collate_tif_djvu.py

#!/usr/bin/python
# -*- coding: utf-8  -*-
"""One trick script to
collate sets of tiff files across multiple
directories into smaller DJVUs.
Used for upload of manuscripts
from Musikverket 2021-05.
"""
import os
from shutil import which  # used for djvu conversion
from subprocess import run  # used for djvu conversion
from tqdm import tqdm


def list_subdirs(path):
    return [f.path for f in os.scandir(path) if f.is_dir()]


def can_djvu():
    """
    Check if DjVu files can be created.
    Check whether DjVuLibre is installed,
    on PATH and marked as executable.
    """
    return which('djvm') is not None and which('c44') is not None


def create_work_djvu(path):
    parent_path = os.path.basename(os.path.dirname(path))
    tmp_djvu = os.path.join(path, "tmp.djvu")
    book_djvu = parent_path + "---" + os.path.basename(os.path.normpath(path)) + ".djvu"
    files_to_process = sorted([x for x in os.listdir(path) if x.endswith(".tif")])
    print("====== Starting {} ======".format(book_djvu))
    for i, page in tqdm(enumerate(files_to_process, 1), total=len(files_to_process)):
        tmp_jpg = os.path.join(path, "{}.jpg".format(page))
        run(['convert', os.path.join(path, page), tmp_jpg], check=True)
        run(['c44', '-crcbfull', tmp_jpg, tmp_djvu], check=True)
        if i == 1:
            run(['djvm', '-c', book_djvu, tmp_djvu], check=True)
        else:
            run(['djvm', '-i', book_djvu, tmp_djvu], check=True)
        os.remove(tmp_jpg)
    os.remove(tmp_djvu)
    print("====== Completed {} ======".format(book_djvu))


def main():
    exclude = ["system volume information", "$", "mediaexplorer",
               "recycle", "trash", "!"] # irrelevant directories
    for directory in sorted(list_subdirs('.')):
        if any(substring in directory.lower() for substring in exclude):
            continue
        for subdirectory in list_subdirs(directory):
            create_work_djvu(subdirectory)


if __name__ == "__main__":
    if not can_djvu():
        raise Exception('Djvu utils djvm and c44 not found.')
    main()