Write a first-ownership collection script

[andymeneely]

In lib/scripts write a new data collection script that collects the first time a given developer was added to a directory for OWNERS. We'll use this to compute time-to-ownership (which will be a separate issue).

The output of the script should be a CSV like this:

email	directory	first-owner-commit
foo@chromium.org	crypto/	abcd3234
bar@chromium.org	crypto/	2309c3ff3

Some issues to watch out for:

• OWNERs files come and go, just as directories do
• Developers can become OWNERs and then leave as OWNERs,
• Handling the OWNER hierarchy of files is going to be tricky.

Here's how we do this:

• Given a list that I'll give you of all commits to OWNERS, parse it
• For each commit, do a git checkout COMMIT
• For each OWNERS file in that tree, get a list of all email addresses mentioned
• Keep a hash of OWNER-OWNERFILE (e.g. brett@chromium.org-src/" => [abc123, 'Aug 1 2009']). Keep track of the earliest date for each directory and owner.
• Then save it out to a CSV

I used this query to get the list of commits: CommitFilepath.where("filepath LIKE '%/OWNERS' ").pluck(:commit_hash).uniq

Please write this in Ruby. Remember that backticks allow you to execute command line commands and save the input, like this:

out = `git log`

[andymeneely]

Next up: merge into master and finish the verifies.