From e831cab7a4c31da02c72044190e9afc1a9ed584c Mon Sep 17 00:00:00 2001
From: Liam Dyer <liamcdyer@gmail.com>
Date: Fri, 3 Jan 2025 14:34:30 -0500
Subject: [PATCH] fix: get full unicode char at cursor position

Closes #864
---
 lua/blink/cmp/completion/trigger/init.lua  |  4 ++-
 lua/blink/cmp/completion/trigger/utils.lua | 29 ++++++++++++++++++++++
 lua/blink/cmp/fuzzy/keyword.rs             | 22 +++++++++++++---
 3 files changed, 50 insertions(+), 5 deletions(-)
 create mode 100644 lua/blink/cmp/completion/trigger/utils.lua

diff --git a/lua/blink/cmp/completion/trigger/init.lua b/lua/blink/cmp/completion/trigger/init.lua
index e463daf2..03d43720 100644
--- a/lua/blink/cmp/completion/trigger/init.lua
+++ b/lua/blink/cmp/completion/trigger/init.lua
@@ -23,6 +23,7 @@
 
 local config = require('blink.cmp.config').completion.trigger
 local context = require('blink.cmp.completion.trigger.context')
+local utils = require('blink.cmp.completion.trigger.utils')
 
 --- @type blink.cmp.CompletionTrigger
 --- @diagnostic disable-next-line: missing-fields
@@ -71,7 +72,8 @@ function trigger.activate()
 
     local cursor = context.get_cursor()
     local cursor_col = cursor[2]
-    local char_under_cursor = context.get_line():sub(cursor_col, cursor_col)
+
+    local char_under_cursor = utils.get_char_at_cursor()
     local is_on_trigger_for_show = trigger.is_trigger_character(char_under_cursor)
 
     local insert_enter_on_trigger_character = config.show_on_trigger_character
diff --git a/lua/blink/cmp/completion/trigger/utils.lua b/lua/blink/cmp/completion/trigger/utils.lua
new file mode 100644
index 00000000..3ceb6435
--- /dev/null
+++ b/lua/blink/cmp/completion/trigger/utils.lua
@@ -0,0 +1,29 @@
+local context = require('blink.cmp.completion.trigger.context')
+local utils = {}
+
+--- Gets the full Unicode character at cursor position
+--- @return string
+function utils.get_char_at_cursor()
+  local line = context.get_line()
+  local cursor_col = context.get_cursor()[2]
+
+  -- Find the start of the UTF-8 character
+  local start_col = cursor_col
+  while start_col > 1 do
+    local char = string.byte(line:sub(start_col, start_col))
+    if char < 0x80 or char > 0xBF then break end
+    start_col = start_col - 1
+  end
+
+  -- Find the end of the UTF-8 character
+  local end_col = cursor_col
+  while end_col < #line do
+    local char = string.byte(line:sub(end_col + 1, end_col + 1))
+    if char < 0x80 or char > 0xBF then break end
+    end_col = end_col + 1
+  end
+
+  return line:sub(start_col, end_col)
+end
+
+return utils
diff --git a/lua/blink/cmp/fuzzy/keyword.rs b/lua/blink/cmp/fuzzy/keyword.rs
index c8045a7e..94aeb78a 100644
--- a/lua/blink/cmp/fuzzy/keyword.rs
+++ b/lua/blink/cmp/fuzzy/keyword.rs
@@ -8,14 +8,16 @@ lazy_static! {
 
 /// Given a line and cursor position, returns the start and end indices of the keyword
 pub fn get_keyword_range(line: &str, col: usize, match_suffix: bool) -> (usize, usize) {
-    let line_before = line.chars().take(col).collect::<String>();
-    let before_match_start = BACKWARD_REGEX.find(&line_before).map(|m| m.start());
+    let before_match_start = BACKWARD_REGEX
+        .find(&line[0..col.min(line.len())])
+        .map(|m| m.start());
     if !match_suffix {
         return (before_match_start.unwrap_or(col), col);
     }
 
-    let line_after = line.chars().skip(col).collect::<String>();
-    let after_match_end = FORWARD_REGEX.find(&line_after).map(|m| m.end() + col);
+    let after_match_end = FORWARD_REGEX
+        .find(&line[col.min(line.len())..])
+        .map(|m| m.end() + col);
     (
         before_match_start.unwrap_or(col),
         after_match_end.unwrap_or(col),
@@ -63,3 +65,15 @@ pub fn guess_keyword_from_item(
     let (start, end) = guess_keyword_range_from_item(item_text, line, cursor_col, match_suffix);
     line[start..end].to_string()
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_get_keyword_range_unicode() {
+        let line = "'вest'";
+        let col = line.len() - 1;
+        assert_eq!(get_keyword_range(line, col, false), (1, line.len() - 1));
+    }
+}