#!/usr/bin/env python
"""
open/dulcinea/bin/gc_stored_files.py

Delete files that are not referenced by StoredFile instances.
"""
import sys, os, glob, time
from durus.storage import gen_durus_id_class
from qp.lib.site import Site

# new files are not deleted since someone could be in the process of
# uploading them or attaching them in another transaction
MIN_AGE = 60*60*24

size = 0

def maybe_remove(filename):
    global size
    if (time.time() - os.stat(filename).st_mtime) > MIN_AGE:
        print(filename)
        size += os.stat(filename).st_size
        os.unlink(filename)


def main (prog, args):
    usage = "usage: %s [site]" % prog

    if len(args) != 1:
        sys.exit(usage)

    site = Site(args[0])
    connection = site.make_file_connection(readonly=True)
    used = {}
    for durus_id, class_name in gen_durus_id_class(connection.get_storage(), 'StoredFile'):
        s = connection.get(durus_id)
        used[s.get_full_path()] = 1
    now = time.time()
    root_directory = site.get('file_store')
    print('root_directory %s' % root_directory)
    for filename in glob.glob(os.path.join(root_directory, '?/??/*')):
        if filename not in used:
            maybe_remove(filename)
    for filename in glob.glob(os.path.join(root_directory, 'tmp/upload.*')):
        maybe_remove(filename)
    print("%s MB" % (size / (1024*1024)))

if __name__ == '__main__':
    main(sys.argv[0], sys.argv[1:])