From 4e8720a40dff785e1e79e89e8985072e5d95d967 Mon Sep 17 00:00:00 2001
From: Hojun-Cho <ghwns6743@gmail.com>
Date: Wed, 12 Aug 2026 16:43:57 +0900
Subject: [PATCH] fix: terminate XIM UTF-8 text runs

---
 tests/xim_test.c | 21 +++++++++++++++++++++
 xim/ximtext.c    | 38 ++++++++++++++++++++++++++------------
 xim/ximtext.h    |  1 +
 3 files changed, 48 insertions(+), 12 deletions(-)

diff --git a/tests/xim_test.c b/tests/xim_test.c
index ffb2aee..17ea467 100644
--- a/tests/xim_test.c
+++ b/tests/xim_test.c
@@ -32,6 +32,26 @@ checktext(const char *s)
 	free(ct);
 }
 
+static void
+checkfallback(void)
+{
+	static const char s[] = "😀";
+	static const char want[] = "\033%G😀\033%@";
+	char *ct, *utf8;
+	size_t clen, ulen;
+
+	ct = ximutf8compound(s, sizeof s - 1, &clen);
+	assert(ct != NULL);
+	assert(clen == sizeof want - 1);
+	assert(memcmp(ct, want, clen) == 0);
+	utf8 = xcb_compound_text_to_utf8(ct, clen, &ulen);
+	assert(utf8 != NULL);
+	assert(ulen == sizeof s - 1);
+	assert(memcmp(utf8, s, ulen) == 0);
+	free(utf8);
+	free(ct);
+}
+
 int
 main(void)
 {
@@ -61,5 +81,6 @@ main(void)
 	checktext("😀");
 	checktext("❤️");
 	checktext("A😀한");
+	checkfallback();
 	return 0;
 }
diff --git a/xim/ximtext.c b/xim/ximtext.c
index 5e59db2..f6087b5 100644
--- a/xim/ximtext.c
+++ b/xim/ximtext.c
@@ -4,25 +4,39 @@
 #include <xcb-imdkit/encoding.h>
 #include "ximtext.h"
 
+char*
+ximutf8compound(const char *s, size_t len, size_t *outlen)
+{
+	static const char begin[] = "\033%G";
+	static const char end[] = "\033%@";
+	char *ct;
+	size_t n;
+
+	if(s == NULL || len > SIZE_MAX - (sizeof begin + sizeof end - 1))
+		return NULL;
+	n = len + sizeof begin + sizeof end - 2;
+	ct = malloc(n + 1);
+	if(ct == NULL)
+		return NULL;
+	memcpy(ct, begin, sizeof begin - 1);
+	memcpy(ct + sizeof begin - 1, s, len);
+	memcpy(ct + sizeof begin - 1 + len, end, sizeof end - 1);
+	ct[n] = '\0';
+	if(outlen != NULL)
+		*outlen = n;
+	return ct;
+}
+
 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)
+	if(s == NULL)
 		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;
+	/* X.Org Compound Text extension for characters in no legacy charset. */
+	return ximutf8compound(s, len, outlen);
 }
diff --git a/xim/ximtext.h b/xim/ximtext.h
index 0e793af..4f1ed4b 100644
--- a/xim/ximtext.h
+++ b/xim/ximtext.h
@@ -1,3 +1,4 @@
 #include <stddef.h>
 
 char *ximcompound(const char*, size_t, size_t*);
+char *ximutf8compound(const char*, size_t, size_t*);