blame_info.sh

#!/bin/bash

#details: this file is intended to be called from inside generate_csv.sh
#usage: ./file_info.sh <Path to root> <Path to file> <blank lines int> <comment lines int> <code lines int>

#example "<ABSOLUTE_PATH>/../../"
OFFSET=$1 # how to get to the directory to search

#example:'Source/Engine/main.cpp'
SOURCEFILE=${2//###/ }
# note: currently as a workaround spaces in filepaths are passed from generate_csv.sh as '###'
#       it's handled automatically so you shouldn't need to worry about it.

LINES_BLANK=$3
LINES_COMMENT=$4
LINES_CODE=$5

LINES_TOTAL=$(($LINES_BLANK + $LINES_COMMENT + $LINES_CODE))
CODE_RATIO=$(echo $LINES_CODE / $LINES_TOTAL | bc -l)
CODE_RATIO_ROUND=$(printf %.3f $CODE_RATIO)

# using git blame on a single file path to figure out who worked on it (and line numbers)
cd $OFFSET

git blame "$SOURCEFILE" -e | awk -v file="$SOURCEFILE" '
{
  email_begin = index($0, "<") + 1;
  email_len  = index($0, ">") - email_begin;
  email = substr($0,email_begin, email_len);
  print  email " " file;
}
' | sort | uniq -c | awk -v file="$SOURCEFILE" -v total=$LINES_TOTAL -v ratio=$CODE_RATIO '
{ 
  lines_normalized=($1 * ratio);
  email = $2
  printf "%s, %s, %.0f\n", file, email, lines_normalized
}'

generate_csv.sh

#!/bin/bash
cd ${0%/*}

# please edit #
NAME="Premake5_LibTest"

# edit if necessary
OUTPUT_FILE="GGEngine_Prototypes_"$NAME".csv"
DIR="../"$NAME"/"

# don't edit #
SCRIPTPATH=`pwd -P`
SOURCE_FILES='temp_source_files.txt'
BLAME_SCRIPT="blame_info.sh"
cd $DIR;

# Step 1: use cloc for paths and line statistics; Stored in temporary file
cloc . --by-file | grep "./" | grep -v "http://" | awk -F"             " '
{
  path = substr($1,3,length($1)-2);
  vals = substr($0,length($1)+1, length($0));
  sub(" ", "###", path);
  print path  vals
}' > $SCRIPTPATH/$SOURCE_FILES;

# Run git blame on each path
cd $SCRIPTPATH
while read p; do 
    sh $BLAME_SCRIPT $DIR $p
done < $SOURCE_FILES | sort > $OUTPUT_FILE

# Clean up temporary file
rm $SCRIPTPATH/$SOURCE_FILES;