Merge branch 'export-metadata-version' into 'master'

Include worker versions from metadata in corpus export

Closes #818

See merge request !1449

arkindex/documents/export/worker_version.sql

@@ -26,4 +26,9 @@ UNION
     FROM documents_transcriptionentity te
     INNER JOIN documents_entity entity ON (te.entity_id = entity.id)
     WHERE entity.corpus_id = '{corpus_id}'::uuid
+UNION
+    SELECT md.worker_version_id
+    FROM documents_metadata md
+    INNER JOIN documents_element element ON (md.element_id = element.id)
+    WHERE element.corpus_id = '{corpus_id}'::uuid
 )

arkindex/documents/tests/tasks/test_export.py

@@ -4,11 +4,12 @@ import sqlite3
 from datetime import datetime, timezone
 from textwrap import dedent
 from unittest.mock import call, patch
+from uuid import uuid4
 
 from django.core import mail
 from django.test import override_settings
 
-from arkindex.dataimport.models import WorkerVersion
+from arkindex.dataimport.models import Repository, RepositoryType, WorkerVersion, WorkerVersionState
 from arkindex.documents.export import export_corpus
 from arkindex.documents.models import (
     Classification,
@@ -17,15 +18,33 @@ from arkindex.documents.models import (
     EntityLink,
     EntityType,
     MetaData,
+    MetaType,
     Transcription,
     TranscriptionEntity,
 )
 from arkindex.images.models import Image
 from arkindex.project.tests import FixtureTestCase
+from ponos.models import Artifact
 
 
 class TestExport(FixtureTestCase):
 
+    @classmethod
+    def setUpTestData(cls):
+        super().setUpTestData()
+        cls.repo = Repository.objects.get(type=RepositoryType.Worker)
+        cls.revision = cls.repo.revisions.first()
+        # The fixtures have two worker versions, only one of them is used in existing elements
+        cls.recognizer = cls.repo.workers.get(slug='reco').versions.get()
+
+    def _make_worker_version(self):
+        return self.revision.versions.create(
+            worker=self.repo.workers.create(slug=str(uuid4())),
+            configuration={},
+            state=WorkerVersionState.Available,
+            docker_image=Artifact.objects.first(),
+        )
+
     @override_settings(PUBLIC_HOSTNAME='https://darkindex.lol')
     @patch('arkindex.documents.export.datetime')
     @patch('arkindex.documents.export.os.unlink')
@@ -37,6 +56,15 @@ class TestExport(FixtureTestCase):
         transcription = Transcription.objects.first()
         version = WorkerVersion.objects.get(worker__slug='reco')
 
+        metadata_version = self._make_worker_version()
+        element.metadatas.create(
+            type=MetaType.Numeric,
+            name='Score',
+            value=1,
+            index=0,
+            worker_version=metadata_version,
+        )
+
         element.classifications.create(
             ml_class=self.corpus.ml_classes.create(name='Blah'),
             confidence=.55555555,
@@ -117,6 +145,13 @@ class TestExport(FixtureTestCase):
                     version.worker.slug,
                     version.worker.type,
                     version.revision.hash
+                ),
+                (
+                    str(metadata_version.id),
+                    metadata_version.worker.name,
+                    metadata_version.worker.slug,
+                    metadata_version.worker.type,
+                    metadata_version.revision.hash
                 )
             ]
         )