Phase 1 project prototype

tests/test_validation.py

@@ -0,0 +1,102 @@
+"""Tests for data validation module."""
+
+from admin_analytics.validation import (
+    validate_row_counts,
+    validate_null_rates,
+    validate_year_coverage,
+    validate_cross_source_consistency,
+    format_report,
+)
+
+
+class TestRowCounts:
+    def test_empty_tables(self, db_conn):
+        counts = validate_row_counts(db_conn)
+        assert counts["raw_cpi_u"] == 0
+        assert counts["raw_institution"] == 0
+
+    def test_with_data(self, db_conn):
+        db_conn.execute(
+            "INSERT INTO raw_cpi_u (year, month, value, series_id) VALUES (2023, 1, 299.17, 'CUUR0000SA0')"
+        )
+        db_conn.execute(
+            "INSERT INTO raw_cpi_u (year, month, value, series_id) VALUES (2023, 2, 300.84, 'CUUR0000SA0')"
+        )
+        counts = validate_row_counts(db_conn)
+        assert counts["raw_cpi_u"] == 2
+
+
+class TestNullRates:
+    def test_empty_tables_excluded(self, db_conn):
+        results = validate_null_rates(db_conn)
+        assert "raw_cpi_u" not in results
+
+    def test_no_nulls(self, db_conn):
+        db_conn.execute(
+            "INSERT INTO raw_cpi_u (year, month, value, series_id) VALUES (2023, 1, 299.17, 'CUUR0000SA0')"
+        )
+        results = validate_null_rates(db_conn)
+        assert results["raw_cpi_u"]["year"] == 0.0
+        assert results["raw_cpi_u"]["value"] == 0.0
+
+    def test_with_nulls(self, db_conn):
+        db_conn.execute(
+            "INSERT INTO raw_990_filing (object_id, ein, tax_year) VALUES ('f1', '123', 2023)"
+        )
+        db_conn.execute(
+            "INSERT INTO raw_990_filing (object_id, ein, tax_year, total_revenue) VALUES ('f2', '123', 2023, 100)"
+        )
+        results = validate_null_rates(db_conn)
+        assert results["raw_990_filing"]["total_revenue"] == 50.0
+        assert results["raw_990_filing"]["ein"] == 0.0
+
+
+class TestYearCoverage:
+    def test_empty_tables(self, db_conn):
+        coverage = validate_year_coverage(db_conn)
+        assert coverage["raw_cpi_u"]["years"] == []
+
+    def test_with_data(self, db_conn):
+        for year in [2020, 2021, 2023]:
+            db_conn.execute(
+                "INSERT INTO raw_cpi_u (year, month, value, series_id) VALUES (?, 1, 100.0, 'CUUR0000SA0')",
+                [year],
+            )
+        coverage = validate_year_coverage(db_conn)
+        assert coverage["raw_cpi_u"]["years"] == [2020, 2021, 2023]
+        assert 2022 in coverage["raw_cpi_u"]["gaps"]
+
+
+class TestCrossSource:
+    def test_empty(self, db_conn):
+        result = validate_cross_source_consistency(db_conn)
+        assert result["years_in_all_sources"] == []
+
+    def test_overlap(self, db_conn):
+        # Add IPEDS finance
+        db_conn.execute(
+            "INSERT INTO raw_ipeds_finance (unitid, year) VALUES (130943, 2022)"
+        )
+        db_conn.execute(
+            "INSERT INTO raw_ipeds_finance (unitid, year) VALUES (130943, 2023)"
+        )
+        # Add 990 filing
+        db_conn.execute(
+            "INSERT INTO raw_990_filing (object_id, tax_year) VALUES ('f1', 2023)"
+        )
+        # Add CPI
+        db_conn.execute(
+            "INSERT INTO raw_cpi_u (year, month, value, series_id) VALUES (2023, 1, 299.0, 'X')"
+        )
+        result = validate_cross_source_consistency(db_conn)
+        assert 2023 in result["years_in_all_sources"]
+        assert 2022 not in result["years_in_all_sources"]
+
+
+class TestFormatReport:
+    def test_runs_on_empty_db(self, db_conn):
+        report = format_report(db_conn)
+        assert "Row Counts" in report
+        assert "NULL Rates" in report
+        assert "Year Coverage" in report
+        assert "Cross-Source" in report