YAML specification

This page outlines the specification for valid Executor YAML files.

Such YAML configurations can be used in a Flow via Flow().add(uses='exec.yml'), or loaded directly via Executor.load_config('exec.yml').

Note that Executor YAML configurations always refer back to an Executor defined in a Python file.

Example

The following constitutes an example Executor configuration:

jtype: MyExecutor
with:
  match_args: {}
py_modules:
  - executor.py
metas:
  name: Indexer
  description: Indexes all documents
  url: https://github.com/janedoe/indexer
  keywords: ["indexer", "executor"]

Keywords

jtype

String specifying the Python type of the Executor. Used to locate the correct class in the Python files given by py_modules.

with

Collection containing keyword arguments passed to the Executor’s __init__() method. Valid values depend on the Executor.

py_modules

List of strings defining the Python dependencies of the Executor. Most notably this must include the Python file that contains the Executor definition itself, as well as any other files imported by this.

metas

Collection containing meta information about the Executor.

Your executor will also be annoted with this information when publishing to Jina Hub. To get better appeal on Jina Hub, you may want to carefully set the metas fields to the correct values:

• name: Human-readable name of the Executor.

• description: Human-readable description of the Executor.

• url: URL of where to find more information about the Executor, normally a GitHub repo URL.

• keywords: A list of strings to help users filter and locate your package.