Just some feedback to the updated Excel reader. I like it, but came across one minor issue:
- I deal with messy Excel data a lot and use KNIME to clean it up.
- In a typical case, I get a .xlsx file with 100+ sheets, there is no real data schema, it’s all over the place.
- My prefered solution is to create a loop and append all sheets into one file (as strings) and go on from there.
- Worked fine with the old Excel reader.
- The new Excel reader tries to force a schema (columns and data types) on me, based on the first sheet he finds, or I select.
- No bueno
- Let the users create a desired target schema in the transformation tab:
- Currently I don’t see a way how to add new columns
- Force numbers to strings
My current workaround is to append a “target schema sheet” in the original file before reading all sheets. It works, but not exactly proud of that.
- Also, “any unknown new column” probably works fine for new files, but doesn’t really do much in a loop.