[BUG] Fix source dataset issue when running link jobs #1193
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Test: test_2_rounds_1k_duckdbPercentage change: -29.9%
Test: test_2_rounds_1k_sqlitePercentage change: -24.1%
Click here for vega lite time series charts |
ADBond
approved these changes
Apr 20, 2023
| ), | ||
| ], | ||
| ) | ||
| def test_link_only(input, source_l, source_r): |
There was a problem hiding this comment.
Nice test! Really helps understand the change
This was referenced Nov 8, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Quick Summary
It looks as if there was a bug in vertically_concatenate.py that was causing two
source_datasetcolumns to be produced in various steps, wheresource_datasetwas outlined in the settings object.This was causing issues in DuckDB, where I couldn't use a vertically concatenated dataframe with a
source_datasetcolumn for a link job -- i.e. a single df withsource_datasetwhich outlines which dataset a record belongs to.But, more troubling than this, it appears that this was just outright breaking link only jobs in spark. I'll post some code that breaks below.
Essentially though, it was causing this behaviour in
concat_with_tf, which was causing spark to throw a wobbly.This code fixes the issue by migrating
_source_dataset_column_nameto the linker class and checking whether the column already exists within the user's database.Internals and why I opted to go down this path:
To start - these changes don't adjust the underlying SQL/logic that's being used wherever I've replaced
linker._settings_obj_source_dataset_column_namewithlinker._source_dataset_column_name.The workflow is still:
concat_with_tfThe change is in the naming convention used and what we output in
predict().Now if the users provides a dataframe that already contains
source_dataset, splink will adjust step 1 to use the alias__splink_source_dataset. This will then be scrapped when it's no longer needed and the user will be left with their originalsource_datasetin the output.If the users provides a df without
source_dataset, splink won't fall over and will just call step 1source_dataset.Why this way?
source_datasetcasesource_datasetcolumn, the link job will still run.On point 3 - We may want to check the source dataset column in preprocessing and establish if it's valid.