I have a script that relies heavily on pandas file i/o functions like read_csv(), to_csv(), etc. with the many custom parameters they offer.
Can these be used in a luigi pipeline for local file targets?
I want to convert it to a luigi pipeline but every example I have seen uses open/close python i/o with custom parsing.
I hear that the luigi target object functionality relies specifically on target.close() or 'with' for the dependency tracking.
Yes. LocalTarget
can be used with the output of arbitrary code. The file tracking relies on the exists()
method. open()
is just a convenience method.