ArgumentOutOfRangeException using IndexOf with CultureInfo 1031

Question

string s = "Gewerbegebiet Waldstraße"; //other possible input "Waldstrasse"

int iFoundStart = s.IndexOf("strasse", StringComparison.CurrentCulture);
if (iFoundStart > -1)
    s = s.Remove(iFoundStart, 7);

I'm running CultureInfo 1031 (german).

IndexOf matches 'straße' or 'strasse' with defined 'strasse' and returns 18 as position.

Neither Remove nor Replace got any overload for setting a culture.

If I remove 6 chars using Remove 1 character will be left if input-string is 'strasse' and 'straße' will work. If input-string is 'straße' and I remove 7 chars I get ArgumentOutOfRangeException.

Is there a way to safely remove the found string? Any method which provides the last index of IndexOf? I stepped closer into IndexOf and it's native code under the hood as expected - so no way to do something own...

Insolvency

How about replacing it with empty string? s = s.Replace("strasse",""); – Shrier

@dotctor I believe the OP is saying that string.Replace doesn't take the culture into account, so "ss" doesn't match "ß". – Acadia

Im running on en-US and got this problem.the thing is IndexOf behaves different. – Frostbite

@M.kazemAkhgary Because in English "ß" and "ss" are not the same. – Scurlock

What if you first did s.Replace("ß", "ss");? – Acadia

@JakubLortz they are not same. but even i get the correct index when i use indexOf. if i use StringComparison.Ordinal then it gives me -1 of course but .Net guys should think about adding this overloads for Replace and Remove methods too and they should behave like IndexOf do. – Frostbite

life is too short to learn German! – Shrier

@juharr: Well in my case there may be more then one 'ß' in my input-string not related to 'straße'. I can replace 'straße' with 'strasse', but there may be a 'sTraße' or a 'stRaße' ... you just never know what was inputted. What I wanted to achieve was replace any '*strasse' to '*str.' and any 'Straße' to 'Str.' – Insolvency

Answer 1

The native Win32 API does expose the length of the string that was found. You can use P/Invoke to call FindNLSStringEx directly:

static class CompareInfoExtensions
{
    [DllImport("kernel32.dll", CharSet = CharSet.Unicode, ExactSpelling = true)]
    private static extern int FindNLSStringEx(string lpLocaleName, uint dwFindNLSStringFlags, string lpStringSource, int cchSource, string lpStringValue, int cchValue, out int pcchFound, IntPtr lpVersionInformation, IntPtr lpReserved, int sortHandle);

    const uint FIND_FROMSTART = 0x00400000;

    public static int IndexOfEx(this CompareInfo compareInfo, string source, string value, int startIndex, int count, CompareOptions options, out int length)
    {
        // Argument validation omitted for brevity
        return FindNLSStringEx(compareInfo.Name, FIND_FROMSTART, source, source.Length, value, value.Length, out length, IntPtr.Zero, IntPtr.Zero, 0);
    }
}

static class Program
{
    static void Main()
    {
        var s = "<<Gewerbegebiet Waldstraße>>";
        //var s = "<<Gewerbegebiet Waldstrasse>>";
        int length;
        int start = new CultureInfo("de-DE").CompareInfo.IndexOfEx(s, "strasse", 0, s.Length, CompareOptions.None, out length);
        Console.WriteLine(s.Substring(0, start) + s.Substring(start + length));
    }
}

I'm not seeing a way to do this using purely the BCL.

Wye

If I want to match 'Berliner Straße' and use CompareOptions.IgnoreCase this fails - any idea why? – Insolvency

@Insolvency I left out the conversion from CompareOptions to FindNLSStringEx option values: you can see that the options parameter does not get used. You will need to add a conversion from CompareOptions.IgnoreCase to either LINGUISTIC_IGNORECASE or NORM_IGNORECASE (to be determined by you). – Wye

This seems to work. I defined another flag LINGUISTIC_IGNORECASE and combined it with FIND_FROMSTART using |. Thank you, too bad that there is no high level approach doing this. – Insolvency