Dialogflow-chatbot-tester

Disclaimer
This test is suitable for chatbots that walk through a decision tree.

Requirements

1. python 3.6 or above
2. numpy
3. pandas
4. requests

Getting Ready

1. You need to first build csv formatted file(s) that represents your decision tree.
   The first line of the CSV file must contain the following header:
   Questions,Yes,No,Clarification,Unique Identifier
   You can take a closer look on how the form looks like and how it can be parsed and used from chatbot-builder from tree

2. For Unit Testing, place part of your CSV tree under tree folder.
   In my case, under tree folder, I have MHRA.csv, NICE.csv and NHSD.csv (named as <organisation_name>.csv ).

3. For Integration Testing, place the integrated CSV file under the tree folder.

4. Get your chatbot's back-end running.
   To test our chatbot, Clone Chatbot Backend and run the server.
   This tester will send a POST request to inputText API and retrieve the bot's response.
   You must finish all the steps to run your Dialogflow chatbot (e.g. filling up your hidden .env file with Dialogflow credentials).

Getting Started

1. If you are running our project, simply run test.py and left points below as options. Otherwise, start from step 2.

2. Navigate to QandA.py and modify various data as you want. The more data you put in to the dictionaries, the more the generated conversations will be randomised, thus adding more stress to your chatbot.

3. Navigate to generate.py and place appropriate CSV file name in GenerateConversation().generate_body() and check if GenerateConversation().generate() works. You should see that target directory, which contains randomly generated convos as csv, in your root folder.

4. Navigate to test.py and place appropriate CSV file name in files list and run test.py. You can see SUCCESS or FAIL on your terminal for each test case.

Extensibility

1. Include DB checking into testing
2. Generate an HTML report
3. Make this repository as a package

Author

• To Eun Kim

License

This project is licensed under the MIT License. Check LICENSE.md for more details.