perf: Convert between 1D and 2D indices in a non-stupid way.

Before I was just concatenating all of the previous lines in the buffer
in order to compute the right indices. This PR changes things to use
builtin Vim functions which are hopefully much more performant.

Author: kylechui

lua/nvim-surround/patterns.lua

@@ -3,18 +3,13 @@ local buffer = require("nvim-surround.buffer")
 local M = {}
 
 -- Converts a 1D index into the buffer to the corresponding 2D buffer position.
----@param index integer The index of the character in the string.
+---@param index integer The index of the character in the buffer (viewed as a single `\n`-joined string).
 ---@return position @The position of the character in the buffer.
 ---@nodiscard
 M.index_to_pos = function(index)
-    local buffer_text = table.concat(buffer.get_lines(1, -1), "\n")
-    -- Counts the number of newline characters, plus one for the final character before the current line
-    local lnum = select(2, buffer_text:sub(1, math.max(1, index - 1)):gsub("\n", "\n")) + 1
-    -- Special case for first line, as there are no newline characters preceding it
-    if lnum == 1 then
-        return { 1, index }
-    end
-    local col = index - #table.concat(buffer.get_lines(1, lnum - 1), "\n") - 1
+    local lnum = vim.fn.byte2line(index)
+    local lnum_bytes = vim.fn.line2byte(lnum)
+    local col = index - lnum_bytes + 1
     return { lnum, col }
 end
 
@@ -23,11 +18,7 @@ end
 ---@return integer @The index of the character into the buffer.
 ---@nodiscard
 M.pos_to_index = function(pos)
-    -- Special case for first line, as there are no newline characters preceding it
-    if pos[1] == 1 then
-        return pos[2]
-    end
-    return #table.concat(buffer.get_lines(1, pos[1] - 1), "\n") + pos[2] + 1
+    return vim.api.nvim_buf_get_offset(0, pos[1] - 1) + pos[2]
 end
 
 -- Expands a selection to properly contain multi-byte characters.