sa-debug-viewer

This is a command-line debug viwer for Sensors Analytics .

Install

Install using npm.

$ npm install -g sa-debug-viewer

DebugEventStream

DebugEventStream is a standard readable stream in object mode. It connects to SensorsData debug viewer and yield parsed message.

import DebugMessageStream from 'sa-debug-viewer'

const messageStream = new DebugMessageStream('http://my-project.cloud.sensorsdata.cn')
messageStream.on('data', (message) => console.log(message))

Message Object

The following object is the message that yield by DebugMessageStream

{
  "project": "default",
  "dryRun": true,
  "message": {
    "time": 1469086165353,
    "properties": {      
      "$lib_version": "0.0.13",
      "env": "qa",
      "page": "home-page",
      "$lib": "node",
      "$city": "北京",
      "$province": "北京",
      "$country": "中国"
    },
    "lib": {
      "$lib_version": "0.0.13",
      "$lib": "node",
      "$lib_method": "code"
    },
    "type": "track",
    "distinct_id": "23451896",
    "event": "page_view"
  }
}

Rx Observable

DebugMessageStream provides toObservable method that converts DebugMessageStream into a Rx.Observable

For more detail, checkout Microsoft's Rx documentation

import DebugMessageStream from 'sa-debug-viewer'

const messageStream = new DebugMessageStream('http://my-project.cloud.sensorsdata.cn')
messageStream.toObservable()
             .filter((message) => message.project === 'production')
             .filter((message) => message.dryRun)
             .map((message) => message.message)
             .subscribe(
               (event) => console.log('Event: ', event),
               (err) => console.error('Error: ', err),
               () => console.log('Completed')
             )

Used in Tracker Automation Test

Testing tracker is tedious and error prone. Especially to refactoring or other massive code modification, tracker is fragile.

To ensure the tracker works as expected, regression test is necessary. But manually test all the tracker is boring and labor-heavy, so it is a good idea to test them with automated tests.

This library can be useful when doing the tracker automation test.

import fetch from 'node-fetch'
import { purchaseApi } from '../apis'
import DebugMessageStream from 'sa-debug-viewer'

describe('Order API', () => {
  it('should track event', async () => {
    const eventPromise = new DebugMessageStream('http://my-project.cloud.sensorsdata.cn')
                          .toObservable()
                          .filter((message) => message.project === 'qa') // Filter message from QA environment
                          .pluck('message') // Expose SA event object
                          .first((message) => message.event === 'order_created' && message.distinct_id === 'u:12345')
                          .timeout(500) // Timeout in 500ms
                          .toPromise()

    await purchaseApi({ user: 'u12345', product: 'p:23232', quantity: 1 }) // Invoke API to trigger remote server

    const trackedEvent = await eventPromise // If timeout, Timeout exception will be thrown here

    expect(trackedEvent).to.have.property('product_id', 'p:23232') // Assert event custom property
    expect(trackedEvent).to.have.property('quantity', 1)
  })
})

CLI Utility

$ sa-debug [options] <host>

Options:
  --project, -p  Project to monitor                                     [string]
  --event, -e    Event filter                                            [array]
  --userId, -u   User id                                                 [array]
  --help         Show help                                             [boolean]

e.g.

Fetch all click event from default project from debug viewer of http://my-project.cloud.sensorsdata.cn

$ sa-debug -e click -p default http://my-project.cloud.sensorsdata.cn

License

MIT