Skip to content
Merged
Show file tree
Hide file tree
Changes from 26 commits
Commits
Show all changes
30 commits
Select commit Hold shift + click to select a range
b6bf0d5
Add BulkUpsert implementation
LiamHamsters Jul 23, 2025
261d342
hot fix
LiamHamsters Jul 23, 2025
8173dd0
hot fix 2
LiamHamsters Jul 23, 2025
8d170b8
fix issues
LiamHamsters Jul 24, 2025
5ac5127
feat: make BulkUpsert API more user-friendly
LiamHamsters Jul 24, 2025
3db706d
Implement simple BulkUpsert in YdbConnection and YdbDataSource (delet…
LiamHamsters Jul 25, 2025
a2aa86b
refactor: use array-based rows for BulkUpsert, remove dictionary
LiamHamsters Jul 25, 2025
bff5137
fix: format
LiamHamsters Jul 25, 2025
924da51
last hot fix
LiamHamsters Jul 25, 2025
5e52cc1
fix issues
LiamHamsters Jul 25, 2025
3c219ef
rebase + SessionImpl -> Session
LiamHamsters Jul 25, 2025
4af5c80
fast fix
LiamHamsters Jul 25, 2025
18060cd
BulkUpsert: forbid usage inside transaction, add test
LiamHamsters Jul 25, 2025
f207b69
fix + edit CHANGELOG.md
LiamHamsters Jul 25, 2025
816dc9c
add BulkUpsertProtoImporter
LiamHamsters Jul 29, 2025
7ffa2ee
hot fix
LiamHamsters Jul 29, 2025
1879bca
bulk upsert: add object[] support, simplify implementation
LiamHamsters Jul 29, 2025
14a2bfe
fix: handle exceptions in BulkUpsertImporter DisposeAsync, update tests
LiamHamsters Jul 29, 2025
5cee465
fix
LiamHamsters Jul 30, 2025
b203781
hot fix
LiamHamsters Jul 30, 2025
eec3d12
remove Type argument
LiamHamsters Jul 30, 2025
8c32732
hot fix
LiamHamsters Jul 30, 2025
c9fdbb5
fix inspections and autoformat
LiamHamsters Jul 30, 2025
6463d36
feat: switch BulkUpsertImporter to plain proto, remove extra wrappers…
LiamHamsters Jul 31, 2025
bfc6cf5
refactor: correct batch bytes check and flush logic in BulkUpsertImpo…
LiamHamsters Jul 31, 2025
1c28862
fix
LiamHamsters Jul 31, 2025
0bb791c
last fix
LiamHamsters Jul 31, 2025
beec0e0
Finally
LiamHamsters Jul 31, 2025
b08ebd5
change ct in Flush
LiamHamsters Jul 31, 2025
96fef4a
edit: tablePath calculation in YdbConnection
LiamHamsters Aug 1, 2025
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/Ydb.Sdk/CHANGELOG.md
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@
- ADO.NET: Added `BeginBulkUpsertImport` for batch upsert operations with transaction checks and integration tests.
- Optimization: On BadSession, do not invoke the `DeleteSession()` method.
- Canceling AttachStream after calling the `DeleteSession` method.
- Fixed bug: fixed issue where session was not deleted (`ClientTransportTimeout`).
Expand Down
98 changes: 98 additions & 0 deletions src/Ydb.Sdk/src/Ado/BulkUpsert/BulkUpsertImporter.cs
Original file line number Diff line number Diff line change
@@ -0,0 +1,98 @@
using Google.Protobuf.Collections;
using Ydb.Sdk.Ado.Internal;
using Ydb.Sdk.Value;
using Ydb.Table;
using Ydb.Table.V1;

namespace Ydb.Sdk.Ado.BulkUpsert;

public sealed class BulkUpsertImporter : IBulkUpsertImporter
{
private readonly IDriver _driver;
private readonly string _tablePath;
private readonly IReadOnlyList<string> _columns;
private readonly int _maxBytes;
private readonly RepeatedField<Ydb.Value> _rows = new();
private readonly CancellationToken _cancellationToken;
private StructType? _structType;
private int _currentBytes;

public BulkUpsertImporter(
IDriver driver,
string tableName,
IReadOnlyList<string> columns,
CancellationToken cancellationToken = default,
int maxBytes = 64 * 1024 * 1024)
{
_driver = driver;
_tablePath = tableName;
_columns = columns;
_maxBytes = maxBytes;
_cancellationToken = cancellationToken;
_maxBytes = maxBytes / 2;
}

public async ValueTask AddRowAsync(object?[] values)
{
if (values.Length != _columns.Count)
throw new ArgumentException("Values count must match columns count", nameof(values));

var ydbValues = values.Select(v =>
v as YdbValue ?? (v is YdbParameter param ? param.YdbValue : new YdbParameter { Value = v }.YdbValue)
).ToArray();

var protoStruct = new Ydb.Value();
foreach (var value in ydbValues)
protoStruct.Items.Add(value.GetProto().Value);

var rowSize = protoStruct.CalculateSize();

if (_currentBytes + rowSize > _maxBytes && _rows.Count > 0)
{
await FlushAsync(_cancellationToken);
}

_rows.Add(protoStruct);
_currentBytes += rowSize;

_structType ??= new StructType
{
Members =
{
_columns.Select((col, i) => new StructMember
{
Name = col,
Type = ydbValues[i].GetProto().Type
})
}
};
}

public async ValueTask FlushAsync(CancellationToken cancellationToken = default)
{
if (_rows.Count == 0) return;
if (_structType == null)
throw new InvalidOperationException("structType is undefined");

var typedValue = new TypedValue
{
Type = new Type { ListType = new ListType { Item = new Type { StructType = _structType } } },
Value = new Ydb.Value { Items = { _rows } }
};

var req = new BulkUpsertRequest { Table = _tablePath, Rows = typedValue };

var resp = await _driver.UnaryCall(
TableService.BulkUpsertMethod,
req,
new GrpcRequestSettings { CancellationToken = cancellationToken }
).ConfigureAwait(false);

var operation = resp.Operation;
if (operation.Status.IsNotSuccess())
throw YdbException.FromServer(operation.Status, operation.Issues);

_rows.Clear();
_currentBytes = 0;
}
}
7 changes: 7 additions & 0 deletions src/Ydb.Sdk/src/Ado/BulkUpsert/IBulkUpsertImporter.cs
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
namespace Ydb.Sdk.Ado.BulkUpsert;

public interface IBulkUpsertImporter
{
ValueTask AddRowAsync(object?[] rows);
ValueTask FlushAsync(CancellationToken cancellationToken = default);
}
17 changes: 17 additions & 0 deletions src/Ydb.Sdk/src/Ado/YdbConnection.cs
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
using System.Data;
using System.Data.Common;
using System.Diagnostics.CodeAnalysis;
using Ydb.Sdk.Ado.BulkUpsert;
using Ydb.Sdk.Ado.Session;
using Ydb.Sdk.Services.Query;
using static System.Data.IsolationLevel;
Expand Down Expand Up @@ -52,6 +53,22 @@ public YdbConnection(YdbConnectionStringBuilder connectionStringBuilder)
ConnectionStringBuilder = connectionStringBuilder;
}

public IBulkUpsertImporter BeginBulkUpsertImport(
string name,
IReadOnlyList<string> columns,
CancellationToken cancellationToken = default
)
{
ThrowIfConnectionClosed();
if (CurrentTransaction is { Completed: false })
throw new InvalidOperationException("BulkUpsert cannot be used inside an active transaction.");

var database = ConnectionStringBuilder.Database.TrimEnd('/');
var tablePath = string.IsNullOrEmpty(database) ? name : $"{database}/{name}";

return new BulkUpsertImporter(Session.Driver, tablePath, columns, cancellationToken);
}

protected override YdbTransaction BeginDbTransaction(IsolationLevel isolationLevel)
{
ThrowIfConnectionClosed();
Expand Down
183 changes: 183 additions & 0 deletions src/Ydb.Sdk/test/Ydb.Sdk.Ado.Tests/YdbConnectionTests.cs
Original file line number Diff line number Diff line change
Expand Up @@ -308,4 +308,187 @@ private List<Task> GenerateTasks() => Enumerable.Range(0, 100).Select(async i =>

protected override async Task OnDisposeAsync() =>
await YdbConnection.ClearPool(new YdbConnection(_connectionStringTls));

[Fact]
public async Task BulkUpsertImporter_HappyPath_Add_Flush()
{
var tableName = $"BulkImporter_{Guid.NewGuid():N}";

var conn = new YdbConnection(_connectionStringTls);
await conn.OpenAsync();
try
{
await using (var createCmd = conn.CreateCommand())
{
createCmd.CommandText = $@"
CREATE TABLE {tableName} (
Id Int32,
Name Utf8,
PRIMARY KEY (Id)
)";
await createCmd.ExecuteNonQueryAsync();
}

var columns = new[] { "Id", "Name" };

var importer = conn.BeginBulkUpsertImport(tableName, columns);

await importer.AddRowAsync([YdbValue.MakeInt32(1), YdbValue.MakeUtf8("Alice")]);
await importer.AddRowAsync([YdbValue.MakeInt32(2), YdbValue.MakeUtf8("Bob")]);
await importer.FlushAsync();

await using (var checkCmd = conn.CreateCommand())
{
checkCmd.CommandText = $"SELECT COUNT(*) FROM {tableName}";
var count = Convert.ToInt32(await checkCmd.ExecuteScalarAsync());
Assert.Equal(2, count);
}

importer = conn.BeginBulkUpsertImport(tableName, columns);
await importer.AddRowAsync([YdbValue.MakeInt32(3), YdbValue.MakeUtf8("Charlie")]);
await importer.AddRowAsync([YdbValue.MakeInt32(4), YdbValue.MakeUtf8("Diana")]);
await importer.FlushAsync();

await using (var checkCmd = conn.CreateCommand())
{
checkCmd.CommandText = $"SELECT Name FROM {tableName} ORDER BY Id";
var names = new List<string>();
await using var reader = await checkCmd.ExecuteReaderAsync();
while (await reader.ReadAsync())
names.Add(reader.GetString(0));
Assert.Contains("Alice", names);
Assert.Contains("Bob", names);
Assert.Contains("Charlie", names);
Assert.Contains("Diana", names);
}
}
finally
{
await using var dropCmd = conn.CreateCommand();
dropCmd.CommandText = $"DROP TABLE {tableName}";
await dropCmd.ExecuteNonQueryAsync();
}
}

[Fact]
public async Task BulkUpsertImporter_ThrowsOnInvalidRowCount()
{
var tableName = $"BulkImporter_{Guid.NewGuid():N}";
var conn = new YdbConnection(_connectionStringTls);
await conn.OpenAsync();
try
{
await using (var createCmd = conn.CreateCommand())
{
createCmd.CommandText = $@"
CREATE TABLE {tableName} (
Id Int32,
Name Utf8,
PRIMARY KEY (Id)
)";
await createCmd.ExecuteNonQueryAsync();
}

var columns = new[] { "Id", "Name" };

var importer = conn.BeginBulkUpsertImport(tableName, columns);

var badRow = new object?[] { YdbValue.MakeInt32(1) };
await Assert.ThrowsAsync<ArgumentException>(async () => await importer.AddRowAsync([badRow]));

await Assert.ThrowsAsync<ArgumentException>(async () =>
{
await importer.AddRowAsync([
new object?[] { YdbValue.MakeInt32(2) }
]);
});
}
finally
{
await using var dropCmd = conn.CreateCommand();
dropCmd.CommandText = $"DROP TABLE {tableName}";
await dropCmd.ExecuteNonQueryAsync();
}
}

[Fact]
public async Task BulkUpsertImporter_MultipleImporters_Parallel()
{
var table1 = $"BulkImporter_{Guid.NewGuid():N}_1";
var table2 = $"BulkImporter_{Guid.NewGuid():N}_2";

var conn = new YdbConnection(_connectionStringTls);
await conn.OpenAsync();
try
{
foreach (var table in new[] { table1, table2 })
{
await using var createCmd = conn.CreateCommand();
createCmd.CommandText = $@"CREATE TABLE {table} (
Id Int32,
Name Utf8,
PRIMARY KEY (Id)
)";
await createCmd.ExecuteNonQueryAsync();
}

var columns = new[] { "Id", "Name" };

await Task.WhenAll(
Task.Run(async () =>
{
var importer = conn.BeginBulkUpsertImport(table1, columns);
var rows = Enumerable.Range(0, 20)
.Select(i => new object?[] { YdbValue.MakeInt32(i), YdbValue.MakeUtf8($"A{i}") })
.ToArray();
foreach (var row in rows)
await importer.AddRowAsync(row);
await importer.FlushAsync();
}),
Task.Run(async () =>
{
var importer = conn.BeginBulkUpsertImport(table2, columns);
var rows = Enumerable.Range(0, 20)
.Select(i => new object?[] { YdbValue.MakeInt32(i), YdbValue.MakeUtf8($"B{i}") })
.ToArray();
foreach (var row in rows)
await importer.AddRowAsync(row);
await importer.FlushAsync();
})
);

foreach (var table in new[] { table1, table2 })
{
await using var checkCmd = conn.CreateCommand();
checkCmd.CommandText = $"SELECT COUNT(*) FROM {table}";
var count = Convert.ToInt32(await checkCmd.ExecuteScalarAsync());
Assert.Equal(20, count);
}
}
finally
{
foreach (var table in new[] { table1, table2 })
{
await using var dropCmd = conn.CreateCommand();
dropCmd.CommandText = $"DROP TABLE {table}";
await dropCmd.ExecuteNonQueryAsync();
}
}
}

[Fact]
public async Task BulkUpsertImporter_ThrowsOnNonexistentTable()
{
var tableName = $"Nonexistent_{Guid.NewGuid():N}";
var conn = new YdbConnection(_connectionStringTls);
await conn.OpenAsync();

var columns = new[] { "Id", "Name" };

var importer = conn.BeginBulkUpsertImport(tableName, columns);

await importer.AddRowAsync([YdbValue.MakeInt32(1), YdbValue.MakeUtf8("NotExists")]);

await Assert.ThrowsAsync<YdbException>(async () => { await importer.FlushAsync(); });
}
}
Loading