Extract data from csv in RASA

Rasa Open Source
2

You can use custom actions.

In your custom action, you can connect to a database or read a CSV file like in any other Python code.

If you use slots, you can get the value of the slot in the custom action by using tracker.get_slot(slot_name).

You do your stuff with the CSV as with any other Python code and get the output, build a reply based on the output, and use the dispatcher.utter_message(reply) function to make the chatbot utter the message.

For example, in the run() method of your custom action:

# get the location slot
location = tracker.get_slot('location')

# read the CSV file
with open('universities.csv', 'r') as file:
    reader = csv.DictReader(file)
    # get a list of universities in the desired location
    output = [row for row in reader if row['Location'] == location]

if output: # there is at least one value
    # build your reply according to the output
    reply  = f"This is a list of universities in {location}:"
    reply += "\n- " + "\n- ".join([item['Name'] for item in output])
    # utter the message
    dispatcher.utter_message(reply)

else: # the list is empty
    dispatcher.utter_message(f"I could not find universities in {location}")