count_contributors.sh

#!/bin/bash

# Default start and end dates
DEFAULT_START_DATE="2025-01-01"
DEFAULT_END_DATE="2025-05-21"

# Initialize variables with default values
START_DATE="$DEFAULT_START_DATE"
END_DATE="$DEFAULT_END_DATE"
DISPLAY_EMAILS=false

# Parse command-line arguments
while [[ "$#" -gt 0 ]]; do
  case "$1" in
    --display-emails)
      DISPLAY_EMAILS=true
      ;;
    --start-date)
      if [[ -n "$2" ]]; then
        START_DATE="$2"
        shift # Consume the argument value
      else
        echo "Error: --start-date requires a date value (YYYY-MM-DD)."
        exit 1
      fi
      ;;
    --end-date)
      if [[ -n "$2" ]]; then
        END_DATE="$2"
        shift # Consume the argument value
      else
        echo "Error: --end-date requires a date value (YYYY-MM-DD)."
        exit 1
      fi
      ;;
    *)
      echo "Unknown option: $1"
      echo "Usage: $0 [--display-emails] [--start-date YYYY-MM-DD] [--end-date YYYY-MM-DD]"
      exit 1
      ;;
  esac
  shift # Consume the argument name
done

# Create a temporary file to store all collected author emails
TEMP_EMAIL_FILE=$(mktemp)

echo "Searching for Git repositories in subdirectories and collecting unique contributors..."
echo "Date range: $START_DATE to $END_DATE"
echo "--------------------------------------------------"

# Get the current working directory to return to later
CURRENT_DIR=$(pwd)

# Loop through each immediate subdirectory
# -maxdepth 1 ensures we only look at direct children, not nested sub-repos.
# -type d filters for directories only.
for dir in */; do
  # Remove the trailing slash from the directory name for cleaner output
  dir_name="${dir%/}"

  # Change into the subdirectory, suppressing output
  pushd "$dir_name" > /dev/null 2>&1

  # Check if the current subdirectory is a Git repository
  if git rev-parse --is-inside-work-tree > /dev/null 2>&1; then
    echo "Processing repository: $dir_name"
    # Get author emails from commits within the specified date range
    # and append them to the temporary file.
    git log --since="$START_DATE" --until="$END_DATE" --format='%aE' >> "$TEMP_EMAIL_FILE"
  else
    echo "Skipping '$dir_name': Not a Git repository."
  fi

  # Return to the original directory, suppressing output
  popd > /dev/null 2>&1
done

echo "--------------------------------------------------"
echo "Aggregation complete. Calculating unique contributors..."

# Count the total number of unique contributors from the collected emails
# sort -u sorts the emails and removes duplicates.
# wc -l counts the number of lines (which are now unique emails).
TOTAL_UNIQUE_CONTRIBUTORS=$(cat "$TEMP_EMAIL_FILE" | sort -u | wc -l)

echo "--------------------------------------------------"
echo "Total number of unique contributors across all scanned repositories: $TOTAL_UNIQUE_CONTRIBUTORS"
echo "--------------------------------------------------"

# Conditionally display the list of unique contributors
if $DISPLAY_EMAILS; then
  echo ""
  echo "List of all unique contributors (email addresses) across all scanned repositories:"
  echo "--------------------------------------------------"
  # Display the unique email addresses
  cat "$TEMP_EMAIL_FILE" | sort -u
  echo "--------------------------------------------------"
fi

# Clean up the temporary file
rm "$TEMP_EMAIL_FILE"

echo "Script finished."