|
| 1 | +// Copyright (c) Microsoft. All rights reserved. |
| 2 | + |
| 3 | +using System.Collections.Generic; |
| 4 | +using System.Linq; |
| 5 | +using System.Runtime.CompilerServices; |
| 6 | +using System.Threading; |
| 7 | +using System.Threading.Tasks; |
| 8 | +using DuckDB.NET.Data; |
| 9 | + |
| 10 | +namespace Microsoft.SemanticKernel.Connectors.Memory.DuckDB; |
| 11 | + |
| 12 | +internal struct DatabaseEntry |
| 13 | +{ |
| 14 | + public string Key { get; set; } |
| 15 | + |
| 16 | + public string MetadataString { get; set; } |
| 17 | + |
| 18 | + public string EmbeddingString { get; set; } |
| 19 | + |
| 20 | + public string? Timestamp { get; set; } |
| 21 | +} |
| 22 | + |
| 23 | +internal sealed class Database |
| 24 | +{ |
| 25 | + private const string TableName = "SKMemoryTable"; |
| 26 | + |
| 27 | + public Database() { } |
| 28 | + |
| 29 | + public Task CreateTableAsync(DuckDBConnection conn, CancellationToken cancellationToken = default) |
| 30 | + { |
| 31 | + using var cmd = conn.CreateCommand(); |
| 32 | + cmd.CommandText = $@" |
| 33 | + CREATE TABLE IF NOT EXISTS {TableName}( |
| 34 | + collection TEXT, |
| 35 | + key TEXT, |
| 36 | + metadata TEXT, |
| 37 | + embedding TEXT, |
| 38 | + timestamp TEXT, |
| 39 | + PRIMARY KEY(collection, key))"; |
| 40 | + return cmd.ExecuteNonQueryAsync(cancellationToken); |
| 41 | + } |
| 42 | + |
| 43 | + public async Task CreateCollectionAsync(DuckDBConnection conn, string collectionName, CancellationToken cancellationToken = default) |
| 44 | + { |
| 45 | + if (await this.DoesCollectionExistsAsync(conn, collectionName, cancellationToken).ConfigureAwait(false)) |
| 46 | + { |
| 47 | + // Collection already exists |
| 48 | + return; |
| 49 | + } |
| 50 | + |
| 51 | + using var cmd = conn.CreateCommand(); |
| 52 | + cmd.CommandText = $@" |
| 53 | + INSERT INTO {TableName} VALUES (?1,?2,?3,?4,?5 ); "; |
| 54 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 55 | + cmd.Parameters.Add(new DuckDBParameter(string.Empty)); |
| 56 | + cmd.Parameters.Add(new DuckDBParameter(string.Empty)); |
| 57 | + cmd.Parameters.Add(new DuckDBParameter(string.Empty)); |
| 58 | + cmd.Parameters.Add(new DuckDBParameter(string.Empty)); |
| 59 | + |
| 60 | + await cmd.ExecuteNonQueryAsync(cancellationToken).ConfigureAwait(false); |
| 61 | + } |
| 62 | + |
| 63 | + public async Task UpdateOrInsertAsync(DuckDBConnection conn, |
| 64 | + string collection, string key, string? metadata, string? embedding, string? timestamp, CancellationToken cancellationToken = default) |
| 65 | + { |
| 66 | + using var cmd = conn.CreateCommand(); |
| 67 | + cmd.CommandText = $@" |
| 68 | + INSERT INTO {TableName} VALUES(?1, ?2, ?3, ?4, ?5) |
| 69 | + ON CONFLICT (collection, key) DO UPDATE SET metadata=?3, embedding=?4, timestamp=?5; "; |
| 70 | + cmd.Parameters.Add(new DuckDBParameter(collection)); |
| 71 | + cmd.Parameters.Add(new DuckDBParameter(key)); |
| 72 | + cmd.Parameters.Add(new DuckDBParameter(metadata ?? string.Empty)); |
| 73 | + cmd.Parameters.Add(new DuckDBParameter(embedding ?? string.Empty)); |
| 74 | + cmd.Parameters.Add(new DuckDBParameter(timestamp ?? string.Empty)); |
| 75 | + await cmd.ExecuteNonQueryAsync(cancellationToken).ConfigureAwait(false); |
| 76 | + } |
| 77 | + |
| 78 | + public async Task<bool> DoesCollectionExistsAsync(DuckDBConnection conn, |
| 79 | + string collectionName, |
| 80 | + CancellationToken cancellationToken = default) |
| 81 | + { |
| 82 | + var collections = await this.GetCollectionsAsync(conn, cancellationToken).ToListAsync(cancellationToken).ConfigureAwait(false); |
| 83 | + return collections.Contains(collectionName); |
| 84 | + } |
| 85 | + |
| 86 | + public async IAsyncEnumerable<string> GetCollectionsAsync(DuckDBConnection conn, |
| 87 | + [EnumeratorCancellation] CancellationToken cancellationToken = default) |
| 88 | + { |
| 89 | + using var cmd = conn.CreateCommand(); |
| 90 | + cmd.CommandText = $@" |
| 91 | + SELECT DISTINCT collection |
| 92 | + FROM {TableName};"; |
| 93 | + |
| 94 | + using var dataReader = await cmd.ExecuteReaderAsync(cancellationToken).ConfigureAwait(false); |
| 95 | + while (await dataReader.ReadAsync(cancellationToken).ConfigureAwait(false)) |
| 96 | + { |
| 97 | + yield return dataReader.GetString("collection"); |
| 98 | + } |
| 99 | + } |
| 100 | + |
| 101 | + public async IAsyncEnumerable<DatabaseEntry> ReadAllAsync(DuckDBConnection conn, |
| 102 | + string collectionName, |
| 103 | + [EnumeratorCancellation] CancellationToken cancellationToken = default) |
| 104 | + { |
| 105 | + using var cmd = conn.CreateCommand(); |
| 106 | + cmd.CommandText = $@" |
| 107 | + SELECT * FROM {TableName} |
| 108 | + WHERE collection=?1;"; |
| 109 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 110 | + |
| 111 | + using var dataReader = await cmd.ExecuteReaderAsync(cancellationToken).ConfigureAwait(false); |
| 112 | + while (await dataReader.ReadAsync(cancellationToken).ConfigureAwait(false)) |
| 113 | + { |
| 114 | + string key = dataReader.GetString("key"); |
| 115 | + if (string.IsNullOrWhiteSpace(key)) |
| 116 | + { |
| 117 | + continue; |
| 118 | + } |
| 119 | + string metadata = dataReader.GetString("metadata"); |
| 120 | + string embedding = dataReader.GetString("embedding"); |
| 121 | + string timestamp = dataReader.GetString("timestamp"); |
| 122 | + yield return new DatabaseEntry() { Key = key, MetadataString = metadata, EmbeddingString = embedding, Timestamp = timestamp }; |
| 123 | + } |
| 124 | + } |
| 125 | + |
| 126 | + public async Task<DatabaseEntry?> ReadAsync(DuckDBConnection conn, |
| 127 | + string collectionName, |
| 128 | + string key, |
| 129 | + CancellationToken cancellationToken = default) |
| 130 | + { |
| 131 | + using var cmd = conn.CreateCommand(); |
| 132 | + cmd.CommandText = $@" |
| 133 | + SELECT * FROM {TableName} |
| 134 | + WHERE collection=?1 |
| 135 | + AND key=?2; "; |
| 136 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 137 | + cmd.Parameters.Add(new DuckDBParameter(key)); |
| 138 | + |
| 139 | + using var dataReader = await cmd.ExecuteReaderAsync(cancellationToken).ConfigureAwait(false); |
| 140 | + if (await dataReader.ReadAsync(cancellationToken).ConfigureAwait(false)) |
| 141 | + { |
| 142 | + string metadata = dataReader.GetString(dataReader.GetOrdinal("metadata")); |
| 143 | + string embedding = dataReader.GetString(dataReader.GetOrdinal("embedding")); |
| 144 | + string timestamp = dataReader.GetString(dataReader.GetOrdinal("timestamp")); |
| 145 | + return new DatabaseEntry() |
| 146 | + { |
| 147 | + Key = key, |
| 148 | + MetadataString = metadata, |
| 149 | + EmbeddingString = embedding, |
| 150 | + Timestamp = timestamp |
| 151 | + }; |
| 152 | + } |
| 153 | + |
| 154 | + return null; |
| 155 | + } |
| 156 | + |
| 157 | + public Task DeleteCollectionAsync(DuckDBConnection conn, string collectionName, CancellationToken cancellationToken = default) |
| 158 | + { |
| 159 | + using var cmd = conn.CreateCommand(); |
| 160 | + cmd.CommandText = $@" |
| 161 | + DELETE FROM {TableName} |
| 162 | + WHERE collection=?;"; |
| 163 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 164 | + return cmd.ExecuteNonQueryAsync(cancellationToken); |
| 165 | + } |
| 166 | + |
| 167 | + public Task DeleteAsync(DuckDBConnection conn, string collectionName, string key, CancellationToken cancellationToken = default) |
| 168 | + { |
| 169 | + using var cmd = conn.CreateCommand(); |
| 170 | + cmd.CommandText = $@" |
| 171 | + DELETE FROM {TableName} |
| 172 | + WHERE collection=?1 |
| 173 | + AND key=?2; "; |
| 174 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 175 | + cmd.Parameters.Add(new DuckDBParameter(key)); |
| 176 | + return cmd.ExecuteNonQueryAsync(cancellationToken); |
| 177 | + } |
| 178 | + |
| 179 | + public Task DeleteEmptyAsync(DuckDBConnection conn, string collectionName, CancellationToken cancellationToken = default) |
| 180 | + { |
| 181 | + using var cmd = conn.CreateCommand(); |
| 182 | + cmd.CommandText = $@" |
| 183 | + DELETE FROM {TableName} |
| 184 | + WHERE collection=?1 |
| 185 | + AND key IS NULL"; |
| 186 | + cmd.Parameters.Add(new DuckDBParameter(collectionName)); |
| 187 | + return cmd.ExecuteNonQueryAsync(cancellationToken); |
| 188 | + } |
| 189 | +} |
0 commit comments