From 67889a1d1420da9ae4cd5ec0111bb3a1639385ea Mon Sep 17 00:00:00 2001
From: pictuga <contact@pictuga.com>
Date: Fri, 24 Mar 2017 21:49:14 -1000
Subject: [PATCH] readabilite: drop useless tags

This extra cluster actually jams the algorithm
---
 morss/readabilite.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/morss/readabilite.py b/morss/readabilite.py
index 5fb58e6..8c371b9 100644
--- a/morss/readabilite.py
+++ b/morss/readabilite.py
@@ -115,6 +115,13 @@ def clean_html(root):
             item.getparent().remove(item)
             continue
 
+        if item.tag in ['div'] \
+            and len(list(item.iterchildren())) <= 1 \
+            and not (item.text or '').strip() \
+            and not (item.tail or '').strip():
+            item.drop_tag()
+            continue
+
         class_id = item.get('class', '') + item.get('id', '')
         if regex_bad.match(class_id) is not None:
             item.getparent().remove(item)