idgf

pipeline/utils/test_nspl_schema.py

@@ -0,0 +1,57 @@
+import polars as pl
+
+from pipeline.utils.nspl_schema import CODE_COL_PATTERN, code_col_overrides
+
+
+def test_matches_current_and_renamed_suffixes():
+    names = ["pcd", "ruc21ind", "oac11ind", "imd20ind", "lat", "long"]
+    overrides = code_col_overrides(names)
+    assert overrides == {
+        "ruc21ind": pl.String,
+        "oac11ind": pl.String,
+        "imd20ind": pl.String,
+    }
+
+
+def test_catches_future_suffix_bump():
+    # The regression: ONS bumps imd20ind -> imd25ind. A hard-coded dict would
+    # no-op here; the stem match must still catch it.
+    names = ["pcd", "ruc25ind", "oac21ind", "imd25ind"]
+    overrides = code_col_overrides(names)
+    assert set(overrides) == {"ruc25ind", "oac21ind", "imd25ind"}
+    assert all(dtype is pl.String for dtype in overrides.values())
+
+
+def test_is_case_insensitive():
+    overrides = code_col_overrides(["RUC21IND", "Oac11Ind", "IMD20ind"])
+    assert set(overrides) == {"RUC21IND", "Oac11Ind", "IMD20ind"}
+
+
+def test_matches_suffixless_stem():
+    assert set(code_col_overrides(["rucind", "oacind", "imdind"])) == {
+        "rucind",
+        "oacind",
+        "imdind",
+    }
+
+
+def test_ignores_unrelated_columns():
+    names = [
+        "pcd",
+        "laua",
+        "imdscore",  # not an *ind column
+        "indicator",  # ends in nothing relevant, no ruc/oac/imd stem
+        "ruc21indx",  # extra trailing char -> not a code column
+        "xruc21ind",  # leading char -> not anchored at start
+    ]
+    assert code_col_overrides(names) == {}
+
+
+def test_empty_names():
+    assert code_col_overrides([]) == {}
+
+
+def test_pattern_anchored():
+    assert CODE_COL_PATTERN.match("imd25ind")
+    assert not CODE_COL_PATTERN.match("oac11indicator")
+    assert not CODE_COL_PATTERN.match("region_imd20ind")