# Copyright (C) 2022 Anaconda, Inc # Copyright (C) 2023 conda # SPDX-License-Identifier: BSD-3-Clause """ Cache suitable for shards, not allowed to change because they are named after their own sha256 hash. """ from __future__ import annotations import logging import sqlite3 from dataclasses import dataclass from typing import TYPE_CHECKING import msgpack import zstandard from conda.gateways.disk.delete import unlink_or_rename_to_trash if TYPE_CHECKING: from pathlib import Path from .shards_typing import ShardDict log = logging.getLogger(__name__) SHARD_CACHE_NAME = "repodata_shards.db" ZSTD_MAX_SHARD_SIZE = 2**20 * 16 # maximum size necessary when compresed data has no size header @dataclass class AnnotatedRawShard: def __init__(self, url: str, package: str, compressed_shard: bytes): # prevent easy mistake of swapping url, package assert "://" in url assert "://" not in package self.url = url self.package = package # remove this field to avoid confusion? self.compressed_shard = compressed_shard url: str package: str compressed_shard: bytes def connect(dburi="cache.db"): """ Get database connection. dburi: uri-style sqlite database filename; accepts certain ?= parameters. """ conn = sqlite3.connect(dburi, uri=True) conn.row_factory = sqlite3.Row conn.execute("PRAGMA foreign_keys = ON") return conn class ShardCache: """ Handle caching for individual shards (not the index of shards). """ def __init__(self, base: Path, create=True): """ base: directory and filename prefix for cache. """ self.base = base self.connect() def copy(self): """ Copy cache with new connection. Useful for threads. """ return ShardCache(self.base, create=False) def connect(self, create=True): """ Args: create: if True, create table if not exists. """ dburi = (self.base / SHARD_CACHE_NAME).as_uri() self.conn = connect(dburi) if not create: return # this schema will also get confused if we merge packages into a single # shard, but the package name should be advisory. self.conn.execute( "CREATE TABLE IF NOT EXISTS shards (" "url TEXT PRIMARY KEY, package TEXT, shard BLOB, " "timestamp TIMESTAMP DEFAULT CURRENT_TIMESTAMP)" ) def insert(self, raw_shard: AnnotatedRawShard): """ Args: url: of shard package: package name raw_shard: msgpack.zst compressed shard data """ # decompress and return shard for convenience, also to validate? unless # caller would rather retrieve the shard from another thread. with self.conn as c: c.execute( "INSERT OR IGNORE INTO SHARDS (url, package, shard) VALUES (?, ?, ?)", (raw_shard.url, raw_shard.package, raw_shard.compressed_shard), ) def retrieve(self, url) -> ShardDict | None: with self.conn as c: row = c.execute("SELECT shard FROM shards WHERE url = ?", (url,)).fetchone() return ( msgpack.loads( zstandard.decompress(row["shard"], max_output_size=ZSTD_MAX_SHARD_SIZE) ) if row else None ) # type: ignore def retrieve_multiple(self, urls: list[str]) -> dict[str, ShardDict | None]: """ Query database for cached shard urls. Return a dict of urls in cache mapping to the Shard or None if not present. """ if not urls: return {} # this optimization does not save a noticeable amount of time. # In one test reusing the context saves difference between .006s and .01s # We could make this a threadlocal. dctx = zstandard.ZstdDecompressor() query = f"SELECT url, shard FROM shards WHERE url IN ({','.join(('?',) * len(urls))}) ORDER BY url" with self.conn as c: result: dict[str, ShardDict | None] = { row["url"]: msgpack.loads( dctx.decompress(row["shard"], max_output_size=ZSTD_MAX_SHARD_SIZE) ) if row else None for row in c.execute(query, urls) # type: ignore } return result def clear_cache(self): """ Truncate the database by removing all rows from tables """ with self.conn as c: c.execute("DELETE FROM shards") def remove_cache(self): """ Remove the sharded cache database. """ # This function appears to support `Path()` except on Windows # `os.rename(path, path + ".conda_trash")` fails: self.conn.close() unlink_or_rename_to_trash(str(self.base / SHARD_CACHE_NAME))