This script can summarize contributions of each author on given filename extension (defaults .py)

git_contribution_stats.sh

#!/bin/bash
# the script is fully created by Google Gemini
# Function to display usage information
usage() {
  echo "Usage: $0 [file_extension]"
  echo "  [file_extension] - Optional file extension to filter by (default: .py)"
  exit 1
}

# Default file extension
file_extension=".py"

# Check if an argument is provided and is not an option
if [[ $# -gt 0 && ! "$1" =~ ^- ]]; then
  file_extension="$1"
  shift  # Shift arguments to remove the processed file extension
fi

# Check for any unexpected additional arguments
if [[ $# -gt 0 ]]; then
  echo "Error: Invalid argument(s) provided."
  usage
fi

# Get a sorted list of unique author names
author_names=$(git log --pretty=format:"%an" | sort -u)

# Check if there are any authors
if [[ -z "$author_names" ]]; then
  echo "No authors found in the commit history."
  exit 0
fi

# Initialize total counters
total_added_lines=0
total_removed_lines=0
total_changed_lines=0

# Create a temporary file to store author stats
stats_file=$(mktemp)

# Iterate through each author to get their individual stats and accumulate total stats
while IFS= read -r author_name; do
  # Get numstat output for this author, for the specified file extension
  stats=$(git log --author="$author_name" --pretty=tformat: --numstat --textconv -- "*$file_extension")

  # Check if git log command was successful
  if [[ $? -ne 0 ]]; then
    echo "Error: Failed to retrieve git log for author '$author_name'."
    continue
  fi

  # Initialize counters for the author
  added_lines=0
  removed_lines=0

  # Process the numstat output
  while IFS=$'\t' read -r added removed file; do
      if [[ "$file" == *"$file_extension" ]]; then
          if [[ -n "$added" ]]; then
              added_lines=$((added_lines + added))
          fi
          if [[ -n "$removed" ]]; then
              removed_lines=$((removed_lines + removed))
          fi
      fi
  done <<< "$stats"

  # Store author stats in the temporary file
  echo "$author_name,$added_lines,$removed_lines" >> "$stats_file"

  # Accumulate total stats
  total_added_lines=$((total_added_lines + added_lines))
  total_removed_lines=$((total_removed_lines + removed_lines))
  total_changed_lines=$((total_added_lines + removed_lines))

done <<< "$author_names"

# Check if there are any changes
if [[ $total_changed_lines -eq 0 ]]; then
  echo "No changes found for files with extension '$file_extension' in the commit history."
  rm "$stats_file"  # Clean up temporary file
  exit 0
fi

# Iterate through each author to print stats and calculate percentage
while IFS=',' read -r author_name added_lines removed_lines; do
  changed_lines=$((added_lines + removed_lines))

  # Calculate percentage
  if (( total_changed_lines > 0 )); then
    percentage=$(awk "BEGIN {printf \"%.2f\", ($changed_lines*100/$total_changed_lines)}")
  else
    percentage=0
  fi

  # Print the results for the author
  echo "Author: $author_name"
  echo "  Added lines: $added_lines"
  echo "  Removed lines: $removed_lines"
  echo "  Total changed lines: $changed_lines"
  echo "  Contribution Percentage: $percentage%"
  echo ""
done < "$stats_file"

# Print overall total stats
echo "Overall Total Added Lines: $total_added_lines"
echo "Overall Total Removed Lines: $total_removed_lines"
echo "Overall Total Changed Lines: $total_changed_lines"

# Clean up temporary file
rm "$stats_file"