feat: add client side transformers to parallel loader

[ad-claw000]

Summary

Added a transformers parameter to ParallelLoader.ingest and ParallelQuery.query to allow the application of client-side transformers directly when using these APIs, resolving the feature request to support usage like:

qr.ingest(csv, transformers=[ImageProps, Resizer])

Verification

• Added the parameter to ParallelLoader.ingest and ParallelQuery.query signatures.
• Used the loop over the provided transformers to wrap the dataset generator before proceeding with the normal query execution.

Fixes #326

[Copilot]

Pull request overview

This PR adds first-class support for applying client-side dataset transformers directly via the parallel ingestion/query APIs (ParallelLoader.ingest and ParallelQuery.query), targeting the feature request in #326.

Changes:

• Added a transformers parameter to ParallelLoader.ingest(...).
• Added a transformers parameter to ParallelQuery.query(...).
• Implemented transformer application by wrapping the provided generator in each transformer, in sequence.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
aperturedb/ParallelQuery.py	Extends query() with a transformers argument and wraps the generator before query execution.
aperturedb/ParallelLoader.py	Extends ingest() with a transformers argument and wraps the generator before delegating to query().

Comments suppressed due to low confidence (2)

aperturedb/ParallelQuery.py:288

• In use_dask mode this wraps the generator with Transformer instances, but DaskManager.run() assumes generator has .df, .filename, and .blobs_relative_to_csv (and later this method does len(generator.df)). Since Transformer doesn’t expose those attributes, query(..., transformers=...) will break for dask-backed generators. Consider applying transformers inside DaskManager.run() (wrap the per-partition data before calling loader.query) or explicitly rejecting transformers when use_dask is True with a clear error.

This issue also appears on line 279 of the same file.

        use_dask = hasattr(generator, "use_dask") and generator.use_dask
        if use_dask:
            self._reset(batchsize=batchsize, numthreads=numthreads)
            self.daskmanager = DaskManager(num_workers=numthreads)

        if transformers:
            for transformer in transformers:
                generator = transformer(generator)

        if hasattr(self, "query_setup"):
            self.query_setup(generator)

        if use_dask:
            results, self.total_actions_time = self.daskmanager.run(
                self.__class__, self.client, generator, batchsize, stats=stats)

aperturedb/ParallelQuery.py:285

• query_setup() is invoked after wrapping the generator in transformers. This hides generator-specific methods like get_indices() used by ParallelLoader.query_setup() to create indexes, so index creation can silently stop working when transformers are provided. Consider calling query_setup(generator) before applying transformers, or make Transformer forward get_indices/other loader hooks to the wrapped dataset.

        if transformers:
            for transformer in transformers:
                generator = transformer(generator)

        if hasattr(self, "query_setup"):
            self.query_setup(generator)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ad-claw000]

Addressed all review comments (including the low confidence ones) in commit 163c99b:

1. Passed client=self.client when instantiating transformers.
2. Removed wrapping logic from ingest() and passed it directly to query().
3. Added test_transformers and test_transformers_rejects_dask tests.
4. Added an explicit check to reject transformers when use_dask is True, throwing a ValueError as suggested.
5. Invoked query_setup() before wrapping transformers to make sure get_indices() is not hidden.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

test/test_Parallel.py:138

• This equivalence test also relies on GeneratorWithErrors(..., error_pct=0) being fully deterministic, but the generator’s current random error condition can still emit a BadCommand at error_pct=0 (randint==0), which can cause intermittent failures and mismatched succeeded counts. Making GeneratorWithErrors deterministic for error_pct=0 (or seeding random here) will stabilize the comparison.

        elements = 10
        
        # Manual wrapping
        generator1 = GeneratorWithErrors(elements=elements, error_pct=0)
        transformer1 = DummyTransformer(generator1, client=db)
        loader1 = ParallelLoader(db)
        loader1.ingest(transformer1, batchsize=2, numthreads=2, stats=False)
        
        # transformers parameter
        generator2 = GeneratorWithErrors(elements=elements, error_pct=0)
        loader2 = ParallelLoader(db)
        loader2.ingest(generator2, batchsize=2, numthreads=2, stats=False, transformers=[DummyTransformer])

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.