在 Azure 表单识别器中处理文档时出现 HttpRequestException

Question

我们在使用发票 Azure 表单识别器容器时遇到了困难。在我们发送文档进行处理后，我们会收到“失败”状态。我们尝试了两种方式发送：application/json 和 application/octet-stream，但总是出现错误。

具体来说，这是一个响应示例：

{
  "status": "failed",
  "createdDateTime": "2024-01-02T15:53:56Z",
  "lastUpdatedDateTime": "2024-01-02T15:53:56Z",
  "error": {
    "code": "InternalServerError",
    "message": "An unexpected error occurred.",
    "details": [
      {
        "code": "HttpRequestException",
        "message": "Could not connect to upstream host: An invalid request URI was provided. Either the request URI must be an absolute URI or BaseAddress must be set.",
        "target": "1"
      }
    ]
  }
}

我们启动这个容器的方式：

~$ dockerun --rm -it -p 5000:5050 --memory 16g --cpus 6 -e Eula=accept -e Billing=https://xxxx.cognitiveservices.azure.com -e ApiKey=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx mcr.microsoft.com/azure-cognitive-services/form-recognizer/invoice-3.0

控制台显示错误：

fail: formrecognizerinvoice[0]

      {"ClassName":"Microsoft.CloudAI.Containers.VDI.Common.Errors.VdiException","Message":"HttpRequestException","Data":null,"InnerException":null,"HelpURL":null,"StackTraceString":"   at Microsoft.CloudAI.Containers.VDI.OnPremises.Common.Handler.FrBaseHandler.HttpPostAsync(String requestUri, HttpContent content)\n   at Microsoft.CloudAI.Containers.VDI.OnPremises.Common.Handler.Fr30GaUpstreamHandler.PostUpstreamInternalRequest(String modelId, Byte[] stream, ServiceMessage message)\n   at Microsoft.CloudAI.Containers.VDI.OnPremises.Common.Handler.Fr30GaUpstreamHandler.HandleMessageAsync(ServiceMessage message)","RemoteStackTraceString":null,"RemoteStackIndex":0,"ExceptionMethod":null,"HResult":-2146233088,"Source":"Microsoft.CloudAI.Containers.VDI.OnPremises.Common","WatsonBuckets":null} SubscriptionId='' RequestId='ebab790b-3e6e-4c0b-af93-9396cb609fd8' Timestamp='' buildVersion='1.2.367.0-20230707.6-27dcd7ae972075713bb9e4477fe50c6e623f1596' hostId='a71ffbc4bc9f' serviceName='formrecognizerinvoice' namespace='vdi' Namesapce='vdi' fileSizeInBytes='100040' processedPageIndex='0' toSecondaryQueue='False' totalPageCount='2' fileType='Pdf' apiName='invoice-2022-08-31' ocrModelName='2023-07-31' scenario='fr_async_layout' resultNameEnum='UpstreamFRResult' ocrResolution='Standard' errorCode='HttpRequestException'

我们如何解决这个问题？

我们正在使用另一个容器测试同一文件：布局表单识别器，它工作正常。

Answer 1

根据文档，

Invoice

功能容器需要支持

Layout

容器。

因此，该问题可能是由于您在不支持布局容器的情况下启动容器，并且可以通过在请求中包含支持容器来解决。

下面是一个示例 YAML

invoice

配置供参考：

version: "3.9"
services:
  azure-cognitive-service-invoice:
    container_name: azure-cognitive-service-invoice
    image: mcr.microsoft.com/azure-cognitive-services/form-recognizer/invoice-3.0
    environment:
        - EULA=accept
        - billing={FORM_RECOGNIZER_ENDPOINT_URI}
        - apiKey={FORM_RECOGNIZER_KEY}
        - AzureCognitiveServiceLayoutHost=http://azure-cognitive-service-layout:5000
    ports:
      - "5000:5050"
  azure-cognitive-service-layout:
    container_name: azure-cognitive-service-layout
    image: mcr.microsoft.com/azure-cognitive-services/form-recognizer/layout-3.0
    environment:
        - EULA=accept
        - billing={FORM_RECOGNIZER_ENDPOINT_URI}
        - apiKey={FORM_RECOGNIZER_KEY}