Handle worker versions in transcription entities

arkindex_worker/cache.py

@@ -156,6 +156,7 @@ class CachedTranscriptionEntity(Model):
     entity = ForeignKeyField(CachedEntity, backref="transcription_entities")
     offset = IntegerField(constraints=[Check("offset >= 0")])
     length = IntegerField(constraints=[Check("length > 0")])
+    worker_version_id = UUIDField()
 
     class Meta:
         primary_key = CompositeKey("transcription", "entity")

arkindex_worker/worker/entity.py

@@ -111,6 +111,7 @@ class EntityMixin(object):
                 "entity": entity,
                 "length": length,
                 "offset": offset,
+                "worker_version_id": self.worker_version_id,
             },
         )
         # TODO: Report transcription entity creation
@@ -118,15 +119,13 @@ class EntityMixin(object):
         if self.use_cache:
             # Store transcription entity in local cache
             try:
-                to_insert = [
-                    {
-                        "transcription": transcription,
-                        "entity": entity,
-                        "offset": offset,
-                        "length": length,
-                    }
-                ]
-                CachedTranscriptionEntity.insert_many(to_insert).execute()
+                CachedTranscriptionEntity.create(
+                    transcription=transcription,
+                    entity=entity,
+                    offset=offset,
+                    length=length,
+                    worker_version_id=self.worker_version_id,
+                )
             except IntegrityError as e:
                 logger.warning(
                     f"Couldn't save created transcription entity in local cache: {e}"

tests/test_cache.py

@@ -58,7 +58,7 @@ def test_create_tables(tmp_path):
 CREATE TABLE "elements" ("id" TEXT NOT NULL PRIMARY KEY, "parent_id" TEXT, "type" VARCHAR(50) NOT NULL, "image_id" TEXT, "polygon" text, "initial" INTEGER NOT NULL, "worker_version_id" TEXT, FOREIGN KEY ("image_id") REFERENCES "images" ("id"))
 CREATE TABLE "entities" ("id" TEXT NOT NULL PRIMARY KEY, "type" VARCHAR(50) NOT NULL, "name" TEXT NOT NULL, "validated" INTEGER NOT NULL, "metas" text, "worker_version_id" TEXT NOT NULL)
 CREATE TABLE "images" ("id" TEXT NOT NULL PRIMARY KEY, "width" INTEGER NOT NULL, "height" INTEGER NOT NULL, "url" TEXT NOT NULL)
-CREATE TABLE "transcription_entities" ("transcription_id" TEXT NOT NULL, "entity_id" TEXT NOT NULL, "offset" INTEGER NOT NULL CHECK (offset >= 0), "length" INTEGER NOT NULL CHECK (length > 0), PRIMARY KEY ("transcription_id", "entity_id"), FOREIGN KEY ("transcription_id") REFERENCES "transcriptions" ("id"), FOREIGN KEY ("entity_id") REFERENCES "entities" ("id"))
+CREATE TABLE "transcription_entities" ("transcription_id" TEXT NOT NULL, "entity_id" TEXT NOT NULL, "offset" INTEGER NOT NULL CHECK (offset >= 0), "length" INTEGER NOT NULL CHECK (length > 0), "worker_version_id" TEXT NOT NULL, PRIMARY KEY ("transcription_id", "entity_id"), FOREIGN KEY ("transcription_id") REFERENCES "transcriptions" ("id"), FOREIGN KEY ("entity_id") REFERENCES "entities" ("id"))
 CREATE TABLE "transcriptions" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "text" TEXT NOT NULL, "confidence" REAL NOT NULL, "worker_version_id" TEXT NOT NULL, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))"""
 
     actual_schema = "\n".join(

tests/test_elements_worker/test_entities.py

@@ -449,6 +449,7 @@ def test_create_transcription_entity(responses, mock_elements_worker):
         "entity": "11111111-1111-1111-1111-111111111111",
         "offset": 5,
         "length": 10,
+        "worker_version_id": "12341234-1234-1234-1234-123412341234",
     }
 
 
@@ -504,6 +505,7 @@ def test_create_transcription_entity_with_cache(
         "entity": "11111111-1111-1111-1111-111111111111",
         "offset": 5,
         "length": 10,
+        "worker_version_id": "12341234-1234-1234-1234-123412341234",
     }
 
     # Check that created transcription entity was properly stored in SQLite cache
@@ -513,5 +515,6 @@ def test_create_transcription_entity_with_cache(
             entity=UUID("11111111-1111-1111-1111-111111111111"),
             offset=5,
             length=10,
+            worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
         )
     ]