# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License 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 annotations
from datetime import datetime
from airflow.providers.amazon.aws.operators.bedrock import BedrockCreateEvaluationJobOperator
from airflow.providers.amazon.aws.operators.s3 import (
S3CreateBucketOperator,
S3CreateObjectOperator,
S3DeleteBucketOperator,
)
from airflow.providers.common.compat.sdk import DAG, chain
from system.amazon.aws.utils import ENV_ID_KEY, SystemTestContextBuilder
from tests_common.test_utils.version_compat import AIRFLOW_V_3_0_PLUS
if AIRFLOW_V_3_0_PLUS:
from airflow.sdk import TriggerRule, task
else:
from airflow.decorators import task # type: ignore[attr-defined,no-redef]
from airflow.utils.trigger_rule import TriggerRule # type: ignore[no-redef,attr-defined]
DAG_ID = "example_bedrock_evaluation"
ROLE_ARN_KEY = "ROLE_ARN"
sys_test_context_task = SystemTestContextBuilder().add_variable(ROLE_ARN_KEY).build()
# Minimal JSONL dataset for Summarization task
EVAL_DATASET = (
'{"prompt": "Summarize: The quick brown fox jumps over the lazy dog.",'
' "referenceResponse": "A fox jumps over a dog."}\n'
)
with DAG(
dag_id=DAG_ID,
schedule=None,
start_date=datetime(2024, 1, 1),
catchup=False,
) as dag:
test_context = sys_test_context_task()
env_id = test_context[ENV_ID_KEY]
role_arn = test_context[ROLE_ARN_KEY]
bucket_name = f"{env_id}-bedrock-eval"
# TEST SETUP
create_bucket = S3CreateBucketOperator(
task_id="create_bucket",
bucket_name=bucket_name,
)
upload_dataset = S3CreateObjectOperator(
task_id="upload_dataset",
s3_bucket=bucket_name,
s3_key="datasets/summarization.jsonl",
data=EVAL_DATASET,
)
# TEST BODY
# [START howto_operator_bedrock_create_evaluation_job]
create_evaluation_job = BedrockCreateEvaluationJobOperator(
task_id="create_evaluation_job",
job_name=f"{env_id}-eval",
role_arn=role_arn,
evaluation_config={
"automated": {
"datasetMetricConfigs": [
{
"taskType": "Summarization",
"dataset": {
"name": "eval-dataset",
"datasetLocation": {"s3Uri": f"s3://{bucket_name}/datasets/summarization.jsonl"},
},
"metricNames": ["Builtin.Accuracy"],
}
]
}
},
inference_config={
"models": [
{
"bedrockModel": {
"modelIdentifier": "us.anthropic.claude-haiku-4-5-20251001-v1:0",
"inferenceParams": "{}",
}
}
]
},
output_data_config={"s3Uri": f"s3://{bucket_name}/output/"},
)
# [END howto_operator_bedrock_create_evaluation_job]
# TEST TEARDOWN
@task(trigger_rule=TriggerRule.ALL_DONE)
def stop_evaluation_job(job_arn: str):
import contextlib
import boto3
with contextlib.suppress(Exception):
boto3.client("bedrock").stop_evaluation_job(jobIdentifier=job_arn)
delete_bucket = S3DeleteBucketOperator(
task_id="delete_bucket",
bucket_name=bucket_name,
force_delete=True,
trigger_rule=TriggerRule.ALL_DONE,
)
chain(
# TEST SETUP
test_context,
create_bucket,
upload_dataset,
# TEST BODY
create_evaluation_job,
# TEST TEARDOWN
stop_evaluation_job(create_evaluation_job.output),
delete_bucket,
)
from tests_common.test_utils.watcher import watcher
list(dag.tasks) >> watcher()
from tests_common.test_utils.system_tests import get_test_run # noqa: E402
test_run = get_test_run(dag)