SQL UPSERT QUERY W/ Duplicate Rows Made Unique Based Upon 3 Fields (C# VisStudio)

Background is I’m making a SQL connection that takes a .csv file and imports it into a SQL Server database table.

The problem I’m running into is, I’m having trouble with the query syntax because there is no unique identifier for a row in the .csv file I’m importing. It takes 3 fields combined to makes a row unique/distinct.

Rough example of the .csv files’ data, the first three of the .csv columns can be considered together to make unique rows:

Order_Id  Product_Id  Date    Other (etc...)
    1         1a       1/9      q
    1         2a       1/9      q
    1         2a       1/10     e
    2         1a       1/9      e
    2         2a       1/10     e

This is my query syntax simplified in Visual Studios, (I’m importing 25 or so columns in reality from the .csv file, and so to keep things straight/simple I made the columns names the exact same in both the .csv file and the SQL-Server table), but the basic syntax looks like this:

private void SaveImportDataToDatabase(DataTable importData)
{
    using (SqlConnection conn = new SqlConnection("Server=localhost;Database=my_Database;Trusted_Connection=True;"))
    {
        conn.Open();
        foreach (DataRow importRow in importData.Rows)
        {
            
            SqlCommand cmd = new SqlCommand("IF EXISTS(SELECT DISTINCT Order_id, Product_Id, Date FROM Sales WHERE Order_id = @Order_id AND Product_Id = @Product_Id AND Date = @Date) UPDATE SQL_Sales SET Order_id = @Order_id WHERE Order_id = @Order_id ELSE INSERT INTO SQL_Sales (order_id, Product_Id, Date)" +
                                            "VALUES (@order_id, @Product_Id, @Date);", conn);
            
            cmd.Parameters.AddWithValue("@Order_id", importRow["Order_id"]);
            cmd.Parameters.AddWithValue("@Product_Id", importRow["Product_Id"]);
            cmd.Parameters.AddWithValue("@Date", importRow["Date"]);
            
            cmd.ExecuteNonQuery();

        }
    }
}

Once imported I see a few problems in the SQL Server table,

1. The order_id field will be null
2. It’s only importing a very small amount of the data, about 50 of 2000 records
3. If I re-import the data with a change in the .csv file, say with a single new row, I get 100 of 2000 records

I’m not sure if what I’m trying to do is possible or worth it. Should I be breaking this down a lot more instead of doing it all in one query? I’m not necessarily new to coding but I don’t code very often/I’m rusty and this is my first C# project so lend me some patience if possible.

Just wanted to add more code in response to @casey crookston, it’s possible problems 2 and 3 are related to my loop

private void btnImport_Click(object sender, EventArgs e)
    {
       Cursor = Cursors.WaitCursor;            
       DataTable importData = GetDataFromFile();

        if (importData == null) return;
        SaveImportDataToDatabase(importData);

        MessageBox.Show("Import Successful");
        txtFileName.Text = string.Empty;

        Cursor = Cursors.Default;
    }

    private DataTable GetDataFromFile()
    {
        DataTable importedData = new DataTable();
       try
        {
            using (StreamReader sr = new StreamReader(txtFileName.Text))
            {
                string header = sr.ReadLine();
                if (string.IsNullOrEmpty(header))
                {
                    MessageBox.Show("No File Data");
                    return null;
                }

                string[] headerColumns = header.Split(',');
                foreach (string headerColumn in headerColumns)
                {
                    importedData.Columns.Add(headerColumn);
                }

                while (!sr.EndOfStream)
                {
                    string line = sr.ReadLine();

                    if (string.IsNullOrEmpty(line)) continue;

                    string[] fields = line.Split(',');
                    DataRow importedRow = importedData.NewRow();

                    for(int i = 1; i < fields.Count(); i++)
                    {
                        importedRow[i] = fields[i];
                    }

                    importedData.Rows.Add(importedRow);
                }
            }
        }
        catch (Exception e)
        {
            Console.WriteLine("The file could not be read:");
            Console.WriteLine(e.Message);
        }

        return importedData;
    }

Answer

This looks like a good spot to use SQL Server’s MERGE syntax:

merge sales s
using (values(@product_id, @order_id, @date, @other_1, @other_2)) 
    as p(order_id, product_id, date, other_1, other_2)
on (s.product_id = p.product_id and s.order_id = p.order_id and s.date = p.date)
when matched then 
    update set s.other_1 = p.other_1, s.other_2 = p.other_2
when not matched by target then 
    insert(order_id, product_id, date, other_1, other_2)
    values(p.order_id, p.product_id, p.date, p.other_1, p.other_2)

This uses the first 3 columns as primary key; when a tuple already exists, then colums other_1 and other_2 are updated with the values that would have been otherwise inserted.