🐛 unarchiving_utils are now based on 7zip cli

services/dynamic-sidecar/src/simcore_service_dynamic_sidecar/modules/nodeports.py

@@ -20,7 +20,12 @@
 from models_library.projects_nodes_io import NodeIDStr
 from models_library.services_types import ServicePortKey
 from pydantic import ByteSize
-from servicelib.archiving_utils import PrunableFolder, archive_dir, unarchive_dir
+from servicelib.archiving_utils import (
+    ArchiveError,
+    PrunableFolder,
+    archive_dir,
+    unarchive_dir,
+)
 from servicelib.async_utils import run_sequentially_in_context
 from servicelib.file_utils import remove_directory
 from servicelib.logging_utils import log_context
@@ -46,7 +51,7 @@ class PortTypeName(str, Enum):
 _FILE_TYPE_PREFIX = "data:"
 _KEY_VALUE_FILE_NAME = "key_values.json"
 
-logger = logging.getLogger(__name__)
+_logger = logging.getLogger(__name__)
 
 # OUTPUTS section
 
@@ -95,7 +100,7 @@ async def upload_outputs(  # pylint:disable=too-many-statements  # noqa: PLR0915
     port_notifier: PortNotifier,
 ) -> None:
     # pylint: disable=too-many-branches
-    logger.debug("uploading data to simcore...")
+    _logger.debug("uploading data to simcore...")
     start_time = time.perf_counter()
 
     settings: ApplicationSettings = get_settings()
@@ -138,7 +143,7 @@ async def upload_outputs(  # pylint:disable=too-many-statements  # noqa: PLR0915
             if is_file_type(port.property_type):
                 src_folder = outputs_path / port.key
                 files_and_folders_list = list(src_folder.rglob("*"))
-                logger.debug("Discovered files to upload %s", files_and_folders_list)
+                _logger.debug("Discovered files to upload %s", files_and_folders_list)
 
                 if not files_and_folders_list:
                     ports_values[port.key] = (None, None)
@@ -213,9 +218,9 @@ async def _archive_dir_notified(
                     if port.key in data and data[port.key] is not None:
                         ports_values[port.key] = (data[port.key], None)
                     else:
-                        logger.debug("Port %s not found in %s", port.key, data)
+                        _logger.debug("Port %s not found in %s", port.key, data)
                 else:
-                    logger.debug("No file %s to fetch port values from", data_file)
+                    _logger.debug("No file %s to fetch port values from", data_file)
 
         if archiving_tasks:
             await limited_gather(*archiving_tasks, limit=4)
@@ -228,8 +233,8 @@ async def _archive_dir_notified(
 
         elapsed_time = time.perf_counter() - start_time
         total_bytes = sum(_get_size_of_value(x) for x in ports_values.values())
-        logger.info("Uploaded %s bytes in %s seconds", total_bytes, elapsed_time)
-        logger.debug(_CONTROL_TESTMARK_DY_SIDECAR_NODEPORT_UPLOADED_MESSAGE)
+        _logger.info("Uploaded %s bytes in %s seconds", total_bytes, elapsed_time)
+        _logger.debug(_CONTROL_TESTMARK_DY_SIDECAR_NODEPORT_UPLOADED_MESSAGE)
 
 
 # INPUTS section
@@ -243,14 +248,32 @@ def _is_zip_file(file_path: Path) -> bool:
 _shutil_move = aiofiles.os.wrap(shutil.move)
 
 
+async def _move_file_to_input_port(
+    final_path: Path, downloaded_file: Path, dest_folder: PrunableFolder
+) -> None:
+    _logger.debug("moving %s", downloaded_file)
+    final_path = final_path / downloaded_file.name
+
+    # ensure parent exists
+    final_path.parent.mkdir(exist_ok=True, parents=True)
+
+    await _shutil_move(downloaded_file, final_path)
+
+    # NOTE: after the download the current value of the port
+    # makes sure previously downloaded files are removed
+    dest_folder.prune(exclude={final_path})
+
+    _logger.debug("file moved to %s", final_path)
+
+
 async def _get_data_from_port(
     port: Port, *, target_dir: Path, progress_bar: ProgressBarData
 ) -> tuple[Port, ItemConcreteValue | None, ByteSize]:
     async with progress_bar.sub_progress(
         steps=2 if is_file_type(port.property_type) else 1,
         description=IDStr("getting data"),
     ) as sub_progress:
-        with log_context(logger, logging.DEBUG, f"getting {port.key=}"):
+        with log_context(_logger, logging.DEBUG, f"getting {port.key=}"):
             port_data = await port.get(sub_progress)
 
         if is_file_type(port.property_type):
@@ -261,7 +284,7 @@ async def _get_data_from_port(
             if not downloaded_file or not downloaded_file.exists():
                 # the link may be empty
                 # remove files all files from disk when disconnecting port
-                logger.debug("removing contents of dir %s", final_path)
+                _logger.debug("removing contents of dir %s", final_path)
                 await remove_directory(
                     final_path, only_children=True, ignore_errors=True
                 )
@@ -270,33 +293,38 @@ async def _get_data_from_port(
             transferred_bytes = downloaded_file.stat().st_size
 
             # in case of valid file, it is either uncompressed and/or moved to the final directory
-            with log_context(logger, logging.DEBUG, "creating directory"):
+            with log_context(_logger, logging.DEBUG, "creating directory"):
                 final_path.mkdir(exist_ok=True, parents=True)
             port_data = f"{final_path}"
             dest_folder = PrunableFolder(final_path)
 
             if _is_zip_file(downloaded_file):
                 # unzip updated data to dest_path
-                logger.debug("unzipping %s", downloaded_file)
-                unarchived: set[Path] = await unarchive_dir(
-                    archive_to_extract=downloaded_file,
-                    destination_folder=final_path,
-                    progress_bar=sub_progress,
-                )
-
-                dest_folder.prune(exclude=unarchived)
+                _logger.debug("unzipping %s", downloaded_file)
+                try:
+                    unarchived: set[Path] = await unarchive_dir(
+                        archive_to_extract=downloaded_file,
+                        destination_folder=final_path,
+                        progress_bar=sub_progress,
+                    )
+                    dest_folder.prune(exclude=unarchived)
+
+                    _logger.debug("all unzipped in %s", final_path)
+                except ArchiveError:
+                    _logger.warning(
+                        "Could not extract archive '%s' to '%s' moving it to: '%s'",
+                        downloaded_file,
+                        final_path,
+                        final_path / downloaded_file.name,
+                    )
+                    await _move_file_to_input_port(
+                        final_path, downloaded_file, dest_folder
+                    )
 
-                logger.debug("all unzipped in %s", final_path)
             else:
-                logger.debug("moving %s", downloaded_file)
-                final_path = final_path / Path(downloaded_file).name
-                await _shutil_move(str(downloaded_file), final_path)
-
-                # NOTE: after the download the current value of the port
-                # makes sure previously downloaded files are removed
-                dest_folder.prune(exclude={final_path})
+                await _move_file_to_input_port(final_path, downloaded_file, dest_folder)
 
-                logger.debug("all moved to %s", final_path)
+                _logger.debug("all moved to %s", final_path)
         else:
             transferred_bytes = sys.getsizeof(port_data)
 
@@ -312,7 +340,7 @@ async def download_target_ports(
     progress_bar: ProgressBarData,
     port_notifier: PortNotifier | None,
 ) -> ByteSize:
-    logger.debug("retrieving data from simcore...")
+    _logger.debug("retrieving data from simcore...")
     start_time = time.perf_counter()
 
     settings: ApplicationSettings = get_settings()
@@ -386,7 +414,7 @@ async def _get_date_from_port_notified(
         data_file.write_text(json.dumps(data))
 
     elapsed_time = time.perf_counter() - start_time
-    logger.info(
+    _logger.info(
         "Downloaded %s in %s seconds",
         total_transfered_bytes.human_readable(decimal=True),
         elapsed_time,