thanks for sharing this issue with us! I just had a quick look into your workflow and you are right there seems something wrong with Cell Splitter Node. I would expect from:
String1,,
String1,,String2
The first line should be a collection with the form of [String1,,] and line should be and already will be outputted as [String1,*,String2]. The * should represent an empty String. I will do further investigations and if it turns out as bug (what I think it is) I will open a ticket and will let you know here in the forum.
thanks for picking it up so quickly. Iβve spend some time to further understand the issue and come up with a possible interims solution.
The problem, form my perspective, comes from the fact that the Cell Splitter node automatically re-creates the table based on the max. size of a row. In my screenshot there were indeed missing values and they were created as the data, by counting the cell separators, was indeed missing.
I create a new table to test the scenario further out (? equals missing):
column1
column2
column3
column4
string1
?
?
?
string1
string2
?
?
string1
string2
string3
?
string1
string2
string3
string4
string1
?
string3
?
string1
?
?
string4
?
string2
?
string4
?
?
?
string4
?
?
string3
string4
?
?
?
string4
Converted to string via collection column it is (1st column equals row index):
If that is converted back but removing the trailing β|?β it created real missing values. So itβs not really a but but the cell splitter does not interpret β?β properly. The new workflow, same from above, has a few scenarios which try to remove all β?β but it breaks the table structure.