WeChat Bot Skeleton - Python version

Introduction

A WeChat (and Weixin) chatbot skeleton in Python with queue/delayed messages support.

This application was written following up on the Building Chatbots For WeChat — Part #1 article and in anticipation of Part #2.

Configuration

In order to successfully run this application, the following environment variables must be set:

Variable name	Description
REDIS_URL	A Redis connection string.
WECHAT_TOKEN	See: WeChat docs
WECHAT_APPID	Provided by the WeChat admin console.
WECHAT_APPSECRET	Provided by the WeChat admin console.

When running the application on the local machine for testing, a .env can be used instead of environment variables. See Running locally for details.

Running locally

Ensure that a working instance of Redis is available, either on the local machine or on a remote server.

Clone the repository and enter the directory:

Install the required dependencies running the following command from the project root:

$ pip install -r requirements.txt

Rename the .env.template file to .env, edit it and set the required parameters. Alternatively, set the environment variables listed in .env.template. The .gitignore file is already configured to ignore the .env file upon commits and pushes.

Run the bot:

$ gunicorn -b 127.0.0.1:8000 src.bot

then configure WeChat to forward messages to it. The Building Chatbots For WeChat — Part #1 article contains information on how to setup the bot on WeChat and use tunneling tools such as ngrok in case a public IP address to connect the server to the local machine is not available.

Set up a cron job to run the queue worker automatically, or do it manually when needed during development and testing:

$ python src/worker.py

Deploying to Heroku

In case you are not familiar with the Heroku platform, please read the Getting Started on Heroku with Python article first.

Ensure that the Heroku CLI is installed and configured with the proper credentials, then create a new app on Heroku to host the bot code.

Provision the required Heroku Redis and Heroku Scheduler add-on elements:

$ heroku addons:create heroku-redis:hobby-dev -a {HEROKU_APP_NAME}
$ heroku addons:create scheduler:standard -a {HEROKU_APP_NAME}

This applications has been also tested with Redis Cloud.

Set all the required environment variables:

$ heroku config:set REDIS_URL="redis://{USERNAME}:{PASSWORD}@{HOST}:{PORT"" -a {HEROKU_APP_NAME}
$ heroku config:set WECHAT_TOKEN={WECHAT_SHARED_TOKEN} -a {HEROKU_APP_NAME}
$ heroku config:set WECHAT_APPID={WECHAT_APP_ID} -a {HEROKU_APP_NAME}
$ heroku config:set WECHAT_APPSECRET={WECHAT_APP_SECRET} -a {HEROKU_APP_NAME}

Heroku Redis automatically sets the REDIS_URL variable, it might be necessary to set it when using different Redis add-ons.

Enter the project root and add a new git remote pointing to the Heroku repository (see also: Creating a Heroku remote):

$ heroku git:remote -r heroku -a {HEROKU_APP_NAME}

Push the code repository to Heroku:

$ git push heroku master

For more information about deplyoing to Heroku with Git, see: Deploying with Git

Get the app URL:

$ heroku info -a {HEROKU_APP_NAME}

then configure WeChat to forward messages to it as described in Running locally.

Tests

This application uses pytest. Run:

$ pytest

from the project root to run tests.

WARNING

The Queue Manager implements the FIFO pattern, so in order for tests to succeed, they MUST be ran against a private Redis instance where no other processes are running I/O operations. Otherwise, another process could alter the queue by pushing and pulling other messages, thus making the test fail.

You can specify a different Redis connection string at runtime:

$ REDIS_URL={test_redis_instance_URL} pytest