Hello friends. 
I would like to share an idea for a flow I created, but I feel like I took too many detours and it could have been better.
The idea is simple:
Read several .csv files from a folder.
But there is a catch:
- Some CSV files need to be deleted from the folder before they are read. These are the files with a size of zero.
- The CSV files do not have headers; therefore, when a file has “zero” size, the CSV Reader node triggers an error because it cannot build the Data Type structure.
- After some files have been read, the flow must also work in cases where no files are found to be deleted.
What is the ultimate goal?
To simplify my flow.
But why?
Because I think I took too many detours. When the folder contains files with greater than zero size, the flow works; however, when it does contain files with size of “zero”, the flow fails due to the Row Filter nodes I used. In other words, no zero-size files can reach the CSV node because those files has no header and are empty.
One idea led to another, and, with each attempt, I kept changing the flow. It took me about 3.5 hours to reach the final model to not breaks.
But as I said, there were so many detours that I became intrigued and would like a more professional perspective.
I am attaching the final flow I completed, but I would like to know if there is a simpler approach.
csv.Help_01.knwf (958.4 KB)
