DataFlow

The following documentation explains how to configure and deploy DataFlows via the HTTP service.

HttpChannelConfig

• asynchronous: bool

  • Unused right now but will be accessible over a websocket in the future and used for long running flows.

• dataflow : DataFlow

  • Flow to which inputs from request to path is forwarded too.

• input_mode : str

  • Mode according to which input data is passed to the dataflow, default:default.

    • default

      Inputs are expected to be mapping of context to list of input to definition mappings eg:

      {
          "insecure-package": [
              {
                  "value": "insecure-package",
                  "definition": "package"
              }
          ]
      }
      

    • preprocess:definition_name

      • Input as whole is treated as value with the given definition after preprocessing. Supported preprocess tags : [json,text,bytes,stream]

• path : str

  • Route in server.

• output_mode : str

  • Mode according to which output from dataflow is treated.

    • bytes:content_type:OUTPUT_KEYS

      • OUTPUT_KEYS are . separated string which is used as keys to traverse the ouput of the flow. eg:

        results = {
            "post_input": {
                "hex": b'speak'
            }
        }
        

        then bytes:post_input.hex will return b'speak'.

    • text:OUTPUT_KEYS

    • json

      • Output of dataflow (Dict) is passes as json

• immediate_response: Dict[str,Any]

  • If provided with a response, server responds immediately with it, whilst scheduling to run the dataflow. Expected keys:

    • status: HTTP status code for the response

    • content_type: MIME type, if not given, determined from the presence of body/text/json

    • body/text/json: One of this according to content_type

    • headers

    {
        "status": 200,
        "content_type": "application/json",
        "data": {"text": "ok"}
    }