Script to insert missing `preInstall` and `postInstall` hook within `installPhase` of nixpkgs derivations

add_missing_hooks.py

import io
import re

import pathlib
import sys


HOOK_PATTERN = r'(?P<ident>[ ]+)?(?P<decl>installPhase\s+=\s+)"(?P<block>.*)";'
HOOK_PATTERN_MULTILINE = (
    r"(?P<decl>installPhase\s+=\s+(?:let(?:(?!\s+in\s+)(?:.|\n))+\s+in\s+)?)"
    r"'{2,3}\n?(?P<ident>\s+)?"
    r"(?P<block>(?:(?!'{2,3};)(?:\n|.)?)+)"
    r"(?P<quotes>'{2,3});"
)


def _replace_match_helper(pattern, func):
    def wrapped(content: str):
        patched = io.StringIO()
        last = 0
        hit = 0

        for matched in re.finditer(pattern, content):
            full = matched.group()

            if "preInstall" in full or "postInstall" in full:
                continue

            prev, post = matched.span()
            patched.write(content[last:prev])

            func(patched, matched)
            last = post
            hit += 1

        patched.write(content[last:])
        return patched.getvalue(), hit

    return wrapped


def replace_match(pattern):
    def wrapper(func):
        return _replace_match_helper(pattern, func)

    return wrapper



@replace_match(HOOK_PATTERN)
def add_runhook(out: io.StringIO, matched: re.Match) -> bool:
    ident, decl, block = matched.groups()

    ident_out = ident or ''
    ident = f"{ident_out}  "

    # block = f'\n{ident}'.join(line.strip() for line in block.split(';'))

    for part in (
        ident_out, decl,
        f'"runHook preInstall; {block}; runHook postInstall";'
    ):
        out.write(part)
    return True


@replace_match(HOOK_PATTERN_MULTILINE)
def add_runhook_multiline(out: io.StringIO, matched: re.Match):
    decl, ident, block, quotes =  matched.groups()

    ident = ident or '  '
    post_ident = ' ' * (len(ident) - 2)

    for part in (
        decl, f"{quotes}\n",
        ident, "runHook preInstall\n\n",
        ident, block.strip(), "\n\n",
        ident, "runHook postInstall\n",
        post_ident, f"{quotes};",
    ):
        out.write(part)
    return True


def main():
    hits = 0

    for filepath in (line.rstrip("\r\n") for line in sys.stdin):
        content = pathlib.Path(filepath).read_text()

        print("processing:", filepath)

        patched, hit = add_runhook(content)
        hits += hit

        patched, hit = add_runhook_multiline(patched)
        hits += hit

        with open(filepath, 'w+') as f:
           f.write(patched)

    print(hits)


if __name__ == "__main__":
    main()

command.sh

grep -rP "installPhase =" \
    | cut -d ':' -f 1 \
    | xargs -i sh -c "grep -P 'runHook\s+preInstall' {} >/dev/null || echo '{}'" \
    | python add_missing_hooks.py