Bug #79455 [Fbk->Nab]: scandir() mishandles UTF-8 character(s)

Edit report at https://bugs.php.net/bug.php?id=79455&edit=1 ID: 79455 Updated by: cmb@php.net Reported by: pguest at meaa dot mea dot com Summary: scandir() mishandles UTF-8 character(s) -Status: Feedback +Status: Not a bug Type: Bug Package: Directory function related Operating System: Windows 10 PHP Version: 7.4.4 Assigned To: cmb Block user comment: N Private report: N New Comment: > I use CLI SAPI within Eclipse PDT IDE. Ah, that custom console layer likely explains the issue. Great that you have been able to resolve the problem! Previous Comments: ------------------------------------------------------------------------ [2020-04-09 14:47:34] pguest at meaa dot mea dot com I use CLI SAPI within Eclipse PDT IDE. I do not explicitly involve an auto_prepend_file within php.ini. I am not sure whether the Eclipse environment invokes certain configuration files prior to a run configuration launch. However I am thankful for your suggestion for use of sapi_windows_cp_set() within my scripts. This dialog with you has helped educate me in regard to character encoding and code pages. Thank you! ------------------------------------------------------------------------ [2020-04-08 07:37:30] cmb@php.net Thanks for the info! So apparently your script is UTF-8 encoded, and you're using the default INI settings, which is supposed to produce the desired filenames, but it works as if there was a call to sapi_windows_cp_set(1252) at the beginning of the script. Is there a respective auto_prepend_file? Anyhow, I suppose that adding sapi_windows_cp_set(65001); at the top of the script should enforce the desired behavior. By the way, which SAPI do you use? ------------------------------------------------------------------------ [2020-04-07 14:53:27] pguest at meaa dot mea dot com cmb, Here is output from: <?php var_dump( ini_get('internal_encoding'), ini_get('default_charset'), ini_get('zend.multibyte'), sapi_windows_cp_get('ansi'), sapi_windows_cp_get('oem') ); ?> C:\Workspace\PDT\__CORE\php_cmb.php:7: string(0) "" C:\Workspace\PDT\__CORE\php_cmb.php:7: string(5) "UTF-8" C:\Workspace\PDT\__CORE\php_cmb.php:7: string(1) "0" C:\Workspace\PDT\__CORE\php_cmb.php:7: int(1252) C:\Workspace\PDT\__CORE\php_cmb.php:7: int(437) ------------------------------------------------------------------------ [2020-04-07 14:47:26] cmb@php.net I cannot reproduce this. Please provide the output of the following script: <?php var_dump( ini_get('internal_encoding'), ini_get('default_charset'), ini_get('zend.multibyte'), sapi_windows_cp_get('ansi'), sapi_windows_cp_get('oem') ); ?> ------------------------------------------------------------------------ [2020-04-07 14:04:34] pguest at meaa dot mea dot com Description: ------------ A Windows filesystem directory containing the character: á, 225, 0xE1 is returned from scandir as two characters: Ã, 195, 0xC3 ¡, 161, 0xA1 In otherwords: 'Arquivo dos Gráficos' becomes: 'Arquivo dos Gráficos' Test script: --------------- <?php $rootdir = 'C:/temp/'; $portuguese_string = 'Arquivo dos Gráficos'; $newdir = $rootdir . $portuguese_string; if (is_dir($newdir)) { rmdir ($newdir); } if (is_dir($rootdir)) { rmdir ($rootdir); } mkdir($rootdir); mkdir($newdir); echo sprintf("'%s' encoding: %s\n", $portuguese_string, mb_detect_encoding($portuguese_string)); $scandir_return = scandir($rootdir); echo "scandir() returns: " . print_r($scandir_return, true); echo sprintf("'%s' encoding: %s\n", $scandir_return[2], mb_detect_encoding($scandir_return[2])); ?> Expected result: ---------------- The expectation is that scandir() should read back the directory string with the same UTF-8 characters involved in its creation. Actual result: -------------- 'Arquivo dos Gráficos' becomes: 'Arquivo dos Gráficos' ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=79455&edit=1