Reading research Parquet files¶
market_data_platform.standardize.parquet provides shared file-reading helpers for research data. They inspect Parquet/CSV column names, detect fields and values in Hive-partitioned directories, project requested columns, and fall back to reading Parquet files individually while restoring partition columns when dataset-level reads fail.
These helpers handle file formats and dataset layout only. They do not implement strategy, features, stock selection, or backtests. Higher-level code can reuse them through read_parquet_dataset_compat and select_available_columns.