CODEWITHSUNDEEP
apache-nifi
ponponke
ponponke

Reputation: 113

Pass a directory as an argument to ExecuteStreamCommand

I have a Java program that is designed to process a directory full of data, passed as an argument to the JAR.

input_dir/
  file1
  file2

How can I tell NiFi to pass a directory to a ExecuteStreamCommand, as an argument, instead of an individual FlowFile ?

Is there a way to model a directory as a FlowFile ?

I tried to use GetFile just before ExecuteStreamCommand on input_dir parent directory in order to get ìnput_dir`, so it would be passed to the stream command.

It didn't work, as GetFile just crawls all the directories looking for actual files when "Recurse Subdirectories" attribute is set to true.

When set to false, GetFile doesn't get any files.

To summarize, I would like to find a way to pass a directory containing data to a ExecuteStreamCommand, not just a single FlowFile.

Hope it makes sens, thank you for your suggestions.

Upvotes: 0

Views: 1021

Answers (1)

Bryan Bende
Bryan Bende

Reputation: 18660

A flow file does not have to be a file from disk, it can be anything. If I am understanding you correctly, you just need a flow file to trigger your ExecuteStreamCommand. You should be able to do this with GenerateFlowFile (set the scheduling strategy appropriately). You can put the directory directly into ExecuteStreamCommand, or if you want it to be more dynamic you can add it as a flow file attribute in GenerateFlowFile, then reference it in ExecuteStreamCommand like ${my.dir} (assuming you called it my.dir in GenerateFlowFile).

Upvotes: 2

Related Questions