From 64e41b807d5605dbe52b2320fd017f9534506022 Mon Sep 17 00:00:00 2001
From: pictuga <contact@pictuga.com>
Date: Sat, 2 May 2020 19:17:15 +0200
Subject: [PATCH] crawler: handle http:/ (single slash)

Fixing one more corner case! malayalam.oneindia.com
---
 morss/crawler.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/morss/crawler.py b/morss/crawler.py
index cc98838..8adf495 100644
--- a/morss/crawler.py
+++ b/morss/crawler.py
@@ -129,6 +129,9 @@ def sanitize_url(url):
     if url.split(':', 1)[0] not in PROTOCOL:
         url = 'http://' + url
 
+    # turns out some websites have really badly fomatted urls (fix http:/badurl)
+    url = re.sub('^(https?):/([^/])', r'\1://\2', url)
+
     url = url.replace(' ', '%20')
 
     # Escape non-ascii unicode characters