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']
17 from Robots.bkmk_robot_base import robot_base, get_error
20 class RedirectException(Exception):
21 def __init__(self, errcode, newurl):
22 Exception.__init__(self)
23 self.errcode = errcode
27 class MyURLopener(urllib.URLopener):
28 # Error 301 -- relocated (permanently)
29 def http_error_301(self, url, fp, errcode, errmsg, headers, data=None):
30 if 'location' in headers:
31 newurl = headers['location']
32 elif 'uri' in headers:
33 newurl = headers['uri']
36 raise RedirectException(errcode, newurl)
38 # Error 302 -- relocated (temporarily)
39 http_error_302 = http_error_301
40 # Error 303 -- relocated (see other)
41 http_error_303 = http_error_301
42 # Error 307 -- relocated (temporarily)
43 http_error_307 = http_error_301
44 # Error 308 -- relocated (permanently)
45 http_error_308 = http_error_301
47 # Error 401 -- authentication required
48 def http_error_401(self, url, fp, errcode, errmsg, headers, data=None):
50 ('http error', errcode, "Authentication required ", headers))
52 def http_error_default(self, url, fp, errcode, errmsg, headers):
56 raise IOError(('http error', errcode, errmsg, headers))
59 urllib._urlopener = MyURLopener()
61 # Fake headers to pretend this is a real browser
62 _user_agent = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en)"
63 " Gecko/20001221 Firefox/2.0.0"
64 urllib._urlopener.addheaders[0] = ('User-Agent', _user_agent)
65 _x_user_agent = "bookmarks_db (Python %d.%d.%d; urllib/%s)" % (
66 sys.version_info[0], sys.version_info[1],
67 sys.version_info[2], urllib.__version__
69 urllib._urlopener.addheader('X-User-Agent', _x_user_agent)
70 urllib._urlopener.addheader('Referer', '')
72 urllib._urlopener.addheader('Accept', '*/*')
73 urllib._urlopener.addheader('Accept-Language', 'ru,en')
74 urllib._urlopener.addheader('Cache-Control', 'max-age=300')
75 urllib._urlopener.addheader('Connection', 'close')
78 urllib_ftpwrapper = urllib.ftpwrapper
82 class myftpwrapper(urllib_ftpwrapper):
83 def __init__(self, user, passwd, host, port, dirs):
84 urllib_ftpwrapper.__init__(self, user, passwd, host, port, dirs)
86 ftpcache_key = (user, host, port, '/'.join(dirs))
89 urllib.ftpwrapper = myftpwrapper
92 class robot_urllib(robot_base):
93 def get(self, bookmark, url, accept_charset=False):
95 # Set fake referer to the base URL
96 urllib._urlopener.addheaders[2] = ('Referer', url)
98 if accept_charset and bookmark.charset:
99 urllib._urlopener.addheader('Accept-Charset', bookmark.charset)
101 fname, headers = urllib.urlretrieve(url)
103 if accept_charset and bookmark.charset:
104 # Remove Accept-Charset
105 del urllib._urlopener.addheaders[-1]
107 infile = open(fname, 'rt')
108 content = infile.read()
111 return None, None, None, headers, content
113 except RedirectException as e:
114 return None, e.errcode, e.newurl, None, None
117 if (e[0] == "http error") and (e[1] == -1):
119 bookmark.no_error = "The server did not return any header - "
120 "it is not an error, actually"
121 self.log(' no headers: %s' % bookmark.no_error)
124 self.log(' Error: %s' % error)
126 return error, None, None, None, None
128 def get_ftp_welcome(self):
130 _welcome = urllib._urlopener.ftpcache[ftpcache_key].ftp.welcome
131 # I am assuming there are no duplicate ftp URLs in db.
132 # If there are - ftpcache_key in next line is invalid.
136 def finish_check_url(self, bookmark):
137 robot_base.finish_check_url(self, bookmark)