builtin: fix rune.to_title(), add tests (#22808)

vlib/builtin/rune.v

@@ -106,7 +106,13 @@ pub fn (c rune) to_lower() rune {
 
 // `to_title` convert to title mode.
 pub fn (c rune) to_title() rune {
-	return c.to_upper()
+	if c < 0x80 {
+		if c >= `a` && c <= `z` {
+			return c - 32
+		}
+		return c
+	}
+	return c.map_to(.to_title)
 }
 
 // `map_to` rune map mode: .to_upper/.to_lower/.to_title
@@ -119,19 +125,27 @@ fn (c rune) map_to(mode MapMode) rune {
 		middle := (start + end) / 2
 		cur_map := unsafe { &rune_maps[middle * rune_maps_columns_in_row] }
 		if c >= u32(unsafe { *cur_map }) && c <= u32(unsafe { *(cur_map + 1) }) {
-			offset := if mode == .to_upper {
+			offset := if mode in [.to_upper, .to_title] {
 				unsafe { *(cur_map + 2) }
 			} else {
 				unsafe { *(cur_map + 3) }
 			}
 			if offset == rune_maps_ul {
 				is_odd := (c - unsafe { *cur_map }) % 2 == 1
-				if mode == .to_upper && is_odd {
+				if mode in [.to_upper, .to_title] && is_odd {
 					return c - 1
 				} else if mode == .to_lower && !is_odd {
 					return c + 1
 				}
 				return c
+			} else if offset == rune_maps_utl {
+				cnt := (c - unsafe { *cur_map }) % 3
+				if mode == .to_upper {
+					return c - cnt
+				} else if mode == .to_lower {
+					return c + 2 - cnt
+				}
+				return c + 1 - cnt
 			}
 			return c + offset
 		}

vlib/builtin/rune_map.v

@@ -3,6 +3,7 @@ module builtin
 enum MapMode {
 	to_upper
 	to_lower
+	to_title
 }
 
 // vfmt off
@@ -16,7 +17,10 @@ struct RuneMap {
 }
 */
 const rune_maps_columns_in_row = 4
+// upper, lower, upper, lower, ... sequence
 const rune_maps_ul = -3 // NOTE: this should *NOT* be used anywhere in rune_maps, as a normal offset.
+// upper, title, lower, upper, title, lower, ... sequence
+const rune_maps_utl = -2 // NOTE: this should *NOT* be used anywhere in rune_maps, as a normal offset.
 // The rune_maps table below, has rows, each containing 4 integers, equivalent to the RuneMap struct from above.
 // It is represented that way, instead of the more natural array of structs, to save on the .c encoding used for the initialisation.
 // The overhead for representing it as an array of structs was ~28KB in .c, while with the flat array of ints, it is ~7.5KB.
@@ -72,21 +76,11 @@ const rune_maps = [
 	0x1B8, 0x1B9, -3, -3,
 	0x1BC, 0x1BD, -3, -3,
 	0x1BF, 0x1BF, 56, 0,
-	0x1C4, 0x1C4, 0, 2,
+	0x1C4, 0x1CC, -2, -2,
-	0x1C5, 0x1C5, -1, 1,
-	0x1C6, 0x1C6, -2, 0,
-	0x1C7, 0x1C7, 0, 2,
-	0x1C8, 0x1C8, -1, 1,
-	0x1C9, 0x1C9, -2, 0,
-	0x1CA, 0x1CA, 0, 2,
-	0x1CB, 0x1CB, -1, 1,
-	0x1CC, 0x1CC, -2, 0,
 	0x1CD, 0x1DC, -3, -3,
 	0x1DD, 0x1DD, -79, 0,
 	0x1DE, 0x1EF, -3, -3,
-	0x1F1, 0x1F1, 0, 2,
+	0x1F1, 0x1F3, -2, -2,
-	0x1F2, 0x1F2, -1, 1,
-	0x1F3, 0x1F3, -2, 0,
 	0x1F4, 0x1F5, -3, -3,
 	0x1F6, 0x1F6, 0, -97,
 	0x1F7, 0x1F7, 0, -56,

vlib/builtin/rune_test.v

@@ -50,6 +50,9 @@ fn test_to_upper() {
 	assert `Ā`.to_upper() == `Ā`
 	assert `Я`.to_upper() == `Я`
 	assert `я`.to_upper() == `Я`
+	assert `Dž`.to_upper() == `DŽ`
+	assert `dž`.to_upper() == `DŽ`
+	assert `DŽ`.to_upper() == `DŽ`
 }
 
 fn test_to_lower() {
@@ -61,6 +64,9 @@ fn test_to_lower() {
 	assert `ā`.to_lower() == `ā`
 	assert `я`.to_lower() == `я`
 	assert `Я`.to_lower() == `я`
+	assert `Dž`.to_lower() == `dž`
+	assert `DŽ`.to_lower() == `dž`
+	assert `dž`.to_lower() == `dž`
 }
 
 fn test_to_title() {
@@ -72,4 +78,7 @@ fn test_to_title() {
 	assert `Ā`.to_title() == `Ā`
 	assert `я`.to_title() == `Я`
 	assert `Я`.to_title() == `Я`
+	assert `DŽ`.to_title() == `Dž`
+	assert `dž`.to_title() == `Dž`
+	assert `Dž`.to_title() == `Dž`
 }