lastra-convert 1.0.0

Python CLI to convert between Lastra and Parquet / CSV / Arrow.
Mirror of the Java lastra-convert tool.

Install

pip install lastra-convert

CLI

Six commands, one per direction. Each takes input output plus a small flag set; run any with -h for the full list.

# Parquet ↔ Lastra
parquet2lastra ticks.parquet ticks.lastra
parquet2lastra ticks.parquet ticks.lastra --columns "ts:long:delta_varint,close:double:pongo"
lastra2parquet ticks.lastra  ticks.parquet --compression zstd

# CSV ↔ Lastra (delimiter auto-detected: comma / tab / semicolon / pipe)
csv2lastra     ticks.csv     ticks.lastra
lastra2csv     ticks.lastra  ticks.csv

# Arrow IPC ↔ Lastra
arrow2lastra   ticks.arrow   ticks.lastra
lastra2arrow   ticks.lastra  ticks.arrow

Type / codec auto-detection (override with --columns):

CSV column types are inferred per-column from up to the first 256 cells: integer if every non-empty cell parses as int; double if every non-empty cell parses as float; binary otherwise.

Library API

from lastra_convert import (
    parquet_to_lastra, lastra_to_parquet,
    csv_to_lastra,     lastra_to_csv,
    arrow_to_lastra,   lastra_to_arrow,
)

# Auto-detect every column's type + default codec.
parquet_to_lastra("ticks.parquet", "ticks.lastra")

# Per-column override.
parquet_to_lastra(
    "ticks.parquet", "ticks.lastra",
    columns="ts:long:delta_varint,close:double:pongo",
)

# Decode every series column back into a Parquet (ZSTD by default).
lastra_to_parquet("ticks.lastra", "ticks.parquet", compression="zstd")

Each function returns the row count written.

Reference implementation

This package mirrors QTSurfer/lastra-convert (Java, JVM CLI). Output is byte-equivalent for the same input + codec choices.

License

Copyright 2026 Wualabs LTD. Apache License 2.0 — see LICENSE.