feat(help): use treesitter for table of contents

Problem: Creating the table of contents for `gO` is complicated.

Solution: Use treesitter instead.

runtime/doc/help.txt

@@ -188,7 +188,7 @@ Local additions ~
 							*local-additions*
 
 ------------------------------------------------------------------------------
-*bars*		Bars example
+Bars example								*bars*
 
 Now that you've jumped here with CTRL-] or a double mouse click, you can use
 CTRL-T, CTRL-O, g<RightMouse>, or <C-RightMouse> to go back to where you were.
@@ -200,5 +200,5 @@ You can use CTRL-] on any word (even if it is not within "|") and Nvim will
 try to find help for it.  Especially for options in single quotes, e.g.
 'hlsearch'.
 
-------------------------------------------------------------------------------
+
  vim:tw=78:isk=!-~,^*,^\|,^\":ts=8:noet:ft=help:norl:

runtime/ftplugin/help.lua

@@ -26,3 +26,7 @@ elseif vim.endswith(bufname, '/doc/lsp.txt') then
     { start = [[\*lsp-semantic-highlight\*]], stop = '^======', match = '^@[%w%p]+' },
   })
 end
+
+vim.keymap.set('n', 'gO', function()
+  require('vim.vimhelp').show_toc()
+end, { buffer = 0, silent = true })

runtime/ftplugin/help.vim

@@ -21,77 +21,5 @@ endif
 " Prefer Vim help instead of manpages.
 setlocal keywordprg=:help
 
-if !exists('g:no_plugin_maps')
-  function! s:show_toc() abort
-    let bufname = bufname('%')
-    let info = getloclist(0, {'winid': 1})
-    if !empty(info) && getwinvar(info.winid, 'qf_toc') ==# bufname
-      lopen
-      return
-    endif
-
-    let toc = []
-    let lnum = 2
-    let last_line = line('$') - 1
-    let last_added = 0
-    let has_section = 0
-    let has_sub_section = 0
-
-    while lnum && lnum <= last_line
-      let level = 0
-      let add_text = ''
-      let text = getline(lnum)
-
-      if text =~# '^=\+$' && lnum + 1 < last_line
-        " A de-facto section heading.  Other headings are inferred.
-        let has_section = 1
-        let has_sub_section = 0
-        let lnum = nextnonblank(lnum + 1)
-        let text = getline(lnum)
-        let add_text = text
-        while add_text =~# '\*[^*]\+\*\s*$'
-          let add_text = matchstr(add_text, '.*\ze\*[^*]\+\*\s*$')
-        endwhile
-      elseif text =~# '^[A-Z0-9][-A-ZA-Z0-9 .][-A-Z0-9 .():]*\%([ \t]\+\*.\+\*\)\?$'
-        " Any line that's yelling is important.
-        let has_sub_section = 1
-        let level = has_section
-        let add_text = matchstr(text, '.\{-}\ze\s*\%([ \t]\+\*.\+\*\)\?$')
-      elseif text =~# '\~$'
-            \ && matchstr(text, '^\s*\zs.\{-}\ze\s*\~$') !~# '\t\|\s\{2,}'
-            \ && getline(lnum - 1) =~# '^\s*<\?$\|^\s*\*.*\*$'
-            \ && getline(lnum + 1) =~# '^\s*>\?$\|^\s*\*.*\*$'
-        " These lines could be headers or code examples.  We only want the
-        " ones that have subsequent lines at the same indent or more.
-        let l = nextnonblank(lnum + 1)
-        if getline(l) =~# '\*[^*]\+\*$'
-          " Ignore tag lines
-          let l = nextnonblank(l + 1)
-        endif
-
-        if indent(lnum) <= indent(l)
-          let level = has_section + has_sub_section
-          let add_text = matchstr(text, '\S.\{-}\ze\s\=\~$')
-        endif
-      endif
-
-      let add_text = substitute(add_text, '\s\+$', '', 'g')
-      if !empty(add_text) && last_added != lnum
-        let last_added = lnum
-        call add(toc, {'bufnr': bufnr('%'), 'lnum': lnum,
-              \ 'text': repeat("\u00a0\u00a0", level) . add_text})
-      endif
-      let lnum = nextnonblank(lnum + 1)
-    endwhile
-
-    call setloclist(0, toc, ' ')
-    call setloclist(0, [], 'a', {'title': 'Help TOC'})
-    lopen
-    let w:qf_toc = bufname
-  endfunction
-
-  nnoremap <silent><buffer> gO :call <sid>show_toc()<cr>
-endif
-
 let &cpo = s:cpo_save
 unlet s:cpo_save

runtime/lua/vim/vimhelp.lua

@@ -30,4 +30,42 @@ function M.highlight_groups(patterns)
   vim.fn.setpos('.', save_cursor)
 end
 
+--- Show a table of contents for the help buffer in a loclist
+function M.show_toc()
+  local bufnr = vim.api.nvim_get_current_buf()
+  local parser = vim.treesitter.get_parser(bufnr, 'vimdoc')
+  local query = vim.treesitter.query.parse(
+    parser:lang(),
+    [[
+    (h1 (heading) @h1)
+    (h2 (heading) @h2)
+    (h3 (heading) @h3)
+    (column_heading (heading) @h4)
+  ]]
+  )
+  local root = parser:parse()[1]:root()
+  local headings = {}
+  for id, node, _, _ in query:iter_captures(root, bufnr) do
+    local text = vim.treesitter.get_node_text(node, bufnr)
+    local capture = query.captures[id]
+    local row, col = node:start()
+    -- only column_headings at col 1 are headings, otherwise it's code examples
+    local is_code = (capture == 'h4' and col > 0)
+    -- ignore tabular material
+    local is_table = (capture == 'h4' and (text:find('\t') or text:find('  ')))
+    -- ignore tag-only headings
+    local is_tag = node:child_count() == 1 and node:child(0):type() == 'tag'
+    if not (is_code or is_table or is_tag) then
+      table.insert(headings, {
+        bufnr = bufnr,
+        lnum = row + 1,
+        text = (capture == 'h3' or capture == 'h4') and '  ' .. text or text,
+      })
+    end
+  end
+  vim.fn.setloclist(0, headings, ' ')
+  vim.fn.setloclist(0, {}, 'a', { title = 'Help TOC' })
+  vim.cmd.lopen()
+end
+
 return M