Architecture of Autocomplete

[Istar-Eldritch]

• Separation of concerns
• Interfaces for request & Response
• Errors and exceptions

[Xhark-Blues]

- Separation of concerns:

It will be separated into two services:

• Huginn: It will implement the logic to resolve what should be autocompleted base on the current state of the text in the search box.

• Muninn: It will storage all the vocabularies and reply with a list of possible words within each vocabulary base on a string. The implementation of the autocompletion method will be configurable.

  • Vocabularies identified so far:
    • Diseases
    • Assay type
    • Technology
    • Tissue type

- Interfaces for Request & Response:

It will work as a single service and respond to the following pattern and payload:

• Muninn:

  • Interface:
    • action.{vocabulary}.autocomplete
    • action.autocomplete
    • Request:
      • payload: {prefix: string, limit?:number}
    • Response:
      • {vocabulary:string[], vocabulary:string[], ...}

• Huginn:

  • Interface:
    • action.sugest
    • Request:
      • payload: {tree: Node}
    • Response:
      • string[]

- Errors and exceptions:

• Muninn:
  • Lists not created on redis.
  • Wrong payload.
• Huginn:
  • No node flagged as autocomplete:true on the tree.
  • Wrong payload.

[Istar-Eldritch]

Can we identify potential errors and exceptions?

[Istar-Eldritch]

@Xhark-Blues we need to add payload validation and error handling to the service.