#!/usr/bin/python3

# clean up our git repository used by lilac

import os
import pathlib
import subprocess
from collections import defaultdict
import shutil
import sys
import time

# repodir = pathlib.Path('/ldata/src/archgitrepo/archlinuxcn').expanduser()
repodir = pathlib.Path('/usr/share/lilac/Packages/BioArchLinux').expanduser()

DRY_RUN = True

def get_git_managed():
  cmd = ['git', 'ls-files']
  out = subprocess.check_output(cmd, text = True)

  pkgbase_to_files = defaultdict(list)
  for line in out.splitlines():
    if line == '.gitignore':
      continue

    try:
      pkgbase, file = line.split('/', 1)
    except ValueError:
      # skip subproject commits
      continue
    pkgbase_to_files[pkgbase].append(file)

  return pkgbase_to_files

def rmdir(d):
  if DRY_RUN:
    print('Would remove dir ', d)
  else:
    print('Removing dir ', d)
    shutil.rmtree(d, ignore_errors=True)

def rmfile(f):
  if DRY_RUN:
    print('Would remove file', f)
  else:
    print('Removing file', f)
    f.unlink()

def is_vcs(dir):
  files = [f.name for f in dir.iterdir()]
  return any(x in files for x in ['.git', '.hg', '.svn', 'packed-refs'])

def process(dir, git_files):
  files = list(dir.iterdir())
  mtimes = {f: f.stat().st_mtime for f in files}

  # don't delete files touched near last update
  try:
    protected_mtime = max(x for x in (
      y for f, y in mtimes.items() if f.name in git_files)
    ) - 86400
  except ValueError: # max() arg is an empty sequence
    protected_mtime = time.time()

  for file in files:
    if file.name == '__pycache__':
      continue

    if file.name.endswith('.log'):
      # logs are being handled by find -delete using crontab
      # and keeped longer than source code
      continue

    if file.name.endswith((
      '.pkg.tar.zst', '.pkg.tar.zst.sig',
      '.pkg.tar.xz', '.pkg.tar.xz.sig',
      )):
      continue

    is_dir = file.is_dir()

    if is_dir and is_vcs(file):
      continue

    if file.name in git_files:
      continue

    if mtimes[file] > protected_mtime:
      continue

    if is_dir:
      rmdir(file)
    else:
      rmfile(file)

def main():
  global DRY_RUN

  if '-f' in sys.argv:
    DRY_RUN = False

  os.chdir(repodir)
  pkgbase_to_files = get_git_managed()

  for dir in repodir.iterdir():
    if dir.name == '.gitignore':
      continue

    if not dir.is_dir():
      rmfile(dir)
      continue

    if not (dir / 'lilac.yaml').exists() and dir.name not in pkgbase_to_files:
      rmdir(dir)
      continue

    process(dir, pkgbase_to_files[dir.name])

if __name__ == '__main__':
  main()