awardsBot

The NSF awards bot regularly contacts principal investigators with reminders on their project specific deadlines. Please read the schedule readme for more detailed information on automated correspondences.

How the bot works

Every 24 hours the bot queries NSF's award API for newly awarded grants and stores this information in a pre-existing database. When it finds a new award it creates a new ticket in Request Tracker and sends an initial correspondence that outlines project-specific expectations and deadlines. It sends reminders to submit annual reports, submit data for Arctic Observing Network (AON) projects, and that the award is expiring soon. The bot sends a summary of the number of each type of correspondence sent and error messages to a slack channel.

Setup

• Copy the bot cron script to a directory

• Create a file called .Renviron in the same directory as the script. Include the following variables:

  DATABASE_PATH                            # Path to the database of awards and correspondences
  LASTRUN_PATH=LASTRUN                     # Determines where the bot stores its state
  SLACK_WEBHOOK_URL="{URL}"                # Your Slack webhook URL
  SLACK_OAUTH_TOKEN="TOKEN"                # Slack OAUTH token (see your bot admin for access)
  RT_BASE_URL="https://example.com/rt"          # The URL of your RT install
  RT_USER="your_rt_user"                   # Your RT username
  RT_PASSWORD="your_rt_password"               # Your RT password
  INITIAL_ANNUAL_REPORT_OFFSET=8           # Number of months after award startDate to send annaul report reminder
  INITIAL_AON_OFFSET=11                    # Number of months after award startDate to send first AON data due reminder
  AON_RECURRING_INTERVAL=6                 # Number of months to send recurring emails for AON data due
  

• see the server vignette for more details

Running

Run the bot cron script every 24 hours.
Example crontab: 0 15 * * * Rscript ~/home/awardsBot/main_cron_script.R

How the database works

The awards bot uses a csv file as a database. It stores metadata about each award harvested from NSF's award API, along with dates at which specific correspondences should be sent. For instance, the database includes the column contact_3mo, which lists the date when the bot should send a reminder that there are 3 months remaining until an award expires. This value is initialized to 3 months before the expDate field. Each time the bot runs, it checks whether the system date is equivalent to contact_3mo and sends a reminder email if this is true. For recurring correspondences, such as annual report reminders, the database contains previous and next response columns. For example, contact_annual_report_next specifies what date to send the reminder, and when contact_annual_report_previous is equivalent to contact_annual_report_next it will update the latter field forward by one year.

Adding a correspondence

In this example we will assume the submission policies changed to require an initial metadata submission within the first two years of an award startDate. The following steps illustrate how to send a one-time correspondence. In order to send a recurring correspondence copy the logic used by set_first_aon_data_due_date() and update_aon_data_due_date(), in addition to the following steps.

• Add the email template to the inst/emails folder
• Add an appropriately named NA column to the database, such as contact_two_year.
• Add a function to database_functions.R that initializes contact_two_year to two years after startDate
  • Add this to the update_contact_dates wrapper function, which is used in main().
• Add an RT correspondence function to RT_functions.R. Use the existing correspondence functions as templates.
  • Add this to the send_correspondences wrapper function, which is used in main()
• Unit tests!!

Testing the bot

The awards-bot package contains modular unit tests, however, many of these don't run, by default, unless the R session is connected to RT and Slack. A thorough test of the bot would involve signing in to RT and Slack, and running devtools::check(); although, if the test_main unit test passes it's generally safe to assume the more modular tests will pass as well.

General notes:

• only use awardsBot:::test_main() in testing. This a wrapper for awardsBot::main() except with an additional email argument
• set email = your test email address in the unit test script
• modify test_main() calls in the unit test script accordingly, including any additional arguments

Testing Locally

Please follow this closely so no emails are sent to researchers by mistake while testing.

1. Add the following lines to the .Renviron (can be accessed by usethis::edit_r_environ())

  SLACK_WEBHOOK_URL="{URL}"                # Your Slack webhook URL
  SLACK_OAUTH_TOKEN="TOKEN"
  RT_BASE_URL="https://example.com/rt"          # The URL of your RT install
  RT_USER="your_rt_user"                   # Your RT username
  RT_PASSWORD="your_rt_password"               # Your RT password

2. Set up slackbot: slackr::slackr_setup(channel = "#awardbot", username = 'awardbot', incoming_webhook_url = Sys.getenv("SLACK_WEBHOOK_URL"), token = Sys.getenv("SLACK_OAUTH_TOKEN"))

3. Login to RT: rt::rt_login()

4. Run devtools::load_all() to source the scripts for testing so you can run the test file line by line

5. Run the test_main unit test locally, ideally line by line.

6. Two test tickets should be created in RT.

Style

Code generally follows the tidyverse style conventions, with the following specific style preferences:

• underscore for all variable names unless referring to an NSF awards API return field (i.e. expDate, startDate, etc.)

Acknowledgements

Work on this package was supported by:

• The Arctic Data Center: NSF-PLR grant #1546024 to M. B. Jones, S. Baker-Yeboah, J. Dozier, M. Schildhauer, and A. Budden

Additional support was provided by the National Center for Ecological Analysis and Synthesis, a Center funded by the University of California, Santa Barbara, and the State of California.