added new exception in case of incorrect usage of inputs

Author: sanderegg

services/dask-sidecar/src/simcore_service_dask_sidecar/computational_sidecar/core.py

@@ -13,7 +13,10 @@
 from aiodocker import Docker
 from common_library.json_serialization import json_dumps
 from dask_task_models_library.container_tasks.docker import DockerBasicAuth
-from dask_task_models_library.container_tasks.errors import ServiceRuntimeError
+from dask_task_models_library.container_tasks.errors import (
+    ServiceInputsUseFileToKeyMapButReceivesZipDataError,
+    ServiceRuntimeError,
+)
 from dask_task_models_library.container_tasks.io import FileUrl, TaskOutputData
 from dask_task_models_library.container_tasks.protocol import ContainerTaskParameters
 from models_library.progress_bar import ProgressReport
@@ -27,7 +30,11 @@
 
 from ..settings import ApplicationSettings
 from ..utils.dask import TaskPublisher
-from ..utils.files import pull_file_from_remote, push_file_to_remote
+from ..utils.files import (
+    check_need_unzipping,
+    pull_file_from_remote,
+    push_file_to_remote,
+)
 from .docker_utils import (
     create_container_config,
     get_computational_shared_data_mount_point,
@@ -75,6 +82,17 @@ async def _write_input_data(
 
                 destination_path = task_volumes.inputs_folder / file_name
 
+                need_extraction = check_need_unzipping(
+                    input_params.url, input_params.file_mime_type, destination_path
+                )
+                if input_params.file_mapping and need_extraction:
+                    raise ServiceInputsUseFileToKeyMapButReceivesZipDataError(
+                        service_key=self.task_parameters.image,
+                        service_version=self.task_parameters.tag,
+                        input_key=input_key,
+                        file_to_key_map=input_params.file_mapping,
+                    )
+
                 if destination_path.parent != task_volumes.inputs_folder:
                     # NOTE: only 'task_volumes.inputs_folder' part of 'destination_path' is guaranteed,
                     # if extra subfolders via file-mapping,

services/dask-sidecar/src/simcore_service_dask_sidecar/computational_sidecar/task_shared_volume.py

@@ -1,4 +1,3 @@
-import asyncio
 import logging
 import shutil
 from dataclasses import dataclass
@@ -23,7 +22,7 @@ def __post_init__(self) -> None:
 
             assert not folder_path.exists()  # nosec
             folder_path.mkdir(parents=True)
-            logger.debug(
+            logger.critical(
                 "created %s in %s",
                 f"{folder=}",
                 f"{self.base_path=}",
@@ -61,4 +60,5 @@ async def __aexit__(
         exc: BaseException | None,
         tb: TracebackType | None,
     ) -> None:
-        await asyncio.get_event_loop().run_in_executor(None, self.cleanup)
+        ...
+        # await asyncio.get_event_loop().run_in_executor(None, self.cleanup)

services/dask-sidecar/src/simcore_service_dask_sidecar/utils/files.py

@@ -141,6 +141,22 @@ async def _copy_file(
 _ZIP_MIME_TYPE: Final[str] = "application/zip"
 
 
+def check_need_unzipping(
+    src_url: AnyUrl,
+    target_mime_type: str | None,
+    dst_path: Path,
+) -> bool:
+    """
+    Checks if the source URL points to a zip file and if the target mime type is not zip.
+    If so, extraction is needed.
+    """
+    src_mime_type, _ = mimetypes.guess_type(f"{src_url.path}")
+    dst_mime_type = target_mime_type
+    if not dst_mime_type:
+        dst_mime_type, _ = mimetypes.guess_type(dst_path)
+    return (src_mime_type == _ZIP_MIME_TYPE) and (dst_mime_type != _ZIP_MIME_TYPE)
+
+
 async def pull_file_from_remote(
     src_url: AnyUrl,
     target_mime_type: str | None,
@@ -157,17 +173,11 @@ async def pull_file_from_remote(
         msg = f"{dst_path.parent=} does not exist. It must be created by the caller"
         raise ValueError(msg)
 
-    src_mime_type, _ = mimetypes.guess_type(f"{src_url.path}")
-    if not target_mime_type:
-        target_mime_type, _ = mimetypes.guess_type(dst_path)
-
     storage_kwargs: S3FsSettingsDict | dict[str, Any] = {}
     if s3_settings and src_url.scheme in S3_FILE_SYSTEM_SCHEMES:
         storage_kwargs = _s3fs_settings_from_s3_settings(s3_settings)
 
-    need_extraction = (src_mime_type == _ZIP_MIME_TYPE) and (
-        target_mime_type != _ZIP_MIME_TYPE
-    )
+    need_extraction = check_need_unzipping(src_url, target_mime_type, dst_path)
     async with AsyncExitStack() as exit_stack:
         if need_extraction:
             # we need to extract the file, so we create a temporary directory