#!/usr/bin/python import re import os import os.path import sys import errno import time import urllib2 import subprocess import argparse import configobj # 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() # read config config = configobj.ConfigObj(args.config_file, file_error=True) if 'include' in config: config_included = configobj.ConfigObj(config['include']) config.merge(config_included) if 'DJANGO_PROJECT_PATH' in config: sys.path.insert(1, config['DJANGO_PROJECT_PATH']) else: sys.stderr.write('"DJANGO_PROJECT_PATH" is missing in config\n') sys.exit(1) from web.lib.utils import get_logger # configure logger if args.debug == True: log_level = 'debug' log_file = None else: log_level = config['log_level'] log_file = config['log_file'] log = get_logger(filename=log_file, log_level=log_level) # we want open urls through proxy if 'proxyurl' in config: proxy_handler = urllib2.ProxyHandler({'http': config['proxyurl']}) else: log.critical('"proxyurl" is missing in config') sys.exit(1) opener = urllib2.build_opener(proxy_handler) if not 'sources' in config: log.critical('"sources" is missing in config') sys.exit(1) # FIXME: use as_list() here for source in config['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: log.critical('fatal error: %s', e) sys.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: log.debug('fetching hosts from: %s', source_hostname) resp = opener.open(source, timeout=60) except urllib2.URLError, e: if hasattr(e, 'reason'): log.warning('failed to reach server %s, reason: %s', source_hostname, e.reason) elif hasattr(e, 'code'): if e.code == 304: log.info('%s: not modified', source_hostname) else: log.warning('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 try: content = resp.read() except: log.warning('failed to read data from %s', source_hostname) continue 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') log.debug('%s Last-Modified: %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]) log.info('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()