SharePoint indexer - DocumentIntelligenceLayoutSkill produces error

Question

SharePoint indexer - DocumentIntelligenceLayoutSkill produces error

Nikita DE BROUX (Sagacify) 0

Hello, I am trying to build a sharepoint indexer. The data from the sharepoint is then passed through a skillset with DocumentIntelligenceLayoutSkill. This skill is preceded by a custom skill to pass the correct data format (url format) to DocumentIntelligenceLayoutSkill (https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-document-intelligence-layout#data-limits).

Here are my skills:

  "skills": [
    {
      "@odata.type": "#Microsoft.Skills.Custom.WebApiSkill",
      "name": "#1",
      "description": "Custom skill to prepare data for layout intelligence",
      "context": "/document",
      "uri": "https://my-custom-skill.azurewebsites.net/api/-prepare-data-fct?code=[my_key]",
      "httpMethod": "POST",
      "timeout": "PT30S",
      "batchSize": 1000,
      "inputs": [
        {
          "name": "document_uri",
          "source": "/document/document_uri",
          "inputs": []
        }
      ],
      "outputs": [
        {
          "name": "layout_intelligence_data",
          "targetName": "layout_intelligence_data"
        }
      ],
      "httpHeaders": {}
    },
    {
      "@odata.type": "#Microsoft.Skills.Util.DocumentIntelligenceLayoutSkill",
      "name": "my_document_intelligence_layout_skill",
      "context": "/document",
      "outputMode": "oneToMany",
      "markdownHeaderDepth": "h3",
      "inputs": [
        {
          "name": "file_data",
          "source": "/document/layout_intelligence_data",
          "inputs": []
        }
      ],
      "outputs": [
        {
          "name": "markdown_document",
          "targetName": "markdown_document"
        }
      ]
    }
  ]

The indexer produces this error:

User's image

Does somebody have any hints/possibilities to solve the problem ? I tested my indexer with only the custom skill and it works just fine.

Thanks

Laxman Reddy Revuri 5,475 Reputation points Microsoft External Staff Moderator

2025-03-21T06:53:27.57+00:00

Hi @Nikita DE BROUX (Sagacify)
The "Could not execute skill" error in your SharePoint indexer with DocumentIntelligenceLayoutSkill typically occurs due to issues with input data, document accessibility, or service limitations.
1.Ensure the data passed from the custom skill matches the expected format for DocumentIntelligenceLayoutSkill.
2.Verify that SharePoint documents are publicly accessible or properly authenticated.
3.Ensure documents do not exceed the 4MB restriction for processing.
4.. Increase timeout settings if needed and monitor performance.

5.Check Azure logs for any temporary issues affecting skill execution. To resolve the issue, enable detailed skill execution logs, validate the custom skill output, and test with smaller batches before scaling. If the problem persists, try running the custom skill independently to isolate the failure point.

6.Verify that the input field specified in the DocumentIntelligenceLayoutSkill matches the output name from your custom skill. In your case, the custom skill outputs layout_intelligence_data, which should be correctly referenced in the DocumentIntelligenceLayoutSkill.
references:
Troubleshooting common indexer errors and warnings in Azure AI Search
Nikita DE BROUX (Sagacify) 0 Reputation points

2025-03-21T09:23:43.1833333+00:00
Hi @Laxman Reddy Revuri , thanks for your answer!

Here is the format output of my custom skill:

{ "values": [ { "recordId": "a1", "data": { "layout_intelligence_data": { "$type": "file", "url": "url1 " } } }, { "recordId": "b5", "data": { "layout_intelligence_data": { "$type": "file", "url": "url1 " } } } ] }

From what I understand, this is the right format I should have. I checked the logs of my function during indexer execution and the url are valid (I can access them). Do I need to give special rights in my skillset to access the sharepoint ? Knowing that that my Azure AI Search resource already has the permissions. How to increase timeout settings for a built-in skill ? All I see is running the indexer on a schedule.

But since my indexer always stops on the first doc going through the DocumentLayoutSkill, I do not think thats the issue.
Nikita DE BROUX (Sagacify) 0 Reputation points

2025-03-21T09:27:16.19+00:00

I did try to run the custom skill indepedently and it works just fine. Error always occurs on the first doc using DocumentLayoutIntelligence (I am not using free tier so size limit is 500 MB)
Laxman Reddy Revuri 5,475 Reputation points Microsoft External Staff Moderator

2025-03-24T04:23:23.3+00:00

Hi @Nikita DE BROUX (Sagacify)
apologize for any inconvenience caused on this.
Unfortunately, built-in skills such as DocumentIntelligenceLayoutSkill do not have a configurable timeout option like custom skills.

However, you can increase the timeout of the overall indexer:

In the indexer definition,

"parameters": {

``"maxFailedItems": -1,

``"maxFailedItemsPerBatch": -1

}

This prevents the indexer from stopping after the first failure. Ensure layout_intelligence_data is being passed exactly as expected by DocumentIntelligenceLayoutSkill.
If the indexer consistently fails on the first document, you may want to test with different documents, such as smaller files or those in different formats. Additionally, consider running the skill on a single document to help identify the issue.
https://learn.microsoft.com/en-us/azure/search/cognitive-search-how-to-debug-skillset
Nikita DE BROUX (Sagacify) 0 Reputation points

2025-03-24T11:05:31.1766667+00:00
Hello, I tested with
"parameters": {

``"maxFailedItems": -1,

``"maxFailedItemsPerBatch": -1

}

and the skillset (DocumentLayoutIntelligence) is not working on a single one of them. Thery are files of various size, extensions (around 19000 files) so I am sure the issue is not related to a problem of extension or size. As mentionned before:

Here is the format output of my custom skill:

{ "values": [ { "recordId": "a1", "data": { "layout_intelligence_data": { "$type": "file", "url": "url1 " } } }, { "recordId": "b5", "data": { "layout_intelligence_data": { "$type": "file", "url": "url1 " } } } ] }

From what I understand, this is the right format I should have. I checked the logs of my function during indexer execution and the url are valid (I can access them). Do I need to give special rights in my skillset to access the sharepoint ? Knowing that that my Azure AI Search resource already has the permissions.

I am pretty sure the problem occurs because the skillset cannot access the url but how to confirm that and manage the problem ? You mention debug plateform but it is not accessible for SharePoint datasource...
Sampath 3,880 Reputation points Microsoft External Staff Moderator

2025-03-27T09:32:26.7433333+00:00

Hello @Nikita DE BROUX (Sagacify) , I noticed the same question was raised by a user. https://learn.microsoft.com/fr-fr/answers/questions/2239060/sharepoint-indexer-documentintelligencelayoutskill which results in a duplicate question which can be closed as a duplicate. Thank you.
Nikita DE BROUX (Sagacify) 0 Reputation points

2025-03-27T09:40:22.07+00:00

Yes I am the one who ask the question both times, you can close this one and keep the other open
Sampath 3,880 Reputation points Microsoft External Staff Moderator

2025-03-27T10:04:56.1766667+00:00

Closing this as a duplicate of https://learn.microsoft.com/fr-fr/answers/questions/2239060/sharepoint-indexer-documentintelligencelayoutskill?comment=question#comment-1986878 . Thank you.