Windows Service non-Unicode Encoding Problem

Question

Hi,

I'm working on a simple windows service which reads csv data and transfer the data to ms-sql server. Csv contains non-unicode chars(Ç窺İıÖöÜüĞğ). ODBC doesn't transfer the right endocing. I try the copy data without ODBC, with text reading, encoding doesn't right too. But text encoding is right when I debug the service. Csv read code:

string[] csvLines = File.ReadAllLines(csvFile, Encoding.GetEncoding(new System.Globalization.CultureInfo("tr-TR").TextInfo.ANSICodePage));

Service is running as LocalSystem and I can't change "System Locale" for non-Unicode programs, I have to use system locale as United States(English)

Answer 1

From memory, creating a new culture in the way you're doing it will use system defaults (i.e. "en-US" in your case).

So, rather than creating a new CultureInfo use the pre-cached one:

CultureInfo.GetCultureInfo("tr-TR")

It works when you debug because the code is running as you, not LocalSystem, and I assume your locale is Turkish.

---

Edit: Oops, should have been GetCultureInfo instead of GetCulture.

This works on my machine in a console app:

Console.WriteLine("en-US: {0}",
        CultureInfo.GetCultureInfo("en-US").TextInfo.ANSICodePage);
Console.WriteLine("tr-TR: {0}",
        CultureInfo.GetCultureInfo("tr-TR").TextInfo.ANSICodePage);

Outputs 1252 and 1254.

Answer 2

The file reading code looks fine and I believe the problem is in how you call the ODBC function. Looks like the encoding information is lost somewhere in between the calls.

Can you post a code snippet showing your ODBC calls? Thanks