import sys, os
import time, urllib
-from Robots.bkmk_robot_base import robot_base, RedirectException, get_error
+from Robots.bkmk_robot_base import robot_base, get_error
+class RedirectException(Exception):
+ def __init__(self, errcode, newurl):
+ Exception.__init__(self)
+ self.errcode = errcode
+ self.newurl = newurl
+
class MyURLopener(urllib.URLopener):
- # Error 302 -- relocated (temporarily)
- def http_error_302(self, url, fp, errcode, errmsg, headers, data=None):
+ # Error 301 -- relocated (permanently)
+ def http_error_301(self, url, fp, errcode, errmsg, headers, data=None):
if headers.has_key('location'):
newurl = headers['location']
elif headers.has_key('uri'):
newurl = "Nowhere"
raise RedirectException(errcode, newurl)
- # Error 301 -- also relocated (permanently)
- http_error_301 = http_error_302
- # Error 307 -- also relocated (temporary)
- http_error_307 = http_error_302
+ # Error 302 -- relocated (temporarily)
+ http_error_302 = http_error_301
+ # Error 303 -- relocated (see other)
+ http_error_303 = http_error_301
+ # Error 307 -- relocated (temporarily)
+ http_error_307 = http_error_301
# Error 401 -- authentication required
def http_error_401(self, url, fp, errcode, errmsg, headers, data=None):
urllib._urlopener = MyURLopener()
# Fake headers to pretend this is a real browser
-_version = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en) Gecko/20001221 Firefox/2.0.0"
-urllib._urlopener.addheaders[0] = ('User-Agent', _version)
-_version = "bookmarks_db (Python %d.%d.%d; urllib/%s)" % (
+_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__)
-urllib._urlopener.addheader('X-User-Agent', _version)
+urllib._urlopener.addheader('X-User-Agent', _x_user_agent)
urllib._urlopener.addheader('Referer', '')
-urllib._urlopener.addheader('Connection', 'close')
urllib._urlopener.addheader('Accept', '*/*')
urllib._urlopener.addheader('Accept-Language', 'ru,en')
urllib._urlopener.addheader('Cache-Control', 'max-age=300')
+urllib._urlopener.addheader('Connection', 'close')
urllib_ftpwrapper = urllib.ftpwrapper
class robot_urllib(robot_base):
- def urlretrieve(self, bookmark, url, accept_charset=False):
+ def get(self, bookmark, url, accept_charset=False):
try:
# Set fake referer to the base URL
urllib._urlopener.addheaders[2] = ('Referer', url)
if accept_charset and bookmark.charset:
urllib._urlopener.addheader('Accept-Charset', bookmark.charset)
- fname, headers = urllib.urlretrieve(url)
- if accept_charset and bookmark.charset:
- del urllib._urlopener.addheaders[-1]
+ try:
+ fname, headers = urllib.urlretrieve(url)
+ finally:
+ if accept_charset and bookmark.charset:
+ del urllib._urlopener.addheaders[-1] # Remove Accept-Charset
infile = open(fname, 'rb')
content = infile.read()
infile.close()
- return headers, content
+ return None, None, None, headers, content
+
+ except RedirectException, e:
+ return None, e.errcode, e.newurl, None, None
except IOError, msg:
if (msg[0] == "http error") and (msg[1] == -1):
+ error = None
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:
- bookmark.error = get_error(msg)
- self.log(' Error: %s' % bookmark.error)
+ error = get_error(msg)
+ self.log(' Error: %s' % error)
- return None, None
+ return error, None, None, None, None
def get_ftp_welcome(self):
global ftpcache_key
# If there are - ftpcache_key in prev line is invalid.
return _welcome
- def cleanup(self):
+ def finish_check_url(self, bookmark):
+ robot_base.finish_check_url(self, bookmark)
urllib.urlcleanup()