Be sure to return a valid URL in ParsedTextEntity.url

The URL received from Telegram in message entities is the raw one
posted by the sender of the message, which might lack a protocol. This
commit enforces a protocol to be present in ParsedTextEntity's URLs,
defaulting to HTTP if none is provided.

Author: Pietro Albini

botogram/objects/messages.py

@@ -6,6 +6,7 @@
     Released under the MIT license
 """
 
+import re
 
 from .base import BaseObject, _itself
 from . import mixins
@@ -16,6 +17,9 @@
     Location
 
 
+_url_protocol_re = re.compile(r"^https?:\/\/|s?ftp:\/\/|mailto:", re.I)
+
+
 def _require_message(func):
     """Decorator which forces the object to have an attached message"""
     @utils.wraps(func)
@@ -142,13 +146,12 @@ def text(self):
     @_require_message
     def url(self):
         """Get the URL attached to the message"""
-        # Use the provided if available
         if self._url is not None:
-            return self._url
-
-        if self.type == "link":
+            # Use the provided if available
+            url = self._url
+        elif self.type == "link":
             # Standard URLs
-            return self.text
+            url = self.text
         elif self.type == "mention":
             # telegram.me URLs
             return "https://telegram.me/%s" % self.text[1:]
@@ -159,6 +162,13 @@ def url(self):
             # Sorry!
             return None
 
+        # Be sure to have a protocol in the URL (default to HTTP)
+        # Apparently Telegram doesn't provide always a valid URL, but whatever
+        # the user sends in the message
+        if not _url_protocol_re.match(url):
+            url = "http://%s" % url
+        return url
+
 
 class ParsedText:
     """Collection of ParsedTextEntity.

tests/test_objects_messages.py

@@ -160,6 +160,13 @@ def test_parsed_text_entity_url():
     entity._url = "https://www.example.com"
     entity.set_message(msg)
     assert entity.url == "https://www.example.com"
+    entity._url = None
+
+    # Url entity without a scheme
+    msg = get_dummy_message("www.example.com/abcdefg")
+    entity.type = "url"
+    entity.set_message(msg)
+    assert entity.url == "http://www.example.com/abcdefg"
 
 
 def test_parsed_text():