input: add new Unicode input mode

This mode is activated through the new key-bindings.unicode-input and
search-bindings.unicode-input key bindings.

When active, the user can “build” a Unicode codepoint by typing its
hexadecimal value.

Note that there’s no visual feedback in this mode. This is
intentional. This mode is intended to be a fallback for users that
don’t use an IME.

Closes #1116

CHANGELOG.md

@@ -52,11 +52,16 @@
   `ctrl`+`shift`+`x` respectively ([#30][30]).
 * `colors.search-box-no-match` and `colors.search-box-match` options
   to `foot.ini` ([#1112][1112]).
+* Very basic Unicode input mode via the new
+  `key-bindings.unicode-input` and `search-bindings.unicode-input` key
+  bindings. Note that there is no visual feedback, as the preferred
+  way of entering Unicode characters is with an IME ([#1116][1116]).
 
 [1058]: https://codeberg.org/dnkl/foot/issues/1058
 [1070]: https://codeberg.org/dnkl/foot/issues/1070
 [30]: https://codeberg.org/dnkl/foot/issues/30
 [1112]: https://codeberg.org/dnkl/foot/issues/1112
+[1116]: https://codeberg.org/dnkl/foot/issues/1116
 
 
 ### Changed

config.c

@@ -117,6 +117,7 @@ static const char *const binding_action_map[] = {
     [BIND_ACTION_TEXT_BINDING] = "text-binding",
     [BIND_ACTION_PROMPT_PREV] = "prompt-prev",
     [BIND_ACTION_PROMPT_NEXT] = "prompt-next",
+    [BIND_ACTION_UNICODE_INPUT] = "unicode-input",
 
     /* Mouse-specific actions */
     [BIND_ACTION_SELECT_BEGIN] = "select-begin",
@@ -148,6 +149,7 @@ static const char *const search_binding_action_map[] = {
     [BIND_ACTION_SEARCH_EXTEND_WORD_WS] = "extend-to-next-whitespace",
     [BIND_ACTION_SEARCH_CLIPBOARD_PASTE] = "clipboard-paste",
     [BIND_ACTION_SEARCH_PRIMARY_PASTE] = "primary-paste",
+    [BIND_ACTION_SEARCH_UNICODE_INPUT] = "unicode-input",
 };
 
 static const char *const url_binding_action_map[] = {

doc/foot.ini.5.scd

@@ -791,6 +791,32 @@ e.g. *search-start=none*.
 	Jump the next prompt (requires shell integration, see
 	*foot*(1)). Default: _Control+Shift+x_.
 
+*unicode-input*
+	Input a Unicode character by typing its codepoint in hexadecimal,
+	followed by *Enter* or *Space*.
+	
+	For example, to input the character _ö_ (LATIN SMALL LETTER O WITH
+	DIAERESIS, Unicode codepoint 0xf6), you would first activate this
+	key binding, then type: *f*, *6*, *Enter*.
+	
+	Another example: to input 😍 (SMILING FACE WITH HEART-SHAPED EYES,
+	Unicode codepoint 0x1f60d), activate this key binding, then type:
+	*1*, *f*, *6*, *0*, *d*, *Enter*.
+	
+	Recognized key bindings in Unicode input mode:
+	
+	- Enter, Space: commit the Unicode character, then exit this mode.
+	- Escape, Ctrl+c, Ctrl+d, Ctrl+g: abort input, then exit this mode.
+	- 0-9, a-f: append next digit to the Unicode's codepoint.
+	- Backspace: undo the last digit.
+	
+	Note that there is no visual feedback while in this mode. This is
+	by design; foot's Unicode input mode is considered to be a
+	fallback. The preferred way of entering Unicode characters, emojis
+	etc is by using an IME.
+	
+	Default: _none_.
+
 # SECTION: search-bindings
 
 This section lets you override the default key bindings used in
@@ -869,6 +895,10 @@ scrollback search mode. The syntax is exactly the same as the regular
 	Paste from the _primary selection_ into the search
 	buffer. Default: _Shift+Insert_.
 
+*unicode-input*
+	Unicode input mode. See _key-bindings.unicode-input_ for
+	details. Default: _none_.
+
 # SECTION: url-bindings
 
 This section lets you override the default key bindings used in URL

foot.ini

@@ -150,6 +150,7 @@
 # show-urls-persistent=none
 # prompt-prev=Control+Shift+z
 # prompt-next=Control+Shift+x
+# unicode-input=none
 # noop=none
 
 [search-bindings]
@@ -171,6 +172,7 @@
 # extend-to-next-whitespace=Control+Shift+w
 # clipboard-paste=Control+v Control+Shift+v Control+y XF86Paste
 # primary-paste=Shift+Insert
+# unicode-input=none
 
 [url-bindings]
 # cancel=Control+g Control+c Control+d Escape

input.c

@@ -36,6 +36,7 @@
 #include "spawn.h"
 #include "terminal.h"
 #include "tokenize.h"
+#include "unicode-mode.h"
 #include "url-mode.h"
 #include "util.h"
 #include "vt.h"
@@ -416,6 +417,10 @@ execute_binding(struct seat *seat, struct terminal *term,
         return true;
     }
 
+    case BIND_ACTION_UNICODE_INPUT:
+        unicode_mode_activate(seat);
+        return true;
+
     case BIND_ACTION_SELECT_BEGIN:
         selection_start(
             term, seat->mouse.col, seat->mouse.row, SELECTION_CHAR_WISE, false);
@@ -1405,7 +1410,69 @@ key_press_release(struct seat *seat, struct terminal *term, uint32_t serial,
     xassert(bindings != NULL);
 
     if (pressed) {
-        if (term->is_searching) {
+        if (seat->unicode_mode.active) {
+            if (sym == XKB_KEY_Return ||
+                sym == XKB_KEY_space ||
+                sym == XKB_KEY_KP_Enter ||
+                sym == XKB_KEY_KP_Space)
+            {
+                char utf8[MB_CUR_MAX];
+                size_t chars = c32rtomb(
+                    utf8, seat->unicode_mode.character, &(mbstate_t){0});
+
+                LOG_DBG("Unicode input: 0x%06x -> %.*s",
+                        seat->unicode_mode.character, (int)chars, utf8);
+
+                if (chars != (size_t)-1) {
+                    if (term->is_searching)
+                        search_add_chars(term, utf8, chars);
+                    else
+                        term_to_slave(term, utf8, chars);
+                }
+
+                unicode_mode_deactivate(seat);
+            }
+
+            else if (sym == XKB_KEY_Escape ||
+                     (seat->kbd.ctrl && (sym == XKB_KEY_c ||
+                                         sym == XKB_KEY_d ||
+                                         sym == XKB_KEY_g)))
+            {
+                unicode_mode_deactivate(seat);
+            }
+
+            else if (sym == XKB_KEY_BackSpace) {
+                if (seat->unicode_mode.count > 0) {
+                    seat->unicode_mode.character >>= 4;
+                    seat->unicode_mode.count--;
+                    unicode_mode_updated(seat);
+                }
+            }
+
+            else if (seat->unicode_mode.count < 6) {
+                int digit = -1;
+
+                /* 0-9, a-f, A-F */
+                if (sym >= XKB_KEY_0 && sym <= XKB_KEY_9)
+                    digit = sym - XKB_KEY_0;
+                else if (sym >= XKB_KEY_a && sym <= XKB_KEY_f)
+                    digit = 0xa + (sym - XKB_KEY_a);
+                else if (sym >= XKB_KEY_A && sym <= XKB_KEY_F)
+                    digit = 0xa + (sym - XKB_KEY_A);
+
+                if (digit >= 0) {
+                    xassert(digit >= 0 && digit <= 0xf);
+                    seat->unicode_mode.character <<= 4;
+                    seat->unicode_mode.character |= digit;
+                    seat->unicode_mode.count++;
+                    unicode_mode_updated(seat);
+                }
+            }
+
+            return;
+        }
+
+        else if (term->is_searching) {
             if (should_repeat)
                 start_repeater(seat, key);
 

key-binding.h

@@ -38,6 +38,7 @@ enum bind_action_normal {
     BIND_ACTION_TEXT_BINDING,
     BIND_ACTION_PROMPT_PREV,
     BIND_ACTION_PROMPT_NEXT,
+    BIND_ACTION_UNICODE_INPUT,
 
     /* Mouse specific actions - i.e. they require a mouse coordinate */
     BIND_ACTION_SELECT_BEGIN,
@@ -48,7 +49,7 @@ enum bind_action_normal {
     BIND_ACTION_SELECT_WORD_WS,
     BIND_ACTION_SELECT_ROW,
 
-    BIND_ACTION_KEY_COUNT = BIND_ACTION_PROMPT_NEXT + 1,
+    BIND_ACTION_KEY_COUNT = BIND_ACTION_UNICODE_INPUT + 1,
     BIND_ACTION_COUNT = BIND_ACTION_SELECT_ROW + 1,
 };
 
@@ -72,6 +73,7 @@ enum bind_action_search {
     BIND_ACTION_SEARCH_EXTEND_WORD_WS,
     BIND_ACTION_SEARCH_CLIPBOARD_PASTE,
     BIND_ACTION_SEARCH_PRIMARY_PASTE,
+    BIND_ACTION_SEARCH_UNICODE_INPUT,
     BIND_ACTION_SEARCH_COUNT,
 };
 

meson.build

@@ -222,6 +222,7 @@ executable(
   'slave.c', 'slave.h',
   'spawn.c', 'spawn.h',
   'tokenize.c', 'tokenize.h',
+  'unicode-mode.c', 'unicode-mode.h',
   'url-mode.c', 'url-mode.h',
   'user-notification.c', 'user-notification.h',
   'wayland.c', 'wayland.h', 'shm-formats.h',

search.c

@@ -18,6 +18,7 @@
 #include "render.h"
 #include "selection.h"
 #include "shm.h"
+#include "unicode-mode.h"
 #include "util.h"
 #include "xmalloc.h"
 
@@ -993,6 +994,10 @@ execute_binding(struct seat *seat, struct terminal *term,
         *update_search_result = *redraw = true;
         return true;
 
+    case BIND_ACTION_SEARCH_UNICODE_INPUT:
+        unicode_mode_activate(seat);
+        return true;
+
     case BIND_ACTION_SEARCH_COUNT:
         BUG("Invalid action type");
         return true;

unicode-mode.c

@@ -0,0 +1,38 @@
+#include "unicode-mode.h"
+
+#include "render.h"
+
+void
+unicode_mode_activate(struct seat *seat)
+{
+    if (seat->unicode_mode.active)
+        return;
+
+    seat->unicode_mode.active = true;
+    seat->unicode_mode.character = u'\0';
+    seat->unicode_mode.count = 0;
+    unicode_mode_updated(seat);
+}
+
+void
+unicode_mode_deactivate(struct seat *seat)
+{
+    if (!seat->unicode_mode.active)
+        return;
+
+    seat->unicode_mode.active = false;
+    unicode_mode_updated(seat);
+}
+
+void
+unicode_mode_updated(struct seat *seat)
+{
+    struct terminal *term = seat->kbd_focus;
+    if (term == NULL)
+        return;
+
+    if (term->is_searching)
+        render_refresh_search(term);
+    else
+        render_refresh(term);
+}

unicode-mode.h

@@ -0,0 +1,7 @@
+#pragma once
+
+#include "wayland.h"
+
+void unicode_mode_activate(struct seat *seat);
+void unicode_mode_deactivate(struct seat *seat);
+void unicode_mode_updated(struct seat *seat);

wayland.h

@@ -206,6 +206,12 @@ struct seat {
         uint32_t serial;
     } ime;
 #endif
+
+    struct {
+        bool active;
+        int count;
+        char32_t character;
+    } unicode_mode;
 };
 
 enum csd_surface {