Initial build out

src/admin_analytics/db/schema.py

@@ -0,0 +1,122 @@
+import duckdb
+
+TABLES = {
+    "raw_institution": """
+        CREATE TABLE IF NOT EXISTS raw_institution (
+            unitid INTEGER NOT NULL,
+            year INTEGER NOT NULL,
+            ein VARCHAR,
+            institution_name VARCHAR,
+            city VARCHAR,
+            state VARCHAR,
+            sector INTEGER,
+            control INTEGER,
+            carnegie_class INTEGER,
+            enrollment_total INTEGER,
+            PRIMARY KEY (unitid, year)
+        )
+    """,
+    "raw_ipeds_finance": """
+        CREATE TABLE IF NOT EXISTS raw_ipeds_finance (
+            unitid INTEGER NOT NULL,
+            year INTEGER NOT NULL,
+            reporting_standard VARCHAR,
+            total_expenses BIGINT,
+            instruction_expenses BIGINT,
+            research_expenses BIGINT,
+            public_service_expenses BIGINT,
+            academic_support_expenses BIGINT,
+            student_services_expenses BIGINT,
+            institutional_support_expenses BIGINT,
+            auxiliary_expenses BIGINT,
+            hospital_expenses BIGINT,
+            other_expenses BIGINT,
+            salaries_wages BIGINT,
+            benefits BIGINT,
+            PRIMARY KEY (unitid, year)
+        )
+    """,
+    "raw_ipeds_staff": """
+        CREATE TABLE IF NOT EXISTS raw_ipeds_staff (
+            unitid INTEGER NOT NULL,
+            year INTEGER NOT NULL,
+            total_staff INTEGER,
+            faculty_total INTEGER,
+            management_total INTEGER,
+            PRIMARY KEY (unitid, year)
+        )
+    """,
+    "raw_ipeds_enrollment": """
+        CREATE TABLE IF NOT EXISTS raw_ipeds_enrollment (
+            unitid INTEGER NOT NULL,
+            year INTEGER NOT NULL,
+            total_enrollment INTEGER,
+            PRIMARY KEY (unitid, year)
+        )
+    """,
+    "raw_990_filing": """
+        CREATE TABLE IF NOT EXISTS raw_990_filing (
+            object_id VARCHAR PRIMARY KEY,
+            ein VARCHAR,
+            tax_year INTEGER,
+            organization_name VARCHAR,
+            return_type VARCHAR,
+            filing_date DATE,
+            total_revenue BIGINT,
+            total_expenses BIGINT,
+            total_assets BIGINT
+        )
+    """,
+    "raw_990_schedule_j": """
+        CREATE SEQUENCE IF NOT EXISTS seq_990_schedule_j START 1;
+        CREATE TABLE IF NOT EXISTS raw_990_schedule_j (
+            id INTEGER PRIMARY KEY DEFAULT nextval('seq_990_schedule_j'),
+            object_id VARCHAR,
+            ein VARCHAR,
+            tax_year INTEGER,
+            person_name VARCHAR,
+            title VARCHAR,
+            base_compensation BIGINT,
+            bonus_compensation BIGINT,
+            other_compensation BIGINT,
+            deferred_compensation BIGINT,
+            nontaxable_benefits BIGINT,
+            total_compensation BIGINT,
+            compensation_from_related BIGINT
+        )
+    """,
+    "raw_990_part_vii": """
+        CREATE SEQUENCE IF NOT EXISTS seq_990_part_vii START 1;
+        CREATE TABLE IF NOT EXISTS raw_990_part_vii (
+            id INTEGER PRIMARY KEY DEFAULT nextval('seq_990_part_vii'),
+            object_id VARCHAR,
+            ein VARCHAR,
+            tax_year INTEGER,
+            person_name VARCHAR,
+            title VARCHAR,
+            avg_hours_per_week DOUBLE,
+            reportable_comp_from_org BIGINT,
+            reportable_comp_from_related BIGINT,
+            other_compensation BIGINT
+        )
+    """,
+    "raw_cpi_u": """
+        CREATE TABLE IF NOT EXISTS raw_cpi_u (
+            year INTEGER NOT NULL,
+            month INTEGER NOT NULL,
+            value DOUBLE,
+            series_id VARCHAR,
+            PRIMARY KEY (year, month)
+        )
+    """,
+}
+
+
+def ensure_schema(conn: duckdb.DuckDBPyConnection) -> None:
+    """Create all raw tables if they don't exist."""
+    for ddl in TABLES.values():
+        # Some DDL blocks contain multiple statements (e.g. CREATE SEQUENCE + CREATE TABLE)
+        for stmt in ddl.split(";"):
+            stmt = stmt.strip()
+            if stmt:
+                conn.execute(stmt)