arkindex_worker/reporting.py

@@ -10,11 +10,13 @@ from arkindex_worker import logger
 
 
 class Reporter(object):
-    def __init__(self, name):
-        # TODO: use real data from workers
+    def __init__(
+        self, name="Unknown worker", slug="unknown-slug", version=None, **kwargs
+    ):
         self.report_data = {
-            "slug": name,
-            "version": "0.0",
+            "name": name,
+            "slug": slug,
+            "version": version,
             "started": datetime.utcnow().isoformat(),
             "elements": {},
         }

arkindex_worker/worker/__init__.py

@@ -39,9 +39,6 @@ class ElementsWorker(
     def __init__(self, description="Arkindex Elements Worker", support_cache=False):
         super().__init__(description, support_cache)
 
-        # Add report concerning elements
-        self.report = Reporter("unknown worker")
-
         # Add mandatory argument to process elements
         self.parser.add_argument(
             "--elements-list",
@@ -93,6 +90,13 @@ class ElementsWorker(
         ), "Worker must be configured to access its process activity state"
         return self.process_information.get("activity_state") == "ready"
 
+    def configure(self):
+        super().configure()
+        # Add report concerning elements
+        self.report = Reporter(
+            **self.worker_details, version=getattr(self, "worker_version_id", None)
+        )
+
     def run(self):
         """
         Process every elements from the provided list

arkindex_worker/worker/base.py

@@ -145,16 +145,19 @@ class BaseWorker(object):
                 f"Loaded worker {worker_version['worker']['name']} revision {worker_version['revision']['hash'][0:7]} from API"
             )
             self.config = worker_version["configuration"]["configuration"]
+            self.worker_details = worker_version["worker"]
             required_secrets = worker_version["configuration"].get("secrets", [])
         elif self.args.config:
             # Load config from YAML file
             self.config = yaml.safe_load(self.args.config)
+            self.worker_details = {"name": "Local worker"}
             required_secrets = self.config.get("secrets", [])
             logger.info(
                 f"Running with local configuration from {self.args.config.name}"
             )
         else:
             self.config = {}
+            self.worker_details = {}
             required_secrets = []
             logger.warning("Running without any extra configuration")
 

arkindex_worker/worker/element.py

@@ -188,12 +188,11 @@ class ElementMixin(object):
     def list_element_children(
         self,
         element,
-        best_class=None,
         folder=None,
         name=None,
         recursive=None,
         type=None,
-        with_best_classes=None,
+        with_classes=None,
         with_corpus=None,
         with_has_children=None,
         with_zone=None,
@@ -206,11 +205,6 @@ class ElementMixin(object):
             element, (Element, CachedElement)
         ), "element shouldn't be null and should be an Element or CachedElement"
         query_params = {}
-        if best_class is not None:
-            assert isinstance(best_class, str) or isinstance(
-                best_class, bool
-            ), "best_class should be of type str or bool"
-            query_params["best_class"] = best_class
         if folder is not None:
             assert isinstance(folder, bool), "folder should be of type bool"
             query_params["folder"] = folder
@@ -223,11 +217,9 @@ class ElementMixin(object):
         if type:
             assert isinstance(type, str), "type should be of type str"
             query_params["type"] = type
-        if with_best_classes is not None:
-            assert isinstance(
-                with_best_classes, bool
-            ), "with_best_classes should be of type bool"
-            query_params["with_best_classes"] = with_best_classes
+        if with_classes is not None:
+            assert isinstance(with_classes, bool), "with_classes should be of type bool"
+            query_params["with_classes"] = with_classes
         if with_corpus is not None:
             assert isinstance(with_corpus, bool), "with_corpus should be of type bool"
             query_params["with_corpus"] = with_corpus

tests/test_elements_worker/test_elements.py

@@ -976,17 +976,6 @@ def test_list_element_children_wrong_element(mock_elements_worker):
     )
 
 
-def test_list_element_children_wrong_best_class(mock_elements_worker):
-    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
-
-    with pytest.raises(AssertionError) as e:
-        mock_elements_worker.list_element_children(
-            element=elt,
-            best_class=1234,
-        )
-    assert str(e.value) == "best_class should be of type str or bool"
-
-
 def test_list_element_children_wrong_folder(mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
@@ -1031,15 +1020,15 @@ def test_list_element_children_wrong_type(mock_elements_worker):
     assert str(e.value) == "type should be of type str"
 
 
-def test_list_element_children_wrong_with_best_classes(mock_elements_worker):
+def test_list_element_children_wrong_with_classes(mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
 
     with pytest.raises(AssertionError) as e:
         mock_elements_worker.list_element_children(
             element=elt,
-            with_best_classes="not bool",
+            with_classes="not bool",
         )
-    assert str(e.value) == "with_best_classes should be of type bool"
+    assert str(e.value) == "with_classes should be of type bool"
 
 
 def test_list_element_children_wrong_with_corpus(mock_elements_worker):

tests/test_reporting.py

 # -*- coding: utf-8 -*-
+import json
+import uuid
+from datetime import datetime
+from tempfile import NamedTemporaryFile
+
 import pytest
 from apistar.exceptions import ErrorResponse
 
@@ -6,10 +11,16 @@ from arkindex_worker.reporting import Reporter
 
 
 def test_init():
-    reporter = Reporter("worker")
+    version_id = str(uuid.uuid4())
+    reporter = Reporter(name="Worker", slug="worker-slug", version=version_id)
     assert "started" in reporter.report_data
     del reporter.report_data["started"]
-    assert reporter.report_data == {"slug": "worker", "version": "0.0", "elements": {}}
+    assert reporter.report_data == {
+        "name": "Worker",
+        "slug": "worker-slug",
+        "version": version_id,
+        "elements": {},
+    }
 
 
 def test_process():
@@ -213,3 +224,32 @@ def test_error():
             "content": '{"coffee": "Can\'t touch this"}',
         },
     ]
+
+
+def test_reporter_save(mocker):
+    datetime_mock = mocker.MagicMock()
+    datetime_mock.utcnow.return_value = datetime(2000, 1, 1)
+    mocker.patch("arkindex_worker.reporting.datetime", datetime_mock)
+    version_id = str(uuid.uuid4())
+    reporter = Reporter(name="Worker", slug="worker-slug", version=version_id)
+    reporter.add_element("myelement", type="text_line", type_count=4)
+    with NamedTemporaryFile() as f:
+        reporter.save(f.name)
+        exported_data = json.load(f)
+    assert exported_data == {
+        "name": "Worker",
+        "slug": "worker-slug",
+        "started": "2000-01-01T00:00:00",
+        "version": version_id,
+        "elements": {
+            "myelement": {
+                "classifications": {},
+                "elements": {"text_line": 4},
+                "entities": [],
+                "errors": [],
+                "metadata": [],
+                "started": "2000-01-01T00:00:00",
+                "transcriptions": 0,
+            }
+        },
+    }