CSV Generator for Graphing Bytes of Code Since a Date in the Past

code_timeline.sh

#!/bin/bash
#run this like: DAYS=365 ./code_timeline.sh https://github.com/org/repo1.git https://github.com/org/repo2.git
#note that the types of files it checks for code are hardcoded below
code='\.cc$|\.h$|\.cpp$|\.hpp$|\.lua$|\.py$|\.js'

: ${DAYS:?Please set DAYS environment variable to a number of days in the past youd like to capture}

#setup each repo
echo -n "day,"
for url in ${@}; do
	name=$(basename ${url})
	repo="${name%.*}"
	echo -n "${repo},"
	rm -rf ${repo}
	git clone $url &> /dev/null #purposely dont get submodules
done

#for each day
current=-${DAYS}
while [ ${current} -lt 1 ]; do
	#date to look for
	echo
	echo -n "$((DAYS+current)),"
	d=$(date -d "${current} days" +%Y-%m-%d)
	((current++))
	#for each repo
	for url in ${@}; do
		#get in there and find the revision matching this date
		name=$(basename ${url})
		pushd "${name%.*}" &> /dev/null
		branch=$(git rev-parse --abbrev-ref HEAD)
		revision=$(git rev-list -n 1 --before="${d} 23:59" ${branch})
		bytes=0
		#do we have a revision that old
		if [ ${revision} ]; then
			#checkit out and see if we had code
			git checkout ${revision} &> /dev/null
			if [ "$(find . -type f | grep -E ${code})" != "" ]; then
				#count the bytes
				bytes=$(wc -c $(find . -type f | grep -E ${code}) | tail -n 1 | awk '{printf $1}')
			fi
			git checkout ${branch} &> /dev/null
		fi
		popd &> /dev/null
		#add that number to the file
		echo -n "${bytes},"
	done
done