1 """Robot based on urllib2
3 This file is a part of Bookmarks database and Internet robot.
7 __author__ = "Oleg Broytman <phd@phdru.name>"
8 __copyright__ = "Copyright (C) 2014-2023 PhiloSoft Design"
9 __license__ = "GNU GPL"
11 __all__ = ['robot_urllib2']
17 from Robots.bkmk_robot_base import robot_base, get_error
23 class FTPHandler(urllib2.FTPHandler):
24 def connect_ftp(self, user, passwd, host, port, dirs, timeout):
26 _fw = urllib2.FTPHandler.connect_ftp(self, user, passwd, host, port,
31 opener = urllib2.OpenerDirector()
32 default_classes = [urllib2.HTTPHandler, urllib2.HTTPDefaultErrorHandler,
33 FTPHandler, urllib2.HTTPErrorProcessor]
34 if hasattr(httplib, 'HTTPS'):
35 default_classes.insert(0, urllib2.HTTPSHandler)
36 for klass in default_classes:
37 opener.add_handler(klass())
39 urllib2.install_opener(opener)
42 # Fake headers to pretend this is a real browser
43 _user_agent = "Mozilla/5.0 (X11; Linux i686; rv:30.0)"
44 " Gecko/20100101 Firefox/30.0"
45 _x_user_agent = "bookmarks_db (Python %d.%d.%d; urllib2)" % (
46 sys.version_info[0], sys.version_info[1], sys.version_info[2])
49 class robot_urllib2(robot_base):
50 def get(self, bookmark, url, accept_charset=False):
51 request = urllib2.Request(url)
52 request.add_header('Accept', '*/*')
53 if accept_charset and bookmark.charset:
54 request.add_header('Accept-Charset', bookmark.charset)
55 request.add_header('Accept-Language', 'ru,en')
56 request.add_header('Cache-Control', 'max-age=300')
57 request.add_header('Connection', 'close')
58 request.add_header('Referer', url)
59 request.add_header('User-agent', _user_agent)
60 request.add_header('X-User-Agent', _x_user_agent)
66 response = urllib2.urlopen(request)
68 except urllib2.HTTPError as e:
69 if e.code in (301, 302, 303, 307):
70 return None, e.code, e.hdrs['Location'], None, None
72 self.log(' HTTP Error %s: %s' % (e.code, e.msg))
73 return ("HTTP Error %s: %s" % (e.code, e.msg),
74 None, None, None, None)
76 except urllib2.URLError as e:
77 self.log(' URL Error: %s' % e.reason)
78 return "URL Error: %s" % e.reason, None, None, None, None
80 except httplib.HTTPException as e:
82 self.log(' HTTP Exception: %s' % error)
83 return "HTTP Exception: %s" % error, None, None, None, None
87 self.log(' I/O Error: %s' % error)
88 return "I/O Error: %s" % error, None, None, None, None
91 return None, None, None, response.info(), response.read()
93 def get_ftp_welcome(self):
96 return _fw.ftp.welcome