青空文庫の「公開中　作家別作品一覧拡充版」から URL を抽出し、テキストファイルを一括取得して、AozoraEpub3 https://github.com/hmdev/AozoraEpub3/ を利用して epub に変換するスクリプト

aozora_epub.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# by Akihiro Uchida, CC0 dedicated to the public domain
# see http://creativecommons.org/publicdomain/zero/1.0/
import os.path
import time
import re
import subprocess
import atexit
import cPickle as pickle
from datetime import datetime
from zipfile import ZipFile
from StringIO import StringIO
import requests
import unicodecsv

basedir = os.path.dirname(__file__)
pkl_fpath = os.path.join(basedir, 'exit.pkl')
list_all_url = 'http://www.aozora.gr.jp/index_pages/list_person_all_extended_utf8.zip'
datefmt = '%Y-%m-%d'


def normalize_name(family, given):
    name = family + given
    katakana_pattern = re.compile(ur'[ァ-ヶー]+')
    if katakana_pattern.match(name):
        name = u'・'.join([s for s in [family, given] if s])
    return name


class AozoraList(object):

    def __init__(self):
        self.last_exited = datetime.fromtimestamp(0)
        if os.path.exists(pkl_fpath):
            with open(pkl_fpath) as f:
                self.last_exited = pickle.load(f)
        req = requests.get(list_all_url)
        zipdata = StringIO(req.content)
        with ZipFile(zipdata) as zf:
            basename = os.path.splitext(os.path.basename(list_all_url))[0]
            encoding = 'utf-8' if basename.endswith('utf8') else cp932
            with zf.open(basename + '.csv') as csv:
                self.csv_reader = unicodecsv.DictReader(csv, encoding=encoding)
        return

    def __iter__(self):
        return self

    def next(self):
        while True:
            row = self.csv_reader.next()
            item = {'author': normalize_name(row[u'姓'], row[u'名']),
                    'title': row[u'作品名'],
                    'url': row[u'テキストファイルURL'],
                    'modified': row[u'テキストファイル最終更新日'],
                    'sylab': row[u'姓読みソート用'][0],
                    'exited': self.last_exited}
            if all(item.values()) and os.path.splitext(item['url'])[-1] == '.zip':
                item['modified'] = datetime.strptime(item['modified'], datefmt)
                return item
        return


class ZipData(object):

    def __init__(self, item):
        self.fpath = os.path.join(basedir, 'zip', item['sylab'],
                                  item['author'], item['title'] + '.zip')
        if os.path.exists(self.fpath) and item['exited'] > item['modified']:
            return
        r = requests.get(item['url'])
        time.sleep(30)
        if r.status_code != 200:
            return
        parent_dir = os.path.dirname(self.fpath)
        if not os.path.isdir(parent_dir):
            os.makedirs(parent_dir)
        with open(self.fpath, 'w') as f:
            f.write(r.content)
        return

    def to_epub(self):
        fpath = os.path.join(basedir, 'epub', item['sylab'],
                             item['author'], item['title'] + '.epub')
        if os.path.exists(fpath) and item['exited'] > item['modified']:
            return
        parent_dir = os.path.dirname(fpath)
        if not os.path.isdir(parent_dir):
            os.makedirs(parent_dir)
        subprocess.check_call(['java', '-Dfile.encoding=UTF-8',
                               '-cp', 'AozoraEpub3/AozoraEpub3.jar', 'AozoraEpub3',
                               '-i', 'AozoraEpub3/presets/reader.ini',
                               '-d', parent_dir, '-of', self.fpath])
        return


@atexit.register
def save_exit_date():
    with open(pkl_fpath, 'w') as f:
        pickle.dump(datetime.now(), f)

if __name__ == '__main__':
    for item in AozoraList():
        zipped = ZipData(item)
        zipped.to_epub()