如何提高正则表达式的性能？

Question

I have some problem. I need replace some text inside string:

var str = "<tr class=\"fieldType\"><td><a href=\"#\" onclick=\"javascript:removeNestedForm(this,&#39;tr.fieldType&#39;,&#39;.mark-for-delete&#39;,false);return false;\">Удалить</a><input data-val=\"true\" data-val-number=\"The field Id must be a number.\" data-val-required=\"The Id field is required.\" id=\"FieldTypes[0]_635503397304941429__Id\" name=\"nestedObject.Id\" type=\"hidden\" value=\"0\" /><input data-val=\"true\" data-val-number=\"The field DocTypeId must be a number.\" data-val-required=\"The DocTypeId field is required.\" id=\"FieldTypes[0]_635503397304941429__DocTypeId\" name=\"nestedObject.DocTypeId\" type=\"hidden\" value=\"0\" /><input class=\"mark-for-delete\" data-val=\"true\" data-val-required=\"The IsDel field is required.\" id=\"FieldTypes[0]_635503397304941429__IsDel\" name=\"nestedObject.IsDel\" type=\"hidden\" value=\"False\" /><input data-val=\"true\" data-val-required=\"The CanDel field is required.\" id=\"FieldTypes[0]_635503397304941429__CanDel\" name=\"nestedObject.CanDel\" type=\"hidden\" value=\"True\" />    </td>    <td>        <select id=\"FieldTypes[0]_635503397304941429__Convertion\" name=\"nestedObject.Convertion\" style=\"width:99%;\"><option value=\"int\">int</option><option value=\"string\">string</option></select></td><td><input data-val=\"true\" data-val-required=\"Название не может быть пустым\" id=\"FieldTypes[0]_635503397304941429__Name\" name=\"nestedObject.Name\" style=\"width:99%;\" type=\"text\" value=\"\"></td><td><input id=\"FieldTypes[0]_635503397304941429__Description\" name=\"nestedObject.Description\" style=\"width:99%;\" type=\"text\" value=\"\" /></td></tr>";

I use method:

private static string ReplaceAttribute(string source, string name, string found, string replaced)
{
    string pattern = string.Format(@"({0}=[\\""]*(\w*[._\[\]]?)*)({1})", name, found);
    string replacement = "$1" + replaced;

    var theRegex = new Regex(pattern, RegexOptions.Compiled | RegexOptions.Singleline);
    var result = theRegex.Replace(source, replacement);

    return result;
}

My code works for a long time:

strPartial = ReplaceAttribute(strPartial, "id", propertyNameFake, collectionProperty + "_" + ticks + "_");
strPartial = ReplaceAttribute(strPartial, "name", propertyNameFake, collectionProperty + "[" + ticks + "]");
strPartial = ReplaceAttribute(strPartial, "data-valmsg-for", propertyNameFake, collectionProperty + "[" + ticks + "]");

How to improve the performance of the regular expression? Thanks.

Answer 1

You are looking for backslashes and quotation marks, but there are no backslashes before the quotation marks in the code, that's only in the C# string literal. Just look for the quotation marks, ie ""? instead of [\\\\""]* . (Note also that \\\\ in a @ delimited string ends up as \\\\ in the string, not \\ ).

But here comes the real speadup; You have conditional values nested inside each other, ie optional alphanumerics followed by an optional separator, repeated zero or more times: (\\w*[._\\[\\]]?)*) . Instead you should just use a set with the characters: [\\w\\._\\[\\]]* .

When the string is matched, the conditional values will start by matching as much as possible, then backtrack to find the longest match where the rest of the pattern matches. With nested conditionals there will be an huge amount of backtracking.

When I tested the changes with your example string the code runs about 600 times faster (11 ms instead of 6240 ms).