为什么此正则表达式在.NET中失败？

Question

Take a look at this test method:

[ Test ]
public static void TestRegex() {
    var goodTextsToTest = new List<string>
        {
            "http://google.com",
            "https://google.com/",
            "ftp://bugger!!!one1",
            "ftss://shoot",
            "somelongergibberish://flkjd",
            "thescheme://green"
        };
    var badTextsToTest = new List<string> { "bad432:4h//orange", "1ftp://1bugger!!!one1", "IAmTheVeryModelOfAModernMajorGeneral", "" };

    var regex = new Regex( "^([a-z][a-z0-9+\\.\\-]*)*://", RegexOptions.IgnoreCase );

    foreach( var txt in badTextsToTest )
        Assert.IsFalse( regex.IsMatch( txt ), "Passed but should have failed: " + txt );

    foreach( var txt in goodTextsToTest )
        Assert.IsTrue( regex.IsMatch( txt ), "Failed but should have passed: " + txt );
}

As it is currently written, this code never returns from var regex = new Regex( "^([az][a-z0-9+\\\\.\\\\-]*)*://", RegexOptions.IgnoreCase ); . The input the code gets stuck on is "IAmTheVeryModelOfAModernMajorGeneral".

Why does this regular expression cause an infinite loop when the input is "IAmTheVeryModelOfAModernMajorGeneral" ?

Bonus question: This code does finish executing if you remove "://" from the regular expression. Ie var regex = new Regex( "^([az][a-z0-9+\\\\.\\\\-]*)*", RegexOptions.IgnoreCase ); Why does this fix it?

Answer 1

Your regex, designed as it is now will take more than 87500 steps to complete, because of backtracking. See the debbuger here . This is what we call catastrophic backtracking