X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rurllib.py;h=b41722fade3af41f245429e11689c3ffae55bb58;hb=71900f3630cb51580964038b78100d60e3671981;hp=f0a614e3799e1ccae9750bd57fbbf5c4f99ccd90;hpb=9d989c8c9ebd7f13e9f0fdc95fa139c99ed83387;p=bookmarks_db.git

diff --git a/Robots/bkmk_rurllib.py b/Robots/bkmk_rurllib.py
index f0a614e..b41722f 100644
--- a/Robots/bkmk_rurllib.py
+++ b/Robots/bkmk_rurllib.py
@@ -5,7 +5,7 @@ This file is a part of Bookmarks database and Internet robot.
 """
 
 __author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 2000-2014 PhiloSoft Design"
+__copyright__ = "Copyright (C) 2000-2017 PhiloSoft Design"
 __license__ = "GNU GPL"
 
 __all__ = ['robot_urllib']
@@ -13,12 +13,18 @@ __all__ = ['robot_urllib']
 
 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'):
@@ -27,36 +33,38 @@ class MyURLopener(urllib.URLopener):
          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): 
-      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.close()
-      raise IOError, ('http error', errcode, errmsg, headers)
+      raise IOError(('http error', errcode, errmsg, headers))
 
 
 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
@@ -72,32 +80,38 @@ urllib.ftpwrapper = myftpwrapper
 
 
 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):
+      except IOError, 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"
             self.log('   no headers: %s' % bookmark.no_error)
          else:
-            bookmark.error = get_error(msg)
-            self.log('   Error: %s' % bookmark.error)
+            error = get_error(e)
+            self.log('   Error: %s' % error)
 
-         return None, None
+         return error, None, None, None, None
 
    def get_ftp_welcome(self):
       global ftpcache_key
@@ -106,5 +120,6 @@ class robot_urllib(robot_base):
                           # 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()