DuckDB ODBC UTF-8 Character Encoding Fix for Excel VBA

DuckDB.bas

Function ConvertUtf8ToUnicode(bytes() As Byte) As String

    Dim ostream As Object
    Set ostream = CreateObject("ADODB.Stream")
    With ostream
        .Type = 1 ' Binary
        .Open
        .Write bytes
        .Position = 0
        .Type = 2 'Text
        .Charset = "UTF-8"
        ConvertUtf8ToUnicode = .ReadText(-1)
        .Close
    End With

End Function

Public Sub Execute(sql As String, range As range)

    Dim connection As Object
    Set connection = CreateObject("ADODB.Connection")
    connection.Open "Driver={DuckDB Driver};Database=:memory:;"

    Dim rs As Object
    Set rs = CreateObject("ADODB.Recordset")
    rs.Open sql, connection

    Dim data As Variant
    data = rs.GetRows()
    Dim rows As Long, cols As Long
    cols = UBound(data, 1)
    rows = UBound(data, 2)

    Dim cells As Variant
    ReDim cells(rows, cols)

    rs.MoveFirst
    Dim row As Long, col As Long, bytes() As Byte
    For row = 0 To rows
        For col = 0 To cols
            If adVarChar <= rs.Fields(col).Type And rs.Fields(col).Type <= adLongVarBinary And Not IsNull(rs.Fields(col).value) Then
                bytes = data(col, row)
                cells(row, col) = ConvertUtf8ToUnicode(bytes)
            Else
                cells(row, col) = data(col, row)
            End If
        Next col
        rs.MoveNext
    Next row

    range.Resize(rows + 1, cols + 1).value = cells

    rs.Close
    Set rs = Nothing

    connection.Close
    Set connection = Nothing

End Sub

Usage:

DuckDB.Execute "select encode('你好') as Chinese", Range("A1")