demo.py

#!/usr/bin/env python3
"""
I attempt to showcase and fix a problem in `urllib.parse.urlunsplit` by fixing it and comparing it to the original.
See `DEMO`, below for results.

TL;DR: It's currently impossible to obtain relative URIs (like `file:selfie.png`) from urlunsplit, but we could fix it.
"""

from urllib.parse import urlsplit, urlunsplit

SplitResult = tuple[str, str, str, str, str]


def urlunsplit_new(components: SplitResult, strict: bool = False):
    scheme, netloc, path, query, fragment = components
    if scheme:
        url = [scheme, ":"]
    else:
        url = []

    if netloc or path.startswith("//"):
        url.extend(("//", netloc))
        if path and not path.startswith("/"):
            if strict:
                raise ValueError("URI can't represent netloc with relative path")
            else:
                url.append("/")  # status quo: path quietly coerced to absolute

    url.append(path)

    if query:
        url.extend(("?", query))
    if fragment:
        url.extend(("#", fragment))
    return "".join(url)


def test(scheme: str, path: str, slash_count: int):
    url = scheme + "/" * slash_count + path
    split = urlsplit(url)
    unsplit_old = urlunsplit(split)
    unsplit_new = urlunsplit_new(split)

    resplit_old = urlsplit(unsplit_old)
    resplit_new = urlsplit(unsplit_new)

    if resplit_old == resplit_new:
        return
    elif resplit_new == split:
        old = "wrong"
        new = "fixed"
    elif resplit_old == split:
        old = "okay!"
        new = "derp!"
    else:
        old = "WRONG!"
        new = "BORKD!"

    print(repr(url), "-split->", split, "-unsplit->", "...")
    print(f"(old, {old})", repr(unsplit_old), "-split->", resplit_old)
    print(f"(new, {new})", repr(unsplit_new), "-split->", resplit_new)
    print()


def main():
    for scheme in ("", "file:"):
        for path in ("", "x"):
            for slash_count in range(5):
                test(scheme, path, slash_count)


DEMO = """
$ python3 demo.py
'////' -split-> SplitResult(scheme='', netloc='', path='//', query='', fragment='') -unsplit-> ...
(old, wrong) '//' -split-> SplitResult(scheme='', netloc='', path='', query='', fragment='')
(new, fixed) '////' -split-> SplitResult(scheme='', netloc='', path='//', query='', fragment='')

'////x' -split-> SplitResult(scheme='', netloc='', path='//x', query='', fragment='') -unsplit-> ...
(old, wrong) '//x' -split-> SplitResult(scheme='', netloc='x', path='', query='', fragment='')
(new, fixed) '////x' -split-> SplitResult(scheme='', netloc='', path='//x', query='', fragment='')

'file:////' -split-> SplitResult(scheme='file', netloc='', path='//', query='', fragment='') -unsplit-> ...
(old, wrong) 'file://' -split-> SplitResult(scheme='file', netloc='', path='', query='', fragment='')
(new, fixed) 'file:////' -split-> SplitResult(scheme='file', netloc='', path='//', query='', fragment='')

'file:x' -split-> SplitResult(scheme='file', netloc='', path='x', query='', fragment='') -unsplit-> ...
(old, wrong) 'file:///x' -split-> SplitResult(scheme='file', netloc='', path='/x', query='', fragment='')
(new, fixed) 'file:x' -split-> SplitResult(scheme='file', netloc='', path='x', query='', fragment='')

'file:////x' -split-> SplitResult(scheme='file', netloc='', path='//x', query='', fragment='') -unsplit-> ...
(old, wrong) 'file://x' -split-> SplitResult(scheme='file', netloc='x', path='', query='', fragment='')
(new, fixed) 'file:////x' -split-> SplitResult(scheme='file', netloc='', path='//x', query='', fragment='')
"""

if __name__ == "__main__":
    main()