Load configuration from API or local file

arkindex_worker/worker.py

@@ -7,6 +7,7 @@ import sys
 import uuid
 from enum import Enum
 
+import yaml
 from apistar.exceptions import ErrorResponse
 
 from arkindex import ArkindexClient, options_from_env
@@ -33,16 +34,27 @@ class BaseWorker(object):
 
         self.worker_version_id = os.environ.get("WORKER_VERSION_ID")
         if not self.worker_version_id:
-            raise Exception(
-                "Missing WORKER_VERSION_ID environment variable to start the Worker"
+            logger.warning(
+                "Missing WORKER_VERSION_ID environment variable, worker is in read-only mode"
             )
 
         logger.info(f"Worker will use {self.work_dir} as working directory")
 
+    @property
+    def is_read_only(self):
+        """Worker cannot publish anything without a worker version ID"""
+        return self.worker_version_id is None
+
     def configure(self):
         """
         Configure worker using cli args and environment variables
         """
+        self.parser.add_argument(
+            "-c",
+            "--config",
+            help="Alternative configuration file when running without a Worker Version ID",
+            type=open,
+        )
         self.parser.add_argument(
             "-v",
             "--verbose",
@@ -66,6 +78,25 @@ class BaseWorker(object):
         self.api_client = ArkindexClient(**options_from_env())
         logger.debug(f"Setup Arkindex API client on {self.api_client.document.url}")
 
+        if self.worker_version_id:
+            # Retrieve initial configuration from API
+            worker_version = self.api_client.request(
+                "RetrieveWorkerVersion", id=self.worker_version_id
+            )
+            logger.info(
+                f"Loaded worker {worker_version['worker']['name']} revision {worker_version['revision']} from API"
+            )
+            self.config = worker_version["configuration"]
+        elif self.args.config:
+            # Load config from YAML file
+            self.config = yaml.safe_load(self.args.config)
+            logger.info(
+                f"Running with local configuration from {self.args.config.name}"
+            )
+        else:
+            self.config = {}
+            logger.warning("Running without any extra configuration")
+
     def add_arguments(self):
         """Override this method to add argparse argument to this worker"""
 
@@ -208,6 +239,9 @@ class ElementsWorker(BaseWorker):
         assert all(
             isinstance(coord, (int, float)) for point in polygon for coord in point
         ), "polygon points should be lists of two numbers"
+        if self.is_read_only:
+            logger.warning("Cannot create element as this worker is in read-only mode")
+            return
 
         sub_element = self.api_client.request(
             "CreateElement",
@@ -241,6 +275,11 @@ class ElementsWorker(BaseWorker):
         assert (
             score and isinstance(score, float) and 0 <= score <= 1
         ), "score shouldn't be null and should be a float in [0..1] range"
+        if self.is_read_only:
+            logger.warning(
+                "Cannot create transcription as this worker is in read-only mode"
+            )
+            return
 
         self.api_client.request(
             "CreateTranscription",
@@ -272,6 +311,11 @@ class ElementsWorker(BaseWorker):
         assert high_confidence and isinstance(
             high_confidence, bool
         ), "high_confidence shouldn't be null and should be of type bool"
+        if self.is_read_only:
+            logger.warning(
+                "Cannot create classification as this worker is in read-only mode"
+            )
+            return
 
         self.api_client.request(
             "CreateClassification",
@@ -306,6 +350,9 @@ class ElementsWorker(BaseWorker):
             assert isinstance(metas, dict), "metas should be of type dict"
         if validated:
             assert isinstance(validated, bool), "validated should be of type bool"
+        if self.is_read_only:
+            logger.warning("Cannot create entity as this worker is in read-only mode")
+            return
 
         entity = self.api_client.request(
             "CreateEntity",
@@ -365,6 +412,11 @@ class ElementsWorker(BaseWorker):
             assert all(
                 isinstance(coord, (int, float)) for point in polygon for coord in point
             ), f"Transcription at index {index} in transcriptions: polygon points should be lists of two numbers"
+        if self.is_read_only:
+            logger.warning(
+                "Cannot create transcriptions as this worker is in read-only mode"
+            )
+            return
 
         annotations = self.api_client.request(
             "CreateElementTranscriptions",