feat(rules): allow non-unique area

tests/__init__.py

@@ -124,7 +124,10 @@ def df_with_row(
 
     srs = pd.Series(data=row.values(), index=row.keys(), name=area_name)
 
-    df = pd.concat([df, srs.to_frame().T])
+    concat_dfs = [srs.to_frame().T]
+    if df is not None:
+        concat_dfs.append(df)
+    df = pd.concat(concat_dfs)
 
     traces_path = (
         Path(rules.PathNames.UNORGANIZED.value) / f"{traces_name}.{rules.FILETYPE}"

tests/test_rules.py

@@ -1,10 +1,15 @@
 """
 Tests for rules.py.
 """
+from contextlib import nullcontext
 
+import pandas as pd
+import pandera as pa
+import pytest
 from hypothesis import given
 
 import tests
+from tracerepo import rules
 
 
 @given(tests.name_regex(geom_type=None))
@@ -14,3 +19,23 @@ def test_filename_regex(name: str):
     """
     assert isinstance(name, str)
     assert len(name) >= 2
+
+
+@pytest.mark.parametrize(
+    "df,raises",
+    [
+        (pd.DataFrame({}), pytest.raises(pa.errors.SchemaError)),
+        (tests.df_with_row(None)[0], nullcontext()),
+        (pd.concat([tests.df_with_row(None)[0]] * 2), nullcontext()),
+    ],
+)
+def test_database_schema(df, raises):
+    """
+    Test database_schema.
+    """
+    result = None
+    with raises:
+        result = rules.database_schema().validate(df)
+    if result is None:
+        return
+    assert isinstance(result, pd.DataFrame)

tracerepo/rules.py

@@ -176,7 +176,7 @@ def database_schema() -> pa.DataFrameSchema:
     """
     schema = pa.DataFrameSchema(
         # Index is the area name
-        index=pa.Index(**name_column_kwargs(unique=True, geom_type=ColumnNames.AREA)),
+        index=pa.Index(**name_column_kwargs(unique=False, geom_type=ColumnNames.AREA)),
         coerce=True,
         # Columns
         columns={