Expand \u and \U escape sequences in command strings, from Christopher

Hunt in GitHub issue 1443.

2 files changed, 33 insertions, 7 deletions

diff --git a/cmd-string.c b/cmd-string.c
index 9c8f10c1..440a0231 100644
--- a/cmd-string.c
+++ b/cmd-string.c
@@ -54,6 +54,22 @@ cmd_string_ungetc(size_t *p)
 	(*p)--;
 }
 
+static int
+cmd_string_unicode(wchar_t *wc, const char *s, size_t *p, char ch)
+{
+	int	size = (ch == 'u') ? 4 : 8;
+	u_int	tmp;
+
+	if (size == 4 && sscanf(s + *p, "%4x", &tmp) != 1)
+		return (-1);
+	if (size == 8 && sscanf(s + *p, "%8x", &tmp) != 1)
+		return (-1);
+	*p += size;
+
+	*wc = (wchar_t)tmp;
+	return (0);
+}
+
 int
 cmd_string_split(const char *s, int *rargc, char ***rargv)
 {
@@ -191,12 +207,11 @@ cmd_string_copy(char **dst, char *src, size_t *len)
 static char *
 cmd_string_string(const char *s, size_t *p, char endch, int esc)
 {
-	int	ch;
-	char   *buf, *t;
-	size_t	len;
-
-	buf = NULL;
-	len = 0;
+	int			ch;
+	wchar_t			wc;
+	struct utf8_data	ud;
+	char		       *buf = NULL, *t;
+	size_t			len = 0;
 
 	while ((ch = cmd_string_getc(s, p)) != endch) {
 		switch (ch) {
@@ -220,6 +235,18 @@ cmd_string_string(const char *s, size_t *p, char endch, int esc)
 			case 't':
 				ch = '\t';
 				break;
+			case 'u':
+			case 'U':
+				if (cmd_string_unicode(&wc, s, p, ch) != 0)
+					goto error;
+				if (utf8_split(wc, &ud) != UTF8_DONE)
+					goto error;
+				if (len >= SIZE_MAX - ud.size - 1)
+					goto error;
+				buf = xrealloc(buf, len + ud.size);
+				memcpy(buf + len, ud.data, ud.size);
+				len += ud.size;
+				continue;
 			}
 			break;
 		case '$':
diff --git a/window-tree.c b/window-tree.c
index 52ab195e..2eb31612 100644
--- a/window-tree.c
+++ b/window-tree.c
@@ -461,7 +461,6 @@ window_tree_build(void *modedata, u_int sort_type, uint64_t *tag,
 	}
 }
 
-
 static void
 window_tree_draw_label(struct screen_write_ctx *ctx, u_int px, u_int py,
     u_int sx, u_int sy, const struct grid_cell *gc, const char *label)