Update DatasetState logic to match naming standard

https://statistics-norway.atlassian.net/wiki/spaces/MPD/pages/3170140229/Navnestandard+og+versjonering+av+datasett+i+DAPLA#Standard-%E2%80%9Cbuckets%E2%80%9D-(b%C3%B8tter)-og-mapper-som-opprettes-for-alle-statistikkprodukt-team-i-SSB
statisticsnorway · Sep 2, 2022 · 6c4eb7c · 6c4eb7c
1 parent 4ad1b7a
commit 6c4eb7c
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 9 deletions.
diff --git a/datadoc/backend/DataDocMetadata.py b/datadoc/backend/DataDocMetadata.py
@@ -79,12 +79,18 @@ def get_dataset_state(self, dataset: str) -> Optional[DatasetState]:
         if dataset is None:
             return None
         dataset_path_parts = list(pathlib.Path(dataset).parts)
-        if "kildedata" in dataset_path_parts:
+        if "utdata" in dataset_path_parts:
+            return DatasetState.OUTPUT_DATA
+        elif "statistikk" in dataset_path_parts:
+            return DatasetState.STATISTIC
+        elif "klargjorte-data" in dataset_path_parts:
+            return DatasetState.PROCESSED_DATA
+        elif "klargjorte_data" in dataset_path_parts:
+            return DatasetState.PROCESSED_DATA
+        elif "kildedata" in dataset_path_parts:
             return DatasetState.SOURCE_DATA
         elif "inndata" in dataset_path_parts:
             return DatasetState.INPUT_DATA
-        elif "klargjorte_data" in dataset_path_parts:
-            return DatasetState.PROCESSED_DATA
         else:
             return None
 

diff --git a/datadoc/tests/test_datadoc_metadata.py b/datadoc/tests/test_datadoc_metadata.py
@@ -2,6 +2,7 @@
 import shutil
 from copy import copy
 from pathlib import PurePath
+from typing import List, Tuple
 
 import pytest
 from datadoc_model import Enums
@@ -23,13 +24,15 @@ def metadata():
     yield DataDocMetadata(TEST_PARQUET_FILEPATH)
 
 
-@pytest.fixture
-def make_paths():
+def make_paths() -> List[Tuple[str, DatasetState]]:
     split_path = list(PurePath(TEST_PARQUET_FILEPATH).parts)
     initial_data = [
         ("kildedata", DatasetState.SOURCE_DATA),
         ("inndata", DatasetState.INPUT_DATA),
+        ("roskildedata/klargjorte-data", DatasetState.PROCESSED_DATA),
         ("klargjorte_data", DatasetState.PROCESSED_DATA),
+        ("klargjorte-data", DatasetState.PROCESSED_DATA),
+        ("statistikk", DatasetState.STATISTIC),
         ("", None),
     ]
     test_data = []
@@ -57,10 +60,12 @@ def existing_metadata_file():
     yield None  # Dummy value, No need to return anything in particular here
 
 
-def test_get_dataset_state(metadata, make_paths):
-    for path, expected_result in make_paths:
-        actual_state = metadata.get_dataset_state(path)
-        assert actual_state == expected_result
+@pytest.mark.parametrize(("path", "expected_result"), make_paths())
+def test_get_dataset_state(
+    path: str, expected_result: DatasetState, metadata: DataDocMetadata
+):
+    actual_state = metadata.get_dataset_state(path)
+    assert actual_state == expected_result
 
 
 def test_get_dataset_state_none(metadata):