[AL-5120][AL-5121] Integration tests for Audio and HTML import

labelbox/__init__.py

@@ -8,6 +8,7 @@
 from labelbox.schema.annotation_import import MALPredictionImport, MEAPredictionImport, LabelImport, MEAToMALPredictionImport
 from labelbox.schema.dataset import Dataset
 from labelbox.schema.data_row import DataRow
+from labelbox.schema.enums import AnnotationImportState
 from labelbox.schema.label import Label
 from labelbox.schema.batch import Batch
 from labelbox.schema.review import Review

tests/integration/annotation_import/test_conversation_import.py

@@ -39,16 +39,4 @@ def test_conversation_entity(client, configured_project_without_data_rows,
         predictions=labels)
     import_annotations.wait_until_done()
 
-    assert import_annotations.errors == []
-
-    exported_labels = configured_project_without_data_rows.label_generator()
-    for label in exported_labels:
-        assert len(
-            label.annotations) == 1  # we have created only 1 annotation above
-        annotation = label.annotations[0]
-
-        assert type(annotation) is ConversationEntity
-        assert annotation.name == "named-entity"
-        assert annotation.value.message_id == "4"
-        assert annotation.value.start == 0
-        assert annotation.value.end == 8
+    assert import_annotations.errors == []

tests/integration/annotation_import/test_data_types.py

@@ -1,10 +1,69 @@
+import itertools
 import pytest
+import uuid
+
 import labelbox as lb
 import labelbox.types as lb_types
 from labelbox.data.annotation_types.data import AudioData, ConversationData, DicomData, DocumentData, HTMLData, ImageData, TextData
 from labelbox.data.serialization import NDJsonConverter
 from labelbox.schema.annotation_import import AnnotationImportState
 
+radio_annotation = lb_types.ClassificationAnnotation(
+    name="radio",
+    value=lb_types.Radio(answer=lb_types.ClassificationAnswer(
+        name="second_radio_answer")))
+checklist_annotation = lb_types.ClassificationAnnotation(
+    name="checklist",
+    value=lb_types.Checklist(answer=[
+        lb_types.ClassificationAnswer(name="option1"),
+        lb_types.ClassificationAnswer(name="option2")
+    ]))
+text_annotation = lb_types.ClassificationAnnotation(
+    name="text", value=lb_types.Text(answer="sample text"))
+
+
+def get_annotation_comparison_dicts_from_labels(labels):
+    labels_ndjson = list(NDJsonConverter.serialize(labels))
+    for annotation in labels_ndjson:
+        annotation.pop('uuid')
+        annotation.pop('dataRow')
+    return labels_ndjson
+
+
+def get_annotation_comparison_dicts_from_export(export_result, data_row_id,
+                                                project_id):
+    exported_data_row = [
+        dr for dr in export_result if dr['data_row']['id'] == data_row_id
+    ][0]
+    exported_annotations = exported_data_row['projects'][project_id]['labels'][
+        0]['annotations']
+    flat_exported_annotations = list(
+        itertools.chain(*exported_annotations.values()))
+
+    converted_annotations = []
+    for annotation in flat_exported_annotations:
+        if annotation['name'] == 'radio':
+            converted_annotations.append({
+                'name': annotation['name'],
+                'answer': {
+                    'name': annotation['radio_answer']['name']
+                }
+            })
+        elif annotation['name'] == 'checklist':
+            converted_annotations.append({
+                'name':
+                    annotation['name'],
+                'answer': [{
+                    'name': answer['name']
+                } for answer in annotation['checklist_answers']]
+            })
+        elif annotation['name'] == 'text':
+            converted_annotations.append({
+                'name': annotation['name'],
+                'answer': annotation['text_answer']['content']
+            })
+    return converted_annotations
+
 
 # TODO: Add VideoData. Currently label import job finishes without errors but project.export_labels() returns empty list.
 @pytest.mark.parametrize('data_type_class', [
@@ -44,3 +103,90 @@ def test_import_data_types(client, configured_project,
     classifications = exported_labels[0]['Label']['classifications']
     assert len(objects) + len(classifications) == len(labels)
     data_row.delete()
+
+
+@pytest.mark.parametrize('data_type, data_class, annotations', [
+    [
+        'html', lb_types.HTMLData,
+        [radio_annotation, checklist_annotation, text_annotation]
+    ],
+    [
+        'audio', lb_types.AudioData,
+        [radio_annotation, checklist_annotation, text_annotation]
+    ],
+])
+def test_import_label_annotations(client, configured_project,
+                                  data_row_json_by_data_type, data_type,
+                                  data_class, annotations):
+
+    dataset = next(configured_project.datasets())
+    data_row_json = data_row_json_by_data_type[data_type]
+    data_row = dataset.create_data_row(data_row_json)
+
+    labels = [
+        lb_types.Label(data=data_class(uid=data_row.uid),
+                       annotations=annotations)
+    ]
+
+    label_import = lb.LabelImport.create_from_objects(client,
+                                                      configured_project.uid,
+                                                      f'test-import-html',
+                                                      labels)
+    label_import.wait_until_done()
+
+    assert label_import.state == lb.AnnotationImportState.FINISHED
+    assert len(label_import.errors) == 0
+    export_params = {
+        "attachments": False,
+        "metadata_fields": False,
+        "data_row_details": False,
+        "project_details": False,
+        "performance_details": False
+    }
+    export_task = configured_project.export_v2(params=export_params)
+    export_task.wait_till_done()
+    assert export_task.errors is None
+    expected_annotations = get_annotation_comparison_dicts_from_labels(labels)
+    actual_annotations = get_annotation_comparison_dicts_from_export(
+        export_task.result, data_row.uid, configured_project.uid)
+    assert actual_annotations == expected_annotations
+    data_row.delete()
+
+
+@pytest.mark.parametrize('data_type, data_class, annotations', [
+    [
+        'html', lb_types.HTMLData,
+        [radio_annotation, checklist_annotation, text_annotation]
+    ],
+    [
+        'audio', lb_types.AudioData,
+        [radio_annotation, checklist_annotation, text_annotation]
+    ],
+])
+def test_import_mal_annotations(client, configured_project_without_data_rows,
+                                data_row_json_by_data_type, data_type,
+                                data_class, annotations, rand_gen):
+
+    dataset = client.create_dataset(name=rand_gen(str))
+    data_row_json = data_row_json_by_data_type[data_type]
+    data_row = dataset.create_data_row(data_row_json)
+
+    configured_project_without_data_rows.create_batch(
+        rand_gen(str),
+        [data_row.uid],
+    )
+
+    labels = [
+        lb_types.Label(data=data_class(uid=data_row.uid),
+                       annotations=annotations)
+    ]
+
+    import_annotations = lb.MALPredictionImport.create_from_objects(
+        client=client,
+        project_id=configured_project_without_data_rows.uid,
+        name=f"import {str(uuid.uuid4())}",
+        predictions=labels)
+    import_annotations.wait_until_done()
+
+    assert import_annotations.errors == []
+    # MAL Labels cannot be exported and compared to input labels