fractal-server 2.14.0a36__tar.gz → 2.14.2__tar.gz

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: fractal-server
-Version: 2.14.0a36
+Version: 2.14.2
 Summary: Backend component of the Fractal analytics platform
 License: BSD-3-Clause
 Author: Tommaso Comparin

fractal_server-2.14.2/fractal_server/__init__.py

@@ -0,0 +1 @@
+__VERSION__ = "2.14.2"

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/routes/admin/v2/job.py

@@ -154,7 +154,7 @@ async def update_job(
     if job_update.status != JobStatusTypeV2.FAILED:
         raise HTTPException(
             status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
-            detail=f"Cannot set job status to {job_update.status}",
+            detail=f"Cannot set job status to {job_update.status.value}",
         )
 
     setattr(job, "status", job_update.status)

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/routes/api/v2/__init__.py

@@ -7,6 +7,7 @@ from .dataset import router as dataset_router_v2
 from .history import router as history_router_v2
 from .images import router as images_routes_v2
 from .job import router as job_router_v2
+from .pre_submission_checks import router as pre_submission_checks_router
 from .project import router as project_router_v2
 from .status_legacy import router as status_legacy_router_v2
 from .submit import router as submit_job_router_v2
@@ -15,7 +16,6 @@ from .task_collection import router as task_collection_router_v2
 from .task_collection_custom import router as task_collection_router_v2_custom
 from .task_group import router as task_group_router_v2
 from .task_group_lifecycle import router as task_group_lifecycle_router_v2
-from .verify_image_types import router as verify_image_types_router
 from .workflow import router as workflow_router_v2
 from .workflow_import import router as workflow_import_router_v2
 from .workflowtask import router as workflowtask_router_v2
@@ -26,7 +26,7 @@ from fractal_server.syringe import Inject
 router_api_v2 = APIRouter()
 
 router_api_v2.include_router(dataset_router_v2, tags=["V2 Dataset"])
-router_api_v2.include_router(verify_image_types_router, tags=["V2 Job"])
+router_api_v2.include_router(pre_submission_checks_router, tags=["V2 Job"])
 router_api_v2.include_router(job_router_v2, tags=["V2 Job"])
 router_api_v2.include_router(images_routes_v2, tags=["V2 Images"])
 router_api_v2.include_router(project_router_v2, tags=["V2 Project"])

fractal_server-2.14.2/fractal_server/app/routes/api/v2/pre_submission_checks.py

@@ -0,0 +1,144 @@
+from typing import Optional
+
+from fastapi import APIRouter
+from fastapi import Depends
+from fastapi import status
+from fastapi.responses import JSONResponse
+from pydantic import BaseModel
+from pydantic import Field
+from sqlmodel import select
+
+from ._aux_functions import _get_dataset_check_owner
+from ._aux_functions import _get_workflow_task_check_owner
+from .images import ImageQuery
+from fractal_server.app.db import AsyncSession
+from fractal_server.app.db import get_async_db
+from fractal_server.app.models import UserOAuth
+from fractal_server.app.models.v2 import HistoryImageCache
+from fractal_server.app.models.v2 import HistoryUnit
+from fractal_server.app.routes.auth import current_active_user
+from fractal_server.app.schemas.v2 import HistoryUnitStatus
+from fractal_server.images.models import AttributeFiltersType
+from fractal_server.images.tools import aggregate_types
+from fractal_server.images.tools import filter_image_list
+
+router = APIRouter()
+
+
+@router.post(
+    "/project/{project_id}/dataset/{dataset_id}/images/verify-unique-types/",
+    status_code=status.HTTP_200_OK,
+)
+async def verify_unique_types(
+    project_id: int,
+    dataset_id: int,
+    query: Optional[ImageQuery] = None,
+    user: UserOAuth = Depends(current_active_user),
+    db: AsyncSession = Depends(get_async_db),
+) -> list[str]:
+    # Get dataset
+    output = await _get_dataset_check_owner(
+        project_id=project_id, dataset_id=dataset_id, user_id=user.id, db=db
+    )
+    dataset = output["dataset"]
+
+    # Filter images
+    if query is None:
+        filtered_images = dataset.images
+    else:
+        filtered_images = filter_image_list(
+            images=dataset.images,
+            attribute_filters=query.attribute_filters,
+            type_filters=query.type_filters,
+        )
+
+    # Get actual values for each available type
+    available_types = aggregate_types(filtered_images)
+    values_per_type: dict[str, set] = {
+        _type: set() for _type in available_types
+    }
+    for _img in filtered_images:
+        for _type in available_types:
+            values_per_type[_type].add(_img["types"].get(_type, False))
+
+    # Find types with non-unique value
+    non_unique_types = [
+        key for key, value in values_per_type.items() if len(value) > 1
+    ]
+    non_unique_types = sorted(non_unique_types)
+
+    return non_unique_types
+
+
+class NonProcessedImagesPayload(BaseModel):
+    attribute_filters: AttributeFiltersType = Field(default_factory=dict)
+    type_filters: dict[str, bool] = Field(default_factory=dict)
+
+
+@router.post(
+    "/project/{project_id}/dataset/{dataset_id}/images/non-processed/",
+    status_code=status.HTTP_200_OK,
+)
+async def check_workflowtask(
+    project_id: int,
+    dataset_id: int,
+    workflow_id: int,
+    workflowtask_id: int,
+    filters: NonProcessedImagesPayload,
+    user: UserOAuth = Depends(current_active_user),
+    db: AsyncSession = Depends(get_async_db),
+) -> JSONResponse:
+
+    db_workflow_task, db_workflow = await _get_workflow_task_check_owner(
+        project_id=project_id,
+        workflow_task_id=workflowtask_id,
+        workflow_id=workflow_id,
+        user_id=user.id,
+        db=db,
+    )
+
+    if db_workflow_task.order == 0:
+        # Skip check for first task in the workflow
+        return JSONResponse(status_code=200, content=[])
+
+    previous_wft = db_workflow.task_list[db_workflow_task.order - 1]
+
+    if previous_wft.task.output_types != {}:
+        # Skip check if previous task has non-trivial `output_types`
+        return JSONResponse(status_code=200, content=[])
+    elif previous_wft.task.type in [
+        "converter_compound",
+        "converter_non_parallel",
+    ]:
+        # Skip check if previous task is converter
+        return JSONResponse(status_code=200, content=[])
+
+    res = await _get_dataset_check_owner(
+        project_id=project_id,
+        dataset_id=dataset_id,
+        user_id=user.id,
+        db=db,
+    )
+    dataset = res["dataset"]
+    filtered_images = filter_image_list(
+        images=dataset.images,
+        type_filters=filters.type_filters,
+        attribute_filters=filters.attribute_filters,
+    )
+
+    filtered_zarr_urls = [image["zarr_url"] for image in filtered_images]
+
+    res = await db.execute(
+        select(HistoryImageCache.zarr_url)
+        .join(HistoryUnit)
+        .where(HistoryImageCache.zarr_url.in_(filtered_zarr_urls))
+        .where(HistoryImageCache.dataset_id == dataset_id)
+        .where(HistoryImageCache.workflowtask_id == previous_wft.id)
+        .where(HistoryImageCache.latest_history_unit_id == HistoryUnit.id)
+        .where(HistoryUnit.status == HistoryUnitStatus.DONE)
+    )
+    done_zarr_urls = res.scalars().all()
+
+    missing_zarr_urls = list(set(filtered_zarr_urls) - set(done_zarr_urls))
+
+    return JSONResponse(status_code=200, content=missing_zarr_urls)

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/routes/api/v2/status_legacy.py

@@ -118,8 +118,8 @@ async def get_workflowtask_status(
             )
         except ValueError:
             logger.warning(
-                f"Job {running_job.id} is submitted but its task list does "
-                f"not contain a {WorkflowTaskStatusTypeV2.SUBMITTED} task."
+                f"Job {running_job.id} is submitted but its task list does not"
+                f" contain a {WorkflowTaskStatusTypeV2.SUBMITTED.value} task."
             )
             first_submitted_index = 0
 

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/routes/api/v2/task_collection_custom.py

@@ -1,3 +1,4 @@
+import os
 import shlex
 import subprocess  # nosec
 from pathlib import Path
@@ -65,25 +66,36 @@ async def collect_task_custom(
                 detail="Cannot infer 'package_root' with 'slurm_ssh' backend.",
             )
     else:
-        if not Path(task_collect.python_interpreter).is_file():
+        if not os.access(
+            task_collect.python_interpreter, os.X_OK
+        ) or not os.access(task_collect.python_interpreter, os.R_OK):
             raise HTTPException(
                 status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
                 detail=(
                     f"{task_collect.python_interpreter=} "
-                    "doesn't exist or is not a file."
+                    "is not accessible to the Fractal user "
+                    "or it is not executable."
                 ),
             )
-        if (
-            task_collect.package_root is not None
-            and not Path(task_collect.package_root).is_dir()
-        ):
+        if not Path(task_collect.python_interpreter).is_file():
             raise HTTPException(
                 status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
-                detail=(
-                    f"{task_collect.package_root=} "
-                    "doesn't exist or is not a directory."
-                ),
+                detail=f"{task_collect.python_interpreter=} is not a file.",
             )
+        if task_collect.package_root is not None:
+            if not os.access(task_collect.package_root, os.R_OK):
+                raise HTTPException(
+                    status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                    detail=(
+                        f"{task_collect.package_root=} "
+                        "is not accessible to the Fractal user."
+                    ),
+                )
+            if not Path(task_collect.package_root).is_dir():
+                raise HTTPException(
+                    status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                    detail=f"{task_collect.package_root=} is not a directory.",
+                )
 
     if task_collect.package_root is None:
 

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/runner/compress_folder.py

@@ -80,6 +80,7 @@ def _remove_temp_subfolder(subfolder_path_tmp_copy: Path, logger_name: str):
 def compress_folder(
     subfolder_path: Path,
     filelist_path: str | None,
+    default_logging_level: int | None = None,
 ) -> str:
     """
     Compress e.g. `/path/archive` into `/path/archive.tar.gz`
@@ -90,13 +91,17 @@ def compress_folder(
     Args:
         subfolder_path: Absolute path to the folder to compress.
         remote_to_local: If `True`, exclude some files from the tar.gz archive.
+        default_logging_level:
 
     Returns:
         Absolute path to the tar.gz archive.
     """
 
     logger_name = "compress_folder"
-    logger = set_logger(logger_name)
+    logger = set_logger(
+        logger_name,
+        default_logging_level=default_logging_level,
+    )
 
     logger.debug("START")
     logger.debug(f"{subfolder_path=}")
@@ -132,7 +137,10 @@ def compress_folder(
         )
 
 
-def main(sys_argv: list[str]):
+def main(
+    sys_argv: list[str],
+    default_logging_level: int | None = None,
+):
 
     help_msg = (
         "Expected use:\n"
@@ -146,15 +154,19 @@ def main(sys_argv: list[str]):
         compress_folder(
             subfolder_path=Path(sys_argv[1]),
             filelist_path=None,
+            default_logging_level=default_logging_level,
         )
     elif num_args == 3 and sys_argv[2] == "--filelist":
         compress_folder(
             subfolder_path=Path(sys_argv[1]),
             filelist_path=sys_argv[3],
+            default_logging_level=default_logging_level,
         )
     else:
         sys.exit(f"Invalid argument.\n{help_msg}\nProvided: {sys_argv[1:]=}")
 
 
 if __name__ == "__main__":
-    main(sys.argv)
+    import logging
+
+    main(sys.argv, default_logging_level=logging.DEBUG)

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/runner/executors/slurm_common/base_slurm_runner.py

@@ -524,12 +524,6 @@ class BaseSlurmRunner(BaseRunner):
                 slurm_job_ids=self.job_ids,
             )
 
-            # NOTE: see issue 2444
-            settings = Inject(get_settings)
-            sleep_time = settings.FRACTAL_SLURM_INTERVAL_BEFORE_RETRIEVAL
-            logger.warning(f"[submit] Now sleep {sleep_time} seconds.")
-            time.sleep(sleep_time)
-
             # Retrieval phase
             logger.debug("[submit] START retrieval phase")
             scancelled_job_ids = []
@@ -705,10 +699,6 @@ class BaseSlurmRunner(BaseRunner):
                 slurm_job_ids=self.job_ids,
             )
 
-            settings = Inject(get_settings)
-            sleep_time = settings.FRACTAL_SLURM_INTERVAL_BEFORE_RETRIEVAL
-            logger.warning(f"[multisubmit] Now sleep {sleep_time} seconds.")
-            time.sleep(sleep_time)
         except Exception as e:
             logger.error(
                 "[multisubmit] Unexpected exception during submission."

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/runner/executors/slurm_common/remote.py

@@ -148,10 +148,8 @@ def worker(
         out = cloudpickle.dumps(result)
 
     # Write the output pickle file
-    tempfile = out_fname + ".tmp"
-    with open(tempfile, "wb") as f:
+    with open(out_fname, "wb") as f:
         f.write(out)
-    os.rename(tempfile, out_fname)
 
 
 if __name__ == "__main__":

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/runner/executors/slurm_sudo/runner.py

@@ -151,7 +151,7 @@ class SudoSlurmRunner(BaseSlurmRunner):
         """
         Fetch artifacts for a list of SLURM jobs.
         """
-        MAX_NUM_THREADS = 4
+        MAX_NUM_THREADS = 12
         THREAD_NAME_PREFIX = "fetch_artifacts"
         logger.debug(
             "[_fetch_artifacts] START "

{fractal_server-2.14.0a36 → fractal_server-2.14.2}/fractal_server/app/runner/extract_archive.py

@@ -25,7 +25,10 @@ def _remove_suffix(*, string: str, suffix: str) -> str:
         raise ValueError(f"Cannot remove {suffix=} from {string=}.")
 
 
-def extract_archive(archive_path: Path):
+def extract_archive(
+    archive_path: Path,
+    default_logging_level: int | None = None,
+):
     """
     Extract e.g. `/path/archive.tar.gz` archive into `/path/archive` folder
 
@@ -34,10 +37,14 @@ def extract_archive(archive_path: Path):
 
     Arguments:
         archive_path: Absolute path to the archive file.
+        default_logging_level
     """
 
     logger_name = "extract_archive"
-    logger = set_logger(logger_name)
+    logger = set_logger(
+        logger_name,
+        default_logging_level=default_logging_level,
+    )
 
     logger.debug("START")
     logger.debug(f"{archive_path.as_posix()=}")
@@ -65,7 +72,10 @@ def extract_archive(archive_path: Path):
     logger.debug("END")
 
 
-def main(sys_argv: list[str]):
+def main(
+    sys_argv: list[str],
+    default_logging_level: int | None = None,
+):
     help_msg = (
         "Expected use:\n"
         "python -m fractal_server.app.runner.extract_archive "
@@ -76,8 +86,14 @@ def main(sys_argv: list[str]):
         sys.exit(f"Invalid argument.\n{help_msg}\nProvided: {sys_argv[1:]=}")
     else:
         tarfile_path = Path(sys_argv[1])
-        extract_archive(tarfile_path)
+        extract_archive(
+            tarfile_path,
+            default_logging_level=default_logging_level,
+        )
 
 
 if __name__ == "__main__":
-    main(sys.argv)
+
+    import logging
+
+    main(sys.argv, default_logging_level=logging.DEBUG)