arkindex-base-worker 0.5.0b3__py3-none-any.whl → 0.5.0rc1__py3-none-any.whl

tests/test_elements_worker/test_metadata.py

@@ -119,19 +119,6 @@ def test_create_metadata_wrong_value(mock_elements_worker):
         )
 
 
-def test_create_metadata_wrong_entity(mock_elements_worker):
-    elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
-
-    with pytest.raises(AssertionError, match="entity should be of type str"):
-        mock_elements_worker.create_metadata(
-            element=elt,
-            type=MetaType.Location,
-            name="Teklia",
-            value="La Turbine, Grenoble 38000",
-            entity=1234,
-        )
-
-
 def test_create_metadata_api_error(responses, mock_elements_worker):
     elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
     responses.add(
@@ -188,7 +175,6 @@ def test_create_metadata(responses, mock_elements_worker):
         "type": "location",
         "name": "Teklia",
         "value": "La Turbine, Grenoble 38000",
-        "entity_id": None,
         "worker_run_id": "56785678-5678-5678-5678-567856785678",
     }
     assert metadata_id == "12345678-1234-1234-1234-123456789123"
@@ -223,7 +209,6 @@ def test_create_metadata_cached_element(responses, mock_elements_worker_with_cac
         "type": "location",
         "name": "Teklia",
         "value": "La Turbine, Grenoble 38000",
-        "entity_id": None,
         "worker_run_id": "56785678-5678-5678-5678-567856785678",
     }
     assert metadata_id == "12345678-1234-1234-1234-123456789123"
@@ -239,7 +224,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
             "type": MetaType.Text,
             "name": "Year",
             "value": "2024",
-            "entity_id": "entity_id",
         },
     ]
     if batch_size > 1:
@@ -256,7 +240,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
                         "name": metadata_list[0]["name"],
                         "value": metadata_list[0]["value"],
                         "dates": [],
-                        "entity_id": None,
                     },
                     {
                         "id": "fake_metadata_id2",
@@ -264,7 +247,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
                         "name": metadata_list[1]["name"],
                         "value": metadata_list[1]["value"],
                         "dates": [],
-                        "entity_id": metadata_list[1]["entity_id"],
                     },
                 ],
             },
@@ -284,7 +266,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
                             "name": meta["name"],
                             "value": meta["value"],
                             "dates": [],
-                            "entity_id": meta.get("entity_id"),
                         }
                     ],
                 },
@@ -316,7 +297,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
     first_meta = {
         **metadata_list[0],
         "type": metadata_list[0]["type"].value,
-        "entity_id": None,
     }
     second_meta = {**metadata_list[1], "type": metadata_list[1]["type"].value}
     empty_payload = {
@@ -346,7 +326,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
             "name": metadata_list[0]["name"],
             "value": metadata_list[0]["value"],
             "dates": [],
-            "entity_id": None,
         },
         {
             "id": "fake_metadata_id2",
@@ -354,7 +333,6 @@ def test_create_metadata_bulk(batch_size, responses, mock_elements_worker):
             "name": metadata_list[1]["name"],
             "value": metadata_list[1]["value"],
             "dates": [],
-            "entity_id": metadata_list[1]["entity_id"],
         },
     ]
 
@@ -373,7 +351,6 @@ def test_create_metadata_bulk_cached_element(
             "type": MetaType.Text,
             "name": "Year",
             "value": "2024",
-            "entity_id": "entity_id",
         },
     ]
     if batch_size > 1:
@@ -390,7 +367,6 @@ def test_create_metadata_bulk_cached_element(
                         "name": metadata_list[0]["name"],
                         "value": metadata_list[0]["value"],
                         "dates": [],
-                        "entity_id": None,
                     },
                     {
                         "id": "fake_metadata_id2",
@@ -398,7 +374,6 @@ def test_create_metadata_bulk_cached_element(
                         "name": metadata_list[1]["name"],
                         "value": metadata_list[1]["value"],
                         "dates": [],
-                        "entity_id": metadata_list[1]["entity_id"],
                     },
                 ],
             },
@@ -418,7 +393,6 @@ def test_create_metadata_bulk_cached_element(
                             "name": meta["name"],
                             "value": meta["value"],
                             "dates": [],
-                            "entity_id": meta.get("entity_id"),
                         }
                     ],
                 },
@@ -450,7 +424,6 @@ def test_create_metadata_bulk_cached_element(
     first_meta = {
         **metadata_list[0],
         "type": metadata_list[0]["type"].value,
-        "entity_id": None,
     }
     second_meta = {**metadata_list[1], "type": metadata_list[1]["type"].value}
     empty_payload = {
@@ -480,7 +453,6 @@ def test_create_metadata_bulk_cached_element(
             "name": metadata_list[0]["name"],
             "value": metadata_list[0]["value"],
             "dates": [],
-            "entity_id": None,
         },
         {
             "id": "fake_metadata_id2",
@@ -488,7 +460,6 @@ def test_create_metadata_bulk_cached_element(
             "name": metadata_list[1]["name"],
             "value": metadata_list[1]["value"],
             "dates": [],
-            "entity_id": metadata_list[1]["entity_id"],
         },
     ]
 
@@ -552,23 +523,6 @@ def test_create_metadata_bulk_wrong_value(mock_elements_worker, wrong_value):
         )
 
 
-@pytest.mark.parametrize("wrong_entity", [[1, 2, 3, 4], 1234, 12.5])
-def test_create_metadata_bulk_wrong_entity(mock_elements_worker, wrong_entity):
-    element = Element({"id": "fake_element_id"})
-    wrong_metadata_list = [
-        {
-            "type": MetaType.Text,
-            "name": "fake_name",
-            "value": "fake_value",
-            "entity_id": wrong_entity,
-        }
-    ]
-    with pytest.raises(AssertionError, match="entity_id should be None or a str"):
-        mock_elements_worker.create_metadata_bulk(
-            element=element, metadata_list=wrong_metadata_list
-        )
-
-
 def test_create_metadata_bulk_api_error(responses, mock_elements_worker):
     element = Element({"id": "12341234-1234-1234-1234-123412341234"})
     metadata_list = [
@@ -576,7 +530,6 @@ def test_create_metadata_bulk_api_error(responses, mock_elements_worker):
             "type": MetaType.Text,
             "name": "fake_name",
             "value": "fake_value",
-            "entity_id": "fake_entity_id",
         }
     ]
     responses.add(

tests/test_merge.py

@@ -7,7 +7,6 @@ from arkindex_worker.cache import (
     SQL_VERSION,
     CachedClassification,
     CachedElement,
-    CachedEntity,
     CachedImage,
     CachedTranscription,
     CachedTranscriptionEntity,
@@ -85,7 +84,6 @@ def test_merge_databases(
         assert CachedElement.select().count() == 0
         assert CachedTranscription.select().count() == 0
         assert CachedClassification.select().count() == 0
-        assert CachedEntity.select().count() == 0
         assert CachedTranscriptionEntity.select().count() == 0
 
     # Retrieve parents databases paths
@@ -103,7 +101,6 @@ def test_merge_databases(
         assert CachedElement.select().count() == len(expected_elements)
         assert CachedTranscription.select().count() == len(expected_transcriptions)
         assert CachedClassification.select().count() == 0
-        assert CachedEntity.select().count() == 0
         assert CachedTranscriptionEntity.select().count() == 0
         assert [
             e.id for e in CachedElement.select().order_by("id")
@@ -124,7 +121,6 @@ def test_merge_chunk(mock_databases, tmp_path):
         assert CachedElement.select().count() == 0
         assert CachedTranscription.select().count() == 0
         assert CachedClassification.select().count() == 0
-        assert CachedEntity.select().count() == 0
         assert CachedTranscriptionEntity.select().count() == 0
 
     # Check filenames
@@ -144,7 +140,6 @@ def test_merge_chunk(mock_databases, tmp_path):
         assert CachedElement.select().count() == 3
         assert CachedTranscription.select().count() == 0
         assert CachedClassification.select().count() == 0
-        assert CachedEntity.select().count() == 0
         assert CachedTranscriptionEntity.select().count() == 0
         assert [e.id for e in CachedElement.select().order_by("id")] == [
             UUID("42424242-4242-4242-4242-424242424242"),
@@ -171,7 +166,6 @@ def test_merge_from_worker(
         assert CachedElement.select().count() == 0
         assert CachedTranscription.select().count() == 0
         assert CachedClassification.select().count() == 0
-        assert CachedEntity.select().count() == 0
         assert CachedTranscriptionEntity.select().count() == 0
 
     # Configure worker with a specific data directory
@@ -191,7 +185,6 @@ def test_merge_from_worker(
     assert CachedElement.select().count() == 3
     assert CachedTranscription.select().count() == 1
     assert CachedClassification.select().count() == 0
-    assert CachedEntity.select().count() == 0
     assert CachedTranscriptionEntity.select().count() == 0
     assert [e.id for e in CachedElement.select().order_by("id")] == [
         UUID("12341234-1234-1234-1234-123412341234"),

tests/test_elements_worker/test_entity_list_and_check.py

@@ -1,293 +0,0 @@
-import pytest
-from responses import matchers
-
-from arkindex.exceptions import ErrorResponse
-from arkindex_worker.models import Transcription
-from arkindex_worker.worker.entity import MissingEntityType
-from tests import CORPUS_ID
-
-from . import BASE_API_CALLS
-
-
-def test_create_entity_type_wrong_name(mock_elements_worker):
-    with pytest.raises(
-        AssertionError, match="name shouldn't be null and should be of type str"
-    ):
-        mock_elements_worker.create_entity_type(name=None)
-
-    with pytest.raises(
-        AssertionError, match="name shouldn't be null and should be of type str"
-    ):
-        mock_elements_worker.create_entity_type(name=1234)
-
-
-def test_create_entity_type_api_error(responses, mock_elements_worker):
-    responses.add(
-        responses.POST,
-        "http://testserver/api/v1/entity/types/",
-        status=418,
-    )
-
-    with pytest.raises(ErrorResponse):
-        mock_elements_worker.create_entity_type(name="firstname")
-
-    assert len(responses.calls) == len(BASE_API_CALLS) + 1
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [("POST", "http://testserver/api/v1/entity/types/")]
-
-
-def test_create_entity_type_already_exists(responses, mock_elements_worker):
-    assert mock_elements_worker.entity_types == {}
-
-    responses.add(
-        responses.POST,
-        "http://testserver/api/v1/entity/types/",
-        status=400,
-        match=[
-            matchers.json_params_matcher({"name": "firstname", "corpus": CORPUS_ID})
-        ],
-    )
-    responses.add(
-        responses.GET,
-        f"http://testserver/api/v1/corpus/{CORPUS_ID}/entity-types/",
-        status=200,
-        json={
-            "count": 1,
-            "next": None,
-            "results": [
-                {"id": "lastname-id", "name": "lastname", "color": "ffd1b3"},
-                {"id": "firstname-id", "name": "firstname", "color": "ffd1b3"},
-            ],
-        },
-    )
-
-    mock_elements_worker.create_entity_type(name="firstname")
-
-    assert len(responses.calls) == len(BASE_API_CALLS) + 2
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        ("POST", "http://testserver/api/v1/entity/types/"),
-        ("GET", f"http://testserver/api/v1/corpus/{CORPUS_ID}/entity-types/"),
-    ]
-
-    # Make sure the entity_types attribute has been updated
-    assert mock_elements_worker.entity_types == {
-        "lastname": "lastname-id",
-        "firstname": "firstname-id",
-    }
-
-
-def test_create_entity_type(responses, mock_elements_worker):
-    assert mock_elements_worker.entity_types == {}
-
-    responses.add(
-        responses.POST,
-        "http://testserver/api/v1/entity/types/",
-        status=200,
-        match=[
-            matchers.json_params_matcher({"name": "firstname", "corpus": CORPUS_ID})
-        ],
-        json={
-            "id": "firstname-id",
-            "name": "firstname",
-            "corpus": CORPUS_ID,
-            "color": "ffd1b3",
-        },
-    )
-
-    mock_elements_worker.create_entity_type(name="firstname")
-
-    assert len(responses.calls) == len(BASE_API_CALLS) + 1
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        ("POST", "http://testserver/api/v1/entity/types/"),
-    ]
-
-    # Make sure the entity_types attribute has been updated
-    assert mock_elements_worker.entity_types == {"firstname": "firstname-id"}
-
-
-def test_check_required_entity_types_wrong_entity_types(mock_elements_worker):
-    with pytest.raises(
-        AssertionError,
-        match="entity_types shouldn't be null and should be of type list",
-    ):
-        mock_elements_worker.check_required_entity_types(entity_types=None)
-
-    with pytest.raises(
-        AssertionError,
-        match="entity_types shouldn't be null and should be of type list",
-    ):
-        mock_elements_worker.check_required_entity_types(entity_types=1234)
-
-    with pytest.raises(
-        AssertionError,
-        match="Entity type at index 1 in entity_types: Should be of type str",
-    ):
-        mock_elements_worker.check_required_entity_types(
-            entity_types=["firstname", 1234]
-        )
-
-
-def test_check_required_entity_types_wrong_create_missing(mock_elements_worker):
-    with pytest.raises(
-        AssertionError,
-        match="create_missing shouldn't be null and should be of type bool",
-    ):
-        mock_elements_worker.check_required_entity_types(
-            entity_types=["firstname"], create_missing=None
-        )
-
-    with pytest.raises(
-        AssertionError,
-        match="create_missing shouldn't be null and should be of type bool",
-    ):
-        mock_elements_worker.check_required_entity_types(
-            entity_types=["firstname"], create_missing=1234
-        )
-
-
-def test_check_required_entity_types_do_not_create_missing(
-    responses, mock_elements_worker
-):
-    # Set one entity type
-    mock_elements_worker.entity_types = {"lastname": "lastname-id"}
-
-    with pytest.raises(
-        MissingEntityType, match="Entity type `firstname` was not in the corpus."
-    ):
-        mock_elements_worker.check_required_entity_types(
-            entity_types=["lastname", "firstname"], create_missing=False
-        )
-
-    assert len(responses.calls) == len(BASE_API_CALLS)
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS
-
-
-def test_check_required_entity_types(responses, mock_elements_worker):
-    # Set one entity type
-    mock_elements_worker.entity_types = {"lastname": "lastname-id"}
-
-    # Call to create a new entity type
-    responses.add(
-        responses.POST,
-        "http://testserver/api/v1/entity/types/",
-        status=200,
-        match=[
-            matchers.json_params_matcher({"name": "firstname", "corpus": CORPUS_ID})
-        ],
-        json={
-            "id": "firstname-id",
-            "name": "firstname",
-            "corpus": CORPUS_ID,
-            "color": "ffd1b3",
-        },
-    )
-
-    mock_elements_worker.check_required_entity_types(
-        entity_types=["lastname", "firstname"], create_missing=True
-    )
-
-    assert len(responses.calls) == len(BASE_API_CALLS) + 1
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        (
-            "POST",
-            "http://testserver/api/v1/entity/types/",
-        ),
-    ]
-
-    # Make sure the entity_types attribute has been updated
-    assert mock_elements_worker.entity_types == {
-        "lastname": "lastname-id",
-        "firstname": "firstname-id",
-    }
-
-
-def test_list_transcription_entities_deprecation(fake_dummy_worker):
-    transcription = Transcription({"id": "fake_transcription_id"})
-    worker_version = "worker_version_id"
-    fake_dummy_worker.api_client.add_response(
-        "ListTranscriptionEntities",
-        id=transcription.id,
-        worker_version=worker_version,
-        response={"id": "entity_id"},
-    )
-    with pytest.deprecated_call(
-        match="`worker_version` usage is deprecated. Consider using `worker_run` instead."
-    ):
-        assert fake_dummy_worker.list_transcription_entities(
-            transcription, worker_version=worker_version
-        ) == {"id": "entity_id"}
-
-    assert len(fake_dummy_worker.api_client.history) == 1
-    assert len(fake_dummy_worker.api_client.responses) == 0
-
-
-def test_list_transcription_entities(fake_dummy_worker):
-    transcription = Transcription({"id": "fake_transcription_id"})
-    worker_run = "worker_run_id"
-    fake_dummy_worker.api_client.add_response(
-        "ListTranscriptionEntities",
-        id=transcription.id,
-        worker_run=worker_run,
-        response={"id": "entity_id"},
-    )
-    assert fake_dummy_worker.list_transcription_entities(
-        transcription, worker_run=worker_run
-    ) == {"id": "entity_id"}
-
-    assert len(fake_dummy_worker.api_client.history) == 1
-    assert len(fake_dummy_worker.api_client.responses) == 0
-
-
-def test_list_corpus_entities(responses, mock_elements_worker):
-    responses.add(
-        responses.GET,
-        f"http://testserver/api/v1/corpus/{CORPUS_ID}/entities/",
-        json={
-            "count": 1,
-            "next": None,
-            "results": [
-                {
-                    "id": "fake_entity_id",
-                }
-            ],
-        },
-    )
-
-    mock_elements_worker.list_corpus_entities()
-
-    assert mock_elements_worker.entities == {
-        "fake_entity_id": {
-            "id": "fake_entity_id",
-        }
-    }
-
-    assert len(responses.calls) == len(BASE_API_CALLS) + 1
-    assert [
-        (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        (
-            "GET",
-            f"http://testserver/api/v1/corpus/{CORPUS_ID}/entities/",
-        ),
-    ]
-
-
-@pytest.mark.parametrize("wrong_name", [1234, 12.5])
-def test_list_corpus_entities_wrong_name(mock_elements_worker, wrong_name):
-    with pytest.raises(AssertionError, match="name should be of type str"):
-        mock_elements_worker.list_corpus_entities(name=wrong_name)
-
-
-@pytest.mark.parametrize("wrong_parent", [{"id": "element_id"}, 12.5, "blabla"])
-def test_list_corpus_entities_wrong_parent(mock_elements_worker, wrong_parent):
-    with pytest.raises(AssertionError, match="parent should be of type Element"):
-        mock_elements_worker.list_corpus_entities(parent=wrong_parent)