Loading pandas DataFrames

By supporting both INSERTs and UPSERTs for DataFrames, when combined with pandas’ read_sql() functionality, PGReaper turns any Postgres database into a robust store for your pandas-based projects. Compared with other methods for loading DataFrames such as to_sql() or a combination of to_csv() and copy_from(), PGReaper provides:

• Automatic type inference for basic Python types and most numpy types
• Properly encoding jsonb (dict or list) and timestamp (datetime) objects
• Automatic correction of problematic column names, e.g. those that are Postgres keywords
• Support for composite primary keys

• Support for both INSERT OR REPLACE and UPSERT operations

  • Faster UPSERT performance using SELECT unnest() rather than slower batched INSERTs