Hi ​@tb09114 ,

A simple solution I can think of is, add a File Copy wirter to the workspace so that the file created by the Python script will be moved to a location, then publish the workspace to FME Flow as Data Download Service including the writer as a download writer.

I suspect that the simplest and most robust solution would be to write the PDF to a temporary folder in your PythonCaller (e.g. using the tempfile Python module or the TempPathnameCreator transformer), then read it back into the workspace with a FeatureReader using the DATAFILE binary format. Then send it to a regular writer, also of type DATAFILE, which you publish for the Download service. Be sure to set the mime-type accordingly, e.g. application/pdf.

​@takashi What my PythonCaller returns is the path of the PDF report. The File Copy writer sounds like an easy solution but I have a hard time to set it up so it actually is using the path string to copy/move the file to another destination.

If the output feature from the PythonCaller has an attribute storing the file path of the PDF report, just rename it to “filecopy_source_dataset” then send the feature to the File Copy writer.

​@takashi your tip about the filecopy_source_dataset has ended months of frustration!!!!  Thank you!

I’ve been facing the same problem with an Excel I generate with Python due to a lot of advanced formatting, I’ve tried so many things but no joy until your tip!!!