Pass existing dataset to post_load_files method

ingestify/domain/models/ingestion/ingestion_job.py

@@ -137,7 +137,7 @@ def run(self):
                 for file_id, file_resource in self.dataset_resource.files.items()
             }
 
-            self.dataset_resource.run_post_load_files(files)
+            self.dataset_resource.run_post_load_files(files, self.dataset)
 
             try:
                 revision = self.store.update_dataset(

ingestify/domain/models/resources/dataset_resource.py

@@ -3,6 +3,7 @@
 from pydantic import Field
 
 from ingestify.domain.models.base import BaseModel
+from ingestify.domain.models.dataset import Dataset
 from ingestify.domain.models.dataset.dataset_state import DatasetState
 from ingestify.exceptions import DuplicateFile
 
@@ -51,17 +52,19 @@ class DatasetResource(BaseModel):
     state: DatasetState = Field(default_factory=lambda: DatasetState.COMPLETE)
     files: dict[str, FileResource] = Field(default_factory=dict)
     post_load_files: Optional[
-        Callable[["DatasetResource", Dict[str, DraftFile]], None]
+        Callable[["DatasetResource", Dict[str, DraftFile], Optional[Dataset]], None]
     ] = None
 
-    def run_post_load_files(self, files: Dict[str, DraftFile]):
+    def run_post_load_files(
+        self, files: Dict[str, DraftFile], existing_dataset: Optional[Dataset] = None
+    ):
         """Hook to modify dataset attributes based on loaded file content.
 
         Useful for setting state based on file content, e.g., keep state=SCHEDULED
         when files contain '{}', change to COMPLETE when they contain actual data.
         """
         if self.post_load_files:
-            self.post_load_files(self, files)
+            self.post_load_files(self, files, existing_dataset)
 
     def add_file(
         self,

ingestify/tests/test_engine.py

@@ -460,7 +460,9 @@ def test_dev_engine():
     assert datasets.first().name == "Test Dataset"
 
 
-def post_load_hook(dataset_resource: DatasetResource, files: dict[str, DraftFile]):
+def post_load_hook(
+    dataset_resource: DatasetResource, files: dict[str, DraftFile], existing_dataset
+):
     # Change state to COMPLETE if file content is not '{}'
     for file in files.values():
         if file.size > 2:
@@ -545,3 +547,13 @@ def test_force_save_creates_revision(config_file):
     dataset_without_files = engine.store.get_dataset_collection(season_id=3).first()
     assert len(dataset_without_files.revisions) == 1
     assert len(dataset_without_files.current_revision.modified_files) == 0
+
+    dataset_with_last_modified = engine.store.get_dataset_collection(
+        season_id=2
+    ).first()
+
+    dataset_without_files = engine.store.get_dataset_collection(metadata_only=True)
+    assert (
+        dataset_without_files.metadata.last_modified
+        == dataset_with_last_modified.last_modified_at
+    )