# SPDX-License-Identifer: GPL-3.0-or-later

import base64
import bz2
import gzip
import itertools
import lzma
import os
import shutil
import subprocess
from abc import ABC, abstractmethod
from collections.abc import Generator, Iterable, Sequence
from contextlib import contextmanager
from dataclasses import dataclass
from enum import Enum, EnumMeta
from mmap import MADV_SEQUENTIAL, MAP_PRIVATE, mmap
from pathlib import Path
from tempfile import NamedTemporaryFile
from typing import IO, Any

from debian.deb822 import GPGV_DEFAULT_KEYRINGS, GPGV_EXECUTABLE, GpgInfo, Release

from .download import URL, DownloadFile, FileCompression, HashSum, HashType
from .filter import PackageFilter
from .logs import LoggerFactory

# https://github.com/pypy/pypy/issues/4991
# TODO: replace again with `from mmap import MAP_POPULATE`
MAP_POPULATE = 0x008000


def should_ignore_errors(ignored_paths: set[str], path: Path):
    return any(path.is_relative_to(ignored_path) for ignored_path in ignored_paths)


def is_safe_path(root_path: Path, path: Path):
    return (root_path / path).resolve().is_relative_to(root_path.resolve())


class InvalidReleaseFilesException(RuntimeError):
    pass


class InvalidSignatureError(RuntimeError):
    def __init__(self, message: str, error: str | None = None) -> None:
        if error:
            message += f"\nGPG error output:\n{error}"

        super().__init__(message)


@dataclass
class PackageFile:
    path: Path
    size: int
    hashes: dict[HashType, HashSum]

    def to_download_file(self, directory: Path):
        download_file = DownloadFile.from_path(
            path=directory / self.path, check_size=True
        )
        for hash_type, hashsum in self.hashes.items():
            download_file.add_compression_variant(
                directory / self.path,
                size=self.size,
                hash_type=hash_type,
                hash_sum=hashsum,
                use_by_hash=False,
            )

        return download_file


class IndexFileParser(ABC):
    def __init__(
        self,
        repository_path: Path,
        index_files: set[Path],
        ignore_errors: set[str],
        logger_id: Any | None = None,
    ) -> None:
        super().__init__()

        self._log = LoggerFactory.get_logger(self, logger_id=logger_id)
        self._repository_path = repository_path
        self._index_files = index_files
        self._ignore_errors = ignore_errors
        self._pool_files: dict[Path, DownloadFile] = {}

    def parse(self) -> set[DownloadFile]:
        for index_file_relative_path in self._index_files:
            index_file = self._repository_path / index_file_relative_path

            if not self._unpack_index(index_file):
                if "binary-all" not in str(index_file):
                    self._log.warning(
                        f"Unable to unpack index file {index_file}. Skipping"
                    )
                continue

            index_file_size = index_file.stat().st_size
            if index_file_size < 1:
                continue

            with open(index_file, "rb") as fp:
                mfp = fp
                if index_file_size > 1 * 1024 * 1024:
                    mfp = mmap(
                        fp.fileno(),
                        length=0,
                        flags=MAP_PRIVATE | MAP_POPULATE,
                    )
                    mfp.madvise(MADV_SEQUENTIAL)

                try:
                    self._do_parse_index(mfp)
                finally:
                    if isinstance(mfp, mmap):
                        mfp.close()

        return set(self._pool_files.values())

    def _is_safe_path(self, path: Path):
        return is_safe_path(self._repository_path, path)

    def _unpack_index(self, file: Path) -> bool:
        for suffix, open_function in BaseRepository.COMPRESSION_SUFFIXES.items():
            compressed_file = file.with_name(f"{file.name}{suffix}")
            if not compressed_file.exists():
                continue

            try:
                with (
                    open_function(compressed_file, "rb") as source_fp,
                    open(file, "wb") as target_fp,
                ):
                    shutil.copyfileobj(source_fp, target_fp)
                    shutil.copystat(compressed_file, file)

                    return True
            except (lzma.LZMAError, OSError):
                return False

        return file.exists()

    def _should_ignore_errors(self, path: Path):
        return should_ignore_errors(self._ignore_errors, path)

    @abstractmethod
    def _do_parse_index(self, fp: IO[bytes] | mmap): ...


# https://github.com/pylint-dev/pylint/issues/5214
# pylint: disable=W0201
class SourcesParser(IndexFileParser):
    def __init__(
        self,
        repository_path: Path,
        index_files: set[Path],
        ignore_errors: set[str],
        package_filter: PackageFilter,
        logger_id: Any | None = None,
    ) -> None:
        super().__init__(
            repository_path, index_files, ignore_errors, logger_id=logger_id
        )

        self._package_filter = package_filter

        self._reset_block_parser()

    # https://github.com/pylint-dev/pylint/issues/5214
    def _reset_block_parser(self):
        self._package: str | None = None
        self._directory: Path | None = None
        self._hash_type = None
        self._package_files: dict[Path, PackageFile] = {}
        self._section: str | None = None

    def _do_parse_index(self, fp: IO[bytes] | mmap):
        self._reset_block_parser()

        for bytes_line in itertools.chain(iter(fp.readline, b""), (b"\n",)):
            if bytes_line[0] == ord(" "):
                if not self._hash_type:
                    continue

                try:
                    hashsum, _size, filename = (
                        bytes_line.decode().strip().split(maxsplit=2)
                    )
                except ValueError:
                    continue

                if " " in filename:
                    continue

                file_path = Path(filename)
                if not self._is_safe_path(file_path):
                    self._log.warning(f"Skipping unsafe path: {file_path}")
                    continue

                self._package_files.setdefault(
                    file_path,
                    PackageFile(path=file_path, size=int(_size), hashes={}),
                ).hashes[self._hash_type] = HashSum(type=self._hash_type, hash=hashsum)
            elif bytes_line[0] != ord("\n"):
                match bytes_line:
                    case line if line.startswith(b"Package:"):
                        _, self._package = line.decode().strip().split()
                    case line if line.startswith(b"Directory:"):
                        _, directory = line.decode().strip().split()

                        directory = Path(directory)
                        if not self._is_safe_path(directory):
                            self._log.warning(f"Skipping unsafe Directory: {directory}")
                            continue

                        self._directory = directory
                    case line if line.startswith(b"Section:"):
                        _, self._section = line.decode().strip().split()
                        continue
                    case line if line.startswith(b"Files:"):
                        self._hash_type = HashType.MD5
                        continue
                    case line if line.startswith(b"Checksums-"):
                        match line[len(b"Checksums-") : -1]:
                            case b"Sha1:":
                                self._hash_type = HashType.SHA1
                            case b"Sha256:":
                                self._hash_type = HashType.SHA256
                            case b"Sha512:":
                                self._hash_type = HashType.SHA512
                            case _:
                                self._hash_type = None

                        continue
                    case _:
                        self._hash_type = None
                        continue
            else:
                if not self._package or not self._directory:
                    self._reset_block_parser()
                    continue

                if not self._package_filter.package_allowed(
                    self._package, section=self._section
                ):
                    self._reset_block_parser()
                    continue

                for package_file in self._package_files.values():
                    download_file = package_file.to_download_file(self._directory)
                    download_file.ignore_errors = self._should_ignore_errors(
                        download_file.path
                    )
                    self._pool_files[download_file.path] = download_file

                self._reset_block_parser()


class PackagesParser(IndexFileParser):
    def __init__(
        self,
        repository_path: Path,
        index_files: set[Path],
        ignore_errors: set[str],
        package_filter: PackageFilter,
        logger_id: Any | None = None,
    ) -> None:
        super().__init__(
            repository_path, index_files, ignore_errors, logger_id=logger_id
        )

        self._package_filter = package_filter

        self._reset_block_parser()

    def _reset_block_parser(self):
        self._package = None
        self._source = None
        self._section: str | None = None
        self._tags: set[str] = set()
        self._file_path = None
        self._size = 0
        self._hashes: dict[HashType, HashSum] = {}
        self._multiline: bytes | None = None

    def _do_parse_index(self, fp: IO[bytes] | mmap):
        for bytes_line in itertools.chain(iter(fp.readline, b""), (b"\n",)):
            if self._multiline is not None:
                if bytes_line[0] == ord(" "):
                    self._multiline += bytes_line
                else:
                    match self._multiline:
                        case line if line.startswith(b"Tag:"):
                            self._tags = {
                                v.strip() for v in self._get_line_value(line).split(",")
                            }

                    self._multiline = None

            if bytes_line[0] != ord("\n"):
                match bytes_line:
                    case line if line.startswith(b"Package:"):
                        self._package = self._get_line_value(line)
                    case line if line.startswith(b"Source:"):
                        self._source = self._get_line_value(line).split()[0]
                    case line if line.startswith(b"Filename:"):
                        file_path = Path(self._get_line_value(line))

                        if not self._is_safe_path(file_path):
                            self._log.warning(f"Skipping unsafe path: {file_path}")
                            continue

                        self._file_path = file_path
                    case line if line.startswith(b"Size:"):
                        self._size = int(self._get_line_value(line))
                    case line if line.startswith(b"Section:"):
                        self._section = self._get_line_value(line)
                        continue
                    case line if line.startswith(b"Tag:"):
                        self._multiline = line
                        continue
                    case line if line.startswith(b"%s:" % HashType.MD5.value.encode()):
                        self._hashes[HashType.MD5] = HashSum(
                            type=HashType.MD5, hash=self._get_line_value(line)
                        )
                    case line if line.startswith(b"%s:" % HashType.SHA1.value.encode()):
                        self._hashes[HashType.MD5] = HashSum(
                            type=HashType.SHA1, hash=self._get_line_value(line)
                        )
                    case line if line.startswith(
                        b"%s:" % HashType.SHA256.value.encode()
                    ):
                        self._hashes[HashType.SHA256] = HashSum(
                            type=HashType.SHA256, hash=self._get_line_value(line)
                        )
                    case line if line.startswith(
                        b"%s:" % HashType.SHA512.value.encode()
                    ):
                        self._hashes[HashType.SHA512] = HashSum(
                            type=HashType.SHA512, hash=self._get_line_value(line)
                        )
                    case _:
                        self._hash_type = None
                        continue
            else:
                if not self._package or not self._file_path or not self._size:
                    self._reset_block_parser()
                    continue

                source_name = self._source if self._source else self._package

                if not self._package_filter.package_allowed(
                    source_name, self._package, section=self._section, tags=self._tags
                ):
                    self._reset_block_parser()
                    continue

                download_file = DownloadFile.from_path(self._file_path, check_size=True)
                download_file.ignore_errors = self._should_ignore_errors(
                    download_file.path
                )
                for hash_type, hashsum in self._hashes.items():
                    download_file.add_compression_variant(
                        path=self._file_path,
                        size=self._size,
                        hash_type=hash_type,
                        hash_sum=hashsum,
                        use_by_hash=False,
                    )

                self._pool_files[self._file_path] = download_file

                self._reset_block_parser()

    def _get_line_value(self, line: bytes):
        return line.decode().strip().split(":", maxsplit=1)[1].strip()


# pylint: enable=W0201


class ByHash(Enum):
    YES = "yes"
    NO = "no"
    FORCE = "force"

    @staticmethod
    def default():
        return ByHash.YES


class GpgInfoExtended(GpgInfo):
    @classmethod
    def from_file(
        cls, target, keyrings: Iterable[str] | None, *args: Any, **kwargs: Any
    ):
        """Create a new GpgInfo object from the given file or
        a tuple of file signature and file.

        See GpgInfo.from_sequence.
        Based on the GpgInfo from python-debian
        TODO: send patch upstream
        """
        if isinstance(target, str):
            return super().from_file(target, *args, keyrings=keyrings, **kwargs)

        keyrings = keyrings or GPGV_DEFAULT_KEYRINGS
        executable = [GPGV_EXECUTABLE]

        process_args = list(executable)
        process_args.extend(["--status-fd", "1"])
        for k in keyrings:
            process_args.extend(["--keyring", k])

        if "--keyring" not in process_args:
            raise OSError("cannot access any of the given keyrings")

        process_args.extend(target)

        with subprocess.Popen(
            process_args,
            stdout=subprocess.PIPE,
            stderr=subprocess.PIPE,
            universal_newlines=False,
        ) as p:
            out, err = p.communicate()

        return cls.from_output(out.decode("utf-8"), err.decode("utf-8"))


class GPGVerifyMeta(EnumMeta):
    def __call__(cls, value, *args: Any, **kwds: Any) -> Any:
        try:
            return super().__call__(value, *args, **kwds)
        except ValueError:
            LoggerFactory.get_logger(cls).warning(
                f"Wrong `gpg_verify` value: {value}. Using default value."
            )
            return GPGVerify.default()


class GPGVerify(Enum, metaclass=GPGVerifyMeta):
    ON = "on"
    OFF = "off"
    FORCE = "force"

    @staticmethod
    def default():
        return GPGVerify.OFF


@dataclass
class BaseRepositoryMetadata(ABC):
    by_hash: ByHash
    sign_by: list[Path] | None

    @abstractmethod
    def as_path(self) -> Path: ...

    def as_string(self) -> str:
        return str(self)

    def __hash__(self) -> int:
        return hash(str(self))


@dataclass(eq=False)
class Codename(BaseRepositoryMetadata):
    @dataclass
    class Component:
        name: str
        mirror_source: bool
        arches: list[str]

    codename: str
    components: dict[str, Component]

    def as_path(self) -> Path:
        return Path("dists") / self.codename

    def should_mirror_source(self):
        return any(c.mirror_source for c in self.components.values())

    def should_mirror_binaries(self):
        return any(c.arches for c in self.components.values())

    def __str__(self) -> str:
        return self.codename


@dataclass(eq=False)
class FlatDirectory(BaseRepositoryMetadata):
    directory: Path
    mirror_source: bool
    mirror_binaries: bool

    def as_path(self) -> Path:
        return self.directory

    def __str__(self) -> str:
        return str(self.directory)


@dataclass
class BaseRepository(ABC):
    COMPRESSION_SUFFIXES = {
        FileCompression.XZ.file_extension: lzma.open,
        FileCompression.GZ.file_extension: gzip.open,
        FileCompression.BZ2.file_extension: bz2.open,
    }

    ALL_INDEX_SUFFIXES = list(COMPRESSION_SUFFIXES.keys()) + [""]

    RELEASE_FILES = (
        "InRelease",
        "Release",
        "Release.gpg",
    )

    url: URL

    clean: bool
    skip_clean: set[Path]
    http2_disable: bool
    mirror_dist_upgrader: bool
    mirror_path: Path | None
    ignore_errors: set[str]
    gpg_verify: GPGVerify

    def __post_init__(self):
        self._log = LoggerFactory.get_logger(self)

        self.package_filter = PackageFilter()

    def get_mirror_path(self, encode_tilde: bool):
        if self.mirror_path:
            return self.mirror_path

        return self.url.as_filesystem_path(encode_tilde)

    def as_filename(self, encode_tilde: bool) -> Path:
        return Path(
            str(self.get_mirror_path(encode_tilde=encode_tilde)).replace(os.sep, "_")
        )

    def get_clean_script_name(self, encode_tilde: bool):
        path = Path(str(self.get_mirror_path(encode_tilde)).replace("/", "_"))
        return path.with_name(f"{path.name}.sh")

    def get_metadata_files(
        self, repository_root: Path, encode_tilde: bool, missing_sources: set[Path]
    ) -> set[DownloadFile]:
        metadata_files: set[DownloadFile] = set()
        mirror_path = repository_root / self.get_mirror_path(encode_tilde)

        for (
            metadata,
            metadata_release_files,
        ) in self.release_files_per_metadata.items():
            codename_metadata_files: dict[Path, DownloadFile] = {}

            for release_file_relative_path in metadata_release_files:
                if release_file_relative_path.suffix == ".gpg":
                    continue

                if release_file_relative_path in missing_sources:
                    continue

                release_file = mirror_path / release_file_relative_path

                if not release_file.exists():
                    continue

                with open(release_file, "rt", encoding="utf-8") as fp:
                    release = Release(fp)

                use_hash = release.get("Acquire-By-Hash") == "yes"
                if use_hash:
                    if self.get_by_hash_policy(metadata) == ByHash.NO:
                        use_hash = False
                elif self.get_by_hash_policy(metadata) == ByHash.FORCE:
                    use_hash = True

                for hash_type in HashType:
                    for file in release.get(hash_type.value, []):
                        path = Path(file["name"])

                        if not is_safe_path(mirror_path, path):
                            self._log.warning(f"Skipping unsafe path: {path}")
                            continue

                        try:
                            size = int(file["size"])
                        except ValueError:
                            size = 0

                        if size <= 0:
                            continue

                        # Ignore release files in release files
                        if file["name"] in self.RELEASE_FILES:
                            continue

                        if not self._metadata_file_allowed(metadata, path):
                            continue

                        hash_sum = HashSum(type=hash_type, hash=file[hash_type.value])

                        repository_path = release_file_relative_path.parent / path
                        uncompressed_path = DownloadFile.uncompressed_path(path)
                        if uncompressed_path in codename_metadata_files:
                            codename_metadata_files[
                                uncompressed_path
                            ].add_compression_variant(
                                path=repository_path,
                                size=size,
                                hash_type=hash_type,
                                hash_sum=hash_sum,
                                use_by_hash=use_hash,
                            )
                        else:
                            codename_metadata_files[uncompressed_path] = (
                                DownloadFile.from_hashed_path(
                                    repository_path,
                                    size=size,
                                    hash_type=hash_type,
                                    hash_sum=hash_sum,
                                    use_by_hash=use_hash,
                                )
                            )

                        codename_metadata_files[
                            uncompressed_path
                        ].ignore_errors = should_ignore_errors(
                            self.ignore_errors, uncompressed_path
                        )

            if not codename_metadata_files:
                self._log.warning(f"No metadata files found for codename {metadata}")

            metadata_files.update(codename_metadata_files.values())

        return metadata_files

    def _get_sources_files_field(self, hash_type: HashType):
        if hash_type == HashType.MD5:
            return "Files"

        return f"Checksum-{hash_type.value.capitalize()}"

    def get_pool_files(
        self, repository_root: Path, encode_tilde: bool, missing_sources: set[Path]
    ) -> Iterable[DownloadFile]:
        pool_files: set[DownloadFile] = set()

        if self.is_source_enabled:
            pool_files.update(
                SourcesParser(
                    repository_root / self.get_mirror_path(encode_tilde),
                    set(self.sources_files) - missing_sources,
                    ignore_errors=self.ignore_errors,
                    package_filter=self.package_filter,
                    logger_id=self.url,
                ).parse()
            )

        if self.is_binaries_enabled:
            pool_files.update(
                PackagesParser(
                    repository_root / self.get_mirror_path(encode_tilde),
                    set(self.packages_files) - missing_sources,
                    ignore_errors=self.ignore_errors,
                    package_filter=self.package_filter,
                    logger_id=self.url,
                ).parse()
            )

        return pool_files

    def _get_gpg_keys(
        self,
        sign_by: list[Path] | None,
        etc_trusted: Path,
        etc_trusted_parts: Path,
    ):
        if sign_by:
            yield from sign_by
        else:
            for file in itertools.chain(
                [etc_trusted],
                etc_trusted_parts.iterdir() if etc_trusted_parts.exists() else [],
            ):
                if not file.is_file() or not os.access(file, os.R_OK):
                    continue

                yield file

    @contextmanager
    def _get_merged_gpg_keyring(
        self,
        sign_by: list[Path] | None,
        etc_trusted: Path,
        etc_trusted_parts: Path,
    ) -> Generator[str | None, Any, None]:
        if self.gpg_verify == GPGVerify.OFF:
            yield None
            return

        # Mimic apt behavior
        # https://salsa.debian.org/apt-team/apt/-/blob/63919b628a9bf386136f708f06c1a8a7d4f09fca/apt-pkg/contrib/gpgv.cc#L311
        with NamedTemporaryFile(prefix="apt-mirror2.", suffix=".gpg") as keyring:
            for file in self._get_gpg_keys(sign_by, etc_trusted, etc_trusted_parts):
                if file.suffix == ".asc":
                    with open(file, "rt", encoding="ascii") as fp:
                        if not next(fp, "").startswith(
                            "-----BEGIN PGP PUBLIC KEY BLOCK-----"
                        ):
                            continue

                        if next(fp, "-").strip() != "":
                            continue

                        base64_data = ""
                        for line in fp:
                            line = line.strip()

                            if line.startswith("-----END"):
                                if base64_data:
                                    keyring.write(base64.b64decode(base64_data))

                                break

                            if not line or line[0] in ("=", "-"):
                                continue

                            base64_data += line
                else:
                    with open(file, "rb") as fp:
                        header = fp.read(1)

                        if len(header) != 1:
                            continue

                        # OpenPGP public key packets
                        # https://salsa.debian.org/apt-team/apt/-/blob/63919b628a9bf386136f708f06c1a8a7d4f09fca/apt-pkg/contrib/gpgv.cc#L352
                        if header[0] not in (0x98, 0x99, 0xC6):
                            continue

                        keyring.write(header)
                        shutil.copyfileobj(fp, keyring)

            keyring.flush()
            yield keyring.name

    def _validate_release_file_signature(self, release_file: Path, keyring: str | None):
        if self.gpg_verify != GPGVerify.OFF:
            gpg_info = None
            gpg_target = None
            keyrings = [keyring] if keyring else None

            if release_file.name == "InRelease":
                gpg_target = str(release_file)
            else:
                release_gpg_file = release_file.with_name(f"{release_file.name}.gpg")
                if release_gpg_file.is_file():
                    gpg_target = (str(release_gpg_file), str(release_file))

            if gpg_target:
                gpg_info = GpgInfoExtended.from_file(
                    gpg_target,
                    keyrings=keyrings,
                )

            if gpg_info is None:
                if self.gpg_verify == GPGVerify.FORCE:
                    raise InvalidSignatureError(
                        f"Unable to find GPG signature for file {release_file}"
                    )
            elif not gpg_info.valid():
                raise InvalidSignatureError(
                    f"Unable to verify release file signature: {release_file}",
                    "\n".join(gpg_info.err) if gpg_info.err else None,
                )

    def validate_release_files(
        self,
        repository_root: Path,
        encode_tilde: bool,
        etc_trusted: Path,
        etc_trusted_parts: Path,
    ):
        release_files_exists = False

        for (
            metadata,
            metadata_release_files,
        ) in self.release_files_per_metadata.items():
            metadata_sizes: dict[str, list[tuple[int, Path]]] = {}
            metadata_hashes: dict[str, dict[HashType, list[tuple[str, Path]]]] = {}

            with self._get_merged_gpg_keyring(
                metadata.sign_by, etc_trusted, etc_trusted_parts
            ) as keyring:
                for release_file_relative_path in metadata_release_files:
                    if release_file_relative_path.suffix == ".gpg":
                        continue

                    release_file = (
                        repository_root
                        / self.get_mirror_path(encode_tilde)
                        / release_file_relative_path
                    )

                    if not release_file.exists():
                        continue

                    with open(release_file, "rt", encoding="utf-8") as fp:
                        release = Release(fp)

                    self._validate_release_file_signature(release_file, keyring)

                    release_files_exists = True
                    for hash_type in HashType:
                        for file in release.get(hash_type.value, []):
                            try:
                                size = int(file["size"])
                            except ValueError:
                                size = 0

                            if size <= 0:
                                continue

                            # Ignore release files in release files
                            if file["name"] in self.RELEASE_FILES:
                                continue

                            path = file["name"]
                            hash_sum = file[hash_type.value]

                            if any(size != s[0] for s in metadata_sizes.get(path, [])):
                                raise InvalidReleaseFilesException(
                                    f"Size of file {path} in release file"
                                    f" {release_file_relative_path} differs from size"
                                    f" in release file {metadata_sizes[path][0][1]}"
                                )

                            if any(
                                hash_sum != s[0]
                                for s in metadata_hashes.get(path, {}).get(
                                    hash_type, []
                                )
                            ):
                                raise InvalidReleaseFilesException(
                                    f"Hashsum of type {hash_type} of file {path} in"
                                    f" release file {release_file_relative_path}"
                                    " differs from hashsum in release file"
                                    f" {metadata_hashes[path][hash_type][0][1]}"
                                )

                            metadata_sizes.setdefault(path, []).append(
                                (size, release_file_relative_path)
                            )
                            metadata_hashes.setdefault(path, {}).setdefault(
                                hash_type, []
                            ).append((hash_sum, release_file_relative_path))

        if not release_files_exists:
            raise InvalidReleaseFilesException("No release files were found")

    @property
    def release_files(self) -> Sequence[Path]:
        return [
            path
            for _, paths in self.release_files_per_metadata.items()
            for path in paths
        ]

    @abstractmethod
    def _metadata_file_allowed(
        self, metadata: BaseRepositoryMetadata, file_path: Path
    ) -> bool: ...

    @abstractmethod
    def get_by_hash_policy(self, metadata: BaseRepositoryMetadata) -> ByHash: ...

    @abstractmethod
    def get_sign_by(self, metadata: BaseRepositoryMetadata) -> list[Path] | None: ...

    @property
    @abstractmethod
    def is_source_enabled(self) -> bool: ...

    @property
    @abstractmethod
    def is_binaries_enabled(self) -> bool: ...

    @property
    @abstractmethod
    def release_files_per_metadata(
        self,
    ) -> dict[BaseRepositoryMetadata, Sequence[Path]]: ...

    @property
    @abstractmethod
    def sources_files(self) -> Sequence[Path]: ...

    @property
    @abstractmethod
    def packages_files(self) -> Sequence[Path]: ...


@dataclass
class Repository(BaseRepository):
    class Codenames(dict[str, Codename]):
        def get_codename(self, codename: str) -> Codename:
            if codename not in self:
                raise RuntimeError(f"Requested codename was not found: {codename}")

            return self[codename]

        def __str__(self) -> str:
            return str([c for c in self])

    DISTS = Path("dists")
    DIST_UPGRADER_ANNOUNCEMENTS = [
        "DevelReleaseAnnouncement",
        "EOLReleaseAnnouncement",
        "ReleaseAnnouncement",
    ]

    codenames: Codenames

    def get_metadata_files(
        self, repository_root: Path, encode_tilde: bool, missing_sources: set[Path]
    ) -> set[DownloadFile]:
        metadata_files = super().get_metadata_files(
            repository_root, encode_tilde, missing_sources
        )

        if not self.mirror_dist_upgrader:
            return metadata_files

        # Add Ubuntu's dist-upgrader paths
        for codename in self.codenames.values():
            if "main" not in codename.components:
                continue

            dist_upgrader_path = Path(
                f"dists/{codename.codename}/main/dist-upgrader-all/current"
            )
            for file in self.DIST_UPGRADER_ANNOUNCEMENTS:
                announcement_path = dist_upgrader_path / file
                metadata_files.add(
                    DownloadFile.from_path(announcement_path, False, True)
                )
                metadata_files.add(
                    DownloadFile.from_path(
                        announcement_path.with_suffix(".html"), False, True
                    )
                )

            metadata_filename = codename.codename.split("-", maxsplit=1)[0]
            metadata_files.add(
                DownloadFile.from_path(
                    dist_upgrader_path / f"{metadata_filename}.tar.gz", False, True
                )
            )
            metadata_files.add(
                DownloadFile.from_path(
                    dist_upgrader_path / f"{metadata_filename}.tar.gz.gpg", False, True
                )
            )

        return metadata_files

    def _metadata_file_allowed(
        self, metadata: BaseRepositoryMetadata, file_path: Path
    ) -> bool:
        codename = self.codenames.get_codename(metadata.as_string())
        file_path_str = str(file_path)

        # Skip source metadata if not needed
        if not codename.should_mirror_source() and (
            "/source/" in file_path_str or file_path.name.startswith("Contents-source")
        ):
            return False

        # Skip binary metadata if not needed
        if not codename.should_mirror_binaries() and any(
            part in file_path_str for part in ("/binary-", "/cnf/", "/dep11/", "/i18n/")
        ):
            return False

        # Skip redundand components
        components_split = min(file_path_str.count("/"), 2)
        if components_split >= 1:
            file_component = file_path_str.rsplit("/", maxsplit=components_split)[0]

            if not any(
                file_component == component.name
                for component in codename.components.values()
            ):
                return False

            if (
                "/binary-" in file_path_str
                and "source" not in file_path_str
                and (
                    not codename.should_mirror_binaries()
                    or not any(
                        arch in file_path_str
                        for arch in itertools.chain(
                            codename.components[file_component].arches, ["-all"]
                        )
                    )
                )
            ):
                return False

        all_arches = set(
            arch
            for component in codename.components.values()
            for arch in component.arches
        )
        if all_arches:
            all_arches.add("all")

        if (
            any(
                file_path.name.startswith(suffix)
                for suffix in ("Commands-", "Components-", "Contents-")
            )
            and "source" not in file_path.name
            and not any(arch in file_path.name for arch in all_arches)
        ):
            return False

        if (
            "Contents-" in file_path_str
            and ".diff" in file_path_str
            and not any(arch in file_path.name for arch in all_arches)
        ):
            return False

        # Allow any metadata not explicitly disallowed
        return True

    @property
    def is_source_enabled(self) -> bool:
        return any(c.should_mirror_source() for c in self.codenames.values())

    @property
    def is_binaries_enabled(self) -> bool:
        return any(c.should_mirror_binaries() for c in self.codenames.values())

    @property
    def release_files_per_metadata(
        self,
    ) -> dict[BaseRepositoryMetadata, Sequence[Path]]:
        return {
            codename: [
                self.DISTS / codename.as_string() / file for file in self.RELEASE_FILES
            ]
            for codename in self.codenames.values()
        }

    @property
    def sources_files(self) -> Sequence[Path]:
        return [
            self.DISTS / codename.as_string() / component.name / "source" / "Sources"
            for codename in self.codenames.values()
            for component in codename.components.values()
            if component.mirror_source
        ]

    @property
    def packages_files(self) -> Sequence[Path]:
        return [
            path
            for path in itertools.chain.from_iterable(
                itertools.chain(
                    (
                        self.DISTS
                        / codename.as_string()
                        / component.name
                        / f"binary-{arch}"
                        / "Packages"
                        for arch in component.arches
                    ),
                    [
                        self.DISTS
                        / codename.as_string()
                        / component.name
                        / "binary-all"
                        / "Packages"
                    ],
                )
                for codename in self.codenames.values()
                for component in codename.components.values()
                if component.arches
            )
        ]

    def get_by_hash_policy(self, metadata: BaseRepositoryMetadata) -> ByHash:
        return self.codenames.get_codename(metadata.as_string()).by_hash

    def get_sign_by(self, metadata: BaseRepositoryMetadata) -> list[Path] | None:
        return self.codenames.get_codename(metadata.as_string()).sign_by

    def __str__(self) -> str:
        return (
            f"{self.url}, codenames: {self.codenames}, mirror_path: {self.mirror_path}"
        )


@dataclass
class FlatRepository(BaseRepository):
    class FlatDirectories(dict[Path, FlatDirectory]):
        def get_directory(self, directory: Path) -> FlatDirectory:
            if directory not in self:
                raise RuntimeError(f"Requested directory was not found: {directory}")

            return self[directory]

        def __str__(self) -> str:
            return str([d.as_string() for d in self.values()])

    directories: FlatDirectories

    def _metadata_file_allowed(
        self, metadata: BaseRepositoryMetadata, file_path: Path
    ) -> bool:
        directory = self.directories.get_directory(metadata.as_path())

        if not directory.mirror_source and (
            any(
                str(file_path) == f"Sources{suffix}"
                for suffix in self.ALL_INDEX_SUFFIXES
            )
        ):
            return False

        if not directory.mirror_binaries and (
            any(
                str(file_path) == f"Packages{suffix}"
                for suffix in self.ALL_INDEX_SUFFIXES
            )
        ):
            return False

        return True

    @property
    def is_source_enabled(self) -> bool:
        return any(d.mirror_source for d in self.directories.values())

    @property
    def is_binaries_enabled(self) -> bool:
        return any(d.mirror_binaries for d in self.directories.values())

    @property
    def release_files_per_metadata(
        self,
    ) -> dict[BaseRepositoryMetadata, Sequence[Path]]:
        return {
            directory: [directory.as_path() / file for file in self.RELEASE_FILES]
            for directory in self.directories.values()
        }

    @property
    def sources_files(self) -> Sequence[Path]:
        return [
            directory.as_path() / "Sources" for directory in self.directories.values()
        ]

    @property
    def packages_files(self) -> Sequence[Path]:
        return [
            directory.as_path() / "Packages" for directory in self.directories.values()
        ]

    def get_by_hash_policy(self, metadata: BaseRepositoryMetadata) -> ByHash:
        return self.directories.get_directory(metadata.as_path()).by_hash

    def get_sign_by(self, metadata: BaseRepositoryMetadata) -> list[Path] | None:
        return self.directories.get_directory(metadata.as_path()).sign_by

    def __str__(self) -> str:
        return (
            f"{self.url}, directories: {self.directories},"
            f" mirror_path: {self.mirror_path}"
        )