dotfiles/local/bin/git-attributes

#!/usr/bin/env bash
#
# Check git repo's files .gitattributes and ensure all of them are mapped.
#
# Author: Ismo Vuorinen <https://github.com/ivuorinen> 2022
# License: MIT

set -euo pipefail

# Default configuration
VERBOSE=0
CHECK_PATTERN="text: auto"
EXIT_ON_MISSING=0
SUGGEST_RULES=1     # Suggestions enabled by default
WRITE_RULES=0       # Writing to file is opt-in
FORMAT_WIDTH=0      # Auto-width by default (0 means auto)
MIN_FORMAT_WIDTH=20 # Minimum format width

DEBUG="${DEBUG:-0}"

if [ "$DEBUG" -eq 1 ]; then
  set -x
fi

# Output functions
msg_err()
{
  echo -e "\e[31m$*\e[0m" >&2
}

msg_success()
{
  echo -e "\e[32m$*\e[0m"
}

msg_warn()
{
  echo -e "\e[33m$*\e[0m" >&2
}

msg_info()
{
  echo -e "\e[36m$*\e[0m"
}

msg_debug()
{
  [[ $VERBOSE -eq 1 ]] && echo -e "\e[35m$*\e[0m"
}

show_help()
{
  cat << EOF
Usage: $(basename "$0") [OPTIONS]

Check if all git-tracked files have corresponding rules in .gitattributes

Options:
  -h, --help          Display this help message
  -v, --verbose       Enable verbose output
  -e, --exit          Exit with error code if missing attributes found
  -p, --pattern       Pattern to check (default: "text: auto")
  -n, --no-suggest    Don't suggest .gitattributes rules (suggestions are on by default)
  -w, --write         Write suggested rules to .gitattributes file
  -f, --format-width  Specify width for formatting rule patterns (default: auto, min: $MIN_FORMAT_WIDTH)

EOF
}

# Parse arguments
while [[ $# -gt 0 ]]; do
  case "$1" in
    -h | --help)
      show_help
      exit 0
      ;;
    -v | --verbose)
      VERBOSE=1
      shift
      ;;
    -e | --exit)
      EXIT_ON_MISSING=1
      shift
      ;;
    -p | --pattern)
      CHECK_PATTERN="$2"
      shift 2
      ;;
    -n | --no-suggest)
      SUGGEST_RULES=0
      shift
      ;;
    -w | --write)
      WRITE_RULES=1
      shift
      ;;
    -f | --format-width)
      if [[ $2 =~ ^[0-9]+$ ]]; then
        FORMAT_WIDTH=$2
        shift 2
      else
        msg_err "Error: --format-width requires a numeric argument"
        exit 1
      fi
      ;;
    *)
      msg_err "Unknown option: $1"
      show_help
      exit 1
      ;;
  esac
done

# Function to check if git is installed
check_git_installed()
{
  if ! command -v git &> /dev/null; then
    msg_err "git could not be found, please install it first"
    exit 1
  fi
}

# Check if we're in a git repository
check_git_repo()
{
  if ! git rev-parse --is-inside-work-tree &> /dev/null; then
    msg_err "Not inside a git repository"
    exit 1
  fi
}

# Check if we're in the git root directory
check_git_root()
{
  local git_root
  git_root=$(git rev-parse --show-toplevel)
  local current_dir
  current_dir=$(pwd)

  if [[ "$git_root" != "$current_dir" ]]; then
    msg_err "Not in git repository root directory"
    msg_warn "Please run this command from: $git_root"
    exit 1
  fi
}

# Check if .gitattributes exists
check_gitattributes_exists()
{
  if [[ ! -f ".gitattributes" ]]; then
    msg_err ".gitattributes file not found in the repository root"
    msg_warn "Create a .gitattributes file before running this command"
    exit 1
  fi
}

# Format rule with proper alignment
format_rule()
{
  local pattern="$1"
  local attributes="$2"
  local width="$3"

  # If pattern starts with "#", it's a comment, don't format
  if [[ "$pattern" == "#"* ]]; then
    echo "$pattern"
    return
  fi

  # If pattern is empty, return empty
  if [[ -z "$pattern" ]]; then
    echo ""
    return
  fi

  printf "%-${width}s %s\n" "$pattern" "$attributes"
}

# Get the file extension properly, handling special cases
get_file_extension()
{
  local file="$1"
  local basename
  basename=$(basename "$file")
  local extension=""

  # Check if file has no extension or is a dotfile
  if [[ "$basename" == .* && ! "$basename" =~ \..+$ ]]; then
    # It's a dotfile without extension (like .gitignore)
    extension="$basename"
  elif [[ "$basename" =~ \..+$ ]]; then
    # Normal file with extension
    extension="${basename##*.}"

    # Check for special cases like .d/ directories
    if [[ "$extension" == "d" ]]; then
      # This is likely a .d directory - use the full filename as pattern
      if [[ -f "$file" ]]; then
        # For files in .d directories, use the complete path as pattern
        extension=$(basename "$file")
      else
        # For .d directory itself, use *.d
        extension="d"
      fi
    fi
  else
    # No extension at all
    extension="$basename"
  fi

  echo "$extension"
}

# Suggest appropriate gitattributes rules based on file extension
suggest_rule()
{
  local file="$1"
  local extension=""
  local pattern=""
  local attributes=""

  msg_debug "Checking file: $file"

  # Skip directories
  if [[ -d "$file" ]]; then
    return
  fi

  # Get proper file extension
  extension=$(get_file_extension "$file")

  # If file path contains .d/ pattern, we need special handling
  if [[ "$file" =~ \.d/ ]]; then
    # Extract the pattern part that includes the .d/ directory
    local dir_part
    dir_part=$(dirname "$file")

    # Check if it's a config directory pattern worth capturing
    if [[ "$dir_part" =~ /(\.d|[^/]+\.d)$ ]]; then
      pattern="$dir_part/*"
      msg_debug "Detected .d directory pattern: $pattern"
    else
      # Use standard extension pattern
      pattern="*.${extension}"
    fi
  else
    # Standard file with extension
    pattern="*.${extension}"
  fi

  # Common text files
  case "$extension" in
    # Shell scripts
    sh | bash | zsh | fish)
      attributes="text eol=lf diff=shell"
      ;;

    # Web development
    html | htm | xhtml | css | scss | sass | less)
      attributes="text eol=lf diff=html"
      ;;
    js | jsx | ts | tsx | json | json5)
      attributes="text eol=lf diff=javascript"
      ;;

    # Programming languages
    php)
      attributes="text eol=lf diff=php"
      ;;
    py)
      attributes="text eol=lf diff=python"
      ;;
    rb)
      attributes="text eol=lf diff=ruby"
      ;;
    go)
      attributes="text eol=lf diff=golang"
      ;;
    java | kt | scala)
      attributes="text eol=lf diff=java"
      ;;
    c | cpp | h | hpp)
      attributes="text eol=lf diff=cpp"
      ;;

    # Documentation
    md | markdown | txt)
      attributes="text eol=lf"
      ;;

    # Configuration files
    yml | yaml | toml | ini | cfg | conf)
      attributes="text eol=lf"
      ;;

    # Git config files and similar patterns
    git)
      attributes="text eol=lf"
      ;;
    gitignore | gitattributes)
      attributes="text eol=lf"
      ;;

    # Binary files
    png | jpg | jpeg | gif | ico | svg | webp | avif)
      attributes="binary"
      ;;
    pdf | doc | docx | xls | xlsx | ppt | pptx)
      attributes="binary"
      ;;
    zip | tar | gz | 7z | rar)
      attributes="binary"
      ;;
    mp3 | mp4 | avi | mov | wav | ogg)
      attributes="binary"
      ;;
    ttf | otf | woff | woff2 | eot)
      attributes="binary"
      ;;

    # Default for unknown extensions
    *)
      # Try to guess if it's text by checking if it contains null bytes
      if file "$file" | grep -q text; then
        attributes="text eol=lf"
      else
        attributes="binary"
      fi
      ;;
  esac

  msg_debug "...suggesting $pattern $attributes"

  echo "$pattern:$attributes"
}

# Function to check for missing .gitattributes
check_gitattributes()
{
  local missing_attributes
  msg_info "Checking for pattern: $CHECK_PATTERN"

  missing_attributes=$(git ls-files | git check-attr -a --stdin | grep "$CHECK_PATTERN" || true)

  if [[ -n "$missing_attributes" ]]; then
    msg_warn "Missing .gitattributes rules detected"

    if [[ $SUGGEST_RULES -eq 1 ]]; then
      # Generate suggestions
      local suggestions

      # Generate the suggestions
      suggestions=$(suggest_gitattributes "$missing_attributes")

      # Display the suggestions
      echo ""
      echo "$suggestions"
      echo ""

      if [[ $WRITE_RULES -eq 1 ]]; then
        msg_debug "...writing to .gitattributes"
        write_to_gitattributes "$suggestions"
      fi
    else
      msg_err ".gitattributes rule missing for the following files:"
      echo "$missing_attributes"
    fi

    if [[ $EXIT_ON_MISSING -eq 1 ]]; then
      return 1
    fi
  else
    msg_success "All files have a corresponding rule in .gitattributes"
  fi

  return 0
}

# Parse rule string and extract pattern and attributes
parse_rule()
{
  local rule="$1"

  if [[ "$rule" == "#"* ]]; then
    # This is a comment line
    echo "$rule::"
    return
  fi

  if [[ "$rule" =~ ^([^[:space:]]+)[[:space:]]+(.*)$ ]]; then
    echo "${BASH_REMATCH[1]}:${BASH_REMATCH[2]}"
  else
    echo "$rule::"
  fi
}

# Check shell scripts by name regardless of extension
detect_shell_scripts()
{
  msg_debug "Detecting shell scripts by name regardless of extension..."

  local shell_scripts_rules=""
  local shell_scripts_found=0
  local patterns=()
  local attributes=()

  # Get already defined rules in .gitattributes
  local existing_rules
  existing_rules=$(grep -v "^#" .gitattributes || true)

  # Get all shell scripts regardless of extension
  local shell_scripts
  shell_scripts=$(git ls-files | xargs file | grep "shell script" | cut -d: -f1)

  if [[ -n "$shell_scripts" ]]; then
    shell_scripts_found=$(echo "$shell_scripts" | wc -l | tr -d ' ')
    msg_debug "Found $shell_scripts_found potential shell scripts."

    # Track scripts that need rules
    declare -A scripts_by_dir=()
    local need_rule_count=0

    # Process each script
    while IFS= read -r script; do
      local rel_path="${script#./}"

      # Skip if exact path already in .gitattributes
      if grep -q "^${rel_path} " <<< "$existing_rules"; then
        msg_debug "Script already in .gitattributes: $rel_path"
        continue
      fi

      # Skip if file extension already covered
      local extension
      extension=$(get_file_extension "$rel_path")
      if [[ "$extension" != "$rel_path" ]] && grep -q "^\*\.$extension " <<< "$existing_rules"; then
        msg_debug "Script covered by extension rule: $rel_path (*.$extension)"
        continue
      fi

      # Check if any parent directory is already covered
      local is_covered=0
      local dir_path="$rel_path"
      while [[ "$dir_path" != "." && "$dir_path" != "/" ]]; do
        dir_path=$(dirname "$dir_path")
        if [[ "$dir_path" == "." ]]; then
          break
        fi

        # Check if directory or any of its contents are covered
        if grep -q "^${dir_path}/\?" <<< "$existing_rules" || grep -q "^${dir_path}/\*" <<< "$existing_rules"; then
          msg_debug "Script covered by directory rule: $rel_path (${dir_path})"
          is_covered=1
          break
        fi
      done

      if [[ $is_covered -eq 1 ]]; then
        continue
      fi

      # Group by directory
      local dir
      dir=$(dirname "$rel_path")
      if [[ "$dir" == "." ]]; then
        dir="root"
      fi

      # Add to appropriate group
      if [[ -z "${scripts_by_dir[$dir]:-}" ]]; then
        scripts_by_dir[$dir]="$rel_path"
      else
        scripts_by_dir[$dir]="${scripts_by_dir[$dir]}\n$rel_path"
      fi

      ((need_rule_count++))
    done <<< "$shell_scripts"

    # Output grouped results
    if [[ $need_rule_count -gt 0 ]]; then
      patterns+=("# Shell scripts detected by content")
      attributes+=("")

      # Check if we can use directory-based rules instead of individual files
      for dir in "${!scripts_by_dir[@]}"; do
        local files_in_dir
        files_in_dir=$(echo -e "${scripts_by_dir[$dir]}" | wc -l)
        local dir_path="$dir"

        if [[ "$dir" == "root" ]]; then
          # For root directory files, list each individually
          while IFS= read -r file; do
            patterns+=("$file")
            attributes+=("text eol=lf diff=shell")
          done <<< "$(echo -e "${scripts_by_dir[$dir]}")"
        elif [[ $files_in_dir -gt 2 ]]; then
          # If directory has multiple scripts, suggest a directory pattern
          patterns+=("# Found $files_in_dir scripts in $dir_path")
          attributes+=("")

          # Special handling for .d directories
          if [[ "$dir_path" =~ \.d$ || "$dir_path" =~ \.d/ ]]; then
            patterns+=("$dir_path/*")
          else
            patterns+=("$dir_path/*")
          fi
          attributes+=("text eol=lf diff=shell")

          # List the files as comments for reference
          while IFS= read -r file; do
            patterns+=("# - ${file}")
            attributes+=("")
          done <<< "$(echo -e "${scripts_by_dir[$dir]}" | sort)"
        else
          # For directories with few scripts, list them individually
          while IFS= read -r file; do
            patterns+=("$file")
            attributes+=("text eol=lf diff=shell")
          done <<< "$(echo -e "${scripts_by_dir[$dir]}")"
        fi
      done

      msg_debug "Adding $need_rule_count shell scripts to suggestions (grouped by directory)."
    else
      msg_debug "All detected shell scripts already have rules."
    fi
  else
    msg_debug "No shell scripts detected."
  fi

  # Return the formatted arrays
  local rules_count=${#patterns[@]}
  for ((i = 0; i < rules_count; i++)); do
    echo "${patterns[$i]}:${attributes[$i]}"
  done
}

# Function to suggest gitattributes rules
suggest_gitattributes()
{
  local missing_attributes="$1"
  local files
  local extension_suggestions=()
  local formatted_suggestions=""
  local all_patterns=()
  local all_attributes=()
  local max_pattern_length=0

  # Add header to suggestions
  all_patterns+=("# Auto-generated rules - $(date +"%Y-%m-%d %H:%M:%S")")
  all_attributes+=("")

  msg_info "Suggested .gitattributes rules:"

  # First, detect shell scripts and add them to suggestions
  msg_info "Detecting shell scripts by content..."
  local shell_scripts_rules
  shell_scripts_rules=$(detect_shell_scripts)

  # Add shell script rules to patterns and attributes arrays
  if [[ -n "$shell_scripts_rules" ]]; then
    while IFS=':' read -r pattern attributes; do
      if [[ -n "$pattern" ]]; then
        all_patterns+=("$pattern")
        all_attributes+=("$attributes")

        # Update max pattern length (skip comments)
        if [[ "$pattern" != "#"* ]] && [[ ${#pattern} -gt $max_pattern_length ]]; then
          max_pattern_length=${#pattern}
        fi
      fi
    done <<< "$shell_scripts_rules"
  fi

  # Extract filenames from git check-attr output
  files=$(echo "$missing_attributes" | awk -F': ' '{print $1}')

  # Get suggestions for each file
  declare -A seen_patterns=()

  while IFS= read -r file; do
    local suggestion
    suggestion=$(suggest_rule "$file")
    if [[ -n "$suggestion" ]]; then
      IFS=':' read -r pattern attributes <<< "$suggestion"

      # Only add each pattern once
      if [[ -z "${seen_patterns[$pattern]:-}" ]]; then
        extension_suggestions+=("$suggestion")
        seen_patterns[$pattern]=1
      fi
    fi
  done <<< "$files"

  # Remove duplicates and sort
  local unique_extensions=()
  mapfile -t unique_extensions < <(printf '%s\n' "${extension_suggestions[@]}" | sort -u)

  # Add extension-based suggestions header if we have any
  if [[ ${#unique_extensions[@]} -gt 0 ]]; then
    all_patterns+=("")
    all_attributes+=("")

    all_patterns+=("# File extension-based rules")
    all_attributes+=("")

    # Add extension rules to patterns and attributes arrays
    for suggestion in "${unique_extensions[@]}"; do
      IFS=':' read -r pattern attributes <<< "$suggestion"

      all_patterns+=("$pattern")
      all_attributes+=("$attributes")

      # Update max pattern length
      if [[ ${#pattern} -gt $max_pattern_length ]]; then
        max_pattern_length=${#pattern}
      fi
    done
  fi

  # Use user-specified format width if provided, otherwise use max_pattern_length
  # But ensure it's at least MIN_FORMAT_WIDTH
  local format_width=$max_pattern_length
  if [[ $FORMAT_WIDTH -gt 0 ]]; then
    format_width=$FORMAT_WIDTH
  fi

  # Ensure minimum width
  if [[ $format_width -lt $MIN_FORMAT_WIDTH ]]; then
    format_width=$MIN_FORMAT_WIDTH
  fi

  msg_debug "Using format width: $format_width"

  # Format and output all suggestions with proper alignment
  local rule_count=${#all_patterns[@]}

  for ((i = 0; i < rule_count; i++)); do
    local pattern="${all_patterns[$i]}"
    local attributes="${all_attributes[$i]}"

    # Handle comments separately
    if [[ "$pattern" == "#"* ]] || [[ -z "$attributes" ]]; then
      formatted_suggestions+="$pattern\n"
      echo "$pattern"
    else
      local formatted_rule
      formatted_rule=$(printf "%-${format_width}s %s\n" "$pattern" "$attributes")
      formatted_suggestions+="$formatted_rule\n"
      echo "$formatted_rule"
    fi
  done

  # Add final message
  echo ""
  msg_info "Add these rules to your .gitattributes file to resolve missing attributes."

  # Return the full suggestion text so it can be both displayed and written to file
  echo -e "$formatted_suggestions"
}

# Write suggestions to .gitattributes file
write_to_gitattributes()
{
  local suggestions="$1"
  local gitattributes=".gitattributes"

  # Check if file exists and is writable
  if [[ ! -w "$gitattributes" ]]; then
    msg_err "Cannot write to $gitattributes. Check permissions."
    return 1
  fi

  # Add a newline at the end of the file if it doesn't have one
  if [[ -s "$gitattributes" ]] && [[ $(tail -c 1 "$gitattributes" | wc -l) -eq 0 ]]; then
    echo "" >> "$gitattributes"
  fi

  # Append suggestions to the file
  echo -e "$suggestions" >> "$gitattributes"

  msg_success "Added suggested rules to $gitattributes"

  # Remind to check the file
  msg_warn "Please review the changes to ensure they're appropriate for your project."

  return 0
}

# Main function
main()
{
  check_git_installed
  check_git_repo
  check_git_root
  check_gitattributes_exists
  check_gitattributes
}

main "$@"