py-i2phosts/py-i2phosts-fetcher

#!/usr/bin/python

import re
import os
import os.path
import sys
import errno
import time
import urllib2
import subprocess
import argparse

proxyurl = 'http://localhost:4444/'
sources = ['http://www.i2p2.i2p/hosts.txt', 'http://stats.i2p/cgi-bin/newhosts.txt',
		'http://i2host.i2p/cgi-bin/i2hostetag', 'http://tino.i2p/hosts.txt',
		'http://trevorreznik.i2p/hosts.txt', 'http://dream.i2p/hosts.txt',
		'http://biw5iauxm7cjkakqygod3tq4w6ic4zzz5mtd4c7xdvvz54fyhnwa.b32.i2p/uncensoredhosts.txt',
		'http://cipherspace.i2p/addressbook.txt', 'http://hosts.i2p/hosts.cgi?filter=all']
#sources = ['http://www.i2p2.i2p/hosts.txt', 'http://stats.i2p/cgi-bin/newhosts.txt']
#sources = ['http://hiddenchan.i2p/ggfg.txt']
#sources = ['http://stats.i2p/cgi-bin/newhosts.xml']
# parse command line options
parser = argparse.ArgumentParser(
		description='Hosts fetcher for py-i2phosts.',
		epilog='Report bugs to http://zzz.i2p/topics/733')
parser.add_argument('-d', '--debug', action='store_true',
		help='write debug messages to stdout instead of log file'),
parser.add_argument('-c', '--config', default='/etc/py-i2phosts/fetcher.conf', dest='config_file',
		help='config file to use')
args = parser.parse_args()

# we want open urls through proxy
proxy_handler = urllib2.ProxyHandler({'http': proxyurl})
opener = urllib2.build_opener(proxy_handler)

for source in sources:
	# cut hostname.i2p from url
	source_hostname = re.sub(r'.*//(.+?)/.+', r'\1', source)
	# use separate file for each host
	filename = 'hosts.txt.' + source_hostname
	# build last-modified info from file mtime
	try:
		mtime = os.path.getmtime(filename)
	except OSError, e:
		if e.errno == errno.ENOENT:
			pass
		else:
			sys.stderr.write('fatal error: %s', e)
			os.exit(1)
	else:
		last_modified = time.strftime('%a, %d %b %Y %H:%M:%S GMT', time.localtime(mtime))
		# prevent redownloading of hosts-file by passing If-Modified-Since http header
		opener.addheaders = [('If-Modified-Since', last_modified)]
	try:
		resp = opener.open(source, timeout=60)
	except urllib2.URLError, e:
		if hasattr(e, 'reason'):
			print 'failed to reach server %s, reason: %s' % (source_hostname, e.reason)
		elif hasattr(e, 'code'):
			if e.code == 304:
				print '%s: not modified' % source_hostname
			else:
				print 'server %s can\'t finish the request, error code: %s' \
					% (source_hostname, e.code)
		continue
	# read data from remote and write it to local file
	content = resp.read()
	f = open(filename, 'w')
	f.write(content)
	f.close()
	# get last-modified info from header and change file's mtime
	lm = resp.headers.get('Last-Modified')
	print '%s: %s' % (source_hostname, lm)
	if lm:
		target_mtime = int(time.mktime(time.strptime(lm, '%a, %d %b %Y %H:%M:%S GMT')))
		os.utime(filename, (target_mtime, target_mtime))

	# form commnd-line for invoke injector
	path = os.path.dirname(sys.argv[0])
	print 'adding hosts from: %s' % source_hostname
	sp_args = [path + '/py-i2phosts-injector', '-s', '-a', '-f', filename, '-d',
			'Auto-added from ' + source_hostname]
	p = subprocess.Popen(sp_args, shell=False, stdin=None,
			stdout=None, stderr=subprocess.STDOUT)
	p.wait()