From a531d135b5964e673d8c4af70564faa209f8498e Mon Sep 17 00:00:00 2001
From: smilerz <smilerz@gmail.com>
Date: Wed, 14 Apr 2021 10:42:31 -0500
Subject: [PATCH] remove html tags from description

---
 cookbook/helper/recipe_url_import.py | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/cookbook/helper/recipe_url_import.py b/cookbook/helper/recipe_url_import.py
index de1fb624..cd9e33a7 100644
--- a/cookbook/helper/recipe_url_import.py
+++ b/cookbook/helper/recipe_url_import.py
@@ -21,7 +21,7 @@ def get_from_scraper(scrape, space):
     except AttributeError:
         description = ''
 
-    recipe_json['description'] = normalize_string(description)
+    recipe_json['description'] = parse_description(description)
 
     try:
         servings = scrape.yields()
@@ -181,6 +181,14 @@ def parse_ingredients(ingredients):
     return ingredients
 
 
+def parse_description(description):
+    description = re.sub(r'\n\s*\n', '\n\n', description)
+    description = re.sub(' +', ' ', description)
+    description = re.sub('</p>', '\n', description)
+    description = re.sub('<[^<]+?>', '', description)
+    return normalize_string(description)
+
+
 def parse_instructions(instructions):
     instruction_text = ''