Merge pull request #369 from codeforpdx/fix-metadata-null-city

Encode state-level city metadata as "null" string

Author: leekahung

backend/scripts/generate_metadata_jsonl.py

@@ -68,7 +68,13 @@ def in_scope(path: Path) -> bool:
         entries.append(
             Document(
                 id=txt_file.stem,
-                struct_data={"city": city, "state": "or"},
+                # State-level docs use the literal string "null" (not JSON null,
+                # which Vertex drops on import) so the retriever's city: ANY("null")
+                # filter matches them.
+                struct_data={
+                    "city": city if city is not None else "null",
+                    "state": "or",
+                },
                 content=Document.Content(
                     mime_type="text/plain",
                     uri=f"gs://{bucket}/{txt_file.name}",

backend/tests/test_generate_metadata_jsonl.py

@@ -58,7 +58,7 @@ def test_state_entry_structure(self, doc_tree: Path):
         entries = build_entries(doc_tree, "my-bucket", {"or"})
         entry = entries[0]
         assert entry.id == "ORS090"
-        assert entry.struct_data == {"city": None, "state": "or"}
+        assert entry.struct_data == {"city": "null", "state": "or"}
         assert entry.content.mime_type == "text/plain"
         assert entry.content.uri == "gs://my-bucket/ORS090.txt"