From 9918da68eb8ec300eb3c2ec100614fbae3a8c1ef Mon Sep 17 00:00:00 2001
From: mdivecky <matej@divecky.com>
Date: Wed, 16 Aug 2023 12:41:29 +0200
Subject: [PATCH] add dirty exceptions to skip broken RSS feeds

---
 processor/app.py | 57 +++++++++++++++++++++++++++++-------------------
 1 file changed, 34 insertions(+), 23 deletions(-)

diff --git a/processor/app.py b/processor/app.py
index b13b2c7..58c5bac 100644
--- a/processor/app.py
+++ b/processor/app.py
@@ -112,30 +112,41 @@ def process_item(article, rc):
 
 
 for feed in config['feeds']:
-	rss_source = str(feed['rss_source'])
-	unique_tag = str(feed['unique_tag'])
-	name = str(feed['name'])
+	try:
+		rss_source = str(feed['rss_source'])
+		unique_tag = str(feed['unique_tag'])
+		name = str(feed['name'])
 
-	rss = feedparser.parse(rss_source)
+		rss = feedparser.parse(rss_source)
 
-	for item in rss['entries']:
-		rss_id = item[unique_tag]
-		title = item['title']
-		#description = item['description'] ## Don't store description for now, as we don't need it and it's big.
-		published = time.strftime('%Y:%m:%d %H:%M:%S %Z %z', item['published_parsed'])
-		link = item['link']
-		article_data = {
-			'title'		:	title,
-			#'description':	description,
-			'published'	:	published,
-			'link'		:	link,
-			'medium'	:	name
-		}
-		article = {
-			'rss_id'	: rss_id, 
-			'content'	: article_data
-		}
-		article_count += 1
-		process_item(article, rc)
+		for item in rss['entries']:
+			try:
+				rss_id = item[unique_tag]
+				title = item['title']
+				#description = item['description'] ## Don't store description for now, as we don't need it and it's big.
+				published = time.strftime('%Y:%m:%d %H:%M:%S %Z %z', item['published_parsed'])
+				link = item['link']
+				article_data = {
+					'title'		:	title,
+					#'description':	description,
+					'published'	:	published,
+					'link'		:	link,
+					'medium'	:	name
+				}
+				article = {
+					'rss_id'	: rss_id, 
+					'content'	: article_data
+				}
+				article_count += 1
+				process_item(article, rc)
+			except Exception as e:
+				print("Parsing article failed")
+				print(e)
+				print(item)
+	except Exception as e:
+		print("Parsing feed failed.")
+		print(e)
+		print(feed)
+		pass
 
 print("Processed articles: " + str(article_count))