codebase-insights

#!/bin/bash 
cd $1
file_count=$(git ls-tree -r --name-only HEAD | wc -l | xargs)

current_file=1
default_since='2019-01-01'
default_fixes_pattern='fix|revert|bug'
echo "Collecting stats of repository files since ${3:-$default_since}"
echo "--> total files: $file_count"
# START CSV
echo "filename, lines, changes, active_days, authors, updated_at, fixes" > $2
git ls-tree -r --name-only HEAD | while read filename; do
    changes=$(git log --pretty='format: %s' --since=${3:-$default_since} -- "$filename" | wc -l)
    active_days=$(git log --pretty='format: %ad' --date=short --since=${3:-$default_since} -- "$filename" | sort -u | wc -l)
    fixes=$(git log --pretty='format: %s' --since=${3:-$default_since} -- "$filename" | grep -E -i "${3:-$default_fixes_pattern}" | wc -l)
    authors=$(git log --pretty='format: %an' --since=${3:-$default_since} -- "$filename" | sort -u | wc -l)
    lines=$(wc -l "$filename" | awk '{print $1;}')
    updated=$(git log -1 --pretty='format: %ad' --date=short -- "$filename")
    changes=$((changes))
    echo -ne "--> processed:   $current_file \r\c"
    echo "$filename, $lines, $changes, $active_days, $authors, $updated, $fixes" >> $2
    
    current_file=$((current_file+1))
done
# END CSV
echo -ne "\n"
echo "--> DONE"