"""
-import string, os
+import sys, os, string
import time, urllib
+from base64 import b64encode
from urlparse import urljoin
from m_lib.net.www.util import parse_time
from m_lib.md5wrapper import md5wrapper
urllib._urlopener = MyURLopener()
-# Some sites allow only Mozilla-compatible browsers; way to stop robots?
-server_version = "Mozilla/3.0 (compatible; Python-urllib/%s)" % urllib.__version__
+server_version = "bookmarks_db (Python-urllib/%s)" % urllib.__version__
urllib._urlopener.addheaders[0] = ('User-agent', server_version)
from parse_html import parse_html
class robot_simple(Robot):
- def check_url(self, bookmark, url_type, url_rest):
+ def check_url(self, bookmark):
if not self.tempfname:
self.tempfname = bookmark.tempfname
try:
try:
self.start = int(time.time())
+ bookmark.icon = None
+
+ url_type, url_rest = urllib.splittype(bookmark.href)
url_host, url_path = urllib.splithost(url_rest)
url_path, url_tag = urllib.splittag(url_path)
if self.log: self.log(" no charset in Content-Type header")
if content_type == "text/html":
parser = parse_html(fname, charset, self.log)
- title = parser.title.replace('\r', '').replace('\n', ' ').strip()
- bookmark.real_title = parser.unescape(title)
- if self.log: self.log(" final title : %s" % bookmark.real_title)
+ bookmark.real_title = parser.title
if parser.refresh:
refresh = parser.refresh
try:
except IndexError:
url = "self"
try:
- timeout = int(refresh.split(';')[0])
+ timeout = float(refresh.split(';')[0])
except (IndexError, ValueError):
- timeout = None
- if timeout is None:
raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh))
else:
- raise RedirectException("html", "%s (%d sec)" % (url, timeout))
+ raise RedirectException("html", "%s (%s sec)" % (url, timeout))
+
+ # Get favicon.ico
icon = parser.icon
if not icon:
icon = "/favicon.ico"
- icon = urljoin("%s://%s" % (url_type, url_host), icon)
+ icon = urljoin("%s://%s%s" % (url_type, url_host, url_path), icon)
if self.log: self.log(" icon : %s" % icon)
+ try:
+ fname, headers = urllib.urlretrieve(icon)
+ except:
+ etype, emsg, tb = sys.exc_info()
+ if self.log: self.log(" no icon : %s %s" % (etype, emsg))
+ etype = None
+ emsg = None
+ tb = None
+ else:
+ content_type = headers["Content-Type"]
+ if content_type.startswith("image/"):
+ icon_file = open(fname, "rb")
+ icon = icon_file.read()
+ icon_file.close()
+ bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon))
+ else:
+ if self.log: self.log(" no icon : %s" % content_type)
+ os.remove(fname)
+
except KeyError:
pass