__all__ = ['robot_urllib']
-import sys, os
-import time, urllib
+import sys
+import urllib
+
from Robots.bkmk_robot_base import robot_base, get_error
class MyURLopener(urllib.URLopener):
# Error 301 -- relocated (permanently)
def http_error_301(self, url, fp, errcode, errmsg, headers, data=None):
- if headers.has_key('location'):
+ if 'location' in headers:
newurl = headers['location']
- elif headers.has_key('uri'):
+ elif 'uri' in headers:
newurl = headers['uri']
else:
newurl = "Nowhere"
# Error 401 -- authentication required
def http_error_401(self, url, fp, errcode, errmsg, headers, data=None):
- raise IOError(('http error', errcode, "Authentication required ", headers))
+ raise IOError(
+ ('http error', errcode, "Authentication required ", headers))
def http_error_default(self, url, fp, errcode, errmsg, headers):
if fp:
- void = fp.read()
+ fp.read()
fp.close()
raise IOError(('http error', errcode, errmsg, headers))
urllib._urlopener = MyURLopener()
# Fake headers to pretend this is a real browser
-_user_agent = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en) Gecko/20001221 Firefox/2.0.0"
+_user_agent = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en)"
+" Gecko/20001221 Firefox/2.0.0"
urllib._urlopener.addheaders[0] = ('User-Agent', _user_agent)
_x_user_agent = "bookmarks_db (Python %d.%d.%d; urllib/%s)" % (
- sys.version_info[0], sys.version_info[1], sys.version_info[2], urllib.__version__)
+ sys.version_info[0], sys.version_info[1],
+ sys.version_info[2], urllib.__version__
+)
urllib._urlopener.addheader('X-User-Agent', _x_user_agent)
urllib._urlopener.addheader('Referer', '')
fname, headers = urllib.urlretrieve(url)
finally:
if accept_charset and bookmark.charset:
- del urllib._urlopener.addheaders[-1] # Remove Accept-Charset
+ # Remove Accept-Charset
+ del urllib._urlopener.addheaders[-1]
- infile = open(fname, 'rb')
+ infile = open(fname, 'rt')
content = infile.read()
infile.close()
except IOError as e:
if (e[0] == "http error") and (e[1] == -1):
error = None
- bookmark.no_error = "The server did not return any header - it is not an error, actually"
+ bookmark.no_error = "The server did not return any header - "
+ "it is not an error, actually"
self.log(' no headers: %s' % bookmark.no_error)
else:
error = get_error(e)