Nested loops with variables shared between iterations

[giuseppemasullo]

Hello everyone, I’m quite beginner at Kinme or, at least, using these components.
This is a simplified example of my actual requirements, but if I get how those logics work I guess I will be able to apply the same to the real scenario.

Let’s pretend to have 2 different tables:

The first one has the field “Document Number” as key and a string:

The second one has several lines per each “Document Number”; each row reports the changes to be performed in the string of the first table.

I thought to create

• a first loop that processes each row of the first table, creating a variable with the string and selecting the rows of the second table relevant for this iteration;
• a nested loop that should apply the changes to the variable as per the instructions provided by the second table.

At each iteration the changes are applied to the variable, but everything disappear once the loop ended.
I read the several discussions already opened on this same topic and I am sorry if I am asking something already clarified to other users, but I struggle in understanding how I should store the final target string in the variable/table row to be used for being exported in an excel file.

Here you go the workflow created for testing purposes.
I really hope that somebody could provide me the key suggestion to get this done.

Thank you for your support.
Giuseppe

Loop_Test.knwf (113.8 KB)

[ActionAndi]

Hi,
that’s a tricky task you have.
I tried to solve it with a recursive loop. Could you check my Workflow?

250322_Loop_Test – actionandi

Recursive loops can be very slow if the number of iterations is large. But give it a try.

Andi

[giuseppemasullo]

I really think that it works!!!
Thank you very much for that.
I had to upgrade my Knime version but I finally run and checked it.
Now I have to apply the same logic to the more complex and extended scenario but I got the logic behind that.
Thank you again for the support and the quick feedback!

Giuseppe

[ActionAndi]

Glad to hear, that it works.

Another approach is to pass the work to python:

• Use a Python Script node, add another input.
• First Input: First table with XML column
• Second Input: Table with replacements

import knime.scripting.io as knio

# This example script performs an inner join on the node's input tables using pandas.
df1 = knio.input_tables[0].to_pandas()
df2 = knio.input_tables[1].to_pandas()

SearchStr = "String Before"
ReplaceStr = "String After"
ID = 'Document Number'

def replace_strings(df1, df2):
    for index, row in df2.iterrows():
        id_value = row[ID]
        search_str = row[SearchStr]
        replace_str = row[ReplaceStr]

        # Replace of SearchStr with ReplaceStr in corresponding string in df1
        df1.loc[df1[ID] == id_value, 'XML'] = df1.loc[df1[ID] == id_value, 'XML'].str.replace(search_str, replace_str)

    return df1

# Apply function
df1 = replace_strings(df1, df2)


knio.output_tables[0] = knio.Table.from_pandas(df1)

[alpayzeybek]

Hi,

Although you’ve already received an answer, I wanted to share an alternative approach. I’ve added an extra Constant Value Column node to retain the XML string in each iteration of the loop. This ensures that the flow variables you create remain tracked throughout the process.

I’ve also uploaded screenshots and the revised KNIME workflow for your reference. I hope this helps you understand my approach to managing flow variables in loops.

Let me know if you have any questions!

Best,
Alpay

Loop_TestAZ.knwf (31.5 KB)