Loop
On this page
Purpose
Loop processes DataFrame rows one at a time through a repeated path, then returns the collected results after the final item.
When to use it
- Process CSV rows one by one.
- Transform each DataFrame row and collect the transformed rows.
- Run a bounded repeated path before a final aggregation step.
How to use it
- Connect a DataFrame to Loop.
- Connect Item to the components that process one row.
- Return the processed Data item to Loop.
- Connect Done to the component that needs the accumulated DataFrame.
Configuration
| Setting | What it controls | Recommended starting point |
|---|---|---|
| Inputs | Supplies the DataFrame to process. | Start with a small DataFrame. |
| Item | Emits one current item to the repeated path. | Connect it to a focused per-row transformation. |
| Done | Emits the accumulated DataFrame after processing completes. | Connect it after the feedback path is working. |
Expected result
Loop runs sequentially. Item supplies one Data item at a time, and Done supplies the accumulated DataFrame after the last item returns. None and string feedback are dropped, so return processed Data from the repeated path. There is no configurable iteration cap, retry, or per-item recovery. An incomplete loop is rejected by the Builder. Paid steps in the loop run once per item.
Reference details
Keep the per-item path short and make sure it returns to Loop. Loop cannot advance until the current item path completes. Filter large DataFrames before the loop and handle an empty Done DataFrame downstream.
Works well with
Use Split Text or DataFrame Operations to supply rows. Put Parser in the repeated path when each row needs templated text. Use Type Convert set to Data when the final step does not already return the Data feedback that Loop expects.