lint(api): add start method to worker pool

api/onnx_web/main.py

@@ -46,14 +46,16 @@ def main():
         disable_progress_bar()
         disable_progress_bars()
 
-    app = Flask(__name__)
-    CORS(app, origins=context.cors_origin)
-
-    # any is a fake device, should not be in the pool
+    # create workers
+    # any is a fake device and should not be in the pool
     pool = DevicePoolExecutor(
         context, [p for p in get_available_platforms() if p.device != "any"]
     )
 
+    # create server
+    app = Flask(__name__)
+    CORS(app, origins=context.cors_origin)
+
     # register routes
     register_static_routes(app, context, pool)
     register_api_routes(app, context, pool)
@@ -63,6 +65,7 @@ def main():
 
 def run():
     app, pool = main()
+    pool.start()
 
     def quit():
         logger.info("shutting down workers")
@@ -74,6 +77,7 @@ def run():
 
 if __name__ == "__main__":
     app, pool = main()
+    pool.start()
     app.run("0.0.0.0", 5000, debug=is_debug())
     logger.info("shutting down app")
     pool.join()

api/onnx_web/worker/pool.py

@@ -80,12 +80,12 @@ class DevicePoolExecutor:
         self.logs = Queue(self.max_pending_per_worker)
         self.rlock = Lock()
 
-        # TODO: these should be part of a start method
+    def start(self) -> None:
         self.create_health_worker()
         self.create_logger_worker()
         self.create_progress_worker()
 
-        for device in devices:
+        for device in self.devices:
             self.create_device_worker(device)
 
     def create_device_worker(self, device: DeviceParams) -> None:
@@ -439,7 +439,11 @@ class DevicePoolExecutor:
                 else:
                     self.total_jobs[progress.device] = 1
 
-                logger.debug("updating job count for device %s: %s", progress.device, self.total_jobs[progress.device])
+                logger.debug(
+                    "updating job count for device %s: %s",
+                    progress.device,
+                    self.total_jobs[progress.device],
+                )
 
             # check if the job has been cancelled
             if progress.job in self.cancelled_jobs: