strip_php_comments()

Complexity

Conditions	40
Paths	531

Size

Total Lines	157
Code Lines	78

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

<?php
function strip_php_comments($code_str)
{
	// This is the faster, better way.
	if (is_callable('token_get_all'))
	{
		$tokens = token_get_all($code_str);

		$parts = array();
		foreach ($tokens as $token)
		{
			if (is_string($token))
				$parts[] = $token;
			else
			{
				list($id, $text) = $token;

				switch ($id) {
					case T_COMMENT:
					case T_DOC_COMMENT:
						end($parts);
						$prev_part = key($parts);

						// For the sake of tider output, trim any horizontal
						// whitespace that immediately preceded the comment.
						$parts[$prev_part] = rtrim($parts[$prev_part], "\t ");

						// For 'C' style comments, also trim one preceding
						// line break, if present.
						if (strpos($text, '/*') === 0)
						{
							if (substr($parts[$prev_part], -2) === "\r\n")
								$parts[$prev_part] = substr($parts[$prev_part], 0, -2);
							elseif (in_array(substr($parts[$prev_part], -1), array("\r", "\n")))
								$parts[$prev_part] = substr($parts[$prev_part], 0, -1);
						}

						break;

					default:
						$parts[] = $text;
						break;
				}
			}
		}

		$code_str = implode('', $parts);

		return $code_str;
	}

	// If the tokenizer extension has been disabled, do the job manually.

	// Leave any heredocs alone.
	if (preg_match_all('/<<<([\'"]?)(\w+)\1?\R(.*?)\R\h*\2;$/ms', $code_str, $matches))
	{
		$heredoc_replacements = array();

		foreach ($matches[0] as $mkey => $heredoc)
			$heredoc_replacements[$heredoc] = var_export(md5($matches[3][$mkey]), true) . ';';

		$code_str = strtr($code_str, $heredoc_replacements);
	}

	// Split before everything that could possibly delimit a comment or a string.
	$parts = preg_split('~(?=#+|/(?=/|\*)|\*/|\R|(?<!\\\)[\'"])~m', $code_str);

	$in_string = 0;
	$in_comment = 0;
	foreach ($parts as $partkey => $part)
	{
		$one_char = substr($part, 0, 1);
		$two_char = substr($part, 0, 2);
		$to_remove = 0;

		/*
		 * Meaning of $in_string values:
		 *	0: not in a string
		 *	1: in a single quote string
		 *	2: in a double quote string
		 */
		if ($one_char == "'")
		{
			if (!empty($in_comment))
				$in_string = 0;
			elseif (in_array($in_string, array(0, 1)))
				$in_string = ($in_string ^ 1);
		}
		elseif ($one_char == '"')
		{
			if (!empty($in_comment))
				$in_string = 0;
			elseif (in_array($in_string, array(0, 2)))
				$in_string = ($in_string ^ 2);
		}

		/*
		 * Meaning of $in_comment values:
		 * 	0: not in a comment
		 *	1: in a single line comment
		 *	2: in a multi-line comment
		 */
		elseif ($one_char == '#' || $two_char == '//')
		{
			$in_comment = !empty($in_string) ? 0 : (empty($in_comment) ? 1 : $in_comment);

			if ($in_comment == 1)
			{
				$parts[$partkey - 1] = rtrim($parts[$partkey - 1], "\t ");

				if (substr($parts[$partkey - 1], -2) === "\r\n")
					$parts[$partkey - 1] = substr($parts[$partkey - 1], 0, -2);
				elseif (in_array(substr($parts[$partkey - 1], -1), array("\r", "\n")))
					$parts[$partkey - 1] = substr($parts[$partkey - 1], 0, -1);
			}
		}
		elseif ($two_char === "\r\n" || $one_char === "\r" || $one_char === "\n")
		{
			if ($in_comment == 1)
				$in_comment = 0;
		}
		elseif ($two_char == '/*')
		{
			$in_comment = !empty($in_string) ? 0 : (empty($in_comment) ? 2 : $in_comment);

			if ($in_comment == 2)
			{
				$parts[$partkey - 1] = rtrim($parts[$partkey - 1], "\t ");

				if (substr($parts[$partkey - 1], -2) === "\r\n")
					$parts[$partkey - 1] = substr($parts[$partkey - 1], 0, -2);
				elseif (in_array(substr($parts[$partkey - 1], -1), array("\r", "\n")))
					$parts[$partkey - 1] = substr($parts[$partkey - 1], 0, -1);
			}
		}
		elseif ($two_char == '*/')
		{
			if ($in_comment == 2)
			{
				$in_comment = 0;

				// Delete the comment closing.
				$to_remove = 2;
			}
		}

		if (empty($in_comment))
			$parts[$partkey] = strlen($part) > $to_remove ? substr($part, $to_remove) : '';
		else
			$parts[$partkey] = '';
	}

	$code_str = implode('', $parts);

	if (!empty($heredoc_replacements))
		$code_str = strtr($code_str, array_flip($heredoc_replacements));

	return $code_str;
}
?>