Test DEBUG level exists.

Test deeply nested config.

Items without _original_schema default to standard.

When both fields exist, messages takes priority.

Test detect_consecutive_repetition with character repetition.

Test detect_consecutive_repetition with formatting patterns (|, __, --) returns (False, None) even if repeated.

Test detect_consecutive_repetition finds repetition.

Test detect_consecutive_repetition with text >= 1000 chars and 10+ consecutive repeats returns (True, pattern).

Test detect_consecutive_repetition with no repetition.

Test detect_consecutive_repetition skips patterns of length < 2 (the loop starts at length=2).

Test detect_consecutive_repetition with text < 1000 chars returns (False, None).

Test detect_consecutive_repetition with custom threshold.

Test that a pure ASCII letter string is detected as English.

Test detect_language correctly identifies Chinese text.

Test that a single Chinese character is detected as Chinese.

Test detect_language with Chinese text.

Test detect_language with empty string.

Test detect_language with empty string.

Test detect_language with English text.

Test detect_language with English text.

Test detect_language with mixed English and Chinese text.

Items without conversation fields default to standard.

Test dfs with nested dict.

Test dfs with simple dict.

Test dfs with list.

Test that dict config without any file path raises ValueError.

Test that dict config with file_path resolves correctly.

Test that dict config with json_path alias resolves correctly.

Test dumping nested config.

Test dump preserves data types.

Test dumping simple config.

Test DataItem with empty conversations.

Test that an empty checkpoint file is handled gracefully.

Test cleanup with empty list does not raise.

Test rewrite decision with empty message.

Test loading empty Python file.

Test filter decision with empty reason.

Test iterating empty registry.

Test encoding already base64 encoded string.

Test encoding from bytes.

Test encoding PIL Image.

Test encode_image_to_base64.

Test encode_image_to_base64 with resize.

Test end_operator.

Test end_operator without start_operator.

Test end_pipeline.

Test end_pipeline without start_pipeline.

Test equality comparison.

Test error logging.

Test ERROR level exists.

Test _execute_operator pads results when operator returns fewer results than items.

Test that sub-pipelines execute in priority order.

Test execution_time property.

Test exists method.

Test expand_items with empty list.

Test expand_items splits multi-turn into single turns.

Test expand_items preserves conversation data.

Test expand_items with single turn.

Test export_csv.

Test export_csv with custom path.

Test exporting empty collector.

Test export_json.

Test export_json with custom path.

Test that JsonFormat supports .json extension.

Test that JsonlFormat supports .jsonl extension.

Test FilterDecision equality.

Test FilterDecision with global index (-1).

Test that FilterDecision is immutable.

Test FilterDecision for kept item.

Test filters handle empty conversations.

Test filter handles invalid JSON response.

Test filter handles model error.

Test that filter removes items.

Test filters handle special characters.

Test filters handle unicode content.

Test filter with empty conversations.

Test format_all_qa for multi-turn.

Test format_content creates formatted string.

Test formatting ERROR level record.

Test formatting INFO level record.

Test format_qa with question only.

Test format_qa with question and answer.

Test format_qa with original values.

Test formatting with color enabled.