FileSystemEmbeddingStorageService

Namespace: Wisej.AI.Services

Assembly: Wisej.AI

Provides a file system-based implementation of the IEmbeddingStorageService interface.

C#

public class FileSystemEmbeddingStorageService : IEmbeddingStorageService

VB.NET

Public Class FileSystemEmbeddingStorageService
    Inherits IEmbeddingStorageService

This service allows for storing, retrieving, querying, and removing embedded documents from a specified storage path in the file system. It uses JSON files to store metadata and embeddings for each document.

Constructors

FileSystemEmbeddingStorageService()

Initializes a new instance of FileSystemEmbeddingStorageService.

Properties

StoragePath

String: Gets or sets the storage path for embeddings. (Default: "~\AI\Embeddings")

Methods

ExistsAsync(collectionName, documentName)

Checks if a document exists in the specified collection.

Parameter	Type	Description
collectionName	String	The name of the collection.
documentName	String	The name of the document.

Returns: Task<Boolean>. A task that represents the asynchronous operation. The task result contains a boolean indicating whether the document exists.

QueryAsync(collectionName, query, topN, minSimilarity, filter)

Queries the collection for documents similar to the provided query vector.

Parameter	Type	Description
collectionName	String	The name of the collection.
query	Single[]	The query vector.
topN	Int32	The number of top similar documents to retrieve.
minSimilarity	Single	The minimum similarity threshold.
filter	Predicate<EmbeddedDocument>	An optional filter predicate for documents. Default is null.

Returns: Task<EmbeddedDocument[]>. A task that represents the asynchronous operation. The task result contains an array of EmbeddedDocument objects.

QueryAsync(collectionName, documentName, query, topN, minSimilarity)

Queries a specific document in the collection for similarity to the provided query vector.

Parameter	Type	Description
collectionName	String	The name of the collection.
documentName	String	The name of the document.
query	Single[]	The query vector.
topN	Int32	The number of top similar chunks to retrieve.
minSimilarity	Single	The minimum similarity threshold.

Returns: Task<EmbeddedDocument>. A task that represents the asynchronous operation. The task result contains the EmbeddedDocument with the top similar chunks.

Throws:

• ArgumentNullException Thrown when the document is null.

RemoveAsync(collectionName, filter)

Removes documents from the specified collection that match the filter predicate.

Parameter	Type	Description
collectionName	String	The name of the collection.
filter	Predicate<EmbeddedDocument>	An optional filter predicate for documents. Default is null.

Returns: Task. A task that represents the asynchronous operation.

RemoveAsync(collectionName, documentName)

Removes a specific document from the specified collection.

Parameter	Type	Description
collectionName	String	The name of the collection.
documentName	String	The name of the document.

Returns: Task. A task that represents the asynchronous operation.

RetrieveAsync(collectionName, documentName, includeEmbedding)

Retrieves a specific document from the specified collection.

Parameter	Type	Description
collectionName	String	The name of the collection.
documentName	String	The name of the document.
includeEmbedding	Boolean	Indicates whether to include the embedding in the retrieved document.

Returns: Task<EmbeddedDocument>. A task that represents the asynchronous operation. The task result contains the EmbeddedDocument.

RetrieveAsync(collectionName, includeEmbedding, filter)

Retrieves all documents from the specified collection that match the filter predicate.

Parameter	Type	Description
collectionName	String	The name of the collection.
includeEmbedding	Boolean	Indicates whether to include embeddings in the retrieved documents.
filter	Predicate<EmbeddedDocument>	An optional filter predicate for documents. Default is null.

Returns: Task<EmbeddedDocument[]>. A task that represents the asynchronous operation. The task result contains an array of EmbeddedDocument objects.

StoreAsync(collectionName, document)

Stores a document in the specified collection.

Parameter	Type	Description
collectionName	String	The name of the collection.
document	EmbeddedDocument	The document to store.

Returns: Task. A task that represents the asynchronous operation.

Throws:

• ArgumentNullException Thrown when the document is null.

Implements

Name	Description
IEmbeddingStorageService	Represents a service for storing and querying embedded documents within specified collections.