From 6ce808c4994ce3e981a3d3023b7c8f626b03c320 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Mon, 5 Nov 2018 12:43:20 +0530
Subject: [PATCH] Allow saving index html easily

---
 src/calibre/web/feeds/news.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/calibre/web/feeds/news.py b/src/calibre/web/feeds/news.py
index d131524114..3277d957e0 100644
--- a/src/calibre/web/feeds/news.py
+++ b/src/calibre/web/feeds/news.py
@@ -657,7 +657,7 @@ class BasicNewsRecipe(Recipe):
             return frozenset()
         return frozenset([(parts.netloc, (parts.path or '').rstrip('/'))])
 
-    def index_to_soup(self, url_or_raw, raw=False, as_tree=False):
+    def index_to_soup(self, url_or_raw, raw=False, as_tree=False, save_raw=None):
         '''
         Convenience method that takes an URL to the index page and returns
         a `BeautifulSoup <https://www.crummy.com/software/BeautifulSoup/bs3/documentation.html>`_
@@ -692,6 +692,9 @@ class BasicNewsRecipe(Recipe):
         else:
             _raw = xml_to_unicode(_raw, strip_encoding_pats=True, resolve_entities=True)[0]
         _raw = clean_xml_chars(_raw)
+        if save_raw:
+            with lopen(save_raw, 'wb') as f:
+                f.write(_raw.encode('utf-8'))
         if as_tree:
             from html5_parser import parse
             return parse(_raw)