Add filter-out feature to rpmqa query page (adamg?)

[projects/pld-ftp-admin.git] / wwwbin / clean-dups-old.py

#!/usr/bin/python
# arekm, 2008
# remove 

from __future__ import print_function

import os
import re
import time
import rpm
import sys

re_info = re.compile(r'.*\.info$')
re_nvr = re.compile('^(.*)-([^-]*)-([^-]*)\.info$')
dir = '/home/pld/admins/th/ftp/test/SRPMS/.metadata'

def compare(f1, f2):
	m1 = re_nvr.match(f1)
	n1 = m1.group(1)
	v1 = m1.group(2)
	r1 = m1.group(3)

	m2 = re_nvr.match(f2)
	n2 = m2.group(1)
	v2 = m2.group(2)
	r2 = m2.group(3)

	l1 = rpm.labelCompare((n1, v1, r1), (n2, v2, r2))
	l2 = rpm.labelCompare((n2, v2, r2), (n1, v1, r1))

	if l1 > 0 and l2 > 0:
		return 0

	return -l1


def find_old(files):
	return sorted(files, compare)

files = {}
dupes = {}

for file in os.listdir(dir):
	if not re_info.match(file):
		continue

	m = re_nvr.match(file)
	if not m:
		print("problem with: %s" % file)
		sys.exit(1)

	if len(sys.argv) == 0:
		p = os.path.join(dir, file)
		mtime = os.stat(p).st_mtime
		if mtime > time.time() - 3*86400:
			continue

	name = m.group(1)

	if files.has_key(name):
		if dupes.has_key(name):
			dupes[name].append(file)
		else:
			dupes[name] = [ files[name] ]
			dupes[name].append(file)
	else:
		files[name] = file

for i in iter(dupes.keys()):
	for old in find_old(dupes[i])[1:]:
		os.system("/home/pld/admins/th/pld-ftp-admin/scripts/remove.py test %s" % old)