Quick clumsy script to convert an Illumina InterOp/ExtractionMetricsOut.bin v2 file into CSV

extraction_metrics_out.py

#!/usr/bin/env python

"""
Quick clumsy script to convert an Illumina InterOp/ExtractionMetricsOut.bin v2
file into CSV.

http://illumina.github.io/interop/extraction_v2.html
"""

import sys
import csv
from struct import unpack
from pathlib import Path

BASES = ["a", "c", "g", "t"]
FIELDS = ["lane", "tile", "cycle"] + [f"focus_{b}" for b in BASES] + [f"max_int_{b}" for b in BASES] + ["datestamp"]

with open(sys.argv[1], "rb") as f_in:
    writer = csv.writer(sys.stdout, lineterminator="\n")
    version, record_size = unpack("<BB", f_in.read(2))
    sys.stderr.write(f"Version: {version}\n")
    sys.stderr.write(f"Record Size: {record_size}\n")
    if version != 2 or record_size != 38:
        sys.stderr.write(
            "ERROR: Expected version 2 with record size 38, not "
            f"version {version} with record size {record_size}\n")
        sys.exit(1)
    writer.writerow(FIELDS)
    while True:
        data = f_in.read(record_size)
        if len(data) != record_size:
            break
        writer.writerow(unpack("<HHHffffHHHHQ", data))