From 1d794e859028a71d182daf2fa5826aeeeab2876b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Lauren=C8=9Biu=20Nicola?= <lnicola@dend.ro>
Date: Tue, 5 May 2020 20:29:04 +0300
Subject: [PATCH] Fix column conversion for supplementary plane characters

---
 crates/ra_ide_db/src/line_index.rs | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/crates/ra_ide_db/src/line_index.rs b/crates/ra_ide_db/src/line_index.rs
index 212cb7b5bc..c7c744fce1 100644
--- a/crates/ra_ide_db/src/line_index.rs
+++ b/crates/ra_ide_db/src/line_index.rs
@@ -31,9 +31,19 @@ pub(crate) struct Utf16Char {
 }
 
 impl Utf16Char {
+    /// Returns the length in 8-bit UTF-8 code units.
     fn len(&self) -> TextSize {
         self.end - self.start
     }
+
+    /// Returns the length in 16-bit UTF-16 code units.
+    fn len_utf16(&self) -> usize {
+        if self.len() == TextSize::from(4) {
+            2
+        } else {
+            1
+        }
+    }
 }
 
 impl LineIndex {
@@ -110,7 +120,7 @@ impl LineIndex {
         if let Some(utf16_chars) = self.utf16_lines.get(&line) {
             for c in utf16_chars {
                 if c.end <= col {
-                    res -= usize::from(c.len()) - 1;
+                    res -= usize::from(c.len()) - c.len_utf16();
                 } else {
                     // From here on, all utf16 characters come *after* the character we are mapping,
                     // so we don't need to take them into account
@@ -125,7 +135,7 @@ impl LineIndex {
         if let Some(utf16_chars) = self.utf16_lines.get(&line) {
             for c in utf16_chars {
                 if col > u32::from(c.start) {
-                    col += u32::from(c.len()) - 1;
+                    col += u32::from(c.len()) - c.len_utf16() as u32;
                 } else {
                     // From here on, all utf16 characters come *after* the character we are mapping,
                     // so we don't need to take them into account
@@ -204,6 +214,9 @@ const C: char = 'メ';
 
         // UTF-16 to UTF-8
         assert_eq!(col_index.utf16_to_utf8_col(1, 19), TextSize::from(21));
+
+        let col_index = LineIndex::new("a𐐏b");
+        assert_eq!(col_index.utf16_to_utf8_col(0, 3), TextSize::from(5));
     }
 
     #[test]