4chan-thread-archiver

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

# The Hydrus Network developer screwed around with this April 2013
# Part of the JSON-based-chanarchiver by Lawrence Wu, 2012/04/04
# Originally from https://github.com/socketubs/4chandownloader
# Rewritten to save in seperate images folder, download plain HTML, modularization, comments, and code cleanup

#
# Initial release Nov. 5, 2009
# v6 release Jan. 20, 2009
# http://cal.freeshell.org
#
# Refactor, update and Python package
# by Socketubs (http://socketubs.net/)
# 09-08-12
#

from include import threadarchiver

from docopt import docopt

doc = """4chan-thread-archiver, uses 4chan API to download thread images and/or 
thumbnails, along with thread HTML, JSON, and a list of referenced external
links. Uses the py4chan library.

Usage:
  4chan-thread-archiver <url> [--path=<string>] [--delay=<int>] [--nothumbs] [--thumbsonly]
  4chan-thread-archiver -h | --help
  4chan-thread-archiver -v | --version

Options:
  --nothumbs          Don't download thumbnails
  --thumbsonly        Download thumbnails, no images
  --delay=<int>       Delay between thread checks [default: 20]
  -h --help           Show help
  -v --version        Show version
"""

def print_as_function( text ): print( text )

def main(args):
    # Copy data from arguments
    thread = args.get('<url>').split('/')[5]
    board  = args.get('<url>').split('/')[3]
    path   = args.get('--path')
    nothumbs = args.get('--nothumbs', False)
    thumbsonly = args.get('--thumbsonly', False)
    delay  = args.get('--delay')

    archiver = threadarchiver.Archiver( print_as_function )
    
    Archiver.Archive( thread, board, path, notthumbs, thumbsonly, delay )
    
if __name__ == '__main__':
  args = docopt(doc, version=0.3)
  main(args)