validate-metadata

#!/usr/bin/env python

import argparse
import gzip
from BeautifulSoup import BeautifulStoneSoup
import urllib2
import uuid

"""
Looks for duplicate package listings in yum meta-data and prints those
out. They might have different checksums.

./validate-metadata http://some.site.com/repos/repo/repodata/primary.xml.gz
"""

parser = argparse.ArgumentParser(description='Validate remote meta-data')
parser.add_argument(dest='url', action='store', metavar='URL',\
                        help='Link to remote meta-data')

args = parser.parse_args()

response = urllib2.urlopen(args.url)
tmp = "/tmp/%s" % uuid.uuid4()
out_file = open(tmp, 'wb')
data = response.read()
out_file.write(data)
out_file.close()

in_file = gzip.open(tmp, 'rb')
file_content = in_file.read()
in_file.close()

soup = BeautifulStoneSoup(file_content)
packages = soup.findAll('package')
a = {}
for package in packages:
    checksum = package.version.checksum.string
    filename =  package.version.location['href']
    if filename not in a.keys():
        a[filename] = [checksum]
    else:
        a[filename].append(checksum)

for key in a.keys():
    if len(a[key]) > 1:
        print key, [str(x) for x in a[key]]