1 """Simple, strightforward robot based on urllib
3 This file is a part of Bookmarks database and Internet robot.
7 __author__ = "Oleg Broytman <phd@phdru.name>"
8 __copyright__ = "Copyright (C) 2000-2023 PhiloSoft Design"
9 __license__ = "GNU GPL"
11 __all__ = ['robot_urllib']
19 from Robots.bkmk_robot_base import robot_base, get_error
22 class RedirectException(Exception):
23 def __init__(self, errcode, newurl):
24 Exception.__init__(self)
25 self.errcode = errcode
29 class MyURLopener(urllib.URLopener):
30 # Error 301 -- relocated (permanently)
31 def http_error_301(self, url, fp, errcode, errmsg, headers, data=None):
32 if headers.has_key('location'):
33 newurl = headers['location']
34 elif headers.has_key('uri'):
35 newurl = headers['uri']
38 raise RedirectException(errcode, newurl)
40 # Error 302 -- relocated (temporarily)
41 http_error_302 = http_error_301
42 # Error 303 -- relocated (see other)
43 http_error_303 = http_error_301
44 # Error 307 -- relocated (temporarily)
45 http_error_307 = http_error_301
47 # Error 401 -- authentication required
48 def http_error_401(self, url, fp, errcode, errmsg, headers, data=None):
49 raise IOError(('http error', errcode, "Authentication required ", headers))
51 def http_error_default(self, url, fp, errcode, errmsg, headers):
55 raise IOError(('http error', errcode, errmsg, headers))
58 urllib._urlopener = MyURLopener()
60 # Fake headers to pretend this is a real browser
61 _user_agent = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en) Gecko/20001221 Firefox/2.0.0"
62 urllib._urlopener.addheaders[0] = ('User-Agent', _user_agent)
63 _x_user_agent = "bookmarks_db (Python %d.%d.%d; urllib/%s)" % (
64 sys.version_info[0], sys.version_info[1], sys.version_info[2], urllib.__version__)
65 urllib._urlopener.addheader('X-User-Agent', _x_user_agent)
66 urllib._urlopener.addheader('Referer', '')
68 urllib._urlopener.addheader('Accept', '*/*')
69 urllib._urlopener.addheader('Accept-Language', 'ru,en')
70 urllib._urlopener.addheader('Cache-Control', 'max-age=300')
71 urllib._urlopener.addheader('Connection', 'close')
74 urllib_ftpwrapper = urllib.ftpwrapper
78 class myftpwrapper(urllib_ftpwrapper):
79 def __init__(self, user, passwd, host, port, dirs):
80 urllib_ftpwrapper.__init__(self, user, passwd, host, port, dirs)
82 ftpcache_key = (user, host, port, '/'.join(dirs))
85 urllib.ftpwrapper = myftpwrapper
88 class robot_urllib(robot_base):
89 def get(self, bookmark, url, accept_charset=False):
91 # Set fake referer to the base URL
92 urllib._urlopener.addheaders[2] = ('Referer', url)
94 if accept_charset and bookmark.charset:
95 urllib._urlopener.addheader('Accept-Charset', bookmark.charset)
97 fname, headers = urllib.urlretrieve(url)
99 if accept_charset and bookmark.charset:
100 del urllib._urlopener.addheaders[-1] # Remove Accept-Charset
102 infile = open(fname, 'rb')
103 content = infile.read()
106 return None, None, None, headers, content
108 except RedirectException as e:
109 return None, e.errcode, e.newurl, None, None
112 if (e[0] == "http error") and (e[1] == -1):
114 bookmark.no_error = "The server did not return any header - it is not an error, actually"
115 self.log(' no headers: %s' % bookmark.no_error)
118 self.log(' Error: %s' % error)
120 return error, None, None, None, None
122 def get_ftp_welcome(self):
124 _welcome = urllib._urlopener.ftpcache[ftpcache_key].ftp.welcome
125 # I am assuming there are no duplicate ftp URLs in db.
126 # If there are - ftpcache_key in next line is invalid.
130 def finish_check_url(self, bookmark):
131 robot_base.finish_check_url(self, bookmark)