Question

I’m looking for a regular expression library in .Net that supports lazy evaluation.

Note: I’m specifically looking for lazy evaluation (i.e., the library, instead of immediately returning all matches in a document, only consumes as much of the document as necessary to determine the next match per request), NOT support for lazy quantifiers – though if it also supports lazy quantifiers, I wouldn’t object!

Specific details: I want to be able to run regexes against very large documents with potentially hundreds of thousands of regex matches, and iterate across the results using IEnumerable<> semantics, without having to take the up-front cost of finding all matches.

Ideally FOSS in C#, but the only requirement is usability from a .Net 3.5 app.

Answer 1

The Match class’ NextMatch method should meet your needs:

Returns a new Match with the results
for the next match, starting at the
position at which the last match ended
(at the character after the last
matched character).

A quick look at it in Reflector confirms this behavior:

public Match NextMatch()
{
    if (this._regex == null)
    {
        return this;
    }
    return this._regex.Run(false, base._length, base._text, this._textbeg,
        this._textend - this._textbeg, this._textpos);
}

Check out the linked MSDN reference for an example of its usage. Briefly, the flow would resemble:

Match m = rx.Match(input);
while (m.Success) 
{
    // do work
    m = m.NextMatch();
}