After you have created the flow and the datasets within the flow and before applying recipe steps to change the data, create a duplicate of the flow. This becomes a snapshot of your original dataset. Since the imported datasets are not affected, the storage overhead for creating backups is relatively low. See Flow View Page.
Track Source Filepath and Filename
When you first load your dataset in the Transformer page, you can add the following to capture the full path to the original file that is the source of the data:
With a few extra steps, you can extract the filename from the above output. For more information, see Source Metadata References.
Track Source Row Information
You can mark the original row numbers of your source data. In the first step in your recipe after initial parsing, add the following:
This step generates a new column that contains the source row number from the source dataset.
NOTE: Source row information can become invalid if you perform multi-dataset operations such as lookups, unions, and joins. For more precise tracking of source information, you should consider creating multi-column keys, including the source row number information. For more information, see Generate Primary Keys.
See Source Metadata References.
Track Steps Affecting a Column
To see all of the steps in your current recipe that reference a specific column, select Show related steps... from the column menu.
All steps are highlighted in the Recipe panel.
NOTE: If another column is dependent on the selected column, all steps pertaining to that column are highlighted as well.
For more information, see Column Menus.
Track Column Value Changes
|D s product|
NOTE: Use this workflow only if it is important to monitor which values have changed in a column. For most use cases, the Transformer page provides sufficient visibility over your sample data to manage column values.
In the following sequence, the original column is called
String. For numeric columns, you can perform more detailed analysis between original and modified column values.
After you have completed your general setup steps of your transform, create a copy of the original column:
D trans p03Value String_orig Type step p01Name Formula type p01Value Single row formula p02Name Formula p02Value String p03Name New column name SearchTerm New formula
- You now have a copy of the original column before any manipulations were applied to it.
Add any transforms to your recipe, including any that change the values of
String. In the example below, the following transform has been applied:
D trans Type step p01Name Columns p01Value String p02Name Formula p02Value TRIM(String) SearchTerm Edit with formula
At the point in your recipe where you would like to test the column for changes, insert the following:
D trans p03Value String_changes Type step p01Name Formula type p01Value Single row formula p02Name Formula p02Value String <> String_orig p03Name New column name SearchTerm New formula
String_changescolumn now contains
truevalues where the values in
Stringhave been changed from their original values (
To see just the values that are different, sort in descending order.
Tip: You can reposition this test anywhere in your recipe after you have created the
- Before you run your recipe, you may want to remove the tracking columns that you generated (
String_changesin our example).
Example tracking column changes
Track Row Changes
- Create a copy of the flow. In its name, identify that it is your original. See Flow View Page.
- In the other flow, create your recipes as normal.
- When done, you can add the following steps:
- Union the two datasets together.
- Sort them by a key column.
NOTE: This method may not work if your recipe includes joins or added or removed columns.
If the rows are exact duplicates, they are removed. The remaining rows contain data that has been changed.