Merge pull request #24 from JarKz/utf8support

feat!: add utf8 support

README.md

@@ -96,6 +96,9 @@ call.
 
 -- packer
 use { "chrisgrieser/nvim-spider" }
+
+-- lazy.nvim
+{ "chrisgrieser/nvim-spider", lazy = true },
 ```
 
 No keybindings are created by default. Below are the mappings to replace the
@@ -127,6 +130,30 @@ vim.keymap.set(
 > using `function() require("spider").motion("w") end` as third argument of
 > the keymap, dot-repeatability will not work.
 
+## UTF-8 support
+
+For adding UTF-8 support for matching non-ASCII text, add rocks `luautf8` in packer.nvim. Or add [dependency](https://github.com/theHamsta/nvim_rocks) like below, in lazy.nvim example.
+
+```lua
+-- packer
+{ "chrisgrieser/nvim-spider", rocks = "luautf8" }
+
+-- lazy.nvim
+{
+    "chrisgrieser/nvim-spider",
+    lazy = true,
+    dependencies = {
+        "theHamsta/nvim_rocks",
+        event = "VeryLazy",
+        build = "pip3 install --user hererocks && python3 -mhererocks . -j2.1.0-beta3 -r3.0.0 && cp nvim_rocks.lua lua",
+        config = function()
+            local rocks = require("nvim_rocks")
+            rocks.ensure_installed("luautf8")
+        end,
+    }
+},
+```
+
 ## Configuration
 The `.setup()` call is optional.
 

lua/spider/init.lua

@@ -1,6 +1,41 @@
 local M = {}
 local patternVariants = require("spider.pattern-variants")
 
+local str_func = {
+	reverse = string.reverse,
+	find = string.find,
+	gmatch = string.gmatch,
+	len =  string.len,
+	init_pos = function(_, col)
+		col = col + 1 -- from 0-based indexing to 1-based
+		local startCol = col
+		return col, startCol
+	end,
+	offset = function(_, pos)
+		return pos
+	end,
+}
+
+local ok, utf8 = pcall(require, "lua-utf8")
+
+if ok then
+	-- remapping functions to utf8 supported functions
+	for name, _ in pairs(str_func) do
+		if utf8[name] then
+			str_func[name] = utf8[name]
+		end
+	end
+	str_func.init_pos = function(s, col)
+		local offset = 1
+		for p, _ in utf8.codes(s) do
+			if p > col then break end
+			offset = offset + 1
+		end
+		local startOffset = offset
+		return offset, startOffset
+	end
+end
+
 --------------------------------------------------------------------------------
 -- CONFIG
 ---@class (exact) optsObj
@@ -35,18 +70,20 @@ end
 ---@param line string
 ---@param pattern string
 ---@param endOfWord boolean look for the end of the pattern instead of the start
----@param col number look for the first match after this number
+---@param offset number -- look for the first match after this number
 ---@nodiscard
----@return number|nil -- returns nil if none is found
-local function firstMatchAfter(line, pattern, endOfWord, col)
+---@return number|nil returns nil if none is found
+local function firstMatchAfter(line, pattern, endOfWord, offset)
 	-- special case: pattern with ^/$, since there can only be one match
 	-- and since gmatch won't work with them
 	if pattern:find("^%^") or pattern:find("%$$") then
-		if pattern:find("%$$") and col >= #line then return nil end -- checking for high col count for virtualedit
-		if pattern:find("^%^") and col ~= 1 then return nil end
-		local start, endPos = line:find(pattern)
+		if pattern:find("%$$") and offset > str_func.len(line) then return nil end -- checking for high col count for virtualedit
+		if pattern:find("^%^") and offset ~= 0 then return nil end
+
+		local start, endPos = str_func.find(line, pattern)
+		if start == nil or endPos == nil then return nil end
+
 		local pos = endOfWord and endPos or start
-		if pos and not endOfWord then pos = pos - 1 end
 		return pos
 	end
 
@@ -56,47 +93,46 @@ local function firstMatchAfter(line, pattern, endOfWord, col)
 		pattern = "()" .. pattern
 	end
 	-- `:gmatch` will return all locations in the string where the pattern is
-	-- found, the loop looks for the first one that is higher than the col to
-	-- look from
-	for pos in line:gmatch(pattern) do
-		if endOfWord and pos > col then return pos - 1 end
-		if not endOfWord and pos >= col then return pos - 1 end
+	-- found, the loop looks for the first one that is higher than the offset
+	-- to look from
+	for pos in str_func.gmatch(line, pattern) do
+		if type(pos) == "string" then return nil end
+
+		if endOfWord then pos = pos - 1 end
+		if pos > offset then return pos end
 	end
 	return nil
 end
 
 ---@param line string input string where to find the pattern
----@param col number position to start looking from
+---@param offset number position to start looking from
 ---@param key "w"|"e"|"b"|"ge" the motion to perform
 ---@param opts optsObj configuration table as in setup()
 ---@nodiscard
 ---@return number|nil pattern position, returns nil if no pattern was found
-local function getNextPosition(line, col, key, opts)
+local function getNextPosition(line, offset, key, opts)
 	local endOfWord = (key == "ge") or (key == "e")
 	local backwards = (key == "b") or (key == "ge")
 	local patterns = patternVariants.get(opts, backwards)
 
 	if backwards then
-		line = line:reverse()
+		line = str_func.reverse(line)
 		endOfWord = not endOfWord
-		if col == -1 then
-			col = 1
-		else
-			col = #line - col + 1
-		end
+
+		local isSameLine = offset ~= 0
+		if isSameLine then offset = str_func.len(line) - offset + 1 end
 	end
 
 	-- search for patterns, get closest one
 	local matches = {}
 	for _, pattern in pairs(patterns) do
-		local match = firstMatchAfter(line, pattern, endOfWord, col)
+		local match = firstMatchAfter(line, pattern, endOfWord, offset)
 		if match then table.insert(matches, match) end
 	end
 	if vim.tbl_isempty(matches) then return nil end -- none found in this line
 	local nextPos = math.min(unpack(matches))
 
-	if not endOfWord then nextPos = nextPos + 1 end
-	if backwards then nextPos = #line - nextPos + 1 end
+	if backwards then nextPos = str_func.len(line) - nextPos + 1 end
 	return nextPos
 end
 
@@ -118,44 +154,49 @@ function M.motion(key, motionOpts)
 	end
 
 	local row, col = unpack(vim.api.nvim_win_get_cursor(0))
-	local startCol, startRow = col, row
-	local lastRow = vim.api.nvim_buf_line_count(0)
+	local startRow = row
+	local lastRow = vim.fn.line("$")
 	local forwards = key == "w" or key == "e"
 
-	-- loop through counts
-	for i = 1, vim.v.count1, 1 do
-		if forwards then
-			col = col + 2 -- +1 (next position), +1 lua indexing
-		elseif not forwards and i > 1 then
-			col = col - 1 -- next pos
-		end
+	local line = getline(row)
+	local offset, startOffset = str_func.init_pos(line, col)
 
-		-- loop through rows (if next location not found in line)
+	-- looping through counts
+	for _ = 1, vim.v.count1, 1 do
+		-- looping through rows (if next location not found in line)
 		while true do
-			local line = getline(row)
-			col = getNextPosition(line, col, key, opts)
-			local onTheSamePos = (col == startCol + 1 and row == startRow)
-			if col and not onTheSamePos then break end
-			col = forwards and 1 or -1
+			local result = getNextPosition(line, offset, key, opts)
+			if result then
+				offset = result
+				local onTheSamePos = (offset == startOffset and row == startRow)
+				if not onTheSamePos then break end
+			end
+
+			offset = 0
 			row = forwards and row + 1 or row - 1
 			if row > lastRow or row < 1 then return end
+			line = getline(row)
 		end
 	end
 
-	col = col - 1 -- lua string indices different
+	col = str_func.offset(line, offset) - 1 -- lua string indices different
 
 	-- operator-pending specific considerations (see issues #3 and #5)
 	local mode = vim.api.nvim_get_mode().mode
 	local isOperatorPending = mode == "no" -- = [n]ormal & [o]perator, not the word "no"
 	if isOperatorPending then
 		local lastCol = vim.fn.col("$")
-		if key == "e" then col = col + 1 end
+		if key == "e" then
+			offset = offset + 1
+			col = str_func.offset(line, offset) - 1
+		end
 
 		if lastCol - 1 == col then
 			-- HACK columns are end-exclusive, cannot actually target the last character
 			-- in the line otherwise without switching to visual mode?!
 			vim.cmd.normal { "v", bang = true }
-			col = col - 1 -- SIC indices in visual off-by-one compared to normal
+			offset = offset - 1
+			col = str_func.offset(line, offset) - 1 -- SIC indices in visual off-by-one compared to normal
 		end
 	end