futurehouse-client 0.5.1__tar.gz → 0.5.3__tar.gz

{futurehouse_client-0.5.1/src/futurehouse_client.egg-info → futurehouse_client-0.5.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: futurehouse-client
-Version: 0.5.1
+Version: 0.5.3
 Summary: A client for interacting with endpoints of the FutureHouse service.
 Author-email: FutureHouse technical staff <hello@futurehouse.org>
 License:                                  Apache License

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/src/futurehouse_client/clients/data_storage_methods.py

@@ -37,6 +37,7 @@ from futurehouse_client.models.data_storage_methods import (
     DirectoryManifest,
     GetDatasetAndEntriesResponse,
     ManifestEntry,
+    RawFileFetchResponse,
 )
 from futurehouse_client.models.rest import (
     DataStorageSearchPayload,
@@ -504,11 +505,15 @@ class DataStorageMethods:
             raise DataStorageError(f"File or directory not found: {file_path}")
         return file_path
 
-    def _build_zip_path(self, name: str, path: str | None) -> str:
+    def _build_zip_path(
+        self, name: str, path_override: str | Path | None
+    ) -> str | Path:
         """Build GCS path for zip file."""
         zip_filename = name if name.endswith(".zip") else f"{name}.zip"
-        if path:
-            return f"{path.rstrip('/')}/{zip_filename}"
+        if path_override:
+            if isinstance(path_override, str):
+                return f"{path_override.rstrip('/')}/{zip_filename}"
+            return path_override / zip_filename
         return zip_filename
 
     # TODO: methods in here need to be moved to fh tools
@@ -663,7 +668,11 @@ class DataStorageMethods:
             raise DataStorageError(f"Failed to download from GCS: {e}") from e
 
     def _prepare_single_file_upload(
-        self, name: str, file_path: Path, description: str | None, path: str | None
+        self,
+        name: str,
+        file_path: Path,
+        description: str | None,
+        file_path_override: str | Path | None,
     ) -> tuple[int, DataStorageRequestPayload | None]:
         """Prepare single file for upload, return file size and payload if text content."""
         file_size = file_path.stat().st_size
@@ -681,7 +690,7 @@ class DataStorageMethods:
                     name=name,
                     description=description,
                     content=text_content,
-                    path=path,
+                    file_path=file_path_override or file_path,
                     is_collection=False,
                 )
             logger.warning(
@@ -777,7 +786,7 @@ class DataStorageMethods:
         name: str,
         dir_path: Path,
         description: str | None,
-        path: str | None = None,
+        dir_path_override: str | Path | None = None,
         ignore_patterns: list[str] | None = None,
         ignore_filename: str = ".gitignore",
         project_id: UUID | None = None,
@@ -789,7 +798,7 @@ class DataStorageMethods:
             name: Name for the directory collection
             dir_path: Path to directory to zip and upload
             description: Description for the collection
-            path: Optional GCS path for the zip file
+            dir_path_override: Optional GCS path for the zip file
             ignore_patterns: List of patterns to ignore when zipping
             ignore_filename: Name of ignore file to read from directory
             project_id: ID of the project this data storage entry belongs to
@@ -807,11 +816,11 @@ class DataStorageMethods:
                 dir_path, temp_zip_path, ignore_patterns, ignore_filename
             )
 
-            zip_gcs_path = self._build_zip_path(name, path)
+            zip_gcs_path = self._build_zip_path(name, dir_path_override)
             payload = DataStorageRequestPayload(
                 name=name,
                 description=description,
-                path=zip_gcs_path,
+                file_path=zip_gcs_path,
                 is_collection=True,
                 project_id=project_id,
                 tags=tags,
@@ -860,7 +869,7 @@ class DataStorageMethods:
         name: str,
         dir_path: Path,
         description: str | None,
-        path: str | None = None,
+        dir_path_override: str | Path | None = None,
         ignore_patterns: list[str] | None = None,
         ignore_filename: str = ".gitignore",
         project_id: UUID | None = None,
@@ -872,7 +881,7 @@ class DataStorageMethods:
             name: Name for the directory collection
             dir_path: Path to directory to zip and upload
             description: Description for the collection
-            path: Optional GCS path for the zip file
+            dir_path_override: Optional GCS path for the zip file
             ignore_patterns: List of patterns to ignore when zipping
             ignore_filename: Name of ignore file to read from directory
             project_id: ID of the project this data storage entry belongs to
@@ -890,11 +899,11 @@ class DataStorageMethods:
                 dir_path, temp_zip_path, ignore_patterns, ignore_filename
             )
 
-            zip_gcs_path = self._build_zip_path(name, path)
+            zip_gcs_path = self._build_zip_path(name, dir_path_override)
             payload = DataStorageRequestPayload(
                 name=name,
                 description=description,
-                path=zip_gcs_path,
+                file_path=zip_gcs_path,
                 is_collection=True,
                 project_id=project_id,
                 tags=tags,
@@ -940,7 +949,7 @@ class DataStorageMethods:
         name: str,
         file_path: Path,
         description: str | None,
-        path: str | None = None,
+        file_path_override: str | Path | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Upload a single file."""
@@ -960,7 +969,7 @@ class DataStorageMethods:
                     name=name,
                     description=description,
                     content=text_content,
-                    path=path,
+                    file_path=file_path_override or file_path,
                     is_collection=False,
                     project_id=project_id,
                 )
@@ -977,7 +986,7 @@ class DataStorageMethods:
         payload = DataStorageRequestPayload(
             name=name,
             description=description,
-            path=path,
+            file_path=file_path_override or file_path,
             is_collection=False,
             project_id=project_id,
         )
@@ -1028,13 +1037,13 @@ class DataStorageMethods:
         name: str,
         file_path: Path,
         description: str | None,
-        path: str | None = None,
+        file_path_override: str | Path | None = None,
         dataset_id: UUID | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Asynchronously upload a single file."""
         file_size, text_payload = self._prepare_single_file_upload(
-            name, file_path, description, path
+            name, file_path, description, file_path_override
         )
 
         if text_payload:
@@ -1048,7 +1057,7 @@ class DataStorageMethods:
         payload = DataStorageRequestPayload(
             name=name,
             description=description,
-            path=path,
+            file_path=file_path_override or file_path,
             is_collection=False,
             dataset_id=dataset_id,
             project_id=project_id,
@@ -1089,14 +1098,14 @@ class DataStorageMethods:
         name: str,
         file_path: Path,
         description: str | None,
-        path: str | None,
+        file_path_override: str | None,
         parent_id: UUID | None,
         dataset_id: UUID | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Upload a single file with a parent ID (sync version)."""
         file_size, text_payload = self._prepare_single_file_upload(
-            name, file_path, description, path
+            name, file_path, description, file_path_override
         )
 
         if text_payload:
@@ -1112,7 +1121,7 @@ class DataStorageMethods:
         payload = DataStorageRequestPayload(
             name=name,
             description=description,
-            path=path,
+            file_path=file_path_override or file_path,
             is_collection=False,
             parent_id=parent_id,
             dataset_id=dataset_id,
@@ -1167,7 +1176,7 @@ class DataStorageMethods:
                 name=item.name,
                 file_path=item,
                 description=file_description,
-                path=None,
+                file_path_override=None,
                 parent_id=current_parent_id,
                 dataset_id=dataset_id,
                 project_id=project_id,
@@ -1310,14 +1319,14 @@ class DataStorageMethods:
         name: str,
         file_path: Path,
         description: str | None,
-        path: str | None,
+        file_path_override: str | None,
         parent_id: UUID | None,
         dataset_id: UUID | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Asynchronously upload a single file with a parent ID."""
         file_size, text_payload = self._prepare_single_file_upload(
-            name, file_path, description, path
+            name, file_path, description, file_path_override
         )
 
         if text_payload:
@@ -1333,7 +1342,7 @@ class DataStorageMethods:
         payload = DataStorageRequestPayload(
             name=name,
             description=description,
-            path=path,
+            file_path=file_path_override or file_path,
             is_collection=False,
             parent_id=parent_id,
             dataset_id=dataset_id,
@@ -1388,7 +1397,7 @@ class DataStorageMethods:
                 name=item.name,
                 file_path=item,
                 description=file_description,
-                path=None,
+                file_path_override=None,
                 parent_id=current_parent_id,
                 dataset_id=dataset_id,
                 project_id=project_id,
@@ -1519,7 +1528,7 @@ class DataStorageMethods:
         name: str,
         content: str,
         description: str | None = None,
-        path: str | None = None,
+        file_path: str | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Store content as a string in the data storage system.
@@ -1528,7 +1537,7 @@ class DataStorageMethods:
             name: Name of the data storage entry
             content: Content to store as a string
             description: Optional description of the data storage entry
-            path: Optional path for the data storage entry
+            file_path: Optional path for the data storage entry
             project_id: ID of the project this data storage entry belongs to
 
         Returns:
@@ -1544,7 +1553,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -1568,7 +1577,7 @@ class DataStorageMethods:
                 name=name,
                 content=content,
                 description=description,
-                path=path,
+                file_path=file_path,
                 project_id=project_id,
             )
             return self._create_data_storage_entry(payload)
@@ -1590,7 +1599,7 @@ class DataStorageMethods:
         name: str,
         content: str,
         description: str | None = None,
-        path: str | None = None,
+        file_path: str | None = None,
         dataset_id: UUID | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
@@ -1600,7 +1609,7 @@ class DataStorageMethods:
             name: Name of the data storage entry
             content: Content to store as a string
             description: Optional description of the data storage entry
-            path: Optional path for the data storage entry
+            file_path: Optional path for the data storage entry
             dataset_id: Optional dataset ID to add entry to, or None to create new dataset
             project_id: ID of the project this data storage entry belongs to
 
@@ -1617,7 +1626,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -1641,7 +1650,7 @@ class DataStorageMethods:
                 name=name,
                 content=content,
                 description=description,
-                path=path,
+                file_path=file_path,
                 dataset_id=dataset_id,
                 project_id=project_id,
             )
@@ -1788,7 +1797,7 @@ class DataStorageMethods:
         name: str,
         file_path: str | Path,
         description: str | None = None,
-        path: str | None = None,
+        file_path_override: str | Path | None = None,
         as_collection: bool = False,
         manifest_filename: str | None = None,
         ignore_patterns: list[str] | None = None,
@@ -1808,7 +1817,7 @@ class DataStorageMethods:
             name: Name of the data storage entry
             file_path: Path to file or directory to upload
             description: Optional description of the data storage entry
-            path: Optional path for the data storage entry
+            file_path_override: Optional path for the data storage entry
             as_collection: If true, upload directories as a single zip file collection.
             manifest_filename: Name of manifest file (JSON or YAML) containing:
                 - entries - Map of file/directory names to their manifest entries
@@ -1834,7 +1843,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -1861,7 +1870,7 @@ class DataStorageMethods:
                     name,
                     file_path,
                     description,
-                    path,
+                    file_path_override,
                     ignore_patterns,
                     ignore_filename,
                     project_id,
@@ -1883,7 +1892,7 @@ class DataStorageMethods:
                     )
                 return responses[0]
             return self._upload_data_single_file(
-                name, file_path, description, path, project_id
+                name, file_path, description, file_path_override, project_id
             )
 
         except HTTPStatusError as e:
@@ -1904,7 +1913,7 @@ class DataStorageMethods:
         name: str,
         file_path: str | Path,
         description: str | None = None,
-        path: str | None = None,
+        file_path_override: str | Path | None = None,
         as_collection: bool = False,
         manifest_filename: str | None = None,
         ignore_patterns: list[str] | None = None,
@@ -1918,7 +1927,7 @@ class DataStorageMethods:
             name: Name of the data storage entry.
             file_path: Path to the file or directory to upload.
             description: Optional description for the entry.
-            path: Optional GCS path for the entry.
+            file_path_override: Optional GCS path for the entry.
             as_collection: If uploading a directory, `True` zips it into a single collection,
                            `False` uploads it as a hierarchical structure of individual objects.
             manifest_filename: Optional manifest file (JSON or YAML) for hierarchical uploads containing:
@@ -1945,7 +1954,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -1972,7 +1981,7 @@ class DataStorageMethods:
                         name,
                         file_path,
                         description,
-                        path,
+                        file_path_override,
                         ignore_patterns,
                         ignore_filename,
                         project_id,
@@ -1993,7 +2002,7 @@ class DataStorageMethods:
                     )
                 return responses[0]
             return await self._aupload_data_single_file(
-                name, file_path, description, path, dataset_id, project_id
+                name, file_path, description, file_path_override, dataset_id, project_id
             )
 
         except HTTPStatusError as e:
@@ -2015,7 +2024,6 @@ class DataStorageMethods:
         existing_location: DataStorageLocationPayload,
         description: str | None = None,
         as_collection: bool = False,
-        path: str | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Store content as a string in the data storage system.
@@ -2032,7 +2040,7 @@ class DataStorageMethods:
             as_collection: If uploading a directory, `True` creates a single storage entry for
                 the whole directory and multiple storage locations for each file, `False` assumes
                 you are uploading a single file.
-            path: Optional path for the data storage entry
+            file_path: Optional path for the data storage entry
             project_id: ID of the project this data storage entry belongs to
 
         Returns:
@@ -2048,7 +2056,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -2071,7 +2079,6 @@ class DataStorageMethods:
             payload = DataStorageRequestPayload(
                 name=name,
                 description=description,
-                path=path,
                 existing_location=existing_location,
                 project_id=project_id,
                 is_collection=as_collection,
@@ -2101,7 +2108,6 @@ class DataStorageMethods:
         existing_location: DataStorageLocationPayload,
         as_collection: bool = False,
         description: str | None = None,
-        path: str | None = None,
         project_id: UUID | None = None,
     ) -> DataStorageResponse:
         """Store content as a string in the data storage system.
@@ -2118,7 +2124,7 @@ class DataStorageMethods:
             as_collection: If uploading a directory, `True` creates a single storage entry for
                 the whole directory and multiple storage locations for each file, `False` assumes
                 you are uploading a single file.
-            path: Optional path for the data storage entry
+            file_path: Optional path for the data storage entry
             project_id: ID of the project this data storage entry belongs to
 
         Returns:
@@ -2134,7 +2140,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created
@@ -2157,7 +2163,6 @@ class DataStorageMethods:
             payload = DataStorageRequestPayload(
                 name=name,
                 description=description,
-                path=path,
                 existing_location=existing_location,
                 project_id=project_id,
                 is_collection=as_collection,
@@ -2464,7 +2469,7 @@ class DataStorageMethods:
     def fetch_data_from_storage(
         self,
         data_storage_id: UUID | None = None,
-    ) -> str | Path | list[Path] | None:
+    ) -> RawFileFetchResponse | str | Path | list[Path] | None:
         """Fetch data from the storage system (sync version).
 
         Args:
@@ -2527,6 +2532,13 @@ class DataStorageMethods:
                         f"No content found for data storage entry {data_storage_id}"
                     )
                     return None
+
+                if result.data_storage.file_path:
+                    return RawFileFetchResponse(
+                        filename=Path(result.data_storage.file_path),
+                        file_content=content,
+                    )
+
                 return content
 
             raise DataStorageRetrievalError(f"Unsupported storage type: {storage_type}")
@@ -2547,7 +2559,7 @@ class DataStorageMethods:
     async def afetch_data_from_storage(
         self,
         data_storage_id: UUID | None = None,
-    ) -> str | Path | list[Path] | None:
+    ) -> RawFileFetchResponse | str | Path | list[Path] | None:
         """Fetch data from the storage system.
 
         Args:
@@ -2613,6 +2625,13 @@ class DataStorageMethods:
                         f"No content found for data storage entry {data_storage_id}"
                     )
                     return None
+
+                if result.data_storage.file_path:
+                    return RawFileFetchResponse(
+                        filename=Path(result.data_storage.file_path),
+                        file_content=content,
+                    )
+
                 return content
 
             raise DataStorageRetrievalError(f"Unsupported storage type: {storage_type}")
@@ -2798,7 +2817,7 @@ class DataStorageMethods:
                     - parent_id - ID of the parent entry for hierarchical storage
                     - project_id - ID of the project this entry belongs to
                     - dataset_id - ID of the dataset this entry belongs to
-                    - path - Path in the storage system where this entry is located
+                    - file_path - Filepath in the storage system where this entry is located
                     - bigquery_schema - Target BigQuery schema for the entry
                     - user_id - ID of the user who created this entry
                     - created_at - Timestamp when the entry was created

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/src/futurehouse_client/clients/rest_client.py

@@ -46,6 +46,7 @@ from futurehouse_client.clients.data_storage_methods import DataStorageMethods
 from futurehouse_client.models.app import (
     AuthType,
     JobDeploymentConfig,
+    LiteTaskResponse,
     Stage,
     TaskRequest,
     TaskResponse,
@@ -530,8 +531,12 @@ class RestClient(DataStorageMethods):
         before_sleep=before_sleep_log(logger, logging.WARNING),
     )
     def get_task(
-        self, task_id: str | None = None, history: bool = False, verbose: bool = False
-    ) -> "TaskResponse":
+        self,
+        task_id: str | None = None,
+        history: bool = False,
+        verbose: bool = False,
+        lite: bool = False,
+    ) -> "TaskResponse | LiteTaskResponse":
         """Get details for a specific task."""
         task_id = task_id or self.trajectory_id
         url = f"/v0.1/trajectories/{task_id}"
@@ -547,7 +552,9 @@ class RestClient(DataStorageMethods):
                     "job_id": task_id,
                 },
             ),
-            self.client.stream("GET", url, params={"history": history}) as response,
+            self.client.stream(
+                "GET", url, params={"history": history, "lite": lite}
+            ) as response,
         ):
             if response.status_code in {401, 403}:
                 raise PermissionError(
@@ -558,6 +565,10 @@ class RestClient(DataStorageMethods):
             data = json.loads(json_data)
             if "id" not in data:
                 data["id"] = task_id
+
+            if lite:
+                return LiteTaskResponse(**data)
+
             verbose_response = TaskResponseVerbose(**data)
 
         if verbose:
@@ -571,8 +582,12 @@ class RestClient(DataStorageMethods):
         before_sleep=before_sleep_log(logger, logging.WARNING),
     )
     async def aget_task(
-        self, task_id: str | None = None, history: bool = False, verbose: bool = False
-    ) -> "TaskResponse":
+        self,
+        task_id: str | None = None,
+        history: bool = False,
+        verbose: bool = False,
+        lite: bool = False,
+    ) -> "TaskResponse | LiteTaskResponse":
         """Get details for a specific task asynchronously."""
         task_id = task_id or self.trajectory_id
         url = f"/v0.1/trajectories/{task_id}"
@@ -588,7 +603,7 @@ class RestClient(DataStorageMethods):
             },
         ):
             async with self.async_client.stream(
-                "GET", url, params={"history": history}
+                "GET", url, params={"history": history, "lite": lite}
             ) as response:
                 if response.status_code in {401, 403}:
                     raise PermissionError(
@@ -599,6 +614,10 @@ class RestClient(DataStorageMethods):
                 data = json.loads(json_data)
                 if "id" not in data:
                     data["id"] = task_id
+
+                if lite:
+                    return LiteTaskResponse(**data)
+
                 verbose_response = TaskResponseVerbose(**data)
 
         if verbose:
@@ -735,7 +754,9 @@ class RestClient(DataStorageMethods):
         progress_bar: bool = False,
         concurrency: int = 10,
         timeout: int = DEFAULT_AGENT_TIMEOUT,
-    ) -> list[TaskResponse]:
+    ) -> list[
+        LiteTaskResponse | TaskResponse
+    ]:  # return will always be lite because  we always call with lite=True
         all_tasks: Collection[TaskRequest | dict[str, Any]] = (
             cast(Collection[TaskRequest | dict[str, Any]], [task_data])
             if (isinstance(task_data, dict) or not isinstance(task_data, Collection))
@@ -749,7 +770,7 @@ class RestClient(DataStorageMethods):
         )
 
         start_time = time.monotonic()
-        completed_tasks: dict[str, TaskResponse] = {}
+        completed_tasks: dict[str, LiteTaskResponse | TaskResponse] = {}
 
         if progress_bar:
             progress = tqdm(
@@ -760,7 +781,7 @@ class RestClient(DataStorageMethods):
             task_results = await gather_with_concurrency(
                 concurrency,
                 [
-                    self.aget_task(task_id, verbose=verbose)
+                    self.aget_task(task_id, verbose=verbose, lite=True)
                     for task_id in trajectory_ids
                     if task_id not in completed_tasks
                 ],
@@ -807,7 +828,9 @@ class RestClient(DataStorageMethods):
         verbose: bool = False,
         progress_bar: bool = False,
         timeout: int = DEFAULT_AGENT_TIMEOUT,
-    ) -> list[TaskResponse]:
+    ) -> list[
+        LiteTaskResponse | TaskResponse
+    ]:  # return will always be lite because  we always call with lite=True
         """Run multiple tasks and wait for them to complete.
 
         Args:
@@ -828,7 +851,7 @@ class RestClient(DataStorageMethods):
         trajectory_ids = [self.create_task(task) for task in all_tasks]
 
         start_time = time.monotonic()
-        completed_tasks: dict[str, TaskResponse] = {}
+        completed_tasks: dict[str, LiteTaskResponse | TaskResponse] = {}
 
         if progress_bar:
             progress = sync_tqdm(
@@ -842,7 +865,7 @@ class RestClient(DataStorageMethods):
                 if task_id in completed_tasks:
                     continue
 
-                task = self.get_task(task_id, verbose=verbose)
+                task = self.get_task(task_id, verbose=verbose, lite=True)
 
                 if not ExecutionStatus(task.status).is_terminal_state():
                     all_done = False

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/src/futurehouse_client/models/app.py

@@ -749,6 +749,22 @@ class SimpleOrganization(BaseModel):
     display_name: str
 
 
+class LiteTaskResponse(BaseModel):
+    task_id: UUID = Field(description="Identifier for a trajectory")
+    query: str = Field(description="Query executed for the trajectory")
+    status: str = Field(description="Current status of the trajectory")
+
+    @model_validator(mode="before")
+    @classmethod
+    def validate_fields(cls, original_data: Mapping[str, Any]) -> Mapping[str, Any]:
+        data = copy.deepcopy(original_data)  # Avoid mutating the original data
+        if not isinstance(data, dict):
+            return data
+        data["query"] = data.get("task", data.get("query"))
+        data["task_id"] = cast(UUID, data.get("id", data.get("task_id")))
+        return data
+
+
 class TaskResponse(BaseModel):
     """Base class for task responses. This holds attributes shared over all futurehouse jobs."""
 

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/src/futurehouse_client/models/data_storage_methods.py

@@ -56,7 +56,7 @@ class DataStorageEntry(BaseModel):
         default=None,
         description="ID of the dataset this entry belongs to",
     )
-    path: str | None = Field(
+    file_path: str | None = Field(
         default=None,
         description="Path in the storage system where this entry is located, if a file.",
     )
@@ -158,9 +158,9 @@ class DataStorageRequestPayload(BaseModel):
         default=None,
         description="ID of existing dataset to add entry to, or None to create new dataset",
     )
-    path: PathLike | str | None = Field(
+    file_path: PathLike | str | None = Field(
         default=None,
-        description="Path to store in the GCS bucket.",
+        description="Filepath to store in the GCS bucket.",
     )
     existing_location: DataStorageLocationPayload | None = Field(
         default=None, description="Target storage metadata"
@@ -385,5 +385,12 @@ class DirectoryUploadConfig(BaseModel):
         )
 
 
+class RawFileFetchResponse(BaseModel):
+    """Response model for fetching a raw file."""
+
+    filename: Path = Field(description="Name of the file")
+    file_content: str = Field(description="Content of the file")
+
+
 # Forward reference resolution for DirectoryManifest
 DirectoryManifest.model_rebuild()

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/src/futurehouse_client/version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.5.1'
-__version_tuple__ = version_tuple = (0, 5, 1)
+__version__ = version = '0.5.3'
+__version_tuple__ = version_tuple = (0, 5, 3)
 
-__commit_id__ = commit_id = 'g0c21b6efd'
+__commit_id__ = commit_id = 'gaaaa80cba'

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3/src/futurehouse_client.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: futurehouse-client
-Version: 0.5.1
+Version: 0.5.3
 Summary: A client for interacting with endpoints of the FutureHouse service.
 Author-email: FutureHouse technical staff <hello@futurehouse.org>
 License:                                  Apache License

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/tests/test_data_storage_e2e.py

@@ -17,6 +17,7 @@ from futurehouse_client.models.data_storage_methods import (
     DataContentType,
     DataStorageLocationPayload,
     DataStorageType,
+    RawFileFetchResponse,
 )
 from futurehouse_client.models.rest import (
     FilterLogic,
@@ -90,9 +91,10 @@ async def test_store_raw_content_async(admin_client: RestClient):
 
 @pytest.mark.timeout(300)
 def test_store_file_content_sync(admin_client: RestClient):
+    file_path = Path("packages/futurehouse-client/tests/test_data/test_file.txt")
     response = admin_client.store_file_content(
         name=f"E2E test entry file: {uuid4()}",
-        file_path=Path("packages/futurehouse-client/tests/test_data/test_file.txt"),
+        file_path=file_path,
     )
 
     assert response is not None
@@ -102,8 +104,37 @@ def test_store_file_content_sync(admin_client: RestClient):
     fetch_response = admin_client.fetch_data_from_storage(response.data_storage.id)
 
     assert fetch_response is not None
+    assert isinstance(fetch_response, RawFileFetchResponse)
+    assert fetch_response.filename == file_path
     assert (
-        fetch_response
+        fetch_response.file_content
+        == "Here is some random text that shall immortalize Eddie's brain in code.\n"
+    )
+
+    admin_client.delete_data_storage_entry(response.data_storage.id)
+
+
+@pytest.mark.timeout(300)
+def test_store_file_content_with_path_override_sync(admin_client: RestClient):
+    file_path = Path("packages/futurehouse-client/tests/test_data/test_file.txt")
+    file_path_override = Path("test_file_override.txt")
+    response = admin_client.store_file_content(
+        name=f"E2E test entry file: {uuid4()}",
+        file_path=file_path,
+        file_path_override=file_path_override,
+    )
+
+    assert response is not None
+    assert response.data_storage.id is not None
+    assert len(response.storage_locations) > 0
+
+    fetch_response = admin_client.fetch_data_from_storage(response.data_storage.id)
+
+    assert fetch_response is not None
+    assert isinstance(fetch_response, RawFileFetchResponse)
+    assert fetch_response.filename == file_path_override
+    assert (
+        fetch_response.file_content
         == "Here is some random text that shall immortalize Eddie's brain in code.\n"
     )
 
@@ -133,9 +164,40 @@ def test_store_dir_content_sync(admin_client: RestClient):
 @pytest.mark.timeout(300)
 @pytest.mark.asyncio
 async def test_store_file_content_async(admin_client: RestClient):
+    file_path = Path("packages/futurehouse-client/tests/test_data/test_file.txt")
+    response = await admin_client.astore_file_content(
+        name=f"E2E test entry file: {uuid4()}",
+        file_path=file_path,
+    )
+
+    assert response is not None
+    assert response.data_storage.id is not None
+    assert len(response.storage_locations) > 0
+
+    fetch_response = await admin_client.afetch_data_from_storage(
+        response.data_storage.id
+    )
+
+    assert fetch_response is not None
+    assert isinstance(fetch_response, RawFileFetchResponse)
+    assert fetch_response.filename == file_path
+    assert (
+        fetch_response.file_content
+        == "Here is some random text that shall immortalize Eddie's brain in code.\n"
+    )
+
+    await admin_client.adelete_data_storage_entry(response.data_storage.id)
+
+
+@pytest.mark.timeout(300)
+@pytest.mark.asyncio
+async def test_store_file_content_with_path_override_async(admin_client: RestClient):
+    file_path = Path("packages/futurehouse-client/tests/test_data/test_file.txt")
+    file_path_override = Path("test_file_override.txt")
     response = await admin_client.astore_file_content(
         name=f"E2E test entry file: {uuid4()}",
-        file_path=Path("packages/futurehouse-client/tests/test_data/test_file.txt"),
+        file_path=file_path,
+        file_path_override=file_path_override,
     )
 
     assert response is not None
@@ -147,8 +209,10 @@ async def test_store_file_content_async(admin_client: RestClient):
     )
 
     assert fetch_response is not None
+    assert isinstance(fetch_response, RawFileFetchResponse)
+    assert fetch_response.filename == file_path_override
     assert (
-        fetch_response
+        fetch_response.file_content
         == "Here is some random text that shall immortalize Eddie's brain in code.\n"
     )
 

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/tests/test_data_storage_methods.py

@@ -486,7 +486,7 @@ class TestDataStorageMethods:
                 json={
                     "name": "test_dir",
                     "description": "Test directory",
-                    "path": "test_dir.zip",  # The actual path includes .zip extension
+                    "file_path": "test_dir.zip",  # The actual path includes .zip extension
                     "is_collection": True,
                 },
             )
@@ -533,7 +533,7 @@ class TestDataStorageMethods:
                 json={
                     "name": "test_dir",
                     "description": "Test directory",
-                    "path": "test_dir.zip",  # The actual path includes .zip extension
+                    "file_path": "test_dir.zip",  # The actual path includes .zip extension
                     "is_collection": True,
                 },
             )

{futurehouse_client-0.5.1 → futurehouse_client-0.5.3}/tests/test_rest.py

@@ -39,6 +39,7 @@ from futurehouse_client.clients.rest_client import (
     UserAgentRequestFetchError,
 )
 from futurehouse_client.models.app import (
+    LiteTaskResponse,
     PhoenixTaskResponse,
     PQATaskResponse,
     Stage,
@@ -150,10 +151,12 @@ async def test_job_response(  # noqa: PLR0915
     with subtests.test("Test TaskResponse with queued task"):
         task_response = admin_client.get_task(task_id)
         assert task_response.status in {"queued", "in progress"}
+        assert not isinstance(task_response, LiteTaskResponse)
         assert task_response.job_name == pqa_task_req.name
         assert task_response.query == pqa_task_req.query
         task_response = await admin_client.aget_task(atask_id)
         assert task_response.status in {"queued", "in progress"}
+        assert not isinstance(task_response, LiteTaskResponse)
         assert task_response.job_name == pqa_task_req.name
         assert task_response.query == pqa_task_req.query