#! /usr/bin/python

import httplib
import re
import os.path
import sys
import time

# Set this to where you want them saved
localFolder = '/home/joe/tmp/'

def getPage (url):

	""" Return the contents of a url """

	try: 
		parts = url.split('//', 2)[1].split('/', 1)
		conn = httplib.HTTPConnection(parts[0])
		print "Connected to "+parts[0]
		conn.request("GET", '/'+parts[1])
		print "Sent request for "+parts[1]
		r = conn.getresponse()
		print "Got response"
		page = r.read()
		print "Read"
		conn.close()
		print "Closed"
		return page

	except:
		return False

def savePage(url, fn):

	""" save a url to a file """

	# Skip existing
	if (os.path.exists(fn)):
		return True 

	contents = getPage(url)

	if (contents != False): 

		try:
			f = open(fn, 'w')
			f.write(contents)
			f.close()

			print fn+" saved"

			return True
		except:
			print "Error saving file to "+fn
			return False

	else:
		print "Error fetching"
		return False

def name (url):

	""" Simple url compliant basename """ 

	parts = url.split('/')
	return parts[len(parts) - 1]

def main (): 

	""" Do the shit """
	
	# Pick thread from cli arg
	if (len(sys.argv) == 2):
		url = sys.argv[1]
	else:
		print "Thread not specified. Pass url to 4chan thread as the argument to this program"
		return
	
	# Our matching pattern
	pat = re.compile('<span class="commentpostername">([^<]+)<\/span> (?:<span class="postertrip">([^<]+)<\/span> )?(\d+\/\d+\/\d+\([^)]+\)\d+:\d+(?::\d+)?).+quote\(\'(\d+)\'\).+(http:\/\/images\.4chan\.org[^\.]+\.jpg).+<blockquote>(.*)<\/blockquote>')

	# Go until 404 or some other issue
	while (True):

		# get contents of page
		contents = getPage(url)

		# That failed?
		if (contents == False):
			print "Failed fetching thread"
			break

		# make sure thread isn't dead, in which case it 404's
		if (re.search('<title>4chan \- 404<\/title>', contents) != None):
			print "Thread dead"
			break

		# get posts
		posts = re.findall(pat, contents)

		i = 0

		# Get all the images
		for post in posts:
			r = savePage (post[4], localFolder+name(post[4]))
			if (r):
				i = i + 1

		# Stats for this pass
		if (i == 0):
			print "Nothing found with this pass"
		else:
			print "%d found with this pass" % i

		# Wait some time between checks
		time.sleep(5)
		print "Trying again.."

if __name__ == '__main__':
	main()