improve url exists detection for new comment threads

isso/utils.py

@@ -32,12 +32,23 @@ class IssoEncoder(json.JSONEncoder):
         return json.JSONEncoder.default(self, obj)
 
 
-def urlexists(host, path):
+def normalize(host):
+
+    if not host.startswith(('http://', 'https://')):
+        host = 'https://' + host
 
     rv = urlparse(host)
-    http = httplib.HTTPSConnection if rv.scheme == 'https' else httplib.HTTPConnection
+    if rv.scheme == 'https':
+        return (rv.netloc, 443)
+    return (rv.netloc.rsplit(':')[0], rv.port or 80)
 
-    with closing(http(rv.netloc, rv.port, timeout=3)) as con:
+
+def urlexists(host, path):
+
+    host, port = normalize(host)
+    http = httplib.HTTPSConnection if port == 443 else httplib.HTTPConnection
+
+    with closing(http(host, port, timeout=3)) as con:
         try:
             con.request('HEAD', path)
         except (httplib.HTTPException, socket.error):
@@ -49,10 +60,10 @@ def heading(host, path):
     """Connect to `host`, GET path and start from #isso-thread to search for
     a possible heading (h1). Returns `None` if nothing found."""
 
-    rv = urlparse(host)
-    http = httplib.HTTPSConnection if rv.scheme == 'https' else httplib.HTTPConnection
+    host, port = normalize(host)
+    http = httplib.HTTPSConnection if port == 443 else httplib.HTTPConnection
 
-    with closing(http(rv.netloc, rv.port, timeout=15)) as con:
+    with closing(http(host, port, timeout=15)) as con:
         con.request('GET', path)
         html = html5lib.parse(con.getresponse().read(), treebuilder="dom")
 

isso/views/comment.py

@@ -43,7 +43,7 @@ class requires:
 @requires(str, 'uri')
 def create(app, environ, request, uri):
 
-    if not utils.urlexists(app.ORIGIN, uri):
+    if app.db.threads.get(uri) is None and not utils.urlexists(app.ORIGIN, uri):
         return Response('URI does not exist', 404)
 
     try:
@@ -70,13 +70,12 @@ def create(app, environ, request, uri):
 
         hash=hashlib.md5(hash).hexdigest())
 
-    try:
-        title = app.db.threads.get(uri)
-    except KeyError:
+    title = app.db.threads.get(uri)
+    if title is None:
         title = app.db.threads.add(uri, utils.heading(app.ORIGIN, uri))
 
     try:
-        rv = app.db.add(uri, comment, utils.anonymize(unicode(request.remote_addr)))
+        rv = app.db.add(uri, comment, request.remote_addr)
     except sqlite3.Error:
         logging.exception('uncaught SQLite3 exception')
         abort(400)