Avoid consing extra string when processing GTK native input

* src/gtkutil.c (xg_im_context_commit): Use
`decode_string_utf8' to decode input text.

* src/keyboard.c (kbd_buffer_get_event_1): If coding system is
Qt, simply return the string without decoding it.

* src/termhooks.h (enum event_kind): Document meaning of Qt as
coding system in a multibyte keystroke event's string argument.

3 files changed, 19 insertions, 4 deletions

diff --git a/src/gtkutil.c b/src/gtkutil.c
index 27aa28b8902..158c29272f5 100644
--- a/src/gtkutil.c
+++ b/src/gtkutil.c
@@ -6097,11 +6097,21 @@ xg_im_context_commit (GtkIMContext *imc, gchar *str,
   struct input_event ie;
 
   EVENT_INIT (ie);
+  /* This used to use g_utf8_to_ucs4_fast, which led to bad results
+     when STR wasn't actually a UTF-8 string, which some input method
+     modules commit.  */
+
   ie.kind = MULTIBYTE_CHAR_KEYSTROKE_EVENT;
-  ie.arg = build_unibyte_string (str);
+  ie.arg = decode_string_utf_8 (Qnil, str, strlen (str),
+                                Qnil, false, Qnil, Qnil);
+
+  /* STR is invalid and not really encoded in UTF-8.  */
+  if (NILP (ie.arg))
+    ie.arg = build_unibyte_string (str);
 
-  Fput_text_property (make_fixnum (0), make_fixnum (strlen (str)),
-                      Qcoding, Qutf_8_unix, ie.arg);
+  Fput_text_property (make_fixnum (0),
+                      make_fixnum (SCHARS (ie.arg)),
+                      Qcoding, Qt, ie.arg);
 
   XSETFRAME (ie.frame_or_window, f);
   ie.modifiers = 0;
diff --git a/src/keyboard.c b/src/keyboard.c
index 0747ab48205..2aff0f1011d 100644
--- a/src/keyboard.c
+++ b/src/keyboard.c
@@ -3841,6 +3841,9 @@ kbd_buffer_get_event_1 (Lisp_Object arg)
   Lisp_Object coding_system = Fget_text_property (make_fixnum (0),
                                                   Qcoding, arg);
 
+  if (EQ (coding_system, Qt))
+    return arg;
+
   return code_convert_string (arg, (!NILP (coding_system)
                                     ? coding_system
                                     : Vlocale_coding_system),
diff --git a/src/termhooks.h b/src/termhooks.h
index 0a9ab61afab..b7696fed4f8 100644
--- a/src/termhooks.h
+++ b/src/termhooks.h
@@ -100,7 +100,9 @@ enum event_kind
 
                                            If it is nil, then the
                                            locale coding system will
-                                           be used.  */
+                                           be used.  If it is t, then
+                                           no decoding will take
+                                           place.  */
   NON_ASCII_KEYSTROKE_EVENT,    /* .code is a number identifying the
                                    function key.  A code N represents
                                    a key whose name is