feat: support non-latin characters for keyword and buffer source

Closes #130
Closes #388

Author: Saghen

lua/blink/cmp/completion/trigger.lua

@@ -36,6 +36,8 @@
 local keyword_config = require('blink.cmp.config').completion.keyword
 local config = require('blink.cmp.config').completion.trigger
 
+local keyword_regex = vim.regex(keyword_config.regex)
+
 --- @type blink.cmp.CompletionTrigger
 --- @diagnostic disable-next-line: missing-fields
 local trigger = {
@@ -62,7 +64,7 @@ function trigger.activate()
         trigger.show({ trigger_character = char })
 
       -- character is part of a keyword
-      elseif char:match(keyword_config.regex) ~= nil and (config.show_on_keyword or trigger.context ~= nil) then
+      elseif keyword_regex:match_str(char) ~= nil and (config.show_on_keyword or trigger.context ~= nil) then
         trigger.show()
 
       -- nothing matches so hide
@@ -82,7 +84,7 @@ function trigger.activate()
       local char_under_cursor = vim.api.nvim_get_current_line():sub(cursor_col, cursor_col)
       local is_on_trigger_for_show = trigger.is_trigger_character(char_under_cursor)
       local is_on_trigger_for_show_on_insert = trigger.is_trigger_character(char_under_cursor, true)
-      local is_on_keyword_char = char_under_cursor:match(keyword_config.regex) ~= nil
+      local is_on_keyword_char = keyword_regex:match_str(char_under_cursor) ~= nil
 
       local insert_enter_on_trigger_character = config.show_on_trigger_character
         and config.show_on_insert_on_trigger_character
@@ -197,17 +199,16 @@ function trigger.within_query_bounds(cursor)
 end
 
 --- Moves forward and backwards around the cursor looking for word boundaries
---- @param regex string
 --- @return blink.cmp.ContextBounds
-function trigger.get_context_bounds(regex)
+function trigger.get_context_bounds()
   local cursor_line = vim.api.nvim_win_get_cursor(0)[1]
   local cursor_col = vim.api.nvim_win_get_cursor(0)[2]
 
   local line = vim.api.nvim_buf_get_lines(0, cursor_line - 1, cursor_line, false)[1]
   local start_col = cursor_col
   while start_col >= 1 do
     local char = line:sub(start_col, start_col)
-    if char:match(regex) == nil then
+    if keyword_regex:match_str(char) == nil then
       start_col = start_col + 1
       break
     end
@@ -218,7 +219,7 @@ function trigger.get_context_bounds(regex)
   local end_col = cursor_col
   while end_col < #line do
     local char = line:sub(end_col + 1, end_col + 1)
-    if char:match(regex) == nil then break end
+    if keyword_regex:match_str(char) == nil then break end
     end_col = end_col + 1
   end
 
@@ -228,7 +229,7 @@ function trigger.get_context_bounds(regex)
   local length = end_col - start_col + 1
   -- Since sub(1, 1) returns a single char string, we need to check if that single char matches
   -- and otherwise mark the length as 0
-  if start_col == end_col and line:sub(start_col, end_col):match(regex) == nil then length = 0 end
+  if start_col == end_col and keyword_regex:match_str(line:sub(start_col, end_col)) == nil then length = 0 end
 
   return { line_number = cursor_line, start_col = start_col, end_col = end_col, length = length }
 end

lua/blink/cmp/config/completion/keyword.lua

@@ -15,8 +15,8 @@ local keyword = {
   --- @type blink.cmp.CompletionKeywordConfig
   default = {
     range = 'prefix',
-    regex = '[%w_\\-]',
-    exclude_from_prefix_regex = '[\\-]',
+    regex = '[-_]\\|\\k',
+    exclude_from_prefix_regex = '-',
   },
 }
 

lua/blink/cmp/fuzzy/lib.rs

@@ -12,7 +12,7 @@ mod fuzzy;
 mod lsp_item;
 
 lazy_static! {
-    static ref REGEX: Regex = Regex::new(r"\w[\w0-9_\\-]{2,32}").unwrap();
+    static ref REGEX: Regex = Regex::new(r"\p{L}[\p{L}0-9_\\-]{2,32}").unwrap();
     static ref FRECENCY: RwLock<Option<FrecencyTracker>> = RwLock::new(None);
 }
 

lua/blink/cmp/lib/text_edits.lua

@@ -97,7 +97,13 @@ function text_edits.get_from_item(item)
   if text_edit == nil then return text_edits.guess(item) end
 
   -- FIXME: temporarily convert insertReplaceEdit to regular textEdit
-  text_edit.range = text_edit.range or text_edit.insert or text_edit.replace
+  if text_edit.range == nil then
+    if config.completion.keyword.range == 'full' and text_edit.replace ~= nil then
+      text_edit.range = text_edit.replace
+    else
+      text_edit.range = text_edit.insert or text_edit.replace
+    end
+  end
   text_edit.insert = nil
   text_edit.replace = nil
   --- @cast text_edit lsp.TextEdit

lua/blink/cmp/lib/utils.lua

@@ -54,39 +54,41 @@ end
 
 --- Gets characters around the cursor and returns the range, 0-indexed
 --- @param range 'prefix' | 'full'
---- @param regex string
---- @param exclude_from_prefix_regex string
+--- @param regex_str string
+--- @param exclude_from_prefix_regex_str string
 --- @return { start_col: number, length: number }
 --- TODO: switch to return start_col, length to simplify downstream logic
-function utils.get_regex_around_cursor(range, regex, exclude_from_prefix_regex)
+function utils.get_regex_around_cursor(range, regex_str, exclude_from_prefix_regex_str)
+  local backward_regex = vim.regex('\\(' .. regex_str .. '\\)\\+$')
+  local forward_regex = vim.regex('^\\(' .. regex_str .. '\\)\\+')
+
   local current_col = vim.api.nvim_win_get_cursor(0)[2] + 1
   local line = vim.api.nvim_get_current_line()
 
-  -- Search backward for the start of the word
-  local start_col = current_col
   local length = 0
-  while start_col > 0 do
-    local char = line:sub(start_col - 1, start_col - 1)
-    if char:match(regex) == nil then break end
-    start_col = start_col - 1
-    length = length + 1
+  local start_col = current_col
+
+  -- Search backward for the start of the word
+  local line_before = line:sub(1, current_col - 1)
+  local before_match_start, _ = backward_regex:match_str(line_before)
+  if before_match_start ~= nil then
+    start_col = before_match_start + 1
+    length = current_col - start_col
   end
 
   -- Search forward for the end of the word if configured
   if range == 'full' then
-    while start_col + length <= #line do
-      local col = start_col + length
-      local char = line:sub(col, col)
-      if char:match(regex) == nil then break end
-      length = length + 1
-    end
+    local line_after = line:sub(current_col)
+    local _, after_match_end = forward_regex:match_str(line_after)
+    if after_match_end ~= nil then length = length + after_match_end end
   end
 
   -- exclude characters matching exclude_prefix_regex from the beginning of the bounds
-  if exclude_from_prefix_regex ~= nil then
+  if exclude_from_prefix_regex_str ~= nil then
+    local exclude_from_prefix_regex = vim.regex(exclude_from_prefix_regex_str)
     while length > 0 do
       local char = line:sub(start_col, start_col)
-      if char:match(exclude_from_prefix_regex) == nil then break end
+      if exclude_from_prefix_regex:match_str(char) == nil then break end
       start_col = start_col + 1
       length = length - 1
     end