arkindex-base-worker 0.5.1b2__tar.gz → 0.5.1b4__tar.gz

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b2
+Version: 0.5.1b4
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/arkindex_base_worker.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkindex-base-worker
-Version: 0.5.1b2
+Version: 0.5.1b4
 Summary: Base Worker to easily build Arkindex ML workflows
 Author-email: Teklia <contact@teklia.com>
 Maintainer-email: Teklia <contact@teklia.com>

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/arkindex_worker/worker/__init__.py

@@ -33,13 +33,10 @@ from arkindex_worker.worker.transcription import TranscriptionMixin
 
 
 class WorkerActivityIterator:
-    def __init__(self, api_client, types):
+    def __init__(self, api_client):
         # Use same api client as main class
         self.api_client = api_client
 
-        # Index element types by ID
-        self.types = {t["id"]: t["slug"] for t in types}
-
         logger.info(
             "Using StartWorkerActivity instead of reading init_elements JSON file"
         )
@@ -53,7 +50,7 @@ class WorkerActivityIterator:
 
     def __next__(self):
         """
-        Provide a new element from a worker activity upon each iteration
+        Provide a new element ID from a worker activity upon each iteration
         """
         try:
             data = self.api_client.request("StartWorkerActivity")
@@ -67,12 +64,7 @@ class WorkerActivityIterator:
             )
             raise e
 
-        # Find the type slug using API provided type_id
-        type_id = data["type_id"]
-        if type_id not in self.types:
-            raise Exception(f"Unknown type {type_id}")
-
-        return Element(type=self.types[type_id], **data)
+        return data["id"]
 
 
 class ElementsWorker(
@@ -154,14 +146,9 @@ class ElementsWorker(
         elif self.process_mode == ProcessMode.Export:
             # For export mode processes, use list_process_elements and return element IDs
             return {item["id"] for item in self.list_process_elements()}
-        elif self.args.consume_worker_activities:
-            # We need to list corpus types as the StartWorkerActivity endpoint only provide type_id
-            self.list_corpus_types()
-
+        elif self.consume_worker_activities:
             # Consume worker activitives one by one
-            return WorkerActivityIterator(
-                self.api_client, types=self.corpus_types.values()
-            )
+            return WorkerActivityIterator(self.api_client)
 
         invalid_element_ids = list(filter(invalid_element_id, out))
         assert not invalid_element_ids, (
@@ -195,7 +182,7 @@ class ElementsWorker(
          - when running with init_elements, we have a known list
          - when running with StartWorkerActivity, we have a queue of unknown size
         """
-        return self.args.consume_worker_activities
+        return self.consume_worker_activities
 
     def run(self):
         """
@@ -241,7 +228,7 @@ class ElementsWorker(
 
                 # Process the element and report its progress if activities are enabled
                 # We do not update the worker activity to "Started" state when consuming them
-                if self.args.consume_worker_activities or self.update_activity(
+                if self.consume_worker_activities or self.update_activity(
                     element.id, ActivityState.Started
                 ):
                     self.process_element(element)

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/arkindex_worker/worker/element.py

@@ -55,6 +55,17 @@ class ElementMixin:
         )
         super().add_arguments()
 
+    @property
+    def consume_worker_activities(self) -> bool:
+        """
+        Helper to detect if the worker rely on an elements.json or consume directly worker activities
+        Uses the process information when available, fallback to CLI args
+        """
+        if self.process_information is not None:
+            return self.process_information.get("skip_elements_json") is True
+
+        return self.args.consume_worker_activities
+
     def list_corpus_types(self):
         """
         Loads available element types in corpus.

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "arkindex-base-worker"
-version = "0.5.1b2"
+version = "0.5.1b4"
 description = "Base Worker to easily build Arkindex ML workflows"
 license = { file = "LICENSE" }
 dependencies = [

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/tests/conftest.py

@@ -153,6 +153,7 @@ def _mock_worker_run_api(responses):
             "train_folder_id": None,
             "validation_folder_id": None,
             "test_folder_id": None,
+            "skip_elements_json": False,
         },
         "summary": "Worker Fake worker @ 123412",
     }
@@ -346,29 +347,57 @@ def mock_elements_worker_consume_wa(monkeypatch, responses, mock_elements_worker
     instead of reading a JSON file
     """
 
-    # Enable consume worker activities mode from CLI args
-    mock_elements_worker.args.consume_worker_activities = True
-
-    # Worker requires element types from corpus details as they are not provided by StartWorkerActivity
-    responses.add(
+    # Enable consume worker activities through the process configuration
+    responses.replace(
         responses.GET,
-        "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/",
+        "http://testserver/api/v1/process/workers/56785678-5678-5678-5678-567856785678/",
         status=200,
         json={
-            "id": "11111111-1111-1111-1111-111111111111",
-            "name": "Test corpus",
-            "types": [
-                {
-                    "id": "page-aaaa-aaaa-aaaa-aaaaaaaaaaaa",
-                    "slug": "page",
-                    "display_name": "Page",
-                    "folder": False,
-                    "color": "28b62c",
-                }
-            ],
+            "id": "56785678-5678-5678-5678-567856785678",
+            "parents": [],
+            "worker": {
+                "id": "deadbeef-1234-5678-1234-worker",
+                "name": "Fake worker",
+                "slug": "fake_worker",
+                "type": "classifier",
+            },
+            "worker_version": {
+                "id": "12341234-1234-1234-1234-123412341234",
+                "configuration": {
+                    "docker": {"image": "python:3"},
+                    "configuration": {"someKey": "someValue"},
+                    "secrets": [],
+                },
+                "worker": {
+                    "id": "deadbeef-1234-5678-1234-worker",
+                    "name": "Fake worker",
+                    "slug": "fake_worker",
+                    "type": "classifier",
+                },
+            },
+            "configuration": None,
+            "model_version": None,
+            "process": {
+                "name": None,
+                "id": "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeffff",
+                "state": "running",
+                "mode": "workers",
+                "corpus": CORPUS_ID,
+                "use_cache": False,
+                "activity_state": "ready",
+                "model_id": None,
+                "train_folder_id": None,
+                "validation_folder_id": None,
+                "test_folder_id": None,
+                "skip_elements_json": True,
+            },
+            "summary": "Worker Fake worker @ 123412",
         },
     )
 
+    # Call configure again to use updated process infos
+    mock_elements_worker.configure()
+
     return mock_elements_worker
 
 

{arkindex_base_worker-0.5.1b2 → arkindex_base_worker-0.5.1b4}/tests/test_elements_worker/test_worker.py

@@ -698,7 +698,8 @@ def test_run_consuming_worker_activities(
     assert mock_elements_worker_consume_wa.is_read_only is False
 
     # Provide 2 worker activities to run and the corresponding update call
-    for elt_id in ("page_1", "page_2"):
+    # and 2 element details response
+    for i, elt_id in enumerate(("page_1", "page_2"), 1):
         responses.add(
             responses.POST,
             "http://testserver/api/v1/process/start-activity/",
@@ -706,7 +707,7 @@ def test_run_consuming_worker_activities(
             json={
                 "id": elt_id,
                 "type_id": "page-aaaa-aaaa-aaaa-aaaaaaaaaaaa",  # Element type provided by mock corpus
-                "name": "Page XXX",
+                "name": f"Page n°{i}",
             },
         )
         responses.add(
@@ -714,6 +715,16 @@ def test_run_consuming_worker_activities(
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
             status=200,
         )
+        responses.add(
+            responses.GET,
+            f"http://testserver/api/v1/element/{elt_id}/",
+            status=200,
+            json={
+                "id": elt_id,
+                "type": "page",
+                "name": f"Page n°{i}",
+            },
+        )
 
     # Then a 404 to stop iterating
     responses.add(
@@ -725,18 +736,19 @@ def test_run_consuming_worker_activities(
     # Simply run the process
     mock_elements_worker_consume_wa.run()
 
-    assert len(responses.calls) == len(BASE_API_CALLS) + 6
+    # We call twice configure in the conftest
+    assert len(responses.calls) == len(BASE_API_CALLS) * 2 + 7
     assert [
         (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        (
-            "GET",
-            "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/",
-        ),
+    ] == BASE_API_CALLS * 2 + [
         (
             "POST",
             "http://testserver/api/v1/process/start-activity/",
         ),
+        (
+            "GET",
+            "http://testserver/api/v1/element/page_1/",
+        ),
         (
             "PUT",
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
@@ -745,6 +757,10 @@ def test_run_consuming_worker_activities(
             "POST",
             "http://testserver/api/v1/process/start-activity/",
         ),
+        (
+            "GET",
+            "http://testserver/api/v1/element/page_2/",
+        ),
         (
             "PUT",
             "http://testserver/api/v1/workers/versions/56785678-5678-5678-5678-567856785678/activity/",
@@ -756,21 +772,17 @@ def test_run_consuming_worker_activities(
     ]
 
     assert [(record.levelno, record.message) for record in caplog.records] == [
-        (
-            logging.INFO,
-            "Loaded 1 element type in corpus (11111111-1111-1111-1111-111111111111).",
-        ),
         (
             logging.INFO,
             "Using StartWorkerActivity instead of reading init_elements JSON file",
         ),
         (
             logging.INFO,
-            "Processing page Page XXX (page_1) (n°1)",
+            "Processing page Page n°1 (page_1) (n°1)",
         ),
         (
             logging.INFO,
-            "Processing page Page XXX (page_2) (n°2)",
+            "Processing page Page n°2 (page_2) (n°2)",
         ),
         (
             logging.INFO,