fix_shueisha.py

"""
usage:
    $ python3 fix_shueisha.py [EPUB_FILE]...
    $ ls | grep *.epub | python3 fix_shueisha.py
"""
import sys
from zipfile import ZipFile
from io import BytesIO, TextIOWrapper


def fix_container(input_buffer):
    output_buffer = TextIOWrapper(BytesIO(), encoding="utf-8")
    for line in TextIOWrapper(input_buffer, encoding="utf-8"):
        if "advanced.opf" in line and not line.startswith("<!--"):
            line = f"<!--{line[:-1]}-->{line[-1]}"
        output_buffer.write(line)
    output_buffer.seek(0)
    return output_buffer


def fix_opf(input_buffer):
    output_buffer = TextIOWrapper(BytesIO(), encoding="utf-8")
    for line in TextIOWrapper(input_buffer, encoding="utf-8"):
        if "cover.jpg" in line and 'properties="cover-image"' not in line:
            line = line.replace("href", 'properties="cover-image" href')
        output_buffer.write(line)
    output_buffer.seek(0)
    return output_buffer


def fix(epub_name):
    filters = {
        "META-INF/container.xml": fix_container,
        "OPS/standard.opf": fix_opf,
    }
    archive = {}
    with ZipFile(epub_name) as zipfile:
        for info in zipfile.infolist():
            if info.filename in filters:
                with zipfile.open(info) as f:
                    archive[info.filename] = filters[info.filename](f)
            else:
                with zipfile.open(info) as f:
                    archive[info.filename] = TextIOWrapper(BytesIO(f.read()), encoding="utf-8")
    with ZipFile(epub_name, "w") as zipfile:
        for filename, io in archive.items():
            with zipfile.open(filename, "w") as f:
                f.write(io.buffer.read())


def main():
    epub_list = sys.argv[1:] or sys.stdin
    for epub in epub_list:
        print(epub.strip(), end="...")
        fix(epub.strip())
        print("done")


if __name__ == "__main__":
    main()