refine knowledge doc filter

This commit is contained in:
Jicheng Lu 2024-09-20 13:42:08 -05:00
parent b942e7159c
commit e760bd1968
6 changed files with 123 additions and 14 deletions

View file

@ -22,9 +22,37 @@ public interface IKnowledgeService
#endregion
#region Document
/// <summary>
/// Save documents and their contents to knowledgebase
/// </summary>
/// <param name="collectionName"></param>
/// <param name="files"></param>
/// <returns></returns>
Task<UploadKnowledgeResponse> UploadDocumentsToKnowledge(string collectionName, IEnumerable<ExternalFileModel> files);
/// <summary>
/// Save document content to knowledgebase without saving the document
/// </summary>
/// <param name="collectionName"></param>
/// <param name="fileName"></param>
/// <param name="fileSource"></param>
/// <param name="contents"></param>
/// <param name="refData"></param>
/// <returns></returns>
Task<bool> ImportDocumentContentToKnowledge(string collectionName, string fileName, string fileSource, IEnumerable<string> contents, DocMetaRefData? refData = null);
/// <summary>
/// Delete one document and its related knowledge in the collection
/// </summary>
/// <param name="collectionName"></param>
/// <param name="fileId"></param>
/// <returns></returns>
Task<bool> DeleteKnowledgeDocument(string collectionName, Guid fileId);
/// <summary>
/// Delete all documents and their related knowledge in the collection
/// </summary>
/// <param name="collectionName"></param>
/// <param name="filter"></param>
/// <returns></returns>
Task<bool> DeleteKnowledgeDocuments(string collectionName, KnowledgeFileFilter filter);
Task<PagedItems<KnowledgeFileModel>> GetPagedKnowledgeDocuments(string collectionName, KnowledgeFileFilter filter);
Task<FileBinaryDataModel> GetKnowledgeDocumentBinaryData(string collectionName, Guid fileId);
#endregion

View file

@ -4,5 +4,14 @@ public class KnowledgeFileFilter : Pagination
{
public IEnumerable<Guid>? FileIds { get; set; }
public IEnumerable<string>? FileNames { get; set; }
public IEnumerable<string>? ContentTypes { get; set; }
public IEnumerable<string>? FileSources { get; set; }
public KnowledgeFileFilter()
{
}
}

View file

@ -189,10 +189,20 @@ public partial class FileRepository
matched = matched && filter.FileIds.Contains(metaData.FileId);
}
if (!filter.FileNames.IsNullOrEmpty())
{
matched = matched && filter.FileNames.Contains(metaData.FileName);
}
if (!filter.FileSources.IsNullOrEmpty())
{
matched = matched & filter.FileSources.Contains(metaData.FileSource);
}
if (!filter.ContentTypes.IsNullOrEmpty())
{
matched = matched && filter.ContentTypes.Contains(metaData.ContentType);
}
}

View file

@ -149,14 +149,17 @@ public class KnowledgeBaseController : ControllerBase
return response;
}
[HttpDelete("/knowledge/document/{collection}/delete")]
public async Task<bool> DeleteKnowledgeDocuments([FromRoute] string collection, [FromBody] GetKnowledgeDocsRequest request)
{
var response = await _knowledgeService.DeleteKnowledgeDocuments(collection, request);
return response;
}
[HttpPost("/knowledge/document/{collection}/page")]
public async Task<PagedItems<KnowledgeFileViewModel>> GetPagedKnowledgeDocuments([FromRoute] string collection, [FromBody] GetKnowledgeDocsRequest request)
{
var data = await _knowledgeService.GetPagedKnowledgeDocuments(collection, new KnowledgeFileFilter
{
Page = request.Page,
Size = request.Size
});
var data = await _knowledgeService.GetPagedKnowledgeDocuments(collection, request);
return new PagedItems<KnowledgeFileViewModel>
{

View file

@ -3,6 +3,7 @@ using BotSharp.Abstraction.Files.Models;
using BotSharp.Abstraction.Files.Utilities;
using BotSharp.Abstraction.Knowledges.Helpers;
using BotSharp.Abstraction.VectorStorage.Enums;
using System.Collections;
using System.Net.Http;
using System.Net.Mime;
@ -151,8 +152,8 @@ public partial class KnowledgeService
// Get doc meta data
var pageData = db.GetKnowledgeBaseFileMeta(collectionName, vectorStoreProvider, new KnowledgeFileFilter
{
FileIds = [ fileId ],
Size = 1
Size = 1,
FileIds = [ fileId ]
});
// Delete doc
@ -177,6 +178,57 @@ public partial class KnowledgeService
}
}
public async Task<bool> DeleteKnowledgeDocuments(string collectionName, KnowledgeFileFilter filter)
{
if (string.IsNullOrWhiteSpace(collectionName)) return false;
var pageSize = filter.Size;
var innerFilter = new KnowledgeFileFilter
{
Page = 1,
Size = pageSize,
FileIds = filter.FileIds,
FileNames = filter.FileNames,
FileSources = filter.FileSources,
ContentTypes = filter.ContentTypes
};
var pageData = await GetPagedKnowledgeDocuments(collectionName, innerFilter);
var total = pageData.Count;
if (total == 0) return false;
var page = 1;
var totalPages = total % pageSize == 0 ? total / pageSize : total / pageSize + 1;
while (page <= totalPages)
{
if (page > 1)
{
pageData = await GetPagedKnowledgeDocuments(collectionName, innerFilter);
}
var fileIds = pageData.Items.Select(x => x.FileId).ToList();
foreach (var fileId in fileIds)
{
try
{
await DeleteKnowledgeDocument(collectionName, fileId);
}
catch
{
continue;
}
}
page++;
}
return true;
}
public async Task<PagedItems<KnowledgeFileModel>> GetPagedKnowledgeDocuments(string collectionName, KnowledgeFileFilter filter)
{
if (string.IsNullOrWhiteSpace(collectionName))
@ -189,11 +241,7 @@ public partial class KnowledgeService
var vectorStoreProvider = _settings.VectorDb.Provider;
// Get doc meta data
var pagedData = db.GetKnowledgeBaseFileMeta(collectionName, vectorStoreProvider, new KnowledgeFileFilter
{
Page = filter.Page,
Size = filter.Size
});
var pagedData = db.GetKnowledgeBaseFileMeta(collectionName, vectorStoreProvider, filter);
var files = pagedData.Items?.Select(x => new KnowledgeFileModel
{
@ -222,8 +270,8 @@ public partial class KnowledgeService
// Get doc binary data
var pageData = db.GetKnowledgeBaseFileMeta(collectionName, vectorStoreProvider, new KnowledgeFileFilter
{
FileIds = [ fileId ],
Size = 1
Size = 1,
FileIds = [ fileId ]
});
var metaData = pageData?.Items?.FirstOrDefault();
@ -247,6 +295,7 @@ public partial class KnowledgeService
}
#region Private methods
/// <summary>
/// Get file content type and file bytes

View file

@ -191,10 +191,20 @@ public partial class MongoRepository
docFilters.Add(builder.In(x => x.FileId, filter.FileIds));
}
if (!filter.FileNames.IsNullOrEmpty())
{
docFilters.Add(builder.In(x => x.FileName, filter.FileNames));
}
if (!filter.FileSources.IsNullOrEmpty())
{
docFilters.Add(builder.In(x => x.FileSource, filter.FileSources));
}
if (!filter.ContentTypes.IsNullOrEmpty())
{
docFilters.Add(builder.In(x => x.ContentType, filter.ContentTypes));
}
}
var filterDef = builder.And(docFilters);