How do I implement a Peek() function on a DataReader?

Question

There doesn't seem to be a Peek method on the DataReader in ado.net. I would like to be able to perform some one-off processing before I loop through my reader, and it would be nice to be able to look at the data in the first row without causing it to be skipped by the subsequent iteration. What is the best way to accomplish this?

I am using a SqlDataReader, but preferably the implementation would be as general as possible (i.e. apply to IDataReader or DbDataReader).

Answer 1

I would suggest something similar to Jason's solution, but using a wrapper that implements IDataReader instead, so:

sealed public class PeekDataReader : IDataReader
{
    private IDataReader wrappedReader;
    private bool wasPeeked;
    private bool lastResult;

    public PeekDataReader(IDataReader wrappedReader)
    {
        this.wrappedReader = wrappedReader;
    }

    public bool Peek()
    {
        // If the previous operation was a peek, do not move...
        if (this.wasPeeked)
            return this.lastResult;

        // This is the first peek for the current position, so read and tag
        bool result = Read();
        this.wasPeeked = true;
        return result;
    }

    public bool Read()
    {
        // If last operation was a peek, do not actually read
        if (this.wasPeeked)
        {
            this.wasPeeked = false;
            return this.lastResult;
        }

        // Remember the result for any subsequent peeks
        this.lastResult = this.wrappedReader.Read();
        return this.lastResult;
    }

    public bool NextResult()
    {
        this.wasPeeked = false;
        return this.wrappedReader.NextResult();
    }

    // Add pass-through operations for all other IDataReader methods
    // that simply call on 'this.wrappedReader'
}

Note that this does require quite a bit of pass-through code for all the unaffected properties, but the benefit is that it is a generic abstraction that can 'peek' at any position in the result set without moving forward on the subsequent 'read' operation.

To use:

using (IDataReader reader = new PeekDataReader(/* actual reader */))
{
    if (reader.Peek())
    {
        // perform some operations on the first row if it exists...
    }

    while (reader.Read())
    {
        // re-use the first row, and then read the remainder...
    }
}

Note though that every 'Peek()' call will actually move to the next record if the previous operation was not also a 'Peek()'. Keeping this symmetry with the 'Read()' operation provides a simpler implementation and a more elegant API.