Improve byte-safety

5 files changed, 236 insertions, 41 deletions

diff --git a/system/core/Log.php b/system/core/Log.php
index 986121526..cf6c75a95 100644
--- a/system/core/Log.php
+++ b/system/core/Log.php
@@ -104,6 +104,13 @@ class CI_Log {
 	 */
 	protected $_levels = array('ERROR' => 1, 'DEBUG' => 2, 'INFO' => 3, 'ALL' => 4);
 
+	/**
+	 * mbstring.func_override flag
+	 *
+	 * @var	bool
+	 */
+	protected static $func_override;
+
 	// --------------------------------------------------------------------
 
 	/**
@@ -115,6 +122,8 @@ class CI_Log {
 	{
 		$config =& get_config();
 
+		isset(self::$func_override) OR self::$func_override = (extension_loaded('mbstring') && ini_get('mbstring.func_override'));
+
 		$this->_log_path = ($config['log_path'] !== '') ? $config['log_path'] : APPPATH.'logs/';
 		$this->_file_ext = (isset($config['log_file_extension']) && $config['log_file_extension'] !== '')
 			? ltrim($config['log_file_extension'], '.') : 'php';
@@ -208,9 +217,9 @@ class CI_Log {
 
 		$message .= $this->_format_line($level, $date, $msg);
 
-		for ($written = 0, $length = strlen($message); $written < $length; $written += $result)
+		for ($written = 0, $length = self::strlen($message); $written < $length; $written += $result)
 		{
-			if (($result = fwrite($fp, substr($message, $written))) === FALSE)
+			if (($result = fwrite($fp, self::substr($message, $written))) === FALSE)
 			{
 				break;
 			}
@@ -244,4 +253,44 @@ class CI_Log {
 	{
 		return $level.' - '.$date.' --> '.$message."\n";
 	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe strlen()
+	 *
+	 * @param	string	$str
+	 * @return	int
+	 */
+	protected static function strlen($str)
+	{
+		return (self::$func_override)
+			? mb_strlen($str, '8bit')
+			: strlen($str);
+	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe substr()
+	 *
+	 * @param	string	$str
+	 * @param	int	$start
+	 * @param	int	$length
+	 * @return	string
+	 */
+	protected static function substr($str, $start, $length = NULL)
+	{
+		if (self::$func_override)
+		{
+			// mb_substr($str, $start, null, '8bit') returns an empty
+			// string on PHP 5.3
+			isset($length) OR $length = ($start >= 0 ? self::strlen($str) - $start : -$start);
+			return mb_substr($str, $start, $length, '8bit');
+		}
+
+		return isset($length)
+			? substr($str, $start, $length)
+			: substr($str, $start);
+	}
 }
diff --git a/system/core/Output.php b/system/core/Output.php
index 06ff1011c..cf6510ff1 100644
--- a/system/core/Output.php
+++ b/system/core/Output.php
@@ -123,6 +123,13 @@ class CI_Output {
 	public $parse_exec_vars = TRUE;
 
 	/**
+	 * mbstring.func_override flag
+	 *
+	 * @var	bool
+	 */
+	protected static $func_override;
+
+	/**
 	 * Class constructor
 	 *
 	 * Determines whether zLib output compression will be used.
@@ -138,6 +145,8 @@ class CI_Output {
 			&& extension_loaded('zlib')
 		);
 
+		isset(self::$func_override) OR self::$func_override = (extension_loaded('mbstring') && ini_get('mbstring.func_override'));
+
 		// Get mime types for later
 		$this->mimes =& get_mimes();
 
@@ -304,9 +313,9 @@ class CI_Output {
 
 		for ($i = 0, $c = count($headers); $i < $c; $i++)
 		{
-			if (strncasecmp($header, $headers[$i], $l = strlen($header)) === 0)
+			if (strncasecmp($header, $headers[$i], $l = self::strlen($header)) === 0)
 			{
-				return trim(substr($headers[$i], $l+1));
+				return trim(self::substr($headers[$i], $l+1));
 			}
 		}
 
@@ -480,13 +489,13 @@ class CI_Output {
 				if (isset($_SERVER['HTTP_ACCEPT_ENCODING']) && strpos($_SERVER['HTTP_ACCEPT_ENCODING'], 'gzip') !== FALSE)
 				{
 					header('Content-Encoding: gzip');
-					header('Content-Length: '.strlen($output));
+					header('Content-Length: '.self::strlen($output));
 				}
 				else
 				{
 					// User agent doesn't support gzip compression,
 					// so we'll have to decompress our cache
-					$output = gzinflate(substr($output, 10, -8));
+					$output = gzinflate(self::substr($output, 10, -8));
 				}
 			}
 
@@ -601,9 +610,9 @@ class CI_Output {
 
 			$output = $cache_info.'ENDCI--->'.$output;
 
-			for ($written = 0, $length = strlen($output); $written < $length; $written += $result)
+			for ($written = 0, $length = self::strlen($output); $written < $length; $written += $result)
 			{
-				if (($result = fwrite($fp, substr($output, $written))) === FALSE)
+				if (($result = fwrite($fp, self::substr($output, $written))) === FALSE)
 				{
 					break;
 				}
@@ -711,7 +720,7 @@ class CI_Output {
 		}
 
 		// Display the cache
-		$this->_display(substr($cache, strlen($match[0])));
+		$this->_display(self::substr($cache, self::strlen($match[0])));
 		log_message('debug', 'Cache file is current. Sending it to browser.');
 		return TRUE;
 	}
@@ -797,4 +806,43 @@ class CI_Output {
 		}
 	}
 
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe strlen()
+	 *
+	 * @param	string	$str
+	 * @return	int
+	 */
+	protected static function strlen($str)
+	{
+		return (self::$func_override)
+			? mb_strlen($str, '8bit')
+			: strlen($str);
+	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe substr()
+	 *
+	 * @param	string	$str
+	 * @param	int	$start
+	 * @param	int	$length
+	 * @return	string
+	 */
+	protected static function substr($str, $start, $length = NULL)
+	{
+		if (self::$func_override)
+		{
+			// mb_substr($str, $start, null, '8bit') returns an empty
+			// string on PHP 5.3
+			isset($length) OR $length = ($start >= 0 ? self::strlen($str) - $start : -$start);
+			return mb_substr($str, $start, $length, '8bit');
+		}
+
+		return isset($length)
+			? substr($str, $start, $length)
+			: substr($str, $start);
+	}
 }
diff --git a/system/libraries/Email.php b/system/libraries/Email.php
index 7f49c1b3d..676bbcafb 100644
--- a/system/libraries/Email.php
+++ b/system/libraries/Email.php
@@ -374,6 +374,13 @@ class CI_Email {
 		5 => '5 (Lowest)'
 	);
 
+	/**
+	 * mbstring.func_override flag
+	 *
+	 * @var	bool
+	 */
+	protected static $func_override;
+
 	// --------------------------------------------------------------------
 
 	/**
@@ -390,6 +397,8 @@ class CI_Email {
 		$this->initialize($config);
 		$this->_safe_mode = ( ! is_php('5.4') && ini_get('safe_mode'));
 
+		isset(self::$func_override) OR self::$func_override = (extension_loaded('mbstring') && ini_get('mbstring.func_override'));
+
 		log_message('info', 'Email Class Initialized');
 	}
 
@@ -1037,7 +1046,7 @@ class CI_Email {
 	{
 		if (function_exists('idn_to_ascii') && $atpos = strpos($email, '@'))
 		{
-			$email = substr($email, 0, ++$atpos).idn_to_ascii(substr($email, $atpos));
+			$email = self::substr($email, 0, ++$atpos).idn_to_ascii(self::substr($email, $atpos));
 		}
 
 		return (bool) filter_var($email, FILTER_VALIDATE_EMAIL);
@@ -1154,7 +1163,7 @@ class CI_Email {
 		{
 			// Is the line within the allowed character count?
 			// If so we'll join it to the output and continue
-			if (mb_strlen($line) <= $charlim)
+			if (self::strlen($line) <= $charlim)
 			{
 				$output .= $line.$this->newline;
 				continue;
@@ -1170,10 +1179,10 @@ class CI_Email {
 				}
 
 				// Trim the word down
-				$temp .= mb_substr($line, 0, $charlim - 1);
-				$line = mb_substr($line, $charlim - 1);
+				$temp .= self::substr($line, 0, $charlim - 1);
+				$line = self::substr($line, $charlim - 1);
 			}
-			while (mb_strlen($line) > $charlim);
+			while (self::strlen($line) > $charlim);
 
 			// If $temp contains data it means we had to split up an over-length
 			// word into smaller chunks so we'll add it back to our current line
@@ -1385,7 +1394,7 @@ class CI_Email {
 					$this->_header_str .= $hdr;
 				}
 
-				strlen($body) && $body .= $this->newline.$this->newline;
+				self::strlen($body) && $body .= $this->newline.$this->newline;
 				$body .= $this->_get_mime_message().$this->newline.$this->newline
 					.'--'.$last_boundary.$this->newline
 
@@ -1532,7 +1541,7 @@ class CI_Email {
 
 		foreach (explode("\n", $str) as $line)
 		{
-			$length = strlen($line);
+			$length = self::strlen($line);
 			$temp = '';
 
 			// Loop through each character in the line to add soft-wrap
@@ -1567,7 +1576,7 @@ class CI_Email {
 
 				// If we're at the character limit, add the line to the output,
 				// reset our temp variable, and keep on chuggin'
-				if ((strlen($temp) + strlen($char)) >= 76)
+				if ((self::strlen($temp) + self::strlen($char)) >= 76)
 				{
 					$output .= $temp.$escape.$this->crlf;
 					$temp = '';
@@ -1582,7 +1591,7 @@ class CI_Email {
 		}
 
 		// get rid of extra CRLF tacked onto the end
-		return substr($output, 0, strlen($this->crlf) * -1);
+		return self::substr($output, 0, self::strlen($this->crlf) * -1);
 	}
 
 	// --------------------------------------------------------------------
@@ -1624,7 +1633,7 @@ class CI_Email {
 					// iconv_mime_encode() will always put a header field name.
 					// We've passed it an empty one, but it still prepends our
 					// encoded string with ': ', so we need to strip it.
-					return substr($output, 2);
+					return self::substr($output, 2);
 				}
 
 				$chars = iconv_strlen($str, 'UTF-8');
@@ -1636,10 +1645,10 @@ class CI_Email {
 		}
 
 		// We might already have this set for UTF-8
-		isset($chars) OR $chars = strlen($str);
+		isset($chars) OR $chars = self::strlen($str);
 
 		$output = '=?'.$this->charset.'?Q?';
-		for ($i = 0, $length = strlen($output); $i < $chars; $i++)
+		for ($i = 0, $length = self::strlen($output); $i < $chars; $i++)
 		{
 			$chr = ($this->charset === 'UTF-8' && ICONV_ENABLED === TRUE)
 				? '='.implode('=', str_split(strtoupper(bin2hex(iconv_substr($str, $i, 1, $this->charset))), 2))
@@ -1647,11 +1656,11 @@ class CI_Email {
 
 			// RFC 2045 sets a limit of 76 characters per line.
 			// We'll append ?= to the end of each line though.
-			if ($length + ($l = strlen($chr)) > 74)
+			if ($length + ($l = self::strlen($chr)) > 74)
 			{
 				$output .= '?='.$this->crlf // EOL
 					.' =?'.$this->charset.'?Q?'.$chr; // New line
-				$length = 6 + strlen($this->charset) + $l; // Reset the length for the new line
+				$length = 6 + self::strlen($this->charset) + $l; // Reset the length for the new line
 			}
 			else
 			{
@@ -1744,14 +1753,14 @@ class CI_Email {
 
 			if ($i === $float)
 			{
-				$chunk[] = substr($set, 1);
+				$chunk[] = self::substr($set, 1);
 				$float += $this->bcc_batch_size;
 				$set = '';
 			}
 
 			if ($i === $c-1)
 			{
-				$chunk[] = substr($set, 1);
+				$chunk[] = self::substr($set, 1);
 			}
 		}
 
@@ -2109,7 +2118,7 @@ class CI_Email {
 
 		$this->_debug_msg[] = '<pre>'.$cmd.': '.$reply.'</pre>';
 
-		if ((int) substr($reply, 0, 3) !== $resp)
+		if ((int) self::substr($reply, 0, 3) !== $resp)
 		{
 			$this->_set_error_message('lang:email_smtp_error', $reply);
 			return FALSE;
@@ -2196,9 +2205,9 @@ class CI_Email {
 	protected function _send_data($data)
 	{
 		$data .= $this->newline;
-		for ($written = $timestamp = 0, $length = strlen($data); $written < $length; $written += $result)
+		for ($written = $timestamp = 0, $length = self::strlen($data); $written < $length; $written += $result)
 		{
-			if (($result = fwrite($this->_smtp_connect, substr($data, $written))) === FALSE)
+			if (($result = fwrite($this->_smtp_connect, self::substr($data, $written))) === FALSE)
 			{
 				break;
 			}
@@ -2382,4 +2391,44 @@ class CI_Email {
 	{
 		is_resource($this->_smtp_connect) && $this->_send_command('quit');
 	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe strlen()
+	 *
+	 * @param	string	$str
+	 * @return	int
+	 */
+	protected static function strlen($str)
+	{
+		return (self::$func_override)
+			? mb_strlen($str, '8bit')
+			: strlen($str);
+	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe substr()
+	 *
+	 * @param	string	$str
+	 * @param	int	$start
+	 * @param	int	$length
+	 * @return	string
+	 */
+	protected static function substr($str, $start, $length = NULL)
+	{
+		if (self::$func_override)
+		{
+			// mb_substr($str, $start, null, '8bit') returns an empty
+			// string on PHP 5.3
+			isset($length) OR $length = ($start >= 0 ? self::strlen($str) - $start : -$start);
+			return mb_substr($str, $start, $length, '8bit');
+		}
+
+		return isset($length)
+			? substr($str, $start, $length)
+			: substr($str, $start);
+	}
 }
diff --git a/system/libraries/Zip.php b/system/libraries/Zip.php
index 140ad7212..25315c92e 100644
--- a/system/libraries/Zip.php
+++ b/system/libraries/Zip.php
@@ -106,12 +106,21 @@ class CI_Zip {
 	public $compression_level = 2;
 
 	/**
+	 * mbstring.func_override flag
+	 *
+	 * @var	bool
+	 */
+	protected static $func_override;
+
+	/**
 	 * Initialize zip compression class
 	 *
 	 * @return	void
 	 */
 	public function __construct()
 	{
+		isset(self::$func_override) OR self::$func_override = (extension_loaded('mbstring') && ini_get('mbstring.func_override'));
+
 		$this->now = time();
 		log_message('info', 'Zip Compression Class Initialized');
 	}
@@ -182,7 +191,7 @@ class CI_Zip {
 			.pack('V', 0) // crc32
 			.pack('V', 0) // compressed filesize
 			.pack('V', 0) // uncompressed filesize
-			.pack('v', strlen($dir)) // length of pathname
+			.pack('v', self::strlen($dir)) // length of pathname
 			.pack('v', 0) // extra field length
 			.$dir
 			// below is "data descriptor" segment
@@ -197,7 +206,7 @@ class CI_Zip {
 			.pack('V',0) // crc32
 			.pack('V',0) // compressed filesize
 			.pack('V',0) // uncompressed filesize
-			.pack('v', strlen($dir)) // length of pathname
+			.pack('v', self::strlen($dir)) // length of pathname
 			.pack('v', 0) // extra field length
 			.pack('v', 0) // file comment length
 			.pack('v', 0) // disk number start
@@ -206,7 +215,7 @@ class CI_Zip {
 			.pack('V', $this->offset) // relative offset of local header
 			.$dir;
 
-		$this->offset = strlen($this->zipdata);
+		$this->offset = self::strlen($this->zipdata);
 		$this->entries++;
 	}
 
@@ -255,10 +264,10 @@ class CI_Zip {
 	{
 		$filepath = str_replace('\\', '/', $filepath);
 
-		$uncompressed_size = strlen($data);
+		$uncompressed_size = self::strlen($data);
 		$crc32  = crc32($data);
-		$gzdata = substr(gzcompress($data, $this->compression_level), 2, -4);
-		$compressed_size = strlen($gzdata);
+		$gzdata = self::substr(gzcompress($data, $this->compression_level), 2, -4);
+		$compressed_size = self::strlen($gzdata);
 
 		$this->zipdata .=
 			"\x50\x4b\x03\x04\x14\x00\x00\x00\x08\x00"
@@ -267,7 +276,7 @@ class CI_Zip {
 			.pack('V', $crc32)
 			.pack('V', $compressed_size)
 			.pack('V', $uncompressed_size)
-			.pack('v', strlen($filepath)) // length of filename
+			.pack('v', self::strlen($filepath)) // length of filename
 			.pack('v', 0) // extra field length
 			.$filepath
 			.$gzdata; // "file data" segment
@@ -279,7 +288,7 @@ class CI_Zip {
 			.pack('V', $crc32)
 			.pack('V', $compressed_size)
 			.pack('V', $uncompressed_size)
-			.pack('v', strlen($filepath)) // length of filename
+			.pack('v', self::strlen($filepath)) // length of filename
 			.pack('v', 0) // extra field length
 			.pack('v', 0) // file comment length
 			.pack('v', 0) // disk number start
@@ -288,7 +297,7 @@ class CI_Zip {
 			.pack('V', $this->offset) // relative offset of local header
 			.$filepath;
 
-		$this->offset = strlen($this->zipdata);
+		$this->offset = self::strlen($this->zipdata);
 		$this->entries++;
 		$this->file_num++;
 	}
@@ -401,8 +410,8 @@ class CI_Zip {
 			.$this->directory."\x50\x4b\x05\x06\x00\x00\x00\x00"
 			.pack('v', $this->entries) // total # of entries "on this disk"
 			.pack('v', $this->entries) // total # of entries overall
-			.pack('V', strlen($this->directory)) // size of central dir
-			.pack('V', strlen($this->zipdata)) // offset to start of central dir
+			.pack('V', self::strlen($this->directory)) // size of central dir
+			.pack('V', self::strlen($this->zipdata)) // offset to start of central dir
 			."\x00\x00"; // .zip file comment length
 	}
 
@@ -425,9 +434,9 @@ class CI_Zip {
 
 		flock($fp, LOCK_EX);
 
-		for ($result = $written = 0, $data = $this->get_zip(), $length = strlen($data); $written < $length; $written += $result)
+		for ($result = $written = 0, $data = $this->get_zip(), $length = self::strlen($data); $written < $length; $written += $result)
 		{
-			if (($result = fwrite($fp, substr($data, $written))) === FALSE)
+			if (($result = fwrite($fp, self::substr($data, $written))) === FALSE)
 			{
 				break;
 			}
@@ -481,4 +490,43 @@ class CI_Zip {
 		return $this;
 	}
 
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe strlen()
+	 *
+	 * @param	string	$str
+	 * @return	int
+	 */
+	protected static function strlen($str)
+	{
+		return (self::$func_override)
+			? mb_strlen($str, '8bit')
+			: strlen($str);
+	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * Byte-safe substr()
+	 *
+	 * @param	string	$str
+	 * @param	int	$start
+	 * @param	int	$length
+	 * @return	string
+	 */
+	protected static function substr($str, $start, $length = NULL)
+	{
+		if (self::$func_override)
+		{
+			// mb_substr($str, $start, null, '8bit') returns an empty
+			// string on PHP 5.3
+			isset($length) OR $length = ($start >= 0 ? self::strlen($str) - $start : -$start);
+			return mb_substr($str, $start, $length, '8bit');
+		}
+
+		return isset($length)
+			? substr($str, $start, $length)
+			: substr($str, $start);
+	}
 }
diff --git a/user_guide_src/source/changelog.rst b/user_guide_src/source/changelog.rst
index 0a8160acb..4be0b31d3 100644
--- a/user_guide_src/source/changelog.rst
+++ b/user_guide_src/source/changelog.rst
@@ -21,6 +21,7 @@ Bug fixes for 3.1.2
 -  Fixed a regression (#4874) - :doc:`Session Library <libraries/sessions>` didn't take into account ``session.hash_bits_per_character`` when validating session IDs.
 -  Fixed a bug (#4871) - :doc:`Query Builder <database/query_builder>` method ``update_batch()`` didn't properly handle identifier escaping.
 -  Fixed a bug (#4884) - :doc:`Query Builder <database/query_builder>` didn't properly parse field names ending in 'is' when used inside WHERE and HAVING statements.
+-  Fixed a bug where ``CI_Log``, ``CI_Output``, ``CI_Email`` and ``CI_Zip`` didn't handle strings in a byte-safe manner when ``mbstring.func_override`` is enabled.
 
 Version 3.1.1
 =============