.
    • apply_to - A coma separated list of email ids or all.
    • order - Order in which rules will be evaluated for a given email.
    • args - Miscellaneous arguments, coma separated.
      • skip_others - If current rule evaluates to true, other rules will be skipped.

  • Some sample rules

    type query order apply_to id args
    blacklist timediff > day and any(i in labels for i in ['unsubscribe']) 10001 all 1 nan
    label:+unsubscribe 'unsubscribe' in tags 2 all 2 nan
    label:-important True 3 all 3 nan
    label:-unread any(i in labels for i in ['unsubscribe', 'groceries', 'order', 'notification', 'otp', 'investment', 'transaction', 'statement']) 10000 all 4 nan
    whitelist 'starred' in labels 1 all 5 nan