fedora-comps/check-missing

142 lines
5.0 KiB
Plaintext
Raw Normal View History

#!/bin/python3
# This script looks for the comps-fXX.xml.in file for the release version
# specified, grabs lists of all packages that exist in that release of
# Fedora, and looks through the comps `pkgreq` lines for ones that specify
# packages that do not currently exist . It is arch-aware. It expects to
# be run on a Fedora system with network access, as it will try to query
# the dnf repos to get lists of currently-existing packages. It relies on
# the non-Rawhide repo definitions actually working for Rawhide if passed
# the Rawhide-matching release number, which is a kinda undocumented
# feature but we rely on it in other places too. You will need
# fedora-repos and python3-lxml packages installed.
import glob
import argparse
import subprocess
import lxml.etree as ET
from collections import defaultdict
ARCHES = ('aarch64', 'ppc64le', 's390x', 'x86_64')
parser = argparse.ArgumentParser(description='Check Fedora comps files for missing packages and packages missing on architectures')
parser.add_argument('relver', help='Release version to check')
parser.add_argument('--update', dest='update', action='store_true', default=False,
help='Update the comps file with the changes')
args = parser.parse_args()
# gather package lists. this eats lots of RAM. I don't care.
# FIXME: for this script to work for EPEL and ELN someone would have
# to figure out the right tweaks to this command line.
pkgs = {}
for arch in ARCHES:
pkgtext = subprocess.run(('dnf', f'--forcearch={arch}', f'--releasever={args.relver}', '--disablerepo=*', '--enablerepo=fedora', '--enablerepo=updates', '--enablerepo=fedora-cisco-openh264', 'repoquery', '--qf=%{NAME}'), capture_output=True, text=True).stdout
pkgs[arch] = pkgtext.splitlines()
compsfile = f'comps-f{args.relver}.xml.in'
# find package reqs in comps
tree = ET.parse(compsfile) #, ET.XMLParser(target=CommentedTreeBuilder()))
root = tree.getroot()
pkgreqs = root.findall('.//packagereq')
# Check if each package is in the repository for each architecture
removedpkgs = defaultdict(list)
archpkgs = defaultdict(list)
for pkgreq in pkgreqs:
reqtype = pkgreq.get('type', '')
# list of arches the package is missing on
missing = []
present = []
# arches the package is listed for (if no 'arch' key, it's listed for all)
reqarches = pkgreq.get('arch', '')
if reqarches:
reqarches = reqarches.split(',')
else:
reqarches = ARCHES
# do the actual check, append arch to 'missing' if it's not there
for arch in reqarches:
if arch in pkgs and pkgreq.text not in pkgs[arch]:
missing.append(arch)
else:
present.append(arch)
grpid = pkgreq.find('./../../id').text
pkgname = pkgreq.text
# print the result
if missing == list(ARCHES):
if pkgreq.getparent() is not None:
removedpkgs[pkgname].append(grpid)
# pkgreq.getparent().remove(pkgreq)
elif missing and reqtype != 'optional':
archpkgs[pkgname] = ','.join(present)
# Find empty groups after packages not in repositories have been removed
pkglists = root.findall('.//packagelist')
removedgrps = {}
for pkglist in pkglists:
if not len(pkglist):
group = pkglist.getparent()
grpid = group.find('./id').text
removedgrps[grpid] = []
group.getparent().remove(group)
# Remove any empty groups from the environment lists
envlists = root.findall('.//environment//groupid')
for envgrp in envlists:
grpid = envgrp.text
if grpid in removedgrps:
# The groups are inside a grouplist inside the environment
par = envgrp.getparent()
envid = par.getparent().find('./id').text
removedgrps[grpid].append(envid)
par.remove(envgrp)
# Remove any empty groups from the category lists
catlists = root.findall('.//category//groupid')
for catgrp in catlists:
grpid = catgrp.text
if grpid in removedgrps:
# The groups are inside a grouplist inside the category
par = catgrp.getparent()
catid = par.getparent().find('./id').text
removedgrps[grpid].append(catid)
par.remove(catgrp)
# Remove any language packs for packages that don't exist anymore
langpacks = root.find('.//langpacks')
removedlang = []
for lang in langpacks.getchildren():
pkg = lang.get('name')
if pkg in list(removedpkgs):
removedlang.append(pkg)
lang.getparent().remove(lang)
# Print out a summary
print('Packages with incorrect architecture tags:')
for pkg in sorted(archpkgs):
print(' {} only available on {}'.format(pkg, archpkgs[pkg]))
print('\nRemoving packages:')
for pkg in sorted(removedpkgs):
print(' {} in group {}'.format(pkg, ', '.join(removedpkgs[pkg])))
print('\nRemoving empty groups:')
for group in sorted(removedgrps):
print(' {} in {}'.format(group, ', '.join(removedgrps[group])))
print('\nRemoving language packs for:')
for lang in removedlang:
print(' {}'.format(lang))
# Write out the updated XML file if desired
if args.update:
tree.write(latest, encoding="UTF-8", xml_declaration=True)