fix: preserve emoji commits through XIM

tests/xim_test.c

@@ -1,8 +1,12 @@
 #include <assert.h>
 #include <stddef.h>
 #include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <xcb-imdkit/encoding.h>
 #include <xkbcommon/xkbcommon-keysyms.h>
 #include "../xim/keymap.h"
+#include "../xim/ximtext.h"
 
 enum
 {
@@ -11,6 +15,23 @@ enum
 	Group1 = 1<<13,
 };
 
+static void
+checktext(const char *s)
+{
+	char *ct, *utf8;
+	size_t clen, len, ulen;
+
+	len = strlen(s);
+	ct = ximcompound(s, len, &clen);
+	assert(ct != NULL);
+	utf8 = xcb_compound_text_to_utf8(ct, clen, &ulen);
+	assert(utf8 != NULL);
+	assert(ulen == len);
+	assert(memcmp(utf8, s, len) == 0);
+	free(utf8);
+	free(ct);
+}
+
 int
 main(void)
 {
@@ -23,6 +44,8 @@ main(void)
 		XKB_KEY_a, XKB_KEY_A, XKB_KEY_NoSymbol, XKB_KEY_NoSymbol,
 	};
 
+	xcb_compound_text_init();
+
 	assert(keymaplookup(letters, 4, 0) == XKB_KEY_a);
 	assert(keymaplookup(letters, 4, ShiftMask) == XKB_KEY_A);
 	assert(keymaplookup(letters, 4, LockMask) == XKB_KEY_A);
@@ -34,5 +57,9 @@ main(void)
 	assert(keymaplookup(punctuation, 2, LockMask|ShiftMask) == XKB_KEY_exclam);
 	assert(keymaplookup(missing, 4, Group1) == XKB_KEY_a);
 	assert(keymaplookup(NULL, 0, 0) == XKB_KEY_NoSymbol);
+	checktext("한글");
+	checktext("😀");
+	checktext("❤️");
+	checktext("A😀한");
 	return 0;
 }

xim/Makefile

@@ -15,8 +15,8 @@ $(PROG): $(OBJS) ../ipc.c ../ipc.h
 
 $(OBJS): ../ipc.h
 
-$(TEST): ../tests/xim_test.c keymap.c keymap.h
+$(TEST): ../tests/xim_test.c keymap.c keymap.h ximtext.c ximtext.h
-	$(CC) $(CFLAGS) -o $@ ../tests/xim_test.c keymap.c $(XIM_LIBS)
+	$(CC) $(CFLAGS) -o $@ ../tests/xim_test.c keymap.c ximtext.c $(XIM_LIBS)
 
 check: $(TEST)
 	./$(TEST)

xim/main.c

@@ -8,6 +8,7 @@
 #include <xcb-imdkit/encoding.h>
 #include "ipc.h"
 #include "keymap.h"
+#include "ximtext.h"
 
 typedef struct Ic Ic;
 struct Ic
@@ -97,7 +98,7 @@ commit(xcb_im_input_context_t *ic, char *s, int len)
 
 	if(len == 0)
 		return;
-	ct = xcb_utf8_to_compound_text(s, len, &clen);
+	ct = ximcompound(s, len, &clen);
 	if(ct == NULL)
 		return;
 	xcb_im_commit_string(xim, ic, XCB_XIM_LOOKUP_CHARS, ct, clen, 0);

xim/ximtext.c

@@ -0,0 +1,28 @@
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <xcb-imdkit/encoding.h>
+#include "ximtext.h"
+
+char*
+ximcompound(const char *s, size_t len, size_t *outlen)
+{
+	static const char utf8[] = "\033%G";
+	char *ct;
+
+	if(s == NULL || len > SIZE_MAX - sizeof utf8)
+		return NULL;
+	ct = xcb_utf8_to_compound_text(s, len, outlen);
+	if(ct != NULL)
+		return ct;
+	/* X11 Compound Text reserves ESC % G for an embedded UTF-8 run. */
+	ct = malloc(len + sizeof utf8);
+	if(ct == NULL)
+		return NULL;
+	memcpy(ct, utf8, sizeof utf8 - 1);
+	memcpy(ct + sizeof utf8 - 1, s, len);
+	ct[len + sizeof utf8 - 1] = '\0';
+	if(outlen != NULL)
+		*outlen = len + sizeof utf8 - 1;
+	return ct;
+}

xim/ximtext.h

@@ -0,0 +1,3 @@
+#include <stddef.h>
+
+char *ximcompound(const char*, size_t, size_t*);