Feature/pdct 1454 fix status should be a list not a string (#18)

* fix: update metadata types - when validating we expect these fields to be a list * fix: update document variant names - this matches the fields that we expect in the database * fix: update family import id - should contain 'family' not events * test: update tests --------- Co-authored-by: Osneil Drakes <[email protected]>
climatepolicyradar · Sep 11, 2024 · 81f6f3f · 81f6f3f
1 parent aa37326
commit 81f6f3f
Show file tree

Hide file tree

Showing 6 changed files with 9 additions and 9 deletions.
diff --git a/gcf_data_mapper/enums/document.py b/gcf_data_mapper/enums/document.py
@@ -30,5 +30,5 @@ class IgnoreDocumentTypes(Enum):
 
 
 class DocumentVariantNames(Enum):
-    ORIGINAL = "Original Translation"
-    TRANSLATION = "Translated"
+    ORIGINAL = "Original Language"
+    TRANSLATION = "Translation"
diff --git a/gcf_data_mapper/parsers/document.py b/gcf_data_mapper/parsers/document.py
@@ -127,7 +127,7 @@ def map_document_metadata(
     return {
         "import_id": f"GCF.document.{approved_ref}_{projects_id}.{doc_id}",
         "family_import_id": f"GCF.family.{approved_ref}.{projects_id}",
-        "metadata": {"type": doc_type},
+        "metadata": {"type": [doc_type]},
         "title": title,
         "source_url": source_url.strip(),
         "variant_name": variant_name,

diff --git a/gcf_data_mapper/parsers/event.py b/gcf_data_mapper/parsers/event.py
@@ -32,7 +32,7 @@ def append_event(
     gcf_events.append(
         {
             "import_id": f"GCF.event.{approved_ref}_{projects_id}.n{n_value:04}",
-            "family_import_id": f"GCF.event.{approved_ref}.{projects_id}",
+            "family_import_id": f"GCF.family.{approved_ref}.{projects_id}",
             "event_title": event.type,
             "date": row[event.column_name],
             "event_type_value": event.type,

diff --git a/gcf_data_mapper/parsers/family.py b/gcf_data_mapper/parsers/family.py
@@ -149,7 +149,7 @@ def map_family_metadata(row: pd.Series) -> Optional[dict]:
         "result_areas": list(set(areas)),
         "result_types": list(set(types)),
         "sector": [row.at[FamilyColumnsNames.SECTOR.value]],
-        "status": status,
+        "status": [status],
         "theme": [row.at[FamilyColumnsNames.THEME.value]],
     }
 
@@ -192,7 +192,7 @@ def map_family_data(
         # For now we are hard coding the category as MCF
         "category": "MCF",
         "collections": [],
-        "description": summary,
+        "summary": summary,
         "geographies": geographies,
         "import_id": import_id,
         "metadata": family_metadata,

diff --git a/tests/unit_tests/parsers/family/test_map_family.py b/tests/unit_tests/parsers/family/test_map_family.py
@@ -10,7 +10,7 @@ def parsed_family_data():
         {
             "category": "MCF",
             "collections": [],
-            "description": "The Summary of the Project",
+            "summary": "The Summary of the Project",
             "geographies": ["BGD"],
             "import_id": "GCF.family.FP003.12660",
             "metadata": {
@@ -24,7 +24,7 @@ def parsed_family_data():
                 "result_areas": ["Coastal protection and restoration"],
                 "result_types": ["Adaptation"],
                 "sector": ["Environment"],
-                "status": "Under Implementation",
+                "status": ["Under Implementation"],
                 "theme": ["Adaptation"],
             },
             "title": "Enhancing resilience of coastal ecosystems and communities",

diff --git a/tests/unit_tests/parsers/family/test_map_family_metadata.py b/tests/unit_tests/parsers/family/test_map_family_metadata.py
@@ -25,7 +25,7 @@ def parsed_family_metadata():
         "result_areas": ["The Area for the Result Area"],
         "result_types": ["The Type for the Result Area"],
         "sector": ["Private"],
-        "status": "Under Implementation",
+        "status": ["Under Implementation"],
         "theme": ["Adaptation"],
     }