concatAozora.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import codecs, sys, zipfile

CODE = "shift_jis"

firstpage = True
g = codecs.open("output.txt", "wb", CODE)

files = sys.argv[1:]
isSingle = True
if files[0] == "-t":
    title = files[1].decode(CODE)
    files = files[2:]
    isSingle = False

for filename in files:
    if filename.lower().endswith(".zip"):
        z = zipfile.ZipFile(filename)
        targets = filter(lambda x:x.endswith(".txt"), z.namelist())
        if len(targets) == 0:
            sys.err.write("No .txt file in %s\n" % filename)
            continue
        f = codecs.iterdecode(z.open(targets[0]), CODE)
    else:
        f = codecs.open(filename, "rb", CODE)

    if isSingle:
        title = f.next()
    chapter = f.next()
    author = f.next()

    body = ""
    for s in f:
        if s.startswith("-------------------------------------------------------"):
            body = ""
        elif s.startswith(u"底本："):
            break
        else:
            body += s

    if firstpage:
        g.write(title)
        g.write("\n")
        g.write(author)
        g.write("\n")
        firstpage = False

    g.write(u"\n［＃改ページ］\n")
    g.write(chapter)
    g.write("\n")
    g.write(body)

    f.close()

g.close()