arkindex-base-worker 0.3.7rc9__py3-none-any.whl → 0.4.0__py3-none-any.whl

tests/test_elements_worker/{test_entities.py → test_entity_create.py}

@@ -3,9 +3,9 @@ import re
 from uuid import UUID
 
 import pytest
-from apistar.exceptions import ErrorResponse
 from responses import matchers
 
+from arkindex.exceptions import ErrorResponse
 from arkindex_worker.cache import (
     CachedElement,
     CachedEntity,
@@ -13,8 +13,8 @@ from arkindex_worker.cache import (
     CachedTranscriptionEntity,
 )
 from arkindex_worker.models import Transcription
-from arkindex_worker.worker.entity import MissingEntityType
 from arkindex_worker.worker.transcription import TextOrientation
+from tests import CORPUS_ID
 
 from . import BASE_API_CALLS
 
@@ -90,7 +90,7 @@ def test_create_entity_api_error(responses, mock_elements_worker):
     responses.add(
         responses.POST,
         "http://testserver/api/v1/entity/",
-        status=500,
+        status=418,
     )
 
     with pytest.raises(ErrorResponse):
@@ -99,17 +99,10 @@ def test_create_entity_api_error(responses, mock_elements_worker):
             type="person",
         )
 
-    assert len(responses.calls) == len(BASE_API_CALLS) + 5
+    assert len(responses.calls) == len(BASE_API_CALLS) + 1
     assert [
         (call.request.method, call.request.url) for call in responses.calls
-    ] == BASE_API_CALLS + [
-        # We retry 5 times the API call
-        ("POST", "http://testserver/api/v1/entity/"),
-        ("POST", "http://testserver/api/v1/entity/"),
-        ("POST", "http://testserver/api/v1/entity/"),
-        ("POST", "http://testserver/api/v1/entity/"),
-        ("POST", "http://testserver/api/v1/entity/"),
-    ]
+    ] == BASE_API_CALLS + [("POST", "http://testserver/api/v1/entity/")]
 
 
 def test_create_entity(responses, mock_elements_worker):
@@ -139,7 +132,7 @@ def test_create_entity(responses, mock_elements_worker):
         "type_id": "person-entity-type-id",
         "metas": {},
         "validated": None,
-        "corpus": "11111111-1111-1111-1111-111111111111",
+        "corpus": CORPUS_ID,
         "worker_run_id": "56785678-5678-5678-5678-567856785678",
     }
     assert entity_id == "12345678-1234-1234-1234-123456789123"
@@ -152,7 +145,7 @@ def test_create_entity_missing_type(responses, mock_elements_worker):
     # Call to list entity types
     responses.add(
         responses.GET,
-        "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/entity-types/",
+        f"http://testserver/api/v1/corpus/{CORPUS_ID}/entity-types/",
         status=200,
         json={
             "count": 1,
@@ -177,7 +170,7 @@ def test_create_entity_missing_type(responses, mock_elements_worker):
     ] == BASE_API_CALLS + [
         (
             "GET",
-            "http://testserver/api/v1/corpus/11111111-1111-1111-1111-111111111111/entity-types/",
+            f"http://testserver/api/v1/corpus/{CORPUS_ID}/entity-types/",
         ),
     ]
 
@@ -209,7 +202,7 @@ def test_create_entity_with_cache(responses, mock_elements_worker_with_cache):
         "type_id": "person-entity-type-id",
         "metas": {},
         "validated": None,
-        "corpus": "11111111-1111-1111-1111-111111111111",
+        "corpus": CORPUS_ID,
         "worker_run_id": "56785678-5678-5678-5678-567856785678",
     }
     assert entity_id == "12345678-1234-1234-1234-123456789123"
@@ -387,7 +380,7 @@ def test_create_transcription_entity_api_error(responses, mock_elements_worker):
     responses.add(
         responses.POST,
         "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        status=500,
+        status=418,
     )
 
     with pytest.raises(ErrorResponse):
@@ -403,31 +396,14 @@ def test_create_transcription_entity_api_error(responses, mock_elements_worker):
             length=10,
         )
 
-    assert len(responses.calls) == len(BASE_API_CALLS) + 5
+    assert len(responses.calls) == len(BASE_API_CALLS) + 1
     assert [
         (call.request.method, call.request.url) for call in responses.calls
     ] == BASE_API_CALLS + [
-        # We retry 5 times the API call
-        (
-            "POST",
-            "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        ),
         (
             "POST",
             "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        ),
-        (
-            "POST",
-            "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        ),
-        (
-            "POST",
-            "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        ),
-        (
-            "POST",
-            "http://testserver/api/v1/transcription/11111111-1111-1111-1111-111111111111/entity/",
-        ),
+        )
     ]
 
 
@@ -708,159 +684,6 @@ def test_create_transcription_entity_with_confidence_with_cache(
     ]
 
 
-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):
-    corpus_id = "11111111-1111-1111-1111-111111111111"
-    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)
-
-
-def test_check_required_entity_types(responses, mock_elements_worker):
-    # Set one entity type
-    mock_elements_worker.entity_types = {"person": "person-entity-type-id"}
-
-    checked_types = ["person", "new-entity"]
-
-    # Call to create new entity type
-    responses.add(
-        responses.POST,
-        "http://testserver/api/v1/entity/types/",
-        status=200,
-        match=[
-            matchers.json_params_matcher(
-                {
-                    "name": "new-entity",
-                    "corpus": "11111111-1111-1111-1111-111111111111",
-                }
-            )
-        ],
-        json={
-            "id": "new-entity-id",
-            "corpus": "11111111-1111-1111-1111-111111111111",
-            "name": "new-entity",
-            "color": "ffd1b3",
-        },
-    )
-
-    mock_elements_worker.check_required_entity_types(
-        entity_types=checked_types,
-    )
-
-    # Make sure the entity_types entry has been updated
-    assert mock_elements_worker.entity_types == {
-        "person": "person-entity-type-id",
-        "new-entity": "new-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 + [
-        (
-            "POST",
-            "http://testserver/api/v1/entity/types/",
-        ),
-    ]
-
-
-def test_check_required_entity_types_no_creation_allowed(
-    responses, mock_elements_worker
-):
-    # Set one entity type
-    mock_elements_worker.entity_types = {"person": "person-entity-type-id"}
-
-    checked_types = ["person", "new-entity"]
-
-    with pytest.raises(
-        MissingEntityType, match="Entity type `new-entity` was not in the corpus."
-    ):
-        mock_elements_worker.check_required_entity_types(
-            entity_types=checked_types, 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
-
-
 @pytest.mark.parametrize("transcription", [None, "not a transcription", 1])
 def test_create_transcription_entities_wrong_transcription(
     mock_elements_worker, transcription
@@ -1014,6 +837,7 @@ def test_create_transcription_entities_wrong_entity(
 
 def test_create_transcription_entities(responses, mock_elements_worker):
     transcription = Transcription(id="transcription-id")
+
     # Call to Transcription entities creation in bulk
     responses.add(
         responses.POST,
@@ -1030,7 +854,14 @@ def test_create_transcription_entities(responses, mock_elements_worker):
                             "offset": 0,
                             "length": 6,
                             "confidence": 1.0,
-                        }
+                        },
+                        {
+                            "name": "Team Rocket",
+                            "type_id": "22222222-2222-2222-2222-222222222222",
+                            "offset": 7,
+                            "length": 11,
+                            "confidence": 1.0,
+                        },
                     ],
                 }
             )
@@ -1039,8 +870,12 @@ def test_create_transcription_entities(responses, mock_elements_worker):
             "entities": [
                 {
                     "transcription_entity_id": "transc-entity-id",
-                    "entity_id": "entity-id",
-                }
+                    "entity_id": "entity-id1",
+                },
+                {
+                    "transcription_entity_id": "transc-entity-id",
+                    "entity_id": "entity-id2",
+                },
             ]
         },
     )
@@ -1058,11 +893,18 @@ def test_create_transcription_entities(responses, mock_elements_worker):
                 "offset": 0,
                 "length": 6,
                 "confidence": 1.0,
-            }
+            },
+            {
+                "name": "Team Rocket",
+                "type_id": "22222222-2222-2222-2222-222222222222",
+                "offset": 7,
+                "length": 11,
+                "confidence": 1.0,
+            },
         ],
     )
 
-    assert len(created_objects) == 1
+    assert len(created_objects) == 2
 
     assert len(responses.calls) == len(BASE_API_CALLS) + 1
     assert [
@@ -1071,5 +913,5 @@ def test_create_transcription_entities(responses, mock_elements_worker):
         (
             "POST",
             "http://testserver/api/v1/transcription/transcription-id/entities/bulk/",
-        ),
+        )
     ]

tests/test_elements_worker/test_entity_list_and_check.py

@@ -0,0 +1,160 @@
+import pytest
+from responses import matchers
+
+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_check_required_entity_types(responses, mock_elements_worker):
+    # Set one entity type
+    mock_elements_worker.entity_types = {"person": "person-entity-type-id"}
+
+    checked_types = ["person", "new-entity"]
+
+    # Call to create new entity type
+    responses.add(
+        responses.POST,
+        "http://testserver/api/v1/entity/types/",
+        status=200,
+        match=[
+            matchers.json_params_matcher(
+                {
+                    "name": "new-entity",
+                    "corpus": CORPUS_ID,
+                }
+            )
+        ],
+        json={
+            "id": "new-entity-id",
+            "corpus": CORPUS_ID,
+            "name": "new-entity",
+            "color": "ffd1b3",
+        },
+    )
+
+    mock_elements_worker.check_required_entity_types(
+        entity_types=checked_types,
+    )
+
+    # Make sure the entity_types entry has been updated
+    assert mock_elements_worker.entity_types == {
+        "person": "person-entity-type-id",
+        "new-entity": "new-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 + [
+        (
+            "POST",
+            "http://testserver/api/v1/entity/types/",
+        ),
+    ]
+
+
+def test_check_required_entity_types_no_creation_allowed(
+    responses, mock_elements_worker
+):
+    # Set one entity type
+    mock_elements_worker.entity_types = {"person": "person-entity-type-id"}
+
+    checked_types = ["person", "new-entity"]
+
+    with pytest.raises(
+        MissingEntityType, match="Entity type `new-entity` was not in the corpus."
+    ):
+        mock_elements_worker.check_required_entity_types(
+            entity_types=checked_types, 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_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)

tests/test_elements_worker/test_image.py

@@ -0,0 +1,66 @@
+import json
+
+import pytest
+
+from arkindex.exceptions import ErrorResponse
+
+from . import BASE_API_CALLS
+
+
+@pytest.mark.parametrize(
+    ("data", "error_message"),
+    [
+        (None, "url shouldn't be null and should be of type str"),
+        (1234, "url shouldn't be null and should be of type str"),
+    ],
+)
+def test_create_iiif_url_wrong_data(data, error_message, mock_elements_worker):
+    with pytest.raises(AssertionError, match=error_message):
+        mock_elements_worker.create_iiif_url(url=data)
+
+
+def test_create_iiif_url_api_error(responses, mock_elements_worker):
+    responses.add(
+        responses.POST,
+        "http://testserver/api/v1/image/iiif/url/",
+        status=418,
+    )
+
+    with pytest.raises(ErrorResponse):
+        mock_elements_worker.create_iiif_url("http://url/to/my/image")
+
+    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/image/iiif/url/")]
+
+
+def test_create_iiif_url(responses, mock_elements_worker):
+    responses.add(
+        responses.POST,
+        "http://testserver/api/v1/image/iiif/url/",
+        status=201,
+        json={
+            "id": "cafecafe-cafe-cafe-cafe-cafecafecafe",
+            "url": "http://url/to/my/image",
+            "status": "checked",
+            "server": {
+                "id": 5,
+                "display_name": "My server",
+                "url": "http://url/to/my",
+                "max_width": 42,
+                "max_height": 42,
+            },
+        },
+    )
+
+    image = mock_elements_worker.create_iiif_url("http://url/to/my/image")
+
+    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/image/iiif/url/")]
+    assert json.loads(responses.calls[-1].request.body) == {
+        "url": "http://url/to/my/image"
+    }
+    assert image.id == "cafecafe-cafe-cafe-cafe-cafecafecafe"