[gtk+] ime: Add builtin handling of dead keys

[Matthias Clasen <matthiasc src gnome org>]

commit 493ca20912a9a84b3b13f44c1e99b97eb75dd943
Author: Carlos Garnacho <carlosg gnome org>
Date:   Fri Jul 26 17:45:19 2013 +0200

    ime: Add builtin handling of dead keys
    
    The IME input method has been both ignoring keypresses of
    non-spacing characters (ditching these as non displayable),
    and not letting IME do anything about those.
    
    Even though, the sparse documentation on IMM/IME seems to
    hint that applications can't pipe non-spacing characters to
    the input method manager, and experimentation shown that
    those characters are indeed handled differently than how
    it'd be expected.
    
    Then, add basic handling of dead keys on the IME input method
    itself , as it's not mutually exclusive with regular keymaps
    with dead keys.
    
    https://bugzilla.gnome.org/show_bug.cgi?id=704937

 modules/input/gtkimcontextime.c |   97 +++++++++++++++++++++++++++++++++++++--
 1 files changed, 92 insertions(+), 5 deletions(-)
---
diff --git a/modules/input/gtkimcontextime.c b/modules/input/gtkimcontextime.c
index aa7f9a8..0de1b53 100644
--- a/modules/input/gtkimcontextime.c
+++ b/modules/input/gtkimcontextime.c
@@ -31,6 +31,7 @@
 
 #include "imm-extra.h"
 
+#include <gdk/gdkkeysyms.h>
 #include "gdk/win32/gdkwin32.h"
 #include "gdk/gdkkeysyms.h"
 
@@ -49,6 +50,9 @@
 
 /* #define BUFSIZE 4096 */
 
+#define IS_DEAD_KEY(k) \
+    ((k) >= GDK_dead_grave && (k) <= (GDK_dead_dasia+1))
+
 #define FREE_PREEDIT_BUFFER(ctx) \
 {                                \
   g_free((ctx)->priv->comp_str); \
@@ -70,6 +74,8 @@ struct _GtkIMContextIMEPrivate
   DWORD comp_str_len;
   LPVOID read_str;
   DWORD read_str_len;
+
+  guint32 dead_key_keyval;
 };
 
 
@@ -298,6 +304,64 @@ gtk_im_context_ime_set_client_window (GtkIMContext *context,
   context_ime->client_window = client_window;
 }
 
+static gunichar
+_gtk_im_context_ime_dead_key_unichar (guint    keyval,
+                                      gboolean spacing)
+{
+  switch (keyval)
+    {
+#define CASE(keysym, unicode, spacing_unicode) \
+      case GDK_dead_##keysym: return (spacing) ? spacing_unicode : unicode;
+
+      CASE (grave, 0x0300, 0x0060);
+      CASE (acute, 0x0301, 0x00b4);
+      CASE (circumflex, 0x0302, 0x005e);
+      CASE (tilde, 0x0303, 0x007e);    /* Also used with perispomeni, 0x342. */
+      CASE (macron, 0x0304, 0x00af);
+      CASE (breve, 0x0306, 0x02d8);
+      CASE (abovedot, 0x0307, 0x02d9);
+      CASE (diaeresis, 0x0308, 0x00a8);
+      CASE (hook, 0x0309, 0);
+      CASE (abovering, 0x030A, 0x02da);
+      CASE (doubleacute, 0x030B, 0x2dd);
+      CASE (caron, 0x030C, 0x02c7);
+      CASE (abovecomma, 0x0313, 0);         /* Equivalent to psili */
+      CASE (abovereversedcomma, 0x0314, 0); /* Equivalent to dasia */
+      CASE (horn, 0x031B, 0);  /* Legacy use for psili, 0x313 (or 0x343). */
+      CASE (belowdot, 0x0323, 0);
+      CASE (cedilla, 0x0327, 0x00b8);
+      CASE (ogonek, 0x0328, 0);        /* Legacy use for dasia, 0x314.*/
+      CASE (iota, 0x0345, 0);
+
+#undef CASE
+    default:
+      return 0;
+    }
+}
+
+static void
+_gtk_im_context_ime_commit_unichar (GtkIMContextIME *context_ime,
+                                    gunichar         c)
+{
+  gchar utf8[10];
+  int len;
+
+  if (context_ime->priv->dead_key_keyval != 0)
+    {
+      gunichar combining;
+
+      combining =
+        _gtk_im_context_ime_dead_key_unichar (context_ime->priv->dead_key_keyval,
+                                              FALSE);
+      g_unichar_compose (c, combining, &c);
+    }
+
+  len = g_unichar_to_utf8 (c, utf8);
+  utf8[len] = 0;
+
+  g_signal_emit_by_name (context_ime, "commit", utf8);
+  context_ime->priv->dead_key_keyval = 0;
+}
 
 static gboolean
 gtk_im_context_ime_filter_keypress (GtkIMContext *context,
@@ -324,16 +388,39 @@ gtk_im_context_ime_filter_keypress (GtkIMContext *context,
   if (!GDK_IS_WINDOW (context_ime->client_window))
     return FALSE;
 
+  if (event->keyval == GDK_space &&
+      context_ime->priv->dead_key_keyval != 0)
+    {
+      c = _gtk_im_context_ime_dead_key_unichar (context_ime->priv->dead_key_keyval, TRUE);
+      context_ime->priv->dead_key_keyval = 0;
+      _gtk_im_context_ime_commit_unichar (context_ime, c);
+      return TRUE;
+    }
+
   c = gdk_keyval_to_unicode (event->keyval);
+
   if (c)
     {
-      guchar utf8[10];
-      int len = g_unichar_to_utf8 (c, utf8);
-      utf8[len] = 0;
-
-      g_signal_emit_by_name (context_ime, "commit", utf8);
+      _gtk_im_context_ime_commit_unichar (context_ime, c);
       retval = TRUE;
     }
+  else if (IS_DEAD_KEY (event->keyval))
+    {
+      gunichar dead_key;
+
+      dead_key = _gtk_im_context_ime_dead_key_unichar (event->keyval, FALSE);
+
+      /* Emulate double input of dead keys */
+      if (dead_key && event->keyval == context_ime->priv->dead_key_keyval)
+        {
+          c = _gtk_im_context_ime_dead_key_unichar (context_ime->priv->dead_key_keyval, TRUE);
+          context_ime->priv->dead_key_keyval = 0;
+          _gtk_im_context_ime_commit_unichar (context_ime, c);
+          _gtk_im_context_ime_commit_unichar (context_ime, c);
+        }
+      else
+        context_ime->priv->dead_key_keyval = event->keyval;
+    }
 
   return retval;
 }