We have a set of actions, each of which takes in some parameters and outputs a value. We’d like to run a particular workflow, which is just a sequence of actions, with a particular set of parameters for each one, and collect their outputs. The output from one action A can be a parameter to another action B, when this is the case, we say that B depends on A, as B cannot run until A has produced its output.
An example workflow is as follows:
- Get the text of the web page ’https://www.example.com/’
- Reverse the string obtained from step 1
- Write the string obtained from step 2 to file “text.txt”
In this scenario, workflows are specified as graphs (DAG) serialized as json, with actions as nodes, and edges representing dependencies. We want to write a function that can take in the DAG and execute the actions in the correct sequence.
An example graph is below -- this graph the same on that's in tests/graph1.json
{
"steps": {
"reverse-text": {
"type": "reverse-string",
"args": {
"string": {
"type": "reference",
"step": "get-text"
}
}
},
"get-text": {
"type": "http-request",
"args": {
"url": {
"type": "value",
"value": "https://www.example.com"
}
}
},
"save-file": {
"type": "save-file",
"args": {
"filename": {
"type": "value",
"value": "/foo.txt"
},
"contents": {
"type": "reference",
"step": "reverse-text"
}
}
}
}
}
Your solution should be able to run the graph above with solution('./tests/graph1.json'), which should produce the following output:
running get-text
running reverse-text
running save-file
Got filename /foo.txt
Got contents !dlrow ,olleH