I created this simple web tool to assist in Tableau CRM/Einstein Analytics dataflow development. It allows users to view the datasets being registered within a given dataflow, and then isolate only the nodes required to output a selected dataset(s).
If you're in an organization like mine, you've probably seen some wicked large dataflows. We're talking hundreds of nodes and multi-hour runtimes. But if you're working on one particular aspect of the dataflow, say adding a few new variables to a computeExpression node, you'll have to run the entire dataflow just to test and iterate on the node which you're adding.
Doing things that way can equate to multi-hour turnarounds and a very slow process. This is where the dataflow extractor shines. Upload the JSON file for your dataflow into the extractor, and select the dataset you're modifying. The extractor will output a new dataflow JSON containing just the nodes needed for your dataset. Now you can work on your computeExpression in this smaller dataflow, which will have a much shorter runtime and allow you to iterate on your changes much more rapidly. When you're done, simply copy the changes back into your original dataflow.
Of course, you can do all of this manually, too. But manually removing unwanted nodes from a dataflow involves clicking on the little delete icon next to each node you want to remove. Tedious!