url_summary_regex.py

# Adapting https://gist.github.com/angelogladding/c61488d2dd82fb73b9078541a4e2ea13 for library use

import re

REST = r"(?P<rest>.*)"
DOMAIN_PATTERN = r"^http[s]?://(?P<domain>\w.[^/]+)/" + REST

url_summary_templates = [
    (
        r"http[s]?://github.com/(?P<user>\w+)/(?P<project>[\w-]+)/",
        "A comment in the {project} GitHub repository",
    ),
    (
        r"http[s]?://github.com/(?P<user>\w+)/(?P<project>[\w-]+)/issues/(?P<issue>\d+)",
        "A comment on an issue in the {project} GitHub repository",
    ),
    (
        r"http[s]?://github.com/(?P<user>\w+)/(?P<project>[\w-]+)/pulls/(?P<pull>\d+)",
        "A comment on a pull request in the {project} GitHub repository",
    ),
    (
        r"http[s]?://twitter.com/(?P<user>\w+)" + REST,
        "A tweet by @{user}",
    ),
    (
        r"^http[s]?://eventbrite.[com|co.uk]",
        "An event on Eventbrite",
    ),
    (
        r"^http[s]?://upcoming.com",
        "An event on Upcoming",
    ),
    (
        r"^http[s]?://calagator.com",
        "An event on Calagator",
    ),
    (
        r"^http[s]?://events.indieweb.org",
        "An IndieWeb event"
    ),
    (
        r"^http[s]?://indieweb.org/(?P<page>[\w-]+)",
        "The {page} page on the IndieWeb wiki",
    ),
]

def get_summary(url: str, custom_properties: list = []):
    """
    Return a text summary for given `url`.
    
    >>> get_summary("https://github.com/capjamesg/indieweb-utils/pulls/1")
    'A comment on a pull request in the indieweb-utils GitHub repository'
    
    """
    for pattern, summary in url_summary_templates + custom_properties:
        if match := re.match(pattern, url):
            return summary.format(**match.groupdict())

    domain = re.match(DOMAIN_PATTERN, url)

    if domain:
        return "A post on " + domain.groupdict()["domain"]

    return ""

print(get_summary("https://github.com/capjamesg/indieweb-utils/pulls/1"))
print(get_summary("https://twitter.com/indieweb"))
print(get_summary("https://eventbrite.co.uk/test"))
print(get_summary("http://upcoming.com/test"))
print(get_summary("https://indieweb.org/coffee"))