regex-for-bitbucket-url-shortening.py

import re

tests = [
    ("https://bitbucket.org", "bitbucket"),
    ("https://bitbucket.org/atlassian", "atlassian"),
    ("https://bitbucket.org/atlassian/workspace/overview", "atlassian"),
    ("https://bitbucket.org/atlassian/aui", "atlassian/aui"),
    ("https://bitbucket.org/atlassian/aui/", "atlassian/aui"),
    ("https://bitbucket.org/atlassian/aui/pull-requests/4758", "atlassian/aui#4758"),
    ("https://bitbucket.org/atlassian/aui/issues/375583", "atlassian/aui#375583"),
    ("https://bitbucket.org/atlassian/aui/issues", "atlassian/aui/issues"),
    ("https://bitbucket.org/atlassian/aui/issues/", "atlassian/aui/issues"),
    ("https://bitbucket.org/atlassian/aui/pipelines", "atlassian/aui/pipelines"),
    (
        "https://bitbucket.org/atlassian/aui/pipelines/results/14542",
        "atlassian/aui/pipelines/results/14542",
    ),
    (
        "https://bitbucket.org/atlassian/aui/commits/de41ded719e579d0ed4ffb8a81c29bb9ada10011",
        "atlassian/aui@de41ded",
    ),
    (
        "https://bitbucket.org/atlassian/aui/branch/future/10.0.x",
        "atlassian/aui@future/10.0.x",
    ),
    (
        "https://bitbucket.org/atlassian/atlas_fe/branch/Vadym_Kovalskiy/securityassistantyml-edited-online-with--1704961113691",
        "atlassian/atlas_fe@Vadym_Kovalskiy/securityassistantyml-edited-online-with--1704961113691",
    ),
    (
        "https://bitbucket.org/pybtex-devs/pybtex/issues/169/replace-pkg_resources-with",
        "pybtex-devs/pybtex#169",
    ),
    (
        "https://bitbucket.org/pybtex-devs/pybtex/branch/feature/fix-pytest-warning",
        "pybtex-devs/pybtex@feature/fix-pytest-warning",
    ),
]


def shorten_bitbucket_url(url):
    pattern = re.compile(
        r"(?P<scheme>https?:\/\/)?"
        r"(?:(?P<domain>bitbucket)\.org)"
        r"(?:\/(?P<user>[\w-]+))?"
        r"(?:\/workspace\/overview)?"
        r"(?:\/(?P<repo>[\w-]+))?"
        r"(?:\/issues\/(?P<issue>[0-9]+))?"
        r"(?:\/pull-requests\/(?P<pr>[0-9]+))?"
        r"(?:\/commits\/(?P<hash>[0-9a-f]+))?"
        r"(?:\/branch\/(?P<branch>[\/\.\w-]+))?"
        r"(?:\/(?P<leftovers>[\/\.\w-]+))?"
    )
    if res := pattern.match(url):
        gd = res.groupdict()
        if not gd["user"]:
            return gd["domain"]
        if not gd["repo"]:
            return gd["user"]
        result = "{user}/{repo}"
        if gd["hash"]:
            gd["shorthash"] = gd["hash"][:7]
            result += "@{shorthash}"
        elif gd["issue"]:
            result += "#{issue}"
        elif gd["pr"]:
            result += "#{pr}"
        elif gd["branch"]:
            result += "@{branch}"
        elif gd["leftovers"]:
            gd["left"] = gd["leftovers"].rstrip("/")
            result += "/{left}"
        return result.format_map(gd)


for url, expected in tests:
    result = shorten_bitbucket_url(url)
    assert result == expected, f"{result} != {expected}"