py-i2phosts/bin/py-i2phosts-fetcher

#!/usr/bin/python

import re
import os
import os.path
import sys
import errno
import datetime
import urllib2
import subprocess
import argparse
import configobj
import tempfile
import socket

# 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
spec = '''
    proxyurl = string(default='http://localhost:4444/')
    log_file = string(default='/var/log/py-i2phosts/fetcher.log')
    log_level = option('debug', 'info', 'warning', 'error', 'critical', default='info')
    '''
spec = spec.split('\n')
config = configobj.ConfigObj(args.config_file, configspec=spec, file_error=True)
if 'include' in config:
    config_included = configobj.ConfigObj(config['include'])
    config.merge(config_included)

# django setup
DJANGO_SETTINGS_MODULE = 'pyi2phosts.settings'
if 'DJANGO_PROJECT_PATH' in config:
    DJANGO_PROJECT_PATH = config['DJANGO_PROJECT_PATH']
else:
    DJANGO_PROJECT_PATH = os.path.dirname(sys.argv[0]) + '/..'
sys.path.insert(1, DJANGO_PROJECT_PATH)
os.environ['DJANGO_SETTINGS_MODULE'] = DJANGO_SETTINGS_MODULE
from pyi2phosts.lib.utils import get_logger
from pyi2phosts.lib.utils import validate_config
from pyi2phosts.extsources.models import ExternalSource

# validate config
validate_config(config)

# 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
proxy_handler = urllib2.ProxyHandler({'http': config['proxyurl']})
opener = urllib2.build_opener(proxy_handler)

all_sources = ExternalSource.objects.filter(active=True)

for source in all_sources:
    log.debug('%s: starting work', source.name)
    if source.last_modified:
        last_modified = source.last_modified.strftime('%a, %d %b %Y %H:%M:%S GMT')
        # prevent redownloading of hosts-file by passing If-Modified-Since http header
        opener.addheaders = [('If-Modified-Since', last_modified)]
        log.debug('%s: appending If-Modified-Since: %s', source.name, last_modified)
    if source.etag:
        opener.addheaders = [('If-None-Match', source.etag)]
        log.debug('%s: appending If-None-Match: %s', source.name, source.etag)
    try:
        log.debug('%s: sending GET...', source.name)
        resp = opener.open(source.url, timeout=60)
    except socket.timeout:
        log.warning('%s: socket timeout', source.name)
        continue
    except urllib2.HTTPError, e:
        if e.code == 304:
            log.info('%s: not modified', source.name)
            source.last_success = datetime.datetime.utcnow()
            source.save()
        else:
            log.warning('%s: can\'t finish the request, error code: %s, reason: %s', source.name, e.code, e.reason)
        continue
    except urllib2.URLError, e:
        log.warning('%s: failed to reach server, reason: %s', source.name, e.reason)
        continue
    # read data from remote and write it to local file
    try:
        log.debug('%s: reading response data', source.name)
        content = resp.read()
    except:
        log.warning('%s: failed to read data', source.name)
        continue
    # save fetched content into temporary file
    fd, tmpfile = tempfile.mkstemp(text=True)
    f = os.fdopen(fd, 'w')
    f.write(content)
    f.close()
    # get last-modified info from header
    lm = resp.headers.get('Last-Modified')
    if lm:
        log.debug('%s: Last-Modified: %s', source.name, lm)
        source.last_modified = datetime.datetime.strptime(lm, '%a, %d %b %Y %H:%M:%S GMT')
    # get ETag
    etag = resp.headers.get('ETag')
    if etag:
        log.debug('%s: ETag: %s', source.name, etag)
        source.etag = etag
    # form command-line for invoke injector
    log.info('%s: adding hosts...', source.name)
    sp_args = ['py-i2phosts-injector', '-s', '-f', tmpfile, '-d',
            'Auto-added from ' + source.name]
    try:
        p = subprocess.Popen(sp_args, shell=False, stdin=None,
                stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    except OSError, e:
        log.error('failed to exec py-i2phosts-injector: %s', e)
        if e.errno == errno.ENOENT:
            log.error('check your PATH environment variable')
        sys.exit(1)
    out = p.communicate()[0]
    os.remove(tmpfile)
    log.info('%s: injector output: \n%s', source.name, out)
    # update last_success
    source.last_success = datetime.datetime.utcnow()
    log.debug('%s: updating last_success timestamp: %s', source.name, source.last_success)
    source.save()