sources/subs/Language.subs.php ➔ cleanLangString()

Complexity

Conditions	47
Paths	397

Size

Total Lines

202

Duplication

Lines	56
Ratio	27.72 %

Code Coverage

Tests	0
CRAP Score	2256

Importance

Changes

0

<?php
function cleanLangString($string, $to_display = true)
{
	// If going to display we make sure it doesn't have any HTML in it - etc.
	$new_string = '';
	if ($to_display)
	{
		// Are we in a string (0 = no, 1 = single quote, 2 = parsed)
		$in_string = 0;
		$is_escape = false;
		$str_len = strlen($string);
		for ($i = 0; $i < $str_len; $i++)
		{
			// Handle escapes first.
			if ($string[$i] == '\\')
			{
				// Toggle the escape.
				$is_escape = !$is_escape;

				// If we're now escaped don't add this string.
				if ($is_escape)
					continue;
			}
			// Special case - parsed string with line break etc?
			elseif (($string[$i] == 'n' || $string[$i] == 't') && $in_string == 2 && $is_escape)
			{
				// Put the escape back...
				$new_string .= $string[$i] == 'n' ? "\n" : "\t";
				$is_escape = false;
				continue;
			}
			// Have we got a single quote?
			elseif ($string[$i] == '\'')
			{
				// Already in a parsed string, or escaped in a linear string, means we print it - otherwise something special.
				if ($in_string != 2 && ($in_string != 1 || !$is_escape))
				{
					// Is it the end of a single quote string?
					if ($in_string == 1)
						$in_string = 0;
					// Otherwise it's the start!
					else
						$in_string = 1;

					// Don't actually include this character!
					continue;
				}
			}
			// Otherwise a double quote?
			elseif ($string[$i] == '"')
			{
				// Already in a single quote string, or escaped in a parsed string, means we print it - otherwise something special.
				if ($in_string != 1 && ($in_string != 2 || !$is_escape))
				{
					// Is it the end of a double quote string?
					if ($in_string == 2)
						$in_string = 0;
					// Otherwise it's the start!
					else
						$in_string = 2;

					// Don't actually include this character!
					continue;
				}
			}
			// A join/space outside of a string is simply removed.
			elseif ($in_string == 0 && (empty($string[$i]) || $string[$i] == '.'))
				continue;
			// Start of a variable?
			elseif ($in_string == 0 && $string[$i] == '$')
			{
				// Find the whole of it!
				preg_match('~([\$A-Za-z0-9\'\[\]_-]+)~', substr($string, $i), $matches);
				if (!empty($matches[1]))
				{
					// Come up with some pseudo thing to indicate this is a var.
					// @todo Do better than this, please!
					$new_string .= '{%' . $matches[1] . '%}';

					// We're not going to re-parse this.
					$i += strlen($matches[1]) - 1;
				}

				continue;
			}
			// Right, if we're outside of a string we have DANGER, DANGER!
			elseif ($in_string == 0)
			{
				continue;
			}

			// Actually add the character to the string!
			$new_string .= $string[$i];

			// If anything was escaped it ain't any longer!
			$is_escape = false;
		}

		// Un-html then re-html the whole thing!
		$new_string = \ElkArte\Util::htmlspecialchars(un_htmlspecialchars($new_string));
	}
	else
	{
		// Keep track of what we're doing...
		$in_string = 0;

		// This is for deciding whether to HTML a quote.
		$in_html = false;
		$str_len = strlen($string);
		for ($i = 0; $i < $str_len; $i++)
		{
			// We don't do parsed strings apart from for breaks.
			if ($in_string == 2)
			{
				$in_string = 0;
				$new_string .= '"';
			}

			// Not in a string yet?
			if ($in_string != 1)
			{
				$in_string = 1;
				$new_string .= ($new_string ? ' . ' : '') . '\'';
			}

			// Is this a variable?
			if ($string[$i] == '{' && $string[$i + 1] == '%' && $string[$i + 2] == '$')
			{
				// Grab the variable.
				preg_match('~\{%([\$A-Za-z0-9\'\[\]_-]+)%\}~', substr($string, $i), $matches);
				if (!empty($matches[1]))
				{
					if ($in_string == 1)
						$new_string .= '\' . ';
					elseif ($new_string)
						$new_string .= ' . ';

					$new_string .= $matches[1];
					$i += strlen($matches[1]) + 3;
					$in_string = 0;
				}

				continue;
			}
			// Is this a lt sign?
			elseif ($string[$i] == '<')
			{
				// Probably HTML?
				if ($string[$i + 1] != ' ')
					$in_html = true;
				// Assume we need an entity...
				else
				{
					$new_string .= '&lt;';
					continue;
				}
			}
			// What about gt?
			elseif ($string[$i] == '>')
			{
				// Will it be HTML?
				if ($in_html)
					$in_html = false;
				// Otherwise we need an entity...
				else
				{
					$new_string .= '&gt;';
					continue;
				}
			}
			// Is it a slash? If so escape it...
			if ($string[$i] == '\\')
				$new_string .= '\\';
			// The infamous double quote?
			elseif ($string[$i] == '"')
			{
				// If we're in HTML we leave it as a quote - otherwise we entity it.
				if (!$in_html)
				{
					$new_string .= '&quot;';
					continue;
				}
			}
			// A single quote?
			elseif ($string[$i] == '\'')
			{
				// Must be in a string so escape it.
				$new_string .= '\\';
			}

			// Finally add the character to the string!
			$new_string .= $string[$i];
		}

		// If we ended as a string then close it off.
		if ($in_string == 1)
			$new_string .= '\'';
		elseif ($in_string == 2)
			$new_string .= '"';
	}

	return $new_string;
}