bot-express is a chatbot development framework to build tailer-made chatbot lightning fast. Developers can extend the chatbot capability just by adding skills.
- NLU integration.
- Context aware behavior.
- Auto parameter collection based on skill.
- Auto language translation.
- Support multiple messengers.
- Available as npm package.
A chatbot based on bot-express is composed of following components.
- Messenger
- NLU
- Bot instance（Node.js app based on bot-express）
Developers can extend the chatbot capability just by adding skills. 1 skill is simply composed by 1 script file. Developers can make chatbot more painstaking by creating polished skill and make it more capable by adding wide variety of skills.
The very basic workflow of bot-express based chatbot is following.
- A user sends message to a chatbot.
- bot-express forwards the message to NLU and identifies the intent of the message.
- bot-express pickup a skill corresponding to the identified intent.
- The skill is executed. ex: Reply to the user, Update the database, Send signal to IoT devices.
bot-express continues interaction with the user until it accomplish the mission defined by the skill. The interaction is conducted to respond/collect required information. Developers can configure the parser and reaction for every single parameters to collect and they are automatically applied to each messages from the user.
bot-express can be installed by installing a npm package just like below.
$ npm install --save bot-express
Running through the tutorial is a fastest way to learn bot-express since it covers the most important configurations step by step.
Tutorial: Create pizza delivery reception bot using bot-express
Also take a glance at sample_skill directory which contains some sample skills.
As for the complete configurations, spec of the skill script and API, please refer to the following document.
https://nkjm.github.io/bot-express
Set environment variable DEBUG to "bot-express:*" to activate full debugging. In production environment, setting "bot-express:skill-status" is recommended. To make this debugging work properly when you use redis as memory store, you need to enable Redis Keyspace Notification like following to subscribe expired event.
$ redis-cli config set notify-keyspace-events Ex
Also need to enable keyspace_notification of memory option in index.js like follows.
memory: {
type: "redis",
options: {
url: process.env.REDIS_URL,
keyspace_notification: true // default is false.
}
}Webhook supports following event at present.
LINE
- message
- follow
- unfollow
- join
- leave
- postback
- beacon
- messages
- messaging-postbacks
If you deploy cluster of bot-express based application, you need to use redis as context store. Please refer to the document for detail.