#!/usr/bin/env python3
"""
Script to check for leftover files

Checks a collection of temporary or cache directories,
to ensure we aren't wasting image size by forgetting cleanup steps.

This script is run in every local repo image we test
"""

import os
import sys
from subprocess import check_output
from textwrap import indent

# directories larger than this are considered a failure
# a few little files here aren't a problem
THRESHOLD = 1  # in MB

MB = 1024 * 1024

# the paths to check
# all of these locations
# should be cleaned up
# missing is okay
PATHS = [
    "/tmp/",
    # check whole home?
    # this shouldn't be empty, but  for our tests (so far) it should be very small
    # This is the easiest way to ensure we aren't leaving any unexpected files
    # without knowing ahead of time where all possible caches might be (.npm, .cache, etc.)
    "~/",
    "/root/",
]


def du(path):
    """Return disk usage in megabytes of a path"""
    # -ks: get total size, reported in kilobytes
    out = check_output(["du", "-Hks", path])
    return int(out.split(None, 1)[0]) / 1024


def check_dir_size(path):
    """Check the size of a directory

    Returns:

    True: directory size is below THRESHOLD or is missing
    False: directory is larger than THRESHOLD
    """
    path = os.path.expanduser(path)

    if not os.path.exists(path):
        print("{path}: missing OK".format(**locals()))
        return True

    size_mb = du(path)
    print("{path}: {size_mb:.1f} MB".format(**locals()), end=" ")
    if size_mb <= THRESHOLD:
        print("OK")
        return True
    else:
        print("FAIL")
        # check size of files one-level deep (du only reports dirs)
        for name in os.listdir(path):
            subpath = os.path.join(path, name)
            if os.path.isfile(subpath):
                file_sz = os.stat(subpath).st_size / MB
                if file_sz > 0.1:
                    print("  {file_sz:.1f}M {subpath}".format(**locals()))
        # get report on all subdirs that are at least 100k
        print(
            indent(
                check_output(["du", "-Hh", "-t", "100000", path]).decode("utf8"), "  "
            )
        )
        return False


def main():
    results = [check_dir_size(path) for path in PATHS]
    if not all(results):
        sys.exit(1)


if __name__ == "__main__":
    main()