BotSharp/src/Plugins/BotSharp.Plugin.Qdrant/QdrantDb.cs

152 lines
4.7 KiB
C#
Raw Normal View History

2024-08-06 22:43:11 +00:00
using BotSharp.Abstraction.Utilities;
2023-11-14 01:25:25 +00:00
using Qdrant.Client;
using Qdrant.Client.Grpc;
2023-06-18 02:56:22 +00:00
namespace BotSharp.Plugin.Qdrant;
public class QdrantDb : IVectorDb
{
2024-01-06 03:24:13 +00:00
private QdrantClient _client;
2023-06-18 02:56:22 +00:00
private readonly QdrantSetting _setting;
2023-06-27 23:36:50 +00:00
private readonly IServiceProvider _services;
public QdrantDb(QdrantSetting setting,
IServiceProvider services)
2023-06-18 02:56:22 +00:00
{
_setting = setting;
2023-06-27 23:36:50 +00:00
_services = services;
2024-01-06 03:24:13 +00:00
}
private QdrantClient GetClient()
{
if (_client == null)
{
_client = new QdrantClient
(
host: _setting.Url,
2024-07-02 14:52:16 +00:00
https: true,
2024-01-06 03:24:13 +00:00
apiKey: _setting.ApiKey
);
}
return _client;
2023-06-18 02:56:22 +00:00
}
public async Task<List<string>> GetCollections()
{
// List all the collections
2024-07-02 14:52:16 +00:00
var collections = await GetClient().ListCollectionsAsync();
2023-11-14 01:25:25 +00:00
return collections.ToList();
2023-06-18 02:56:22 +00:00
}
2024-08-06 21:02:02 +00:00
public async Task<KnowledgeCollectionInfo> GetCollectionInfo(string collectionName)
{
2024-08-06 22:43:11 +00:00
var client = GetClient();
var exists = await client.CollectionExistsAsync(collectionName);
if (!exists) return new KnowledgeCollectionInfo();
var info = await client.GetCollectionInfoAsync(collectionName);
2024-08-06 21:02:02 +00:00
return new KnowledgeCollectionInfo
{
DataCount = info.PointsCount,
VectorCount = info.VectorsCount
};
}
2024-08-06 22:43:11 +00:00
public async Task<UuidPagedItems<KnowledgeCollectionData>> GetCollectionData(KnowledgeFilter filter)
{
var client = GetClient();
var exists = await client.CollectionExistsAsync(filter.CollectionName);
if (!exists)
{
return new UuidPagedItems<KnowledgeCollectionData>();
}
var totalPointCount = await client.CountAsync(filter.CollectionName);
var response = await client.ScrollAsync(filter.CollectionName, limit: (uint)filter.Size,
offset: !string.IsNullOrWhiteSpace(filter.StartId) ? new PointId { Uuid = filter.StartId } : 0,
vectorsSelector: filter.WithVector);
var points = response?.Result?.Select(x => new KnowledgeCollectionData
{
Id = x.Id?.Uuid ?? string.Empty,
Text = x.Payload.ContainsKey(KnowledgePayloadName.Text) ? x.Payload[KnowledgePayloadName.Text].StringValue : string.Empty,
Answer = x.Payload.ContainsKey(KnowledgePayloadName.Answer) ? x.Payload[KnowledgePayloadName.Answer].StringValue : string.Empty,
Vector = filter.WithVector ? x.Vectors?.Vector?.Data?.ToArray() : null
})?.ToList() ?? new List<KnowledgeCollectionData>();
return new UuidPagedItems<KnowledgeCollectionData>
{
Count = totalPointCount,
NextId = response?.NextPageOffset?.Uuid,
Items = points
};
}
2023-06-18 18:15:00 +00:00
public async Task CreateCollection(string collectionName, int dim)
2023-06-18 02:56:22 +00:00
{
var collections = await GetCollections();
if (!collections.Contains(collectionName))
{
// Create a new collection
2024-01-06 03:24:13 +00:00
await GetClient().CreateCollectionAsync(collectionName, new VectorParams()
2023-11-14 01:25:25 +00:00
{
Size = (ulong)dim,
Distance = Distance.Cosine
});
2023-06-18 02:56:22 +00:00
}
// Get collection info
2023-11-14 01:25:25 +00:00
var collectionInfo = await _client.GetCollectionInfoAsync(collectionName);
2023-06-18 18:15:00 +00:00
if (collectionInfo == null)
2023-06-18 02:56:22 +00:00
{
throw new Exception($"Create {collectionName} failed.");
}
}
2024-07-17 21:03:46 +00:00
public async Task<bool> Upsert(string collectionName, string id, float[] vector, string text, Dictionary<string, string>? payload = null)
2023-06-18 02:56:22 +00:00
{
// Insert vectors
2024-07-02 14:52:16 +00:00
var point = new PointStruct()
2023-06-18 02:56:22 +00:00
{
2024-07-02 14:52:16 +00:00
Id = new PointId()
2023-11-14 01:25:25 +00:00
{
2024-07-02 14:52:16 +00:00
Uuid = id
},
Vectors = vector,
2024-07-02 14:56:15 +00:00
2024-08-06 22:43:11 +00:00
Payload =
2024-07-17 21:03:46 +00:00
{
2024-08-06 19:01:26 +00:00
{ KnowledgePayloadName.Text, text }
2024-07-17 21:03:46 +00:00
}
2024-07-02 14:52:16 +00:00
};
2023-06-27 23:36:50 +00:00
2024-07-17 21:03:46 +00:00
if (payload != null)
2024-07-02 14:52:16 +00:00
{
2024-07-17 21:03:46 +00:00
foreach (var item in payload)
{
point.Payload.Add(item.Key, item.Value);
}
2024-07-02 14:52:16 +00:00
}
2024-07-17 21:03:46 +00:00
var client = GetClient();
var result = await client.UpsertAsync(collectionName, points: new List<PointStruct>
2024-07-02 14:52:16 +00:00
{
point
});
2024-07-17 21:03:46 +00:00
return result.Status == UpdateStatus.Completed;
2023-06-18 02:56:22 +00:00
}
2024-07-17 21:03:46 +00:00
public async Task<List<string>> Search(string collectionName, float[] vector, string returnFieldName, int limit = 5, float confidence = 0.5f)
2023-06-18 02:56:22 +00:00
{
2024-07-17 21:03:46 +00:00
var client = GetClient();
var points = await client.SearchAsync(collectionName, vector,
limit: (ulong)limit,
scoreThreshold: confidence);
2023-06-27 23:36:50 +00:00
2024-07-17 21:03:46 +00:00
return points.Select(x => x.Payload[returnFieldName].StringValue).ToList();
2023-06-18 02:56:22 +00:00
}
}