Phase 1 project prototype

tests/test_bls.py

@@ -0,0 +1,75 @@
+"""Tests for BLS CPI-U download and loader."""
+
+import httpx
+import respx
+
+from admin_analytics.bls.download import download_cpi_file
+from admin_analytics.bls.loader import load_cpi
+from admin_analytics.config import BLS_CPI_URL
+
+
+class TestDownload:
+    @respx.mock
+    def test_download_creates_file(self, tmp_path, monkeypatch):
+        monkeypatch.setattr("admin_analytics.bls.download.BLS_DATA_DIR", tmp_path)
+        respx.get(BLS_CPI_URL).mock(
+            return_value=httpx.Response(200, text="series_id\tyear\tperiod\tvalue\n")
+        )
+        path = download_cpi_file(force=True)
+        assert path.exists()
+        assert path.parent == tmp_path
+
+    @respx.mock
+    def test_download_skips_when_exists(self, tmp_path, monkeypatch):
+        monkeypatch.setattr("admin_analytics.bls.download.BLS_DATA_DIR", tmp_path)
+        existing = tmp_path / "cu.data.0.Current"
+        existing.write_text("cached")
+        path = download_cpi_file(force=False)
+        assert path.read_text() == "cached"
+
+    @respx.mock
+    def test_download_force_overwrites(self, tmp_path, monkeypatch):
+        monkeypatch.setattr("admin_analytics.bls.download.BLS_DATA_DIR", tmp_path)
+        existing = tmp_path / "cu.data.0.Current"
+        existing.write_text("old")
+        respx.get(BLS_CPI_URL).mock(
+            return_value=httpx.Response(200, text="new data")
+        )
+        path = download_cpi_file(force=True)
+        assert path.read_text() == "new data"
+
+
+class TestLoader:
+    def test_load_cpi(self, db_conn, fixtures_dir):
+        fixture = fixtures_dir / "cu_data_sample.tsv"
+        count = load_cpi(db_conn, fixture)
+        # Fixture has 5 valid CUUR0000SA0 monthly rows (M01, M02, M12, M01, M06)
+        # Excludes: M13 (annual avg), CUSR0000SA0 (wrong series), S01 (semi-annual)
+        assert count == 5
+
+    def test_load_cpi_correct_values(self, db_conn, fixtures_dir):
+        fixture = fixtures_dir / "cu_data_sample.tsv"
+        load_cpi(db_conn, fixture)
+        rows = db_conn.execute(
+            "SELECT year, month, value FROM raw_cpi_u ORDER BY year, month"
+        ).fetchall()
+        assert rows[0] == (2022, 1, 281.148)
+        assert rows[-1] == (2023, 6, 305.109)
+
+    def test_load_cpi_types(self, db_conn, fixtures_dir):
+        fixture = fixtures_dir / "cu_data_sample.tsv"
+        load_cpi(db_conn, fixture)
+        row = db_conn.execute(
+            "SELECT year, month, value, series_id FROM raw_cpi_u LIMIT 1"
+        ).fetchone()
+        assert isinstance(row[0], int)
+        assert isinstance(row[1], int)
+        assert isinstance(row[2], float)
+        assert row[3] == "CUUR0000SA0"
+
+    def test_load_cpi_idempotent(self, db_conn, fixtures_dir):
+        fixture = fixtures_dir / "cu_data_sample.tsv"
+        load_cpi(db_conn, fixture)
+        load_cpi(db_conn, fixture)
+        count = db_conn.execute("SELECT COUNT(*) FROM raw_cpi_u").fetchone()[0]
+        assert count == 5