scripts/build_release.py

#!/usr/bin/env python3
from __future__ import annotations

import argparse
import os
import shutil
from pathlib import Path

from gitignore_parser import handle_negation, rule_from_pattern


def _load_gitignore_rules(gitignore_path: Path) -> list:
    """
    Parse a single .gitignore into gitignore_parser IgnoreRule objects.

    We intentionally use gitignore_parser's internal rule representation so we can
    combine multiple .gitignore files (root + subfolders) while keeping correct
    precedence (later rules override earlier ones).
    """
    base_dir = gitignore_path.parent

    try:
        lines = gitignore_path.read_text(encoding="utf-8").splitlines()
    except OSError:
        return []

    rules = []
    for raw in lines:
        line = raw.strip()
        if not line or line.startswith("#"):
            continue
        rules.append(rule_from_pattern(line, base_path=str(base_dir)))
    return rules


def _is_ignored(abs_path: Path, rules: list) -> bool:
    # gitignore_parser expects paths as strings; absolute paths are safest here.
    return bool(handle_negation(str(abs_path), rules))


def _copy_repo_respecting_gitignores(repo_root: Path, out_dir: Path) -> None:
    repo_root = repo_root.resolve()
    out_dir = out_dir.resolve()

    # If out_dir is inside repo_root, compute its rel path so we can prune traversal into it.
    out_rel: Path | None = None
    try:
        out_rel = out_dir.relative_to(repo_root)
    except ValueError:
        out_rel = None

    # Map directory -> cumulative rules that apply inside that directory.
    rules_for_dir: dict[Path, list] = {}

    for current_dir, dirnames, filenames in os.walk(repo_root, topdown=True):
        current_path = Path(current_dir)

        try:
            rel_dir = current_path.resolve().relative_to(repo_root)
        except ValueError:
            continue

        # Hard excludes
        if rel_dir == Path(".git") or Path(".git") in rel_dir.parents:
            dirnames[:] = []
            continue
        if out_rel is not None and (rel_dir == out_rel or out_rel in rel_dir.parents):
            dirnames[:] = []
            continue

        # Build cumulative rules for this directory: parent rules + local .gitignore rules
        parent_rel = rel_dir.parent if rel_dir != Path(".") else None
        parent_rules = rules_for_dir.get(parent_rel, []) if parent_rel is not None else []

        local_rules = []
        local_gitignore = current_path / ".gitignore"
        if local_gitignore.is_file():
            local_rules = _load_gitignore_rules(local_gitignore)

        current_rules = [*parent_rules, *local_rules]
        rules_for_dir[rel_dir] = current_rules

        # Filter directories in-place to control traversal
        kept_dirs: list[str] = []
        for d in dirnames:
            rel = rel_dir / d

            if rel == Path(".git") or Path(".git") in rel.parents:
                continue
            if out_rel is not None and (rel == out_rel or out_rel in rel.parents):
                continue

            abs_path = (repo_root / rel).resolve()
            if _is_ignored(abs_path, current_rules):
                continue

            kept_dirs.append(d)
        dirnames[:] = kept_dirs

        # Ensure destination dir exists
        (out_dir / rel_dir).mkdir(parents=True, exist_ok=True)

        # Copy files
        for f in filenames:
            rel_file = rel_dir / f
            abs_file = (repo_root / rel_file).resolve()

            if _is_ignored(abs_file, current_rules):
                continue

            src = repo_root / rel_file
            dst = out_dir / rel_file
            dst.parent.mkdir(parents=True, exist_ok=True)
            shutil.copy2(src, dst)


def main() -> None:
    parser = argparse.ArgumentParser(
        description="Build a release directory by copying all files except those ignored by .gitignore files."
    )
    parser.add_argument("--out-dir", default="out", help="Output directory.")
    parser.add_argument("--clean", action="store_true", help="Delete output directory before building.")
    args = parser.parse_args()

    repo_root = Path.cwd()
    out_dir = (repo_root / args.out_dir).resolve()

    if args.clean and out_dir.exists():
        shutil.rmtree(out_dir)
    out_dir.mkdir(parents=True, exist_ok=True)

    _copy_repo_respecting_gitignores(repo_root=repo_root, out_dir=out_dir)
    print(f"Release directory built at: {out_dir}")


if __name__ == "__main__":
    main()