digitalhub 0.8.0b0__py3-none-any.whl → 0.8.0b1__py3-none-any.whl

digitalhub/context/context.py

@@ -0,0 +1,136 @@
+from __future__ import annotations
+
+import typing
+from pathlib import Path
+
+if typing.TYPE_CHECKING:
+    from digitalhub.entities.project.entity._base import Project
+
+
+class Context:
+    """
+    Context class built forom a `Project` instance. It contains
+    some information about the project, such as the project name,
+    a client instance (local or non-local), the local context
+    project path and information about client locality.
+    It exposes CRUD operations for the entities and act as a layer
+    between the project object and its client.
+    """
+
+    def __init__(self, project: Project) -> None:
+        self.name = project.name
+        self.client = project._client
+        self.local = project._client.is_local()
+        self.root = Path(project.spec.context)
+
+    def create_object(self, api: str, obj: dict, **kwargs) -> dict:
+        """
+        Create an object.
+
+        Parameters
+        ----------
+        api : str
+            Create API.
+        obj : dict
+            Object to create.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            Response object.
+        """
+        return self.client.create_object(api, obj, **kwargs)
+
+    def read_object(self, api: str, **kwargs) -> dict:
+        """
+        Read an object.
+
+        Parameters
+        ----------
+        api : str
+            Read API.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            Response object.
+        """
+        return self.client.read_object(api, **kwargs)
+
+    def update_object(self, api: str, obj: dict, **kwargs) -> dict:
+        """
+        Update an object.
+
+        Parameters
+        ----------
+        api : str
+            Update API.
+        obj : dict
+            Object to update.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            Response object.
+        """
+        return self.client.update_object(api, obj, **kwargs)
+
+    def delete_object(self, api: str, **kwargs) -> dict:
+        """
+        Delete an object.
+
+        Parameters
+        ----------
+        api : str
+            Delete API.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            Response object.
+        """
+        return self.client.delete_object(api, **kwargs)
+
+    def list_objects(self, api: str, **kwargs) -> dict:
+        """
+        List objects.
+
+        Parameters
+        ----------
+        api : str
+            The api to list the objects with.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            The list of objects.
+        """
+        return self.client.list_objects(api, **kwargs)
+
+    def list_first_object(self, api: str, **kwargs) -> dict:
+        """
+        List first objects.
+
+        Parameters
+        ----------
+        api : str
+            The api to list the objects with.
+        **kwargs : dict
+            Keyword arguments passed to the request.
+
+        Returns
+        -------
+        dict
+            The list of objects.
+        """
+        return self.client.list_first_object(api, **kwargs)

digitalhub/datastores/builder.py

@@ -0,0 +1,134 @@
+from __future__ import annotations
+
+import typing
+
+from digitalhub.datastores.objects.local import LocalDatastore
+from digitalhub.datastores.objects.remote import RemoteDatastore
+from digitalhub.datastores.objects.s3 import S3Datastore
+from digitalhub.datastores.objects.sql import SqlDatastore
+from digitalhub.stores.builder import get_default_store, get_store
+from digitalhub.utils.uri_utils import map_uri_scheme
+
+if typing.TYPE_CHECKING:
+    from digitalhub.datastores.objects.base import Datastore
+    from digitalhub.stores.objects.base import Store
+
+
+REGISTRY_DATASTORES = {"local": LocalDatastore, "remote": RemoteDatastore, "s3": S3Datastore, "sql": SqlDatastore}
+
+
+class DatastoreBuilder:
+    """
+    Datastore builder class.
+    """
+
+    def __init__(self) -> None:
+        self._instances: dict[str, Datastore] = {}
+        self._default: Datastore | None = None
+        self._def_scheme = "s3"
+
+    def build(self, uri: str) -> None:
+        """
+        Build a datastore instance and register it.
+        It overrides any existing instance.
+
+        Parameters
+        ----------
+        uri : str
+            URI to parse.
+
+        Returns
+        -------
+        None
+        """
+        store = get_store(uri)
+        self._instances[store.type] = self.build_datastore(store)
+
+    def get(self, uri: str) -> Datastore:
+        """
+        Get a store instance by URI.
+
+        Parameters
+        ----------
+        uri : str
+            URI to parse.
+
+        Returns
+        -------
+        Datastore
+            The datastore instance.
+        """
+        scheme = map_uri_scheme(uri)
+        if scheme not in self._instances:
+            self.build(uri)
+        return self._instances[scheme]
+
+    def default(self) -> Datastore:
+        """
+        Get the default store instance.
+
+        Returns
+        -------
+        Datastore
+            The default datastore instance.
+        """
+        if self._default is None:
+            store = get_default_store()
+            self._default = self.build_datastore(store)
+        return self._default
+
+    def build_datastore(self, store: Store) -> Datastore:
+        """
+        Build a datastore instance.
+
+        Parameters
+        ----------
+        store : Store
+            Store instance.
+
+        Returns
+        -------
+        Datastore
+            The datastore instance.
+
+        Raises
+        ------
+        NotImplementedError
+            If the store type is not implemented.
+        """
+        try:
+            return REGISTRY_DATASTORES[store.type](store)
+        except KeyError as e:
+            raise NotImplementedError from e
+
+
+builder = DatastoreBuilder()
+
+
+def get_datastore(uri: str) -> Datastore:
+    """
+    Get a datastore instance by URI.
+
+    Parameters
+    ----------
+    uri : str
+        URI to parse.
+
+    Returns
+    -------
+    Datastore
+        The datastore instance.
+    """
+    return builder.get(uri)
+
+
+def get_default_datastore() -> Datastore:
+    """
+    Get the default datastore instance.
+
+    Returns
+    -------
+    Datastore
+        The default datastore instance.
+    """
+    return builder.default()

digitalhub/datastores/objects/base.py

@@ -0,0 +1,85 @@
+from __future__ import annotations
+
+import typing
+from abc import ABCMeta, abstractmethod
+from typing import Any
+
+from digitalhub.readers.builder import get_reader_by_engine
+
+if typing.TYPE_CHECKING:
+    from digitalhub.stores.objects.base import Store
+
+
+class Datastore(metaclass=ABCMeta):
+    """
+    Datastore abstract class.
+    """
+
+    def __init__(self, store: Store, **kwargs) -> None:
+        self.store = store
+
+    ##############################
+    # IO methods
+    ##############################
+
+    @abstractmethod
+    def write_df(self, df: Any, dst: str, extension: str | None = None, **kwargs) -> str:
+        """
+        Write DataFrame as parquet or csv.
+        """
+
+    def read_df(
+        self,
+        path: str | list[str],
+        extension: str,
+        engine: str | None = "pandas",
+        **kwargs,
+    ) -> Any:
+        """
+        Read DataFrame from path.
+
+        Parameters
+        ----------
+        path : str | list[str]
+            Path(s) to read DataFrame from.
+        extension : str
+            Extension of the file.
+        engine : str
+            Dataframe engine (pandas, polars, etc.).
+        **kwargs : dict
+            Keyword arguments.
+
+        Returns
+        -------
+        Any
+            DataFrame.
+        """
+        reader = get_reader_by_engine(engine)
+        self._validate_extension(extension)
+        return reader.read_df(path, extension, **kwargs)
+
+    ##############################
+    # Helper methods
+    ##############################
+
+    @staticmethod
+    def _validate_extension(extension: str) -> None:
+        """
+        Validate extension.
+
+        Parameters
+        ----------
+        extension : str
+            Extension of the file.
+
+        Returns
+        -------
+        None
+
+        Raises
+        ------
+        ValueError
+            If extension is not supported.
+        """
+        if extension not in ["csv", "parquet", "file"]:
+            raise ValueError(f"Extension {extension} not supported.")

digitalhub/datastores/objects/local.py

@@ -0,0 +1,42 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+from digitalhub.datastores.objects.base import Datastore
+from digitalhub.readers.builder import get_reader_by_object
+
+
+class LocalDatastore(Datastore):
+    """
+    Local Datastore class.
+    """
+
+    def write_df(self, df: Any, dst: str, extension: str | None = None, **kwargs) -> str:
+        """
+        Method to write a dataframe to a file. Kwargs are passed to df.to_parquet().
+        If destination is not provided, the dataframe is written to the default
+        store path with generated name.
+
+        Parameters
+        ----------
+        df : Any
+            The dataframe to write.
+        dst : str
+            The destination of the dataframe.
+        **kwargs : dict
+            Keyword arguments.
+
+        Returns
+        -------
+        str
+            Path of written dataframe.
+        """
+        self.store._check_local_dst(dst)
+        self._validate_extension(Path(dst).suffix.removeprefix("."))
+
+        # Write dataframe
+        reader = get_reader_by_object(df)
+        reader.write_df(df, dst, extension=extension, **kwargs)
+
+        return dst

digitalhub/datastores/objects/remote.py

@@ -0,0 +1,23 @@
+from __future__ import annotations
+
+from typing import Any
+
+from digitalhub.datastores.objects.base import Datastore
+
+
+class RemoteDatastore(Datastore):
+    """
+    Remote Datastore class.
+    """
+
+    def write_df(self, df: Any, dst: str, extension: str | None = None, **kwargs) -> str:
+        """
+        Method to write a dataframe to a file. Note that this method is not implemented
+        since the remote store is not meant to write dataframes.
+
+        Raises
+        ------
+        NotImplementedError
+            This method is not implemented.
+        """
+        raise NotImplementedError("Remote store does not support write_df.")

digitalhub/datastores/objects/s3.py

@@ -0,0 +1,38 @@
+from __future__ import annotations
+
+from io import BytesIO
+from typing import Any
+
+from digitalhub.datastores.objects.base import Datastore
+from digitalhub.readers.builder import get_reader_by_object
+
+
+class S3Datastore(Datastore):
+    """
+    S3 Datastore class.
+    """
+
+    def write_df(self, df: Any, dst: str, extension: str | None = None, **kwargs) -> str:
+        """
+        Write a dataframe to S3 based storage. Kwargs are passed to df.to_parquet().
+
+        Parameters
+        ----------
+        df : Any
+            The dataframe.
+        dst : str
+            The destination path on S3 based storage.
+        **kwargs : dict
+            Keyword arguments.
+
+        Returns
+        -------
+        str
+            The S3 path where the dataframe was saved.
+        """
+        fileobj = BytesIO()
+        reader = get_reader_by_object(df)
+        reader.write_df(df, fileobj, extension=extension, **kwargs)
+
+        key = self.store._get_key(dst)
+        return self.store.upload_fileobject(fileobj, key)

digitalhub/datastores/objects/sql.py

@@ -0,0 +1,60 @@
+from __future__ import annotations
+
+from typing import Any
+
+from digitalhub.datastores.objects.base import Datastore
+from digitalhub.readers.builder import get_reader_by_object
+
+
+class SqlDatastore(Datastore):
+    """
+    Sql Datastore class.
+    """
+
+    def write_df(self, df: Any, dst: str, extension: str | None = None, **kwargs) -> str:
+        """
+        Write a dataframe to a database. Kwargs are passed to df.to_sql().
+
+        Parameters
+        ----------
+        df : Any
+            The dataframe to write.
+        dst : str
+            The destination of the dataframe.
+        **kwargs : dict
+            Keyword arguments.
+
+        Returns
+        -------
+        str
+            Path of written dataframe.
+        """
+        schema = self.store._get_schema(dst)
+        table = self.store._get_table_name(dst)
+        return self._upload_table(df, schema, table, **kwargs)
+
+    def _upload_table(self, df: Any, schema: str, table: str, **kwargs) -> str:
+        """
+        Upload a table to SQL based storage.
+
+        Parameters
+        ----------
+        df : DataFrame
+            The dataframe.
+        schema : str
+            Destination schema.
+        table : str
+            Destination table.
+        **kwargs : dict
+            Keyword arguments.
+
+        Returns
+        -------
+        str
+            The SQL URI where the dataframe was saved.
+        """
+        reader = get_reader_by_object(df)
+        engine = self.store._check_factory()
+        reader.write_table(df, table, engine, schema, **kwargs)
+        engine.dispose()
+        return f"sql://{engine.url.database}/{schema}/{table}"