Filter worker versions

arkindex/documents/export/__init__.py

@@ -27,12 +27,31 @@ EXPORT_QUERIES = [(
     """
 ), (
     'worker_version',
-    # TODO: Filter worker versions properly
     """
     SELECT version.id, worker.name, worker.slug, worker.type, revision.hash
     FROM dataimport_workerversion version
     INNER JOIN dataimport_worker worker ON (version.worker_id = worker.id)
     INNER JOIN dataimport_revision revision ON (version.revision_id = revision.id)
+    WHERE version.id IN (
+        SELECT worker_version_id FROM documents_element WHERE corpus_id = '{corpus_id}'::uuid
+    UNION
+        SELECT worker_version_id FROM documents_entity WHERE corpus_id = '{corpus_id}'::uuid
+    UNION
+        SELECT classification.worker_version_id
+        FROM documents_classification classification
+        INNER JOIN documents_element element ON (element.id = classification.element_id)
+        WHERE element.corpus_id = '{corpus_id}'::uuid
+    UNION
+        SELECT transcription.worker_version_id
+        FROM documents_transcription transcription
+        INNER JOIN documents_element element ON (element.id = transcription.element_id)
+        WHERE element.corpus_id = '{corpus_id}'::uuid
+    UNION
+        SELECT te.worker_version_id
+        FROM documents_transcriptionentity te
+        INNER JOIN documents_entity entity ON (te.entity_id = entity.id)
+        WHERE entity.corpus_id = '{corpus_id}'::uuid
+    )
     """
 ), (
     'element',