Implementation: Atomic

Atomic components are implemented by writing a script or program in the language defined by the <launcher> element of component.xml. When the component is executed, it receives a properties file called a command file as an argument that contains values and paths for input files, parameters and output files, as well as various metadata. This properties file is parsed by a language-specific Anduril library, and the values are exposed to the component code.

Provided language-specific libraries and links to their documentation are:

Language Source code folder
Bash lang/bash
Java core/src/main/java/org/anduril/component
Lua lang/lua
Matlab lang/matlab
Octave lang/octave
Perl lang/perl
Python lang/python
R lang/r
Scala core/src/main/scala/org/anduril/runtime

Example: R component

To implement the SimpleCSVFilter component whose interface we defined, we write the following R script (~/my-bundles/demo/components/SimpleCSVFilter/SimpleCSVFilter.r):

library(componentSkeleton)

execute <- function(cf) {

    input <- CSV.read(get.input(cf, 'in'))
    include <- get.parameter(cf, 'include', 'boolean')

    if (input.defined(cf, 'columnNames')) {
        column.names <- CSV.read(get.input(cf, 'columnNames'))[,1]
        if (include) {
            output <- input[, column.names, drop = FALSE]
        } else {
            output <- input[, !(colnames(input) %in% column.names), drop = FALSE]
        }
    } else {
        output <- input
    }

    CSV.write(get.output(cf, 'out'), output)
    return(0)
}

main(execute)

Here we used the componentSkeleton R library that is provided by Anduril (lang/r). It parses the properties files in main into an object called cf and provides read/write functions for the CSV format defined in the tools bundle.