BotSharp/src/Plugins/BotSharp.Plugin.MongoStorage/Repository/MongoRepository.KnowledgeBase.cs

227 lines
8.1 KiB
C#
Raw Normal View History

using BotSharp.Abstraction.Knowledges.Models;
2024-09-10 19:02:25 +00:00
using BotSharp.Abstraction.VectorStorage.Models;
namespace BotSharp.Plugin.MongoStorage.Repository;
public partial class MongoRepository
{
#region Configs
2024-09-10 19:02:25 +00:00
public bool AddKnowledgeCollectionConfigs(List<VectorCollectionConfig> configs, bool reset = false)
{
var filter = Builders<KnowledgeCollectionConfigDocument>.Filter.Empty;
var docs = configs?.Where(x => !string.IsNullOrWhiteSpace(x.Name))
.Select(x => new KnowledgeCollectionConfigDocument
{
Id = Guid.NewGuid().ToString(),
Name = x.Name,
Type = x.Type,
2024-09-12 16:51:23 +00:00
VectorStore = KnowledgeVectorStoreConfigMongoModel.ToMongoModel(x.VectorStore),
2024-09-10 19:02:25 +00:00
TextEmbedding = KnowledgeEmbeddingConfigMongoModel.ToMongoModel(x.TextEmbedding)
})?.ToList() ?? new List<KnowledgeCollectionConfigDocument>();
if (reset)
{
_dc.KnowledgeCollectionConfigs.DeleteMany(filter);
_dc.KnowledgeCollectionConfigs.InsertMany(docs);
return true;
}
// Update if collection already exists, otherwise insert.
var insertDocs = new List<KnowledgeCollectionConfigDocument>();
var updateDocs = new List<KnowledgeCollectionConfigDocument>();
var names = docs.Select(x => x.Name).ToList();
filter = Builders<KnowledgeCollectionConfigDocument>.Filter.In(x => x.Name, names);
var savedConfigs = _dc.KnowledgeCollectionConfigs.Find(filter).ToList();
foreach (var doc in docs)
{
var found = savedConfigs.FirstOrDefault(x => x.Name == doc.Name);
if (found != null)
{
found.Type = doc.Type;
2024-09-12 16:51:23 +00:00
found.VectorStore = doc.VectorStore;
2024-09-10 19:02:25 +00:00
found.TextEmbedding = doc.TextEmbedding;
updateDocs.Add(found);
}
else
{
insertDocs.Add(doc);
}
}
if (!insertDocs.IsNullOrEmpty())
{
_dc.KnowledgeCollectionConfigs.InsertMany(docs);
}
if (!updateDocs.IsNullOrEmpty())
{
foreach (var doc in updateDocs)
{
filter = Builders<KnowledgeCollectionConfigDocument>.Filter.Eq(x => x.Id, doc.Id);
_dc.KnowledgeCollectionConfigs.ReplaceOne(filter, doc);
}
}
return true;
}
public bool DeleteKnowledgeCollectionConfig(string collectionName)
{
if (string.IsNullOrWhiteSpace(collectionName)) return false;
var filter = Builders<KnowledgeCollectionConfigDocument>.Filter.Eq(x => x.Name, collectionName);
var deleted = _dc.KnowledgeCollectionConfigs.DeleteMany(filter);
return deleted.DeletedCount > 0;
}
public IEnumerable<VectorCollectionConfig> GetKnowledgeCollectionConfigs(VectorCollectionConfigFilter filter)
{
if (filter == null)
{
return Enumerable.Empty<VectorCollectionConfig>();
}
var builder = Builders<KnowledgeCollectionConfigDocument>.Filter;
var filters = new List<FilterDefinition<KnowledgeCollectionConfigDocument>> { builder.Empty };
// Apply filters
if (!filter.CollectionNames.IsNullOrEmpty())
{
filters.Add(builder.In(x => x.Name, filter.CollectionNames));
}
if (!filter.CollectionTypes.IsNullOrEmpty())
{
filters.Add(builder.In(x => x.Type, filter.CollectionTypes));
}
if (!filter.VectorStroageProviders.IsNullOrEmpty())
{
2024-09-12 16:51:23 +00:00
filters.Add(builder.In(x => x.VectorStore.Provider, filter.VectorStroageProviders));
2024-09-10 19:02:25 +00:00
}
// Get data
var configs = _dc.KnowledgeCollectionConfigs.Find(Builders<KnowledgeCollectionConfigDocument>.Filter.And(filters)).ToList();
return configs.Select(x => new VectorCollectionConfig
{
Name = x.Name,
Type = x.Type,
2024-09-12 16:51:23 +00:00
VectorStore = KnowledgeVectorStoreConfigMongoModel.ToDomainModel(x.VectorStore),
2024-09-10 19:02:25 +00:00
TextEmbedding = KnowledgeEmbeddingConfigMongoModel.ToDomainModel(x.TextEmbedding)
});
}
#endregion
#region Documents
public bool SaveKnolwedgeBaseFileMeta(KnowledgeDocMetaData metaData)
{
if (metaData == null
|| string.IsNullOrWhiteSpace(metaData.Collection)
2024-09-17 17:50:39 +00:00
|| string.IsNullOrWhiteSpace(metaData.VectorStoreProvider))
{
return false;
}
2024-09-17 17:50:39 +00:00
var doc = new KnowledgeCollectionFileMetaDocument
{
2024-09-14 00:11:41 +00:00
Id = Guid.NewGuid().ToString(),
Collection = metaData.Collection,
FileId = metaData.FileId,
FileName = metaData.FileName,
FileSource = metaData.FileSource,
ContentType = metaData.ContentType,
VectorStoreProvider = metaData.VectorStoreProvider,
VectorDataIds = metaData.VectorDataIds,
2024-09-18 15:35:28 +00:00
RefData = KnowledgeFileMetaRefMongoModel.ToMongoModel(metaData.RefData),
CreateDate = metaData.CreateDate,
CreateUserId = metaData.CreateUserId
};
2024-09-17 17:50:39 +00:00
_dc.KnowledgeCollectionFileMeta.InsertOne(doc);
return true;
}
2024-09-17 17:50:39 +00:00
public bool DeleteKnolwedgeBaseFileMeta(string collectionName, string vectorStoreProvider, Guid? fileId = null)
{
if (string.IsNullOrWhiteSpace(collectionName)
|| string.IsNullOrWhiteSpace(vectorStoreProvider))
{
return false;
}
var builder = Builders<KnowledgeCollectionFileMetaDocument>.Filter;
var filters = new List<FilterDefinition<KnowledgeCollectionFileMetaDocument>>()
{
builder.Eq(x => x.Collection, collectionName),
builder.Eq(x => x.VectorStoreProvider, vectorStoreProvider)
};
if (fileId != null)
{
filters.Add(builder.Eq(x => x.FileId, fileId));
}
var res = _dc.KnowledgeCollectionFileMeta.DeleteMany(builder.And(filters));
return res.DeletedCount > 0;
}
public PagedItems<KnowledgeDocMetaData> GetKnowledgeBaseFileMeta(string collectionName, string vectorStoreProvider, KnowledgeFileFilter filter)
{
if (string.IsNullOrWhiteSpace(collectionName)
|| string.IsNullOrWhiteSpace(vectorStoreProvider))
{
return new PagedItems<KnowledgeDocMetaData>();
}
2024-09-17 17:50:39 +00:00
var builder = Builders<KnowledgeCollectionFileMetaDocument>.Filter;
var docFilters = new List<FilterDefinition<KnowledgeCollectionFileMetaDocument>>()
{
builder.Eq(x => x.Collection, collectionName),
builder.Eq(x => x.VectorStoreProvider, vectorStoreProvider)
};
// Apply filters
2024-09-19 16:43:49 +00:00
if (filter != null)
{
2024-09-19 16:43:49 +00:00
if (!filter.FileIds.IsNullOrEmpty())
{
docFilters.Add(builder.In(x => x.FileId, filter.FileIds));
}
if (!filter.FileSources.IsNullOrEmpty())
{
docFilters.Add(builder.In(x => x.FileSource, filter.FileSources));
}
}
var filterDef = builder.And(docFilters);
2024-09-17 17:50:39 +00:00
var sortDef = Builders<KnowledgeCollectionFileMetaDocument>.Sort.Descending(x => x.CreateDate);
var docs = _dc.KnowledgeCollectionFileMeta.Find(filterDef).Sort(sortDef).Skip(filter.Offset).Limit(filter.Size).ToList();
var count = _dc.KnowledgeCollectionFileMeta.CountDocuments(filterDef);
var files = docs?.Select(x => new KnowledgeDocMetaData
{
Collection = x.Collection,
FileId = x.FileId,
FileName = x.FileName,
FileSource = x.FileSource,
ContentType = x.ContentType,
VectorStoreProvider = x.VectorStoreProvider,
VectorDataIds = x.VectorDataIds,
2024-09-18 15:35:28 +00:00
RefData = KnowledgeFileMetaRefMongoModel.ToDomainModel(x.RefData),
CreateDate = x.CreateDate,
CreateUserId = x.CreateUserId
2024-09-18 15:35:28 +00:00
})?.ToList() ?? new();
return new PagedItems<KnowledgeDocMetaData>
{
Items = files,
Count = (int)count
};
}
#endregion
2024-09-10 19:02:25 +00:00
}