Index of PHP tokens for Emacs and beyond

By Troels Knak-Nielsen

I’ve been using Emacs as my primary editor for a while now. A lot of people prefer IDE’s, but I’ve never been comfortable with them. I kind of like the ability to show a list of classes & functions in a file though. Emacs can use ctags to generate a list of tokens for a file, but I weren’t really satisfied with its output.
As you probably know, PHP has a ridiculous amount of functions for all kinds of things, and as it happens, token_get_all gives access to the Zend Engine tokenizer. In other words, the same chunk of code, which PHP itself uses, when reading a .php file. This provides an excellent base for writing a script, which can parse the socks off ctags.

So without further ado: ~/scripts/tokens.php parses a file and prints the classes and functions found in there. It also prints a small excerpt of any docblocks, which precedes the item. This makes for a nice table of contents.

if (!defined('T_UNSPECIFIED_STRING')) {
  define('T_UNSPECIFIED_STRING', -1);
function token_get_all_improved($data) {
  $tokens = array();
  $line = 1;
  $col = 0;
  $level = 0;
  $scope_level = null;
  $in_scope = false;
  foreach (token_get_all($data) as $token) {
    if (is_array($token)) {
      list ($token, $text) = $token;
    } else if (is_string($token)) {
      $text = $token;
      $token = T_UNSPECIFIED_STRING;
    if ($token === T_CURLY_OPEN || $token === T_DOLLAR_OPEN_CURLY_BRACES || $text === '{') {
      if (is_null($scope_level)) {
        $scope_level = $level;
    } else if ($text == '}') {
      if ($in_scope && $level < $scope_level) {
        $in_scope = false;
    $tmp = $text;
    $num_new_lines = substr_count($tmp, "n");
    if (1 <= $num_new_lines) {
      $line += $num_new_lines;
      $col =  1;
      $tmp = substr($tmp, strrpos($tmp, "n") + 1);
      if ($tmp === false) {
        $tmp = '';
    $col += strlen($tmp);

    if ($token === T_INTERFACE || $token === T_CLASS) {
      $in_scope = true;
      $scope_level = null;

    $xtoken = new StdClass();
    $xtoken->type = $token;
    $xtoken->text = $text;
    $xtoken->line = $line;
    $xtoken->col = $col;
    $xtoken->blockLevel = $level;
    $xtoken->isClassScope = $in_scope && !is_null($scope_level);
    $tokens[] = $xtoken;
  return $tokens;

function docblock_excerpt($str) {
  if (preg_match('~*{2}[sn*]+(.*)~', trim($str, '/'), $matches)) {
    return $matches[1];

function transform_tokens_to_list($tokens) {
  $buffer = null;
  $docblock = null;
  $results = array();
  $access = '+';
  foreach ($tokens as $token) {
    switch ($token->type) {
    case T_PUBLIC:
      $access = '+';
    case T_PRIVATE:
      $access = '-';
    case T_PROTECTED:
      $access = '#';
    case T_DOC_COMMENT:
      $docblock = $token->text;
    case T_INTERFACE:
    case T_CLASS:
    case T_FUNCTION:
      $buffer = $token;
      $buffer->access = $access;
      $access = '+';
    case T_STRING:
      if (!is_null($buffer)) {
        $buffer->isMember = $buffer->isClassScope || ($buffer->type != T_FUNCTION);
        $buffer->docblock = $docblock;
        $buffer->name = $token->text;
        $results[] = $buffer;
        $buffer = null;
        $docblock = null;
  return $results;

function list_to_table($results) {
  $view = array();
  $last = null;
  foreach ($results as $token) {
    if ($last && ((!$token->isMember && $last->isMember) || (in_array($token->type, array(T_INTERFACE, T_CLASS))))) {
      $view[] = array('', '', '', '', '');
    $last = $token;

    $view[] = array(
      $token->isMember ? $token->access : '',
      strtolower(str_replace('T_', '', token_name($token->type))),
  return $view;

function format_table($map) {
  $out = array();
  $column_widths = array_fill(0, count($map[0]), 0);
  foreach ($map as $row) {
    foreach ($row as $num => $col) {
      $column_widths[$num] = max($column_widths[$num], strlen($col));
  foreach ($map as $row) {
    $line = '';
    foreach ($row as $num => $col) {
      $line .= str_pad($col, $column_widths[$num] + 2);
    $out[] = trim($line);
  return implode("n", $out);

            isset($argv[1]) ? $argv[1] : 'php://stdin')))))
  . "n");

You can use the script alone, but I have also written an accompanying emacs-mode, which loads the result into a buffer and makes each line “clickable”, so you can jump to the relevant position in the file. This could probably be improved on — I’m still a novice, when it comes to lisp. Save it as ~/site-lisp/php-tokens.el (Or where your custom elisp scripts reside).

(defvar php-tokens-mode-target-buffer nil)
(defun php-tokens-mode-find-occurrence ()
  (let ((pos (get-text-property (point) 'occur-target)))
    (when pos
      (unless (buffer-live-p (get-buffer php-tokens-mode-target-buffer))
        (error "Buffer for this occurrence was killed"))

(defun php-tokens-mode-goto-occurrence (&optional event)
  "Go to the occurrence the current line describes."
  (interactive (list last-nonmenu-event))
  (let ((pos
         (if (null event)
             ;; Actually `event-end' works correctly with a nil argument as
             ;; well, so we could dispense with this test, but let's not
             ;; rely on this undocumented behavior.
           (with-current-buffer (window-buffer (posn-window (event-end event)))
               (goto-char (posn-point (event-end event)))
    (pop-to-buffer php-tokens-mode-target-buffer)
    (goto-line pos)))

(defun php-tokens-mode (buffer-name regexp)
  (pop-to-buffer buffer-name)
  (setq buffer-read-only nil)
  (let ((matchbeg 0)
        (origpt nil)
        (begpt nil)
        (contloop t)
        (endpt nil))
      (goto-char (point-min)) ;; begin searching in the buffer
      (while (and contloop (not (eobp)))
        (setq origpt (point))
        (if (setq endpt (re-search-forward regexp nil t))
            (add-text-properties (line-beginning-position) (line-end-position)
                                  `(mouse-face (highlight))
                                  `(occur-target ,(string-to-number (match-string 0)))))
          (setq contloop nil)))))
  (local-set-key '[return] 'php-tokens-mode-goto-occurrence)
  (local-set-key '[mouse-1] 'php-tokens-mode-goto-occurrence))

(defgroup php-tokens nil
  "Generates an overview of PHP tokens"
  :group 'convenience)

(defun php-tokens ()
  "Lists tokens for a PHP-buffer"
  (let* ((source-buffername (buffer-name))
         (buffername "*php-tokens*")
         (contents (buffer-substring-no-properties (point-min) (point-max))))
    (when (get-buffer buffername)
      (kill-buffer buffername))
      (pop-to-buffer buffername)
        (insert contents)
        (shell-command-on-region (point-min) (point-max)
                                 (format "php %s" (shell-quote-argument (expand-file-name "~/scripts/tokens.php")))
                                 buffername nil))
      (setq php-tokens-mode-target-buffer source-buffername)
      (php-tokens-mode buffername "^\([0-9]+\)[ ]+"))))

(provide 'php-tokens)

To use it, you must bind the function to a key. Make the following modifications to your .emacs file:

(require 'php-tokens)
(global-set-key '[f7] 'php-tokens)

    Aha! You right. I get idea from you. Thanks

  • pachanga

    I wonder how can this be used in vim ;)

  • kyberfabrikken

    The script takes either a filename as argument or contents as input (over stdin).
    You can invoke an external command in vim, using:

    :! php ~/scripts/tokens.php %

    But this will only work on the current file, so you have to save it first. I’m sure there’s a way to pipe the buffer to stdin of an external command somehow, but as you might have guessed, I don’t use vim that often.

  • Mike Seth

    or you could just use ctags and the native tag jump facility



Learn Coding Online
Learn Web Development

Start learning web development and design for free with SitePoint Premium!

Get the latest in Front-end, once a week, for free.