Rewind: git history analysis scripts

A small library of handy git analysis scripts roughly inspired by Gary Bernhardt's Destroy All Software screencasts, and extensively documented in my ebook Unfuck A Monorail For Great Justice.

Use these when you want to quickly generate meaningful history reports for git projects.

stats

Run a statistics bash script against a git repository. For each filename, discover lines of code, number of commits, and the date of the first and the last commit. Then pull that into a spreadsheet app, sort by each metric, and get an idea which files are the most important files in the history of the project.

bash file_stats.sh [dir] [filetypes] > stats.csv
open -a Numbers stats.csv

For example:

bash file_stats.sh /ember-project "js" "handlebars"

Or:

bash file_stats.sh /rails-project "rb" "haml" "coffee" "scss"

authorship

This is more tricky, and requires Ruby. First run the determine_authors.sh script to get a count of all the authorship events in the repo; then run the tally_authors.rb Ruby script to sort the authorship events by name.

You can do all that as a one-liner in bash like this:

ruby tally_authors.rb <(cd /project && /this_dir/determine_authors.sh "js")

The code isn't tricky at all, and the output is easy to understand, but it's also easy to misconstrue. That's the tricky part. A developer could show up with few authorship events because they're new to a team, because they often do pair programming, because they prefer large commits to small ones, or for several other possible reasons. Exercise good judgement.

Also, where file_stats.sh will examine multiple file types, determine_authors.sh and tally_authors.rb assumes you're only working with one type of file at a time. Pull requests welcome!

Caveat: Running this code against extremely large projects with very long histories (e.g. Rails) might be very slow.

Why Rewind?

Rewind is currently a key character in the IDW Transformers comics.

license

MIT license.