repo-analyzer/analyze-repo-redundancy.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

import os
import sys
import argparse
import sqlite3
import string
import rpm
import re
import gettext
prepare_repodb = __import__("prepare-repodb")

gettext.install('urpm-tools')

DB = 'repo.db'

def parseargs(args):
    parser = argparse.ArgumentParser(description=_('Check repository\'s redundancy for a given list of packages.'))
    parser.add_argument('packages_lst', metavar='packages.lst',
                        help=('path to packages.lst'))
    parser.add_argument('--repo', action='append', nargs='+', required=True,
        metavar='REPO', help=_('URL or name of a repository'))
    opts = parser.parse_args()
    return opts

def main(args):
    options = parseargs(args)
    with open(options.packages_lst, 'r') as f:
        ks_cfg = f.read()
        section = None
        in_packages = False
        sel_packages = []
        for line in ks_cfg.split('\n'):
            ls = line.strip()
            if ls and not ls.startswith('#'):
                sel_packages.append(ls)

    conn = sqlite3.connect(DB)
    c = conn.cursor()
    repodirs = []
    for repo in options.repo[0]:
        print repo
        rid = c.execute("""
SELECT id FROM repodirs WHERE name = ? OR url = ?
""", [repo, repo]).fetchall()
        if not rid:
            print 'Repository "%s" not found.' % repo
            exit(1)
        repodirs.append(rid[0][0])
    if not repodirs:
        print 'No repositories specified.'
        exit(1)
    if not sel_packages:
        print 'No packages specified.'
        exit(1)
    in_repodirs = ','.join(str(id) for id in repodirs)
    package_ids = []
    for spackage in sel_packages:
        packages_found = c.execute("""
SELECT id, nvra, version, release FROM packages WHERE repodir_id IN (%s) AND name = ?
""" % in_repodirs, [spackage]).fetchall()
        if not packages_found:
            print 'Package "%s" not found!' % spackage
            continue
        pid = None
        if len(packages_found) > 1:
            chosen_nvra = None
            chosen_version = None
            for vpid in packages_found:
                better_found = pid is None
                verrel = (vpid[2] if vpid[2] is not None else '') + '-' + \
                         (vpid[3] if vpid[3] is not None else '')
                if not better_found:
                    better_found = prepare_repodb.version_ok(chosen_version,
                                                             RPMSENSE_GREATER,
                                                             verrel)
                if better_found:
                    pid = vpid[0]
                    chosen_nvra = vpid[1]
                    chosen_version = verrel
            print ('Multiple versions of "%s" found (%s). Choosing latest: %s.' %
                   (spackage, ', '.join(cp[1] for cp in packages_found), chosen_nvra))
        else:
            pid = packages_found[0][0]
        package_ids.append(pid)

    package_cnt = 0
    while package_cnt < len(package_ids):
        package_cnt = len(package_ids)
        in_packages = ','.join(str(id) for id in package_ids)
        for required_package in c.execute("""
SELECT packages.id, packages.name, nvra
  FROM packages, package_requires_res
  WHERE packages.id = package_requires_res.dep_package_id AND
        package_requires_res.package_id IN (%s) AND
        packages.repodir_id IN (%s) AND
        packages.id NOT IN (%s)
  ORDER BY packages.name
""" % (in_repodirs, in_packages, in_packages)):
            package_ids.append(required_package[0])

    in_packages = ','.join(str(id) for id in package_ids)
    pre_repodir_id = -1
    repo_packages_cnt = 0
    for reduntant_package in c.execute("""
SELECT repodir_id, repodirs.name, packages.id, packages.nvra
  FROM packages, repodirs
  WHERE packages.repodir_id = repodirs.id AND
        repodirs.id IN (%s) AND packages.id NOT IN (%s)
  ORDER BY repodir_id, nvra
""" % (in_repodirs, in_packages)):
        (repodir_id, repodir_name, package_nvra) = \
        (reduntant_package[0], reduntant_package[1], reduntant_package[3])
        if pre_repodir_id != repodir_id:
            if pre_repodir_id == -1:
                print 'Redundant packages:'
            else:
                print 'Total: %d' % repo_packages_cnt
                print ''
            print '%d) %s' % (repodir_id, repodir_name)
            pre_repodir_id = repodir_id
            repo_packages_cnt = 0
        print '\t%s' % package_nvra
        repo_packages_cnt += 1
    if pre_repodir_id != -1:
        print 'Total: %d' % repo_packages_cnt
        print ''

    conn.rollback()

if __name__ == "__main__":
    main(sys.argv)