X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rsimple.py;h=9ab40a3b7f30c37ee52a9c6fe20039608f5e47c1;hb=49357dd2797d8da9af35718a73dadc45b10b5119;hp=41fa2b33a607fe1db2e1a322520d47abd2cb6d46;hpb=c79561dfd3ff2e2a69b542211bfb7c692210776d;p=bookmarks_db.git diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py index 41fa2b3..9ab40a3 100644 --- a/Robots/bkmk_rsimple.py +++ b/Robots/bkmk_rsimple.py @@ -1,9 +1,16 @@ -""" - Simple, strightforward robot +"""Simple, strightforward robot + +This file is a part of Bookmarks database and Internet robot. - Written by Oleg Broytman. Copyright (C) 2000-2010 PhiloSoft Design. """ +__author__ = "Oleg Broytman " +__copyright__ = "Copyright (C) 2000-2012 PhiloSoft Design" +__license__ = "GNU GPL" + +__all__ = ['robot_simple', 'get_error'] + + import sys, os import time, urllib from base64 import b64encode @@ -19,7 +26,9 @@ from parse_html import parse_html class RedirectException(Exception): reloc_dict = { 301: "perm.", - 302: "temp.", + 302: "temp2.", + 303: "temp3.", + 307: "temp7.", "html": "html" } def __init__(self, errcode, newurl): @@ -187,9 +196,9 @@ class robot_simple(Robot): if icons[icon]: bookmark.icon_href = icon content_type, bookmark.icon = icons[icon] - self.log(" cached icon: %s" % content_type) + self.log(" cached icon: %s" % content_type) else: - self.log(" cached icon: no icon") + self.log(" cached icon: no icon") else: try: _icon = icon @@ -198,7 +207,7 @@ class robot_simple(Robot): icon_fname, headers = urllib.urlretrieve(_icon) except RedirectException, e: _icon = e.url - self.log(" redirect to : %s" % _icon) + self.log(" redirect to : %s" % _icon) else: break else: @@ -206,19 +215,23 @@ class robot_simple(Robot): except: etype, emsg, tb = sys.exc_info() self.log(" no icon : %s %s" % (etype, emsg)) - etype = None - emsg = None - tb = None + etype = emsg = tb = None icons[icon] = None else: content_type = headers["Content-Type"] - if content_type.startswith("image/"): + if content_type.startswith("application/") \ + or content_type.startswith("image/") \ + or content_type.startswith("text/plain"): icon_file = open(icon_fname, "rb") icon_data = icon_file.read() icon_file.close() bookmark.icon_href = icon - bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon_data)) self.log(" got icon : %s" % content_type) + if content_type.startswith("application/") \ + or content_type.startswith("text/plain"): + self.log(" non-image content type, assume x-icon") + content_type = 'image/x-icon' + bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon_data)) icons[icon] = (content_type, bookmark.icon) else: self.log(" no icon : bad content type '%s'" % content_type) @@ -275,7 +288,6 @@ class robot_simple(Robot): # Tested return 1 - def finish_check_url(self, bookmark, fname=None): # Calculate these attributes even in case of an error if fname and os.path.exists(fname):