C# StreamReader, “ReadLine” For Custom Delimiters

问题

What is the best way to have the functionality of the StreamReader.ReadLine() method, but with custom (String) delimiters?

I'd like to do something like:

String text;
while((text = myStreamReader.ReadUntil("my_delim")) != null)
{
   Console.WriteLine(text);
}

I attempted to make my own using Peek() and StringBuilder, but it's too inefficient. I'm looking for suggestions or possibly an open-source solution.

Thanks.

Edit

I should have clarified this earlier...I have seen this answer, however, I'd prefer not to read the entire file into memory.

回答1:

I figured I would post my own solution. It seems to work pretty well and the code is relatively simple. Feel free to comment.

public static String ReadUntil(this StreamReader sr, String delim)
{
    StringBuilder sb = new StringBuilder();
    bool found = false;

    while (!found && !sr.EndOfStream)
    {
       for (int i = 0; i < delim.Length; i++)
       {
           Char c = (char)sr.Read();
           sb.Append(c);

           if (c != delim[i])
               break;

           if (i == delim.Length - 1)
           {
               sb.Remove(sb.Length - delim.Length, delim.Length);
               found = true;
           }
        }
     }

     return sb.ToString();
}

回答2:

This code should work for any string separator.

public static IEnumerable<string> ReadChunks(this TextReader reader, string chunkSep)
{
    var sb = new StringBuilder();

    var sepbuffer = new Queue<char>(chunkSep.Length);
    var sepArray = chunkSep.ToCharArray();

    while (reader.Peek() >= 0)
    {
        var nextChar = (char)reader.Read();
        if (nextChar == chunkSep[sepbuffer.Count])
        {
            sepbuffer.Enqueue(nextChar);
            if (sepbuffer.Count == chunkSep.Length)
            {
                yield return sb.ToString();
                sb.Length = 0;
                sepbuffer.Clear();
            }
        }
        else
        {
            sepbuffer.Enqueue(nextChar);
            while (sepbuffer.Count > 0)
            {
                sb.Append(sepbuffer.Dequeue());
                if (sepbuffer.SequenceEqual(chunkSep.Take(sepbuffer.Count)))
                    break;
            }
        }
    }
    yield return sb.ToString() + new string(sepbuffer.ToArray());
}

Disclaimer:

I made a little testing on this and is actually slower than ReadLine method, but I suspect it is due to the enqueue/dequeue/sequenceEqual calls that in the ReadLine method can be avoided (because the separator is always \r\n).

Again, I made few tests and it should work, but don't take it as perfect, and feel free to correct it. ;)

回答3:

Here is a simple parser I used where needed (usually if streaming is not a paramount just read and .Split does the job), not too optimized but should work fine:
(it's more of a Split like method - and more notes below)

    public static IEnumerable<string> Split(this Stream stream, string delimiter, StringSplitOptions options)
    {
        var buffer = new char[_bufffer_len];
        StringBuilder output = new StringBuilder();
        int read;
        using (var reader = new StreamReader(stream))
        {
            do
            {
                read = reader.ReadBlock(buffer, 0, buffer.Length);
                output.Append(buffer, 0, read);

                var text = output.ToString();
                int id = 0, total = 0;
                while ((id = text.IndexOf(delimiter, id)) >= 0)
                {
                    var line = text.Substring(total, id - total);
                    id += delimiter.Length;
                    if (options != StringSplitOptions.RemoveEmptyEntries || line != string.Empty)
                        yield return line;
                    total = id;
                }
                output.Remove(0, total);
            }
            while (read == buffer.Length);
        }

        if (options != StringSplitOptions.RemoveEmptyEntries || output.Length > 0)
            yield return output.ToString();
    }

...and you can simply switch to char delimiters if needed just replace the

while ((id = text.IndexOf(delimiter, id)) >= 0)

...with

while ((id = text.IndexOfAny(delimiters, id)) >= 0)

(and id++ instead of id+= and a signature this Stream stream, StringSplitOptions options, params char[] delimiters)

...also removes empty etc.
hope it helps

回答4:

    public static String ReadUntil(this StreamReader streamReader, String delimiter)
    {
        StringBuilder stringBuilder = new StringBuilder();

        while (!streamReader.EndOfStream)
        {
            stringBuilder.Append(value: (Char) streamReader.Read());

            if (stringBuilder.ToString().EndsWith(value: delimiter))
            {
                stringBuilder.Remove(stringBuilder.Length - delimiter.Length, delimiter.Length);
                break;
            }
        }

        return stringBuilder.ToString();
    }

来源：https://stackoverflow.com/questions/9873097/c-sharp-streamreader-readline-for-custom-delimiters