Turn on profiling when FLASK_PROFILE=True

CONTRIBUTING.md

@@ -69,6 +69,29 @@ npm run lint
 npm run format
 ```
 
+## Profiling
+
+If you need to do some performance profiling you can run the debug server with
+`FLASK_PROFILE=True`. You’ll need to first create a directory for the profile
+files:
+
+```
+mkdir -p tmp/profiler
+FLASK_PROFILE=True poetry run flask run --debug
+```
+
+There are multiple ways you can view the `.prof` files generated by the
+profiling middleware. One easy way to view them is with snakeviz.
+
+```
+poetry run pip install snakeviz
+poetry run snakeviz tmp/profiler/<endpoint>.prof
+```
+
+That first line installs snakeviz in your virtual env (without adding it as a
+dependency to the project) and the second will load a performance profile into
+your default browser for you to view.
+
 ## Coding Style
 
 ### Python

src/unified_graphics/__init__.py

@@ -35,6 +35,17 @@ def create_app(config: Optional[Mapping[str, Any]] = None):
     app.config.from_prefixed_env()
     app.config.from_mapping(config)
 
+    if app.debug and app.config.get("PROFILE", "False") == "True":
+        from werkzeug.middleware.profiler import ProfilerMiddleware
+
+        app.wsgi_app = ProfilerMiddleware(  # type: ignore
+            app.wsgi_app,
+            restrictions=[30],
+            stream=None,  # Disable stdout
+            profile_dir="tmp/profiler",
+            filename_format="{method}-{path}-{time:.0f}-{elapsed:.0f}ms.prof",
+        )
+
     models.db.init_app(app)
 
     app.register_blueprint(routes.bp)