feat(tests): add integration tests for all api usage scenarios

internal/rest/http/file/di.py

@@ -7,8 +7,12 @@
     get_dataset_repo,
 )
 from internal.uow import UnitOfWork
-from internal.usecase.file import SaveFile, SaveDataset, CheckContentType
-from internal.usecase.file.retrieve_dataset import RetrieveDataset
+from internal.usecase.file import (
+    SaveFile,
+    SaveDataset,
+    CheckContentType,
+    RetrieveDataset,
+)
 from internal.usecase.file.save_dataset import DatasetRepo as SaveDatasetRepo
 from internal.usecase.file.retrieve_dataset import DatasetRepo as RetrieveDatasetRepo
 from internal.usecase.file.save_file import FileRepo, FileMetadataRepo

tests/conftest.py

@@ -229,7 +229,7 @@ def set_task_use_case(unit_of_work, task_repo, dataset_repo, profiling_task_work
 def profile_task_use_case(unit_of_work_without_pool, dataset_repo, file_repo):
     return ProfileTask(
         unit_of_work=unit_of_work_without_pool,
-        dataset_repo=task_repo,
+        dataset_repo=dataset_repo,
         file_repo=file_repo,
     )
 

tests/datasets/university.txt

@@ -0,0 +1,9 @@
+Course,Classroom,Professor,Semester
+Math,512,Dr. Smith,Fall
+Physics,406,Dr. Green,Fall
+English,208,Prof. Turner,Fall
+History,209,Prof. Davis,Fall
+Math,512,Dr. Smith,Spring
+Physics,503,Dr. Gray,Spring
+English,116,Prof. Turner,Spring
+Biology,209,Prof. Light,Spring

tests/integration/common_requests.py

@@ -0,0 +1,36 @@
+from fastapi.testclient import TestClient
+
+
+def upload_csv_dataset(
+    client: TestClient,
+    file_name,
+    mime_type,
+    separator,
+    header,
+):
+    file_path = f"tests/datasets/{file_name}"
+
+    with open(file_path, "rb") as file:
+        form_data = {
+            "separator": separator,
+            "header": header,
+            "file": (file_name, file, mime_type),
+        }
+
+        response = client.post(
+            "api/file/csv",
+            files={"file": form_data["file"]},
+            data={
+                "separator": form_data["separator"],
+                "header": form_data["header"],
+            },
+        )
+
+    return response
+
+
+def set_task(client: TestClient, dataset_id, config):
+    response = client.post(
+        "api/task/set", params={"dataset_id": dataset_id}, json=config.model_dump()
+    )
+    return response

tests/integration/test_ping.py

@@ -0,0 +1,8 @@
+from fastapi.testclient import TestClient
+
+
+def test_retrieve_task(client: TestClient):
+    response = client.get("api/common/ping")
+
+    assert response.status_code == 200
+    assert response.json() == "Pong!"

tests/integration/test_retrieve_dataset_info.py

@@ -0,0 +1,40 @@
+from fastapi.testclient import TestClient
+from uuid import UUID, uuid4
+
+from internal.infrastructure.data_storage.relational.model.file import DatasetORM
+from tests.integration.common_requests import upload_csv_dataset
+
+
+def test_retrieve_dataset(client: TestClient, context):
+    file_name = "university_fd.csv"
+    file_path = f"tests/datasets/{file_name}"
+    mime_type = "text/csv"
+    separator = ","
+    header = [0]
+
+    response = upload_csv_dataset(client, file_name, mime_type, separator, header)
+    assert response.status_code == 200
+
+    dataset_id = UUID(response.json())
+
+    response = client.post(f"api/file/dataset/{dataset_id}")
+
+    assert response.status_code == 200
+
+    response_data = response.json()
+    assert response_data["id"] == str(dataset_id)
+
+    dataset = context.postgres_context.get(DatasetORM, dataset_id)
+    assert dataset is not None
+    assert response_data["file_id"] == str(dataset.file_id)
+    assert response_data["separator"] == dataset.separator
+    assert response_data["header"] == dataset.header
+
+
+def test_retrieve_non_existent_dataset(client: TestClient):
+    dataset_id = uuid4()  # non existen dataset
+
+    response = client.post(f"api/file/dataset/{dataset_id}")
+
+    assert response.status_code == 404
+    assert response.json()["detail"] == "Dataset not found"

tests/integration/test_retrieve_task_info.py

@@ -0,0 +1,58 @@
+import time
+from fastapi.testclient import TestClient
+from uuid import UUID, uuid4
+
+from internal.domain.task.value_objects import FdTaskConfig, PrimitiveName
+from internal.domain.task.value_objects.fd.algo_name import FdAlgoName
+from internal.domain.task.value_objects.fd.algo_config import AidConfig
+from internal.infrastructure.data_storage.relational.model.task import TaskORM
+from tests.integration.common_requests import upload_csv_dataset, set_task
+
+
+def test_retrieve_task(client: TestClient, context):
+    dataset_response = upload_csv_dataset(
+        client, "university_fd.csv", "text/csv", ".", [0]
+    )
+    assert dataset_response.status_code == 200
+    dataset_id = UUID(dataset_response.json())
+
+    algo_config = AidConfig(algo_name=FdAlgoName.Aid, is_null_equal_null=True)
+    task_config = FdTaskConfig(primitive_name=PrimitiveName.fd, config=algo_config)
+
+    task_response = set_task(client, dataset_id, task_config)
+    assert task_response.status_code == 200
+    task_id = UUID(task_response.json())
+
+    time.sleep(1)
+
+    response = client.get(f"api/task/{task_id}")
+    assert response.status_code == 200
+
+    response_data = response.json()
+    assert response_data["dataset_id"] == str(dataset_id)
+
+    task = context.postgres_context.get(TaskORM, task_id)
+    assert task is not None
+    assert response_data["dataset_id"] == str(task.dataset_id)
+    assert response_data["status"] == task.status
+    assert response_data["config"] == task.config
+    assert response_data["result"] == task.result
+    assert response_data["raised_exception_name"] == task.raised_exception_name
+    assert response_data["failure_reason"] == task.failure_reason
+    assert response_data["traceback"] == task.traceback
+
+    if response_data["status"] == "completed":
+        assert response_data["result"] is not None
+    if response_data["status"] == "failure":
+        assert response_data["raised_exception_name"] is not None
+        assert response_data["failure_reason"] is not None
+        assert response_data["traceback"] is not None
+
+
+def test_retrieve_non_existent_task(client: TestClient):
+    task_id = uuid4()  # non existen task
+
+    response = client.get(f"api/task/{task_id}")
+
+    assert response.status_code == 404
+    assert response.json()["detail"] == "Task not found"

tests/integration/test_set_profiling_task.py

@@ -0,0 +1,42 @@
+from fastapi.testclient import TestClient
+from uuid import UUID, uuid4
+
+from internal.domain.task.value_objects import FdTaskConfig, PrimitiveName
+from internal.domain.task.value_objects.fd.algo_name import FdAlgoName
+from internal.domain.task.value_objects.fd.algo_config import AidConfig
+from internal.infrastructure.data_storage.relational.model.task import TaskORM
+from tests.integration.common_requests import upload_csv_dataset, set_task
+
+
+def test_set_profiling_task(client: TestClient, context):
+    dataset_response = upload_csv_dataset(
+        client, "university_fd.csv", "text/csv", ".", [0]
+    )
+    assert dataset_response.status_code == 200
+    dataset_id = UUID(dataset_response.json())
+
+    algo_config = AidConfig(algo_name=FdAlgoName.Aid, is_null_equal_null=True)
+    task_config = FdTaskConfig(primitive_name=PrimitiveName.fd, config=algo_config)
+
+    response = set_task(client, dataset_id, task_config)
+
+    assert response.status_code == 200
+
+    task_id = UUID(response.json())
+
+    task_data = context.postgres_context.get(TaskORM, task_id)
+
+    assert task_data is not None
+    assert task_data.config == task_config.model_dump()
+    assert task_data.dataset_id == dataset_id
+
+
+def test_set_profiling_task_with_non_existent_dataset(client: TestClient):
+    dataset_id = uuid4()
+    algo_config = AidConfig(algo_name=FdAlgoName.Aid, is_null_equal_null=True)
+    task_config = FdTaskConfig(primitive_name=PrimitiveName.fd, config=algo_config)
+
+    response = set_task(client, dataset_id, task_config)
+
+    assert response.status_code == 404
+    assert response.json()["detail"] == "Dataset not found"

tests/integration/test_upload_dataset.py

@@ -0,0 +1,58 @@
+import os
+import pytest
+from fastapi.testclient import TestClient
+
+from uuid import UUID
+
+from internal.infrastructure.data_storage.relational.model.file import DatasetORM
+from tests.integration.common_requests import upload_csv_dataset
+
+
+@pytest.mark.asyncio
+async def test_upload_csv_dataset(client: TestClient, context, tmp_upload_dir):
+    file_name = "university_fd.csv"
+    file_path = f"tests/datasets/{file_name}"
+    mime_type = "text/csv"
+    separator = ","
+    header = [0]
+
+    response = upload_csv_dataset(client, file_name, mime_type, separator, header)
+
+    assert response.status_code == 200
+
+    dataset_id = UUID(response.json())
+
+    data = context.postgres_context.get(DatasetORM, dataset_id)
+    assert data is not None
+    assert data.id == dataset_id
+    assert data.separator == separator
+    assert data.header == header
+    assert data.related_tasks == []
+
+    file = data.file_metadata
+    assert file.original_file_name == "university_fd.csv"
+    assert file.mime_type == mime_type
+
+    saved_file_path = os.path.join(tmp_upload_dir, str(file.file_name))
+    assert os.path.exists(saved_file_path)
+
+    with open(saved_file_path, "rb") as saved_file, open(
+        file_path, "rb"
+    ) as original_file:
+        saved_file_content = saved_file.read()
+        original_file_content = original_file.read()
+        assert saved_file_content == original_file_content
+
+
+@pytest.mark.asyncio
+async def test_upload_csv_dataset_with_incorrect_mime_type(client: TestClient):
+    file_name = "university.txt"
+    file_path = f"tests/datasets/{file_name}"
+    mime_type = "text/plain"
+    separator = ","
+    header = [0]
+
+    response = upload_csv_dataset(client, file_name, mime_type, separator, header)
+
+    assert response.status_code == 400
+    assert response.json()["detail"] == "File is not CSV"