transfer-requirements-git-urls

#! /usr/bin/env python3

import sys
import pathlib
import re


vcs_url_re = re.compile(
        r"(git\+[a-z]+://[-_a-zA-Z0-9./]+)"
        "((?:@[-_a-zA-Z0-9]+)?)"
        r"\#egg=([-a-zA-Z0-9_.]+)"
        )


def main():
    import argparse

    parser = argparse.ArgumentParser()
    parser.add_argument("from_file", metavar="FROM_REQ.TXT")
    parser.add_argument("to_file", metavar="TO_REQ.TXT")
    args = parser.parse_args()

    pkg_to_url_branch = {}
    with pathlib.Path(args.from_file).open() as req_txt:
        for ln in req_txt:
            ln = ln.strip()
            if not ln:
                continue
            if ln.startswith("#"):
                continue

            match = vcs_url_re.search(ln)
            if not match:
                continue

            branch = match.group(2)
            if branch:
                assert branch.startswith("@")
                branch = branch[1:]

            pkg_to_url_branch[match.group(3)] = (
                    match.group(1),
                    branch,
                    )

    with pathlib.Path(args.to_file).open() as req_txt:
        result = []
        for ln in req_txt:
            lnread = ln.strip()
            if not lnread:
                result.append(ln)
            if lnread.startswith("#"):
                result.append(ln)

            match = vcs_url_re.search(lnread)
            if match and match.group(3) in pkg_to_url_branch:
                projname = match.group(3)
                base_url, branch = pkg_to_url_branch[projname]
                pkgspec = f"{base_url}"
                if branch:
                    pkgspec += f"@{branch}"
                pkgspec += f"#egg={projname}"

                result.append(vcs_url_re.sub(pkgspec, ln))
            else:
                result.append(ln)

    with pathlib.Path(args.to_file).open("w") as req_txt:
        req_txt.write("".join(result))


if __name__ == "__main__":
    main()