testbed/aws__sagemaker-python-sdk/tests/integ/test_marketplace.py

# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"). You
# may not use this file except in compliance with the License. A copy of
# the License is located at
#
#     http://aws.amazon.com/apache2.0/
#
# or in the "license" file accompanying this file. This file is
# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
# ANY KIND, either express or implied. See the License for the specific
# language governing permissions and limitations under the License.
from __future__ import absolute_import

import itertools
import os
import time
import requests

import pandas
import pytest
import docker

import sagemaker
import tests.integ
from sagemaker import AlgorithmEstimator, ModelPackage, Model
from sagemaker.serializers import CSVSerializer
from sagemaker.tuner import IntegerParameter, HyperparameterTuner
from sagemaker.utils import sagemaker_timestamp, _aws_partition, unique_name_from_base
from tests.integ import DATA_DIR
from tests.integ.timeout import timeout, timeout_and_delete_endpoint_by_name
from tests.integ.marketplace_utils import REGION_ACCOUNT_MAP
from tests.integ.test_multidatamodel import (
    _ecr_image_uri,
    _ecr_login,
    _create_repository,
    _delete_repository,
)
from tests.integ.retry import retries
import logging

logger = logging.getLogger(__name__)

# All these tests require a manual 1 time subscription to the following Marketplace items:
# Algorithm: Scikit Decision Trees
# https://aws.amazon.com/marketplace/pp/prodview-ha4f3kqugba3u
#
# Pre-Trained Model: Scikit Decision Trees - Pretrained Model
# https://aws.amazon.com/marketplace/pp/prodview-7qop4x5ahrdhe
#
# Both are written by Amazon and are free to subscribe.

ALGORITHM_ARN = (
    "arn:{partition}:sagemaker:{region}:{account}:algorithm/scikit-decision-trees-"
    "15423055-57b73412d2e93e9239e4e16f83298b8f"
)

MODEL_PACKAGE_ARN = (
    "arn:{partition}:sagemaker:{region}:{account}:model-package/scikit-iris-detector-"
    "154230595-8f00905c1f927a512b73ea29dd09ae30"
)


@pytest.mark.release
@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
@pytest.mark.skip(
    reason="This test has always failed, but the failure was masked by a bug. "
    "This test should be fixed. Details in https://github.com/aws/sagemaker-python-sdk/pull/968"
)
def test_marketplace_estimator(sagemaker_session, cpu_instance_type):
    with timeout(minutes=15):
        data_path = os.path.join(DATA_DIR, "marketplace", "training")
        region = sagemaker_session.boto_region_name
        account = REGION_ACCOUNT_MAP[region]
        algorithm_arn = ALGORITHM_ARN.format(
            partition=_aws_partition(region), region=region, account=account
        )

        algo = AlgorithmEstimator(
            algorithm_arn=algorithm_arn,
            role="SageMakerRole",
            instance_count=1,
            instance_type=cpu_instance_type,
            sagemaker_session=sagemaker_session,
        )

        train_input = algo.sagemaker_session.upload_data(
            path=data_path, key_prefix="integ-test-data/marketplace/train"
        )

        algo.fit({"training": train_input})

    endpoint_name = "test-marketplace-estimator{}".format(sagemaker_timestamp())
    with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session, minutes=20):
        predictor = algo.deploy(1, cpu_instance_type, endpoint_name=endpoint_name)
        shape = pandas.read_csv(os.path.join(data_path, "iris.csv"), header=None)

        a = [50 * i for i in range(3)]
        b = [40 + i for i in range(10)]
        indices = [i + j for i, j in itertools.product(a, b)]

        test_data = shape.iloc[indices[:-1]]
        test_x = test_data.iloc[:, 1:]

        print(predictor.predict(test_x.values).decode("utf-8"))


@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
def test_marketplace_attach(sagemaker_session, cpu_instance_type):
    with timeout(minutes=15):
        data_path = os.path.join(DATA_DIR, "marketplace", "training")
        region = sagemaker_session.boto_region_name
        account = REGION_ACCOUNT_MAP[region]
        algorithm_arn = ALGORITHM_ARN.format(
            partition=_aws_partition(region), region=region, account=account
        )

        mktplace = AlgorithmEstimator(
            algorithm_arn=algorithm_arn,
            role="SageMakerRole",
            instance_count=1,
            instance_type=cpu_instance_type,
            sagemaker_session=sagemaker_session,
            base_job_name=unique_name_from_base("test-marketplace"),
        )

        train_input = mktplace.sagemaker_session.upload_data(
            path=data_path, key_prefix="integ-test-data/marketplace/train"
        )

        mktplace.fit({"training": train_input}, wait=False)
        training_job_name = mktplace.latest_training_job.name

        print("Waiting to re-attach to the training job: %s" % training_job_name)
        time.sleep(20)
        endpoint_name = "test-marketplace-estimator{}".format(sagemaker_timestamp())

    with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session, minutes=20):
        print("Re-attaching now to: %s" % training_job_name)
        estimator = AlgorithmEstimator.attach(
            training_job_name=training_job_name, sagemaker_session=sagemaker_session
        )
        predictor = estimator.deploy(
            1, cpu_instance_type, endpoint_name=endpoint_name, serializer=CSVSerializer()
        )
        shape = pandas.read_csv(os.path.join(data_path, "iris.csv"), header=None)
        a = [50 * i for i in range(3)]
        b = [40 + i for i in range(10)]
        indices = [i + j for i, j in itertools.product(a, b)]

        test_data = shape.iloc[indices[:-1]]
        test_x = test_data.iloc[:, 1:]

        print(predictor.predict(test_x.values).decode("utf-8"))


@pytest.mark.release
@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
def test_marketplace_model(sagemaker_session, cpu_instance_type):
    region = sagemaker_session.boto_region_name
    account = REGION_ACCOUNT_MAP[region]
    model_package_arn = MODEL_PACKAGE_ARN.format(
        partition=_aws_partition(region), region=region, account=account
    )

    def predict_wrapper(endpoint, session):
        return sagemaker.Predictor(endpoint, session, serializer=CSVSerializer())

    model = ModelPackage(
        role="SageMakerRole",
        model_package_arn=model_package_arn,
        sagemaker_session=sagemaker_session,
        predictor_cls=predict_wrapper,
    )

    endpoint_name = "test-marketplace-model-endpoint{}".format(sagemaker_timestamp())
    with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session, minutes=20):
        predictor = model.deploy(1, cpu_instance_type, endpoint_name=endpoint_name)
        data_path = os.path.join(DATA_DIR, "marketplace", "training")
        shape = pandas.read_csv(os.path.join(data_path, "iris.csv"), header=None)
        a = [50 * i for i in range(3)]
        b = [40 + i for i in range(10)]
        indices = [i + j for i, j in itertools.product(a, b)]

        test_data = shape.iloc[indices[:-1]]
        test_x = test_data.iloc[:, 1:]

        print(predictor.predict(test_x.values).decode("utf-8"))


@pytest.fixture(scope="module")
def iris_image(sagemaker_session):
    algorithm_name = unique_name_from_base("iris-classifier")
    ecr_image = _ecr_image_uri(sagemaker_session, algorithm_name)
    ecr_client = sagemaker_session.boto_session.client("ecr")
    username, password = _ecr_login(ecr_client)

    docker_client = docker.from_env()

    # Build and tag docker image locally
    path = os.path.join(DATA_DIR, "marketplace", "iris")
    image, build_logs = docker_client.images.build(
        path=path,
        tag=algorithm_name,
        rm=True,
    )
    image.tag(ecr_image, tag="latest")
    _create_repository(ecr_client, algorithm_name)

    # Retry docker image push
    for _ in retries(3, "Upload docker image to ECR repo", seconds_to_sleep=10):
        try:
            docker_client.images.push(
                ecr_image, auth_config={"username": username, "password": password}
            )
            break
        except requests.exceptions.ConnectionError:
            # This can happen when we try to create multiple repositories in parallel, so we retry
            pass

    yield ecr_image

    # Delete repository after the marketplace integration tests complete
    _delete_repository(ecr_client, algorithm_name)


def test_create_model_package(sagemaker_session, boto_session, iris_image):
    MODEL_NAME = "iris-classifier-mp"
    # Prepare
    s3_bucket = sagemaker_session.default_bucket()

    model_name = unique_name_from_base(MODEL_NAME)
    model_description = "This model accepts petal length, petal width, sepal length, sepal width and predicts whether \
    flower is of type setosa, versicolor, or virginica"

    supported_realtime_inference_instance_types = supported_batch_transform_instance_types = [
        "ml.m4.xlarge"
    ]
    supported_content_types = ["text/csv", "application/json", "application/jsonlines"]
    supported_response_MIME_types = ["application/json", "text/csv", "application/jsonlines"]

    validation_input_path = "s3://" + s3_bucket + "/validation-input-csv/"
    validation_output_path = "s3://" + s3_bucket + "/validation-output-csv/"

    iam = boto_session.resource("iam")
    role = iam.Role("SageMakerRole").arn
    sm_client = boto_session.client("sagemaker")
    s3_client = boto_session.client("s3")
    s3_client.put_object(
        Bucket=s3_bucket, Key="validation-input-csv/input.csv", Body="5.1, 3.5, 1.4, 0.2"
    )

    ValidationSpecification = {
        "ValidationRole": role,
        "ValidationProfiles": [
            {
                "ProfileName": "Validation-test",
                "TransformJobDefinition": {
                    "BatchStrategy": "SingleRecord",
                    "TransformInput": {
                        "DataSource": {
                            "S3DataSource": {
                                "S3DataType": "S3Prefix",
                                "S3Uri": validation_input_path,
                            }
                        },
                        "ContentType": supported_content_types[0],
                    },
                    "TransformOutput": {
                        "S3OutputPath": validation_output_path,
                    },
                    "TransformResources": {
                        "InstanceType": supported_batch_transform_instance_types[0],
                        "InstanceCount": 1,
                    },
                },
            },
        ],
    }

    # get pre-existing model artifact stored in ECR
    model = Model(
        image_uri=iris_image,
        model_data=validation_input_path + "input.csv",
        role=role,
        sagemaker_session=sagemaker_session,
        enable_network_isolation=False,
    )

    # Call model.register() - the method under test - to create a model package
    model.register(
        supported_content_types,
        supported_response_MIME_types,
        supported_realtime_inference_instance_types,
        supported_batch_transform_instance_types,
        marketplace_cert=True,
        description=model_description,
        model_package_name=model_name,
        validation_specification=ValidationSpecification,
    )

    # wait for model execution to complete
    time.sleep(60 * 3)

    # query for all model packages with the name <MODEL_NAME>
    response = sm_client.list_model_packages(
        MaxResults=10,
        NameContains=MODEL_NAME,
        SortBy="CreationTime",
        SortOrder="Descending",
    )

    if len(response["ModelPackageSummaryList"]) > 0:
        sm_client.delete_model_package(ModelPackageName=model_name)

    # assert that response is non-empty
    assert len(response["ModelPackageSummaryList"]) > 0


@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
def test_marketplace_tuning_job(sagemaker_session, cpu_instance_type):
    data_path = os.path.join(DATA_DIR, "marketplace", "training")
    region = sagemaker_session.boto_region_name
    account = REGION_ACCOUNT_MAP[region]
    algorithm_arn = ALGORITHM_ARN.format(
        partition=_aws_partition(region), region=region, account=account
    )

    mktplace = AlgorithmEstimator(
        algorithm_arn=algorithm_arn,
        role="SageMakerRole",
        instance_count=1,
        instance_type=cpu_instance_type,
        sagemaker_session=sagemaker_session,
        base_job_name=unique_name_from_base("test-marketplace"),
    )

    train_input = mktplace.sagemaker_session.upload_data(
        path=data_path, key_prefix="integ-test-data/marketplace/train"
    )

    mktplace.set_hyperparameters(max_leaf_nodes=10)

    hyperparameter_ranges = {"max_leaf_nodes": IntegerParameter(1, 100000)}

    tuner = HyperparameterTuner(
        estimator=mktplace,
        base_tuning_job_name=unique_name_from_base("byo"),
        objective_metric_name="validation:accuracy",
        hyperparameter_ranges=hyperparameter_ranges,
        max_jobs=2,
        max_parallel_jobs=2,
    )

    tuner.fit({"training": train_input}, include_cls_metadata=False)
    time.sleep(15)
    tuner.wait()


@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
def test_marketplace_transform_job(sagemaker_session, cpu_instance_type):
    data_path = os.path.join(DATA_DIR, "marketplace", "training")
    region = sagemaker_session.boto_region_name
    account = REGION_ACCOUNT_MAP[region]
    algorithm_arn = ALGORITHM_ARN.format(
        partition=_aws_partition(region), region=region, account=account
    )

    algo = AlgorithmEstimator(
        algorithm_arn=algorithm_arn,
        role="SageMakerRole",
        instance_count=1,
        instance_type=cpu_instance_type,
        sagemaker_session=sagemaker_session,
        base_job_name=unique_name_from_base("test-marketplace"),
    )

    train_input = algo.sagemaker_session.upload_data(
        path=data_path, key_prefix="integ-test-data/marketplace/train"
    )

    shape = pandas.read_csv(data_path + "/iris.csv", header=None).drop([0], axis=1)

    transform_workdir = DATA_DIR + "/marketplace/transform"
    shape.to_csv(transform_workdir + "/batchtransform_test.csv", index=False, header=False)
    transform_input = algo.sagemaker_session.upload_data(
        transform_workdir, key_prefix="integ-test-data/marketplace/transform"
    )

    algo.fit({"training": train_input})

    transformer = algo.transformer(1, cpu_instance_type)
    transformer.transform(transform_input, content_type="text/csv")
    transformer.wait()


@pytest.mark.skipif(
    tests.integ.test_region() in tests.integ.NO_MARKET_PLACE_REGIONS,
    reason="Marketplace is not available in {}".format(tests.integ.test_region()),
)
def test_marketplace_transform_job_from_model_package(sagemaker_session, cpu_instance_type):
    data_path = os.path.join(DATA_DIR, "marketplace", "training")
    shape = pandas.read_csv(data_path + "/iris.csv", header=None).drop([0], axis=1)

    TRANSFORM_WORKDIR = DATA_DIR + "/marketplace/transform"
    shape.to_csv(TRANSFORM_WORKDIR + "/batchtransform_test.csv", index=False, header=False)
    transform_input = sagemaker_session.upload_data(
        TRANSFORM_WORKDIR, key_prefix="integ-test-data/marketplace/transform"
    )

    region = sagemaker_session.boto_region_name
    account = REGION_ACCOUNT_MAP[region]
    model_package_arn = MODEL_PACKAGE_ARN.format(
        partition=_aws_partition(region), region=region, account=account
    )

    model = ModelPackage(
        role="SageMakerRole",
        model_package_arn=model_package_arn,
        sagemaker_session=sagemaker_session,
    )

    transformer = model.transformer(1, cpu_instance_type)
    transformer.transform(transform_input, content_type="text/csv")
    transformer.wait()