Fix ueberzug backend's y position being off by one

[Vonr]

Before:

After:

[3rd]

@Vonr Hey, thank you so much for the PR!
I don't think it's a ueberzug-specific issue, it's probably something related to accounting for user-defined options around here:

image.nvim/lua/image/renderer.lua

Line 5 in 847f5d6

local get_global_offsets = function()

Been seeing the same issue on #5

[Vonr]

That's interesting... So far all the images I've tested with work with this fix but ill look into renderer.lua.

[Vonr]

Might be line 109?

[Vonr]

local get_global_offsets = function()
  local x = 0
  local y = 0
  if vim.opt.number then x = x + math.max(vim.opt.numberwidth:get(), ('' .. vim.api.nvim_buf_line_count(0)):len() + (vim.opt.relativenumber:get() and 1 or 0)) end
  -- if vim.opt.signcolumn:get() ~= "no" then x = x + 2 end
  if vim.opt.showtabline == 2 then y = y + 1 end
  if vim.opt.winbar ~= "none" then y = y + 1 end
  return { x = x, y = y }
end

This fixes my left-right alignment but it probably breaks signcolumn alignment.
I commented that line out because I use signcolumn 'number'.

[Vonr]

local get_global_offsets = function()
  local x = 0
  local y = 0

  local numberwidth = math.max(
      vim.opt.numberwidth:get(),
      ('' .. vim.api.nvim_buf_line_count(0)):len()
        + (vim.opt.relativenumber:get() and 1 or 0))

  if vim.opt.number then x = x + numberwidth end

  local signcolumn = vim.opt.signcolumn:get()
  if signcolumn ~= "no" then
      x = x + math.max(0, 2 - (signcolumn == "number" and numberwidth or 0))
  end

  if vim.opt.showtabline == 2 then y = y + 1 end
  if vim.opt.winbar ~= "none" then y = y + 1 end
  return { x = x, y = y }
end

This has a pretty decent chance of working for most setups, but I honestly have no idea.

[3rd]

@Vonr that looks great, I'll test it out today and also add some unit tests for this, and we can merge it soon.
Thanks a lot!

[3rd]

I think these are buffer-relative and this version applies the offsets for the active buffers to all the visible windows, looking into making this generic enough.

[3rd]

Great catch, didn't see it because the winbar logic above is wrong.

[3rd]

@Vonr think we can rely on textoff for the gutter math, can you test if it works ok for you with the latest changes?
c957678

[Vonr]

Looks like it works, thank you!