Address comments

Author: islean

cg/apps/housekeeper/hk.py

@@ -403,7 +403,7 @@ def get_files_from_latest_version(self, bundle_name: str, tags: list[str]) -> Qu
         return self.files(version=version.id, tags=tags)
 
     def is_fastq_or_spring_in_all_bundles(self, bundle_names: list[str]) -> bool:
-        """Return whether or not all FASTQ/SPRING files are included for the given bundles."""
+        """Return whether all FASTQ/SPRING files are included for the given bundles."""
         sequencing_files_in_hk: dict[str, bool] = {}
         if not bundle_names:
             return False
@@ -423,11 +423,11 @@ def is_fastq_or_spring_in_all_bundles(self, bundle_names: list[str]) -> bool:
         return all(sequencing_files_in_hk.values())
 
     def get_non_archived_files(self, bundle_name: str, tags: Optional[list] = None) -> list[File]:
-        """Returns all non-archived_files from a given bundle, tagged with the given tags"""
+        """Returns all non-archived files from a given bundle, tagged with the given tags"""
         return self._store.get_non_archived_files(bundle_name=bundle_name, tags=tags or [])
 
     def get_archived_files(self, bundle_name: str, tags: list | None = None) -> list[File]:
-        """Returns all archived_files from a given bundle, tagged with the given tags"""
+        """Returns all archived files from a given bundle, tagged with the given tags"""
         return self._store.get_archived_files(bundle_name=bundle_name, tags=tags or [])
 
     def add_archives(self, files: list[Path], archive_task_id: int) -> None:

cg/constants/archiving.py

@@ -5,3 +5,4 @@ class ArchiveLocations(StrEnum):
     """Demultiplexing related directories and files."""
 
     KAROLINSKA_BUCKET: str = "karolinska_bucket"
+    PDC: str = "PDC"

cg/meta/archive/archive.py

@@ -98,16 +98,21 @@ def archive_all_non_archived_spring_files(
         files_and_samples: list[FileAndSample] = self.add_samples_to_files(files_to_archive)
 
         for archive_location in ArchiveLocations:
-            self.archive_to_location(
-                files_and_samples=files_and_samples,
-                archive_location=archive_location,
-            )
+            if archive_location != ArchiveLocations.PDC:
+                self.archive_to_location(
+                    files_and_samples=files_and_samples,
+                    archive_location=archive_location,
+                )
 
-    def retrieve_case(self, case_id: str):
+    def retrieve_case(self, case_id: str) -> None:
+        """Submits jobs to retrieve any archived files belonging to the given case, and updates the Archive entries
+        with the retrieval job id."""
         case = self.status_db.get_case_by_internal_id(case_id)
-        self.retrieve_samples(self.get_archived_samples(case))
+        self.retrieve_samples(self.get_samples_with_archived_spring_files(case))
 
-    def get_archived_samples(self, case: Case):
+    def get_samples_with_archived_spring_files(self, case: Case) -> list[Sample]:
+        """Returns a list of samples which have archived Spring files, i.e. they have entries in the Archive table
+        in Housekeeper."""
         return [
             link.sample
             for link in case.links

cg/meta/workflow/analysis.py

@@ -18,6 +18,7 @@
     Priority,
     SequencingFileTag,
 )
+from cg.constants.archiving import ArchiveLocations
 from cg.constants.constants import AnalysisType, CaseActions, WorkflowManager
 from cg.constants.priority import PRIORITY_TO_SLURM_QOS
 from cg.exc import AnalysisNotReadyError, BundleAlreadyAddedError, CgDataError, CgError
@@ -491,7 +492,7 @@ def ensure_flow_cells_on_disk(self, case_id: str) -> None:
             self.status_db.request_flow_cells_for_case(case_id)
 
     def is_case_ready_for_analysis(self, case_id: str) -> bool:
-        if self.get_archive_location_for_case(case_id) == "PDC":
+        if self.get_archive_location_for_case(case_id) == ArchiveLocations.PDC:
             if self._is_flow_cell_check_applicable(
                 case_id
             ) and not self.status_db.are_all_flow_cells_on_disk(case_id):
@@ -516,9 +517,9 @@ def prepare_fastq_files(self, case_id: str, dry_run: bool) -> None:
         if not self.is_case_ready_for_analysis(case_id):
             raise AnalysisNotReadyError("FASTQ file are not present for the analysis to start")
 
-    def ensure_files_are_not_archived(self, case_id):
+    def ensure_files_are_not_archived(self, case_id: str):
         """Checks if any files are archived and submits a job to retrieve any which are."""
-        if self.get_archive_location_for_case(case_id) == "PDC":
+        if self.get_archive_location_for_case(case_id) == ArchiveLocations.PDC:
             self.ensure_flow_cells_on_disk(case_id)
         else:
             if not self.are_all_files_present(case_id):
@@ -529,7 +530,9 @@ def ensure_files_are_not_archived(self, case_id):
                 )
                 spring_archive_api.retrieve_case(case_id)
 
-    def are_all_files_present(self, case_id):
+    def are_all_files_present(self, case_id: str) -> bool:
+        """Returns true if no Spring files are archived in the data location used by the customer tied to the given
+        case, and false otherwise."""
         case: Case = self.status_db.get_case_by_internal_id(case_id)
         for sample in [link.sample for link in case.links]:
             if (
@@ -540,5 +543,5 @@ def are_all_files_present(self, case_id):
                 return False
         return True
 
-    def get_archive_location_for_case(self, case_id: str):
+    def get_archive_location_for_case(self, case_id: str) -> str:
         return self.status_db.get_case_by_internal_id(case_id).customer.data_archive_location