Does the regular expression engine skip over strings that are shorter than the pattern?

Question

I want to loop through a set of strings. On each string i want to loop through a set of regular expression to determine which expressions match on the string I\'m on. Howe

Answer 1

When you're after implementation details, and when the source code is available, the best way to tell is to simply look at it. :)

The short answer is: not exactly.

The optimization implemented in the .NET regex implementation is a Boyer-Moore string search as the first phase of matching when possible. Take a look at the source code for the gory details.

From the code itself:

// The RegexBoyerMoore object precomputes the Boyer-Moore
// tables for fast string scanning. These tables allow
// you to scan for the first occurance of a string within
// a large body of text without examining every character.
// The performance of the heuristic depends on the actual
// string and the text being searched, but usually, the longer
// the string that is being searched for, the fewer characters
// need to be examined.

This requires an anchoring prefix, which is searched for by this function, whose comment says:

/*
 * This is the one of the only two functions that should be called from outside.
 * It takes a RegexTree and computes the set of chars that can start it.
 */

The matching algorithm contains code which returns a no match result immediately if the input string is shorter than the computed prefix.

Note that it's also looking for anchors and optimizing for these, of course.

I did not find a minimum length optimization in the code, but I admit I didn't read it thoroughly _{(gotta do that one day)}. But I know other regex implementations which do this kind of optimization (PCRE comes to mind). Anyway, the .NET implementation has its own way of optimizing things, you should rely on that.