Some problems cannot be repeated. For example, gbk is converted to utf8, and then converted to utf8. This will cause garbled characters. I originally tried to detect the encoding before conversion, but it seemed to have failed. I specifically tried a file and checked whether it was gbk or utf-8, and both returned true. I don’t understand this.
/**
* Convert file encoding
* Dependent extensions filesystem and mbstring
* @example
*
<br> * include_once 'ConvertEncode.php';<br> * $convert = new ConvertEncode();<br> * try{<br> * $convert->setPath('my', true, true);//Directory<br> * //$convert->setPath('my.php ');//Single file<br> * $convert->setEncode('GBK', 'UTF-8');<br> * $convert->convert();<br> * }catch(ConvertException $e) {<br> * echo $e->getMessage();<br> * }<br> *
*/
class ConvertEncode {
/**
* The encoding to be converted to
* @var string
*/
private $_to_encoding;
/**
* Encoding before conversion
* @var string
*/
private $_from_encoding;
/**
* Directory or single file to be converted
* @var string
*/
private $_path;
/**
* Whether it is a directory, it is only set when the directory is given.
* @var boolean
*/
private $_directory;
/**
* Whether to traverse recursively, only valid for directories
* @var boolean
*/
private $_recursion;
/**
* Save all files to be converted, only used when converting files in the directory
* @var array
*/
private $_files = array();
/**
* Constructor
*/
public function __construct() {
if( ! function_exists('mb_convert_encoding') ) {
throw new ConvertException('mbstring extension be required');
}
}
/**
* Set the directory or single file to be converted
* @param string $path directory or file
* @param boolean whether it is a directory
* @param boolean whether it is a recursive directory
* @return boolean
*/
public function setPath($path, $is_dir = false, $rec = false) {
$this->_path = $path;
$this->_directory = $is_dir;
$this->_recursion = $rec;
return true;
}
/**
* Set the encoding before conversion and the encoding to be converted to
* @param string $encode The encoding before conversion
* @param string $encode The encoding to be converted
* @return boolean
*/
public function setEncode($encode_from, $encode_to) {
$this->_from_encoding = $encode_from;
$this->_to_encoding = $encode_to;
return true;
}
/**
* Convert encoding, convert separately according to whether it is a directory setting
* @return boolean
*/
public function convert() {
if($this->_directory ) {
return $this->_convertDirectory();
}
return $this->_convertFile();
}
/**
* Convert file
* @throws ConvertException
* @return boolean
*/
private function _convertFile() {
if( ! file_exists($this->_path) ) {
$message = $this->_path . ' does not exist.';
throw new ConvertException($message);
}
if( ! is_file($this->_path) ) {
$message = $this->_path . ' is not a file.';
throw new ConvertException($message);
}
if( ! $this->_isWR() ) {
$message = $this->_path . ' must can be read and write.';
throw new ConvertException($message);
}
$file_real_path = realpath($this->_path);
$file_content_from = file_get_contents( $file_real_path );
if( mb_check_encoding($file_content_from, $this->_from_encoding) ) {
$file_content_to = mb_convert_encoding( $file_content_from, $this->_to_encoding, $this->_from_encoding );
file_put_contents( $file_real_path, $file_content_to );
}
return true;
}
/**
* Convert directory
* @throws ConvertException
* @return boolean
*/
private function _convertDirectory() {
if( ! file_exists($this->_path) ) {
$message = $this->_path . ' does not exist.';
throw new ConvertException($message);
}
if( ! is_dir($this->_path) ) {
$message = $this->_path . ' is not a directory.';
throw new ConvertException($message);
}
if( ! $this->_isWR() ) {
$message = $this->_path . ' must can be read and write.';
throw new ConvertException($message);
}
$this->_scanDirFiles();
if( empty($this->_files) ) {
$message = $this->_path . ' is a empty directory.';
throw new ConvertException($message);
}
foreach( $this->_files as $value ) {
$file_content_from = file_get_contents( $value );
if( mb_check_encoding($file_content_from, $this->_from_encoding) ) {
$file_content_to = mb_convert_encoding( $file_content_from, $this->_to_encoding, $this->_from_encoding );
file_put_contents( $value, $file_content_to );
}
}
return true;
}
/**
* Determine whether the file or directory is readable and writable
* @return boolean returns true if it is readable and writable, otherwise returns false
*/
private function _isWR() {
if( is_readable($this->_path) && is_writable($this->_path) ) {
return true;
}
return false;
}
/**
* Traverse the directory, find all files, and add absolute paths
* @return boolean
*/
private function _scanDirFiles($dir = '') {
$base_path = empty( $dir ) ? realpath($this->_path) . DIRECTORY_SEPARATOR : realpath($dir) . DIRECTORY_SEPARATOR;
$files_tmp = empty( $dir ) ? scandir($this->_path) : scandir($dir);
foreach( $files_tmp as $value ) {
if( $value == '.' || $value == '..' || ( strpos($value, '.') === 0 ) ) {
continue;
}
$value = $base_path . $value;
if( is_dir($value) ) {
if( $this->_recursion ) {
$this->_scanDirFiles($value);
}
}
elseif( is_file($value) ) {
$this->_files[] = $value;
}
}
return true;
}
}
/**
* Conversion exception
*
*/
class ConvertException extends Exception {
}