Switch to building stage repository by combining production and diff …

…of Copr

When the version is nightly, this will not pull from production but instead
treat what is in Copr as the source of truth. At the end, the output for
a versioned repository is a list of unsigned packages.

build_stage_repository

@@ -11,6 +11,7 @@ import hashlib
 import sys
 import tempfile
 import dnf.comps
+import argparse
 
 
 def move_rpms_from_copr_to_stage(collection, version, src_folder, dest_folder, dist, arch, source=False):
@@ -40,6 +41,18 @@ def move_rpms_from_copr_to_stage(collection, version, src_folder, dest_folder, d
     shutil.rmtree(src_folder)
 
 
+def filter_packages(target_dir, packages):
+    files = glob.glob(f"{target_dir}/*.rpm")
+
+    for file in files:
+        file_name = os.path.basename(file)
+        rpm_name = get_rpm_name(file)
+
+        if rpm_name not in packages:
+            print(f"Removed {rpm_name}. Package not in comps.")
+            os.remove(os.path.join(target_dir, file_name))
+
+
 def get_rpm_name(rpm):
     command = [
         'rpmquery',
@@ -62,10 +75,12 @@ def modulemd_yaml(collection, version):
 def comps(collection, version, dist):
     branch = 'develop' if version == 'nightly' else version
 
-    if collection == 'foreman-katello':
-        name = 'katello'
-    elif collection == 'foreman-candlepin':
+    if collection == 'candlepin':
         name = 'katello-candlepin'
+    elif collection == 'client':
+        name = 'foreman-client'
+    elif collection == 'plugins':
+        name = 'foreman-plugins'
     else:
         name = collection
 
@@ -121,28 +136,30 @@ def create_modulemd(collection, version, stage_dir):
 
 
 def create_repository(repo_dir):
-    check_output(['createrepo', repo_dir])
+    check_output(['createrepo_c', repo_dir], stderr=STDOUT)
 
 
-def sync_copr_repository(collection, version, target_dir, dist, arch, source=False):
-    if source:
-        print(f"Syncing {collection} {version} Source RPM repository from Copr")
+def sync_prod_repository(collection, version, target_dir, dist, arch):
+    if arch == 'source':
+        print(f"Syncing {collection} {version} Source RPM repository from yum.theforeman.org")
     else:
-        print(f"Syncing {collection} {version} RPM repository from Copr")
+        print(f"Syncing {collection} {version} RPM repository from yum.theforeman.org")
 
     cmd = [
         'dnf',
         'reposync',
-        '--newest-only',
+        '--refresh',
+        '--download-metadata',
+        '--norepopath',
         '--repo',
         f"{dist}-{collection}-{version}-{arch}",
         '--repofrompath',
-        f"{dist}-{collection}-{version}-{arch},https://download.copr.fedorainfracloud.org/results/@theforeman/{collection}-{version}-staging/rhel-{dist.replace('el', '')}-{arch}/",
+        f"{dist}-{collection}-{version}-{arch},{prod_repository(collection, version, dist, arch)}",
         '--download-path',
         target_dir
     ]
 
-    if source:
+    if arch == 'source':
         cmd.extend([
             '--source',
         ])
@@ -172,47 +189,202 @@ def packages_from_comps(comps):
     return packages
 
 
+def copr_repository(collection, version, dist, arch):
+    if collection == 'candlepin':
+        name = 'katello-candlepin'
+    elif collection == 'client':
+        name = 'foreman-client'
+    elif collection == 'plugins':
+        name = 'foreman-plugins'
+    else:
+        name = collection
+
+    return f"https://download.copr.fedorainfracloud.org/results/@theforeman/{name}-{version}-staging/rhel-{dist.replace('el', '')}-{arch}"
+
+
+def prod_repository(collection, version, dist, arch):
+    if collection == "foreman" and version == "nightly":
+        return f"https://yum.theforeman.org/{version}/{dist}/{arch}"
+    else:
+        return f"https://yum.theforeman.org/{collection.replace('foreman-', '')}/{version}/{dist}/{arch}"
+
+
+def copr_repository_urls(repository):
+    cmd = [
+        'dnf',
+        'reposync',
+        '--urls',
+        '--refresh',
+        '--repofrompath',
+        f"copr,{repository}",
+        '--repo',
+        'copr'
+    ]
+
+    urls = check_output(cmd, universal_newlines=True, stderr=STDOUT)
+    return urls.split("\n")
+
+
+def compare_repositories(new_repository, old_repository, source=False):
+    cmd = [
+        'dnf',
+        'repodiff',
+        '--simple',
+        '--refresh',
+        '--compare-arch',
+        '--repofrompath',
+        f"new,{new_repository}",
+        '--repofrompath',
+        f"old,{old_repository}",
+        '--repo-old',
+        'old',
+        '--repo-new',
+        'new'
+    ]
+
+    if not source:
+        cmd.extend(['--arch', 'noarch,x86_64'])
+
+    print(' '.join(cmd))
+    return check_output(cmd, universal_newlines=True, stderr=STDOUT)
+
+
+def parse_repodiff(diff):
+    split_diff = diff.split("\n")
+    packages = []
+
+    for line in split_diff:
+        if line.startswith('Added package  :'):
+            packages.append(line.replace('Added package  : ', ''))
+
+        if ' -> ' in line:
+            packages.append(line.split(' -> ')[1])
+
+    return packages
+
+
+def download_copr_packages(packages, urls, repository, downloads_dir, included_packages):
+    for package in packages:
+        name = f"{package}.rpm".replace("-1:", "-")
+
+        for url in urls:
+            if name in url:
+                download_url = url
+
+        if not os.path.exists(f"{downloads_dir}/{name}"):
+            print(f"Downloading {repository}/{name} to {downloads_dir}")
+            urlretrieve(download_url, f"{downloads_dir}/{name}")
+        else:
+            print(f"Skipping {repository}/{name}. Already downloaded.")
+
+
+def find_unsigned_packages(target_dir, gpgkey):
+    unsigned = []
+    packages = glob.glob(f"{target_dir}/*.rpm")
+
+    for package in packages:
+        cmd = [
+            'rpm',
+            '-qpi',
+            package
+        ]
+
+        output = check_output(cmd, universal_newlines=True, stderr=STDOUT)
+
+        if gpgkey.lower() not in output:
+            unsigned.append(package)
+
+    return unsigned
+
+
+def handle_args():
+    parser = argparse.ArgumentParser(description='Generate a stage repository')
+    parser.add_argument(
+        'collection',
+        help='Repository to generate for (e.g. foreman)'
+    )
+    parser.add_argument(
+        'version',
+        help='Version to generate the repository for (e.g. nightly)'
+    )
+    parser.add_argument(
+        'dist',
+        help='Dist to generate repository for (e.g. el8)'
+    )
+    parser.add_argument(
+        '--gpgkey',
+        help='Short form gpgkey for the version being generated, does not apply to nightly'
+    )
+
+    args = parser.parse_args()
+
+    if not args.gpgkey and args.version != 'nightly':
+        raise SystemExit("When the version is not nightly, --gpgkey must be supplied")
+
+    if args.gpgkey and args.version == 'nightly':
+        raise SystemExit("--gpgkey cannot be used when the version is nightly")
+
+    if args.gpgkey:
+        if len(args.gpgkey) != 8:
+            raise SystemExit("GPG key must be in short form and 8 characters long")
+
+    return args
+
+
 def main():
-    try:
-        collection = sys.argv[1]
-        version = sys.argv[2]
-        dist = sys.argv[3]
-        arch = 'x86_64'
-    except IndexError:
-        raise SystemExit(f"Usage: {sys.argv[0]} collection version os")
+    args = handle_args()
 
-    base_dir = 'tmp'
-    rpm_sync_dir = f"{base_dir}/rpms"
-    srpm_sync_dir = f"{base_dir}/srpms"
+    collection = args.collection
+    version = args.version
+    dist = args.dist
+    arch = 'x86_64'
+    gpgkey = args.gpgkey
 
-    stage_dir = f"{base_dir}/{collection}/{version}/{dist}/"
+    base_dir = 'tmp'
+    stage_dir = f"{base_dir}/{collection}/{version}/{dist}"
     rpm_dir = f"{stage_dir}/{arch}"
     srpm_dir = f"{stage_dir}/source"
 
-    if not os.path.exists(rpm_sync_dir):
-        os.makedirs(rpm_sync_dir)
-
-    if not os.path.exists(srpm_sync_dir):
-        os.makedirs(srpm_sync_dir)
-
     if not os.path.exists(rpm_dir):
         os.makedirs(rpm_dir)
 
     if not os.path.exists(srpm_dir):
         os.makedirs(srpm_dir)
 
-    sync_copr_repository(collection, version, rpm_sync_dir, dist, arch)
-    sync_copr_repository(collection, version, srpm_sync_dir, dist, arch, source=True)
+    if version != 'nightly':
+        sync_prod_repository(collection, version, rpm_dir, dist, arch)
+        sync_prod_repository(collection, version, srpm_dir, dist, 'source')
+
+    packages_to_include = packages_from_comps(comps(collection, version, dist))
+    copr_repo = copr_repository(collection, version, dist, arch)
+    copr_package_urls = copr_repository_urls(copr_repo)
+
+    rpm_diff = compare_repositories(copr_repo, rpm_dir)
+    packages = parse_repodiff(rpm_diff)
+    download_copr_packages(packages, copr_package_urls, copr_repo, rpm_dir, packages_to_include)
+    filter_packages(rpm_dir, packages_to_include)
 
-    move_rpms_from_copr_to_stage(collection, version, srpm_sync_dir, srpm_dir, dist, arch, source=True)
-    move_rpms_from_copr_to_stage(collection, version, rpm_sync_dir, rpm_dir, dist, arch)
+    srpm_diff = compare_repositories(copr_repo, srpm_dir, True)
+    srpm_packages = parse_repodiff(srpm_diff)
+    download_copr_packages(srpm_packages, copr_package_urls, copr_repo, srpm_dir, packages_to_include)
+    filter_packages(srpm_dir, packages_to_include)
 
     create_repository(rpm_dir)
     create_repository(srpm_dir)
 
     if collection in ['foreman', 'foreman-katello'] and dist == 'el8':
         create_modulemd(collection, version, rpm_dir)
 
+    if gpgkey:
+        unsigned_rpms = find_unsigned_packages(rpm_dir, gpgkey)
+        unsigned_srpms = find_unsigned_packages(srpm_dir, gpgkey)
+
+        print("\n# Unsigned RPMs")
+        for rpm in unsigned_rpms:
+            print(f"Unsigned RPM - {rpm}")
+        for srpm in unsigned_srpms:
+            print(f"Unsigned RPM - {rpm}")
+
 
 if __name__ == '__main__':
     main()