Skip to main content

Customized deviceShifu data handler

Shifu allows user to customize the deviceShifu to make the digital twin more efficient.

Default scenario​

By default, data coming from device to deviceShifu will be available in its original format:

With customized processing​

After adding customized logic, deviceShifu can process the data to make them more app-friendly:

  1. convert the data into the format needed by user;
  2. filter out unneeded data;
  3. some calculation and data analysis on-the-fly

Add customized deviceShifu​

Here are 3 steps before we start running Shifu:

  1. Write the data processing logic in customized_handlers.py, a general structure of the processing method is like this: 
    # use instruction/API name as the method name
    def humidity(raw_data): 
        new_data = process(raw_data)
        return new_data
  2. Move customized_handlers.py into pkg/deviceshifu/pythoncustomizedhandlers
  3. Build deviceShifu

Example: humidity detector​

The example device used here is in the directory examples/deviceshifu/customized/humidity_detector.

This device is a virtual humidity detector which will provide the humidity and temperature data in JSON format using HTTP, the instruction/API is just humidity.

0. Build the humidity-detector virtual device​

First step is to build the docker image of humidity-detector with humidity-detector.go and its Dockerfile in examples/deviceshifu/customized/humidity_detector.

1. Add the customized data handler​

In examples/deviceshifu/customized/humidity_detector/pythoncustomizedhandlers, there is a customized_hanlders.py, copy that file into pkg/deviceshifu/pythoncustomizedhandlers.

2. Edit Dockerfile of deviceShifu​

Copy examples/deviceshifu/customized/humidity_detector/sample_deviceshifu_dockerfiles/Dockerfile.deviceshifuHTTP to dockerfiles.

3. Add mapping of data handlers​

In examples/deviceshifu/customized/humidity_detector/configuration,there is a deviceshifu-humidity-detector-configmap.yaml,Map your device commands to the funcName of your data processing program.

(e.g.: instructions is '/humidity_custom' and funcName is 'humidity')

Then you need to set humidity_custom: humidity under customInstructionsPython and set device commands under instructions.instructions and telemetries.telemetries.device_health.properties.instructionInstructions

As follows:

data:
  customInstructionsPython: |
    humidity_custom: humidity 
    #humidity_custom is instructions，humidity is the handler funcName
  instructions: |
    instructions:
      humidity_custom:
  telemetries: |
    telemetries:
      device_health:
        properties:
          instruction: humidity_custom

4. Build deviceShifu docker image​

A new deviceShifu image needs to be built to include the customized data handler.

In the root directory of shifu, run the following command to build the deviceShifu docker image:

make buildx-build-image-deviceshifu-http-http

5. Start Shifu​

This will be exactly the same as in quick install.

After Shifu is running, we need to import the newly built humidity detector virtual device into the kind cluster, like this:

kind load docker-image humidity-detector:v0.0.1

6. Check processed data​

The raw data from this virtual device should be processed by the customized handler defined in customized_hanlders.py.

In the nginx shell we will be able to curl the API humidity and check if the result is processed. An example result should belike this:

[
  {
  "unit": "℃", 
  "code": "20990922009", 
  "exception": "temperature is too high", 
  "name": "atmosphere temperature", 
  "val": "37"
  }, 
  {
    "unit": "%RH", 
    "code": "20990922009", 
    "exception": "humidity is too high", 
    "name": "atmosphere humidity", 
    "val": "88"
  }
]