digitalhub 0.9.1__py3-none-any.whl → 0.10.0b0__py3-none-any.whl

digitalhub/entities/_operations/processor.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import typing
+from typing import Any
 
 from digitalhub.client.api import get_client
 from digitalhub.context.api import delete_context, get_context
@@ -713,46 +714,8 @@ class OperationsProcessor:
             entity_id=entity_id,
             **kwargs,
         )
-        return build_entity_from_dict(obj)
-
-    def read_material_entity(
-        self,
-        identifier: str,
-        entity_type: str | None = None,
-        project: str | None = None,
-        entity_id: str | None = None,
-        **kwargs,
-    ) -> MaterialEntity:
-        """
-        Read object from backend.
-
-        Parameters
-        ----------
-        identifier : str
-            Entity key (store://...) or entity name.
-        entity_type : str
-            Entity type.
-        project : str
-            Project name.
-        entity_id : str
-            Entity ID.
-        **kwargs : dict
-            Parameters to pass to the API call.
-
-        Returns
-        -------
-        MaterialEntity
-            Object instance.
-        """
-        obj: MaterialEntity = self.read_context_entity(
-            identifier,
-            entity_type=entity_type,
-            project=project,
-            entity_id=entity_id,
-            **kwargs,
-        )
-        obj._get_files_info()
-        return obj
+        entity = build_entity_from_dict(obj)
+        return self._post_process_get(entity)
 
     def read_unversioned_entity(
         self,
@@ -1000,42 +963,6 @@ class OperationsProcessor:
             List of object instances.
         """
         context = self._get_context_from_identifier(identifier, project)
-        obj = self._read_context_entity_versions(
-            context,
-            identifier,
-            entity_type=entity_type,
-            project=project,
-            **kwargs,
-        )
-        return [build_entity_from_dict(o) for o in obj]
-
-    def read_material_entity_versions(
-        self,
-        identifier: str,
-        entity_type: str | None = None,
-        project: str | None = None,
-        **kwargs,
-    ) -> list[MaterialEntity]:
-        """
-        Read object versions from backend.
-
-        Parameters
-        ----------
-        identifier : str
-            Entity key (store://...) or entity name.
-        entity_type : str
-            Entity type.
-        project : str
-            Project name.
-        **kwargs : dict
-            Parameters to pass to the API call.
-
-        Returns
-        -------
-        list[MaterialEntity]
-            List of object instances.
-        """
-        context = self._get_context_from_identifier(identifier, project)
         objs = self._read_context_entity_versions(
             context,
             identifier,
@@ -1045,8 +972,8 @@ class OperationsProcessor:
         )
         objects = []
         for o in objs:
-            entity: MaterialEntity = build_entity_from_dict(o)
-            entity._get_files_info()
+            entity: ContextEntity = build_entity_from_dict(o)
+            entity = self._post_process_get(entity)
             objects.append(entity)
         return objects
 
@@ -1106,37 +1033,10 @@ class OperationsProcessor:
         """
         context = self._get_context(project)
         objs = self._list_context_entities(context, entity_type, **kwargs)
-        return [build_entity_from_dict(obj) for obj in objs]
-
-    def list_material_entities(
-        self,
-        project: str,
-        entity_type: str,
-        **kwargs,
-    ) -> list[MaterialEntity]:
-        """
-        List all latest version objects from backend.
-
-        Parameters
-        ----------
-        project : str
-            Project name.
-        entity_type : str
-            Entity type.
-        **kwargs : dict
-            Parameters to pass to the API call.
-
-        Returns
-        -------
-        list[MaterialEntity]
-            List of object instances.
-        """
-        context = self._get_context(project)
-        objs = self._list_context_entities(context, entity_type, **kwargs)
         objects = []
         for o in objs:
-            entity: MaterialEntity = build_entity_from_dict(o)
-            entity._get_files_info()
+            entity: ContextEntity = build_entity_from_dict(o)
+            entity = self._post_process_get(entity)
             objects.append(entity)
         return objects
 
@@ -1324,6 +1224,26 @@ class OperationsProcessor:
             **kwargs,
         )
 
+    def _post_process_get(self, entity: ContextEntity) -> ContextEntity:
+        """
+        Post process get (files, metrics).
+
+        Parameters
+        ----------
+        entity : ContextEntity
+            Entity to post process.
+
+        Returns
+        -------
+        ContextEntity
+            Post processed entity.
+        """
+        if hasattr(entity.status, "metrics"):
+            entity._get_metrics()
+        if hasattr(entity.status, "files"):
+            entity._get_files_info()
+        return entity
+
     ##############################
     # Context entity operations
     ##############################
@@ -1644,6 +1564,82 @@ class OperationsProcessor:
         )
         return context.client.update_object(api, entity_list, **kwargs)
 
+    def read_metrics(
+        self,
+        project: str,
+        entity_type: str,
+        entity_id: str,
+        metric_name: str | None = None,
+        **kwargs,
+    ) -> dict:
+        """
+        Get metrics from backend.
+
+        Parameters
+        ----------
+        project : str
+            Project name.
+        entity_type : str
+            Entity type.
+        entity_id : str
+            Entity ID.
+        **kwargs : dict
+            Parameters to pass to the API call.
+
+        Returns
+        -------
+        dict
+            Response from backend.
+        """
+        context = self._get_context(project)
+        api = context.client.build_api(
+            ApiCategories.CONTEXT.value,
+            BackendOperations.METRICS.value,
+            project=context.name,
+            entity_type=entity_type,
+            entity_id=entity_id,
+            metric_name=metric_name,
+        )
+        return context.client.read_object(api, **kwargs)
+
+    def update_metric(
+        self,
+        project: str,
+        entity_type: str,
+        entity_id: str,
+        metric_name: str,
+        metric_value: Any,
+        **kwargs,
+    ) -> None:
+        """
+        Get single metric from backend.
+
+        Parameters
+        ----------
+        project : str
+            Project name.
+        entity_type : str
+            Entity type.
+        entity_id : str
+            Entity ID.
+        **kwargs : dict
+            Parameters to pass to the API call.
+
+        Returns
+        -------
+        None
+        """
+        context = self._get_context(project)
+        api = context.client.build_api(
+            ApiCategories.CONTEXT.value,
+            BackendOperations.METRICS.value,
+            project=context.name,
+            entity_type=entity_type,
+            entity_id=entity_id,
+            metric_name=metric_name,
+        )
+        context.client.update_object(api, metric_value, **kwargs)
+
     def _search(
         self,
         project: str,

digitalhub/entities/artifact/crud.py

@@ -158,7 +158,7 @@ def get_artifact(
     >>>                    project="my-project",
     >>>                    entity_id="my-artifact-id")
     """
-    return processor.read_material_entity(
+    return processor.read_context_entity(
         identifier=identifier,
         entity_type=ENTITY_TYPE,
         project=project,
@@ -198,7 +198,7 @@ def get_artifact_versions(
     >>> obj = get_artifact_versions("my-artifact-name"
     >>>                             project="my-project")
     """
-    return processor.read_material_entity_versions(
+    return processor.read_context_entity_versions(
         identifier=identifier,
         entity_type=ENTITY_TYPE,
         project=project,
@@ -226,7 +226,7 @@ def list_artifacts(project: str, **kwargs) -> list[Artifact]:
     --------
     >>> objs = list_artifacts(project="my-project")
     """
-    return processor.list_material_entities(
+    return processor.list_context_entities(
         project=project,
         entity_type=ENTITY_TYPE,
         **kwargs,

digitalhub/entities/dataitem/_base/entity.py

@@ -5,7 +5,7 @@ from pathlib import Path
 
 from digitalhub.entities._base.material.entity import MaterialEntity
 from digitalhub.entities._commons.enums import EntityTypes
-from digitalhub.readers._commons.enums import Extensions
+from digitalhub.entities.dataitem.utils import DEFAULT_EXTENSION
 from digitalhub.utils.exceptions import EntityError
 from digitalhub.utils.uri_utils import has_sql_scheme
 
@@ -67,7 +67,7 @@ class Dataitem(MaterialEntity):
             return file_format
 
         if has_sql_scheme(path):
-            return Extensions.PARQUET.value
+            return DEFAULT_EXTENSION
 
         ext = Path(path).suffix[1:]
         if ext is not None:

digitalhub/entities/dataitem/crud.py

@@ -165,7 +165,7 @@ def get_dataitem(
     >>>                    project="my-project",
     >>>                    entity_id="my-dataitem-id")
     """
-    return processor.read_material_entity(
+    return processor.read_context_entity(
         identifier=identifier,
         entity_type=ENTITY_TYPE,
         project=project,
@@ -205,7 +205,7 @@ def get_dataitem_versions(
     >>> objs = get_dataitem_versions("my-dataitem-name",
     >>>                              project="my-project")
     """
-    return processor.read_material_entity_versions(
+    return processor.read_context_entity_versions(
         identifier=identifier,
         entity_type=ENTITY_TYPE,
         project=project,
@@ -233,7 +233,7 @@ def list_dataitems(project: str, **kwargs) -> list[Dataitem]:
     --------
     >>> objs = list_dataitems(project="my-project")
     """
-    return processor.list_material_entities(
+    return processor.list_context_entities(
         project=project,
         entity_type=ENTITY_TYPE,
         **kwargs,

digitalhub/entities/dataitem/table/entity.py

@@ -85,7 +85,7 @@ class DataitemTable(Dataitem):
                 checker = data_path
 
             extension = self._get_extension(checker, file_format)
-            return get_store("").read_df(data_path, extension, engine, **kwargs)
+            return get_store(self.project, "").read_df(data_path, extension, engine, **kwargs)
 
         except Exception as e:
             raise e
@@ -119,7 +119,7 @@ class DataitemTable(Dataitem):
         str
             Path to the written dataframe.
         """
-        return get_store(self.spec.path).write_df(df, self.spec.path, extension=extension, **kwargs)
+        return get_store(self.project, self.spec.path).write_df(df, self.spec.path, extension=extension, **kwargs)
 
     @staticmethod
     def _clean_tmp_path(pth: Path | None, clean: bool) -> None:

digitalhub/{utils/data_utils.py → entities/dataitem/table/utils.py}

@@ -1,49 +1,23 @@
 from __future__ import annotations
 
-import json
+from digitalhub.utils.generic_utils import dump_json
 
 
-def build_data_preview(preview: list[dict] | None = None, rows_count: int | None = None) -> dict:
+def prepare_data(data: list[list], columnar: bool = False) -> list[list]:
     """
-    Build data preview.
+    Prepare data.
 
     Parameters
     ----------
-    preview : list[dict] | None
-        Preview.
-    rows_count : int | None
-        Row count.
-
-    Returns
-    -------
-    dict
-        Data preview.
-    """
-    dict_ = {}
-    if preview is not None:
-        dict_["cols"] = preview
-    if rows_count is not None:
-        dict_["rows"] = rows_count
-    return dict_
-
-
-def get_data_preview(columns: list, data: list[list], columnar: bool = False) -> list[dict]:
-    """
-    Prepare preview.
-
-    Parameters
-    ----------
-    columns : list
-        Columns names.
-    data : list[list]
-        Data to preview.
-    columnar : bool
+    data : list
+        Data.
+    columnar : bool | None
         If data are arranged in columns. If False, data are arranged in rows.
 
     Returns
     -------
-    list[dict]
-        Data preview.
+    list[list]
+        Prepared data.
     """
     # Reduce data to 10 rows
     if not columnar:
@@ -56,17 +30,10 @@ def get_data_preview(columns: list, data: list[list], columnar: bool = False) ->
     if not columnar:
         data = list(map(list, list(zip(*data))))
 
-    # Prepare the preview
-    data_dict = prepare_preview(columns, data)
-
-    # Filter memoryview values
-    filtered_memview = filter_memoryview(data_dict)
-
-    # Check the size of the preview data
-    return check_preview_size(filtered_memview)
+    return data
 
 
-def prepare_preview(column_names: list, data: list[list]) -> list[dict]:
+def prepare_preview(columns: list, data: list[list]) -> list[dict]:
     """
     Get preview.
 
@@ -80,9 +47,10 @@ def prepare_preview(column_names: list, data: list[list]) -> list[dict]:
     list[dict]
         Preview.
     """
-    if len(column_names) != len(data):
+    if len(columns) != len(data):
         raise ValueError("Column names and data must have the same length")
-    return [{"name": column, "value": values} for column, values in zip(column_names, data)]
+    preview = [{"name": column, "value": values} for column, values in zip(columns, data)]
+    return filter_memoryview(preview)
 
 
 def filter_memoryview(data: list[dict]) -> list[dict]:
@@ -91,13 +59,13 @@ def filter_memoryview(data: list[dict]) -> list[dict]:
 
     Parameters
     ----------
-    data : pd.DataFrame
+    data : list[dict]
         Data.
 
     Returns
     -------
-    list[str]
-        Column to filter out from preview.
+    list[dict]
+        Preview.
     """
     key_to_filter = []
     for i in data:
@@ -108,13 +76,13 @@ def filter_memoryview(data: list[dict]) -> list[dict]:
     return data
 
 
-def check_preview_size(preview: list[dict]) -> list:
+def check_preview_size(preview: dict) -> list:
     """
     Check preview size. If it's too big, return empty list.
 
     Parameters
     ----------
-    preview : list[dict]
+    preview : dict
         Preview.
 
     Returns
@@ -122,6 +90,30 @@ def check_preview_size(preview: list[dict]) -> list:
     list
         Preview.
     """
-    if len(json.dumps(preview).encode("utf-8")) >= 64000:
+    if len(dump_json(preview)) >= 64000:
         return []
     return preview
+
+
+def finalize_preview(preview: list[dict] | None = None, rows_count: int | None = None) -> dict:
+    """
+    Finalize preview.
+
+    Parameters
+    ----------
+    preview : list[dict] | None
+        Preview.
+    rows_count : int | None
+        Row count.
+
+    Returns
+    -------
+    dict
+        Data preview.
+    """
+    data = {}
+    if preview is not None:
+        data["cols"] = preview
+    if rows_count is not None:
+        data["rows"] = rows_count
+    return data

digitalhub/entities/dataitem/utils.py

@@ -8,7 +8,6 @@ from digitalhub.context.api import get_context
 from digitalhub.entities._base.entity._constructors.uuid import build_uuid
 from digitalhub.entities._base.material.utils import build_log_path_from_source, eval_local_source
 from digitalhub.entities._commons.enums import EntityKinds, EntityTypes
-from digitalhub.readers._commons.enums import Extensions
 from digitalhub.readers.api import get_reader_by_object
 from digitalhub.utils.generic_utils import slugify_string
 
@@ -16,6 +15,9 @@ if typing.TYPE_CHECKING:
     from digitalhub.entities.dataitem._base.entity import Dataitem
 
 
+DEFAULT_EXTENSION = "parquet"
+
+
 def eval_source(
     source: str | list[str] | None = None,
     data: Any | None = None,
@@ -39,11 +41,12 @@ def eval_source(
         raise ValueError("You must provide source or data.")
 
     if source is not None:
-        return eval_local_source(source)
+        eval_local_source(source)
+        return source
 
     if kind == EntityKinds.DATAITEM_TABLE.value:
         ctx = get_context(project)
-        pth = ctx.root / f"{slugify_string(name)}.{Extensions.PARQUET.value}"
+        pth = ctx.root / f"{slugify_string(name)}.{DEFAULT_EXTENSION}"
         reader = get_reader_by_object(data)
         reader.write_parquet(data, pth)
         return str(pth)