fix: multibyte chars in text lines (#48)

lua/precognition/horizontal_motions.lua

@@ -27,27 +27,26 @@ end
 
 ---@param str string
 ---@param cursorcol integer
----@param _linelen integer
+---@param linelen integer
 ---@param big_word boolean
 ---@return Precognition.PlaceLoc
-function M.next_word_boundary(str, cursorcol, _linelen, big_word)
+function M.next_word_boundary(str, cursorcol, linelen, big_word)
     local offset = cursorcol
-    local len = vim.fn.strcharlen(str)
     local char = vim.fn.strcharpart(str, offset - 1, 1)
     local c_class = utils.char_class(char, big_word)
 
     if c_class ~= cc.whitespace then
-        while utils.char_class(char, big_word) == c_class and offset <= len do
+        while utils.char_class(char, big_word) == c_class and offset <= linelen do
             offset = offset + 1
             char = vim.fn.strcharpart(str, offset - 1, 1)
         end
     end
 
-    while utils.char_class(char, big_word) == cc.whitespace and offset <= len do
+    while utils.char_class(char, big_word) == cc.whitespace and offset <= linelen do
         offset = offset + 1
         char = vim.fn.strcharpart(str, offset - 1, 1)
     end
-    if offset > len then
+    if offset > linelen then
         return 0
     end
 
@@ -114,11 +113,10 @@ end
 
 ---@param str string
 ---@param cursorcol integer
----@param _linelen integer
+---@param linelen integer
 ---@param big_word boolean
 ---@return Precognition.PlaceLoc
-function M.prev_word_boundary(str, cursorcol, _linelen, big_word)
-    local len = vim.fn.strcharlen(str)
+function M.prev_word_boundary(str, cursorcol, linelen, big_word)
     local offset = cursorcol - 1
     local char = vim.fn.strcharpart(str, offset - 1, 1)
     local c_class = utils.char_class(char, big_word)
@@ -141,7 +139,7 @@ function M.prev_word_boundary(str, cursorcol, _linelen, big_word)
         end
     end
 
-    if offset == nil or offset > len or offset < 0 then
+    if offset == nil or offset > linelen or offset < 0 then
         return 0
     end
     return offset + 1

lua/precognition/init.lua

@@ -54,6 +54,10 @@ local M = {}
 ---@field PrevParagraph Precognition.PlaceLoc
 ---@field NextParagraph Precognition.PlaceLoc
 
+---@class Precognition.ExtraPadding
+---@field start integer
+---@field length integer
+
 ---@type Precognition.HintConfig
 local defaultHintConfig = {
     Caret = { text = "^", prio = 2 },
@@ -106,8 +110,9 @@ local gutter_group = "precognition_gutter"
 
 ---@param marks Precognition.VirtLine
 ---@param line_len integer
+---@param extra_padding Precognition.ExtraPadding
 ---@return table
-local function build_virt_line(marks, line_len)
+local function build_virt_line(marks, line_len, extra_padding)
     if not marks then
         return {}
     end
@@ -127,20 +132,26 @@ local function build_virt_line(marks, line_len)
             if existing == " " and existing ~= hint then
                 line_table[col] = hint
             else -- if the character is not a space, then we need to check the prio
-                local existingKey
+                local existing_key
                 for key, value in pairs(config.hints) do
                     if value.text == existing then
-                        existingKey = key
+                        existing_key = key
                         break
                     end
                 end
-                if existing ~= " " and config.hints[mark].prio > config.hints[existingKey].prio then
+                if existing ~= " " and config.hints[mark].prio > config.hints[existing_key].prio then
                     line_table[col] = hint
                 end
             end
         end
     end
 
+    if #extra_padding > 0 then
+        for _, padding in ipairs(extra_padding) do
+            line_table[padding.start] = line_table[padding.start] .. string.rep(" ", padding.length)
+        end
+    end
+
     local line = table.concat(line_table)
     if line:match("^%s+$") then
         return {}
@@ -201,15 +212,17 @@ local function display_marks()
     if utils.is_blacklisted_buffer(bufnr) then
         return
     end
-    local cursorline, cursorcol = unpack(vim.api.nvim_win_get_cursor(0))
-    cursorcol = cursorcol + 1
+    local cursorline = vim.fn.line(".")
+    local cursorcol = vim.fn.charcol(".")
     if extmark and not dirty then
         return
     end
 
     local tab_width = vim.bo.expandtab and vim.bo.shiftwidth or vim.bo.tabstop
     local cur_line = vim.api.nvim_get_current_line():gsub("\t", string.rep(" ", tab_width))
     local line_len = vim.fn.strcharlen(cur_line)
+    ---@type Precognition.ExtraPadding[]
+    local extra_padding = {}
     -- local after_cursor = vim.fn.strcharpart(cur_line, cursorcol + 1)
     -- local before_cursor = vim.fn.strcharpart(cur_line, 0, cursorcol - 1)
     -- local before_cursor_rev = string.reverse(before_cursor)
@@ -232,7 +245,10 @@ local function display_marks()
         Zero = 1,
     }
 
-    local virt_line = build_virt_line(virtual_line_marks, line_len)
+    --multicharacter padding
+    utils.add_multibyte_padding(cur_line, extra_padding, line_len)
+
+    local virt_line = build_virt_line(virtual_line_marks, line_len, extra_padding)
 
     -- TODO: can we add indent lines to the virt line to match indent-blankline or similar (if installed)?
 

lua/precognition/utils.lua

@@ -49,4 +49,18 @@ function M.create_pad_array(len, str)
     return pad_array
 end
 
+---Add extra padding for multi byte character characters
+---@param cur_line string
+---@param extra_padding Precognition.ExtraPadding[]
+---@param line_len integer
+function M.add_multibyte_padding(cur_line, extra_padding, line_len)
+    for i = 1, line_len do
+        local char = vim.fn.strcharpart(cur_line, i - 1, 1)
+        local width = vim.fn.strdisplaywidth(char)
+        if width > 1 then
+            table.insert(extra_padding, { start = i, length = width - 1 })
+        end
+    end
+end
+
 return M

tests/precognition/char_spec.lua

@@ -51,6 +51,16 @@ describe("big_word classing", function()
         eq(utils.char_class("@", true), 1)
         eq(utils.char_class(".", true), 1)
     end)
+
+    it("can class emoji characters", function()
+        eq(utils.char_class("🐱", false), 1)
+        eq(utils.char_class("😸", false), 1)
+    end)
+
+    it("can class nerdfont characters", function()
+        eq(utils.char_class("", false), 1)
+        eq(utils.char_class("", false), 1)
+    end)
 end)
 
 describe("pad arrays", function()

tests/precognition/horizontal_motions_spec.lua

@@ -304,4 +304,13 @@ describe("edge case", function()
         eq(16, hm.end_of_word(str, 15, #str, false))
         eq(22, hm.end_of_word(str, 21, #str, false))
     end)
+
+    it("multibyte characters", function()
+        local str = "# 💭👀precognition.nvim"
+        local len = vim.fn.strcharlen(str)
+        eq(3, hm.next_word_boundary(str, 1, len, false))
+        eq(17, hm.prev_word_boundary(str, 18, len, false))
+        eq(3, hm.prev_word_boundary(str, 18, len, true))
+        eq(3, hm.prev_word_boundary(str, 5, len, false))
+    end)
 end)