I have two expressions in a Column Expression node. I have a simple string manipulation expression to strip each column of whitespace, and a rule expression which removes empty strings. I’m very new to Knime but i think these work ok except the rule expression adds “(#1)” to every column name. The names go into the column expressions node as (e.g.) “Product_code_1” and come out as “Product_code_1 (#1)”.
Clearly I’m doing something wrong but i don’t know what.
Below is a screenshot of the nodes in sequence, the regex nodes change the column names all to ANONYMOUS_COLUMN and then changes it back with a flow variable. I tried to make the workflow without renaming the columns and adding the flow variable of the currentColumnName to the output flow variable in the column expression node, but this worked out even worse as each column was duplicated 3 time. I saw someone using the renaming technique in an example in order to make the Output Column on the expression dynamic (instead of using flow variable).
Have couple of ideas what could be wrong in your current design but it would be best if you share workflow to avoid unnecessary guessing if you are interested to find that out See here how to do that:
I really like the Column expression node as it lets us build some fairly complex logic without having to go to another node.
2 Feature requests:
add a function to let you use regex to parse out parts of a string as a capture group
allow you to use new output columns created in expressions above it to be used in the current expression… I realize the workaround is to use a second node after the column is created with the new column… this one is definitely less important
Thanks for the suggestion, @gab1one . I was mainly asking for that feature so that I could have access to the regex in the same node’s if else logic. I wanted to avoid doing the regex first to create a new column of data just for the logic and then filtering out the column after.
Yes that makes a lot of sense, I will add this to the list of possible improvements of that node. Just wanted to let you know that the regex node exists so you can do what you, even if it is not as convenient as it should be.
Hi there,
I think the thread went a bit off-topic? the main question – “(#1)” being appended to the column name – is not answered, as far as I can see.
I have the same problem as the thread author: the column name is changed after some simple string manipulation within the column expressions node. Is this a general behavior we have to live with?
Hi Ivan,
thanks for your reply! actually I used the node without a loop, but I realized, that the number #1 is only appended, when I use more than one expression inside the node.
The (#1) being appended to the column name happens in the event that you have 2 columns with the same name. Just like you can’t have 2 fields with the same name in a db table, the same rule applies in Knime tables.
There are several cases where this can happen. You can either rename the first column before the process that is going to create/generate another column with the same name, or rename the column after the process has created/generated another column with the same name.
In this case, what probably happened is that the user did not choose the “Replace Column” option during the String Manipulation, but instead chose to “Append new column” and added the same column name as new column. So Knime created a new column with the same name, but since the name already existed, Knime appends (#1) to the name so that there is no name clash/conflict. It will append (#2), (#3), etc is there are more columns with the same name.
See here how to do that:
(We are not that strict about it.)