Media Cloud Metadata Extractor

This is a package to extract a domain, title, publication date, text, and language content from the URL or text of an online news story. The methods for each are extracted from the larger Media Cloud project, but also build on numerous 3rd party libraries. The metadata extracted includes:

the original URL of publication
a normalized URL useful for de-duplication
the canonical domain published on
the date of publication
the primary language used in the article text
the title of the article
a normalized title useful for de-duplication
the text content of the news article
the name of the library used to extract the article content

Other often-reused methods and configuration related to the mediacloud service also live in this package.

Installation

pip install mediacloud-metadata

Usage

If you pass in a URL, it will follow redirects and fetch the HTML for you.

from mcmetadata import extract
metadata = extract(url="https://my.awesome.news/story-path")

You can also pass in HTML you already have on hand. Note that in this case it is also useful to pass in the URL because that is used for some for some of the metadata extraction.

from mcmetadata import extract
metadata = extract(url="https://my.awesome.news/story-path",
                   html_text="<html><head><title>my webpage ... </html>")

Development

If you are interested in adding code to this module, first clone the GitHub repository.

Installing

flit install
pre-commit install

Testing

pytest

Distributing a New Version

Run pytest to make sure all the test pass
Update the version number in pyproject.toml
Make a brief note in the CHANGELOG.md about what changes
Commit the changes
Tag the commit with a semantic version number - v*.*.*
Push to repo to GitHub

Test Cache

Test are run against fixtures by default. This can be changed with the use of '--use-cache=False' when running tests. When adding new tests, re-run 'scripts/get-test-web-content.py'

Contributors

Created as part of the Media Cloud Project. Contributes include:

Rahul Bhargava (Media Cloud, Northeastern University)
Paige Gulley (Media Cloud)
Phil Budne (Media Cloud)
Vangelis Banos (Internet Archive)

mediacloud / metadata-lib

readme