[newlib-cygwin] Cygwin: console: improve replacement char algorithm

[Corinna Vinschen]

https://sourceware.org/git/gitweb.cgi?p=newlib-cygwin.git;h=bf8aabe830d215f13e690b21a682fc37aeb8752c

commit bf8aabe830d215f13e690b21a682fc37aeb8752c
Author: Corinna Vinschen <***@vinschen.de>
Date: Wed Sep 5 23:39:25 2018 +0200

Cygwin: console: improve replacement char algorithm

Try various Unicode characters which may be used as a replacement
character in case an invalid character has to be printed.

Current list is 0xfffd "REPLACEMENT CHARACTER", 0x25a1 "WHITE SQUARE",
and 0x2592 "MEDIUM SHADE" in that order.

Additionally workaround a problem with some fonts (namely DejaVu
Sans Mono) which are returned wit ha broken fontname with trailing
stray characters.

Signed-off-by: Corinna Vinschen <***@vinschen.de>

Diff:
---
winsup/cygwin/autoload.cc | 7 +++
winsup/cygwin/fhandler_console.cc | 95 +++++++++++++++++++++++++++++++++++++--
2 files changed, 99 insertions(+), 3 deletions(-)

diff --git a/winsup/cygwin/autoload.cc b/winsup/cygwin/autoload.cc
index f71abe8..4fac3e3 100644
--- a/winsup/cygwin/autoload.cc
+++ b/winsup/cygwin/autoload.cc
@@ -640,6 +640,12 @@ LoadDLLfunc (LsaRegisterLogonProcess, 12, secur32)

LoadDLLfunc (SHGetDesktopFolder, 4, shell32)

+LoadDLLfunc (CreateFontW, 56, gdi32)
+LoadDLLfunc (DeleteObject, 4, gdi32)
+LoadDLLfunc (EnumFontFamiliesExW, 20, gdi32)
+LoadDLLfunc (GetGlyphIndicesW, 20, gdi32)
+LoadDLLfunc (SelectObject, 8, gdi32)
+
LoadDLLfunc (CloseClipboard, 0, user32)
LoadDLLfunc (CloseDesktop, 4, user32)
LoadDLLfunc (CloseWindowStation, 4, user32)
@@ -651,6 +657,7 @@ LoadDLLfunc (DispatchMessageW, 4, user32)
LoadDLLfunc (EmptyClipboard, 0, user32)
LoadDLLfunc (EnumWindows, 8, user32)
LoadDLLfunc (GetClipboardData, 4, user32)
+LoadDLLfunc (GetDC, 4, user32)
LoadDLLfunc (GetForegroundWindow, 0, user32)
LoadDLLfunc (GetKeyboardLayout, 4, user32)
LoadDLLfunc (GetMessageW, 16, user32)
diff --git a/winsup/cygwin/fhandler_console.cc b/winsup/cygwin/fhandler_console.cc
index b4674b8..c654d66 100644
--- a/winsup/cygwin/fhandler_console.cc
+++ b/winsup/cygwin/fhandler_console.cc
@@ -1971,14 +1971,103 @@ bad_escape:
}
}

+#define NUM_REPLACEMENT_CHARS 3
+
+static const wchar_t replacement_char[NUM_REPLACEMENT_CHARS] =
+{
+ 0xfffd, /* REPLACEMENT CHARACTER */
+ 0x25a1, /* WHITE SQUARE */
+ 0x2592 /* MEDIUM SHADE */
+};
+/* nFont member is always 0 so we have to use the facename. */
+static WCHAR cons_facename[LF_FACESIZE];
+static int rp_char_idx;
+static HDC cdc;
+
+static int CALLBACK
+enum_proc (const LOGFONTW *lf, const TEXTMETRICW *tm,
+ DWORD FontType, LPARAM lParam)
+{
+ int *done = (int *) lParam;
+ *done = 1;
+ return 0;
+}
+
+static void
+check_font (HANDLE hdl)
+{
+ CONSOLE_FONT_INFOEX cfi;
+ LOGFONTW lf;
+
+ cfi.cbSize = sizeof cfi;
+ if (!GetCurrentConsoleFontEx (hdl, 0, &cfi))
+ return;
+ /* Switched font? */
+ if (wcscmp (cons_facename, cfi.FaceName) == 0)
+ return;
+ if (!cdc && !(cdc = GetDC (GetConsoleWindow ())))
+ return;
+ /* Some FaceNames like DejaVu Sans Mono are sometimes returned with stray
+ trailing chars. Fix it. */
+ lf.lfCharSet = ANSI_CHARSET;
+ lf.lfPitchAndFamily = FIXED_PITCH | FF_DONTCARE;
+ wchar_t *cp = wcpcpy (lf.lfFaceName, cfi.FaceName) - 1;
+ int done = 0;
+ do
+ {
+ EnumFontFamiliesExW (cdc, &lf, enum_proc, (LPARAM) &done, 0);
+ if (!done && cp > lf.lfFaceName)
+ *cp-- = L'\0';
+ }
+ while (!done);
+ /* Yes. Check for the best replacement char. */
+ HFONT f = CreateFontW (0, 0, 0, 0,
+ cfi.FontWeight, FALSE, FALSE, FALSE,
+ ANSI_CHARSET, OUT_DEFAULT_PRECIS,
+ CLIP_DEFAULT_PRECIS, DEFAULT_QUALITY,
+ FIXED_PITCH | FF_DONTCARE, lf.lfFaceName);
+ if (!f)
+ return;
+
+ HFONT old_f = (HFONT) SelectObject(cdc, f);
+ if (old_f)
+ {
+ WORD glyph_idx[NUM_REPLACEMENT_CHARS];
+
+ if (GetGlyphIndicesW (cdc, replacement_char,
+ NUM_REPLACEMENT_CHARS, glyph_idx,
+ GGI_MARK_NONEXISTING_GLYPHS) != GDI_ERROR)
+ {
+ int i;
+
+ for (i = 0; i < NUM_REPLACEMENT_CHARS; ++i)
+ if (glyph_idx[i] != 0xffff)
+ break;
+ if (i == NUM_REPLACEMENT_CHARS)
+ i = 0;
+ rp_char_idx = i;
+ /* Note that we copy the original name returned by
+ GetCurrentConsoleFontEx, even if it was broken.
+ This allows an early return, rather than to store
+ the fixed name and then having to enum font families
+ all over again. */
+ wcscpy (cons_facename, cfi.FaceName);
+ }
+ SelectObject (cdc, old_f);
+ }
+ DeleteObject (f);
+}
+
/* This gets called when we found an invalid input character.
- Print Unicode REPLACEMENT CHARACTER (UTF 0xfffd). */
+ Print one of the above Unicode chars as replacement char. */
inline void
fhandler_console::write_replacement_char ()
{
- static const wchar_t replacement_char = 0xfffd; /* REPLACEMENT CHARACTER */
+ check_font (get_output_handle ());
+
DWORD done;
- WriteConsoleW (get_output_handle (), &replacement_char, 1, &done, 0);
+ WriteConsoleW (get_output_handle (), &replacement_char[rp_char_idx], 1,
+ &done, 0);
}

const unsigned char *