mirror-coreos-fedora-coreos-config/build-rootfs

#!/usr/bin/python3

# This script is called by the Containerfile to build FCOS. Here's what it does at a high-level:
# 1. It gathers the list of FCOS-specific packages using the manifests.
# 2. It gathers the list of FCOS-specific overlays using the manifests.
# 3. It runs `bootc-base-imagectl make-rootfs`.
# 4. It injects various metadata (e.g. image.json, live/ bits, platforms.json, and disk.yaml).
# 5. It runs the postprocess scripts defined in the manifest.

import argparse
import glob
import hashlib
import json
import os
import shutil
import subprocess
import sys
import tempfile
import yaml


ARCH = os.uname().machine
INPUTHASH = '/run/inputhash'
HERMETIC_REPO = '/etc/yum.repos.d/hermeto.repo'

# Build-args to propagate as env vars into the postprocess bwrap sandbox.
# Add entries here as needed; they must also be declared as ARG in the
# Containerfile builder stage so they are present in os.environ.
POSTPROCESS_ENV_VARS = ['NAME']
IS_HERMETIC = os.path.exists(HERMETIC_REPO)


def main():
    parser = argparse.ArgumentParser(description='Build a CoreOS rootfs.')
    parser.add_argument('--srcdir', default='/src',
                        help='The source config directory')
    subparsers = parser.add_subparsers(help='Subcommands', required=True)

    cmd_make_rootfs = \
        subparsers.add_parser('make-rootfs',
                              help='Generate a container root filesystem')
    cmd_make_rootfs.add_argument('--target-rootfs', required=True,
                                 help='Path to the target rootfs.')
    cmd_make_rootfs.set_defaults(func=build_rootfs)

    cmd_parse_treefile = \
        subparsers.add_parser('parse-treefile',
                              help='Print flattened treefile to stdout')
    cmd_parse_treefile.set_defaults(func=print_treefile)
    args = parser.parse_args()

    # Convert srcdir arg into absolute path
    args.srcdir = os.path.abspath(args.srcdir)

    variables = {
        'target_rootfs': getattr(args, 'target_rootfs', ''),
        'srcdir': args.srcdir,
        'manifest_name': os.getenv('MANIFEST'),
        'image_config': os.getenv('IMAGE_CONFIG'),
        'version': os.getenv('VERSION'),
        'stream': os.getenv('STREAM'),
        'osid': os.getenv('ID'),
        'mutate_os_release': os.getenv('MUTATE_OS_RELEASE'),
        'strict_mode': os.getenv('STRICT_MODE'),
        'passwd_group_dir': os.getenv('PASSWD_GROUP_DIR')
    }

    args.func(**variables)


def print_treefile(manifest_name, osid, stream, version, srcdir, **kwargs):
    if not manifest_name or not osid or not stream or not version or not srcdir:
        raise Exception("Must set env vars before calling. Source build-args.conf")
    manifest_path = os.path.join(srcdir, manifest_name)
    print(json.dumps(get_treefile(manifest_path, osid, stream, version)))


def get_lockfile_dir(srcdir, stream):
    """
    Returns the directory containing lockfiles.
    First looks in lockfiles/${stream}/, then falls back to srcdir root.
    """
    stream_dir = os.path.join(srcdir, 'lockfiles', stream)
    if os.path.isdir(stream_dir):
        return stream_dir
    return srcdir


def get_treefile(manifest_path, osid, stream, version):
    with tempfile.NamedTemporaryFile(suffix='.json', mode='w') as tmp_manifest:
        # Substitute in a few values from build-args into the treefile.
        ## Split the version to get components for releasever and osversion
        x, y = (version.split('.') + ['0'])[:2]
        osversion = f"{osid}-{x}"
        if osid == "rhel":
            # For RHCOS we add the minor to the osversion
            osversion = f"{osid}-{x}.{y}"
        json.dump({
            "variables": {
                "deriving": True,
                "id": osid,
                "stream": stream,
                "osversion": osversion
            },
            "releasever": int(x),  # Only needed/used by Fedora
            "include": manifest_path
        }, tmp_manifest)
        tmp_manifest.flush()
        data = subprocess.check_output(['rpm-ostree', 'compose', 'tree',
                                        '--print-only', tmp_manifest.name])
    return json.loads(data)


def inject_yumrepos(srcdir):
    # first delete all the default repos
    for repo in glob.glob('/etc/yum.repos.d/*.repo'):
        if repo == HERMETIC_REPO:
            # this is the repo Konflux injects when hermetic build is enabled
            continue
        os.unlink(repo)

    # and now inject our repos
    if not IS_HERMETIC:
        secret_repo = '/run/secrets/yumrepos'
        if os.path.exists(secret_repo):
            shutil.copy(secret_repo, '/etc/yum.repos.d/secret.repo')
        for repo in glob.glob(f'{srcdir}/*.repo'):
            shutil.copy(repo, "/etc/yum.repos.d")


def run_buildroot_prep(srcdir):
    # This allows FCOS/SCOS/RHCOS to do specific things before the main
    # build process. Runs after inject_yumrepos() so repos are available.
    buildroot_prep = os.path.join(srcdir, 'buildroot-prep')
    if os.path.isfile(buildroot_prep):
        subprocess.check_call([buildroot_prep])

def build_rootfs(target_rootfs, srcdir, manifest_name,
                 image_config, osid, stream, version,
                 strict_mode, passwd_group_dir, mutate_os_release):

    # we allow strict_mode and passwd_group_dir to be None. Check all others.
    if not target_rootfs or not manifest_name or not image_config or \
        not osid or not stream or not version or not mutate_os_release:
        raise Exception("Must set env vars before calling. Source build-args.conf")

    manifest_path = os.path.join(srcdir, manifest_name)
    image_cfg_path = os.path.join(srcdir, image_config)

    manifest = get_treefile(
        manifest_path=manifest_path,
        osid=osid,
        stream=stream,
        version=version
    )
    packages = list(manifest['packages'])

    repos = manifest.get('repos', [])
    lockfile_repos = manifest.get('lockfile-repos', [])
    if repos or lockfile_repos:
        inject_yumrepos(srcdir)

    # prepare buildroot (if anything needs to be done)
    run_buildroot_prep(srcdir)

    local_overrides = prepare_local_rpm_overrides(target_rootfs, srcdir)
    if local_overrides:
        repos += ['overrides']

    locked_nevras = get_locked_nevras(local_overrides, srcdir, stream)
    if locked_nevras:
        # Lockfile repos require special handling because we only want locked
        # NEVRAs to appear there. For lack of a generic solution for any repo
        # there, we only special-case the one place where we know we use this.
        if lockfile_repos == ['fedora-coreos-pool']:
            if not IS_HERMETIC:
                modify_pool_repo(locked_nevras)
            repos += lockfile_repos
        elif len(lockfile_repos) > 0:
            raise Exception(f"unknown lockfile-repo found in {lockfile_repos}")

    add_dirs = gather_add_dirs(manifest, srcdir)
    nodocs = (manifest.get('documentation') is False)
    recommends = manifest.get('recommends')
    # We generate the initramfs using dracut ourselves later after our
    # CoreOS postprocess scripts have run. If this version of rpm-ostree
    # supports it we'll tell it to not run dracut in the initial compose.
    no_initramfs = True if no_initramfs_arg_supported() else False

    if passwd_group_dir is not None:
        inject_passwd_group(os.path.join(srcdir, passwd_group_dir))

    with tempfile.NamedTemporaryFile(mode='w') as argsfile:
        for pkg in packages:
            argsfile.write(f"--install={pkg}\n")
        for add_dir in add_dirs:
            argsfile.write(f"--add-dir={add_dir}\n")
        if nodocs:
            argsfile.write("--no-docs\n")
            # temporarily work around https://issues.redhat.com/browse/RHEL-97826
            tmpd = workaround_rhel_97826(argsfile)
        if recommends:
            if not recommends_arg_supported():
                raise Exception(f"Need to set recommends: true but --recommends is unsupported")
            argsfile.write("--recommends\n")
        if no_initramfs:
            argsfile.write("--no-initramfs\n")
        if repos and repo_arg_supported():
            for repo in repos:
                argsfile.write(f"--repo={repo}\n")
        if locked_nevras and lock_arg_supported():
            for locked_nevra in locked_nevras:
                argsfile.write(f"--lock={locked_nevra}\n")
        argsfile.flush()
        # print out argsfile content for debugging
        with open(argsfile.name, "r") as f:
            print(f"Using following argsfile for bootc-base-imagectl:\n{f.read()}", flush=True)
        cache_arg = []
        if os.path.isdir('/cache') and rpm_ostree_has_cachedir_fix():
            cache_arg = ['--cachedir=/cache']
        subprocess.check_call(["/usr/libexec/bootc-base-imagectl",
                               "--args-file", argsfile.name, "build-rootfs",
                               "--manifest", 'minimal-plus',
                               target_rootfs] + cache_arg)
    if nodocs and tmpd is not None:
        del tmpd

    inject_live(target_rootfs, srcdir)
    inject_image_json(target_rootfs, image_cfg_path, stream)
    inject_platforms_json(target_rootfs, srcdir)
    inject_disk_yaml(target_rootfs, srcdir)
    inject_content_manifest(target_rootfs, manifest)

    inject_version_info(
        rootfs=target_rootfs,
        replace_version=mutate_os_release,
        version=version
    )

    if strict_mode == '1':
        verify_strict_mode(target_rootfs, locked_nevras)
    run_postprocess_scripts(target_rootfs, manifest)
    run_dracut(target_rootfs)
    cleanup_extraneous_files(target_rootfs)

    calculate_inputhash(target_rootfs, add_dirs, manifest)

def get_bootc_base_imagectl_help():
    return subprocess.check_output(['/usr/libexec/bootc-base-imagectl', 'build-rootfs', '-h'], encoding='utf-8')


def repo_arg_supported():
    # Detect if we have https://gitlab.com/fedora/bootc/base-images/-/merge_requests/248.
    # If not, then we can't use `--repo`. That's OK because that should only
    # happen on RHEL, where we don't have any default repos anyway and only rely on
    # the mounted secret repo file.
    return '--repo REPO' in get_bootc_base_imagectl_help()


def lock_arg_supported():
    # Detect if we have https://gitlab.com/fedora/bootc/base-images/-/merge_requests/279.
    # If not, then we can't use `--lock`. That should only happen in RHCOS,
    # where we only use this for autolocking and not base lockfile management.
    return '--lock NEVRA' in get_bootc_base_imagectl_help()


def recommends_arg_supported():
    # Detect if we have https://gitlab.com/fedora/bootc/base-images/-/merge_requests/314.
    # If not, then we can't use `--recommends` and should error.
    return '--recommends' in get_bootc_base_imagectl_help()


def no_initramfs_arg_supported():
    # Detect if we have # https://gitlab.com/fedora/bootc/base-images/-/merge_requests/320.
    # If not, then we can't use `--no-initramfs`, but that's OK because it's just
    # an optimization to prevent building the initramfs twice.
    if not '--no-initramfs' in get_bootc_base_imagectl_help():
        return False
    # Detect if we have https://github.com/coreos/rpm-ostree/commit/481fbb034292666578780bacfdbf3dae9d10e6c3
    # At the time of this writing it's unreleased in rpm-ostree but it
    # should be in the next release (2025.13 or 2026.1).
    out = subprocess.check_output(['rpm-ostree', '--version'], encoding='utf-8')
    data = yaml.safe_load(out)
    version_str = data['rpm-ostree']['Version']
    # ideally, we could use `packaging.version`, but that's not in centos-bootc
    # but conveniently, Python list comparisons do the right thing here
    version = [int(c) for c in version_str.split('.')]
    return version >= [2025, 13]


def workaround_rhel_97826(argsfile):
    basedir = 'usr/share/doc/bootc/baseimage/base'
    # Detect if we have https://github.com/bootc-dev/bootc/pull/1352.
    # This is not strictly correct because we're checking the _builder
    # image_ which could be different from the target. In practice, we don't
    # "cross-compose" and we're always going to be fetching equal or newer
    # NEVRAs from the repos so this is fine.
    if os.path.exists(f'/{basedir}/sysroot/.keepdir'):
        return
    print("Working around RHEL-97826", flush=True)
    tmpd = tempfile.TemporaryDirectory()
    shutil.copytree(f'/{basedir}', os.path.join(tmpd.name, basedir), symlinks=True)
    argsfile.write(f'--add-dir={tmpd.name}\n')
    return tmpd


def rpm_ostree_has_cachedir_fix():
    # we can only use --cachedir if we have rpm-ostree >= 2025.9 which has
    # https://github.com/coreos/rpm-ostree/pull/5391
    out = subprocess.check_output(['rpm-ostree', '--version'], encoding='utf-8')
    data = yaml.safe_load(out)
    version_str = data['rpm-ostree']['Version']
    # ideally, we could use `packaging.version`, but that's not in centos-bootc
    # but conveniently, Python list comparisons do the right thing here
    version = [int(c) for c in version_str.split('.')]
    return version >= [2025, 9]


# We want to keep our passwd/group as canonical for now. We should be
# able to clean this up when we migrate them to sysusers instead. See:
# https://github.com/coreos/rpm-ostree/pull/5427
def inject_passwd_group(parent_dir):
    minimal = '/usr/share/doc/bootc-base-imagectl/manifests/minimal'
    dst_passwd = os.path.join(minimal, 'passwd')
    dst_group = os.path.join(minimal, 'group')
    # unlink first instead of overwriting as a way to confirm they're still there
    os.unlink(dst_passwd)
    os.unlink(dst_group)
    print("Overriding passwd/group files", flush=True)
    shutil.copy(os.path.join(parent_dir, 'passwd'), dst_passwd)
    shutil.copy(os.path.join(parent_dir, 'group'), dst_group)



def run_postprocess_scripts(rootfs, manifest):
    # Since we have the derive-only manifest handy, just run the scripts now. An
    # alternative is to run it as a second stage, which would avoid the bwrap,
    # but operating on the raw rootfs means we don't pay for deleted files (nor
    # without requiring another rechunk).
    # Propagate select build-args as env vars for postprocess scripts.
    env = {}
    for var in POSTPROCESS_ENV_VARS:
        val = os.getenv(var)
        if val:
            env[var] = val
    for i, script in enumerate(manifest.get('postprocess', [])):
        name = f'usr/libexec/coreos-postprocess-{i}'
        with open(os.path.join(rootfs, name), mode='w') as f:
            os.fchmod(f.fileno(), 0o755)
            f.write(script)
        print(f"\n<<< Running CoreOS postprocess script {i} >>>\n", flush=True)
        bwrap(rootfs, [f'/{name}'], env=env)
        os.unlink(os.path.join(rootfs, name))


def run_dracut(rootfs):
    print(f"Running dracut to generate the initramfs", flush=True)
    # https://docs.fedoraproject.org/en-US/bootc/initramfs/#_modifying_and_regenerating_the_initrd
    kver = bwrap(rootfs, ['ls', '/usr/lib/modules'], capture=True).strip()
    bwrap(rootfs, env={'DRACUT_NO_XATTR': '1'},
          args=['dracut', '--verbose', '--force',  '--reproducible',
                '--no-hostonly', f"/usr/lib/modules/{kver}/initramfs.img", kver])


def prepare_local_rpm_overrides(rootfs, srcdir):
    overrides_repo = os.path.join(srcdir, 'overrides/rpm')
    if not os.path.isdir(f'{overrides_repo}/repodata'):
        return None

    pkglist = subprocess.check_output(['dnf', 'repoquery', f'--repofrompath=overrides,file://{overrides_repo}',
                                       '--repo=overrides', '--latest-limit=1', f'--arch={ARCH},noarch',
                                       '--qf=pkg,%{name},%{evr},%{arch}\\n'], encoding='utf-8')
    lockfile = {}
    for line in pkglist.splitlines():
        if not line.startswith("pkg"):
            continue
        _, name, evr, arch = line.split(',')
        lockfile[name] = {"evra": f"{evr}.{arch}"}

    if len(lockfile) == 0:
        return None

    with open('/etc/yum.repos.d/overrides.repo', 'w') as f:
        f.write(f'''
[overrides]
name=overrides
baseurl=file://{overrides_repo}
gpgcheck=0
cost=500
priority=1
''')

    print("Injected", len(lockfile), 'package overrides')
    return lockfile


# Could upstream this as e.g. `bootc-base-imagectl runroot /rootfs <cmd>` maybe?
# But we'd need to carry it anyway at least for RHCOS 9.6.
def bwrap(rootfs, args, env=dict(), capture=False):
    env_args = []
    for k, v in env.items():
        env_args.extend(['--setenv', k, v])

    args = ['bwrap', '--bind', f'{rootfs}', '/', '--dev', '/dev',
            '--proc', '/proc', '--tmpfs', '/tmp', '--tmpfs', '/var',
            '--tmpfs', '/var/tmp', '--tmpfs', '/run',
            '--bind', '/run/.containerenv', '/run/.containerenv'] + env_args + ['--'] + args
    if capture:
        return subprocess.check_output(args, encoding='utf-8')
    subprocess.check_call(args)


def get_locked_nevras(local_overrides, srcdir, stream):
    lockfile_dir = get_lockfile_dir(srcdir, stream)
    lockfile_path = os.path.join(lockfile_dir, f"manifest-lock.{ARCH}.json")
    overrides_path = os.path.join(lockfile_dir, "manifest-lock.overrides.yaml")
    overrides_arch_path = os.path.join(lockfile_dir, f"manifest-lock.overrides.{ARCH}.yaml")

    # we go from lowest priority to highest here: base lockfiles, overrides, local overrides
    locks = {}
    for path in [lockfile_path, overrides_path, overrides_arch_path]:
        if os.path.exists(path):
            with open(path) as f:
                if path.endswith('.yaml'):
                    data = yaml.safe_load(f)
                else:
                    data = json.load(f)
                # this essentially re-implements the merge semantics of rpm-ostree
                locks.update({pkgname: v['evra'] if 'evra' in v else v['evr']
                              for (pkgname, v) in data['packages'].items()})
    if local_overrides:
        locks.update({pkgname: v['evra'] if 'evra' in v else v['evr']
                      for (pkgname, v) in local_overrides.items()})
    return [f'{k}-{v}' for (k, v) in locks.items()]


def modify_pool_repo(locked_nevras):
    # When adding the pool, we only want to _filter in_ locked packages;
    # matching `lockfile-repos` semantics. This is abusing pretty hard the
    # `includepkgs=` semantic but... it works.
    repo = os.path.join('/etc/yum.repos.d/fedora-coreos-pool.repo')
    packages = ','.join(locked_nevras)
    with open(repo, 'a') as f:
        f.write(f"\nincludepkgs={packages}\n")


# This re-implements rpm-ostree's mutate-os-release to preserve the historical
# /usr/lib/os-release API, but we may in the future completely sever off of that
# and only rely on org.opencontainers.image.version as argued in:
# https://gitlab.com/fedora/bootc/base-images/-/issues/40
# https://gitlab.com/fedora/bootc/base-images/-/issues/46
def inject_version_info(rootfs, replace_version, version):
    os_release_path = os.path.join(rootfs, 'usr/lib/os-release')
    with open(os_release_path) as f:
        from collections import OrderedDict
        os_release = OrderedDict()
        for line in f:
            line = line.strip()
            if len(line) == 0 or line.startswith('#'):
                continue
            (k, v) = line.split('=', 1)
            os_release[k] = v

    for key in ['VERSION', 'PRETTY_NAME']:
        os_release[key] = os_release[key].replace(replace_version, version)
    os_release['OSTREE_VERSION'] = f"'{version}'"
    os_release['IMAGE_VERSION'] = f"'{version}'"

    with open(os_release_path, mode='w', encoding='utf-8') as f:
        for (k, v) in os_release.items():
            f.write(f'{k}={v}\n')


def gather_add_dirs(manifest, srcdir):
    add_dirs = []
    for layer in manifest.get('ostree-layers', []):
        add_dirs.append(os.path.normpath(os.path.join(srcdir, layer)))

    # sort by the base name of the directory to allow for overriding
    # files in layers that sort earlier. i.e. a file in 03-core/etc/foo.d/file
    # would override a file in 02-core/etc/foo.d/file.
    add_dirs.sort(key=lambda x: os.path.basename(x))

    rootfs_override = os.path.join(srcdir, 'overrides/rootfs')
    if os.path.isdir(rootfs_override) and len(os.listdir(rootfs_override)) > 0:
        print("Injecting rootfs override")
        add_dirs.append(rootfs_override)

    return add_dirs


# Inject live/ bits.
def inject_live(rootfs, srcdir):
    target_path = os.path.join(rootfs, 'usr/share/coreos-assembler/live')
    shutil.copytree(os.path.join(srcdir, "live"), target_path)


def inject_image_json(rootfs, image_cfg_path, stream):
    image = flatten_image_yaml(image_cfg_path, format_args={'stream': stream})
    fn = os.path.join(rootfs, 'usr/share/coreos-assembler/image.json')
    with open(fn, 'w') as f:
        json.dump(image, f, sort_keys=True)


def inject_content_manifest(target_rootfs, manifest):
    content_manifest_path = '/run/secrets/container-image-build/contentsets'
    if not os.path.exists(content_manifest_path):
        return

    with open(content_manifest_path) as f:
        data = yaml.safe_load(f)

    repos = []
    for base_repo in manifest['repos']:
        if base_repo in data['repo_mapping']:
            if data['repo_mapping'][base_repo]['name'] != '':
                repo_name = data['repo_mapping'][base_repo]['name'].replace('$ARCH', ARCH)
                repos.append(repo_name)
            else:
                print('Warning: No corresponding repo in repository-to-cpe.json for ' + base_repo)
        else:
            # Warning message for repositories with no entry in content_sets.yaml
            print('Warning: No corresponding entry in content_sets.yaml for ' + base_repo)

    dest = os.path.join(target_rootfs, "usr/share/buildinfo/content_manifest.json")
    os.makedirs(os.path.dirname(dest), exist_ok=True)
    with open(dest, 'w') as f:
        json.dump(fp=f, obj={
            'metadata': {
                'icm_version': 1,
                'icm_spec': 'https://raw.githubusercontent.com/containerbuildsystem/atomic-reactor/master/atomic_reactor/schemas/content_manifest.json',
                'image_layer_index': 1
            },
            'content_sets': repos,
            'image_contents': []
        })


def verify_strict_mode(rootfs, locked_nevras):
    cmd = ['rpm', '-qa', '--qf', '%{EPOCH}\t%{NVRA}\t%{NVR}\t%{NEVRA}\t%{NEVR}\n']
    rpms = bwrap(rootfs, cmd, capture=True)
    for rpm in rpms.splitlines():
        epoch, nvra, nvr, nevra, nevr = rpm.split()
        if nevra in locked_nevras or nevr in locked_nevras:
            continue
        # Do one more check. If a package has an Epoch explicitly
        # set to 0 [1] (rather than just an undefined Epoch) then
        # rpm-ostree won't write that value into the lockfiles. We
        # need to check just the NVR or NVRA in that case.
        # [1] https://src.fedoraproject.org/rpms/perl/blob/a8ff590c732b326216ab1499780e5964e4b03ddf/f/perl.spec#_2048
        if epoch == '0':
            if nvra in locked_nevras or nvr in locked_nevras:
                continue
        raise Exception(f"found unlocked RPM in strict mode: {rpm}")
    print("Strict mode: all installed packages were locked")


def cleanup_extraneous_files(rootfs):
    def unlink_optional(f):
        try:
            os.unlink(os.path.join(rootfs, f))
        except FileNotFoundError:
            pass

    # for now just this
    unlink_optional('usr/share/rpm/.rpm.lock')


def calculate_inputhash(rootfs, overlays, manifest):
    h = hashlib.sha256()

    # rpms
    rpms = bwrap(rootfs, ['rpm', '-qa', '--qf', '%{NEVRA}\n'], capture=True)
    rpms = sorted(rpms.splitlines())
    h.update(''.join(rpms).encode('utf-8'))

    # overlays
    for overlay in overlays:
        all_files = []
        for root, _, files in os.walk(overlay):
            for file in files:
                all_files.append(os.path.join(root, file))
        all_files = sorted(all_files)
        for file in all_files:
            if os.path.islink(file):
                # For symlinks, hash the link target instead of following it
                link_target = os.readlink(file)
                h.update(link_target.encode('utf-8'))
            else:
                with open(file, 'rb') as f:
                    # When python3.11+ is the minimal version we can use hashlib.file_digest
                    # h.update(hashlib.file_digest(f, 'sha256').digest())
                    h.update(hashlib.sha256(f.read()).digest())
                    has_x_bit = os.stat(f.fileno()).st_mode & 0o111 != 0
                    h.update(bytes([has_x_bit]))

    # postprocess
    for script in manifest.get('postprocess', []):
        h.update(script.encode('utf-8'))

    with open(INPUTHASH, 'w', encoding='utf-8') as f:
        f.write(h.hexdigest())


# Imported from cosa
# Merge two lists, avoiding duplicates. Exact duplicate kargs could be valid
# but we have no use case for them right now in our official images.
def merge_lists(x, y, k):
    x[k] = x.get(k, [])
    assert isinstance(x[k], list)
    y[k] = y.get(k, [])
    assert isinstance(y[k], list)
    x[k].extend([i for i in y[k] if i not in x[k]])


# Imported from cosa
def flatten_image_yaml(srcfile, base=None, format_args={}):
    if base is None:
        base = {}

    with open(srcfile) as f:
        contents = f.read()
        srcyaml = yaml.safe_load(contents.format(**format_args))

    # first, special-case list values
    merge_lists(base, srcyaml, 'extra-kargs')

    # then handle all the non-list values
    base = merge_dicts(base, srcyaml)

    if 'include' not in srcyaml:
        return base

    fn = os.path.join(os.path.dirname(srcfile), srcyaml['include'])
    del base['include']
    return flatten_image_yaml(fn, base=base, format_args=format_args)


# Imported from cosa
# Credit to @arithx
def merge_dicts(x, y):
    """
    Merge two dicts recursively, but based on the difference.
    """
    sd = set(x.keys()).symmetric_difference(y.keys())
    ret = {}
    for d in [x, y]:
        for k, v in d.items():
            if k in sd:
                # the key is only present in one dict, add it directly
                ret.update({k: v})
            elif isinstance(x[k], dict) and isinstance(y[k], dict):
                # recursively merge
                ret.update({k: merge_dicts(x[k], y[k])})
            elif isinstance(x[k], list) and isinstance(y[k], list):
                ret.update({k: x[k]})
                merge_lists(ret, y, k)
            else:
                # first dictionary always takes precedence
                ret.update({k: x[k]})
    return ret


def inject_platforms_json(rootfs, srcdir):
    with open(os.path.join(srcdir, 'platforms.yaml')) as f:
        platforms = yaml.safe_load(f)
    fn = os.path.join(rootfs, 'usr/share/coreos-assembler/platforms.json')
    if ARCH in platforms:
        with open(fn, 'w') as f:
            json.dump(platforms[ARCH], f, sort_keys=True, indent=2)
            f.write('\n')


def inject_disk_yaml(rootfs, srcdir):
    dst = os.path.join(rootfs, 'usr/lib/image-builder/bootc')
    if os.path.exists(dst):
        # Relative symlink as both files are in the same directory
        print(f"Creating symlink to arch specific image-builder config: 'disk.yaml' -> '{ARCH}.yaml'")
        os.symlink(f'{ARCH}.yaml', os.path.join(dst, 'disk.yaml'))
    else:
        print("No image-builder config found. Skipping symlink to arch specific config.")


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