Refiling sections on the same file gives unexpected results

Question

Refiling sections on the same file gives unexpected results

lyz-code opened this issue 6 months ago · comments

Lyz commented 6 months ago

Describe the bug

When refiling a section to another in the same file that is below the result is not the expected

Steps to reproduce

In the file test.org with the contents:

* Heading 1
** Subheading 1 
** Subheading 2
* Heading 2     
**** SubSubheading 2.1

Place the cursor over Subheading 1 and try to refile to test.org:SubSubheading 2.1

Expected behavior

The section is well refined, so the result file is:

* Heading 1
** Subheading 2
* Heading 2     
**** SubSubheading 2.1   
***** Subheading 1

Right now the output is:

* Heading 1
***** Subheading 1
** Subheading 2
* Heading 2     
**** SubSubheading 2.1

Emacs functionality

I guess it works as expected

Minimal init.lua

vim.g.mapleader = ' '

vim.cmd([[set runtimepath=$VIMRUNTIME]])
vim.cmd([[set packpath=/tmp/nvim/site]])

local package_root = '/tmp/nvim/site/pack'
local install_path = package_root .. '/packer/start/packer.nvim'

local function load_plugins()
  require('packer').startup({
    {
      'wbthomason/packer.nvim',
      { 'nvim-treesitter/nvim-treesitter' },
      { 'kristijanhusak/orgmode.nvim', branch = 'master' },
    },
    config = {
      package_root = package_root,
      compile_path = install_path .. '/plugin/packer_compiled.lua',
    },
  })
end

_G.load_config = function()
  require('orgmode').setup_ts_grammar()
  require('nvim-treesitter.configs').setup({
    highlight = {
      enable = true,
      additional_vim_regex_highlighting = { 'org' },
    },
  })

  vim.cmd([[packadd nvim-treesitter]])
  vim.cmd([[runtime plugin/nvim-treesitter.lua]])
  vim.cmd([[TSUpdateSync org]])

  -- Close packer after install
  if vim.bo.filetype == 'packer' then
    vim.api.nvim_win_close(0, true)
  end

  require('orgmode').setup({
    org_agenda_files = {
      './*'
    }
  }
  )

  -- Reload current file if it's org file to reload tree-sitter
  if vim.bo.filetype == 'org' then
    vim.cmd([[edit!]])
  end
end

if vim.fn.isdirectory(install_path) == 0 then
  vim.fn.system({ 'git', 'clone', 'https://github.com/wbthomason/packer.nvim', install_path })
  load_plugins()
  require('packer').sync()
  vim.cmd([[autocmd User PackerCompileDone ++once lua load_config()]])
else
  load_plugins()
  load_config()
end

Screenshots and recordings

No response

OS / Distro

Debian

Neovim version/commit

0.9.1

Additional context

If instead you place the cursor over SubSubheading 2.1 and try to refile to test.org:Subheading 1 the issue is similar. The expected output is:

* Heading 1
** Subheading 1 
*** SubSubheading 2.1  
** Subheading 2
* Heading 2

And right now it returns:

* Heading 1
** Subheading 1
** Subheading 2
* Heading 2
*** SubSubheading 2.1

Sebastian Flügge · Answer 1 · Wed Dec 27 2023 08:50:23 GMT+0800 (China Standard Time)

Yes, the Refiling feature has some unexpected bugs. I'm currently working on understanding the code, because I want to fix this. My current understanding is, that the correct nesting/promotion of the subheading is done, but the "move" part of the refiling is somehow skipped.

I also observed, that when I refile a subheading into a file without a target headline, the contend is moved, but the subheading is not promoted to the topmost-one, which I would actually expect. Otherwise, if I do a lot of refiling of subheadings with different indent levels into one file, I get a completly messed-up structure instead of a flat list.
@kristijanhusak: This behavior is actually encoded in a unit test, so I would like to ask you, if this behaviour is intentional. If so, I would like to make it optional over a configuration variable, because for my daily use it is actually quite annoying.

Kristijan Husak · Answer 2 · Wed Dec 27 2023 15:46:19 GMT+0800 (China Standard Time)

that the correct nesting/promotion of the subheading is done, but the "move" part of the refiling is somehow skipped.

Moving in the same file is done here https://github.com/nvim-orgmode/orgmode/blob/master/lua/orgmode/capture/init.lua#L352. It works differently because there's no need to open same file remotely to move things around.

I also observed, that when I refile a subheading into a file without a target headline, the contend is moved, but the subheading is not promoted to the topmost-one, which I would actually expect

If I'm not mistaken, that's how Emacs orgmode works, but it's been a while. If you can double check how it behaves there that would be great.

Sebastian Flügge · Answer 3 · Wed Dec 27 2023 20:23:25 GMT+0800 (China Standard Time)

@kristijanhusak: Yeah, I actually found this function already and was already pretty sure, that the bug is right there. I'm still investigating, how the surrounding code actually works to implement my convenience wish, but for now here is the fix for the particular problem: #637

Kristijan Husak · Answer 4 · Fri Dec 29 2023 02:01:56 GMT+0800 (China Standard Time)

@lyz-code can you check if the issue is fixed now?

Sebastian Flügge · Answer 5 · Sun Dec 31 2023 05:54:39 GMT+0800 (China Standard Time)

I also observed, that when I refile a subheading into a file without a target headline, the contend is moved, but the subheading is not promoted to the topmost-one, which I would actually expect

If I'm not mistaken, that's how Emacs orgmode works, but it's been a while. If you can double check how it behaves there that would be great.

I actually exhumed and repaired my 2years old emacs config to try this out and I realized the following:

Refiling target files are defined separately from the agenda files, although it's common to just define the agenda files
you cannot refile to a plain file, only to headings which are crawled from the files defined in org-refile-targets
org-refile-targets takes also a deepness level, which defines, which headings are actually considered as refile target

Here is an example:

(setq org-refile-targets '((org-agenda-files :maxlevel . 3))

For simplicity I would keep the possibility to directly refile to files, which actually has also value. But if no Heading is given, we should promote the headline to the first level, because it allows for a mutch cleaner and deterministic workflow.
@kristijanhusak If you agree, I will implement this and open a PR. Otherwise it probably ends up in a personal customization on my fork. 😉

Kristijan Husak · Answer 6 · Sun Dec 31 2023 16:41:03 GMT+0800 (China Standard Time)

If you agree, I will implement this and open a PR. Otherwise it probably ends up in a personal customization on my fork.

Yeah, go for it. Thanks for investigating!

Sebastian Flügge · Answer 7 · Mon Jan 01 2024 02:11:55 GMT+0800 (China Standard Time)

If you agree, I will implement this and open a PR. Otherwise it probably ends up in a personal customization on my fork.

Yeah, go for it. Thanks for investigating!

Done: #641

Sebastian Flügge · Answer 8 · Wed Jan 03 2024 08:57:43 GMT+0800 (China Standard Time)

If you agree, I will implement this and open a PR. Otherwise it probably ends up in a personal customization on my fork.

Yeah, go for it. Thanks for investigating!

I ran the tests on my local machine against the current nightly:

Everything is green here.

Is it possible, to rerun the failed tests in Github actions? It seems, that I don't have the permissions for that.

Kristijan Husak · Answer 9 · Wed Jan 03 2024 16:06:25 GMT+0800 (China Standard Time)

@seflue I just re-run last github action https://github.com/nvim-orgmode/orgmode/actions/runs/7372025475?pr=642 and it failed on the same thing. We'll see once we merge some of these other PRs if the issue persists.

Lyz · Answer 10 · Thu Jan 11 2024 06:47:22 GMT+0800 (China Standard Time)

Sorry for the late response I've been away on holidays ^^.

The fix works perfect for me, thank you so much @seflue <3!!

May we close the issue? or wait for the CI thing to be fixed?

Kristijan Husak · Answer 11 · Thu Jan 11 2024 19:09:30 GMT+0800 (China Standard Time)

Since it's fixed, I'm closing this. Thanks for reporting!