Merge pull request #294 from uhh-lt/fix-database-in-tests

Give tests their own database

.editorconfig

@@ -1,5 +1,8 @@
 root = true
 
+[*]
+insert_final_newline = true
+
 [*.{yml,yaml}]
 indent_size = 2
 indent_style = space

.github/workflows/backend_check_schema.yml

@@ -9,6 +9,10 @@ on:
 jobs:
   check-schema:
     runs-on: ubuntu-latest
+    env:
+      API_PRODUCTION_WORKERS: 1
+      RAY_ENABLED: False
+      COMPOSE_PROFILES: ""
     steps:
       - uses: actions/checkout@v3
       - name: Start Docker Containers
@@ -19,8 +23,6 @@ jobs:
           chmod -R a+rwx backend_repo/ models_cache/ spacy_models/
           python monkey_patch_docker_compose_for_backend_tests.py --disable_ray
           export GID=$(id -g)
-          export API_PRODUCTION_WORKERS=0
-          export RAY_CONFIG="./config_test_no_gpu.yaml"
           docker compose -f compose-test.yml up -d --quiet-pull postgres
           echo Waiting for containers to start...
           sleep 10

.github/workflows/backend_e2e_tests.yml

@@ -84,7 +84,7 @@ jobs:
           chmod -R a+rwx backend_repo/ models_cache/ spacy_models/
           python monkey_patch_docker_compose_for_backend_tests.py
           export GID=$(id -g)
-          export API_PRODUCTION_WORKERS=0
+          export API_PRODUCTION_WORKERS=1
           export RAY_CONFIG="./config_test_no_gpu.yaml"
           docker compose -f compose-test.yml up -d
           echo Waiting for containers to start...

.github/workflows/backend_unit_tests.yml

@@ -47,6 +47,12 @@ jobs:
           echo "Environment: $ENVIRONMENT"
   run-unit-tests:
     runs-on: ubuntu-latest
+    environment:
+      API_PRODUCTION_WORKERS: 1
+      # disable backend and frontend
+      COMPOSE_PROFILES: ""
+      RAY_ENABLED: False
+      POSTRES_DB: dwts-test
     steps:
       - name: Set Swap Space to 10GB
         uses: pierotofy/set-swap-space@master
@@ -61,10 +67,6 @@ jobs:
           chmod -R a+rwx backend_repo/ models_cache/ spacy_models/
           python monkey_patch_docker_compose_for_backend_tests.py --disable_ray
           export GID=$(id -g)
-          export API_PRODUCTION_WORKERS=0
-          # disable backend and frontend
-          export COMPOSE_PROFILES=
-          export RAY_ENABLED=False
           docker compose -f compose-test.yml up -d --quiet-pull
           echo Waiting for containers to start...
           sleep 30

.github/workflows/update-openapi-spec.yml

@@ -10,6 +10,10 @@ on:
 jobs:
   update-openapi-spec:
     runs-on: ubuntu-latest
+    env:
+      API_PRODUCTION_WORKERS: 1
+      RAY_ENABLED: False
+      API_EXPOSED: 5500
     steps:
       - uses: actions/checkout@v3
       - name: Checkout PR
@@ -24,9 +28,6 @@ jobs:
           chmod -R a+rwx backend_repo/ models_cache/ spacy_models/
           python monkey_patch_docker_compose_for_backend_tests.py --disable_ray
           export GID=$(id -g)
-          export API_PRODUCTION_WORKERS=0
-          export RAY_ENABLED=False
-          export API_EXPOSED=5500
           docker compose -f compose-test.yml up -d --quiet-pull
           echo Waiting for containers to start...
       - name: Setup node

backend/.env.testing.example

@@ -0,0 +1,8 @@
+# Copy this file to .env.testing and edit it to match your preferences
+# Use it in conjunction with your normal env:
+# source .env; source .env.testing
+# Before running your tests.
+# This way, you can keep only necessary overrides in .env.testing
+
+RAY_ENABLED=False
+POSTGRES_DB=dwts-testing

backend/.gitignore

@@ -12,6 +12,7 @@ backend/logs
 
 # Local configuration
 .env
+.env.testing
 
 # npm stuff for python readability
 src/node_modules

backend/src/app/core/db/sql_service.py

@@ -3,12 +3,11 @@
 
 from loguru import logger
 from pydantic import PostgresDsn
-from sqlalchemy import create_engine
+from sqlalchemy import create_engine, inspect
 from sqlalchemy.engine import Engine
 from sqlalchemy.orm import Session, sessionmaker
 from sqlalchemy_utils import create_database, database_exists, drop_database
 
-from app.core.data.orm.orm_base import ORMBase
 from app.core.db.import_all_orms import *  # noqa: F401, F403
 from app.util.singleton_meta import SingletonMeta
 from config import conf
@@ -37,6 +36,10 @@ def __new__(cls, *args, **kwargs):
             cls.__engine: Engine = engine
             cls.session_maker = sessionmaker(autoflush=False, bind=engine)
 
+            if kwargs.get("reset_database") is True:
+                logger.warning("Dropping existing DB!")
+                drop_database(cls.__engine.url)
+
             return super(SQLService, cls).__new__(cls)
 
         except Exception as e:
@@ -47,23 +50,29 @@ def __new__(cls, *args, **kwargs):
     def __del__(self):
         self.__engine.dispose()
 
-    # This method is unused and only left here for historic reference
-    def _create_database_and_tables(self, drop_if_exists: bool = False) -> None:
-        logger.info("Setting up PostgresSQL DB and tables...")
-        if drop_if_exists and database_exists(self.__engine.url):
-            logger.warning("Dropping existing DB!")
-            drop_database(self.__engine.url)
+    def drop_database(self):
+        logger.warning("Dropping existing DB!")
+        drop_database(self.__engine.url)
 
+    def create_database_if_not_exists(self):
         if not database_exists(self.__engine.url):
             # create the DB
             create_database(self.__engine.url)
             logger.debug("Created DB!")
 
-            # create all tables from SQLAlchemy ORM Models
-            ORMBase.metadata.create_all(self.__engine)
-            logger.debug("Created Tables!")
+    def database_contains_data(self):
+        if not database_exists(self.__engine.url):
+            return False
+
+        inspector = inspect(self.__engine)
+        schemas = inspector.get_schema_names()
+
+        for schema in schemas:
+            print("schema: %s" % schema)
+            if len(inspector.get_table_names(schema=schema)) > 0:
+                return True
 
-        logger.info("Done setting up PostgresSQL DB and tables!")
+        return False
 
     @contextmanager
     def db_session(self) -> Generator[Session, None, None]:

backend/src/app/core/startup.py

@@ -99,10 +99,10 @@ def __init_services__(
     repos = RepoService()
     if create_root_repo_directory_structure:
         repos._create_root_repo_directory_structure(remove_if_exists=reset_repo)
-    # create SQL DBs and Tables # TODO Flo: Alembic
+    # create SQL DBs and Tables
     from app.core.db.sql_service import SQLService
 
-    SQLService(echo=sql_echo)
+    SQLService(echo=sql_echo, reset_database=reset_database)
     # import and init ElasticSearch
     from app.core.search.elasticsearch_service import ElasticSearchService
 

backend/src/migration/migrate.py

@@ -78,6 +78,7 @@ def run_required_migrations():
 
 
 def __migrate_database_schema() -> None:
+    SQLService().create_database_if_not_exists()
     config = Config("alembic.ini")
     upgrade(config, "head")
     print("MIGRATED DB SCHEMA!")

backend/src/test/conftest.py

@@ -11,17 +11,27 @@
 import pytest
 from fastapi import Request
 from fastapi.datastructures import Headers
+from loguru import logger
 
 from app.core.authorization.authz_user import AuthzUser
 from app.core.data.orm.project import ProjectORM
+from app.core.db.sql_service import SQLService
 from app.core.startup import startup
+from config import conf
 
 os.environ["RAY_ENABLED"] = "False"
 
 # Flo: just do it once. We have to check because if we start the main function, unvicorn will import this
 # file once more manually, so it would be executed twice.
 STARTUP_DONE = bool(int(os.environ.get("STARTUP_DONE", "0")))
 if not STARTUP_DONE:
+    if SQLService().database_contains_data():
+        # Make sure we don't accidentally delete important data
+        logger.error(
+            f"Database '{conf.postgres.db}' is not empty. The tests will only run given a database without any tables in it."
+        )
+        exit(1)
+
     startup(reset_data=True)
     os.environ["STARTUP_DONE"] = "1"
 
@@ -31,7 +41,11 @@
 from app.core.data.dto.code import CodeCreate, CodeRead
 from app.core.data.dto.project import ProjectCreate
 from app.core.data.dto.user import UserCreate, UserRead
-from app.core.db.sql_service import SQLService
+
+
+def pytest_sessionfinish():
+    # Make sure the next test session starts with a clean database
+    SQLService().drop_database()
 
 
 # Always use the asyncio backend for async tests

docker/docker-compose.yml

@@ -6,7 +6,6 @@ services:
     environment:
       POSTGRES_USER: ${POSTGRES_USER:-dwtsuser}
       POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-dwts123}
-      POSTGRES_DB: dwts
     ports:
       - "${POSTGRES_EXPOSED:-5432}:5432"
     volumes: