Merge pull request #12343 from ndreas/fix/lsp-multibyte-indexing

Fix LSP multibyte indexing

runtime/lua/vim/lsp/util.lua

@@ -95,17 +95,23 @@ local edit_sort_key = sort_by_key(function(e)
   return {e.A[1], e.A[2], e.i}
 end)
 
-local function get_line_byte_from_line_character(bufnr, lnum, cnum)
+--- Position is a https://microsoft.github.io/language-server-protocol/specifications/specification-current/#position
-  -- Skip check when the byte and character position is the same
+-- Returns a zero-indexed column, since set_lines() does the conversion to
-  if cnum > 0 then
+-- 1-indexed
-    local lines = api.nvim_buf_get_lines(bufnr, lnum, lnum+1, false)
+local function get_line_byte_from_position(bufnr, position)
-
+  -- LSP's line and characters are 0-indexed
+  -- Vim's line and columns are 1-indexed
+  local col = position.character
+  -- When on the first character, we can ignore the difference between byte and
+  -- character
+  if col > 0 then
+    local line = position.line
+    local lines = api.nvim_buf_get_lines(bufnr, line, line + 1, false)
     if #lines > 0 then
-      return vim.str_byteindex(lines[1], cnum)
+      return vim.str_byteindex(lines[1], col)
     end
   end
-
+  return col
-  return cnum
 end
 
 function M.apply_text_edits(text_edits, bufnr)
@@ -118,15 +124,9 @@ function M.apply_text_edits(text_edits, bufnr)
   for i, e in ipairs(text_edits) do
     -- adjust start and end column for UTF-16 encoding of non-ASCII characters
     local start_row = e.range.start.line
-    local start_col = get_line_byte_from_line_character(
+    local start_col = get_line_byte_from_position(bufnr, e.range.start)
-      bufnr,
-      start_row,
-      e.range.start.character)
     local end_row = e.range["end"].line
-    local end_col = get_line_byte_from_line_character(
+    local end_col = get_line_byte_from_position(bufnr, e.range['end'])
-      bufnr,
-      end_row,
-      e.range["end"].character)
     start_line = math.min(e.range.start.line, start_line)
     finish_line = math.max(e.range["end"].line, finish_line)
     -- TODO(ashkan) sanity check ranges for overlap.
@@ -543,9 +543,7 @@ function M.jump_to_location(location)
   api.nvim_buf_set_option(0, 'buflisted', true)
   local range = location.range or location.targetSelectionRange
   local row = range.start.line
-  local col = range.start.character
+  local col = get_line_byte_from_position(0, range.start)
-  local line = api.nvim_buf_get_lines(0, row, row+1, true)[1]
-  col = vim.str_byteindex(line, col)
   api.nvim_win_set_cursor(0, {row + 1, col})
   return true
 end

test/functional/plugin/lsp_spec.lua

@@ -1317,4 +1317,65 @@ describe('LSP', function()
       eq("Unknown", exec_lua("return vim.lsp.util._get_symbol_kind_name(1000)"))
     end)
   end)
+
+  describe('lsp.util.jump_to_location', function()
+    local target_bufnr
+
+    before_each(function()
+      target_bufnr = exec_lua [[
+        local bufnr = vim.uri_to_bufnr("file://fake/uri")
+        local lines = {"1st line of text", "å å ɧ 汉语 ↥ 🤦 🦄"}
+        vim.api.nvim_buf_set_lines(bufnr, 0, 1, false, lines)
+        return bufnr
+      ]]
+    end)
+
+    local location = function(start_line, start_char, end_line, end_char)
+      return {
+        uri = "file://fake/uri",
+        range = {
+          start = { line = start_line, character = start_char },
+          ["end"] = { line = end_line, character = end_char },
+        },
+      }
+    end
+
+    local jump = function(msg)
+      eq(true, exec_lua('return vim.lsp.util.jump_to_location(...)', msg))
+      eq(target_bufnr, exec_lua[[return vim.fn.bufnr('%')]])
+      return {
+        line = exec_lua[[return vim.fn.line('.')]],
+        col = exec_lua[[return vim.fn.col('.')]],
+      }
+    end
+
+    it('jumps to a Location', function()
+      local pos = jump(location(0, 9, 0, 9))
+      eq(1, pos.line)
+      eq(10, pos.col)
+    end)
+
+    it('jumps to a LocationLink', function()
+      local pos = jump({
+          targetUri = "file://fake/uri",
+          targetSelectionRange = {
+            start = { line = 0, character = 4 },
+            ["end"] = { line = 0, character = 4 },
+          },
+          targetRange = {
+            start = { line = 1, character = 5 },
+            ["end"] = { line = 1, character = 5 },
+          },
+        })
+      eq(1, pos.line)
+      eq(5, pos.col)
+    end)
+
+    it('jumps to the correct multibyte column', function()
+      local pos = jump(location(1, 2, 1, 2))
+      eq(2, pos.line)
+      eq(4, pos.col)
+      eq('å', exec_lua[[return vim.fn.expand('<cword>')]])
+    end)
+  end)
 end)