[PATCH 0/2] MR1846: Fix CUI output on Unix tty.

[eric pouech (＠epo)]

While testing some Windows CUI app (hiew.exe from bug #54141), it appears
that some console API allow to write all the possible control characters
onto the console.
This serie:
- add tests for read/write test of these
- fix 'unix' console to not send some of the control characters to the
  underlying tty, which would reinterpret them. This leads to unwanted
  display (ie hexdump of a file would insert real cr/lf when displaying
  dump of \r\n")

[PATCH 1/2] kernel32/tests: Test that control characters can be read & store.

[Eric Pouech]

From: Eric Pouech

WriteConsole (not in processed mode) and WriteConsoleOutput* functions
allow to write control characters, which can then be read back as is.

Signed-off-by: Eric Pouech 
---
 dlls/kernel32/tests/console.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/dlls/kernel32/tests/console.c b/dlls/kernel32/tests/console.c
index 1f285faf1e2..f8e39c164d2 100644
--- a/dlls/kernel32/tests/console.c
+++ b/dlls/kernel32/tests/console.c
@@ -306,6 +306,7 @@ static void testWriteNotWrappedNotProcessed(HANDLE hCon, COORD sbSize)
     DWORD		len, mode;
     const char*         mytest = "123";
     const int           mylen = strlen(mytest);
+    char                ctrl_buf[32];
     int                 ret;
     int			p;

@@ -336,6 +337,17 @@ static void testWriteNotWrappedNotProcessed(HANDLE hCon, COORD sbSize)
     ok(SetConsoleCursorPosition(hCon, c) != 0, "Cursor in upper-left-3\n");

ok(WriteConsoleA(hCon, mytest, mylen, &len, NULL) != 0 && len == mylen, "WriteConsole\n");
+
+    /* test how control chars are handled. */
+    c.X = c.Y = 0;
+    ok(SetConsoleCursorPosition(hCon, c) != 0, "Cursor in upper-left-3\n");
+    for (p = 0; p < 32; p++) ctrl_buf[p] = (char)p;
+    ok(WriteConsoleA(hCon, ctrl_buf, 32, &len, NULL) != 0 && len == 32, "WriteConsole\n");
+    for (p = 0; p < 32; p++)
+    {
+        c.X = p; c.Y = 0;
+        okCHAR(hCon, c, (char)p, TEST_ATTRIB);
+    }
 }

static void testWriteNotWrappedProcessed(HANDLE hCon, COORD sbSize)
@@ -2378,6 +2390,19 @@ static void test_WriteConsoleOutputCharacterA(HANDLE output_handle)
     ret = WriteConsoleOutputCharacterA(output_handle, output, 0, origin, &count);
     ok(ret == TRUE, "Expected WriteConsoleOutputCharacterA to return TRUE, got %d\n", ret);
     ok(count == 0, "Expected count to be 0, got %lu\n", count);
+
+    for (i = 1; i < 32; i++)
+    {
+        CONSOLE_SCREEN_BUFFER_INFO csbi;
+        char ch = (char)i;
+        COORD c = {1, 2};
+
+        ret = WriteConsoleOutputCharacterA(output_handle, &ch, 1, c, &count);
+        ok(ret == TRUE, "Expected WriteConsoleOutputCharacterA to return TRUE, got %d\n", ret);
+        ok(count == 1, "Expected count to be 1, got %lu\n", count);
+        okCHAR(output_handle, c, (char)i, 7);
+        ret = GetConsoleScreenBufferInfo(output_handle, &csbi);
+    }
 }

static void test_WriteConsoleOutputCharacterW(HANDLE output_handle)

[PATCH 2/2] conhost: Don't print control characters for Unix tty.

[Eric Pouech]

From: Eric Pouech

They will be reinterpreted by Unix tty, leading to wrong display.

Signed-off-by: Eric Pouech 
---
 programs/conhost/conhost.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/programs/conhost/conhost.c b/programs/conhost/conhost.c
index c3a106c9321..769f998f404 100644
--- a/programs/conhost/conhost.c
+++ b/programs/conhost/conhost.c
@@ -358,6 +358,8 @@ static void update_output( struct screen_buffer *screen_buffer, RECT *rect )
     int x, y, size, trailing_spaces;
     char_info_t *ch;
     char buf[8];
+    WCHAR wch;
+    const unsigned int mask = (1u << '\0') | (1u << '\b') | (1u << '\t') | (1u << '\n') | (1u << '\a') | (1u << '\r');

if (!is_active( screen_buffer ) || rect->top > rect->bottom || rect->right < rect->left)
         return;
@@ -393,9 +395,11 @@ static void update_output( struct screen_buffer *screen_buffer, RECT *rect )
                 tty_write( screen_buffer->console, "\x1b[K", 3 );
                 break;
             }
-
+            wch = ch->ch;
+            if (screen_buffer->console->is_unix && wch < L' ' && mask & (1u << wch))
+                wch = L'?';
             size = WideCharToMultiByte( get_tty_cp( screen_buffer->console ), 0,
-                                        &ch->ch, 1, buf, sizeof(buf), NULL, NULL );
+                                        &wch, 1, buf, sizeof(buf), NULL, NULL );
             tty_write( screen_buffer->console, buf, size );
             screen_buffer->console->tty_cursor_x++;
         }

[Jacek Caban (＠jacek)]

FWIW, I think it's weird that it's not handled similarly by MS for pseudo-consoles, they will be broken in this case. Anyway, testing conforms that, so I guess we need to do that only for Unix consoles.

[PATCH 0/2] MR1846: Fix CUI output on Unix tty. - approved

[Jacek Caban (＠jacek)]

This merge request was approved by Jacek Caban.

[eric pouech (＠epo)]

On Tue Dec 20 14:31:28 2022 +0000, Jacek Caban wrote:
> FWIW, I think it's weird that it's not handled similarly by MS for
> pseudo-consoles, they will be broken in this case. Anyway, testing
> conforms that, so I guess we need to do that only for Unix consoles.

yes for pseudo consoles, see https://github.com/microsoft/terminal/issues/4363