DEV Community

Waylon Walker
Waylon Walker

Posted on • Originally published at on

re-create Google reader - 🐍 Parsing RSS feeds with Python

I am looking into a way to replace my google reader experience that I had back in 2013 before google took it from us. I am starting by learning how to parse feeds with python, and without much previous knowledge, it proved to be much easier than anticipated thanks to the feedparser library.


Install the feedparser library.

conda create -n reader python=3.8 -y
source activate reader
pip install feedparser
Enter fullscreen mode Exit fullscreen mode

Get the content

import feedparser
feed = feedparser.parse('')
Enter fullscreen mode Exit fullscreen mode

The feed object

The feed is a feedparser.FeedParserDict. For all intents and purposes this seems to just behave like a dict with the following keys().

['feed', 'entries', 'bozo', 'headers', 'etag', 'href', 'status', 'encoding', 'version', 'namespaces', 'content'])
Enter fullscreen mode Exit fullscreen mode

feed has some general information about the rss feed, but the meat of the feed is in entries. The rest of the keys weren't all that useful for me at the moment.

pulling multiple feeds

I grabbed a few popular RSS feeds that I was familiar with to get started.

urls = ['',
feeds = [feedparser.parse(url)['entries'] for url in urls]
Enter fullscreen mode Exit fullscreen mode

I checked out the keys, all three had the following keys. Mine also had the full post under 'content', this is because I added an extra custom_element for publishing to from an RSS feed.

>>> dict_keys(['title', 'title_detail', 'summary', 'summary_detail', 'links', 'link', 'id', 'guidislink', 'published'
, 'published_parsed'])
Enter fullscreen mode Exit fullscreen mode


I also pulled the Since is it setup for more Authors it had a few extra keys.

>>> dict_keys(['title', 'title_detail', 'authors', 'author', 'author_detail', 'published', 'published_parsed', 'links
', 'link', 'id', 'guidislink', 'summary', 'summary_detail', 'tags'])
Enter fullscreen mode Exit fullscreen mode

Combining Feeds

Now that I have a list of feeds, I can create a single feed sorted by date with a list comprehension. Note I did need to pull in dateutil.parser to convert the date strings to datetime objects to be sorted.

import dateutil.parser

feed = [item for feed in feeds for item in feed]
feed.sort(key=lambda x: dateutil.parser.parse(x['published']), reverse=True)
Enter fullscreen mode Exit fullscreen mode
[ins] In [115]: [{'title': i['title'], 'date': i['published'], 'link': i['link']}  for i in feed[:10]]
[{'title': '🙋\u200d♂️ Can Anyone Explain Twitter Cards to me?',
  'date': 'Sat, 11 Jul 2020 03:00:00 GMT',
  'link': ''},
 {'title': 'How I Built My GitHub Profile',
  'date': 'Fri, 10 Jul 2020 03:00:00 GMT',
  'link': ''},
 {'title': 'Lessons and Regrets from My $25000 Launch',
  'date': 'Fri, 03 Jul 2020 04:06:47 GMT',
  'link': ''},
 {'title': 'SLIDES - understanding python *args and **kwargs',
  'date': 'Thu, 02 Jul 2020 05:00:00 GMT',
  'link': ''},
 {'title': 'Launching the Coding Career Handbook!',
  'date': 'Wed, 01 Jul 2020 13:08:37 GMT',
  'link': ''},
 {'title': 'Gracefully adopt kedro, the catalog',
  'date': 'Mon, 29 Jun 2020 03:00:00 GMT',
  'link': ''},
 {'title': "🤓 What's on your GitHub Profile",
  'date': 'Mon, 29 Jun 2020 03:00:00 GMT',
  'link': ''},
 {'title': "Versioned Docs in 30 Seconds with Amplify Console's Branch Subdomains",
  'date': 'Fri, 26 Jun 2020 16:34:09 GMT',
  'link': ''},
 {'title': "What's New in React",
  'date': 'Wed, 24 Jun 2020 00:00:00 GMT',
  'link': ''},
 {'title': 'Coding Careers - Vincit',
  'date': 'Wed, 24 Jun 2020 00:00:00 GMT',
  'link': ''}]
Enter fullscreen mode Exit fullscreen mode

Decentralized Feed

I think the idea of RSS is super cool, and the idea that I can potentially create my own custom platform-agnostic decentralized feed is pretty cool. I would love to have a google reader like experience back.

This post was super fun to explore. I used an external library (feedparser) to pull in the feeds, but other than that It was all vanilla python 3.8. In DataScience we tend to get very DataFrame heavy and I miss working with vanilla datatypes sometimes.

👀 see an issue, edit this post on GitHub

I have been writing short snippets about my mentality breaking into the tech/data industry in my newsletter, 👇 check it out, and let's get the conversation started.

Sign up for my Newsletter

Top comments (0)