vcsserver/lib/archive_cache/backends/fanout_cache.py

# Copyright (C) 2015-2024 RhodeCode GmbH
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License, version 3
# (only), as published by the Free Software Foundation.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#
# This program is dual-licensed. If you wish to learn more about the
# RhodeCode Enterprise Edition, including its added features, Support services,
# and proprietary license terms, please see https://rhodecode.com/licenses/

import codecs
import hashlib
import logging
import os
import typing

import fsspec

from .base import BaseCache, BaseShard
from ..utils import ShardFileReader, NOT_GIVEN
from ...type_utils import str2bool

log = logging.getLogger(__name__)


class FileSystemShard(BaseShard):

    def __init__(self, index, directory, directory_folder, fs, **settings):
        self._index: int = index
        self._directory: str = directory
        self._directory_folder: str = directory_folder
        self.storage_type: str = 'directory'

        self.fs = fs

    @property
    def directory(self) -> str:
        """Cache directory final path."""
        return os.path.join(self._directory, self._directory_folder)

    def _get_keyfile(self, archive_key) -> tuple[str, str]:
        key_file: str = f'{archive_key}.{self.key_suffix}'
        return key_file, os.path.join(self.directory, key_file)

    def _get_writer(self, path, mode):
        for count in range(1, 11):
            try:
                # Another cache may have deleted the directory before
                # the file could be opened.
                return self.fs.open(path, mode)
            except OSError:
                if count == 10:
                    # Give up after 10 tries to open the file.
                    raise
                continue

    def _write_file(self, full_path, iterator, mode):

        # ensure dir exists
        destination, _ = os.path.split(full_path)
        if not self.fs.exists(destination):
            self.fs.makedirs(destination)

        writer = self._get_writer(full_path, mode)

        digest = hashlib.sha256()
        with writer:
            size = 0
            for chunk in iterator:
                size += len(chunk)
                digest.update(chunk)
                writer.write(chunk)
            writer.flush()
            # Get the file descriptor
            fd = writer.fileno()

            # Sync the file descriptor to disk, helps with NFS cases...
            os.fsync(fd)
            sha256 = digest.hexdigest()
            log.debug('written new archive cache under %s, sha256: %s', full_path, sha256)
            return size, sha256

    def store(self, key, value_reader, metadata: dict | None = None):
        return self._store(key, value_reader, metadata, mode='xb')

    def fetch(self, key, retry=NOT_GIVEN,
              retry_attempts=NOT_GIVEN, retry_backoff=1, **kwargs) -> tuple[ShardFileReader, dict]:
        return self._fetch(key, retry, retry_attempts, retry_backoff)

    def remove(self, key):
        return self._remove(key)

    def random_filename(self):
        """Return filename and full-path tuple for file storage.

        Filename will be a randomly generated 28 character hexadecimal string
        with ".archive_cache" suffixed. Two levels of sub-directories will be used to
        reduce the size of directories. On older filesystems, lookups in
        directories with many files may be slow.
        """

        hex_name = codecs.encode(os.urandom(16), 'hex').decode('utf-8')

        archive_name = hex_name[4:] + '.archive_cache'
        filename = f"{hex_name[:2]}/{hex_name[2:4]}/{archive_name}"

        full_path = os.path.join(self.directory, filename)
        return archive_name, full_path

    def __repr__(self):
        return f'{self.__class__.__name__}(index={self._index}, dir={self.directory})'


class FileSystemFanoutCache(BaseCache):
    shard_name: str = 'shard_{:03d}'
    shard_cls = FileSystemShard

    def __init__(self, locking_url, **settings):
        """
        Initialize file system cache instance.

        :param str locking_url: redis url for a lock
        :param settings: settings dict

        """
        self._locking_url = locking_url
        self._config = settings
        cache_dir = self.get_conf('archive_cache.filesystem.store_dir')
        directory = str(cache_dir)
        directory = os.path.expanduser(directory)
        directory = os.path.expandvars(directory)
        self._directory = directory
        self._storage_path = directory  # common path for all from BaseCache

        self._shard_count = int(self.get_conf('archive_cache.filesystem.cache_shards', pop=True))
        if self._shard_count < 1:
            raise ValueError('cache_shards must be 1 or more')

        self._eviction_policy = self.get_conf('archive_cache.filesystem.eviction_policy', pop=True)
        self._cache_size_limit = self.gb_to_bytes(int(self.get_conf('archive_cache.filesystem.cache_size_gb')))

        self.retry = str2bool(self.get_conf('archive_cache.filesystem.retry', pop=True))
        self.retry_attempts = int(self.get_conf('archive_cache.filesystem.retry_attempts', pop=True))
        self.retry_backoff = int(self.get_conf('archive_cache.filesystem.retry_backoff', pop=True))

        log.debug('Initializing %s archival cache instance', self)
        fs = fsspec.filesystem('file')
        # check if it's ok to write, and re-create the archive cache main dir
        # A directory is the virtual equivalent of a physical file cabinet.
        # In other words, it's a container for organizing digital data.
        # Unlike a folder, which can only store files, a directory can store files,
        # subdirectories, and other directories.
        if not fs.exists(self._directory):
            fs.makedirs(self._directory, exist_ok=True)

        self._shards = tuple(
            self.shard_cls(
                index=num,
                directory=directory,
                directory_folder=self.shard_name.format(num),
                fs=fs,
                **settings,
            )
            for num in range(self._shard_count)
        )
        self._hash = self._shards[0].hash

    def _get_size(self, shard, archive_path):
        return os.stat(archive_path).st_size