xlsx_props_to_csv.py

#! /usr/bin/env python3

# Usage:
#  $ python3 this.py > output.csv
#  $ nkf --overwrite --oc=UTF-8-BOM output.csv

import glob
import hashlib
import os
import sys
from openpyxl import load_workbook

import pandas as pd


def get_xlsx_properties(xlsx_path: str) -> list[str]:
    try:
        print(f"loading: {xlsx_path}", file=sys.stderr)
        with open(xlsx_path, "rb") as f:
            file_hash = hashlib.file_digest(f, "sha256").hexdigest()
        file_size = os.path.getsize(xlsx_path)
        props = load_workbook(xlsx_path).properties
        row = [
            xlsx_path,
            props.creator,
            props.lastModifiedBy,
            props.created,
            props.modified,
            file_size,
            file_hash,
            "ok",
        ]
        return row
    except Exception as e:
        row = [
            xlsx_path,
            None,
            None,
            None,
            None,
            None,
            None,
            str(e),
        ]
        return row


df = pd.DataFrame(
    [get_xlsx_properties(path) for path in glob.glob("*.xlsx")],
    columns=[
        "prop_path",
        "prop_creator",
        "prop_lastModifiedBy",
        "prop_created",
        "prop_modified",
        "file_size",
        "file_hash",
        "status",
    ],
)

df.to_csv(
    path_or_buf=sys.stdout,
)