mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-06-20 12:54:23 -04:00
avoid use of __double_underscore member variables because they're special https://shahriar.svbtle.com/underscores-in-python
This commit is contained in:
parent
becd832ea3
commit
21891476c4
2 changed files with 12 additions and 13 deletions
|
@ -100,8 +100,8 @@ class BrozzlerWorker:
|
||||||
self._max_browsers = max_browsers
|
self._max_browsers = max_browsers
|
||||||
|
|
||||||
# these two settings can be overridden by the job/site configuration
|
# these two settings can be overridden by the job/site configuration
|
||||||
self.__proxy = proxy
|
self._default_proxy = proxy
|
||||||
self.__enable_warcprox_features = enable_warcprox_features
|
self._default_enable_warcprox_features = enable_warcprox_features
|
||||||
|
|
||||||
self._browser_pool = brozzler.browser.BrowserPool(max_browsers,
|
self._browser_pool = brozzler.browser.BrowserPool(max_browsers,
|
||||||
chrome_exe=chrome_exe, ignore_cert_errors=True)
|
chrome_exe=chrome_exe, ignore_cert_errors=True)
|
||||||
|
@ -113,18 +113,17 @@ class BrozzlerWorker:
|
||||||
def _proxy(self, site):
|
def _proxy(self, site):
|
||||||
if site.proxy:
|
if site.proxy:
|
||||||
return site.proxy
|
return site.proxy
|
||||||
elif self.__proxy:
|
elif self._default_proxy:
|
||||||
return self.__proxy
|
return self._default_proxy
|
||||||
elif self._service_registry and (
|
elif self._service_registry and (
|
||||||
site.enable_warcprox_features or
|
site.enable_warcprox_features
|
||||||
self.__enable_warcprox_features):
|
or self._default_enable_warcprox_features):
|
||||||
warcprox_service = self._service_registry.available_service('warcprox')
|
svc = self._service_registry.available_service('warcprox')
|
||||||
site.proxy = '%s:%s' % (warcprox_service['host'],
|
site.proxy = '%s:%s' % (svc['host'], svc['port'])
|
||||||
warcprox_service['port'])
|
|
||||||
self._frontier.update_site(site)
|
self._frontier.update_site(site)
|
||||||
self.logger.info(
|
self.logger.info(
|
||||||
'chose warcprox %s from service registry for site %s',
|
'chose warcprox instance %s from service registry for %s',
|
||||||
site.proxy, site)
|
repr(site.proxy), site)
|
||||||
return site.proxy
|
return site.proxy
|
||||||
return None
|
return None
|
||||||
|
|
||||||
|
@ -133,7 +132,7 @@ class BrozzlerWorker:
|
||||||
if site.enable_warcprox_features is not None:
|
if site.enable_warcprox_features is not None:
|
||||||
return site.enable_warcprox_features
|
return site.enable_warcprox_features
|
||||||
else:
|
else:
|
||||||
return self.__enable_warcprox_features
|
return self._default_enable_warcprox_features
|
||||||
|
|
||||||
def _youtube_dl(self, destdir, site):
|
def _youtube_dl(self, destdir, site):
|
||||||
ydl_opts = {
|
ydl_opts = {
|
||||||
|
|
2
setup.py
2
setup.py
|
@ -32,7 +32,7 @@ def find_package_data(package):
|
||||||
|
|
||||||
setuptools.setup(
|
setuptools.setup(
|
||||||
name='brozzler',
|
name='brozzler',
|
||||||
version='1.1b7.dev104',
|
version='1.1b7.dev105',
|
||||||
description='Distributed web crawling with browsers',
|
description='Distributed web crawling with browsers',
|
||||||
url='https://github.com/internetarchive/brozzler',
|
url='https://github.com/internetarchive/brozzler',
|
||||||
author='Noah Levitt',
|
author='Noah Levitt',
|
||||||
|
|
Loading…
Add table
Add a link
Reference in a new issue