def process_feed(self, feed_obj):
"""Processes sinlge feed"""
parsed_feed = ps.parse_feed(feed_obj)
schema.sanitize(parsed_feed)
for data in ProcessManager.process_feed_content(parsed_feed):
data["rss_id"] = feed_obj.rss_id
data["data_id"] = db.insert_update_data(data)
db.insert_update_tokenized(data)
self.processed_items += 1
parsed_feed["rss_id"] = feed_obj.rss_id
db.insert_update_feed(parsed_feed)
def process_feed_content(parsed_feed):
"""Processes sinlge feed content"""
for parsed_content in ps.parse_content(parsed_feed):
schema.sanitize(parsed_content)
schema.serialize(parsed_content)
yield {
"identifier": parsed_content["identifier"],
"published": parsed_content["published_parsed"],
"description": parsed_content["description"],
"description_hash": parsed_content["description_hash"],
"title": parsed_content["title"],
"link": parsed_content["link"],
"tokenized_count": parsed_content["tokenized_count"],
"tokenized_plain": parsed_content["tokenized_plain"],
}