data-sourcerer 0.1.0__py3-none-any.whl

sourcerer/infrastructure/file_system/services.py

@@ -0,0 +1,147 @@
+"""
+File system service implementation.
+
+This module provides a concrete implementation of the BaseFileSystemService
+interface for interacting with the local file system.
+"""
+
+from pathlib import Path
+
+from sourcerer.domain.file_system.entities import ListDirOutput
+from sourcerer.domain.file_system.services import BaseFileSystemService
+from sourcerer.infrastructure.file_system.exceptions import (
+    ReadFileException,
+    ListDirException,
+)
+from sourcerer.infrastructure.utils import custom_sort_key
+
+
+class FileSystemService(BaseFileSystemService):
+    """
+    Implementation of the file system service for local file operations.
+
+    This class provides methods for interacting with the local file system,
+    implementing the BaseFileSystemService interface. It handles operations
+    like reading, writing, listing, and searching files within a specified
+    working directory.
+    """
+
+    ACCESS_DENIED_ERROR = "Access denied: Path outside work directory"
+    FILE_NOT_FOUND_ERROR = "File does not exist"
+    MOVING_FILE_ERROR = "Moving file error"
+
+    def __init__(self, work_dir: Path | str):
+        """
+        Initialize a FileSystemService instance with the specified working directory.
+
+        Args:
+            work_dir (Path or str): The path to the working directory. If a string is provided,
+                it will be converted to a Path object.
+        """
+        self.work_dir = work_dir if isinstance(work_dir, Path) else Path(work_dir)
+
+    def read(self, path: Path) -> str:
+        """
+        Reads and processes the contents of a file located at the specified path.
+
+        This method will attempt to open the file, read its content, and perform
+        any necessary processing on the data. It assumes that the file is in
+        a format appropriate for the specific application logic and that the
+        given path is accessible.
+
+        Args:
+            path (Path): The path to the file to read.
+
+        Returns:
+            str: The processed data extracted from the file.
+
+        Raises:
+            ReadFileException: An error occurred during the file read operation.
+        """
+
+        if not isinstance(path, Path):
+            raise ReadFileException("file_path must be a Path object")
+        if not path.exists():
+            raise ReadFileException(f"File not found: {path}")
+        self.validate_path_within_work_dir(path, ReadFileException)
+        try:
+            return path.read_text()
+        except Exception as e:
+            raise ReadFileException(f"Error reading file: {e}") from e
+
+    def list_dir(
+        self,
+        path: Path,
+        relative_paths: bool = False,
+        recursively=False,
+        max_items: int | None = None,
+    ) -> ListDirOutput:
+        """
+        List all files and directories within the specified directory.
+
+        Args:
+            path (Path): The path to the directory.
+            relative_paths (bool): Whether to return relative paths or full paths.
+            recursively (bool): Whether iterate recursively over the content.
+            max_items (int | None): Maximum number of items to return.
+
+        Returns:
+             ListDirOutput: A data class containing the list of files and directories within the specified path.
+                - files (list[Path]): Sorted list of file paths
+                - directories (list[Path]): Sorted list of directory paths
+
+        Raises:
+            ListDirException: If the path is invalid, directory doesn't exist, or path is not a directory.
+
+        """
+        if not isinstance(path, Path):
+            raise ListDirException("Path must be a Path object")
+        if not path.exists():
+            raise ListDirException(f"Directory not found: {path}")
+        if not path.is_dir():
+            raise ListDirException(f"Path is not a directory: {path}")
+        try:
+            path.relative_to(self.work_dir)
+        except ValueError as e:
+            raise ListDirException("Access denied: Path outside work directory") from e
+        try:
+            files = []
+            directories = []
+
+            entries = path.rglob("*") if recursively else path.iterdir()
+            items = 0
+            for entry in entries:
+                if max_items and items > max_items:
+                    raise ListDirException(
+                        f"Too many items, max processable dir size: {max_items}"
+                    )
+                items += 1
+                target = files if entry.is_file() else directories
+                target.append(entry.relative_to(path) if relative_paths else entry)
+            return ListDirOutput(
+                files=sorted(files, key=custom_sort_key),
+                directories=sorted(directories, key=custom_sort_key),
+            )
+        except PermissionError as e:
+            raise ListDirException(f"Access denied: {e}") from e
+        except Exception as e:
+            raise ListDirException(f"Error listing directory: {e}") from e
+
+    def validate_path_within_work_dir(self, path: Path, exception_class: type) -> None:
+        """
+        Validates if a path is within the working directory.
+
+        This method checks if the given path is contained within the working directory
+        to prevent unauthorized access to files outside the designated work area.
+
+        Args:
+            path (Path): The path to validate
+            exception_class (type): The exception class to raise if validation fails
+
+        Raises:
+            exception_class: If the path is outside the working directory
+        """
+        try:
+            path.relative_to(self.work_dir)
+        except ValueError as e:
+            raise exception_class(self.ACCESS_DENIED_ERROR) from e

sourcerer/infrastructure/storage_provider/__init__.py

@@ -0,0 +1,7 @@
+"""
+Storage provider infrastructure module.
+
+This package provides concrete implementations of the storage provider
+service interfaces defined in the domain layer, along with
+infrastructure-specific exceptions and utilities.
+"""

sourcerer/infrastructure/storage_provider/exceptions.py

@@ -0,0 +1,78 @@
+"""
+Storage provider exception classes.
+
+This module defines exception classes for handling errors that occur
+during interactions with cloud storage providers.
+"""
+
+from sourcerer.domain.storage_provider.exceptions import BaseStorageProviderException
+
+
+class CredentialsNotFoundException(BaseStorageProviderException):
+    """
+    Exception raised when credentials are not found for a cloud storage provider.
+
+    This exception is a specific case of BaseStorageProviderException
+    and should be used to indicate that the required credentials for
+    accessing a cloud storage service are missing or not configured.
+    """
+
+
+class ListStoragesException(BaseStorageProviderException):
+    """
+    Exception raised for errors that occur during the listing of cloud storage services.
+
+    This exception is a specific case of BaseStorageProviderException
+    and is used to indicate issues encountered when attempting to retrieve
+    a list of available storage services from a cloud provider.
+    """
+
+
+class StoragePermissionException(BaseStorageProviderException):
+    """
+    Exception raised for permission-related errors in cloud storage operations.
+
+    This exception is triggered when there is an issue with permissions
+    while accessing or modifying resources in a cloud storage service.
+    """
+
+
+class ListStorageItemsException(BaseStorageProviderException):
+    """
+    Exception raised when an error occurs while listing items in cloud storage.
+
+    This exception is a specific case of BaseStorageProviderException
+    and is used to indicate issues encountered during the retrieval of
+    storage items from a cloud storage provider.
+    """
+
+
+class ReadStorageItemsException(BaseStorageProviderException):
+    """
+    Exception raised for errors encountered while reading items from cloud storage.
+
+    This exception is a specific case of BaseStorageProviderException
+    and should be used to indicate issues that occur during the retrieval
+    of data from cloud storage services.
+    """
+
+
+class DeleteStorageItemsException(BaseStorageProviderException):
+    """
+    Exception raised for errors that occur during the deletion of storage items.
+
+    This exception is a specific type of BaseStorageProviderException
+    and is used to indicate issues encountered when attempting to delete
+    items from a cloud storage service.
+    """
+
+
+class UploadStorageItemsException(BaseStorageProviderException):
+    """
+    Exception raised for errors that occur during the upload of items
+    to cloud storage.
+
+    This exception is a specific case of BaseStorageProviderException
+    and is used to signal issues encountered while uploading data to
+    cloud storage services.
+    """

sourcerer/infrastructure/storage_provider/registry.py

@@ -0,0 +1,84 @@
+"""
+Registry for storage provider service implementations.
+
+This module provides a registry system for managing different types of
+storage provider services.
+"""
+
+import functools
+from typing import Type, Dict
+
+from sourcerer.domain.storage_provider.services import BaseStorageProviderService
+from sourcerer.infrastructure.utils import Singleton
+
+
+class StorageProviderRegistry(metaclass=Singleton):
+    """
+    Registry for storage provider service implementations.
+
+    This singleton class maintains a registry of storage provider service
+    implementations organized by provider identifier.
+    """
+
+    def __init__(self, *args, **kwargs):
+        """Initialize the registry with an empty dictionary."""
+        super().__init__(*args, **kwargs)
+        self._items_: Dict[str, type] = {}
+
+    def register(self, storage_provider: str, cls: Type[BaseStorageProviderService]):
+        """
+        Register a storage provider service implementation.
+
+        Args:
+            storage_provider (str): The provider identifier
+            cls (Type[BaseStorageProviderService]): The service class to register
+        """
+        self._items_[storage_provider] = cls
+
+    def get(self):
+        """
+        Get all registered storage provider services.
+
+        Returns:
+            dict: Dictionary of all registered storage provider services
+        """
+        return self._items_
+
+    def get_by_provider(self, provider):
+        """
+        Get a storage provider service by provider identifier.
+
+        Args:
+            provider (str): The provider identifier
+
+        Returns:
+            Type[BaseStorageProviderService]: The storage provider service class
+        """
+        return self._items_.get(provider)
+
+
+def storage_provider(provider: str):
+    """
+    Decorator for registering storage provider service implementations.
+
+    Args:
+        provider (str): The provider identifier
+
+    Returns:
+        callable: Decorator function
+    """
+
+    def wrapper(cls):
+        storage_provider_registry.register(provider, cls)
+
+        @functools.wraps(cls)
+        def wrapped(*args, **kwargs):
+            return cls(*args, **kwargs)
+
+        return wrapped
+
+    return wrapper
+
+
+# Singleton registry instance
+storage_provider_registry = StorageProviderRegistry()