sc2tcbook.py

#!/usr/bin/env python
import opencc
import os
import sys
import glob
import zipfile

converter = opencc.OpenCC('s2hk.json')

def expand_wildcards(pattern):
    # Use glob to find all matching files recursively
    return glob.glob(pattern, recursive=True)

def sc2tc(book_path):
    if not book_path.endswith('.epub'):
        return
    print(f"Reading file: {book_path}")
    temp_book_path = book_path.replace('.epub', '_temp.epub')

    with zipfile.ZipFile(book_path, 'r') as zip_ref:
        with zipfile.ZipFile(temp_book_path, 'w') as temp_book:
            for file_info in zip_ref.infolist():
                # Read the content of the file
                with zip_ref.open(file_info) as file:
                    if file_info.filename.endswith('.html') or file_info.filename.endswith('.ncx') or file_info.filename.endswith('.opf'):
                        content = file.read()
                        converted_content = converter.convert(content)
                        converted_content = converted_content.replace('zh-CN', 'zh-TW')
                        temp_book.writestr(file_info.filename, converted_content)
                    else:
                        temp_book.writestr(file_info, zip_ref.read(file_info.filename))

    os.remove(book_path)
    os.rename(temp_book_path, book_path)
    head, tail = os.path.split(book_path)
    tc_path = converter.convert(tail)
    if head:
        tc_path = head + '/' + tc_path
    os.rename(book_path, tc_path)
    print(f"Writing file: {tc_path}")

if len(sys.argv) < 2:
    print("Usage: sc2tcbook.py <filename(s)>")
    sys.exit(1)

# Collect all matching files
all_matching_files = set()  # Use a set to avoid duplicates

for pattern in sys.argv[1:]:
    matching_files = expand_wildcards(pattern)
    all_matching_files.update(matching_files)

if not all_matching_files:
    print("No files found matching the provided patterns.")
    sys.exit(1)

# Process each matching file
for file_path in all_matching_files:
    if os.path.isfile(file_path):
        sc2tc(file_path)