CSharpDB.ImportExport 4.3.0

CSharpDB.ImportExport

Native table archive import/export support for CSharpDB.

CSharpDB.ImportExport provides the shared table archive format used by the execution engine, client API, admin import/export tools, and DevOps comparison features. It is a low-level package for tools that need to read or write CSharpDB table archives directly.

Features

• native CSharpDB table archive reader and writer
• schema and manifest metadata models
• row streaming through IAsyncEnumerable<DbValue[]>
• primary-key index metadata and indexed integer primary-key lookup
• conversion between archive schema models and CSharpDB.Primitives.TableSchema

The integer primary-key lookup index is an optional acceleration structure. Writers retain at most 65,536 index entries in memory. Archives with more rows are written without the physical lookup index; their schema and rows are still complete, and external-table point lookups use the streaming scan fallback.

Format v5 archives carry required SHA-256 digests for their schema, rows, and optional physical index. Every reader path verifies those section digests before exposing archive data; v3 and v4 remain readable for compatibility.

The Admin restore workflow adds an independent post-load check before a staged table becomes visible. It compares archive rows with the loaded table using the csharpdb-canon-v1 logical encoding and an order-independent, duplicate-preserving 256-partition checksum. Regenerated rowversion values are explicitly excluded. The target scan and final rename share one transaction, and metadata plus rows are read from one immutable private archive snapshot. A mismatch, validation error, or cancellation before the activation boundary leaves no activated table. Once validation passes, the short rename/commit sequence is intentionally non-cancelable and its result is reconciled if commit reporting fails through a durable activation receipt written in the same transaction. The immutable snapshot and memory-bounded checksum spill workspace each have a configurable 4-GiB default safety limit through CSharpDB.Admin.ImportExport.Services.TableArchiveRestoreOptions; exceeding either limit fails the restore safely. This activation contract currently requires a direct/local CSharpDB transport.

Reading Archives

using CSharpDB.ImportExport.TableArchives;

var metadata = await TableArchiveReader.ReadMetadataAsync("customers.cdbtable");
Console.WriteLine($"{metadata.Manifest.SourceTableName}: {metadata.Manifest.RowCount} rows");

await foreach (var row in TableArchiveReader.ReadRowsAsync("customers.cdbtable"))
{
    Console.WriteLine(row[0]);
}

Writing Archives

using CSharpDB.ImportExport.TableArchives;
using CSharpDB.Primitives;

TableSchema schema = new()
{
    TableName = "customers",
    Columns =
    [
        new ColumnDefinition { Name = "id", Type = DbType.Integer, IsPrimaryKey = true },
        new ColumnDefinition { Name = "name", Type = DbType.Text },
    ],
};

DbValue[][] rows =
[
    [DbValue.Integer(1), DbValue.Text("Ada")],
    [DbValue.Integer(2), DbValue.Text("Grace")],
];

await TableArchiveWriter.WriteAsync(
    "customers.cdbtable",
    schema,
    TableArchiveWriter.ToAsyncRows(rows));

Dependencies

• CSharpDB.Primitives - shared type system and schema contracts
• CSharpDB.Storage - record serialization used by native table archives

Related Packages

License

MIT - see LICENSE for details.