时间戳

首页 > 解决方案 > 使用 boto3 调用 AWS SageMaker 终端节点时出错：“无法将数据解析为 JSON。确保 Content-Type 标头设置为“application/json”

amazon-web-services - 使用 boto3 调用 AWS SageMaker 终端节点时出错：“无法将数据解析为 JSON。确保 Content-Type 标头设置为“application/json”

问题描述

我正在尝试使用 boto3 通过以下简单代码调用 AWS SageMaker 端点

import boto3

session = boto3.Session(profile_name='mlacc',
                        region_name='us-west-2')

sagemaker_client = session.client('sagemaker-runtime')

request_body = "{\n    \"requestSource\": \"unittest\",\n    \"clusters\": [{\n        \"clusterMetadata\": {\n "
"\"clusterId\": \"id1\",\n            \"topic\": [\"corona virus\", \"Donald Trump\"],\n            "
"\"clusterSize\": 2\n        },\n        \"documents\": [{\n            \"uid\": \"1\",\n            "
"\"content\": \"content2\",\n            \"domain\": \"CNN.com\",\n            \"title\": \"This is a "
"title\",\n            \"similarityScore\": 1.3,\n            \"published_at\": 1566264017,\n            "
"\"domain_rank\": 1,\n            \"trust_domain_score\": 1\n        }, {\n            \"uid\": \"2\","
"\n            \"content\": \"content2\",\n            \"domain\": \"CNN.com\",\n            \"title\": "
"\"This is a title\",\n            \"similarityScore\": 1.3,\n            \"published_at\": 1566264017,"
"\n            \"domain_rank\": 1,\n            \"trust_domain_score\": 1\n        }, {\n            \"uid\": "
"\"2\",\n            \"content\": \"content3\",\n            \"domain\": \"CNN.com\",\n            \"title\": "
"\"This is a title\",\n            \"similarityScore\": 1.3,\n            \"published_at\": 1566264017,"
"\n            \"domain_rank\": 1,\n            \"trust_domain_score\": 1\n        }]\n    }]\n}"


response = sagemaker_client.invoke_endpoint(
    EndpointName='myEndpoint22',
    Body=request_body,
    ContentType='application/json',
)

response_json = response['Body'].read().decode('utf-8')

print(response_json)

运行此代码时出现以下错误

Traceback (most recent call last):
  File "/Users/rppatwa/Desktop/WorkDocs/CodePlayground/SimplePythonProject/src/PrototypeTesting/SummarizationLocal.py", line 205, in <module>
    main()
  File "/Users/rppatwa/Desktop/WorkDocs/CodePlayground/SimplePythonProject/src/PrototypeTesting/SummarizationLocal.py", line 186, in main
    ContentType='application/json',
  File "/Users/rppatwa/anaconda3/lib/python3.7/site-packages/botocore/client.py", line 316, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "/Users/rppatwa/anaconda3/lib/python3.7/site-packages/botocore/client.py", line 635, in _make_api_call
    raise error_class(parsed_response, operation_name)
botocore.errorfactory.ModelError: An error occurred (ModelError) when calling the InvokeEndpoint operation: Received client error (400) from model with message "Unable to parse data as JSON. Make sure the Content-Type header is set to "application/json"". See https://us-west-2.console.aws.amazon.com/cloudwatch/home?region=us-west-2#logEventViewer:group=/aws/sagemaker/Endpoints/KeyurshaASMLModel in account 753843489946 for more information.

如果我内联正文（不使用 request_json），则此调用成功。请让我知道我错过了什么。

谢谢

标签： amazon-web-servicesboto3amazon-sagemaker

解决方案

您需要删除尾随逗号， ContentType='application/json',然后尝试下面的代码段以将 JSON 传递到正文字段。

import json 
json.dumps(request_body) 
test=json.dumps(request_body).encode()

这也将验证您正在传递的 JSON。现在将测试传递给 body 以调用端点。

推荐阅读