Skip to content

Latest commit

 

History

History
35 lines (27 loc) · 1.73 KB

CODE_STRUCTURE.md

File metadata and controls

35 lines (27 loc) · 1.73 KB
Raw

Code Structure

This is a brief outline of the Catalyst code structure and data flow. In app/, there are the following directories-

  • models: Fairly standard model specifying the data about the annotators. The annotators are saved in the database for convenience- they do not change at runtime and are loaded from db/seeds.rb where all data about them is specified.

  • controllers: Handles inbound requests to the Catalyst API, either from apps requesting info on annotators or a JSON specifying which annotators to run (as specified above).

  • api: Handles outbound requests from Catalyst to the DocManager API

  • pipeline_setup: The controllers call these methods to setup a series of methods to run based on the JSON specifying which annotators to run. The pipeline setup loops through the annotators and calls the annotator methods with the input parameters specified in the JSON. The annotators return blocks of code (closures including the parameters specified), which are chained together into one big block of code- a "recipe" combining all the annotators Catalyst is supposed to run.

  • annotators: These are the methods Catalyst uses to analyze the documents. Each is a class with a block of code that performs the specified extraction or analysis function on the document. They vary in complexity and purpose. Some are ones we created, some call external lbraries.

  • run_pipeline: This queries DocManager to get the documents Catalyst should run on (using the parameters to filter the document set as needed specified above). It then loops through each document, and calls the whole recipe block on the document, running all Catalyst methods specified. It then saves the results by sending them back to DocManager.