S3/DynamoDB: Improve PartiQL tests (#6897)

.github/workflows/tests_real_aws.yml

@@ -42,4 +42,4 @@ jobs:
       env:
         MOTO_TEST_ALLOW_AWS_REQUEST: ${{ true }}
       run: |
-        pytest -sv tests/test_ec2/ tests/test_lakeformation/ tests/test_ses/ tests/test_s3 -m aws_verified
+        pytest -sv tests/test_dynamodb/ tests/test_ec2/ tests/test_lakeformation/ tests/test_ses/ tests/test_s3 -m aws_verified

setup.cfg

@@ -52,7 +52,7 @@ all =
     openapi-spec-validator>=0.2.8
     pyparsing>=3.0.7
     jsondiff>=1.1.2
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
     aws-xray-sdk!=0.96,>=0.93
     setuptools
     multipart
@@ -67,7 +67,7 @@ proxy =
     openapi-spec-validator>=0.2.8
     pyparsing>=3.0.7
     jsondiff>=1.1.2
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
     aws-xray-sdk!=0.96,>=0.93
     setuptools
     multipart
@@ -82,7 +82,7 @@ server =
     openapi-spec-validator>=0.2.8
     pyparsing>=3.0.7
     jsondiff>=1.1.2
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
     aws-xray-sdk!=0.96,>=0.93
     setuptools
     flask!=2.2.0,!=2.2.1
@@ -116,7 +116,7 @@ cloudformation =
     openapi-spec-validator>=0.2.8
     pyparsing>=3.0.7
     jsondiff>=1.1.2
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
     aws-xray-sdk!=0.96,>=0.93
     setuptools
 cloudfront =
@@ -139,10 +139,10 @@ dms =
 ds = sshpubkeys>=3.1.0
 dynamodb =
     docker>=3.0.0
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
 dynamodbstreams =
     docker>=3.0.0
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
 ebs = sshpubkeys>=3.1.0
 ec2 = sshpubkeys>=3.1.0
 ec2instanceconnect =
@@ -205,15 +205,15 @@ resourcegroupstaggingapi =
     openapi-spec-validator>=0.2.8
     pyparsing>=3.0.7
     jsondiff>=1.1.2
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
 route53 =
 route53resolver = sshpubkeys>=3.1.0
 s3 =
     PyYAML>=5.1
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
 s3crc32c =
     PyYAML>=5.1
-    py-partiql-parser==0.3.7
+    py-partiql-parser==0.3.8
     crc32c
 s3control =
 sagemaker =

tests/test_dynamodb/__init__.py

@@ -1 +1,57 @@
-# This file is intentionally left blank.
+import boto3
+import os
+from functools import wraps
+from moto import mock_dynamodb
+from uuid import uuid4
+
+
+def dynamodb_aws_verified(func):
+    """
+    Function that is verified to work against AWS.
+    Can be run against AWS at any time by setting:
+      MOTO_TEST_ALLOW_AWS_REQUEST=true
+
+    If this environment variable is not set, the function runs in a `mock_dynamodb` context.
+
+    This decorator will:
+      - Create a table
+      - Run the test and pass the table_name as an argument
+      - Delete the table
+    """
+
+    @wraps(func)
+    def pagination_wrapper():
+        client = boto3.client("dynamodb", region_name="us-east-1")
+        table_name = str(uuid4())[0:6]
+
+        allow_aws_request = (
+            os.environ.get("MOTO_TEST_ALLOW_AWS_REQUEST", "false").lower() == "true"
+        )
+
+        if allow_aws_request:
+            print(f"Test {func} will create DynamoDB Table {table_name}")
+            resp = create_table_and_test(table_name, client)
+        else:
+            with mock_dynamodb():
+                resp = create_table_and_test(table_name, client)
+        return resp
+
+    def create_table_and_test(table_name, client):
+        client.create_table(
+            TableName=table_name,
+            KeySchema=[{"AttributeName": "pk", "KeyType": "HASH"}],
+            AttributeDefinitions=[{"AttributeName": "pk", "AttributeType": "S"}],
+            ProvisionedThroughput={"ReadCapacityUnits": 1, "WriteCapacityUnits": 5},
+            Tags=[{"Key": "environment", "Value": "moto_tests"}],
+        )
+        waiter = client.get_waiter("table_exists")
+        waiter.wait(TableName=table_name)
+        try:
+            resp = func(table_name)
+        finally:
+            ### CLEANUP ###
+            client.delete_table(TableName=table_name)
+
+        return resp
+
+    return pagination_wrapper

tests/test_dynamodb/test_dynamodb_statements.py

@@ -1,68 +1,79 @@
 import boto3
+import pytest
 
 from moto import mock_dynamodb
 from unittest import TestCase
 
+from . import dynamodb_aws_verified
 
-class TestSelectStatements:
-    mock = mock_dynamodb()
 
-    @classmethod
+item1 = {
-    def setup_class(cls):
+    "pk": {"S": "msg1"},
-        cls.mock.start()
+    "body": {"S": "some text"},
-        cls.client = boto3.client("dynamodb", "us-east-1")
+    "nested_attrs": {"M": {"some": {"S": "key"}}},
-        cls.client.create_table(
+    "list_attrs": {"L": [{"BOOL": True}, {"BOOL": False}]},
-            TableName="messages",
+    "bool_attr": {"BOOL": True},
-            KeySchema=[{"AttributeName": "id", "KeyType": "HASH"}],
+}
-            AttributeDefinitions=[{"AttributeName": "id", "AttributeType": "S"}],
+item2 = {"pk": {"S": "msg2"}, "body": {"S": "n/a"}, "unique_key": {"S": "key"}}
-            ProvisionedThroughput={"ReadCapacityUnits": 1, "WriteCapacityUnits": 5},
-        )
-        cls.item1 = {"id": {"S": "msg1"}, "body": {"S": "some text"}}
-        cls.item2 = {"id": {"S": "msg2"}, "body": {"S": "n/a"}, "unique": {"S": "key"}}
-        cls.client.put_item(TableName="messages", Item=cls.item1)
-        cls.client.put_item(TableName="messages", Item=cls.item2)
 
-    @classmethod
-    def teardown_class(cls):
-        try:
-            cls.mock.stop()
-        except RuntimeError:
-            pass
 
-    def test_execute_statement_select_star(self):
+def create_items(table_name):
-        items = TestSelectStatements.client.execute_statement(
+    client = boto3.client("dynamodb", "us-east-1")
-            Statement="select * from messages"
+    client.put_item(TableName=table_name, Item=item1)
-        )["Items"]
+    client.put_item(TableName=table_name, Item=item2)
-        assert TestSelectStatements.item1 in items
-        assert TestSelectStatements.item2 in items
 
-    def test_execute_statement_select_unique(self):
+
-        items = TestSelectStatements.client.execute_statement(
+@pytest.mark.aws_verified
-            Statement="select unique from messages"
+@dynamodb_aws_verified
-        )["Items"]
+def test_execute_statement_select_star(table_name=None):
+    client = boto3.client("dynamodb", "us-east-1")
+    create_items(table_name)
+    items = client.execute_statement(Statement=f"select * from {table_name}")["Items"]
+    assert item1 in items
+    assert item2 in items
+
+
+@pytest.mark.aws_verified
+@dynamodb_aws_verified
+def test_execute_statement_select_unique(table_name=None):
+    client = boto3.client("dynamodb", "us-east-1")
+    create_items(table_name)
+    items = client.execute_statement(Statement=f"select unique_key from {table_name}")[
+        "Items"
+    ]
     assert {} in items
-        assert {"unique": {"S": "key"}} in items
+    assert {"unique_key": {"S": "key"}} in items
 
-    def test_execute_statement_with_parameter(self):
+
-        stmt = "select * from messages where id = ?"
+@pytest.mark.aws_verified
-        items = TestSelectStatements.client.execute_statement(
+@dynamodb_aws_verified
-            Statement=stmt, Parameters=[{"S": "msg1"}]
+def test_execute_statement_with_parameter(table_name=None):
-        )["Items"]
+    client = boto3.client("dynamodb", "us-east-1")
+    create_items(table_name)
+    stmt = f"select * from {table_name} where pk = ?"
+    items = client.execute_statement(Statement=stmt, Parameters=[{"S": "msg1"}])[
+        "Items"
+    ]
     assert len(items) == 1
-        assert TestSelectStatements.item1 in items
+    assert item1 in items
 
-        stmt = "select id from messages where id = ?"
+    stmt = f"select pk from {table_name} where pk = ?"
-        items = TestSelectStatements.client.execute_statement(
+    items = client.execute_statement(Statement=stmt, Parameters=[{"S": "msg1"}])[
-            Statement=stmt, Parameters=[{"S": "msg1"}]
+        "Items"
-        )["Items"]
+    ]
     assert len(items) == 1
-        assert {"id": {"S": "msg1"}} in items
+    assert {"pk": {"S": "msg1"}} in items
 
-    def test_execute_statement_with_no_results(self):
+
-        stmt = "select * from messages where id = ?"
+@pytest.mark.aws_verified
-        items = TestSelectStatements.client.execute_statement(
+@dynamodb_aws_verified
-            Statement=stmt, Parameters=[{"S": "msg3"}]
+def test_execute_statement_with_no_results(table_name=None):
-        )["Items"]
+    client = boto3.client("dynamodb", "us-east-1")
+    create_items(table_name)
+    stmt = f"select * from {table_name} where pk = ?"
+    items = client.execute_statement(Statement=stmt, Parameters=[{"S": "msg3"}])[
+        "Items"
+    ]
     assert items == []
 
 

tests/test_ec2/test_fleets.py

@@ -57,6 +57,7 @@ class launch_template_context:
         self.ec2.delete_launch_template(LaunchTemplateId=self.lt_id)
 
 
+@pytest.mark.aws_verified
 @ec2_aws_verified
 def test_launch_template_is_created_properly():
     with launch_template_context() as ctxt:
@@ -379,6 +380,7 @@ def test_create_fleet_using_launch_template_config__overrides():
     assert instance["SubnetId"] == subnet_id
 
 
+@pytest.mark.aws_verified
 @ec2_aws_verified
 def test_delete_fleet():
     with launch_template_context() as ctxt:

tests/test_lakeformation/test_resource_tags_integration.py

@@ -1,10 +1,12 @@
 import boto3
+import pytest
 
 from uuid import uuid4
 
 from . import lakeformation_aws_verified
 
 
+@pytest.mark.aws_verified
 @lakeformation_aws_verified
 def test_add_unknown_lf_tags(
     bucket_name=None,  # pylint: disable=unused-argument
@@ -32,6 +34,7 @@ def test_add_unknown_lf_tags(
     }
 
 
+@pytest.mark.aws_verified
 @lakeformation_aws_verified
 def test_tag_lakeformation_database(
     bucket_name=None,  # pylint: disable=unused-argument
@@ -115,6 +118,7 @@ def test_tag_lakeformation_database(
     )
 
 
+@pytest.mark.aws_verified
 @lakeformation_aws_verified
 def test_tag_lakeformation_table(
     bucket_name=None,  # pylint: disable=unused-argument
@@ -198,6 +202,7 @@ def test_tag_lakeformation_table(
     )
 
 
+@pytest.mark.aws_verified
 @lakeformation_aws_verified
 def test_tag_lakeformation_columns(
     bucket_name=None,  # pylint: disable=unused-argument
@@ -361,6 +366,7 @@ def test_tag_lakeformation_columns(
     )
 
 
+@pytest.mark.aws_verified
 @lakeformation_aws_verified
 def test_lf_tags(
     bucket_name=None, db_name=None, table_name=None, column_name=None

tests/test_s3/test_s3_select.py

@@ -1,20 +1,22 @@
-import json
-from unittest import TestCase
-from uuid import uuid4
-
 import boto3
+import json
 import pytest
 
-from moto import mock_s3
+from . import s3_aws_verified
 
 
 SIMPLE_JSON = {"a1": "b1", "a2": "b2", "a3": None}
 SIMPLE_JSON2 = {"a1": "b2", "a3": "b3"}
+NESTED_JSON = {"a1": {"b1": "b2"}, "a2": [True, False], "a3": True, "a4": [1, 5]}
 EXTENSIVE_JSON = [
     {
         "staff": [
-            {"name": "Janelyn M", "city": "Chicago", "kids": 2},
+            {
-            {"name": "Stacy P", "city": "Seattle", "kids": 1},
+                "name": "Janelyn M",
+                "city": "Chicago",
+                "kids": [{"Name": "Josh"}, {"Name": "Jay"}],
+            },
+            {"name": "Stacy P", "city": "Seattle", "kids": {"Name": "Josh"}},
         ],
         "country": "USA",
     }
@@ -26,37 +28,32 @@ y,u,i
 q,w,y"""
 
 
-@mock_s3
+def create_test_files(bucket_name):
-class TestS3Select(TestCase):
+    client = boto3.client("s3", "us-east-1")
-    def setUp(self) -> None:
+    client.put_object(
-        self.client = boto3.client("s3", "us-east-1")
+        Bucket=bucket_name, Key="simple.json", Body=json.dumps(SIMPLE_JSON)
-        self.bucket_name = str(uuid4())
-        self.client.create_bucket(Bucket=self.bucket_name)
-        self.client.put_object(
-            Bucket=self.bucket_name, Key="simple.json", Body=json.dumps(SIMPLE_JSON)
     )
-        self.client.put_object(
+    client.put_object(Bucket=bucket_name, Key="list.json", Body=json.dumps(SIMPLE_LIST))
-            Bucket=self.bucket_name, Key="list.json", Body=json.dumps(SIMPLE_LIST)
+    client.put_object(Bucket=bucket_name, Key="simple_csv", Body=SIMPLE_CSV)
-        )
+    client.put_object(
-        self.client.put_object(
+        Bucket=bucket_name,
-            Bucket=self.bucket_name, Key="simple_csv", Body=SIMPLE_CSV
-        )
-        self.client.put_object(
-            Bucket=self.bucket_name,
         Key="extensive.json",
         Body=json.dumps(EXTENSIVE_JSON),
     )
+    client.put_object(
+        Bucket=bucket_name,
+        Key="nested.json",
+        Body=json.dumps(NESTED_JSON),
+    )
 
-    def tearDown(self) -> None:
-        self.client.delete_object(Bucket=self.bucket_name, Key="list.json")
-        self.client.delete_object(Bucket=self.bucket_name, Key="simple.json")
-        self.client.delete_object(Bucket=self.bucket_name, Key="simple_csv")
-        self.client.delete_object(Bucket=self.bucket_name, Key="extensive.json")
-        self.client.delete_bucket(Bucket=self.bucket_name)
 
-    def test_query_all(self):
+@pytest.mark.aws_verified
-        content = self.client.select_object_content(
+@s3_aws_verified
-            Bucket=self.bucket_name,
+def test_query_all(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="simple.json",
         Expression="SELECT * FROM S3Object",
         ExpressionType="SQL",
@@ -70,20 +67,20 @@ class TestS3Select(TestCase):
     json.loads(result[0]["Records"]["Payload"][0:-1].decode("utf-8"))
 
     # Verify result contains metadata
-        assert {
+    stats = [res for res in result if "Stats" in res][0]["Stats"]
-            "Stats": {
+    assert "BytesScanned" in stats["Details"]
-                "Details": {
+    assert "BytesProcessed" in stats["Details"]
-                    "BytesScanned": 24,
+    assert "BytesReturned" in stats["Details"]
-                    "BytesProcessed": 24,
-                    "BytesReturned": 22,
-                }
-            }
-        } in result
     assert {"End": {}} in result
 
-    def test_count_function(self):
+
-        content = self.client.select_object_content(
+@pytest.mark.aws_verified
-            Bucket=self.bucket_name,
+@s3_aws_verified
+def test_count_function(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="simple.json",
         Expression="SELECT count(*) FROM S3Object",
         ExpressionType="SQL",
@@ -93,10 +90,15 @@ class TestS3Select(TestCase):
     result = list(content["Payload"])
     assert {"Records": {"Payload": b'{"_1":1},'}} in result
 
+
+@pytest.mark.aws_verified
+@s3_aws_verified
 @pytest.mark.xfail(message="Not yet implement in our parser")
-    def test_count_as(self):
+def test_count_as(bucket_name=None):
-        content = self.client.select_object_content(
+    client = boto3.client("s3", "us-east-1")
-            Bucket=self.bucket_name,
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="simple.json",
         Expression="SELECT count(*) as cnt FROM S3Object",
         ExpressionType="SQL",
@@ -106,10 +108,15 @@ class TestS3Select(TestCase):
     result = list(content["Payload"])
     assert {"Records": {"Payload": b'{"cnt":1},'}} in result
 
+
+@pytest.mark.aws_verified
+@s3_aws_verified
 @pytest.mark.xfail(message="Not yet implement in our parser")
-    def test_count_list_as(self):
+def test_count_list_as(bucket_name=None):
-        content = self.client.select_object_content(
+    client = boto3.client("s3", "us-east-1")
-            Bucket=self.bucket_name,
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="list.json",
         Expression="SELECT count(*) as cnt FROM S3Object",
         ExpressionType="SQL",
@@ -119,38 +126,49 @@ class TestS3Select(TestCase):
     result = list(content["Payload"])
     assert {"Records": {"Payload": b'{"cnt":1},'}} in result
 
-    def test_count_csv(self):
+
-        content = self.client.select_object_content(
+@pytest.mark.aws_verified
-            Bucket=self.bucket_name,
+@s3_aws_verified
+def test_count_csv(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="simple_csv",
         Expression="SELECT count(*) FROM S3Object",
         ExpressionType="SQL",
-            InputSerialization={
+        InputSerialization={"CSV": {"FileHeaderInfo": "USE", "FieldDelimiter": ","}},
-                "CSV": {"FileHeaderInfo": "USE", "FieldDelimiter": ","}
-            },
         OutputSerialization={"JSON": {"RecordDelimiter": ","}},
     )
     result = list(content["Payload"])
     assert {"Records": {"Payload": b'{"_1":3},'}} in result
 
-    def test_default_record_delimiter(self):
+
-        content = self.client.select_object_content(
+@pytest.mark.aws_verified
-            Bucket=self.bucket_name,
+@s3_aws_verified
+def test_default_record_delimiter(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="simple_csv",
         Expression="SELECT count(*) FROM S3Object",
         ExpressionType="SQL",
-            InputSerialization={
+        InputSerialization={"CSV": {"FileHeaderInfo": "USE", "FieldDelimiter": ","}},
-                "CSV": {"FileHeaderInfo": "USE", "FieldDelimiter": ","}
-            },
         # RecordDelimiter is not specified - should default to new line (\n)
         OutputSerialization={"JSON": {}},
     )
     result = list(content["Payload"])
     assert {"Records": {"Payload": b'{"_1":3}\n'}} in result
 
-    def test_extensive_json__select_list(self):
+
-        content = self.client.select_object_content(
+@pytest.mark.aws_verified
-            Bucket=self.bucket_name,
+@s3_aws_verified
+def test_extensive_json__select_list(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="extensive.json",
         Expression="select * from s3object[*].staff[*] s",
         ExpressionType="SQL",
@@ -160,9 +178,14 @@ class TestS3Select(TestCase):
     result = list(content["Payload"])
     assert {"Records": {"Payload": b"{},"}} in result
 
-    def test_extensive_json__select_all(self):
+
-        content = self.client.select_object_content(
+@pytest.mark.aws_verified
-            Bucket=self.bucket_name,
+@s3_aws_verified
+def test_extensive_json__select_all(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
         Key="extensive.json",
         Expression="select * from s3object s",
         ExpressionType="SQL",
@@ -170,8 +193,36 @@ class TestS3Select(TestCase):
         OutputSerialization={"JSON": {"RecordDelimiter": ","}},
     )
     result = list(content["Payload"])
-        assert {
+    records = [res for res in result if "Records" in res][0]["Records"][
-            "Records": {
+        "Payload"
-                "Payload": b'{"_1":[{"staff":[{"name":"Janelyn M","city":"Chicago","kids":2},{"name":"Stacy P","city":"Seattle","kids":1}],"country":"USA"}]},'
+    ].decode("utf-8")
-            }
+
-        } in result
+    # For some reason, AWS returns records with a comma at the end
+    assert records[-1] == ","
+
+    # Because the original doc is a list, it is returned like this
+    assert json.loads(records[:-1]) == {"_1": EXTENSIVE_JSON}
+
+
+@pytest.mark.aws_verified
+@s3_aws_verified
+def test_nested_json__select_all(bucket_name=None):
+    client = boto3.client("s3", "us-east-1")
+    create_test_files(bucket_name)
+    content = client.select_object_content(
+        Bucket=bucket_name,
+        Key="nested.json",
+        Expression="select * from s3object s",
+        ExpressionType="SQL",
+        InputSerialization={"JSON": {"Type": "DOCUMENT"}},
+        OutputSerialization={"JSON": {"RecordDelimiter": ","}},
+    )
+    result = list(content["Payload"])
+    records = [res for res in result if "Records" in res][0]["Records"][
+        "Payload"
+    ].decode("utf-8")
+
+    # For some reason, AWS returns records with a comma at the end
+    assert records[-1] == ","
+
+    assert json.loads(records[:-1]) == NESTED_JSON

tests/test_ses/test_ses_boto3.py

@@ -1297,6 +1297,7 @@ def test_render_template():
     )
 
 
+@pytest.mark.aws_verified
 @ses_aws_verified
 def test_render_template__advanced():
     conn = boto3.client("ses", region_name="us-east-1")