Create a .zip archive of a .NET solution directory with build artifacts omitted

ziplab.py

"""Create a .zip archive of a .NET solution directory with build artifacts omitted.

If the solution directory is tracked in a Git repository that has a .gitignore to
exclude .NET build artifacts, consider `git archive <branch> -o <filename>` instead.

"""

import argparse
import fnmatch
import hashlib
import sys
from pathlib import Path, PurePosixPath
from typing import IO, Iterator
from zipfile import ZIP_DEFLATED, ZipFile

IGNORE_GLOBAL_PATTERNS = (".*", "*.doc", "*.docx", "*.docx#", "*.pdf")
IGNORE_PROJECT_PATTERNS = ("bin", "obj")


def main() -> None:
    parser = argparse.ArgumentParser(
        description=__doc__,
        formatter_class=argparse.RawDescriptionHelpFormatter,
    )
    collision_group = parser.add_mutually_exclusive_group()
    collision_group.add_argument(
        "-c",
        "--check",
        action="store_true",
        help="If dest exists, compare contents with current files",
    )
    collision_group.add_argument(
        "-f",
        "--force",
        action="store_true",
        help="Write archive even if dest exists",
    )
    parser.add_argument("solution", help="The solution directory to archive", type=Path)
    parser.add_argument(
        "dest",
        help="Where to write the archive (defaults to name of solution directory)",
        nargs="?",
        type=Path,
    )

    args = parser.parse_args()
    check: bool = args.check
    force: bool = args.force
    solution: Path = args.solution
    dest: Path | None = args.dest

    check_solution_path(solution)

    if dest is None:
        dest = solution.with_suffix(".zip")

    dest_exists = dest.is_file()
    if dest_exists and not check and not force:
        sys.exit(
            f"Destination {dest} already exists\n"
            f"(use -f/--force to overwrite, or -c/--check to compare contents)"
        )
    elif dest_exists and check:
        compare_solution_archive(solution, dest)
    elif dest_exists and force:
        # NOTE: consider renaming dest before writing
        create_solution_archive(solution, dest)
        print("Overwritten", dest)
    elif check:
        sys.exit(
            f"Destination {dest} does not exist (omit -c/--check to write archive)"
        )
    else:
        create_solution_archive(solution, dest)
        print("Saved to", dest)


def check_solution_path(solution: Path) -> None:
    if not solution.is_dir():
        sys.exit(f"Solution must be a directory, not {dir}")

    sln = (".sln", ".slnx")
    sln = [solution / solution.with_suffix(s) for s in sln]
    if not any(p.is_file() for p in sln):
        sys.exit(f"Directory is missing solution file, {' or '.join(p.name for p in sln)}")

    for child in solution.iterdir():
        if child.is_dir() and any(child.glob("*.csproj")):
            break
    else:
        sys.exit("No projects found inside solution directory (.csproj)")


def create_solution_archive(src: Path, dest: Path | IO[bytes]) -> None:
    with ZipFile(dest, "w", compression=ZIP_DEFLATED) as dest_zip:
        for path in filter_solution_files(src):
            print(path)
            dest_zip.write(path)


def filter_solution_files(solution: Path) -> Iterator[Path]:
    for path in solution.iterdir():
        if fnmatch_any(path.name, *IGNORE_GLOBAL_PATTERNS):
            continue
        elif path.is_dir():
            yield from filter_project_files(path)
        else:
            yield path


def filter_project_files(project: Path) -> Iterator[Path]:
    for path in project.iterdir():
        if fnmatch_any(path.name, *IGNORE_PROJECT_PATTERNS, *IGNORE_GLOBAL_PATTERNS):
            continue
        else:
            yield path


def fnmatch_any(name: str, *patterns: str) -> bool:
    return any(fnmatch.fnmatch(name, p) for p in patterns)


def hexdigest(data: bytes | IO[bytes]) -> str:
    m = hashlib.sha3_256()
    if isinstance(data, (bytes, bytearray, memoryview)):
        m.update(data)
    else:
        while chunk := data.read(2**20):
            m.update(chunk)

    return m.hexdigest()


def compare_solution_archive(src: Path, dest: Path) -> None:
    lines: list[str] = []
    with ZipFile(dest) as dest_zip:
        deleted = {Path(info.filename) for info in dest_zip.filelist}

        for path in filter_solution_files(src):
            deleted.discard(path)

            try:
                with dest_zip.open(str(PurePosixPath(path))) as f:
                    old = hexdigest(f)
            except KeyError:
                lines.append(f"A {path}")
                continue

            with path.open("rb") as f:
                new = hexdigest(f)

            if old != new:
                lines.append(f"M {path}")

    for path in deleted:
        lines.append(f"D {path}")

    if not lines:
        lines.append("No changes in content")

    print("\n".join(lines))


if __name__ == "__main__":
    main()

> py ziplab.py -h
usage: ziplab.py [-h] [-c | -f] solution [dest]

Create a .zip archive of a .NET solution directory with build artifacts omitted.

If the solution directory is tracked in a Git repository that has a .gitignore to
exclude .NET build artifacts, consider `git archive <branch> -o <filename>` instead.

positional arguments:
  solution     The solution directory to archive
  dest         Where to write the archive (defaults to name of solution directory)

options:
  -h, --help   show this help message and exit
  -c, --check  If dest exists, compare contents with current files
  -f, --force  Write archive even if dest exists

> py ziplab.py Lab1
Lab1\Lab1.slnx
Lab1\Medals.csv
Lab1\Question1\Program.cs
Lab1\Question1\Question1.csproj
Lab1\Question2\Program.cs
Lab1\Question2\Question2.csproj
Lab1\Question2\SinglyLinkedList.cs
Saved to Lab1.zip

> py ziplab.py Lab1
Destination Lab1.zip already exists
(use -f/--force to overwrite, or -c/--check to compare contents)

> py ziplab.py Lab1 --check
No changes in content

> rm Lab1\Medals.csv
> py ziplab.py Lab1 --check
D Lab1\Medals.csv

> py ziplab.py Lab1 --force
Lab1\Lab1.slnx
Lab1\Question1\Program.cs
Lab1\Question1\Question1.csproj
Lab1\Question2\Program.cs
Lab1\Question2\Question2.csproj
Lab1\Question2\SinglyLinkedList.cs
Overwritten Lab1.zip