Interchange / GeneralValidated with real files

Apache Parquet

.parquet.pq

Any data pipeline writing columnar analytics data — Python/pandas, Spark, big-data/ETL tools

What it is

A column-oriented binary format built for large analytics datasets. AltaiPlot reads it via a bundled Python bridge that delegates directly to pandas' read_parquet (backed by Apache Arrow's own mature parquet codec) — not a hand-written parser.

Who produces it

Data engineering and analytics pipelines (pandas, Spark, dbt, cloud data warehouses) that export intermediate or final results in Parquet instead of CSV for size/speed reasons.

What we read / what we don't read

What we read

  • All columns and rows in the file, via pandas.read_parquet

What we don't read

  • This path hasn't been independently verified against a real-world third-party Parquet file this session — it inherits pandas/Arrow's own well-established parquet support, but AltaiPlot's own test coverage so far only round-trips a file the same library wrote, not one produced by an external tool. Treat it as spec-compliant rather than field-verified until we've confirmed otherwise

How to import it into AltaiPlot

  1. Open AltaiPlot and drag your .parquet/.pq file onto the window, or use File → Import.
  2. AltaiPlot detects the Parquet magic bytes (PAR1) and hands it to the bundled Python reader automatically.
  3. All columns land as plotting-ready data — no manual schema mapping needed.

Auditability note: this page's capability claims are sourced fromresources/python/app_sidecar.py:1579 — see the full format list for every extension AltaiPlot recognizes.