module table_io

irx.table_io

Source: packages/irx/src/irx/table_io.py:1

Simple native CSV and Parquet table I/O.

Eager reads return independently owned PyArrow tables; writes borrow their input and check finalization. Importing this module performs no native build or optional Parquet loading. Format engines remain in the IRx native runtime.

Classes

• ParquetFilter — Immutable AND predicate group with independent composition.

Functions

• read_csv — Read all CSV rows with strict parsing before optional projection.

• read_parquet — Read exact rows, discovering files or Hive datasets by default.

• where — Create a checked predicate without serialized values or type tags.

• write_csv — Write all chunks and check finalization without consuming the table.

• write_parquet — Write exact table types using explicit overrides and checked close.