AzureSearchDocumentStore<T>

Models & Types · AiDotNet.RetrievalAugmentedGeneration.DocumentStores

Azure AI Search document store backed by the real Azure AI Search REST API.

Inheritance: DocumentStoreBase<T> → AzureSearchDocumentStore<T>

Implements: IDisposable

For Beginners

Azure AI Search is Microsoft’s managed search-and-vector service. This class is a real client for it - every method here makes an HTTP call to your search service. Authentication uses the api-key header and every request carries an api-version query-string parameter.

How It Works

This store talks to an Azure AI Search service over HTTP. It manages a single index with a vector field and an HNSW vector-search profile, upserts documents via the mergeOrUpload action, performs vector search with an OData $filter, deletes documents via the delete action and pages the whole index. The original id and content are stored under id and content; the full metadata dictionary is stored as a JSON string under metadata_json (for lossless round-tripping) while each scalar metadata entry is also flattened to a top-level field so it can be used in $filter expressions.

API deviation: Azure indexes have a fixed schema, so metadata keys used in filters must exist as filterable fields in the index. This client creates the index with the base fields (id, content, metadata_json, embedding); flattened metadata fields are still sent on upload and referenced by $filter, but for production filtering the caller should ensure those fields are declared in the index (or pre-create the index).

Constructors

Constructor Summary
AzureSearchDocumentStore(string indexName, string endpoint, string apiKey, DistanceMetricType distanceMetric, int vectorDimension, string apiVersion, HttpClient httpClient, HttpMessageHandler handler) Initializes a new instance of the AzureSearchDocumentStore class.

Properties

Property Summary
int DocumentCount { get; } Gets the number of documents currently stored in the index.
int VectorDimension { get; } Gets the dimensionality of vectors stored in this index.
string IndexName { get; } Gets the name of the Azure AI Search index this store is bound to.

Methods

Method Summary
Task ClearAsync(CancellationToken cancellationToken)
void Clear()
void Dispose() Releases the HTTP client when this store created it.