Issue with FILE_kill

[dco045]

Hello

I encounter a problem with File_Kill function.

The part of code I encounter the problem :

function clean_dir
  nFiles  = DIR_ListArray(MyList, ".", "*", %FILE_NORMAL + %FILE_ADDPATH)

  printl "                                                     nombre de fichier(s)   trouve(s) " , nFiles in c_green
  for ii = 1 to nFiles
    print mylist(ii) , "     " in c_magenta
    printl file_kill(mylist(ii) )
  Next
end function

This code works correctly with most files, but fails with filename containing accented letters, some are from files transferred fro MAC.

With unaccented files the retcode is 0 and the file is removed.
With accented files the retcode is 53 = File not found and obviously the file is NOT removed.

I suppose that is as the issue I posted 29-01-2024, 19:35 , Eros replyed 14-02-2023, 12:09 about ANSI/OEM/UTF8 conversion ?



Another little flaw with File_Kill

If the file name to remove is a subdir, the subdir is not removed, normal it is designed for files.....
But the return code is 0, which means : No error ...

Maybe , would it be someting as 75 = Wrong path ?

Regards,

Dany

[ReneMiner]

The accented-non-accented thing appears to me as a problem of the codepage more than a thinbasic problem
we will find out,
open windows explorer, where you can type the path, type %COMSPEC% [enter]
the windows console will open,into console window type CHCP[enter]
you will be informed about the currently used codepage, ANSI i guess 850 or 1252 (latin or western europe) if french speaking

Now thinbasic default works unicode-aware using UTF8 which is a codepage number 65001

 
%CP_utf7 = 65000
%CP_utf8= 65001
%CP_utf16_LE = 1200   ' this is the WIDECHAR 16 bit on windows
%CP_utf16_BE = 1201  ' UNIX,Linux,Solarix,Androidix, Macintoshix,ComputerLäuftNix,..

ANSI/OEM/EBCDIC/BOCU/SCSU are a couple of codepage-collections mostly deprecated

When you are working with a visible console window, i.e. your code contains

Uses "console"

then it were applicable to let

Console_SetoutputCP( %CP_UTF8 )
' Console_SetInputCP( %CP_UTF8 ) might set this to the number <XXX>
Console_ShowWindow( %Console_SW_SHOWNORMAL )

and it should go conform with content of thuinbasic UTF8-encoded strings.

When no visible console window either you use

string FileToRun = App_Scriptpath & "myBatchfile.bat"
string FileResult= App_Scriptpath & "reply.txt"
String myBatchfile ="@echo off" & $CRLF & "chcp >" & FileResult & $crlf & $spc


File_Save(FileToRun, myBatchfile) 
Shell FileToRun, "", True
msgbox "Returned was " & File_load(FileResult)

' .bat-scripts are usually ANSI. if your batchfile contains accented chars or the result is unreadable convert

myBatchfile=utf8ToAnsi$(myBatchfile)

before saving it and vice versa

msgbox "Returned was " & AnsiToUtf8$(File_load(FileResult))

Shell_CaptureOutput can do this without the files on a more direct way since thinbasic takes care of incoming data to be UTF8 (at least i hope so)

 


Begin Const
$_CMD= "%COMSPEC% /C "
End Const 

string sAll = Shell_captureOutput( $_CMD & "Dir /a:-d /b" ,"", %SW_HIDE, 20)
string sCurrDir = Shell_CaptureOutput( $_CMD & "CD", "",%SW_HIDE,10)

msgbox "Found Files" & $crlf & sAll, %mb_iconInformation, sCurrDir
String sFile()
Long numFiles = parse sAll, sFile, $CRLF 
' will make an array of all files in sAll

string isTB = Shell_CaptureOutput( $_CMD & "IF EXIST " & $DQ &  App_Path & "thinbasic.exe" & $DQ & " (ECHO TRUE) ELSE (ECHO FALSE)", "", %SW_HIDE,10)
msgbox App_Path & "thinbasic.exe EXIST = " & isTB 

'The parenthesis surrounding ECHO TRUE  and ECHO FALSE prevent from  additional whitespace that might follow after "FALSE" so it will not  return is as "FALSE "
' For directories the path must end with backslash when EXIST is used

gives you an idea.
Of course you can combine batchfiles and result-files with the direct-method , just for a reminder

Echo Hello>myFile.txt
Echo World >>myFile.txt

the single "arrow >" ovewrites the content of an existing file
and double "arrow >>" appends it to the existing content.


Not a programmatic solution were to lauch cmd,
click on the icon(system-menu) upper left corner and
check settings,
then the rightmost tab should also tell the current by the system used codepage that i would try as number <XXX> mentioned above

[dco045]

The accented-non-accented thing appears to me as a problem of the codepage more than a thinbasic problem
we will find out,
open windows explorer, where you can type the path, type %COMSPEC% [enter]
the windows console will open,into console window type CHCP[enter]
you will be informed about the currently used codepage, ANSI i guess 850 or 1252 (latin or western europe) if french speaking

Now thinbasic default works unicode-aware using UTF8 which is a codepage number 65001
[code]
%CP_utf7 = 65000
%CP_utf8= 65001
%CP_utf16_LE = 1200 ' this is the WIDECHAR 16 bit on windows
%CP_utf16_BE = 1201 ' UNIX,Linux,Solarix,Androidix, Macintoshix,ComputerLäuftNix,..

Hello Rene,

I'm perfectly aware of code pages.
The post is to point that two functions of ThinBasic, used in the same script, in the same sub, don't agree with code pages.
Someware, the same kind of problem was pointed to, and Eros corrected the flaw.

Echo Hello>myFile.txt
Echo World >>myFile.txt

the single "arrow >" ovewrites the content of an existing file
and double "arrow >>" appends it to the existing content.

Thanks for the explain of redirections. As you can read in my profile, I was Unix sys-admin for 20 years


Dany.

[ErosOlmi]

Hello

I encounter a problem with File_Kill function.

The part of code I encounter the problem :

function clean_dir
  nFiles  = DIR_ListArray(MyList, ".", "*", %FILE_NORMAL + %FILE_ADDPATH)

  printl "                                                     nombre de fichier(s)   trouve(s) " , nFiles in c_green
  for ii = 1 to nFiles
    print mylist(ii) , "     " in c_magenta
    printl file_kill(mylist(ii) )
  Next
end function

This code works correctly with most files, but fails with filename containing accented letters, some are from files transferred fro MAC.

With unaccented files the retcode is 0 and the file is removed.
With accented files the retcode is 53 = File not found and obviously the file is NOT removed.

I suppose that is as the issue I posted 29-01-2024, 19:35 , Eros replyed 14-02-2023, 12:09 about ANSI/OEM/UTF8 conversion ?



Another little flaw with File_Kill

If the file name to remove is a subdir, the subdir is not removed, normal it is designed for files.....
But the return code is 0, which means : No error ...

Maybe , would it be someting as 75 = Wrong path ?

Regards,

Dany

Hi Dany, can you please attach a file (even empty file) with the name creating problems in File_Kill?
I'm trying to reproduce the problem on my computer but File_Kill seems working fine with accented chars like èùàòéì.
So I think I'm missing something

[Petr Schreiber]

Hi Dany,

tried it on my PC and also could not get the files to remain undeleted.

If you can pack a directory with the offending files, it would help greatly!


Petr