def setTorProxy():
# TOR Socks proxy
isTorEnabled = CFG_CRAWLER.get("clientConfig").get("tor_enabled", False)
if isTorEnabled:
torProxyAdress = CFG_CRAWLER.get("clientConfig").get("tor_proxyadress", "localhost")
torProxyPort = CFG_CRAWLER.get("clientConfig").get("tor_proxyport", 9050)
# Route an HTTP request through the SOCKS proxy
socks.setdefaultproxy(socks.PROXY_TYPE_SOCKS5, torProxyAdress, torProxyPort)
socket.socket = socks.socksocket
def send_supergros(self, xml_file):
if config.use_proxy:
logger.info('connecting to proxy')
socks.setdefaultproxy(socks.PROXY_TYPE_SOCKS5,"127.0.0.1", 9999)
socket.socket = socks.socksocket
logger.info('ftp-ing ro supergros')
ftp = ftplib.FTP(config.supergros_server)
ftp.login(config.supergros_user, config.supergros_password)
ftp.cwd('PROD/EDISGR/D0901-FROM-EDISGR')
self.do_ftp(ftp, xml_file)
def setTorProxy():
# TOR Socks proxy
isTorEnabled = CFG_CRAWLER.get("clientConfig").get("tor_enabled", False)
if isTorEnabled:
torProxyAdress = CFG_CRAWLER.get("clientConfig").get(
"tor_proxyadress", "localhost")
torProxyPort = CFG_CRAWLER.get("clientConfig").get(
"tor_proxyport", 9050)
# Route an HTTP request through the SOCKS proxy
socks.setdefaultproxy(socks.PROXY_TYPE_SOCKS5, torProxyAdress,
torProxyPort)
socket.socket = socks.socksocket
def runCrawler():
mapURL = {}
cfgCrawler = Config(os.path.join(RAGPICKER_ROOT, 'config', 'crawler.conf'))
#TOR Socks proxy
isTorEnabled = cfgCrawler.get("clientConfig").get("tor_enabled", False)
if isTorEnabled:
torProxyAdress = cfgCrawler.get("clientConfig").get("tor_proxyadress", "localhost")
torProxyPort = cfgCrawler.get("clientConfig").get("tor_proxyport", 9050)
# Route an HTTP request through the SOCKS proxy
socks.setdefaultproxy(socks.PROXY_TYPE_SOCKS5, torProxyAdress, torProxyPort)
socket.socket = socks.socksocket
# Build the PluginManager
crawlerPluginManager = PluginManager()
crawlerPluginManager.setPluginPlaces(["crawler"])
crawlerPluginManager.collectPlugins()
# Trigger run from the "Crawler" plugins
for pluginInfo in crawlerPluginManager.getAllPlugins():
crawlerModul = pluginInfo.plugin_object
#Config for crawler module
try:
options = cfgCrawler.get(pluginInfo.name)
crawlerModul.set_options(options)
except Exception:
log.error("Crawler module %s not found in configuration file", pluginInfo.name)
# If the crawler module is disabled in the config, skip it.
if not options.enabled:
continue
log.info("Run Crawler: " + pluginInfo.name)
try:
returnMap = crawlerModul.run()
mapURL.update(returnMap)
except Exception as e:
log.error('Error (%s) in %s', e, pluginInfo.name)
return mapURL