Advanced news feeds extractor and finder library. Helps to automatically extract news from websites without RSS/ATOM feeds
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


newsworker -- advanced automatic news extractor using HTML scraping

pypi version Documentation Status

newsworker is a Python 3 lib that extracts feeds from html pages. It's useful when you need to subscribe to a news from website that doesn't publish RSS/ATOM feeds and you don't want to use page change monitoring tools since it's not so accurate.

An idea behind this algorithm is simple. Most pages with news contain date and time information about certain news. These dates could look like "2017-09-27" or "1 jul 2016" or many other ways. First of all we needed to find all dates, second is to understand when date is just a date of this webpage and when date is on webpage area dedicated for news.

This tool helps to find news by locating news blocks on html webpage and parsing them for further usage.

Usage examples

Extract news from html page from EIB website and Bulgarian government website

>>> feed, session = f.get_feed(url="")
>>> feed
>>> from newsworker.extractor import FeedExtractor
>>> f = FeedExtractor(filtered_text_length=150)
>>> feed, session = f.get_feed(url="")
>>> feed
{'title': 'European Investment Bank (EIB)', 'language': 'en', 'link': '', 'description': 'European Investment Bank (EIB)', 'items': [{'title': 'Blockchain Challenge: coders at the EIB', 'description': 'Blockchain Challenge: coders at the EIB', 'pubdate': datetime.datetime(2018, 6, 18, 0, 0), 'unique_id': 'f9d359f76118076c5331ffec3cdb82eb', 'raw_html': b'<div class="first-column col-xs-12 col-sm-12 col-md-8 col-lg-8 no-padding-left-right"><div class="video-box no-padding-left-right"><a class="video-youtube" href=""><div class="img-item-1" style="background-image:url(\'/img/movies/blockchain-video-hp.png\');"><span class="video-icon"><img height="100" src="/img/site/play.png" width="100"/></span><div class="video-container"><div class="left-box col-lg-8 col-xs-12"><div class="video-date-time"><small>18/06/2018</small><span class="space-separator"> | </span><small>02:12</small></div><div class="video-title col-xs-12 col-lg-12 no-padding-left-right">Blockchain Challenge: coders at the EIB</div></div></div></div></a></div></div>', 'extra': {'links': [''], 'images': ['']}, 'link': ''}, {'title': 'A brighter life for Kenyan women', 'description': 'Jujuy Verde – new horizons for women waste-pickers in Argentina', 'pubdate': datetime.datetime(2018, 6, 5, 0, 0), 'unique_id': '9caef61535352d2734d122c0e092b011', 'raw_html': b'<div class="second-column col-xs-12 col-sm-12 col-md-4 col-lg-4 no-padding-left-right"><div class="video-box no-padding-left-right"><a class="video-youtube  fancybox.iframe" href=""><div class="img-item-2" style="background-image:url(\'/img/kenya-dlight-video-hp.png\');"><span class="video-icon"><img height="100" src="/img/site/play.png" width="100"/></span><div class="video-container"><div class="left-box col-lg-8 col-xs-12"><div class="video-date-time"><small>04/06/2018</small><span class="space-separator"> | </span><small>01:32</small></div><div class="video-title col-xs-12 col-lg-12 no-padding-left-right">A brighter life for Kenyan women</div></div></div></div></a></div><div class="video-box no-padding-left-right"><a class="video-youtube fancybox.iframe" href=""><div class="img-item-3" style="background-image:url(\'/img/jujuy-video-hp.png\');"><span class="video-icon"><img height="100" src="/img/site/play.png" width="100"/></span><div class="video-container"><div class="left-box col-lg-8 col-xs-12"><div class="video-date-time"><small>05/06/2018</small><span class="space-separator"> | </span><small>03:12</small></div><div class="video-title col-xs-12 col-lg-12 no-padding-left-right">Jujuy Verde \xc3\xa2\xe2\x82\xac\xe2\x80\x9c new horizons for women waste-pickers in Argentina</div></div></div></div></a></div></div>', 'extra': {'links': ['', ''], 'images': ['']}, 'link': ''}], 'cache': {'pats': ['dt:date:date_1']}}

Reuse cached patterns to speed up further news extraction. It could greatly improve page parsing speed since it minimizes number of date comparsion up to 100x times (2-3 date patterns instead of 350 patterns)

>>> pats = feeds['cache']['pats']
>>> feed, session = f.get_feed(url="", cached_p=pats)
Change user agent if needed
>>> USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.11 (KHTML, like Gecko) Chrome/23.0.1271.64 Safari/537.11'
>>> feed, session = f.get_feed(url="", user_agent=USER_AGENT)
Initialize feed finder on webpage
>>> from newsworker.finder import FeedsFinder
>>> f = FeedsFinder()
Try to extract feeds if no one feed exists
>>> feeds = f.find_feeds('')
{'url': '', 'items': []}
Add "extractrss" param launches FeedExtractor
>>> feeds = f.find_feeds('', extractrss=True)
>>> feeds
{'url': '', 'items': [{'feedtype': 'html', 'title': 'Министерски съвет :: Новини', 'num_entries': 12, 'url': ''}]}
Find all feeds and more info from feeds. With "noverify=False" each feed parsed
>>> feeds = f.find_feeds('', noverify=False)
>>> feeds
{'url': '', 'items': [{'title': 'Digital Transformation Agency', 'url': '', 'feedtype': 'rss', 'num_entries': 10}]}
Addind "include_entries=True" returns feeds and all parsed feed entries
>>> feeds = f.find_feeds('', noverify=False, include_entries=True)
>>> feeds


Documentation is built automatically and can be found on Read the Docs.


  • Identifies news blocks on webpages using date patterns. More than 348 date patterns supported. Uses qddate
  • Extremely fast, uses pyparsing to identify dates on webpages
  • Includes function to find feeds on html page and if no feed found, than extract news


  • Not all language-specific dates supported
  • Right aligned dates like "Published - 27-01-2018" not supported. It's not hard to add it but it greatly increases false acceptance rate.
  • Some news pages has no dates with urls or texts. These pages are not supported yet

Speed optimization

  • qddate date parsing lib was created for this algorithm. Right now pattern marching is really fast
  • date patterns could be cached to speed up parsing speed for the same website
  • feed finder without verification of feeds works fast, but if verification enabled than it's slowed down


  • Support more date formats and improve qddate lib
  • Support news pages without dates


The easiest way is to use the newsworker.FeedExtractor class, and it's get_feed function.

.. automodule:: newsworker.extractor
   :members: FeedExtractor
.. automodule:: newsworker.finder
   :members: FeedsFinder


newsworker relies on following libraries in some ways:

  • qddate_ is a module for data processing
  • pyparsing is a module for advanced text processing.
  • lxml is a module for xml parsing.

Supported languages specific dates

  • Bulgarian
  • Czech
  • English
  • French
  • German
  • Portuguese
  • Russian
  • Spanish


I wrote this news extraction code at 2008 year and later only updated it several times, migrating from regular expressions to pyparsing. Initial project was divided between qddate date parsing lib and newsworker intended to news identification on html pages

Feel free to ask question

Join the chat at