tests: added integration tests (#120)

docker-compose.yml

@@ -19,5 +19,6 @@ services:
       - S3_HOST=$S3_HOST
       - LAMBDA_HOST=data-test
       - LAMBDA_PORT=8080
+      - BUCKET=dqg-settings-local
     depends_on:
       - data-test

tests/integration_tests/data_test/test_data_test.py

@@ -1,9 +1,8 @@
 from jsonschema import validate
-import boto3
 import pytest
-import requests
 import json
 import os
+import test_utils as test_utils
 
 schema = {
     "$schema": "http://json-schema.org/draft-04/schema#",
@@ -52,65 +51,55 @@
 }
 
 
-@pytest.fixture(scope="function")
-def s3_test_data(request):
+@pytest.fixture
+def run_lambda(request):
     bucket_name = "dataqa"
     file_name = request.param
     file_path = f"{bucket_name}/{file_name}"
-    s3 = _create_boto_s3_resource()
-    _upload_file_to_s3(s3, bucket_name, file_path, file_name)
-    response = _invoke_lambda(file_path)
+    s3 = test_utils.create_boto_s3_resource()
+    test_utils.upload_file_to_s3(s3, bucket_name, file_path, file_name)
+    response = test_utils.invoke_lambda(file_path)
     json_response = json.loads(response.text)
-    validate(instance=json_response, schema=schema)
-    yield file_path
-    _delete_s3_file(s3, bucket_name, file_path)
+    yield json_response
+    test_utils.delete_s3_file(s3, bucket_name, file_path)
 
 
-@pytest.mark.parametrize("s3_test_data",
+@pytest.mark.parametrize("run_lambda",
                          ["titanic.csv",
                           "titanic.parquet",
                           "titanic.json",
                           "titanic_nested.json"],
                          indirect=True)
-def test_data_test(s3_test_data: str):
-    pass
-
-
-def _delete_s3_file(s3, bucket_name: str, file_path: str):
-    s3.Object(bucket_name, file_path).delete()
-
-
-def _upload_file_to_s3(s3, bucket_name: str, file_path: str, file_name: str):
-    local_path = f"./test_data/{file_name}"
-    s3.create_bucket(Bucket=bucket_name)
-    s3.Object(bucket_name, file_path).put(Body=open(local_path, 'rb'))
-
-
-def _create_boto_s3_resource():
-    host = os.environ["S3_HOST"]
-    url = f"http://{host}:4566"
-    s3 = boto3.resource("s3", endpoint_url=url,
-                        aws_access_key_id="test",
-                        aws_secret_access_key="test")
-    return s3
+def test_data_test(run_lambda: str):
+    json_response = run_lambda
+    validate(instance=json_response, schema=schema)
 
 
-def _invoke_lambda(file_path: str):
-    lambda_host = os.environ["LAMBDA_HOST"]
-    lambda_port = os.environ["LAMBDA_PORT"]
-    lambda_url = f"http://{lambda_host}:{lambda_port}/2015-03-31/functions/function/invocations"
+@pytest.mark.parametrize("run_lambda", ['titanic.csv'], indirect=True)
+def test_profile_result_check(run_lambda: str):
+    qa_bucket = os.environ['BUCKET']
+    gx_dir = "great_expectations/expectations"
+    s3 = test_utils.create_boto_s3_resource()
+    json_response = run_lambda
+    suite_name = json_response['suite_name']
+    result_path = f"{qa_bucket}/{gx_dir}/{suite_name}.json"
+    expectations_suite = test_utils.read_json_file(s3, qa_bucket, result_path)
+    expectations = expectations_suite["expectations"]
+    expectations_type = [e['expectation_type'] for e in expectations]
+    assert set(expectations_type) == {
+        'expect_column_values_to_be_in_type_list',
+        'expect_column_value_z_scores_to_be_less_than',
+        'expect_column_values_to_be_unique',
+        'expect_table_columns_to_match_set',
+        'expect_column_mean_to_be_between',
+        'expect_column_values_to_be_between',
+        'expect_column_stdev_to_be_between',
+        'expect_column_to_exist',
+        'expect_column_values_to_be_increasing',
+        'expect_column_median_to_be_between',
+        'expect_column_values_to_be_in_set',
+        'expect_column_quantile_values_to_be_between',
+        'expect_table_row_count_to_equal',
+        'expect_column_values_to_not_be_null'}
+    assert len(expectations) == 73
 
-    payload = json.dumps({
-        "run_name": "local_test",
-        "source_root": "dataqa",
-        "source_data": f"{file_path}",
-        "engine": "s3"
-    })
-    headers = {
-        'Content-Type': 'application/json'
-    }
-    response = requests.request("POST",
-                                lambda_url,
-                                headers=headers,
-                                data=payload)
-    return response

tests/integration_tests/data_test/test_utils.py

@@ -0,0 +1,50 @@
+import boto3
+import requests
+import json
+import os
+
+
+def read_json_file(s3, bucket_name,  file_path):
+    content_object = s3.Object(bucket_name, file_path)
+    file_content = content_object.get()['Body'].read().decode('utf-8')
+    return json.loads(file_content)
+
+
+def delete_s3_file(s3, bucket_name: str, file_path: str):
+    s3.Object(bucket_name, file_path).delete()
+
+
+def upload_file_to_s3(s3, bucket_name: str, file_path: str, file_name: str):
+    local_path = f"./test_data/{file_name}"
+    s3.create_bucket(Bucket=bucket_name)
+    s3.Object(bucket_name, file_path).put(Body=open(local_path, 'rb'))
+
+
+def create_boto_s3_resource():
+    host = os.environ["S3_HOST"]
+    url = f"http://{host}:4566"
+    s3 = boto3.resource("s3", endpoint_url=url,
+                        aws_access_key_id="test",
+                        aws_secret_access_key="test")
+    return s3
+
+
+def invoke_lambda(file_path: str):
+    lambda_host = os.environ["LAMBDA_HOST"]
+    lambda_port = os.environ["LAMBDA_PORT"]
+    lambda_url = f"http://{lambda_host}:{lambda_port}/2015-03-31/functions/function/invocations"
+
+    payload = json.dumps({
+        "run_name": "local_test",
+        "source_root": "dataqa",
+        "source_data": f"{file_path}",
+        "engine": "s3"
+    })
+    headers = {
+        'Content-Type': 'application/json'
+    }
+    response = requests.request("POST",
+                                lambda_url,
+                                headers=headers,
+                                data=payload)
+    return response