Hi @Kamusett , I re-read what you wrote. I would not say that it’s a bug, it’s how it’s supposed to work. At least that’s how the other nodes work.
I thought the File Reader always worked like that too, but from the deprecated version, it actually recognizes string “?” as empty, which it should not. Empty is empty, and string “?” is data.
The deprecated version had an advanced setting where you could actually define any string pattern as empty:
So you could define string “?” as empty (which it seems to do by default).
This setting does not seem to exist in the new version.
As an alternative for the new version, you can replace the string “?” to missing with a String Manipulation (Multi Column) with this expression:
toNull(replace($$CURRENTCOLUMN$$, "?", ""))
Here are the 3 things you need to configure (highlighted in blue):
- Make sure you remove all the non-string columns
- Add the above expression
- Choose “Replace selected input columns”
This will change these string “?” (black ?):
to missing values (red ?):
