How do I calculate the code-to-comment ratio of a C# project?

Question

Note: I am not asking what the golden code-to-comment ratio is, nor am I attempting to affix a particular ratio to our team. Instead, we would like to improve how well our codebase is documented (we started with a "code should document itself" mentality), which can be accomplished either by removing dead code or by adding comments to live code, and we would like to measure how well we are going about doing that by measuring this ratio multiple times over the course of several months. Also note that I would like to actually measure the amount of comments we have, so something that gets LOC from the generated IL won't work.

How would I go about getting the code-to-comments ratio for a C# project? Would I need to write my own parsing script, or is there something in Roslyn I can leverage? Do any major IDEs carry this functionality directly? As a bonus, can I filter out "punctuation", such as extra whitespace, comment delimiters (// and /* */), and opening/closing curly brackets?

For what it's worth, writing clearer code is more valuable than adding more comments. — Robert Harvey, Nov 21 '18 at 20:40
[This page has some regular expressions that should help.](http://www.spoiledtechie.com/post/2011/11/22/How-To-Count-Lines-of-Code-in-Visual-Studio.aspx) — Robert Harvey, Nov 21 '18 at 20:44
@RobertHarvey I tried those regexes (following all instructions) and they didn't match any lines. — TheHans255, Nov 21 '18 at 20:54
Yeah, that page seems hopelessly out of date. They're not even valid regexes. Try the same technique, using this regex: `(/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+/)|(//.*)` — Robert Harvey, Nov 21 '18 at 21:08
@RobertHarvey OK, this looks like a good start. Nice website, too! — TheHans255, Nov 21 '18 at 21:30

score 2 · Answer 1 · answered Nov 21 '18 at 23:16

2

You can identify the comment lines in your code using this regex:

(/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+/)|(//.*)

Try plugging it into the "Find in Files" functionality in Visual Studio to see it in action.

https://regex101.com/r/GCrfzc/1

answered Nov 21 '18 at 23:16

Robert Harvey

178,213
47
333
501

This regex is correct, but a few of the files I've used it on end up timing out the regex engine due to catastrophic backtracking. – TheHans255 Nov 26 '18 at 23:32
1

Actually (and this is related to the test cases it's having issues with), this regex also does not account for comment delimiters appearing in strings (such as `"/path/to/folder/*"`). – TheHans255 Nov 26 '18 at 23:51

score 1 · Accepted Answer · edited Nov 28 '18 at 20:42

Using Robert Harvey's regex, I managed to create a short C# method that calculates this metric from an input string. It goes character by character in order to properly account for lines that have both code and comments on them, and also excludes additional whitespace from the metric so that things like line indentations don't count.

To prevent catastrophic backtracking, I simplified the regex (I found you don't need the newline checks, since character exclude groups already take care of those) and also made the body of the block comment a non-backtracking group.

public static double CodeToCommentRatio(
    string text, 
    out int codeChars, 
    out int commentChars, 
    out int blankChars)
{
    // First, filter out excess whitespace, reporting the number of characters removed this way
    Regex lineStartRegex = new Regex(@"(^|\n)[ \t]+");
    Regex blanksRegex = new Regex(@"[ \t]+");
    string minWhitespaceText = blanksRegex.Replace(lineStartRegex.Replace(text, "\n"), " ");
    blankChars = text.Length - minWhitespaceText.Length;

    // Then, match all comments and report the number of characters in comments
    Regex commentsRegex = new Regex(@"(/\*(?>[^*]|(\*+[^*/]))*\*+/)|(//.*)");
    MatchCollection comments = commentsRegex.Matches(minWhitespaceText);
    commentChars = 0;
    foreach (Match comment in comments)
    {
        commentChars += comment.Length;
    }
    codeChars = minWhitespaceText.Length - commentChars;

    // Finally, return the ratio
    return (double)codeChars / commentChars;
}

How do I calculate the code-to-comment ratio of a C# project?

2 Answers2