gmarkup: Fix unvalidated UTF-8 read in markup parsing error paths

When formatting the error messages for markup parsing errors, the parser
was unconditionally reading a UTF-8 character from the input buffer —
but the buffer might end with a partial code sequence, resulting in
reading off the end of the buffer by up to three bytes.

Fix this and add a test case, courtesy of pdknsk.
Signed-off-by: Philip Withnall <withnall@endlessm.com>

https://gitlab.gnome.org/GNOME/glib/issues/1462

glib/gmarkup.c

@@ -557,11 +557,23 @@ char_str (gunichar c,
   return buf;
 }
 
+/* Format the next UTF-8 character as a gchar* for printing in error output
+ * when we encounter a syntax error. This correctly handles invalid UTF-8,
+ * emitting it as hex escapes. */
 static gchar*
 utf8_str (const gchar *utf8,
           gchar       *buf)
 {
-  char_str (g_utf8_get_char (utf8), buf);
+  gunichar c = g_utf8_get_char_validated (utf8, -1);
+  if (c == (gunichar) -1 || c == (gunichar) -2)
+    {
+      gchar *temp = g_strdup_printf ("\\x%02x", (guint)(guchar)*utf8);
+      memset (buf, 0, 8);
+      memcpy (buf, temp, strlen (temp));
+      g_free (temp);
+    }
+  else
+    char_str (c, buf);
   return buf;
 }
 

glib/tests/Makefile.am

@@ -158,7 +158,7 @@ markup_tests = \
 	fail-31 fail-32 fail-33 fail-34 fail-35 \
 	fail-36 fail-37 fail-38 fail-39 fail-40 \
 	fail-41 fail-42 fail-43 fail-44 fail-45 \
-	fail-46 fail-47 fail-48 fail-49 \
+	fail-46 fail-47 fail-48 fail-49 fail-50 \
 	valid-1 valid-2 valid-3 valid-4 valid-5 \
 	valid-6 valid-7 valid-8 valid-9 valid-10 \
 	valid-11 valid-12 valid-13 valid-14 valid-15 \

glib/tests/markups/fail-50.expected

+ERROR Error on line 1 char 5: Odd character “\xfc”, expected an open quote mark after the equals sign when giving value for attribute “r” of element “”

glib/tests/markups/fail-50.gmarkup

+<	r=
\ No newline at end of file