Question

I am trying to parse through some log files and put them into a database for analysis. A single line looks something like this:

2012-09-30 17:16:27,213 [39] (boxes) ERROR Assembly.Places [(null)] - Error while displaying a thing

I have made a regular expression that works well for pulling out the date in front and breaking up the lines that way, but I lose the date itself. This is a pretty important bit of data, and I don’t want to lose it!

I cannot just do this by \r\n, because some logs are fatal errors that include stack traces for the developers. Those, obviously, use \r\n to make them readable.

My current code looks like this for reference:

var logpath = Directory.GetFiles(@"C:\a\directory", "*.log");
foreach (var log in logpath)
{
    var fileStream = new StreamReader(log);
    var fileString = fileStream.ReadToEnd();
    var records = Regex.Split(fileString, "[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}:[0-9]{2},[0-9]{3}");
            ...
}

Answer 1

Split() will always remove the matched delimiter. The trick is not to match any actual text, but rather a position in the string.

This is done through zero-width look-ahead:

var datePattern = "^(?=[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}:[0-9]{2},[0-9]{3})";
var datePositions = new Regex(datePattern, RegexOptions.Multiline);

// ...

Regex.Split(fileString, datePositions);