A debugging script for checking Uyuni repodata Packages files for invalid Filename paths (uyuni#7788)

uyuni_repodata_packages_filename_validator.py

#!/usr/bin/python3
import os

# This script is published under public domain.
# Use at your own risk! No warranties provided!
# Author: Santeri Pikarinen (github.com/santeri3700)

# Get list of channel labels (subdirectories) under the repodata directory
# These directories and Packages files are genereated by Taskomatic DebPackageWriter
repodata_channel_labels = os.listdir("/var/cache/rhn/repodata/")

# print(f"Checking Packages files under the following channel labels: {repodata_channel_labels}\n")

# Loop through each channel label and open the Packages file for reading only
all_invalid_packages = []
for repodata_channel_label in repodata_channel_labels:
    try:
        with open("/var/cache/rhn/repodata/" + repodata_channel_label + "/Packages", "r") as packages_file:
            invalid_packages = []
            for line in packages_file:
                if line.startswith("Filename: "):
                    # Example line: "Filename: mychannel-ubuntu2204_amd64/getPackage/mypackage_1.2.3-4.amd64-deb.deb"

                    # Strip the "Filename: " part without being length specific
                    filepath = line.replace("Filename: ", "").strip()

                    # Split the filepath per "/" into a list
                    filepath_parts = filepath.split("/")

                    # Get the channel label from the filepath
                    filepath_channel_label = filepath_parts[0]

                    # Get the package name from the filepath
                    filepath_package_name = filepath_parts[2]

                    # The package filename is invalid if the channel label is not the same as the repodata channel label
                    if repodata_channel_label != filepath_channel_label:
                        invalid_package = {
                            'real_channel': repodata_channel_label,
                            'wrong_channel': filepath_channel_label,
                            'package_name': filepath_package_name,
                            'filepath': filepath
                        }
                        invalid_packages.append(invalid_package)
                        all_invalid_packages.append(invalid_package)

            if invalid_packages:
                print(f"### {repodata_channel_label} ###")
                print("# Downloading these packages will fail if not subscribed to the channel they are pointing to.")
                for p in invalid_packages:
                    print(f"# - Package '{p['package_name']}' present in '{p['real_channel']}' but points to '{p['wrong_channel']}'")
                print("#"*15+"\n")
    except FileNotFoundError:
        # This channel is probably not a DEB channel if the "Release" file doesn't exist
        pass

if all_invalid_packages:
    print(f"## There are total of {len(all_invalid_packages)} packages which are pointing to a wrong channel in the repodata/Packages file.\n")
print("Done!")