Windows: readPassword

[bheads]

readPassword is returning a string of keys pressed not characters.

entering youHooHoo as a password returns
YOU►HOO►HOO

We might want to use _getch for the password input
(https://msdn.microsoft.com/en-us/library/078sfkak.aspx)

Unix is working as expected.

[bheads]

Nim's password input ran into this as well
nim-lang/Nim#2137

Will see if I can get a PR today

[bheads]

Testing with this:

int getch(bool echo = false)
    {
        import std.algorithm : among;

        INPUT_RECORD ir;
        DWORD count;
        int result;
        auto m = mode;

        mode = ConsoleInputMode.None;
        scope(exit) mode = m;

        while(ReadConsoleInputW(hInput, &ir, 1, &count)) {
            if(ir.EventType == KEY_EVENT &&
                ir.KeyEvent.bKeyDown &&
                ir.KeyEvent.wVirtualKeyCode.among(VK_SHIFT, VK_MENU, VK_CONTROL) == 0) {
                return ir.KeyEvent.UnicodeChar;
            }
        }

        return '\0';
    }

[robik]

Sorry for this delay and thanks for reporting. Any chance this will be a PR? :)

[AntonOks]

The below code (based on Abscissa/scriptlike#24) works a lot better (NOTE: I use the "original" terminal.d from Adam! and flush the STDOUT to get things printed ;) ):

string readPassword (char mask = '*')
{
// import terminal;
import arsd.terminal;
auto term = Terminal(ConsoleOutputType.linear);
auto input = RealTimeConsoleInput(&term, ConsoleInputFlags.raw);
string pass;
dchar ch;
ch = input.getch();
while(ch != '\r' && ch != '\n')
{
import std.range;
if(ch == '\b' && !pass.empty)
{
pass = pass[0..$-1];
write('\b');
stdout.flush;
}
else
{
pass ~= ch;
write(mask);
stdout.flush();
}
ch = input.getch();
}
return pass;
}

[adamdruppe]

On Sun, Mar 05, 2017 at 01:20:28PM -0800, notna123 wrote: flush the STDOUT to get things printed ;) ):

The getch function should flush for you btw.

[AntonOks]

nope, it doesn't... just tried again and still need the flush ;)

[AntonOks]

fixed in #36

[adamdruppe]

On Sun, Mar 05, 2017 at 01:33:50PM -0800, notna123 wrote: nope, it doesn't... just tried again and still need the flush ;)

Weird. I see it in the source, it should flush before any events :S whelp, even if it is a bug I fix later, no big deal to flush twice, if the buffer is empty anyway it will just noop.