Collect directories in one place

collect

#!/usr/bin/env python3

import shutil
import subprocess
import sys
import time
from pathlib import Path


save_dir = Path.home() / "Downloads" / "to_upload"

def get_all_git_repositories():
    cmd = ["find", "/", "-name", ".git"]
    res = subprocess.run(cmd, capture_output=True)
    repos = res.stdout.decode("utf-8").split("\n")
    return [Path(p).parent for p in repos if p]


def sanitize_path(path):
    return str(path).replace("/", "..").strip(".")


def get_size(path):
    cmd = ["du", "-ms", str(path)]  # unit: MB
    res = subprocess.run(cmd, capture_output=True)
    size, _ = res.stdout.decode("utf-8").split(maxsplit=1)
    return int(size) * 1e6


def ask(msg, target):
    print(msg, ":", target)
    while True:
        ans = input("Continue? [y/n] : ")
        if ans.lower().startswith("y"):
            break
        if ans.lower().startswith("n"):
            print(f"Skipping {target}")
            return False
        print("Please choose either of [y/n]")
    return True

def write(path):
    if not save_dir.exists():
        save_dir.mkdir()
    out_path = save_dir / sanitize_path(path)
    if path.is_file():
        if out_path.exists():
            print("Output already exists, skipping")
            return
        shutil.copyfile(path, out_path)
        return
    if (out_path.parent / (out_path.name + ".tar.gz")).exists():
        print("Output already exists, skipping")
        return
    size = get_size(path)
    if size > 100e6:
        if not ask(f"Large package (size={size / 1e6}MB)", path):
            return
    shutil.make_archive(out_path, "gztar", path)

def get_dotfiles():
    _candidates = [".bashrc", ".bash_profile", ".zshrc", ".zprofile", ".profile"]
    ret = []
    for filename in _candidates:
        path = Path.home() / filename
        if path.exists():
            ret.append(path)
    return ret


def get_all_python_file():
    out_dir = save_dir / ".." / "_pyfiles"
    if not out_dir.exists():
        out_dir.mkdir()
    cmd = ["find", "/", "-name", '"*.py"']
    res = subprocess.run(cmd, capture_output=True)
    repos = res.stdout.decode("utf-8").split("\n")
    files = [Path(p).parent for p in repos if p]
    _ = [shutil.copyfile(p, out_dir / sanitize_path(p)) for p in files]
    return out_dir


if __name__ == "__main__":
    import argparse
    p = argparse.ArgumentParser()
    p.add_argument("--git", action="store_true")
    p.add_argument("--here", action="store_true")
    p.add_argument("--add", type=str, nargs="*")
    p.add_argument("--dots", action="store_true")
    p.add_argument("--py", action="store_true")

    args = p.parse_args()

    if sum(map(lambda x: bool(x), [args.git, args.here, args.add, args.dots, args.py])) > 1:
        sys.exit(print("Conflicting options specified"))

    paths = []
    if args.git:
        paths = get_all_git_repositories()
    elif args.add:
        paths = [Path(p) for p in args.add]
    elif args.dots:
        paths = get_dotfiles()
    elif args.py:
        paths = [get_all_python_file()]
    elif args.here:
        paths = [Path(".")]

    paths = list(map(lambda p: p.resolve(), paths))

    for path in paths.copy():
        try:
            save_dir.relative_to(path)
            print("Output path recursively included, skipping : ", path)
            paths.remove(path)
        except ValueError:
            pass

    if paths:
        if not ask("About to process", paths):
            sys.exit("Quit")

    for path in paths:
        print("Processing", path)
        write(path)
    print(f"All {len(paths)} files completed")