X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rsimple.py;h=c34a12728d136307da774e5a6ac086622bf51515;hb=d655a6006176ecd02b8bbf6fe942a8f518f30e13;hp=2a18beb72edad934fa1120c1f9e9fb3035130c75;hpb=199a8ce519d73ee4e7629657b2119b3213efb3e9;p=bookmarks_db.git diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py index 2a18beb..c34a127 100644 --- a/Robots/bkmk_rsimple.py +++ b/Robots/bkmk_rsimple.py @@ -1,10 +1,17 @@ """ - Simple, strightforward robot; guaranteed to has problems with timeouts :) + Simple, strightforward robot - Written by BroytMann, Mar 2000 - Aug 2002. Copyright (C) 2000-2002 PhiloSoft Design + Written by Oleg BroytMann. Copyright (C) 2000-2007 PhiloSoft Design. """ +import string, os +import time, urllib +from urlparse import urljoin +from m_lib.net.www.util import parse_time +from m_lib.md5wrapper import md5wrapper + + class RedirectException(Exception): reloc_dict = { 301: "perm.", @@ -15,12 +22,6 @@ class RedirectException(Exception): Exception.__init__(self, "(%s) to %s" % (self.reloc_dict[errcode], newurl)) -import string, os -import time, urllib -from m_lib.net.www.util import parse_time -from m_lib.md5wrapper import md5wrapper - - class MyURLopener(urllib.URLopener): # Error 302 -- relocated (temporarily) def http_error_302(self, url, fp, errcode, errmsg, headers, data=None): @@ -91,7 +92,9 @@ class robot_simple(Robot): url_host, url_path = urllib.splithost(url_rest) url_path, url_tag = urllib.splittag(url_path) + if bookmark.charset: urllib._urlopener.addheader('Accept-Charset', bookmark.charset) fname, headers = urllib.urlretrieve("%s://%s%s" % (url_type, url_host, url_path), self.tempfname) + if bookmark.charset: del urllib._urlopener.addheaders[-1] size = 0 last_modified = None @@ -155,6 +158,11 @@ class robot_simple(Robot): raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh)) else: raise RedirectException("html", "%s (%d sec)" % (url, timeout)) + icon = parser.icon + if not icon: + icon = "/favicon.ico" + icon = urljoin("%s://%s" % (url_type, url_host), icon) + if self.log: self.log(" icon : %s" % icon) except KeyError: pass