Random Access with StreamReader - Seeking and File Position with ReadLine

I spent some time researching how to have dynamic access with StreamReader while using ReadLine and the solution turned out to more complicated that I would expect for .NET at this stage of its development.

My problem is that I have a text file I am reading with a StreamReader object and I am using ReadLine to get one line of text at a time. Nothing special here, but periodically I need to go back to a previous position in the file and read that part again.

If you use the BaseStream property of the StreamReader Object, then you can use the Position property and Seek method to get and set the current position in the file. These two items would appear to be all that you need to get the random access you need; however, when you use Position to get the offset of the file after the ReadLine, you don’t get the position of the file after the last line read, you get the position of the end of the buffer (usually 1024 bytes in size). So, when you use Seek to go back to that position, you will unlikely get back to the position you want unless the buffer boundary just happens to work out right.

I searched around for a simple solution to this problem (there may be and I haven’t found it yet). There were lots of posting about using FileStream, but it doesn't have ReadLine to read a line of text, so they were suggesting that you implement your own version of ReadLine for FileStream. I also found information on DiscardBufferedData that can be used with StreamReader, but it doesn’t help you get the correct offset after using ReadLine.
There were several suggestions on writing your own version of StreamReader:

http://bytes.com/forum/thread238508.html

http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=6720&SiteID=1

I finally bit the bullet and created my own class to accomplish what I need (and only what I need). The previous two posts didn’t provide me with a working solution, so I am going to post what I have come up with. It may need some more refinement, but it appears to work okay. This is in VB.NET, but should be easily translatable into C# if necessary.

The first code snippet is an example of using the FileClass to read a file. In this example, I am looking for the string “*****StartOfData******”. When this string is found, I get the position and then call my PreprocessData function to read the rest of the data. I then go back to the position and run ProcessData from the same point:

Dim list() As String
Dim sep() As Char = {""c}

Try
     Dim s As New FileClass
     s.Open(TextBoxDataFile.Text)

     Dim buffer As String = ""

     Do

        If Not s.GetNextLine(buffer) Then
           Exit Do
        End If

        list = buffer.Split(sep)

        If buffer = "*****StartOfData*****" Then

        Dim startOfData As Integer = s.GetCurrentOffset()

        PreprocessData(s)

        s.SetCurrentOffset(startOfData)

        ProcessData(s)

    End If

Loop Until s.EOF()

    s.Close()

Catch ex As Exception

    Return False

End Try

The FileClass is shown below:

Imports System.IO
Imports System.text

Public Class FileClass

    Const BUFFER_SIZE As Integer = 1024

    Private g_file As StreamReader = Nothing
    Private g_line As Integer = 0
    Private g_position As Integer = 0
    Private g_buffer(BUFFER_SIZE) As Char
    Private g_bufferSize As Integer = 0
    Private g_offset As Integer = 0
    Private g_eofFlag As Boolean = True
    Private g_lineBuffer As New StringBuilder(BUFFER_SIZE)
    Private g_bufferOffset As Integer = 0

    Public Function Open(ByVal filename As String) As Boolean

        If Not g_file Is Nothing Then Close()

        g_file = New StreamReader(filename)
        g_line = 0
        g_position = 0
        g_eofFlag = False
        g_bufferSize = 0
        g_bufferOffset = 0

        LoadBuffer()

    End Function

    Public Function Close() As Boolean

        g_file.Close()
        g_file = Nothing
        g_line = 0
        g_position = 0
        g_eofFlag = True
        g_bufferSize = 0

        Return True

    End Function

    Public Function GetCurrentOffset() As Integer
        Return g_offset
    End Function

    Public Function SetCurrentOffset(ByVal offset As Integer) As Boolean

        Dim pos As Long = g_file.BaseStream.Seek(offset, SeekOrigin.Begin)

        g_file.DiscardBufferedData()

        LoadBuffer()

        Return offset = pos

    End Function

    Public Function GetNextLine(ByRef data As String) As Boolean

        g_lineBuffer.Length = 0

        Dim ch As Char
        Dim flag As Boolean = False

        While Not flag

            ch = g_buffer(g_position)

            If ch = vbCr Then
                ' do nothing - skip cr
            ElseIf ch = vbLf Then
                flag = True
            Else
                g_lineBuffer.Append(ch)
            End If

            g_position = g_position + 1

            If g_position = g_bufferSize Then
                If Not LoadBuffer() Then
                    Exit While
                End If
            End If

        End While

        If flag Then
            g_offset = g_bufferOffset + g_position
            data = g_lineBuffer.ToString
            Return True
        End If

        Return False

    End Function

    Private Function LoadBuffer() As Boolean

        g_bufferOffset = Convert.ToInt32(g_file.BaseStream.Position)
        g_position = 0
        g_bufferSize = g_file.Read(g_buffer, 0, BUFFER_SIZE)

        If g_bufferSize = 0 Then
            g_eofFlag = True
            Return False
        End If

        Return True

    End Function

    Public Function EOF() As Boolean
        Return g_eofFlag
    End Function

End Class

The FileClass is pretty simple, it just has to fill a buffer and look for the carriage returns and linefeeds itself to make generate the line it reads.