[BUG] Azure Search result deserialization cannot handle document debug info

[mattgotteiner]

Library name and version

Azure.Search.Documents 11.7.0-beta.1

Describe the bug

Attempting to use QueryDebugMode results in the following error:

System.InvalidOperationException
  HResult=0x80131509
  Message=The requested operation requires an element of type 'Array', but the target element has type 'Object'.
  Source=System.Text.Json
  StackTrace:
   at System.Text.Json.ThrowHelper.ThrowJsonElementWrongTypeException(JsonTokenType expectedType, JsonTokenType actualType)
   at System.Text.Json.JsonElement.EnumerateArray()
   at Azure.Search.Documents.Models.SearchResult`1.<DeserializeAsync>d__21.MoveNext()
   at Azure.Search.Documents.Models.SearchResults`1.<DeserializeAsync>d__37.MoveNext()
   at Azure.Search.Documents.SearchClient.<SearchInternal>d__41`1.MoveNext()
   at Azure.Search.Documents.SearchClient.<SearchInternal>d__40`1.MoveNext()
   at Azure.Core.Pipeline.TaskExtensions.EnsureCompleted[T](Task`1 task)
   at Azure.Search.Documents.SearchClient.Search[T](String searchText, SearchOptions options, CancellationToken cancellationToken)
   at Program.<Main>$(String[] args) in C:\Users\magottei\source\repos\minrepro\Program.cs:line 7

Expected behavior

1. debug is a parameter on the search request. It should not be required to set SemanticSearch on SearchOptions to use this property. Here's a sample REST request demonstrating this:

{
  "search": "*",
  "top": 1,
  "vectorQueries": [
    {
      "kind": "text",
      "text": "test",
      "fields": "contentVector"
    }
  ],
  "debug": "vector"
}

2. The "vector" value of QueryDebugMode is missing despite being present in the swagger. It should not be required to create a new instance of QueryDebugMode to use this value
3. DocumentDebugInfo is deserialized as an array when it should be treated as an object. Here's a sample REST response corresponding to the query in the repro below:

{
  "value": [
    {
      "@search.score": 0.61361045,
      "@search.documentDebugInfo": {
        "semantic": null,
        "vectors": {
          "subscores": {
            "documentBoost": null,
            "text": null,
            "vectors": [
              {
                "contentVector": {
                  "searchScore": 0.61361045,
                  "vectorSimilarity": 0.37030154880836874
                }
              }
            ]
          }
        }
      },
....

Actual behavior

The repro crashes before the search results are fully deserialized, preventing use of debug property

Reproduction Steps

1. Use the sample to populate a sample index with vector values
2. Run the following C# code to reproduce the bug using the index from the sample above:

using Azure.Identity;
using Azure.Search.Documents;
using Azure.Search.Documents.Models;

var client = new SearchClient(new Uri("https://service.search.windows.net"), "test-dotnet-vector-sample", new DefaultAzureCredential());

client.Search<SearchDocument>(
    searchText: "*",
    new SearchOptions
    {
        Size = 1,
        VectorSearch = new VectorSearchOptions
        {
            Queries =
            {
                new VectorizableTextQuery("test") { Fields = { "contentVector" } }
            }
        },
        SemanticSearch = new SemanticSearchOptions { Debug = new QueryDebugMode("vector") }
    });

Environment

.NET SDK:
Version: 8.0.403
Commit: c64aa40a71
Workload version: 8.0.400-manifests.e99c892e
MSBuild version: 17.11.9+a69bbaaf5

Runtime Environment:
OS Name: Windows
OS Version: 10.0.26100
OS Platform: Windows
RID: win-x64
Base Path: C:\Program Files\dotnet\sdk\8.0.403\

[github-actions]

Thank you for your feedback. Tagging and routing to the team member best able to assist.

[ShivangiReja]

@mattgotteiner We’ve fixed the issue in Azure.Search.Documents version 11.7.0-beta.2. Could you please check if it works for you?