Test pipeline preserves extra fields in data.

Test __call__ returns a batch.

Test that rejected items are separated.

Test pipeline call with single item.

Test __call__ with data where all items get rejected.

Test __call__ with empty list returns empty list.

Test __call__ with empty list.

Test pipeline with filter operator.

Test pipeline with rewriter operator.

Test __call__ with single item.

Test that BaseAPI cannot be instantiated directly.

Test that Filter subclass can be instantiated (check has default impl).

Test that Operator cannot be instantiated.

Test that Rewriter subclass can be instantiated (rewrite has default impl).

Test check detects consecutive character repetition.

Test check_content with text string.

Test check_content with dict.

Test check_content with list of dicts.

Test check_content with list of strings.

Test check handles empty extension.

Test check rejects excluded extension.

Test check rejects extreme aspect ratio.

Test check rejects too small image.

Test check that keeps.

Test check passes when no image.

Test check passes when no image (item-level filter returns False).

Test check passes when no image and not required.

Test check rejects when no image but required.

Test check passes for non-repetitive text.

Test check passes when no image/video tags.

Test checking already processed dataset.

Test checking question length.

Test check that rejects.

Test check returns (rejected, reason) tuple.

Test check allows tags in question.

Test process rejects too long conversation (qa_count > max_length).

Test check rejects too long text.

Test process rejects too short conversation (qa_count < min_length).

Test check rejects too short text.

Test checking unprocessed dataset.

Test check passes for valid extension.

Test check passes for valid image size.

Test check passes for valid length.

Test check passes for valid length.

Test check passes for valid aspect ratio.

Test check rejects response with image tag.

Test check rejects repetitive text (needs 1000+ chars with 10+ repeats).

Test check rejects response with video tag.

Test checkpoint file is valid JSON.

Test checkpoint persists across manager instances.

Test that clean clears the data buffer.

Test cleaning data item.

Test clearing all checkpoints.

Test clearing nonexistent checkpoint.

Test clearing checkpoint for specific file.

Test closing logger.

Test close_all method.

Test closing logger multiple times.

Test cnt_data increments after save.

Test cnt_rejected_data increments after save with rejected items.

Test concurrent generation.

Test handling concurrent updates (basic).

Test __contains__ for existing module.

Test __contains__ for nonexistent module.

Test ConvLengthFilter is registered.

Test conversations property.

Test copy_image with bytes.

Test copy_image with None.

Test copy_image with PIL Image returns the same reference.

Test that after corrupt file, new updates work normally.

Test that a corrupt file logs a warning when logger is provided.

Test that a corrupt checkpoint file results in empty checkpoints.

Test counters accumulate across multiple saves.

Test creating basic OperatorStats.

Test creating basic DatasetStats.

Test creating basic PipelineStats.

Test creating config from file paths.

Test creating config with data sizes.

Test creating config with source mapping.

Test creating empty StandardConfig.

Test creating FilterDecision.

Test creating OperatorResult.

Test creating QA instance.

Test creating Result.

Test creating Result with existing decisions.

Test creating RewriteDecision.

Test creating DatasetConfig with all fields.

Test creating with dataset configs.

Test creating DatasetConfig with required fields only.

Test critical logging.

Test CRITICAL level exists.

Test data property returns underlying dict.

Test DATALOADER contains StandardDataLoader.

Test DATALOADER registry has StandardDataLoader.

Test DATALOADER registry exists.

Test DATASAVER contains StandardDataSaver.

Test DATASAVER registry has StandardDataSaver.

Test DATASAVER registry exists.

Test debug logging.