path: root/langbreak

#!/usr/bin/env bash
#
# Language Breakdown 
# Displays a colored percentage bar showing programming language breakdown
# in the current directory, similar to GitHub/GitLab's linguist feature.
#
# Author: n0tori 
# Date: January 08, 2026
# License: GPLv3

# Color constants (ANSI 256-color codes)
readonly COLOR_RESET='\033[0m'
declare -A LANGUAGE_COLORS
declare -A LANGUAGE_BYTES
declare -A EXTENSION_MAP

#######################################
# Initialize language-to-color mapping (GitHub linguist colors)
# Globals:
#   LANGUAGE_COLORS
# Arguments:
#   None
#######################################
initialize_color_map() {
  # Map language names to ANSI 256-color codes
  LANGUAGE_COLORS["JavaScript"]="226"
  LANGUAGE_COLORS["Python"]="33"
  LANGUAGE_COLORS["TypeScript"]="69"
  LANGUAGE_COLORS["Shell"]="34"
  LANGUAGE_COLORS["Go"]="80"
  LANGUAGE_COLORS["Java"]="208"
  LANGUAGE_COLORS["Ruby"]="160"
  LANGUAGE_COLORS["C"]="244"
  LANGUAGE_COLORS["C++"]="205"
  LANGUAGE_COLORS["Rust"]="166"
  LANGUAGE_COLORS["PHP"]="62"
  LANGUAGE_COLORS["HTML"]="196"
  LANGUAGE_COLORS["CSS"]="92"
  LANGUAGE_COLORS["C#"]="28"
  LANGUAGE_COLORS["Kotlin"]="167"
  LANGUAGE_COLORS["Scala"]="196"
  LANGUAGE_COLORS["Swift"]="214"
  LANGUAGE_COLORS["Objective-C"]="69"
  LANGUAGE_COLORS["Perl"]="27"
  LANGUAGE_COLORS["Lua"]="27"
  LANGUAGE_COLORS["R"]="27"
  LANGUAGE_COLORS["Haskell"]="99"
  LANGUAGE_COLORS["Elixir"]="93"
  LANGUAGE_COLORS["Erlang"]="160"
  LANGUAGE_COLORS["Clojure"]="28"
  LANGUAGE_COLORS["YAML"]="250"
  LANGUAGE_COLORS["JSON"]="227"
  LANGUAGE_COLORS["TOML"]="172"
  LANGUAGE_COLORS["XML"]="172"
  LANGUAGE_COLORS["SQL"]="250"
  LANGUAGE_COLORS["PowerShell"]="27"
  LANGUAGE_COLORS["Vim Script"]="28"
  LANGUAGE_COLORS["Markdown"]="248"
}

#######################################
# Initialize file extension to language mapping
# Globals:
#   EXTENSION_MAP
# Arguments:
#   None
#######################################
initialize_extension_map() {
  # Map file extensions to language names
  EXTENSION_MAP["js"]="JavaScript"
  EXTENSION_MAP["jsx"]="JavaScript"
  EXTENSION_MAP["mjs"]="JavaScript"
  EXTENSION_MAP["py"]="Python"
  EXTENSION_MAP["ts"]="TypeScript"
  EXTENSION_MAP["tsx"]="TypeScript"
  EXTENSION_MAP["sh"]="Shell"
  EXTENSION_MAP["bash"]="Shell"
  EXTENSION_MAP["go"]="Go"
  EXTENSION_MAP["java"]="Java"
  EXTENSION_MAP["rb"]="Ruby"
  EXTENSION_MAP["c"]="C"
  EXTENSION_MAP["h"]="C"
  EXTENSION_MAP["cpp"]="C++"
  EXTENSION_MAP["hpp"]="C++"
  EXTENSION_MAP["cc"]="C++"
  EXTENSION_MAP["cxx"]="C++"
  EXTENSION_MAP["rs"]="Rust"
  EXTENSION_MAP["php"]="PHP"
  EXTENSION_MAP["html"]="HTML"
  EXTENSION_MAP["htm"]="HTML"
  EXTENSION_MAP["css"]="CSS"
  EXTENSION_MAP["cs"]="C#"
  EXTENSION_MAP["kt"]="Kotlin"
  EXTENSION_MAP["kts"]="Kotlin"
  EXTENSION_MAP["scala"]="Scala"
  EXTENSION_MAP["swift"]="Swift"
  EXTENSION_MAP["m"]="Objective-C"
  EXTENSION_MAP["mm"]="Objective-C"
  EXTENSION_MAP["pl"]="Perl"
  EXTENSION_MAP["pm"]="Perl"
  EXTENSION_MAP["lua"]="Lua"
  EXTENSION_MAP["r"]="R"
  EXTENSION_MAP["R"]="R"
  EXTENSION_MAP["hs"]="Haskell"
  EXTENSION_MAP["ex"]="Elixir"
  EXTENSION_MAP["exs"]="Elixir"
  EXTENSION_MAP["erl"]="Erlang"
  EXTENSION_MAP["hrl"]="Erlang"
  EXTENSION_MAP["clj"]="Clojure"
  EXTENSION_MAP["cljs"]="Clojure"
  EXTENSION_MAP["yml"]="YAML"
  EXTENSION_MAP["yaml"]="YAML"
  EXTENSION_MAP["json"]="JSON"
  EXTENSION_MAP["toml"]="TOML"
  EXTENSION_MAP["xml"]="XML"
  EXTENSION_MAP["sql"]="SQL"
  EXTENSION_MAP["ps1"]="PowerShell"
  EXTENSION_MAP["psm1"]="PowerShell"
  EXTENSION_MAP["vim"]="Vim Script"
  EXTENSION_MAP["md"]="Markdown"
  EXTENSION_MAP["markdown"]="Markdown"
}

#######################################
# Detect language from shebang 
# Arguments:
#   File path
# Outputs:
#   Language name or empty string
#######################################
detect_language_from_shebang() {
  local file="$1"
  local first_line
  first_line=$(head -n 1 "${file}" 2>/dev/null || echo "")

  if [[ "${first_line}" =~ ^#! ]]; then
    if [[ "${first_line}" =~ bash ]]; then
      echo "Shell"
    elif [[ "${first_line}" =~ python ]]; then
      echo "Python"
    elif [[ "${first_line}" =~ ruby ]]; then
      echo "Ruby"
    elif [[ "${first_line}" =~ node ]]; then
      echo "JavaScript"
    fi
  fi
}

#######################################
# Detect language from file
# Globals:
#   EXTENSION_MAP
# Arguments:
#   File path
# Outputs:
#   Language name or empty string
#######################################
detect_language() {
  local file="$1"
  local filename
  local extension

  filename=$(basename "${file}")

  if [[ "${filename}" == *.* ]]; then
    extension="${filename##*.}"
    if [[ -n "${EXTENSION_MAP[${extension}]:-}" ]]; then
      echo "${EXTENSION_MAP[${extension}]}"
      return 0
    fi
  fi

  # Fall back to shebang for extensionless files
  detect_language_from_shebang "${file}"
}

#######################################
# Count bytes for all files by language
# Globals:
#   LANGUAGE_BYTES
# Arguments:
#   None
#######################################
count_language_bytes() {
  local file
  local language
  local bytes

  # Find files with exclusions
  while IFS= read -r file; do
    language=$(detect_language "${file}")

    if [[ -n "${language}" ]]; then
      bytes=$(wc -c < "${file}" 2>/dev/null || echo "0")
      LANGUAGE_BYTES["${language}"]=$((${LANGUAGE_BYTES[${language}]:-0} + bytes))
    fi
  done < <(find . -type f \
    ! -path '*/\.*' \
    ! -path '*/node_modules/*' \
    ! -path '*/vendor/*' \
    ! -path '*/venv/*' \
    ! -path '*/__pycache__/*' \
    ! -path '*/dist/*' \
    ! -path '*/build/*' \
    ! -path '*/out/*' \
    ! -path '*/target/*')
}

#######################################
# Render the language breakdown bar and legend
# Globals:
#   LANGUAGE_BYTES
#   LANGUAGE_COLORS
#   COLOR_RESET
# Arguments:
#   None
#######################################
render_output() {
  local total_bytes=0
  local terminal_width
  local language
  local sorted_languages

  for language in "${!LANGUAGE_BYTES[@]}"; do
    total_bytes=$((total_bytes + LANGUAGE_BYTES[${language}]))
  done

  if [[ ${total_bytes} -eq 0 ]]; then
    echo "No recognized programming language files found."
    return 0
  fi

  terminal_width=$(tput cols 2>/dev/null || echo "80")

  # Sort languages by byte count (descending)
  sorted_languages=$(for lang in "${!LANGUAGE_BYTES[@]}"; do
    echo "${LANGUAGE_BYTES[${lang}]} ${lang}"
  done | sort -rn | awk '{print $2}')

  # Render the bar
  for language in ${sorted_languages}; do
    local bytes="${LANGUAGE_BYTES[${language}]}"
    local percentage=$((bytes * 100 / total_bytes))
    local bar_width=$((bytes * terminal_width / total_bytes))
    local color="${LANGUAGE_COLORS[${language}]:-7}"

    # Print colored bar segment
    printf "\033[48;5;%sm" "${color}"
    printf "%${bar_width}s" "" | tr ' ' '█'
    printf "${COLOR_RESET}"
  done
  echo ""

  # Render the legend
  for language in ${sorted_languages}; do
    local bytes="${LANGUAGE_BYTES[${language}]}"
    local percentage=$(awk "BEGIN {printf \"%.1f\", ${bytes} * 100.0 / ${total_bytes}}")
    local color="${LANGUAGE_COLORS[${language}]:-7}"

    printf "\033[38;5;%sm■\033[0m %-15s %6s%%\n" "${color}" "${language}" "${percentage}"
  done
}

#######################################
# Main function
# Arguments:
#   None
#######################################
main() {
  initialize_color_map
  initialize_extension_map
  count_language_bytes
  render_output
}

main "$@"