get volumes and chapters included in a manga archive

get_vapters.py

# really low effort, assumes internal page names follow something close to the Madokami naming scheme
# originally started to be able to auto update tracking sites before I realized I don't care enough
# > python get_vapters.py path/to/archive.cbz [path/to/additional/archives.cbz]
# path/to/archive.cbz: v01, c001-005

import os
import re
import sys
import zipfile


def get_zip_file(file_path):
    with zipfile.ZipFile(file_path) as z:
        return z


def is_zip(file_path):
    zip_exts = ("zip", "cbz")
    ext = file_path.rsplit(".", maxsplit=1)[1]
    return ext in zip_exts


def is_image(file_name):
    img_exts = ("png", "jpg", "webp", "avif")
    ext = file_name.rsplit(".", maxsplit=1)[1]
    return ext in img_exts


def page_names_from_zip(zip_file):
    files = zip_file.filelist.copy()
    page_names = [x.filename for x in files if not x.is_dir() and is_image(x.filename)]
    return page_names


def parse_page_names(page_names):
    volumes, chapters = [], []
    pattern = r"([\w\d ]+) - c(\d{3,4}) \(v(\d{2,3})\).+"
    for name in page_names:
        match = re.search(pattern, name)
        if match:
            title, chapter, volume = match.group(1, 2, 3)
            if volume not in volumes:
                volumes.append(int(volume))
            if chapter not in chapters:
                chapters.append(int(chapter))
    return (volumes, chapters)


def get_ranges(numbers):
    numbers = sorted(set(numbers))
    gaps = [[s, e] for s, e in zip(numbers, numbers[1:]) if s + 1 < e]
    edges = iter(numbers[:1] + sum(gaps, []) + numbers[-1:])
    return list(zip(edges, edges))


def format_ranges(ranges, range_type):
    formatted = ""
    pad = 2 if range_type == "volume" else 3
    for r in ranges:
        start = str(r[0]).zfill(pad)
        end = str(r[1]).zfill(pad)
        if start == end:
            formatted += f"{start}, "
        else:
            formatted += f"{start}-{end}, "
    formatted = re.sub(", $", "", formatted)
    return formatted


if __name__ == "__main__":
    if len(sys.argv) > 1:
        for f in sys.argv[1:]:
            if os.path.isfile(f) and is_zip(f):
                zip_file = get_zip_file(f)
                page_names = page_names_from_zip(zip_file)
                volumes, chapters = parse_page_names(page_names)
                volume_range = get_ranges(volumes)
                chapter_range = get_ranges(chapters)
                volume_str = format_ranges(volume_range, "volume")
                chapter_str = format_ranges(chapter_range, "chapter")
                print(f"{zip_file.filename}: v{volume_str}, c{chapter_str}")