X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rsimple.py;h=ae7c39e481be8e30b144981bf18a1a8aa32b19d2;hb=b9b055b270749b6af5d83f1e6390793d7b951888;hp=77c7a150109509583c2eb3d5fb5f538d45806283;hpb=6dedb3549bfb4accfd29f7ccc5533c1e3f5a7348;p=bookmarks_db.git diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py index 77c7a15..ae7c39e 100644 --- a/Robots/bkmk_rsimple.py +++ b/Robots/bkmk_rsimple.py @@ -21,6 +21,7 @@ class RedirectException(Exception): } def __init__(self, errcode, newurl): Exception.__init__(self, "(%s) to %s" % (self.reloc_dict[errcode], newurl)) + self.url = newurl class MyURLopener(urllib.URLopener): @@ -156,6 +157,10 @@ class robot_simple(Robot): except (IndexError, ValueError): raise RedirectException("html", "Bad redirect to %s (%s)" % (url, refresh)) else: + try: + timeout = int(refresh.split(';')[0]) + except ValueError: + pass # float timeout raise RedirectException("html", "%s (%s sec)" % (url, timeout)) # Get favicon.ico @@ -163,9 +168,18 @@ class robot_simple(Robot): if not icon: icon = "/favicon.ico" icon = urljoin("%s://%s%s" % (url_type, url_host, url_path), icon) - if self.log: self.log(" icon : %s" % icon) + if self.log: self.log(" looking for icon at : %s" % icon) try: - fname, headers = urllib.urlretrieve(icon) + for i in range(8): + try: + fname, headers = urllib.urlretrieve(icon) + except RedirectException, e: + icon = e.url + if self.log: self.log(" redirect to : %s" % icon) + else: + break + else: + raise IOError("Too many redirects") except: etype, emsg, tb = sys.exc_info() if self.log: self.log(" no icon : %s %s" % (etype, emsg)) @@ -179,8 +193,9 @@ class robot_simple(Robot): icon = icon_file.read() icon_file.close() bookmark.icon = "data:%s;base64,%s" % (content_type, b64encode(icon)) + if self.log: self.log(" got icon : %s" % content_type) else: - if self.log: self.log(" no icon : %s" % content_type) + if self.log: self.log(" no icon : bad content type %s" % content_type) os.remove(fname) except KeyError: