nuke_emojis.sh

#!/bin/bash

# ==============================================================================
# nuke_emojis.sh
# ==============================================================================
#
# This script interactively helps you remove emoji characters and normalize
# whitespace in your codebase. It's designed to be Git-aware, processing
# only tracked or untracked (but not ignored) text files.
#
# Key Features:
# - **Emoji Removal:** Strips `\p{Extended_Pictographic}` emojis, Zero Width
#   Joiners (\x{200d}), and Variation Selectors (\x{fe0f}).
# - **Whitespace Normalization:** Collapses multiple horizontal spaces, trims
#   leading/trailing whitespace from lines, and collapses multiple blank lines.
# - **Interactive Mode:** Prompts for confirmation for each file with a diff
#   preview, allowing you to accept, skip, or apply to all remaining.
# - **Dry Run Mode:** Shows changes without modifying any files.
# - **Automatic Application:** Can apply changes to all files without prompting.
# - **Git Integration:** Respects `.gitignore` and can automatically `git add`
#   modified files.
# - **Safety:** Creates `.bak` backups of original files and cleans up its
#   temporary files automatically.
#
# Usage:
#   ./nuke_emojis.sh                        # Interactive mode
#   ./nuke_emojis.sh --dry-run / -d         # Show changes, don't modify files
#   ./nuke_emojis.sh --yes / -y             # Apply all changes automatically
#
# Ensure you have `git`, `grep`, `perl`, and `mktemp` installed.
# It's always recommended to back up your repository before running:
# `git commit -am "Pre-emoji cleanup backup"`
#
# ==============================================================================

# Exit immediately if a command exits with a non-zero status.
set -e

# ANSI colors (self-contained)
YELLOW='\033[0;33m'
BLUE='\033[0;34m'
GREEN='\033[0;32m'
RED='\033[0;31m'
NC='\033[0m'

strip_emoji_interactive() {
  # --- Temporary File Cleanup Trap ---
  # Array to store paths of temporary files created by mktemp
  local tmp_files=()

  # Define cleanup function to be executed on script exit
  cleanup() {
    echo -e "${BLUE}Cleaning up temporary files...${NC}"
    for f in "${tmp_files[@]}"; do
      if [[ -f "$f" ]]; then
        rm -f "$f"
        echo -e "${BLUE}  Removed: $f${NC}"
      fi
    done
  }

  # Register the cleanup function to run on EXIT (normal or abnormal termination)
  trap cleanup EXIT

  # --- Parse Command Line Options ---
  local auto_apply=false # Apply changes automatically (non-interactive)
  local dry_run=false    # Show diffs but do not modify files
  local apply_all_remaining=false # Batch apply mode for subsequent files

  # Loop through arguments to parse flags
  while [[ $# -gt 0 ]]; do
    case "$1" in
      -y|--yes)
        auto_apply=true
        ;;
      -d|--dry-run)
        dry_run=true
        ;;
      *)
        echo -e "${RED}Error: Unknown option '$1'${NC}"
        echo "Usage: strip_emoji_interactive [--yes|-y] [--dry-run|-d]"
        return 2 # Indicate an error
        ;;
    esac
    shift # Move to next argument
  done

  echo -e "${YELLOW}Starting emoji stripper.${NC}"
  echo "Files will be checked based on Git status (tracked or untracked/non-ignored text files)."
  if [[ "$dry_run" == "true" ]]; then
    echo -e "${YELLOW}!!! DRY RUN MODE: No files will be modified. !!!${NC}"
  else
    echo "A .bak file will be created for each modified file."
  fi
  echo "----------------------------------------------------"

  local files_processed=0
  local files_cleaned=0
  local exit_status=0 # Default to success

  # Use git ls-files for comprehensive file discovery (tracked and untracked but not ignored).
  # -z: null-separated output for safety with filenames containing spaces/newlines.
  # -c: tracked files, -o: other (untracked) files, --exclude-standard: respects .gitignore.
  git ls-files -z -c -o --exclude-standard |
    while IFS= read -r -d '' file; do
      # Speed optimization: Combine binary check and emoji regex check into one grep -IqP
      # This performs one pass over the file instead of two.
      if grep -IqP '\p{Extended_Pictographic}|\x{200d}|\x{fe0f}' "$file"; then
        files_processed=$((files_processed + 1))
        # Handle spaces and weird filenames more safely by using printf
        printf "${BLUE}File: %s${NC}\n" "$file"

        # Create a temporary "would-be" cleaned version
        local tmp_clean="$(mktemp)"
        # Push mktemp result into tmp_files array for cleanup
        tmp_files+=("$tmp_clean")

        # Apply Perl transformations to create the cleaned version in the temp file
        perl -CSDA -pe \
          's/\p{Extended_Pictographic}|\x{200d}|\x{fe0f}//g; # Strip emojis
           s/\h{2,}/ /g;                                   # Collapse multiple horizontal whitespace to single space
           s/^\s+|\s+$//g;                                 # Trim leading/trailing whitespace
           s/\n{3,}/\n\n/g;' "$file" > "$tmp_clean"

        # Show diff preview (side-by-side, colorized)
        git --no-pager diff --no-index --color=always "$file" "$tmp_clean"

        echo

        local reply=""
        if [[ "$dry_run" == "true" ]]; then
          echo -e "${YELLOW}DRY RUN: Changes shown above. File not modified.${NC}"
          reply="n" # In dry-run, always treat as "no" to actual modification
        elif [[ "$auto_apply" == "true" || "$apply_all_remaining" == "true" ]]; then
          reply="y" # Auto-apply if --yes or --all-remaining is set
        else
          read -rp "$(echo -e ${YELLOW}Apply these changes to this file? [y/N/a (all remaining)]${NC} ) " reply
        fi

        if [[ $reply =~ ^[Yy]$ ]]; then
          if [[ "$dry_run" == "false" ]]; then
            # Apply changes: Create a backup and move the cleaned temp file over the original.
            cp "$file" "$file.bak"
            mv "$tmp_clean" "$file"
            echo -e "${GREEN}→ Cleaned: $file${NC}"
            files_cleaned=$((files_cleaned + 1))

            # Git staging awareness: Stage the modified file
            git add "$file"
            echo -e "${GREEN}→ Cleaned and staged: $file${NC}"
          fi
        elif [[ $reply =~ ^[Aa]$ ]]; then
          apply_all_remaining=true
          if [[ "$dry_run" == "false" ]]; then
            cp "$file" "$file.bak"
            mv "$tmp_clean" "$file"
            echo -e "${GREEN}→ Cleaned: $file${NC}"
            files_cleaned=$((files_cleaned + 1))
            git add "$file"
            echo -e "${GREEN}→ Cleaned and staged: $file${NC}"
          fi
        else
          # Cleanup temp file directly if not moving it
          # No need to remove via rm "$tmp_clean" here, trap will handle it on exit
          echo -e "${RED}→ Skipped: $file${NC}"
          exit_status=1 # User aborted at least one file
        fi
        echo "---------------------------------------"
      fi # End of grep -IqP
    done # End of while read file

  echo -e "${GREEN}Emoji stripping interactive process completed.${NC}"
  echo -e "${BLUE}Processed $files_processed files, cleaned $files_cleaned files.${NC}"

  # Set final exit status based on operations
  if [[ $files_processed -eq 0 && $exit_status -eq 0 ]]; then
    # No relevant files found, or no changes needed, and no errors/aborts.
    return 0
  elif [[ $exit_status -ne 0 ]]; then
    # Some files were skipped by user.
    return 1
  else
    # All relevant files processed/cleaned successfully.
    return 0
  fi
}

# --- Main Execution ---
# Call the function with all command-line arguments passed to the script
strip_emoji_interactive "$@"