scruby 0.9.3__py3-none-any.whl → 0.17.0__py3-none-any.whl

scruby/db.py

@@ -1,301 +1,756 @@
-"""Creation and management of the database."""
-
-from __future__ import annotations
-
-__all__ = ("Scruby",)
-
-import concurrent.futures
-import contextlib
-import logging
-import zlib
-from collections.abc import Callable
-from pathlib import Path as SyncPath
-from shutil import rmtree
-from typing import Any, Never, TypeVar, assert_never
-
-import orjson
-from anyio import Path, to_thread
-
-from scruby import constants
-
-logger = logging.getLogger(__name__)
-
-T = TypeVar("T")
-
-
-class Scruby[T]:
-    """Creation and management of database.
-
-    Args:
-        class_model: Class of Model (Pydantic).
-    """
-
-    def __init__(  # noqa: D107
-        self,
-        class_model: T,
-    ) -> None:
-        self.__class_model = class_model
-        self.__db_root = constants.DB_ROOT
-        self.__length_reduction_hash = constants.LENGTH_REDUCTION_HASH
-        # The maximum number of keys.
-        match self.__length_reduction_hash:
-            case 0:
-                self.__max_num_keys = 4294967296
-            case 2:
-                self.__max_num_keys = 16777216
-            case 4:
-                self.__max_num_keys = 65536
-            case 6:
-                self.__max_num_keys = 256
-            case _ as unreachable:
-                msg: str = f"{unreachable} - Unacceptable value for LENGTH_REDUCTION_HASH."
-                logger.critical(msg)
-                assert_never(Never(unreachable))
-
-    async def get_leaf_path(self, key: str) -> Path:
-        """Asynchronous method for getting path to collection cell by key.
-
-        Args:
-            key: Key name.
-        """
-        if not isinstance(key, str):
-            logger.error("The key is not a type of `str`.")
-            raise KeyError("The key is not a type of `str`.")
-        if len(key) == 0:
-            logger.error("The key should not be empty.")
-            raise KeyError("The key should not be empty.")
-        # Key to crc32 sum.
-        key_as_hash: str = f"{zlib.crc32(key.encode('utf-8')):08x}"[self.__length_reduction_hash :]
-        # Convert crc32 sum in the segment of path.
-        separated_hash: str = "/".join(list(key_as_hash))
-        # The path of the branch to the database.
-        branch_path: Path = Path(
-            *(
-                self.__db_root,
-                self.__class_model.__name__,
-                separated_hash,
-            ),
-        )
-        # If the branch does not exist, need to create it.
-        if not await branch_path.exists():
-            await branch_path.mkdir(parents=True)
-        # The path to the database cell.
-        leaf_path: Path = Path(*(branch_path, "leaf.json"))
-        return leaf_path
-
-    async def set_key(
-        self,
-        key: str,
-        value: T,
-    ) -> None:
-        """Asynchronous method for adding and updating keys to collection.
-
-        Args:
-            key: Key name.
-            value: Value of key.
-        """
-        # The path to the database cell.
-        leaf_path: Path = await self.get_leaf_path(key)
-        value_json: str = value.model_dump_json()
-        # Write key-value to the database.
-        if await leaf_path.exists():
-            # Add new key or update existing.
-            data_json: bytes = await leaf_path.read_bytes()
-            data: dict = orjson.loads(data_json) or {}
-            data[key] = value_json
-            await leaf_path.write_bytes(orjson.dumps(data))
-        else:
-            # Add new key to a blank leaf.
-            await leaf_path.write_bytes(orjson.dumps({key: value_json}))
-
-    async def get_key(self, key: str) -> T:
-        """Asynchronous method for getting value of key from collection.
-
-        Args:
-            key: Key name.
-        """
-        # The path to the database cell.
-        leaf_path: Path = await self.get_leaf_path(key)
-        # Get value of key.
-        if await leaf_path.exists():
-            data_json: bytes = await leaf_path.read_bytes()
-            data: dict = orjson.loads(data_json) or {}
-            obj: T = self.__class_model.model_validate_json(data[key])
-            return obj
-        msg: str = "`get_key` - The unacceptable key value."
-        logger.error(msg)
-        raise KeyError()
-
-    async def has_key(self, key: str) -> bool:
-        """Asynchronous method for checking presence of key in collection.
-
-        Args:
-            key: Key name.
-        """
-        # The path to the database cell.
-        leaf_path: Path = await self.get_leaf_path(key)
-        # Checking whether there is a key.
-        if await leaf_path.exists():
-            data_json: bytes = await leaf_path.read_bytes()
-            data: dict = orjson.loads(data_json) or {}
-            try:
-                data[key]
-                return True
-            except KeyError:
-                return False
-        return False
-
-    async def delete_key(self, key: str) -> None:
-        """Asynchronous method for deleting key from collection.
-
-        Args:
-            key: Key name.
-        """
-        # The path to the database cell.
-        leaf_path: Path = await self.get_leaf_path(key)
-        # Deleting key.
-        if await leaf_path.exists():
-            data_json: bytes = await leaf_path.read_bytes()
-            data: dict = orjson.loads(data_json) or {}
-            del data[key]
-            await leaf_path.write_bytes(orjson.dumps(data))
-            return
-        msg: str = "`delete_key` - The unacceptable key value."
-        logger.error(msg)
-        raise KeyError()
-
-    @staticmethod
-    async def napalm() -> None:
-        """Asynchronous method for full database deletion.
-
-        The main purpose is tests.
-
-        Warning:
-            - `Be careful, this will remove all keys.`
-        """
-        with contextlib.suppress(FileNotFoundError):
-            await to_thread.run_sync(rmtree, constants.DB_ROOT)
-        return
-
-    @staticmethod
-    def search_task(
-        key: int,
-        filter_fn: Callable,
-        length_reduction_hash: str,
-        db_root: str,
-        class_model: T,
-    ) -> dict[str, Any] | None:
-        """Search task."""
-        key_as_hash: str = f"{key:08x}"[length_reduction_hash:]
-        separated_hash: str = "/".join(list(key_as_hash))
-        leaf_path: SyncPath = SyncPath(
-            *(
-                db_root,
-                class_model.__name__,
-                separated_hash,
-                "leaf.json",
-            ),
-        )
-        if leaf_path.exists():
-            data_json: bytes = leaf_path.read_bytes()
-            data: dict[str, str] = orjson.loads(data_json) or {}
-            for _, val in data.items():
-                doc = class_model.model_validate_json(val)
-                if filter_fn(doc):
-                    return doc
-        return None
-
-    def find_one(
-        self,
-        filter_fn: Callable,
-        max_workers: int | None = None,
-        timeout: float | None = None,
-    ) -> T | None:
-        """Find a single document matching the filter.
-
-        The search is based on the effect of a quantum loop.
-        The search effectiveness depends on the number of processor threads.
-        Ideally, hundreds and even thousands of threads are required.
-
-        Args:
-            filter_fn: A function that execute the conditions of filtering.
-            max_workers: The maximum number of processes that can be used to
-                         execute the given calls. If None or not given then as many
-                         worker processes will be created as the machine has processors.
-            timeout: The number of seconds to wait for the result if the future isn't done.
-                     If None, then there is no limit on the wait time.
-        """
-        keys: range = range(1, self.__max_num_keys)
-        search_task_fn: Callable = self.search_task
-        length_reduction_hash: int = self.__length_reduction_hash
-        db_root: str = self.__db_root
-        class_model: T = self.__class_model
-        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
-            for key in keys:
-                future = executor.submit(
-                    search_task_fn,
-                    key,
-                    filter_fn,
-                    length_reduction_hash,
-                    db_root,
-                    class_model,
-                )
-                doc = future.result(timeout)
-                if doc is not None:
-                    return doc
-        return None
-
-    def find(
-        self,
-        filter_fn: Callable,
-        db_query_docs_limit: int = 1000,
-        max_workers: int | None = None,
-        timeout: float | None = None,
-    ) -> list[T] | None:
-        """Find one or more documents matching the filter.
-
-        The search is based on the effect of a quantum loop.
-        The search effectiveness depends on the number of processor threads.
-        Ideally, hundreds and even thousands of threads are required.
-
-        Args:
-            filter_fn: A function that execute the conditions of filtering.
-            db_query_docs_limit: Limiting the number of request results. By default = 1000.
-            max_workers: The maximum number of processes that can be used to
-                         execute the given calls. If None or not given then as many
-                         worker processes will be created as the machine has processors.
-            timeout: The number of seconds to wait for the result if the future isn't done.
-                     If None, then there is no limit on the wait time.
-        """
-        keys: range = range(1, self.__max_num_keys)
-        search_task_fn: Callable = self.search_task
-        length_reduction_hash: int = self.__length_reduction_hash
-        db_root: str = self.__db_root
-        class_model: T = self.__class_model
-        counter: int = 0
-        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
-            results = []
-            for key in keys:
-                if counter == db_query_docs_limit:
-                    break
-                future = executor.submit(
-                    search_task_fn,
-                    key,
-                    filter_fn,
-                    length_reduction_hash,
-                    db_root,
-                    class_model,
-                )
-                doc = future.result(timeout)
-                if doc is not None:
-                    results.append(doc)
-                    counter += 1
-        return results or None
-
-    def collection_name(self) -> str:
-        """Get collection name."""
-        return self.__class_model.__name__
-
-    def collection_full_name(self) -> str:
-        """Get full name of collection."""
-        return f"{self.__db_root}/{self.__class_model.__name__}"
+"""Creation and management of the database."""
+
+from __future__ import annotations
+
+__all__ = ("Scruby",)
+
+import concurrent.futures
+import contextlib
+import logging
+import zlib
+from collections.abc import Callable
+from pathlib import Path as SyncPath
+from shutil import rmtree
+from typing import Any, Literal, Never, TypeVar, assert_never
+
+import orjson
+from anyio import Path, to_thread
+from pydantic import BaseModel
+
+from scruby import constants
+from scruby.errors import (
+    KeyAlreadyExistsError,
+    KeyNotExistsError,
+)
+
+logger = logging.getLogger(__name__)
+
+T = TypeVar("T")
+
+
+class _Meta(BaseModel):
+    """Metadata of Collection."""
+
+    counter_documents: int
+
+
+class Scruby[T]:
+    """Creation and management of database.
+
+    Args:
+        class_model: Class of Model (Pydantic).
+    """
+
+    def __init__(  # noqa: D107
+        self,
+        class_model: T,
+    ) -> None:
+        self.__meta = _Meta
+        self.__class_model = class_model
+        self.__db_root = constants.DB_ROOT
+        self.__hash_reduce_left = constants.HASH_REDUCE_LEFT
+        # The maximum number of branches.
+        match self.__hash_reduce_left:
+            case 0:
+                self.__max_branch_number = 4294967296
+            case 2:
+                self.__max_branch_number = 16777216
+            case 4:
+                self.__max_branch_number = 65536
+            case 6:
+                self.__max_branch_number = 256
+            case _ as unreachable:
+                msg: str = f"{unreachable} - Unacceptable value for HASH_REDUCE_LEFT."
+                logger.critical(msg)
+                assert_never(Never(unreachable))
+        # Caching a pati for metadata in the form of a tuple.
+        # The zero branch is reserved for metadata.
+        branch_number: int = 0
+        branch_number_as_hash: str = f"{branch_number:08x}"[constants.HASH_REDUCE_LEFT :]
+        separated_hash: str = "/".join(list(branch_number_as_hash))
+        self.__meta_path_tuple = (
+            constants.DB_ROOT,
+            class_model.__name__,
+            separated_hash,
+            "meta.json",
+        )
+        # Create metadata for collection, if required.
+        branch_path = SyncPath(
+            *(
+                self.__db_root,
+                self.__class_model.__name__,
+                separated_hash,
+            ),
+        )
+        if not branch_path.exists():
+            branch_path.mkdir(parents=True)
+            meta = _Meta(
+                counter_documents=0,
+            )
+            meta_json = meta.model_dump_json()
+            meta_path = SyncPath(*(branch_path, "meta.json"))
+            meta_path.write_text(meta_json, "utf-8")
+
+    async def _get_meta(self) -> _Meta:
+        """Asynchronous method for getting metadata of collection.
+
+        This method is for internal use.
+
+        Returns:
+            Metadata object.
+        """
+        meta_path = Path(*self.__meta_path_tuple)
+        meta_json = await meta_path.read_text()
+        meta: _Meta = self.__meta.model_validate_json(meta_json)
+        return meta
+
+    async def _set_meta(self, meta: _Meta) -> None:
+        """Asynchronous method for updating metadata of collection.
+
+        This method is for internal use.
+
+        Returns:
+            None.
+        """
+        meta_json = meta.model_dump_json()
+        meta_path = Path(*self.__meta_path_tuple)
+        await meta_path.write_text(meta_json, "utf-8")
+
+    async def _counter_documents(self, step: Literal[1, -1]) -> None:
+        """Asynchronous method for management of documents in metadata of collection.
+
+        This method is for internal use.
+
+        Returns:
+            None.
+        """
+        meta_path = Path(*self.__meta_path_tuple)
+        meta_json = await meta_path.read_text("utf-8")
+        meta: _Meta = self.__meta.model_validate_json(meta_json)
+        meta.counter_documents += step
+        meta_json = meta.model_dump_json()
+        await meta_path.write_text(meta_json, "utf-8")
+
+    def _sync_counter_documents(self, number: int) -> None:
+        """Management of documents in metadata of collection.
+
+        This method is for internal use.
+        """
+        meta_path = SyncPath(*self.__meta_path_tuple)
+        meta_json = meta_path.read_text("utf-8")
+        meta: _Meta = self.__meta.model_validate_json(meta_json)
+        meta.counter_documents += number
+        meta_json = meta.model_dump_json()
+        meta_path.write_text(meta_json, "utf-8")
+
+    async def _get_leaf_path(self, key: str) -> Path:
+        """Asynchronous method for getting path to collection cell by key.
+
+        This method is for internal use.
+
+        Args:
+            key: Key name.
+
+        Returns:
+            Path to cell of collection.
+        """
+        if not isinstance(key, str):
+            logger.error("The key is not a type of `str`.")
+            raise KeyError("The key is not a type of `str`.")
+        if len(key) == 0:
+            logger.error("The key should not be empty.")
+            raise KeyError("The key should not be empty.")
+        # Key to crc32 sum.
+        key_as_hash: str = f"{zlib.crc32(key.encode('utf-8')):08x}"[self.__hash_reduce_left :]
+        # Convert crc32 sum in the segment of path.
+        separated_hash: str = "/".join(list(key_as_hash))
+        # The path of the branch to the database.
+        branch_path: Path = Path(
+            *(
+                self.__db_root,
+                self.__class_model.__name__,
+                separated_hash,
+            ),
+        )
+        # If the branch does not exist, need to create it.
+        if not await branch_path.exists():
+            await branch_path.mkdir(parents=True)
+        # The path to the database cell.
+        leaf_path: Path = Path(*(branch_path, "leaf.json"))
+        return leaf_path
+
+    async def add_key(
+        self,
+        key: str,
+        value: T,
+    ) -> None:
+        """Asynchronous method for adding key to collection.
+
+        Args:
+            key: Key name. Type `str`.
+            value: Value of key. Type `BaseModel`.
+
+        Returns:
+            None.
+        """
+        # The path to cell of collection.
+        leaf_path: Path = await self._get_leaf_path(key)
+        value_json: str = value.model_dump_json()
+        # Write key-value to collection.
+        if await leaf_path.exists():
+            # Add new key.
+            data_json: bytes = await leaf_path.read_bytes()
+            data: dict = orjson.loads(data_json) or {}
+            try:
+                data[key]
+            except KeyError:
+                data[key] = value_json
+                await leaf_path.write_bytes(orjson.dumps(data))
+            else:
+                err = KeyAlreadyExistsError()
+                logger.error(err.message)
+                raise err
+        else:
+            # Add new key to a blank leaf.
+            await leaf_path.write_bytes(orjson.dumps({key: value_json}))
+        await self._counter_documents(1)
+
+    async def update_key(
+        self,
+        key: str,
+        value: T,
+    ) -> None:
+        """Asynchronous method for updating key to collection.
+
+        Args:
+            key: Key name. Type `str`.
+            value: Value of key. Type `BaseModel`.
+
+        Returns:
+            None.
+        """
+        # The path to cell of collection.
+        leaf_path: Path = await self._get_leaf_path(key)
+        value_json: str = value.model_dump_json()
+        # Update the existing key.
+        if await leaf_path.exists():
+            # Update the existing key.
+            data_json: bytes = await leaf_path.read_bytes()
+            data: dict = orjson.loads(data_json) or {}
+            try:
+                data[key]
+                data[key] = value_json
+                await leaf_path.write_bytes(orjson.dumps(data))
+            except KeyError:
+                err = KeyNotExistsError()
+                logger.error(err.message)
+                raise err from None
+        else:
+            logger.error("The key not exists.")
+            raise KeyError()
+
+    async def get_key(self, key: str) -> T:
+        """Asynchronous method for getting value of key from collection.
+
+        Args:
+            key: Key name.
+
+        Returns:
+            Value of key or KeyError.
+        """
+        # The path to the database cell.
+        leaf_path: Path = await self._get_leaf_path(key)
+        # Get value of key.
+        if await leaf_path.exists():
+            data_json: bytes = await leaf_path.read_bytes()
+            data: dict = orjson.loads(data_json) or {}
+            obj: T = self.__class_model.model_validate_json(data[key])
+            return obj
+        msg: str = "`get_key` - The unacceptable key value."
+        logger.error(msg)
+        raise KeyError()
+
+    async def has_key(self, key: str) -> bool:
+        """Asynchronous method for checking presence of key in collection.
+
+        Args:
+            key: Key name.
+
+        Returns:
+            True, if the key is present.
+        """
+        # Get path to cell of collection.
+        leaf_path: Path = await self._get_leaf_path(key)
+        # Checking whether there is a key.
+        if await leaf_path.exists():
+            data_json: bytes = await leaf_path.read_bytes()
+            data: dict = orjson.loads(data_json) or {}
+            try:
+                data[key]
+                return True
+            except KeyError:
+                return False
+        return False
+
+    async def delete_key(self, key: str) -> None:
+        """Asynchronous method for deleting key from collection.
+
+        Args:
+            key: Key name.
+
+        Returns:
+            None.
+        """
+        # The path to the database cell.
+        leaf_path: Path = await self._get_leaf_path(key)
+        # Deleting key.
+        if await leaf_path.exists():
+            data_json: bytes = await leaf_path.read_bytes()
+            data: dict = orjson.loads(data_json) or {}
+            del data[key]
+            await leaf_path.write_bytes(orjson.dumps(data))
+            await self._counter_documents(-1)
+            return
+        msg: str = "`delete_key` - The unacceptable key value."
+        logger.error(msg)
+        raise KeyError()
+
+    @staticmethod
+    async def napalm() -> None:
+        """Asynchronous method for full database deletion.
+
+        The main purpose is tests.
+
+        Warning:
+            - `Be careful, this will remove all keys.`
+
+        Returns:
+            None.
+        """
+        with contextlib.suppress(FileNotFoundError):
+            await to_thread.run_sync(rmtree, constants.DB_ROOT)
+        return
+
+    @staticmethod
+    def _task_find(
+        branch_number: int,
+        filter_fn: Callable,
+        hash_reduce_left: str,
+        db_root: str,
+        class_model: T,
+    ) -> list[T] | None:
+        """Task for find documents.
+
+        This method is for internal use.
+
+        Returns:
+            List of documents or None.
+        """
+        branch_number_as_hash: str = f"{branch_number:08x}"[hash_reduce_left:]
+        separated_hash: str = "/".join(list(branch_number_as_hash))
+        leaf_path: SyncPath = SyncPath(
+            *(
+                db_root,
+                class_model.__name__,
+                separated_hash,
+                "leaf.json",
+            ),
+        )
+        docs: list[T] = []
+        if leaf_path.exists():
+            data_json: bytes = leaf_path.read_bytes()
+            data: dict[str, str] = orjson.loads(data_json) or {}
+            for _, val in data.items():
+                doc = class_model.model_validate_json(val)
+                if filter_fn(doc):
+                    docs.append(doc)
+        return docs or None
+
+    def find_one(
+        self,
+        filter_fn: Callable,
+        max_workers: int | None = None,
+        timeout: float | None = None,
+    ) -> T | None:
+        """Finds a single document matching the filter.
+
+        The search is based on the effect of a quantum loop.
+        The search effectiveness depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            filter_fn: A function that execute the conditions of filtering.
+            max_workers: The maximum number of processes that can be used to
+                         execute the given calls. If None or not given then as many
+                         worker processes will be created as the machine has processors.
+            timeout: The number of seconds to wait for the result if the future isn't done.
+                     If None, then there is no limit on the wait time.
+
+        Returns:
+            Document or None.
+        """
+        branch_numbers: range = range(1, self.__max_branch_number)
+        search_task_fn: Callable = self._task_find
+        hash_reduce_left: int = self.__hash_reduce_left
+        db_root: str = self.__db_root
+        class_model: T = self.__class_model
+        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
+            for branch_number in branch_numbers:
+                future = executor.submit(
+                    search_task_fn,
+                    branch_number,
+                    filter_fn,
+                    hash_reduce_left,
+                    db_root,
+                    class_model,
+                )
+                docs = future.result(timeout)
+                if docs is not None:
+                    return docs[0]
+        return None
+
+    def find_many(
+        self,
+        filter_fn: Callable,
+        limit_docs: int = 1000,
+        max_workers: int | None = None,
+        timeout: float | None = None,
+    ) -> list[T] | None:
+        """Finds one or more documents matching the filter.
+
+        The search is based on the effect of a quantum loop.
+        The search effectiveness depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            filter_fn: A function that execute the conditions of filtering.
+            limit_docs: Limiting the number of documents. By default = 1000.
+            max_workers: The maximum number of processes that can be used to
+                         execute the given calls. If None or not given then as many
+                         worker processes will be created as the machine has processors.
+            timeout: The number of seconds to wait for the result if the future isn't done.
+                     If None, then there is no limit on the wait time.
+
+        Returns:
+            List of documents or None.
+        """
+        branch_numbers: range = range(1, self.__max_branch_number)
+        search_task_fn: Callable = self._task_find
+        hash_reduce_left: int = self.__hash_reduce_left
+        db_root: str = self.__db_root
+        class_model: T = self.__class_model
+        counter: int = 0
+        result: list[T] = []
+        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
+            for branch_number in branch_numbers:
+                if counter >= limit_docs:
+                    return result[:limit_docs]
+                future = executor.submit(
+                    search_task_fn,
+                    branch_number,
+                    filter_fn,
+                    hash_reduce_left,
+                    db_root,
+                    class_model,
+                )
+                docs = future.result(timeout)
+                if docs is not None:
+                    for doc in docs:
+                        if counter >= limit_docs:
+                            return result[:limit_docs]
+                        result.append(doc)
+                        counter += 1
+        return result or None
+
+    def collection_name(self) -> str:
+        """Get collection name.
+
+        Returns:
+            Collection name.
+        """
+        return self.__class_model.__name__
+
+    def collection_full_name(self) -> str:
+        """Get full name of collection.
+
+        Returns:
+            Full name of collection.
+        """
+        return f"{self.__db_root}/{self.__class_model.__name__}"
+
+    async def estimated_document_count(self) -> int:
+        """Get an estimate of the number of documents in this collection using collection metadata.
+
+        Returns:
+            The number of documents.
+        """
+        meta = await self._get_meta()
+        return meta.counter_documents
+
+    def count_documents(
+        self,
+        filter_fn: Callable,
+        max_workers: int | None = None,
+        timeout: float | None = None,
+    ) -> int:
+        """Count the number of documents a matching the filter in this collection.
+
+        The search is based on the effect of a quantum loop.
+        The search effectiveness depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            filter_fn: A function that execute the conditions of filtering.
+            max_workers: The maximum number of processes that can be used to
+                         execute the given calls. If None or not given then as many
+                         worker processes will be created as the machine has processors.
+            timeout: The number of seconds to wait for the result if the future isn't done.
+                     If None, then there is no limit on the wait time.
+
+        Returns:
+            The number of documents.
+        """
+        branch_numbers: range = range(1, self.__max_branch_number)
+        search_task_fn: Callable = self._task_find
+        hash_reduce_left: int = self.__hash_reduce_left
+        db_root: str = self.__db_root
+        class_model: T = self.__class_model
+        counter: int = 0
+        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
+            for branch_number in branch_numbers:
+                future = executor.submit(
+                    search_task_fn,
+                    branch_number,
+                    filter_fn,
+                    hash_reduce_left,
+                    db_root,
+                    class_model,
+                )
+                if future.result(timeout) is not None:
+                    counter += 1
+        return counter
+
+    @staticmethod
+    def _task_delete(
+        branch_number: int,
+        filter_fn: Callable,
+        hash_reduce_left: int,
+        db_root: str,
+        class_model: T,
+    ) -> int:
+        """Task for find and delete documents.
+
+        This method is for internal use.
+
+        Returns:
+            The number of deleted documents.
+        """
+        branch_number_as_hash: str = f"{branch_number:08x}"[hash_reduce_left:]
+        separated_hash: str = "/".join(list(branch_number_as_hash))
+        leaf_path: SyncPath = SyncPath(
+            *(
+                db_root,
+                class_model.__name__,
+                separated_hash,
+                "leaf.json",
+            ),
+        )
+        counter: int = 0
+        if leaf_path.exists():
+            data_json: bytes = leaf_path.read_bytes()
+            data: dict[str, str] = orjson.loads(data_json) or {}
+            new_state: dict[str, str] = {}
+            for key, val in data.items():
+                doc = class_model.model_validate_json(val)
+                if filter_fn(doc):
+                    counter -= 1
+                else:
+                    new_state[key] = val
+            leaf_path.write_bytes(orjson.dumps(new_state))
+        return counter
+
+    def delete_many(
+        self,
+        filter_fn: Callable,
+        max_workers: int | None = None,
+        timeout: float | None = None,
+    ) -> int:
+        """Delete one or more documents matching the filter.
+
+        The search is based on the effect of a quantum loop.
+        The search effectiveness depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            filter_fn: A function that execute the conditions of filtering.
+            max_workers: The maximum number of processes that can be used to
+                         execute the given calls. If None or not given then as many
+                         worker processes will be created as the machine has processors.
+            timeout: The number of seconds to wait for the result if the future isn't done.
+                     If None, then there is no limit on the wait time.
+
+        Returns:
+            The number of deleted documents.
+        """
+        branch_numbers: range = range(1, self.__max_branch_number)
+        search_task_fn: Callable = self._task_delete
+        hash_reduce_left: int = self.__hash_reduce_left
+        db_root: str = self.__db_root
+        class_model: T = self.__class_model
+        counter: int = 0
+        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
+            for branch_number in branch_numbers:
+                future = executor.submit(
+                    search_task_fn,
+                    branch_number,
+                    filter_fn,
+                    hash_reduce_left,
+                    db_root,
+                    class_model,
+                )
+                counter += future.result(timeout)
+        if counter < 0:
+            self._sync_counter_documents(counter)
+        return abs(counter)
+
+    @staticmethod
+    def _task_get_docs(
+        branch_number: int,
+        hash_reduce_left: int,
+        db_root: str,
+        class_model: T,
+    ) -> list[Any]:
+        """Get documents for custom task.
+
+        This method is for internal use.
+
+        Returns:
+            List of documents.
+        """
+        branch_number_as_hash: str = f"{branch_number:08x}"[hash_reduce_left:]
+        separated_hash: str = "/".join(list(branch_number_as_hash))
+        leaf_path: SyncPath = SyncPath(
+            *(
+                db_root,
+                class_model.__name__,
+                separated_hash,
+                "leaf.json",
+            ),
+        )
+        docs: list[str, T] = []
+        if leaf_path.exists():
+            data_json: bytes = leaf_path.read_bytes()
+            data: dict[str, str] = orjson.loads(data_json) or {}
+            for _, val in data.items():
+                docs.append(class_model.model_validate_json(val))
+        return docs
+
+    def run_custom_task(self, custom_task_fn: Callable, limit_docs: int = 1000) -> Any:
+        """Running custom task.
+
+        This method running a task created on the basis of a quantum loop.
+        Effectiveness running task depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            custom_task_fn: A function that execute the custom task.
+            limit_docs: Limiting the number of documents. By default = 1000.
+
+        Returns:
+            The result of a custom task.
+        """
+        kwargs = {
+            "get_docs_fn": self._task_get_docs,
+            "branch_numbers": range(1, self.__max_branch_number),
+            "hash_reduce_left": self.__hash_reduce_left,
+            "db_root": self.__db_root,
+            "class_model": self.__class_model,
+            "limit_docs": limit_docs,
+        }
+        return custom_task_fn(**kwargs)
+
+    @staticmethod
+    def _task_update(
+        branch_number: int,
+        filter_fn: Callable,
+        hash_reduce_left: str,
+        db_root: str,
+        class_model: T,
+        new_data: dict[str, Any],
+    ) -> int:
+        """Task for find documents.
+
+        This method is for internal use.
+
+        Returns:
+            The number of updated documents.
+        """
+        branch_number_as_hash: str = f"{branch_number:08x}"[hash_reduce_left:]
+        separated_hash: str = "/".join(list(branch_number_as_hash))
+        leaf_path: SyncPath = SyncPath(
+            *(
+                db_root,
+                class_model.__name__,
+                separated_hash,
+                "leaf.json",
+            ),
+        )
+        counter: int = 0
+        if leaf_path.exists():
+            data_json: bytes = leaf_path.read_bytes()
+            data: dict[str, str] = orjson.loads(data_json) or {}
+            new_state: dict[str, str] = {}
+            for _, val in data.items():
+                doc = class_model.model_validate_json(val)
+                if filter_fn(doc):
+                    for key, value in new_data.items():
+                        doc.__dict__[key] = value
+                        new_state[key] = doc.model_dump_json()
+                    counter += 1
+            leaf_path.write_bytes(orjson.dumps(new_state))
+        return counter
+
+    def update_many(
+        self,
+        filter_fn: Callable,
+        new_data: dict[str, Any],
+        max_workers: int | None = None,
+        timeout: float | None = None,
+    ) -> int:
+        """Updates one or more documents matching the filter.
+
+        The search is based on the effect of a quantum loop.
+        The search effectiveness depends on the number of processor threads.
+        Ideally, hundreds and even thousands of threads are required.
+
+        Args:
+            filter_fn: A function that execute the conditions of filtering.
+            new_data: New data for the fields that need to be updated.
+            max_workers: The maximum number of processes that can be used to
+                         execute the given calls. If None or not given then as many
+                         worker processes will be created as the machine has processors.
+            timeout: The number of seconds to wait for the result if the future isn't done.
+                     If None, then there is no limit on the wait time.
+
+        Returns:
+            The number of updated documents.
+        """
+        branch_numbers: range = range(1, self.__max_branch_number)
+        update_task_fn: Callable = self._task_update
+        hash_reduce_left: int = self.__hash_reduce_left
+        db_root: str = self.__db_root
+        class_model: T = self.__class_model
+        counter: int = 0
+        with concurrent.futures.ThreadPoolExecutor(max_workers) as executor:
+            for branch_number in branch_numbers:
+                future = executor.submit(
+                    update_task_fn,
+                    branch_number,
+                    filter_fn,
+                    hash_reduce_left,
+                    db_root,
+                    class_model,
+                    new_data,
+                )
+                counter += future.result(timeout)
+        return counter