<?php

define('API_PARSE_BEGIN', 0);
define('API_PARSE_AFTER_DOXYGEN', 1);
define('API_PARSE_AFTER_FUNCTION', 2);
define('API_PARSE_AFTER_DEFINE', 3);
define('API_PARSE_FUNCTION_ARGUMENTS_BEFORE', 4);
define('API_PARSE_FUNCTION_ARGUMENTS', 5);
define('API_PARSE_FUNCTION_BODY', 6);

function parse_string($string) {
  $string = preg_replace('#\r\n?#', '\n', $string);
  $state = API_PARSE_BEGIN;
  $last_doxygen = array();
  $objects = array();
  foreach (token_get_all($string) as $token) {
    if (is_array($token)) {
      list($type, $value, $line) = $token;
    }
    else {
      $type = NULL;
      $value = $token;
    }

    if ($type === T_WHITESPACE) {
      // New line handling allows one line break. The common case is just a
      // single line break so we optimize for that.
      if ($state === API_PARSE_AFTER_DOXYGEN && strlen($value) > 1 && substr_count($value, "\n") > 1) {
        _parse_finish_one($objects, $last_doxygen, $state);
      }
      continue;
    }
    if ($type === T_FUNCTION) {
      $state = API_PARSE_AFTER_FUNCTION;
      continue;
    }
    if ($value === 'define') {
      $state = API_PARSE_AFTER_DEFINE;
      continue;
    }

    switch ($state) {
      case API_PARSE_BEGIN:
        if ($type === T_DOC_COMMENT) {
          $last_doxygen = parse_doxygen($value);
          $last_doxygen['start'] = $line;
          $state = API_PARSE_AFTER_DOXYGEN;
        }
        elseif (!empty($last_doxygen)) {
          _parse_finish_one($objects, $last_doxygen, $state);
        }
        break;

      case API_PARSE_AFTER_FUNCTION:
        $last_doxygen += array(
          'type' => 'function',
          'start' => $line,
          'identifer' => $value,
          'arguments' => '',
        );
        $state = API_PARSE_FUNCTION_ARGUMENTS_BEFORE;
        break;
      case API_PARSE_FUNCTION_ARGUMENTS_BEFORE:
        if ($value == '(') {
          // Skipping the whitespace (automatically) and the first (
          // by using this state.
          $state = API_PARSE_FUNCTION_ARGUMENTS;
        }
        else {
          // The previous character was a &
          $last_doxygen['identifer'] .= $value;
        }
        break;
      case API_PARSE_FUNCTION_ARGUMENTS:
        if ($value == '{') {
          // cut off the last )
          $last_doxygen['arguments'] = substr($last_doxygen['arguments'], 0, -1);
          $state = API_PARSE_FUNCTION_BODY;
          $count = 1;
        }
        else {
          $last_doxygen['arguments'] .= $value;
        }
      case API_PARSE_AFTER_DEFINE:
        if ($type === T_CONSTANT_ENCAPSED_STRING) {
          $last_doxygen += array(
            'type' => 'constant',
            'start' => $line,
            'identifer' => trim($value, "'\""),
          );
          _parse_finish_one($objects, $last_doxygen, $state);
        }
        break;
      case API_PARSE_FUNCTION_BODY:
        if ($value == '{') {
          $count++;
        }
        if ($value == '}') {
          $count--;
        }
        if (!$count) {
          $last_doxygen['finish'] = $line;
          _parse_finish_one($objects, $last_doxygen, $state);
        }
        break;
      default:
        _parse_finish_one($objects, $last_doxygen, $state);
        break;
    }
  }
  return $objects;
}

function _parse_finish_one(&$objects, &$last_doxygen, &$state) {
  if ($last_doxygen) {
    $objects[] = $last_doxygen;
    $last_doxygen = array();
  }
  $state = API_PARSE_BEGIN;
}

function parse_doxygen($doxygen) {
  $in_description = FALSE;
  $data = array('summary' => '');
  $doxygen = substr($doxygen, 4, -2);
  $description = array();
  $data['description'] = &$description;
  $needs_identifier = array();
  foreach (explode("\n", $doxygen) as $line) {
    // Remove the spaces and the asterisk space from the beginning of the
    // line.
    $line = substr(trim($line), 2);
    if ($line[0] == '@') {
      // We do not handle these yet.
      if ($line[1] == '{' || $line[1] == '}') continue;
      list($command, $argument) = explode(' ', substr($line, 1), 2);
      switch ($command) {
        case 'file': case 'defgroup': case 'ingroup': case 'addgroup': case 'mainpage':
          list($argument, $summary) = explode(' ', $argument, 2);
          $data += array(
            'type' => $command,
            'identifer' => $argument,
            'summary' => $summary,
          );
          break;
        case 'see':
          $data['see'][] = $argument;
          break;
        case 'param':
          $description = implode(' ', $description);
          unset($description);
          $description = array();
          $data['param'][$argument] = &$description;
          $in_description = TRUE;
          break;
        case 'return':
          $description = implode(' ', $description);
          unset($description);
          $description = array();
          $data['return'] = array(
            'return_type' => $summary,
            'description' => &$description,
          );
          $in_description = TRUE;
          break;
      }
    }
    else {
      $line = trim($line);
      if ($line) {
        if ($in_description) {
          $description[] = $line;
        }
        else {
          $data['summary'] = $line;
          $in_description = TRUE;
        }
      }
    }
  }
  $description = implode(' ', $description);
  return $data;
}

function test() {
  $teststring = <<<__END__
  <?php

  /**
  * Testing constant.
  */
  define('CONST', 42);

  /**
  * Testing line break.
  */

  function foo() { }
__END__;
  $objects = parse_string($teststring);
  var_export($objects);
  var_dump($objects['foo'] == '');
  $teststring = <<<__END__
  <?php

  /**
  * Define.
  */
  define('CONST', 42);

  function undocumented() {
  }

  /**
  * Doxygen 1
  */
  junk
  /**
  * Doxygen 2
  */
  function foo() {}
__END__;
  $objects = parse_string($teststring);
  var_dump(!strpos($objects['foo'], '1'));
  var_dump(strpos($objects['foo'], '2'));
}

$objects = parse_string(file_get_contents('/var/www/drupal/includes/form.inc'));
var_export($objects);