Git: How to estimate a contribution of a person to my project in terms of added/changed lines of code?

I have a GIT repository and I want to calculate how many lines of code were added/changed by one person or a group of persons during some period of time. Is it possible to calculate with git?

Solution 1:

You can use git log and some shell-fu:

git log --shortstat --author "Aviv Ben-Yosef" --since "2 weeks ago" --until "1 week ago" \
    | grep "files\? changed" \
    | awk '{files+=$1; inserted+=$4; deleted+=$6} END \
           {print "files changed", files, "lines inserted:", inserted, "lines deleted:", deleted}'

Explanation: git log --shortstat displays a short statistic about each commit, which, among other things, shows the number of changed files, inserted and deleted lines. We can then filter it for a specific committer (--author "Your Name") and a time range (--since "2 weeks ago" --until "1 week ago").

Now, in order to actually sum up the stats instead of seeing the entry per commit, we do some shell scripting to do it. First, we use grep to filter only the lines with the diffs. These lines look like this:

 8 files changed, 169 insertions(+), 81 deletions(-)

or this:

 1 file changed, 4 insertions(+), 4 deletions(-)

We then sum these using awk: for each line we add the files changed (1st word), inserted lines (4th word) and deleted lines (6th word) and then print them after summing it all up.

Edit: forward slashes were added in the top snippet so it can be copy and pasted into a command line.

Solution 2:

You can generate stats using Gitstats. It has an 'Authors' section which includes number of lines add/removed by the top 20 authors (top 20 by commit count).

Edit: There's also Git: Blame Statistics

Solution 3:

Run this command:

git log --pretty=format:'' --numstat --author 'Lu4' | awk 'NF' | awk '{insertions+=$1; deletions+=$2} END {print NR, "files changed,", insertions, "insertions(+),", deletions, "deletions(+)"}';

This command is very close to the clever one in abyx's answer, but it also handles the edge case found by Wallace Sidhrée. Sometimes, a commit involves deletions only (i.e., no insertions). The command in abyx's answer incorrectly reads those deletions as insertions. The command here reads them correctly because it uses --numstat instead of --shortstat. Unlike --shortstat, --numstat includes both the insertions and deletions for those commits.

Note that both commands include binary files in the file count but exclude the number of lines inserted and deleted inside those binaries.

Here is another useful trick. Create a file called gitstats with this content:

#!/usr/bin/env bash

git log --pretty=format:'' --numstat "$@" | awk 'NF' | awk '{insertions+=$1; deletions+=$2} END {print NR, "files changed,", insertions, "insertions(+),", deletions, "deletions(+)"}';

Then you can run that command with any extra options to git log you want. Here are some examples:

./gitstats;
./gitstats --since '1 month ago';
./gitstats --since '1 month ago' --until '1 day ago';
./gitstats --author 'Lu4' --since '1 month ago' --until '1 day ago';

(The file can be named something other than gitstats, of course.)

Solution 4:

For particular dates, you can use --since "2012-08-27" --until "2012-09-01"

Like

git log --shortstat --author "Fabian" --since "2012-08-27" --until "2012-09-01" | grep "files changed" | awk '{files+=$1; inserted+=$4; deleted+=$6} END {print "files changed", files, "lines inserted:", inserted, "lines deleted:", deleted}'

Check this explanation

http://gitref.org/inspect/