Why does FileStream.Position increment in multiples of 1024?

Question

I have a text file that I want to read line by line and record the position in the text file as I go. After reading any line of the file the program can exit, and I need to resume reading the file at the next line when it resumes.

Here is some sample code:

using (FileStream fileStream = new FileStream("Sample.txt", FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
{
    fileStream.Seek(GetLastPositionInFile(), SeekOrigin.Begin);
    using (StreamReader streamReader = new StreamReader(fileStream))
    {
        while (!streamReader.EndOfStream)
        {
            string line = streamReader.ReadLine();
            DoSomethingInteresting(line);
            SaveLastPositionInFile(fileStream.Position);

            if (CheckSomeCondition())
            {
                break;
            }
        }
    }
}

When I run this code, the value of fileStream.Position does not change after reading each line, it only advances after reading a couple of lines. When it does change, it increases in multiples of 1024. Now I assume that there is some buffering going on under the covers, but how can I record the exact position in the file?

Answer 1

It's not FileStream that's responsible - it's StreamReader. It's reading 1K at a time for efficiency.

Keeping track of the effective position of the stream as far as the StreamReader is concerned is tricky... particularly as ReadLine will discard the line ending, so you can't accurately reconstruct the original data (it could have ended with "\n" or "\r\n"). It would be nice if StreamReader exposed something to make this easier (I'm pretty sure it could do so without too much difficulty) but I don't think there's anything in the current API to help you :(

By the way, I would suggest that instead of using EndOfStream, you keep reading until ReadLine returns null. It just feels simpler to me:

string line;
while ((line = reader.ReadLine()) != null)
{
    // Process the line
}