patch 9.0.1485: no functions for converting from/to UTF-16 indexv9.0.1485

Problem:    no functions for converting from/to UTF-16 index.
Solution:   Add UTF-16 flag to existing funtions and add strutf16len() and
            utf16idx(). (Yegappan Lakshmanan, closes #12216)

5 files changed, 559 insertions, 45 deletions

diff --git a/src/evalfunc.c b/src/evalfunc.c
index 5c10f1ec9..10d00d5a1 100644
--- a/src/evalfunc.c
+++ b/src/evalfunc.c
@@ -1751,9 +1751,9 @@ static funcentry_T global_functions[] =
 			ret_number,	    f_bufwinnr},
     {"byte2line",	1, 1, FEARG_1,	    arg1_number,
 			ret_number,	    f_byte2line},
-    {"byteidx",		2, 2, FEARG_1,	    arg2_string_number,
+    {"byteidx",		2, 3, FEARG_1,	    arg3_string_number_bool,
 			ret_number,	    f_byteidx},
-    {"byteidxcomp",	2, 2, FEARG_1,	    arg2_string_number,
+    {"byteidxcomp",	2, 3, FEARG_1,	    arg3_string_number_bool,
 			ret_number,	    f_byteidxcomp},
     {"call",		2, 3, FEARG_1,	    arg3_any_list_dict,
 			ret_any,	    f_call},
@@ -1803,7 +1803,7 @@ static funcentry_T global_functions[] =
 			ret_number,	    f_charclass},
     {"charcol",		1, 2, FEARG_1,	    arg2_string_or_list_number,
 			ret_number,	    f_charcol},
-    {"charidx",		2, 3, FEARG_1,	    arg3_string_number_bool,
+    {"charidx",		2, 4, FEARG_1,	    arg3_string_number_bool,
 			ret_number,	    f_charidx},
     {"chdir",		1, 1, FEARG_1,	    arg1_string,
 			ret_string,	    f_chdir},
@@ -2601,6 +2601,8 @@ static funcentry_T global_functions[] =
 			ret_number,	    f_strridx},
     {"strtrans",	1, 1, FEARG_1,	    arg1_string,
 			ret_string,	    f_strtrans},
+    {"strutf16len",	1, 2, FEARG_1,	    arg2_string_bool,
+			ret_number,	    f_strutf16len},
     {"strwidth",	1, 1, FEARG_1,	    arg1_string,
 			ret_number,	    f_strwidth},
     {"submatch",	1, 2, FEARG_1,	    arg2_number_bool,
@@ -2785,6 +2787,8 @@ static funcentry_T global_functions[] =
 			ret_dict_any,	    f_undotree},
     {"uniq",		1, 3, FEARG_1,	    arg13_sortuniq,
 			ret_first_arg,	    f_uniq},
+    {"utf16idx",	2, 4, FEARG_1,	    arg3_string_number_bool,
+			ret_number,	    f_utf16idx},
     {"values",		1, 1, FEARG_1,	    arg1_dict_any,
 			ret_list_member,    f_values},
     {"virtcol",		1, 2, FEARG_1,	    arg2_string_or_list_bool,
diff --git a/src/proto/strings.pro b/src/proto/strings.pro
index 602208831..a72e1ff5e 100644
--- a/src/proto/strings.pro
+++ b/src/proto/strings.pro
@@ -36,12 +36,14 @@ void f_string(typval_T *argvars, typval_T *rettv);
 void f_strlen(typval_T *argvars, typval_T *rettv);
 void f_strcharlen(typval_T *argvars, typval_T *rettv);
 void f_strchars(typval_T *argvars, typval_T *rettv);
+void f_strutf16len(typval_T *argvars, typval_T *rettv);
 void f_strdisplaywidth(typval_T *argvars, typval_T *rettv);
 void f_strwidth(typval_T *argvars, typval_T *rettv);
 void f_strcharpart(typval_T *argvars, typval_T *rettv);
 void f_strpart(typval_T *argvars, typval_T *rettv);
 void f_strridx(typval_T *argvars, typval_T *rettv);
 void f_strtrans(typval_T *argvars, typval_T *rettv);
+void f_utf16idx(typval_T *argvars, typval_T *rettv);
 void f_tolower(typval_T *argvars, typval_T *rettv);
 void f_toupper(typval_T *argvars, typval_T *rettv);
 void f_tr(typval_T *argvars, typval_T *rettv);
diff --git a/src/strings.c b/src/strings.c
index 7c868bf1f..7d4281dcd 100644
--- a/src/strings.c
+++ b/src/strings.c
@@ -1006,10 +1006,6 @@ string_reduce(
     static void
 byteidx(typval_T *argvars, typval_T *rettv, int comp UNUSED)
 {
-    char_u	*t;
-    char_u	*str;
-    varnumber_T	idx;
-
     rettv->vval.v_number = -1;
 
     if (in_vim9script()
@@ -1017,20 +1013,42 @@ byteidx(typval_T *argvars, typval_T *rettv, int comp UNUSED)
 		|| check_for_number_arg(argvars, 1) == FAIL))
 	return;
 
-    str = tv_get_string_chk(&argvars[0]);
-    idx = tv_get_number_chk(&argvars[1], NULL);
+    char_u *str = tv_get_string_chk(&argvars[0]);
+    varnumber_T	idx = tv_get_number_chk(&argvars[1], NULL);
     if (str == NULL || idx < 0)
 	return;
 
-    t = str;
+    varnumber_T	utf16idx = FALSE;
+    if (argvars[2].v_type != VAR_UNKNOWN)
+    {
+	utf16idx = tv_get_bool(&argvars[2]);
+	if (utf16idx < 0 || utf16idx > 1)
+	{
+	    semsg(_(e_using_number_as_bool_nr), utf16idx);
+	    return;
+	}
+    }
+
+    int (*ptr2len)(char_u *);
+    if (enc_utf8 && comp)
+	ptr2len = utf_ptr2len;
+    else
+	ptr2len = mb_ptr2len;
+
+    char_u *t = str;
     for ( ; idx > 0; idx--)
     {
 	if (*t == NUL)		// EOL reached
 	    return;
-	if (enc_utf8 && comp)
-	    t += utf_ptr2len(t);
-	else
-	    t += (*mb_ptr2len)(t);
+	if (utf16idx)
+	{
+	    int clen = ptr2len(t);
+	    int c = (clen > 1) ? utf_ptr2char(t) : *t;
+	    if (c > 0xFFFF)
+		idx--;
+	}
+	if (idx > 0)
+	    t += ptr2len(t);
     }
     rettv->vval.v_number = (varnumber_T)(t - str);
 }
@@ -1059,42 +1077,49 @@ f_byteidxcomp(typval_T *argvars, typval_T *rettv)
     void
 f_charidx(typval_T *argvars, typval_T *rettv)
 {
-    char_u	*str;
-    varnumber_T	idx;
-    varnumber_T	countcc = FALSE;
-    char_u	*p;
-    int		len;
-    int		(*ptr2len)(char_u *);
-
     rettv->vval.v_number = -1;
 
-    if ((check_for_string_arg(argvars, 0) == FAIL
+    if (check_for_string_arg(argvars, 0) == FAIL
 		|| check_for_number_arg(argvars, 1) == FAIL
-		|| check_for_opt_bool_arg(argvars, 2) == FAIL))
+		|| check_for_opt_bool_arg(argvars, 2) == FAIL
+		|| (argvars[2].v_type != VAR_UNKNOWN
+		    && check_for_opt_bool_arg(argvars, 3) == FAIL))
 	return;
 
-    str = tv_get_string_chk(&argvars[0]);
-    idx = tv_get_number_chk(&argvars[1], NULL);
+    char_u *str = tv_get_string_chk(&argvars[0]);
+    varnumber_T	idx = tv_get_number_chk(&argvars[1], NULL);
     if (str == NULL || idx < 0)
 	return;
 
+    varnumber_T	countcc = FALSE;
+    varnumber_T	utf16idx = FALSE;
     if (argvars[2].v_type != VAR_UNKNOWN)
-	countcc = tv_get_bool(&argvars[2]);
-    if (countcc < 0 || countcc > 1)
     {
-	semsg(_(e_using_number_as_bool_nr), countcc);
-	return;
+	countcc = tv_get_bool(&argvars[2]);
+	if (argvars[3].v_type != VAR_UNKNOWN)
+	    utf16idx = tv_get_bool(&argvars[3]);
     }
 
+    int (*ptr2len)(char_u *);
     if (enc_utf8 && countcc)
 	ptr2len = utf_ptr2len;
     else
 	ptr2len = mb_ptr2len;
 
-    for (p = str, len = 0; p <= str + idx; len++)
+    char_u	*p;
+    int		len;
+    for (p = str, len = 0; utf16idx ? idx >= 0 : p <= str + idx; len++)
     {
 	if (*p == NUL)
 	    return;
+	if (utf16idx)
+	{
+	    idx--;
+	    int clen = ptr2len(p);
+	    int c = (clen > 1) ? utf_ptr2char(p) : *p;
+	    if (c > 0xFFFF)
+		idx--;
+	}
 	p += ptr2len(p);
     }
 
@@ -1359,6 +1384,38 @@ f_strchars(typval_T *argvars, typval_T *rettv)
 }
 
 /*
+ * "strutf16len()" function
+ */
+    void
+f_strutf16len(typval_T *argvars, typval_T *rettv)
+{
+    rettv->vval.v_number = -1;
+
+    if (check_for_string_arg(argvars, 0) == FAIL
+	    || check_for_opt_bool_arg(argvars, 1) == FAIL)
+	return;
+
+    varnumber_T countcc = FALSE;
+    if (argvars[1].v_type != VAR_UNKNOWN)
+	countcc = tv_get_bool(&argvars[1]);
+
+    char_u		*s = tv_get_string(&argvars[0]);
+    varnumber_T		len = 0;
+    int			(*func_mb_ptr2char_adv)(char_u **pp);
+    int			ch;
+
+    func_mb_ptr2char_adv = countcc ? mb_cptr2char_adv : mb_ptr2char_adv;
+    while (*s != NUL)
+    {
+	ch = func_mb_ptr2char_adv(&s);
+	if (ch > 0xFFFF)
+	    ++len;
+	++len;
+    }
+    rettv->vval.v_number = len;
+}
+
+/*
  * "strdisplaywidth()" function
  */
     void
@@ -1619,6 +1676,61 @@ f_strtrans(typval_T *argvars, typval_T *rettv)
     rettv->vval.v_string = transstr(tv_get_string(&argvars[0]));
 }
 
+
+/*
+ *
+ * "utf16idx()" function
+ */
+    void
+f_utf16idx(typval_T *argvars, typval_T *rettv)
+{
+    rettv->vval.v_number = -1;
+
+    if (check_for_string_arg(argvars, 0) == FAIL
+	    || check_for_opt_number_arg(argvars, 1) == FAIL
+	    || check_for_opt_bool_arg(argvars, 2) == FAIL
+	    || (argvars[2].v_type != VAR_UNKNOWN
+		    && check_for_opt_bool_arg(argvars, 3) == FAIL))
+	    return;
+
+    char_u *str = tv_get_string_chk(&argvars[0]);
+    varnumber_T	idx = tv_get_number_chk(&argvars[1], NULL);
+    if (str == NULL || idx < 0)
+	return;
+
+    varnumber_T	countcc = FALSE;
+    varnumber_T	charidx = FALSE;
+    if (argvars[2].v_type != VAR_UNKNOWN)
+    {
+	countcc = tv_get_bool(&argvars[2]);
+	if (argvars[3].v_type != VAR_UNKNOWN)
+	    charidx = tv_get_bool(&argvars[3]);
+    }
+
+    int (*ptr2len)(char_u *);
+    if (enc_utf8 && countcc)
+	ptr2len = utf_ptr2len;
+    else
+	ptr2len = mb_ptr2len;
+
+    char_u	*p;
+    int		len;
+    for (p = str, len = 0; charidx ? idx >= 0 : p <= str + idx; len++)
+    {
+	if (*p == NUL)
+	    return;
+	int clen = ptr2len(p);
+	int c = (clen > 1) ? utf_ptr2char(p) : *p;
+	if (c > 0xFFFF)
+	    len++;
+	p += ptr2len(p);
+	if (charidx)
+	    idx--;
+    }
+
+    rettv->vval.v_number = len > 0 ? len - 1 : 0;
+}
+
 /*
  * "tolower(string)" function
  */
diff --git a/src/testdir/test_functions.vim b/src/testdir/test_functions.vim
index 3bea88df1..e32c4f5ff 100644
--- a/src/testdir/test_functions.vim
+++ b/src/testdir/test_functions.vim
@@ -1192,19 +1192,14 @@ func Test_byte2line_line2byte()
   bw!
 endfunc
 
-" Test for byteidx() and byteidxcomp() functions
+" Test for byteidx() using a character index
 func Test_byteidx()
   let a = '.é.' " one char of two bytes
   call assert_equal(0, byteidx(a, 0))
-  call assert_equal(0, byteidxcomp(a, 0))
   call assert_equal(1, byteidx(a, 1))
-  call assert_equal(1, byteidxcomp(a, 1))
   call assert_equal(3, byteidx(a, 2))
-  call assert_equal(3, byteidxcomp(a, 2))
   call assert_equal(4, byteidx(a, 3))
-  call assert_equal(4, byteidxcomp(a, 3))
   call assert_equal(-1, byteidx(a, 4))
-  call assert_equal(-1, byteidxcomp(a, 4))
 
   let b = '.é.' " normal e with composing char
   call assert_equal(0, b->byteidx(0))
@@ -1212,18 +1207,184 @@ func Test_byteidx()
   call assert_equal(4, b->byteidx(2))
   call assert_equal(5, b->byteidx(3))
   call assert_equal(-1, b->byteidx(4))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  call assert_equal(0, byteidx(str, 0))
+  call assert_equal(1, byteidx(str, 1))
+  call assert_equal(6, byteidx(str, 2))
+  call assert_equal(7, byteidx(str, 3))
+  call assert_equal(12, byteidx(str, 4))
+  call assert_equal(-1, byteidx(str, 5))
+
+  " empty string
+  call assert_equal(0, byteidx('', 0))
+  call assert_equal(-1, byteidx('', 1))
+
+  " error cases
   call assert_fails("call byteidx([], 0)", 'E730:')
+  call assert_fails("call byteidx('abc', [])", 'E745:')
+endfunc
+
+" Test for byteidxcomp() using a character index
+func Test_byteidxcomp()
+  let a = '.é.' " one char of two bytes
+  call assert_equal(0, byteidxcomp(a, 0))
+  call assert_equal(1, byteidxcomp(a, 1))
+  call assert_equal(3, byteidxcomp(a, 2))
+  call assert_equal(4, byteidxcomp(a, 3))
+  call assert_equal(-1, byteidxcomp(a, 4))
 
+  let b = '.é.' " normal e with composing char
   call assert_equal(0, b->byteidxcomp(0))
   call assert_equal(1, b->byteidxcomp(1))
   call assert_equal(2, b->byteidxcomp(2))
   call assert_equal(4, b->byteidxcomp(3))
   call assert_equal(5, b->byteidxcomp(4))
   call assert_equal(-1, b->byteidxcomp(5))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  call assert_equal(0, byteidxcomp(str, 0))
+  call assert_equal(1, byteidxcomp(str, 1))
+  call assert_equal(2, byteidxcomp(str, 2))
+  call assert_equal(4, byteidxcomp(str, 3))
+  call assert_equal(6, byteidxcomp(str, 4))
+  call assert_equal(7, byteidxcomp(str, 5))
+  call assert_equal(8, byteidxcomp(str, 6))
+  call assert_equal(10, byteidxcomp(str, 7))
+  call assert_equal(12, byteidxcomp(str, 8))
+  call assert_equal(-1, byteidxcomp(str, 9))
+
+  " empty string
+  call assert_equal(0, byteidxcomp('', 0))
+  call assert_equal(-1, byteidxcomp('', 1))
+
+  " error cases
   call assert_fails("call byteidxcomp([], 0)", 'E730:')
+  call assert_fails("call byteidxcomp('abc', [])", 'E745:')
 endfunc
 
-" Test for charidx()
+" Test for byteidx() using a UTF-16 index
+func Test_byteidx_from_utf16_index()
+  " string with single byte characters
+  let str = "abc"
+  for i in range(3)
+    call assert_equal(i, byteidx(str, i, v:true))
+  endfor
+  call assert_equal(3, byteidx(str, 3, v:true))
+  call assert_equal(-1, byteidx(str, 4, v:true))
+
+  " string with two byte characters
+  let str = "a©©b"
+  call assert_equal(0, byteidx(str, 0, v:true))
+  call assert_equal(1, byteidx(str, 1, v:true))
+  call assert_equal(3, byteidx(str, 2, v:true))
+  call assert_equal(5, byteidx(str, 3, v:true))
+  call assert_equal(6, byteidx(str, 4, v:true))
+  call assert_equal(-1, byteidx(str, 5, v:true))
+
+  " string with two byte characters
+  let str = "a😊😊b"
+  call assert_equal(0, byteidx(str, 0, v:true))
+  call assert_equal(1, byteidx(str, 1, v:true))
+  call assert_equal(1, byteidx(str, 2, v:true))
+  call assert_equal(5, byteidx(str, 3, v:true))
+  call assert_equal(5, byteidx(str, 4, v:true))
+  call assert_equal(9, byteidx(str, 5, v:true))
+  call assert_equal(10, byteidx(str, 6, v:true))
+  call assert_equal(-1, byteidx(str, 7, v:true))
+
+  " string with composing characters
+  let str = '-á-b́'
+  call assert_equal(0, byteidx(str, 0, v:true))
+  call assert_equal(1, byteidx(str, 1, v:true))
+  call assert_equal(4, byteidx(str, 2, v:true))
+  call assert_equal(5, byteidx(str, 3, v:true))
+  call assert_equal(8, byteidx(str, 4, v:true))
+  call assert_equal(-1, byteidx(str, 5, v:true))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  call assert_equal(0, byteidx(str, 0, v:true))
+  call assert_equal(1, byteidx(str, 1, v:true))
+  call assert_equal(6, byteidx(str, 2, v:true))
+  call assert_equal(7, byteidx(str, 3, v:true))
+  call assert_equal(12, byteidx(str, 4, v:true))
+  call assert_equal(-1, byteidx(str, 5, v:true))
+
+  " empty string
+  call assert_equal(0, byteidx('', 0, v:true))
+  call assert_equal(-1, byteidx('', 1, v:true))
+
+  " error cases
+  call assert_fails('call byteidx(str, 0, [])', 'E745:')
+endfunc
+
+" Test for byteidxcomp() using a UTF-16 index
+func Test_byteidxcomp_from_utf16_index()
+  " string with single byte characters
+  let str = "abc"
+  for i in range(3)
+    call assert_equal(i, byteidxcomp(str, i, v:true))
+  endfor
+  call assert_equal(3, byteidxcomp(str, 3, v:true))
+  call assert_equal(-1, byteidxcomp(str, 4, v:true))
+
+  " string with two byte characters
+  let str = "a©©b"
+  call assert_equal(0, byteidxcomp(str, 0, v:true))
+  call assert_equal(1, byteidxcomp(str, 1, v:true))
+  call assert_equal(3, byteidxcomp(str, 2, v:true))
+  call assert_equal(5, byteidxcomp(str, 3, v:true))
+  call assert_equal(6, byteidxcomp(str, 4, v:true))
+  call assert_equal(-1, byteidxcomp(str, 5, v:true))
+
+  " string with two byte characters
+  let str = "a😊😊b"
+  call assert_equal(0, byteidxcomp(str, 0, v:true))
+  call assert_equal(1, byteidxcomp(str, 1, v:true))
+  call assert_equal(1, byteidxcomp(str, 2, v:true))
+  call assert_equal(5, byteidxcomp(str, 3, v:true))
+  call assert_equal(5, byteidxcomp(str, 4, v:true))
+  call assert_equal(9, byteidxcomp(str, 5, v:true))
+  call assert_equal(10, byteidxcomp(str, 6, v:true))
+  call assert_equal(-1, byteidxcomp(str, 7, v:true))
+
+  " string with composing characters
+  let str = '-á-b́'
+  call assert_equal(0, byteidxcomp(str, 0, v:true))
+  call assert_equal(1, byteidxcomp(str, 1, v:true))
+  call assert_equal(2, byteidxcomp(str, 2, v:true))
+  call assert_equal(4, byteidxcomp(str, 3, v:true))
+  call assert_equal(5, byteidxcomp(str, 4, v:true))
+  call assert_equal(6, byteidxcomp(str, 5, v:true))
+  call assert_equal(8, byteidxcomp(str, 6, v:true))
+  call assert_equal(-1, byteidxcomp(str, 7, v:true))
+  call assert_fails('call byteidxcomp(str, 0, [])', 'E745:')
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  call assert_equal(0, byteidxcomp(str, 0, v:true))
+  call assert_equal(1, byteidxcomp(str, 1, v:true))
+  call assert_equal(2, byteidxcomp(str, 2, v:true))
+  call assert_equal(4, byteidxcomp(str, 3, v:true))
+  call assert_equal(6, byteidxcomp(str, 4, v:true))
+  call assert_equal(7, byteidxcomp(str, 5, v:true))
+  call assert_equal(8, byteidxcomp(str, 6, v:true))
+  call assert_equal(10, byteidxcomp(str, 7, v:true))
+  call assert_equal(12, byteidxcomp(str, 8, v:true))
+  call assert_equal(-1, byteidxcomp(str, 9, v:true))
+
+  " empty string
+  call assert_equal(0, byteidxcomp('', 0, v:true))
+  call assert_equal(-1, byteidxcomp('', 1, v:true))
+
+  " error cases
+  call assert_fails('call byteidxcomp(str, 0, [])', 'E745:')
+endfunc
+
+" Test for charidx() using a byte index
 func Test_charidx()
   let a = 'xáb́y'
   call assert_equal(0, charidx(a, 0))
@@ -1232,17 +1393,20 @@ func Test_charidx()
   call assert_equal(3, charidx(a, 7))
   call assert_equal(-1, charidx(a, 8))
   call assert_equal(-1, charidx(a, -1))
-  call assert_equal(-1, charidx('', 0))
-  call assert_equal(-1, charidx(test_null_string(), 0))
 
   " count composing characters
-  call assert_equal(0, charidx(a, 0, 1))
-  call assert_equal(2, charidx(a, 2, 1))
-  call assert_equal(3, charidx(a, 4, 1))
-  call assert_equal(5, charidx(a, 7, 1))
-  call assert_equal(-1, charidx(a, 8, 1))
+  call assert_equal(0, a->charidx(0, 1))
+  call assert_equal(2, a->charidx(2, 1))
+  call assert_equal(3, a->charidx(4, 1))
+  call assert_equal(5, a->charidx(7, 1))
+  call assert_equal(-1, a->charidx(8, 1))
+
+  " empty string
+  call assert_equal(-1, charidx('', 0))
   call assert_equal(-1, charidx('', 0, 1))
 
+  " error cases
+  call assert_equal(-1, charidx(test_null_string(), 0))
   call assert_fails('let x = charidx([], 1)', 'E1174:')
   call assert_fails('let x = charidx("abc", [])', 'E1210:')
   call assert_fails('let x = charidx("abc", 1, [])', 'E1212:')
@@ -1250,6 +1414,237 @@ func Test_charidx()
   call assert_fails('let x = charidx("abc", 1, 2)', 'E1212:')
 endfunc
 
+" Test for charidx() using a UTF-16 index
+func Test_charidx_from_utf16_index()
+  " string with single byte characters
+  let str = "abc"
+  for i in range(3)
+    call assert_equal(i, charidx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, charidx(str, 3, v:false, v:true))
+
+  " string with two byte characters
+  let str = "a©©b"
+  call assert_equal(0, charidx(str, 0, v:false, v:true))
+  call assert_equal(1, charidx(str, 1, v:false, v:true))
+  call assert_equal(2, charidx(str, 2, v:false, v:true))
+  call assert_equal(3, charidx(str, 3, v:false, v:true))
+  call assert_equal(-1, charidx(str, 4, v:false, v:true))
+
+  " string with four byte characters
+  let str = "a😊😊b"
+  call assert_equal(0, charidx(str, 0, v:false, v:true))
+  call assert_equal(1, charidx(str, 1, v:false, v:true))
+  call assert_equal(1, charidx(str, 2, v:false, v:true))
+  call assert_equal(2, charidx(str, 3, v:false, v:true))
+  call assert_equal(2, charidx(str, 4, v:false, v:true))
+  call assert_equal(3, charidx(str, 5, v:false, v:true))
+  call assert_equal(-1, charidx(str, 6, v:false, v:true))
+
+  " string with composing characters
+  let str = '-á-b́'
+  for i in str->strcharlen()->range()
+    call assert_equal(i, charidx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, charidx(str, 4, v:false, v:true))
+  for i in str->strchars()->range()
+    call assert_equal(i, charidx(str, i, v:true, v:true))
+  endfor
+  call assert_equal(-1, charidx(str, 6, v:true, v:true))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  for i in str->strcharlen()->range()
+    call assert_equal(i, charidx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, charidx(str, 4, v:false, v:true))
+  for i in str->strchars()->range()
+    call assert_equal(i, charidx(str, i, v:true, v:true))
+  endfor
+  call assert_equal(-1, charidx(str, 8, v:true, v:true))
+
+  " empty string
+  call assert_equal(-1, charidx('', 0, v:false, v:true))
+  call assert_equal(-1, charidx('', 0, v:true, v:true))
+
+  " error cases
+  call assert_equal(-1, charidx('', 0, v:false, v:true))
+  call assert_equal(-1, charidx('', 0, v:true, v:true))
+  call assert_equal(-1, charidx(test_null_string(), 0, v:false, v:true))
+  call assert_fails('let x = charidx("abc", 1, v:false, [])', 'E1212:')
+  call assert_fails('let x = charidx("abc", 1, v:true, [])', 'E1212:')
+endfunc
+
+" Test for utf16idx() using a byte index
+func Test_utf16idx_from_byteidx()
+  " UTF-16 index of a string with single byte characters
+  let str = "abc"
+  for i in range(3)
+    call assert_equal(i, utf16idx(str, i))
+  endfor
+  call assert_equal(-1, utf16idx(str, 3))
+
+  " UTF-16 index of a string with two byte characters
+  let str = 'a©©b'
+  call assert_equal(0, str->utf16idx(0))
+  call assert_equal(1, str->utf16idx(1))
+  call assert_equal(1, str->utf16idx(2))
+  call assert_equal(2, str->utf16idx(3))
+  call assert_equal(2, str->utf16idx(4))
+  call assert_equal(3, str->utf16idx(5))
+  call assert_equal(-1, str->utf16idx(6))
+
+  " UTF-16 index of a string with four byte characters
+  let str = 'a😊😊b'
+  call assert_equal(0, utf16idx(str, 0))
+  call assert_equal(2, utf16idx(str, 1))
+  call assert_equal(2, utf16idx(str, 2))
+  call assert_equal(2, utf16idx(str, 3))
+  call assert_equal(2, utf16idx(str, 4))
+  call assert_equal(4, utf16idx(str, 5))
+  call assert_equal(4, utf16idx(str, 6))
+  call assert_equal(4, utf16idx(str, 7))
+  call assert_equal(4, utf16idx(str, 8))
+  call assert_equal(5, utf16idx(str, 9))
+  call assert_equal(-1, utf16idx(str, 10))
+
+  " UTF-16 index of a string with composing characters
+  let str = '-á-b́'
+  call assert_equal(0, utf16idx(str, 0))
+  call assert_equal(1, utf16idx(str, 1))
+  call assert_equal(1, utf16idx(str, 2))
+  call assert_equal(1, utf16idx(str, 3))
+  call assert_equal(2, utf16idx(str, 4))
+  call assert_equal(3, utf16idx(str, 5))
+  call assert_equal(3, utf16idx(str, 6))
+  call assert_equal(3, utf16idx(str, 7))
+  call assert_equal(-1, utf16idx(str, 8))
+  call assert_equal(0, utf16idx(str, 0, v:true))
+  call assert_equal(1, utf16idx(str, 1, v:true))
+  call assert_equal(2, utf16idx(str, 2, v:true))
+  call assert_equal(2, utf16idx(str, 3, v:true))
+  call assert_equal(3, utf16idx(str, 4, v:true))
+  call assert_equal(4, utf16idx(str, 5, v:true))
+  call assert_equal(5, utf16idx(str, 6, v:true))
+  call assert_equal(5, utf16idx(str, 7, v:true))
+  call assert_equal(-1, utf16idx(str, 8, v:true))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  call assert_equal(0, utf16idx(str, 0))
+  call assert_equal(1, utf16idx(str, 1))
+  call assert_equal(1, utf16idx(str, 2))
+  call assert_equal(1, utf16idx(str, 3))
+  call assert_equal(1, utf16idx(str, 4))
+  call assert_equal(1, utf16idx(str, 5))
+  call assert_equal(2, utf16idx(str, 6))
+  call assert_equal(3, utf16idx(str, 7))
+  call assert_equal(3, utf16idx(str, 8))
+  call assert_equal(3, utf16idx(str, 9))
+  call assert_equal(3, utf16idx(str, 10))
+  call assert_equal(3, utf16idx(str, 11))
+  call assert_equal(-1, utf16idx(str, 12))
+  call assert_equal(0, utf16idx(str, 0, v:true))
+  call assert_equal(1, utf16idx(str, 1, v:true))
+  call assert_equal(2, utf16idx(str, 2, v:true))
+  call assert_equal(2, utf16idx(str, 3, v:true))
+  call assert_equal(3, utf16idx(str, 4, v:true))
+  call assert_equal(3, utf16idx(str, 5, v:true))
+  call assert_equal(4, utf16idx(str, 6, v:true))
+  call assert_equal(5, utf16idx(str, 7, v:true))
+  call assert_equal(6, utf16idx(str, 8, v:true))
+  call assert_equal(6, utf16idx(str, 9, v:true))
+  call assert_equal(7, utf16idx(str, 10, v:true))
+  call assert_equal(7, utf16idx(str, 11, v:true))
+  call assert_equal(-1, utf16idx(str, 12, v:true))
+
+  " empty string
+  call assert_equal(-1, utf16idx('', 0))
+  call assert_equal(-1, utf16idx('', 0, v:true))
+
+  " error cases
+  call assert_equal(-1, utf16idx("", 0))
+  call assert_equal(-1, utf16idx("abc", -1))
+  call assert_equal(-1, utf16idx(test_null_string(), 0))
+  call assert_fails('let l = utf16idx([], 0)', 'E1174:')
+  call assert_fails('let l = utf16idx("ab", [])', 'E1210:')
+  call assert_fails('let l = utf16idx("ab", 0, [])', 'E1212:')
+endfunc
+
+" Test for utf16idx() using a character index
+func Test_utf16idx_from_charidx()
+  let str = "abc"
+  for i in str->strcharlen()->range()
+    call assert_equal(i, utf16idx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 3, v:false, v:true))
+
+  " UTF-16 index of a string with two byte characters
+  let str = "a©©b"
+  for i in str->strcharlen()->range()
+    call assert_equal(i, utf16idx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 4, v:false, v:true))
+
+  " UTF-16 index of a string with four byte characters
+  let str = "a😊😊b"
+  call assert_equal(0, utf16idx(str, 0, v:false, v:true))
+  call assert_equal(2, utf16idx(str, 1, v:false, v:true))
+  call assert_equal(4, utf16idx(str, 2, v:false, v:true))
+  call assert_equal(5, utf16idx(str, 3, v:false, v:true))
+  call assert_equal(-1, utf16idx(str, 4, v:false, v:true))
+
+  " UTF-16 index of a string with composing characters
+  let str = '-á-b́'
+  for i in str->strcharlen()->range()
+    call assert_equal(i, utf16idx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 4, v:false, v:true))
+  for i in str->strchars()->range()
+    call assert_equal(i, utf16idx(str, i, v:true, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 6, v:true, v:true))
+
+  " string with multiple composing characters
+  let str = '-ą́-ą́'
+  for i in str->strcharlen()->range()
+    call assert_equal(i, utf16idx(str, i, v:false, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 4, v:false, v:true))
+  for i in str->strchars()->range()
+    call assert_equal(i, utf16idx(str, i, v:true, v:true))
+  endfor
+  call assert_equal(-1, utf16idx(str, 8, v:true, v:true))
+
+  " empty string
+  call assert_equal(-1, utf16idx('', 0, v:false, v:true))
+  call assert_equal(-1, utf16idx('', 0, v:true, v:true))
+
+  " error cases
+  call assert_equal(-1, utf16idx(test_null_string(), 0, v:true, v:true))
+  call assert_fails('let l = utf16idx("ab", 0, v:false, [])', 'E1212:')
+endfunc
+
+" Test for strutf16len()
+func Test_strutf16len()
+  call assert_equal(3, strutf16len('abc'))
+  call assert_equal(3, 'abc'->strutf16len(v:true))
+  call assert_equal(4, strutf16len('a©©b'))
+  call assert_equal(4, strutf16len('a©©b', v:true))
+  call assert_equal(6, strutf16len('a😊😊b'))
+  call assert_equal(6, strutf16len('a😊😊b', v:true))
+  call assert_equal(4, strutf16len('-á-b́'))
+  call assert_equal(6, strutf16len('-á-b́', v:true))
+  call assert_equal(4, strutf16len('-ą́-ą́'))
+  call assert_equal(8, strutf16len('-ą́-ą́', v:true))
+  call assert_equal(0, strutf16len(''))
+
+  " error cases
+  call assert_fails('let l = strutf16len([])', 'E1174:')
+  call assert_fails('let l = strutf16len("a", [])', 'E1212:')
+  call assert_equal(0, strutf16len(test_null_string()))
+endfunc
+
 func Test_count()
   let l = ['a', 'a', 'A', 'b']
   call assert_equal(2, count(l, 'a'))
@@ -3074,5 +3469,4 @@ func Test_delfunc_while_listing()
   call StopVimInTerminal(buf)
 endfunc
 
-
 " vim: shiftwidth=2 sts=2 expandtab
diff --git a/src/version.c b/src/version.c
index e5099dd47..0ce90ebc5 100644
--- a/src/version.c
+++ b/src/version.c
@@ -696,6 +696,8 @@ static char *(features[]) =
 static int included_patches[] =
 {   /* Add new patch number below this line */
 /**/
+    1485,
+/**/
     1484,
 /**/
     1483,