Download in-memory database as duck or sqlite file

[andy-wrks]

What happens?

I'm using DuckDB Wasm 1.29.0 through JavaScript in the browser (Chrome).
I create some tables in the "memory" database and populate them with data from various sources. Now I'd like to download the in-memory db to a single file (i.e. preferrably not as single csv, json or parquet tables).
There seems to be the possibility to download the db as a duck or sqlite file, however I struggle to succeed with this.

Here's what I tried for sqlite:

async exportSQLiteFile() {

    const fileName = "my.db";
    const aliasName = "mydb";

    await this.#db.registerEmptyFileBuffer(fileName);

    await this.#conn.query(`ATTACH '${fileName}' AS ${aliasName} (READ_WRITE, TYPE SQLITE);`);
    // Copy the in-memory database to the new database file
    await this.#conn.query(`COPY FROM DATABASE memory TO ${aliasName};`);
    await this.#conn.query(`DETACH ${aliasName};`);

    const buffer = await this.#db.copyFileToBuffer(fileName);
    fileIo.downloadFile(buffer, fileName);

    await this.#db.dropFile(fileName);
}

The COPY process seems to work. When I change to mydb, the tables and data from the memory db are there. However, the downloaded file "my.db" is 1 byte in size...

I then tried to download a duckdb file:

async exportDuckFile() {

    const fileName = "my.db";
    const aliasName = "mydb";

    await this.#db.registerEmptyFileBuffer(fileName);

    await this.#conn.query(`ATTACH '${fileName}' AS ${aliasName} (READ_WRITE, TYPE DUCKDB);`);
    // Copy the in-memory database to the new database file
    await this.#conn.query(`COPY FROM DATABASE memory TO ${aliasName};`);
    await this.#conn.query(`DETACH ${aliasName};`);
	
    const buffer = await this.#db.copyFileToBuffer(fileName);
    fileIo.downloadFile(buffer, fileName);
	
    await this.#db.dropFile(fileName);
}

This results in an "Uncaught (in promise) Error" when using copyFileToBuffer.

The only thing that has worked for me so far, is exporting the tables to single parquet-files like this:

async exportParquetFiles(dbName) {

    // Note: This is a custom function for getting the names of existing tables.
    // Not to be confused with conn.getTableNames from the API, which does something else.
    const tableNames = await this.getTableNames(dbName);

    for (const tableName of tableNames) {
        const fileName = `${tableName}.parquet`;
        await this.#db.registerEmptyFileBuffer(fileName);
        await this.#conn.query(`COPY (SELECT * FROM ${tableName}) TO '${fileName}' (FORMAT PARQUET);`);
        const buffer = await this.#db.copyFileToBuffer(fileName);
        fileIo.downloadFile(buffer, fileName);
        await this.#db.dropFile(fileName);
    }
}

This still trows a warning for every exported file: "Buffering missing file: *.parquet"
I'm not sure if this is a problem or what needs to be done in order to not get this warning.

What am I missing in the first 2 examples, how is it possible to export the in.memory db to a duckdb or sqlite file?

To Reproduce

See code examples and description above.

Browser/Environment:

Chrome 134.0.6998.178

Device:

Win11 Laptop, i7-8850-H CPU

DuckDB-Wasm Version:

1.29.0

DuckDB-Wasm Deployment:

locally via VSCode Live Server

Full Name:

Andy

Affiliation:

none