From 67c8929db37c0ea29f15fabc157dc339555e9c2b Mon Sep 17 00:00:00 2001
From: deepsikha-dash <deepsikhadash326@gmail.com>
Date: Wed, 8 Jul 2026 22:55:52 +0530
Subject: [PATCH] feat(backend): add migration tracking and schema version
 validation

---
 backend/secuscan/database.py             | 61 ++++++++++++++++++++-
 testing/backend/unit/test_saved_views.py | 68 ++++++++++++++++++++++++
 2 files changed, 128 insertions(+), 1 deletion(-)

diff --git a/backend/secuscan/database.py b/backend/secuscan/database.py
index 98048393b..29fede081 100644
--- a/backend/secuscan/database.py
+++ b/backend/secuscan/database.py
@@ -44,6 +44,8 @@ async def connect(self):
         conn.row_factory = aiosqlite.Row
         await conn.execute("PRAGMA foreign_keys = ON")
         await self._create_schema()
+        await self._ensure_schema_migrations_table()
+        await self._validate_schema_version()
         await self._run_migrations()
 
     async def disconnect(self):
@@ -701,6 +703,54 @@ async def _create_schema(self):
             )
 
 
+    async def _ensure_schema_migrations_table(self):
+        """Create the migration tracking table if it does not already exist."""
+        await self.connection.execute(
+            """
+            CREATE TABLE IF NOT EXISTS schema_migrations (
+                version TEXT PRIMARY KEY,
+                applied_at TIMESTAMP NOT NULL DEFAULT (datetime('now'))
+            )
+            """
+        )
+        await self.connection.commit()
+
+    async def _applied_migrations(self) -> set[str]:
+        """Return the set of migration filenames already applied."""
+        rows = await self.fetchall(
+            "SELECT version FROM schema_migrations"
+        )
+        return {row["version"] for row in rows}
+
+    async def _validate_schema_version(self):
+        """Ensure the database was not created by a newer application."""
+
+        applied = await self._applied_migrations()
+
+        available = {
+            migration.name
+            for migration in (Path(__file__).parent / "migrations").glob("*.sql")
+        }
+
+        unknown = applied - available
+
+        if unknown:
+            raise RuntimeError(
+                "Database schema is newer than this application. "
+                f"Unknown migration(s): {', '.join(sorted(unknown))}"
+            )
+
+    async def _record_migration(self, version: str):
+        """Record a successfully applied migration."""
+        await self.execute(
+            """
+            INSERT INTO schema_migrations(version)
+            VALUES (?)
+            """,
+            (version,),
+        )
+
+
     async def _run_migrations(self):
         migrations_dir = Path(__file__).parent / "migrations"
 
@@ -710,13 +760,22 @@ async def _run_migrations(self):
                 "ensure the backend package is installed correctly."
             )
 
+        applied = await self._applied_migrations()
+
         for migration_file in sorted(migrations_dir.glob("*.sql")):
+            migration_name = migration_file.name
+
+            if migration_name in applied:
+                continue
+
             sql = migration_file.read_text(encoding="utf-8")
+
             try:
                 await self.connection.executescript(sql)
+                await self._record_migration(migration_name)
             except Exception as exc:
                 raise RuntimeError(
-                    f"Migration {migration_file.name} failed — startup aborted: {exc}"
+                    f"Migration {migration_name} failed — startup aborted: {exc}"
                 ) from exc
 
         await self._backfill_risk_scores()
diff --git a/testing/backend/unit/test_saved_views.py b/testing/backend/unit/test_saved_views.py
index b196e46b2..e52b55f3c 100644
--- a/testing/backend/unit/test_saved_views.py
+++ b/testing/backend/unit/test_saved_views.py
@@ -9,6 +9,7 @@
 from backend.secuscan.saved_views import saved_views_router
 from backend.secuscan.database import Database, get_db
 import backend.secuscan.database as _db_module
+from pathlib import Path
 
 
 # ─── Fixtures ─────────────────────────────────────────────────────────────────
@@ -442,3 +443,70 @@ def test_malformed_json_raises(self):
             SavedViewCreate(name="v", filter_json="not json")
         # pydantic raises an error for invalid JSON in field_validator
         assert "validation error" in str(exc_info.value).lower()
+
+@pytest.mark.asyncio
+async def test_migrations_are_idempotent(tmp_path):
+    db_file = tmp_path / "idempotent.db"
+
+    db = Database(str(db_file))
+    await db.connect()
+    await db.disconnect()
+
+    db = Database(str(db_file))
+    await db.connect()
+
+    rows = await db.fetchall(
+        "SELECT COUNT(*) AS count FROM schema_migrations"
+    )
+
+    migration_count = len(
+        list((Path(_db_module.__file__).parent / "migrations").glob("*.sql"))
+    )
+
+    assert rows[0]["count"] == migration_count
+
+    await db.disconnect()
+
+
+@pytest.mark.asyncio
+async def test_schema_version_is_recorded(tmp_path):
+    db_file = tmp_path / "schema.db"
+
+    db = Database(str(db_file))
+    await db.connect()
+
+    rows = await db.fetchall(
+        "SELECT version FROM schema_migrations ORDER BY version"
+    )
+
+    assert rows
+
+    assert any(
+        row["version"] == "001_add_performance_indexes.sql"
+        for row in rows
+    )
+
+    await db.disconnect()
+
+
+@pytest.mark.asyncio
+async def test_database_newer_than_application_fails(tmp_path):
+    db_file = tmp_path / "future.db"
+
+    db = Database(str(db_file))
+    await db.connect()
+
+    await db.execute(
+        """
+        INSERT INTO schema_migrations(version)
+        VALUES (?)
+        """,
+        ("999_future.sql",),
+    )
+
+    await db.disconnect()
+
+    db = Database(str(db_file))
+
+    with pytest.raises(RuntimeError, match="Database schema is newer"):
+        await db.connect()