Excel cell-values are truncated by OLEDB-provider

Question

I'm using the OleDbConnection class to retrieve data from an Excel 2000/2003 workbook:

string connectionString = "Provider=Microsoft.Jet.OLEDB.4.0;" +
                          "Data Source=" + filename + ";" +
                          "Extended Properties=\"Excel 8.0;IMEX=1\";";

OleDbConnection connection = new OleDbConnection(connectionString);
connection.Open();

// code to get table name from schema omitted

var dataAdapter = new OleDbDataAdapter(string.Format("SELECT * FROM [{0}]", name),connection);
var myDataSet = new DataSet();
dataAdapter.Fill(myDataSet, "ExcelInfo");

Now it turns out that cells in the worksheet with length greater than 255 characters are being truncated. Is this a limitation in the Microsoft.Jet.OLEDB-provider, or is there something I can do about it?

Anyone?

Answer 1

Try this OleDBAdapter Excel QA I posted via stack overflow.

I populated an worksheet cell (Rows[0][4]) w/ 445 characters and it worked fine... Add this to the end of the code for the ouput

// DataSet:          
Object row0Col3 = ds.Tables["xlsImport"].Rows[0][2];
Object row0Col4 = ds.Tables["xlsImport"].Rows[0][4];

string rowZeroColumn3 = row0Col3.ToString();
string rowZeroColumn4 = row0Col4.ToString();

Console.WriteLine("Row 0, Col 4 string length: {0} " + Environment.NewLine + "Excel content: {1}", rowZeroColumn4.Length, rowZeroColumn4);           

Answer 2

The OLEDB provider for excel will attempt to automatically determine the DataTypes based off of the first 8 rows of data, this can be set with the HDR=Yes/No property in the connection string. Additionally, there are multiple types that it can apply to text columns. The memo type holds over 255 characters, so if none of the first 8 rows have that then it will incorrectly set the data type.

The way to change this is by changing a registry setting called TypeGuessRows, as described here: Microsoft Support

NOTE: The valid range of values for the TypeGuessRows key is 0 to 16. However, if the value is 0, the number of source rows scanned is 16384. So if you have a very large file make sure the biggest rows are first.