build: move all generator scripts to src/gen/

- Move all generator Lua scripts to the `src/gen/`
- Add a `.luarc.json` to `src/gen/`
- Add a `preload.lua` to `src/gen/`
  - Add `src` to `package.path` so it aligns with `.luarc.json'
- Fix all `require` statements in `src/gen/` so they are consistent:
    - `require('scripts.foo')` -> `require('gen.foo')`
    - `require('src.nvim.options')` -> `require('nvim.options')`
    - `require('api.dispatch_deprecated')` -> `require('nvim.api.dispatch_deprecated')`

src/.luarc.json

@@ -0,0 +1,24 @@
+{
+  "$schema": "https://raw.githubusercontent.com/LuaLS/vscode-lua/master/setting/schema.json",
+  "runtime": {
+    "version": "LuaJIT"
+  },
+  "workspace": {
+    "library": [
+      "../runtime/lua",
+      "${3rd}/luv/library"
+    ],
+    "checkThirdParty": "Disable"
+  },
+  "diagnostics": {
+    "groupFileStatus": {
+      "strict": "Opened",
+      "strong": "Opened"
+    },
+    "groupSeverity": {
+      "strong": "Warning",
+      "strict": "Warning"
+    },
+    "unusedLocalExclude": [ "_*" ]
+  }
+}

src/gen/cdoc_grammar.lua

@@ -0,0 +1,87 @@
+--[[!
+LPEG grammar for C doc comments
+]]
+
+--- @class nvim.cdoc.Param
+--- @field kind 'param'
+--- @field name string
+--- @field desc? string
+
+--- @class nvim.cdoc.Return
+--- @field kind 'return'
+--- @field desc string
+
+--- @class nvim.cdoc.Note
+--- @field desc? string
+
+--- @alias nvim.cdoc.grammar.result
+--- | nvim.cdoc.Param
+--- | nvim.cdoc.Return
+--- | nvim.cdoc.Note
+
+--- @class nvim.cdoc.grammar
+--- @field match fun(self, input: string): nvim.cdoc.grammar.result?
+
+local lpeg = vim.lpeg
+local P, R, S = lpeg.P, lpeg.R, lpeg.S
+local Ct, Cg = lpeg.Ct, lpeg.Cg
+
+--- @param x vim.lpeg.Pattern
+local function rep(x)
+  return x ^ 0
+end
+
+--- @param x vim.lpeg.Pattern
+local function rep1(x)
+  return x ^ 1
+end
+
+--- @param x vim.lpeg.Pattern
+local function opt(x)
+  return x ^ -1
+end
+
+local nl = P('\r\n') + P('\n')
+local ws = rep1(S(' \t') + nl)
+
+local any = P(1) -- (consume one character)
+local letter = R('az', 'AZ') + S('_$')
+local ident = letter * rep(letter + R('09'))
+
+local io = P('[') * (P('in') + P('out') + P('inout')) * P(']')
+
+--- @param x string
+local function Pf(x)
+  return opt(ws) * P(x) * opt(ws)
+end
+
+--- @type table<string,vim.lpeg.Pattern>
+local v = setmetatable({}, {
+  __index = function(_, k)
+    return lpeg.V(k)
+  end,
+})
+
+local grammar = P {
+  rep1(P('@') * v.ats),
+
+  ats = v.at_param + v.at_return + v.at_deprecated + v.at_see + v.at_brief + v.at_note + v.at_nodoc,
+
+  at_param = Ct(
+    Cg(P('param'), 'kind') * opt(io) * ws * Cg(ident, 'name') * opt(ws * Cg(rep(any), 'desc'))
+  ),
+
+  at_return = Ct(Cg(P('return'), 'kind') * opt(S('s')) * opt(ws * Cg(rep(any), 'desc'))),
+
+  at_deprecated = Ct(Cg(P('deprecated'), 'kind')),
+
+  at_see = Ct(Cg(P('see'), 'kind') * ws * opt(Pf('#')) * Cg(rep(any), 'desc')),
+
+  at_brief = Ct(Cg(P('brief'), 'kind') * ws * Cg(rep(any), 'desc')),
+
+  at_note = Ct(Cg(P('note'), 'kind') * ws * Cg(rep(any), 'desc')),
+
+  at_nodoc = Ct(Cg(P('nodoc'), 'kind')),
+}
+
+return grammar --[[@as nvim.cdoc.grammar]]

src/gen/cdoc_parser.lua

@@ -0,0 +1,223 @@
+local cdoc_grammar = require('gen.cdoc_grammar')
+local c_grammar = require('gen.c_grammar')
+
+--- @class nvim.cdoc.parser.param
+--- @field name string
+--- @field type string
+--- @field desc string
+
+--- @class nvim.cdoc.parser.return
+--- @field name string
+--- @field type string
+--- @field desc string
+
+--- @class nvim.cdoc.parser.note
+--- @field desc string
+
+--- @class nvim.cdoc.parser.brief
+--- @field kind 'brief'
+--- @field desc string
+
+--- @class nvim.cdoc.parser.fun
+--- @field name string
+--- @field params nvim.cdoc.parser.param[]
+--- @field returns nvim.cdoc.parser.return[]
+--- @field desc string
+--- @field deprecated? true
+--- @field since? string
+--- @field attrs? string[]
+--- @field nodoc? true
+--- @field notes? nvim.cdoc.parser.note[]
+--- @field see? nvim.cdoc.parser.note[]
+
+--- @class nvim.cdoc.parser.State
+--- @field doc_lines? string[]
+--- @field cur_obj? nvim.cdoc.parser.obj
+--- @field last_doc_item? nvim.cdoc.parser.param|nvim.cdoc.parser.return|nvim.cdoc.parser.note
+--- @field last_doc_item_indent? integer
+
+--- @alias nvim.cdoc.parser.obj
+--- | nvim.cdoc.parser.fun
+--- | nvim.cdoc.parser.brief
+
+--- If we collected any `---` lines. Add them to the existing (or new) object
+--- Used for function/class descriptions and multiline param descriptions.
+--- @param state nvim.cdoc.parser.State
+local function add_doc_lines_to_obj(state)
+  if state.doc_lines then
+    state.cur_obj = state.cur_obj or {}
+    local cur_obj = assert(state.cur_obj)
+    local txt = table.concat(state.doc_lines, '\n')
+    if cur_obj.desc then
+      cur_obj.desc = cur_obj.desc .. '\n' .. txt
+    else
+      cur_obj.desc = txt
+    end
+    state.doc_lines = nil
+  end
+end
+
+--- @param line string
+--- @param state nvim.cdoc.parser.State
+local function process_doc_line(line, state)
+  line = line:gsub('^%s+@', '@')
+
+  local parsed = cdoc_grammar:match(line)
+
+  if not parsed then
+    if line:match('^ ') then
+      line = line:sub(2)
+    end
+
+    if state.last_doc_item then
+      if not state.last_doc_item_indent then
+        state.last_doc_item_indent = #line:match('^%s*') + 1
+      end
+      state.last_doc_item.desc = (state.last_doc_item.desc or '')
+        .. '\n'
+        .. line:sub(state.last_doc_item_indent or 1)
+    else
+      state.doc_lines = state.doc_lines or {}
+      table.insert(state.doc_lines, line)
+    end
+    return
+  end
+
+  state.last_doc_item_indent = nil
+  state.last_doc_item = nil
+
+  local kind = parsed.kind
+
+  state.cur_obj = state.cur_obj or {}
+  local cur_obj = assert(state.cur_obj)
+
+  if kind == 'brief' then
+    state.cur_obj = {
+      kind = 'brief',
+      desc = parsed.desc,
+    }
+  elseif kind == 'param' then
+    state.last_doc_item_indent = nil
+    cur_obj.params = cur_obj.params or {}
+    state.last_doc_item = {
+      name = parsed.name,
+      desc = parsed.desc,
+    }
+    table.insert(cur_obj.params, state.last_doc_item)
+  elseif kind == 'return' then
+    cur_obj.returns = { {
+      desc = parsed.desc,
+    } }
+    state.last_doc_item_indent = nil
+    state.last_doc_item = cur_obj.returns[1]
+  elseif kind == 'deprecated' then
+    cur_obj.deprecated = true
+  elseif kind == 'nodoc' then
+    cur_obj.nodoc = true
+  elseif kind == 'since' then
+    cur_obj.since = parsed.desc
+  elseif kind == 'see' then
+    cur_obj.see = cur_obj.see or {}
+    table.insert(cur_obj.see, { desc = parsed.desc })
+  elseif kind == 'note' then
+    state.last_doc_item_indent = nil
+    state.last_doc_item = {
+      desc = parsed.desc,
+    }
+    cur_obj.notes = cur_obj.notes or {}
+    table.insert(cur_obj.notes, state.last_doc_item)
+  else
+    error('Unhandled' .. vim.inspect(parsed))
+  end
+end
+
+--- @param item table
+--- @param state nvim.cdoc.parser.State
+local function process_proto(item, state)
+  state.cur_obj = state.cur_obj or {}
+  local cur_obj = assert(state.cur_obj)
+  cur_obj.name = item.name
+  cur_obj.params = cur_obj.params or {}
+
+  for _, p in ipairs(item.parameters) do
+    local param = { name = p[2], type = p[1] }
+    local added = false
+    for _, cp in ipairs(cur_obj.params) do
+      if cp.name == param.name then
+        cp.type = param.type
+        added = true
+        break
+      end
+    end
+
+    if not added then
+      table.insert(cur_obj.params, param)
+    end
+  end
+
+  cur_obj.returns = cur_obj.returns or { {} }
+  cur_obj.returns[1].type = item.return_type
+
+  for _, a in ipairs({
+    'fast',
+    'remote_only',
+    'lua_only',
+    'textlock',
+    'textlock_allow_cmdwin',
+  }) do
+    if item[a] then
+      cur_obj.attrs = cur_obj.attrs or {}
+      table.insert(cur_obj.attrs, a)
+    end
+  end
+
+  cur_obj.deprecated_since = item.deprecated_since
+
+  -- Remove some arguments
+  for i = #cur_obj.params, 1, -1 do
+    local p = cur_obj.params[i]
+    if p.name == 'channel_id' or vim.tbl_contains({ 'lstate', 'arena', 'error' }, p.type) then
+      table.remove(cur_obj.params, i)
+    end
+  end
+end
+
+local M = {}
+
+--- @param filename string
+--- @return {} classes
+--- @return nvim.cdoc.parser.fun[] funs
+--- @return string[] briefs
+function M.parse(filename)
+  local funs = {} --- @type nvim.cdoc.parser.fun[]
+  local briefs = {} --- @type string[]
+  local state = {} --- @type nvim.cdoc.parser.State
+
+  local txt = assert(io.open(filename, 'r')):read('*all')
+
+  local parsed = c_grammar.grammar:match(txt)
+  for _, item in ipairs(parsed) do
+    if item.comment then
+      process_doc_line(item.comment, state)
+    else
+      add_doc_lines_to_obj(state)
+      if item[1] == 'proto' then
+        process_proto(item, state)
+        table.insert(funs, state.cur_obj)
+      end
+      local cur_obj = state.cur_obj
+      if cur_obj and not item.static then
+        if cur_obj.kind == 'brief' then
+          table.insert(briefs, cur_obj.desc)
+        end
+      end
+      state = {}
+    end
+  end
+
+  return {}, funs, briefs
+end
+
+-- M.parse('src/nvim/api/vim.c')
+
+return M

src/gen/gen_api_dispatch.lua

@@ -7,7 +7,7 @@
 
 local mpack = vim.mpack
 
-local hashy = require 'generators.hashy'
+local hashy = require 'gen.hashy'
 
 local pre_args = 7
 assert(#arg >= pre_args)
@@ -31,7 +31,7 @@ local headers = {}
 -- set of function names, used to detect duplicates
 local function_names = {}
 
-local c_grammar = require('generators.c_grammar')
+local c_grammar = require('gen.c_grammar')
 
 local startswith = vim.startswith
 
@@ -150,7 +150,7 @@ end
 
 -- Export functions under older deprecated names.
 -- These will be removed eventually.
-local deprecated_aliases = require('api.dispatch_deprecated')
+local deprecated_aliases = require('nvim.api.dispatch_deprecated')
 for _, f in ipairs(shallowcopy(functions)) do
   local ismethod = false
   if startswith(f.name, 'nvim_') then
@@ -300,7 +300,7 @@ for i, item in ipairs(types) do
 end
 
 local packed = table.concat(pieces)
-local dump_bin_array = require('generators.dump_bin_array')
+local dump_bin_array = require('gen.dump_bin_array')
 dump_bin_array(api_metadata_output, 'packed_api_metadata', packed)
 api_metadata_output:close()
 

src/gen/gen_api_ui_events.lua

@@ -7,10 +7,10 @@ local remote_output = io.open(arg[3], 'wb')
 local metadata_output = io.open(arg[4], 'wb')
 local client_output = io.open(arg[5], 'wb')
 
-local c_grammar = require('generators.c_grammar')
+local c_grammar = require('gen.c_grammar')
 local events = c_grammar.grammar:match(input:read('*all'))
 
-local hashy = require 'generators.hashy'
+local hashy = require 'gen.hashy'
 
 local function write_signature(output, ev, prefix, notype)
   output:write('(' .. prefix)

src/gen/gen_declarations.lua

@@ -1,4 +1,4 @@
-local grammar = require('generators.c_grammar').grammar
+local grammar = require('gen.c_grammar').grammar
 
 --- @param fname string
 --- @return string?

src/gen/gen_eval.lua

@@ -8,7 +8,7 @@ local funcsfname = autodir .. '/funcs.generated.h'
 
 --Will generate funcs.generated.h with definition of functions static const array.
 
-local hashy = require 'generators.hashy'
+local hashy = require 'gen.hashy'
 
 local hashpipe = assert(io.open(funcsfname, 'wb'))
 
@@ -47,7 +47,7 @@ hashpipe:write([[
 
 ]])
 
-local funcs = require('eval').funcs
+local funcs = require('nvim.eval').funcs
 for _, func in pairs(funcs) do
   if func.float_func then
     func.func = 'float_op_wrapper'

src/gen/gen_events.lua

@@ -1,7 +1,7 @@
 local fileio_enum_file = arg[1]
 local names_file = arg[2]
 
-local auevents = require('auevents')
+local auevents = require('nvim.auevents')
 local events = auevents.events
 
 local enum_tgt = io.open(fileio_enum_file, 'w')

src/gen/gen_ex_cmds.lua

@@ -11,7 +11,7 @@ local enumfile = io.open(enumfname, 'w')
 local defsfile = io.open(defsfname, 'w')
 
 local bit = require 'bit'
-local ex_cmds = require('ex_cmds')
+local ex_cmds = require('nvim.ex_cmds')
 local defs = ex_cmds.cmds
 local flags = ex_cmds.flags
 

src/gen/gen_filetype.lua

@@ -0,0 +1,209 @@
+local do_not_run = true
+if do_not_run then
+  print([[
+    This script was used to bootstrap the filetype patterns in runtime/lua/vim/filetype.lua. It
+    should no longer be used except for testing purposes. New filetypes, or changes to existing
+    filetypes, should be ported manually as part of the vim-patch process.
+  ]])
+  return
+end
+
+local filetype_vim = 'runtime/filetype.vim'
+local filetype_lua = 'runtime/lua/vim/filetype.lua'
+
+local keywords = {
+  ['for'] = true,
+  ['or'] = true,
+  ['and'] = true,
+  ['end'] = true,
+  ['do'] = true,
+  ['if'] = true,
+  ['while'] = true,
+  ['repeat'] = true,
+}
+
+local sections = {
+  extension = { str = {}, func = {} },
+  filename = { str = {}, func = {} },
+  pattern = { str = {}, func = {} },
+}
+
+local specialchars = '%*%?\\%$%[%]%{%}'
+
+local function add_pattern(pat, ft)
+  local ok = true
+
+  -- Patterns that start or end with { or } confuse splitting on commas and make parsing harder, so just skip those
+  if not string.find(pat, '^%{') and not string.find(pat, '%}$') then
+    for part in string.gmatch(pat, '[^,]+') do
+      if not string.find(part, '[' .. specialchars .. ']') then
+        if type(ft) == 'string' then
+          sections.filename.str[part] = ft
+        else
+          sections.filename.func[part] = ft
+        end
+      elseif string.match(part, '^%*%.[^%./' .. specialchars .. ']+$') then
+        if type(ft) == 'string' then
+          sections.extension.str[part:sub(3)] = ft
+        else
+          sections.extension.func[part:sub(3)] = ft
+        end
+      else
+        if string.match(part, '^%*/[^' .. specialchars .. ']+$') then
+          -- For patterns matching */some/pattern we want to easily match files
+          -- with path /some/pattern, so include those in filename detection
+          if type(ft) == 'string' then
+            sections.filename.str[part:sub(2)] = ft
+          else
+            sections.filename.func[part:sub(2)] = ft
+          end
+        end
+
+        if string.find(part, '^[%w-_.*?%[%]/]+$') then
+          local p = part:gsub('%.', '%%.'):gsub('%*', '.*'):gsub('%?', '.')
+          -- Insert into array to maintain order rather than setting
+          -- key-value directly
+          if type(ft) == 'string' then
+            sections.pattern.str[p] = ft
+          else
+            sections.pattern.func[p] = ft
+          end
+        else
+          ok = false
+        end
+      end
+    end
+  end
+
+  return ok
+end
+
+local function parse_line(line)
+  local pat, ft
+  pat, ft = line:match('^%s*au%a* Buf[%a,]+%s+(%S+)%s+setf%s+(%S+)')
+  if pat then
+    return add_pattern(pat, ft)
+  else
+    local func
+    pat, func = line:match('^%s*au%a* Buf[%a,]+%s+(%S+)%s+call%s+(%S+)')
+    if pat then
+      return add_pattern(pat, function()
+        return func
+      end)
+    end
+  end
+end
+
+local unparsed = {}
+local full_line
+for line in io.lines(filetype_vim) do
+  local cont = string.match(line, '^%s*\\%s*(.*)$')
+  if cont then
+    full_line = full_line .. ' ' .. cont
+  else
+    if full_line then
+      if not parse_line(full_line) and string.find(full_line, '^%s*au%a* Buf') then
+        table.insert(unparsed, full_line)
+      end
+    end
+    full_line = line
+  end
+end
+
+if #unparsed > 0 then
+  print('Failed to parse the following patterns:')
+  for _, v in ipairs(unparsed) do
+    print(v)
+  end
+end
+
+local function add_item(indent, key, ft)
+  if type(ft) == 'string' then
+    if string.find(key, '%A') or keywords[key] then
+      key = string.format('["%s"]', key)
+    end
+    return string.format([[%s%s = "%s",]], indent, key, ft)
+  elseif type(ft) == 'function' then
+    local func = ft()
+    if string.find(key, '%A') or keywords[key] then
+      key = string.format('["%s"]', key)
+    end
+    -- Right now only a single argument is supported, which covers
+    -- everything in filetype.vim as of this writing
+    local arg = string.match(func, '%((.*)%)$')
+    func = string.gsub(func, '%(.*$', '')
+    if arg == '' then
+      -- Function with no arguments, call the function directly
+      return string.format([[%s%s = function() vim.fn["%s"]() end,]], indent, key, func)
+    elseif string.match(arg, [[^(["']).*%1$]]) then
+      -- String argument
+      if func == 's:StarSetf' then
+        return string.format([[%s%s = starsetf(%s),]], indent, key, arg)
+      else
+        return string.format([[%s%s = function() vim.fn["%s"](%s) end,]], indent, key, func, arg)
+      end
+    elseif string.find(arg, '%(') then
+      -- Function argument
+      return string.format(
+        [[%s%s = function() vim.fn["%s"](vim.fn.%s) end,]],
+        indent,
+        key,
+        func,
+        arg
+      )
+    else
+      assert(false, arg)
+    end
+  end
+end
+
+do
+  local lines = {}
+  local start = false
+  for line in io.lines(filetype_lua) do
+    if line:match('^%s+-- END [A-Z]+$') then
+      start = false
+    end
+
+    if not start then
+      table.insert(lines, line)
+    end
+
+    local indent, section = line:match('^(%s+)-- BEGIN ([A-Z]+)$')
+    if section then
+      start = true
+      local t = sections[string.lower(section)]
+
+      local sorted = {}
+      for k, v in pairs(t.str) do
+        table.insert(sorted, { [k] = v })
+      end
+
+      table.sort(sorted, function(a, b)
+        return a[next(a)] < b[next(b)]
+      end)
+
+      for _, v in ipairs(sorted) do
+        local k = next(v)
+        table.insert(lines, add_item(indent, k, v[k]))
+      end
+
+      sorted = {}
+      for k, v in pairs(t.func) do
+        table.insert(sorted, { [k] = v })
+      end
+
+      table.sort(sorted, function(a, b)
+        return next(a) < next(b)
+      end)
+
+      for _, v in ipairs(sorted) do
+        local k = next(v)
+        table.insert(lines, add_item(indent, k, v[k]))
+      end
+    end
+  end
+  local f = io.open(filetype_lua, 'w')
+  f:write(table.concat(lines, '\n') .. '\n')
+  f:close()
+end

src/gen/gen_lsp.lua

@@ -0,0 +1,514 @@
+-- Generates lua-ls annotations for lsp.
+
+local USAGE = [[
+Generates lua-ls annotations for lsp.
+
+USAGE:
+nvim -l src/gen/gen_lsp.lua gen  # by default, this will overwrite runtime/lua/vim/lsp/_meta/protocol.lua
+nvim -l src/gen/gen_lsp.lua gen --version 3.18 --out runtime/lua/vim/lsp/_meta/protocol.lua
+nvim -l src/gen/gen_lsp.lua gen --version 3.18 --methods --capabilities
+]]
+
+local DEFAULT_LSP_VERSION = '3.18'
+
+local M = {}
+
+local function tofile(fname, text)
+  local f = io.open(fname, 'w')
+  if not f then
+    error(('failed to write: %s'):format(f))
+  else
+    print(('Written to: %s'):format(fname))
+    f:write(text)
+    f:close()
+  end
+end
+
+--- The LSP protocol JSON data (it's partial, non-exhaustive).
+--- https://raw.githubusercontent.com/microsoft/language-server-protocol/gh-pages/_specifications/lsp/3.18/metaModel/metaModel.schema.json
+--- @class vim._gen_lsp.Protocol
+--- @field requests vim._gen_lsp.Request[]
+--- @field notifications vim._gen_lsp.Notification[]
+--- @field structures vim._gen_lsp.Structure[]
+--- @field enumerations vim._gen_lsp.Enumeration[]
+--- @field typeAliases vim._gen_lsp.TypeAlias[]
+
+---@param opt vim._gen_lsp.opt
+---@return vim._gen_lsp.Protocol
+local function read_json(opt)
+  local uri = 'https://raw.githubusercontent.com/microsoft/language-server-protocol/gh-pages/_specifications/lsp/'
+    .. opt.version
+    .. '/metaModel/metaModel.json'
+  print('Reading ' .. uri)
+
+  local res = vim.system({ 'curl', '--no-progress-meter', uri, '-o', '-' }):wait()
+  if res.code ~= 0 or (res.stdout or ''):len() < 999 then
+    print(('URL failed: %s'):format(uri))
+    vim.print(res)
+    error(res.stdout)
+  end
+  return vim.json.decode(res.stdout)
+end
+
+-- Gets the Lua symbol for a given fully-qualified LSP method name.
+local function to_luaname(s)
+  -- "$/" prefix is special: https://microsoft.github.io/language-server-protocol/specification/#dollarRequests
+  return s:gsub('^%$', 'dollar'):gsub('/', '_')
+end
+
+---@param protocol vim._gen_lsp.Protocol
+---@param gen_methods boolean
+---@param gen_capabilities boolean
+local function write_to_protocol(protocol, gen_methods, gen_capabilities)
+  if not gen_methods and not gen_capabilities then
+    return
+  end
+
+  local indent = (' '):rep(2)
+
+  --- @class vim._gen_lsp.Request
+  --- @field deprecated? string
+  --- @field documentation? string
+  --- @field messageDirection string
+  --- @field clientCapability? string
+  --- @field serverCapability? string
+  --- @field method string
+  --- @field params? any
+  --- @field proposed? boolean
+  --- @field registrationMethod? string
+  --- @field registrationOptions? any
+  --- @field since? string
+
+  --- @class vim._gen_lsp.Notification
+  --- @field deprecated? string
+  --- @field documentation? string
+  --- @field errorData? any
+  --- @field messageDirection string
+  --- @field clientCapability? string
+  --- @field serverCapability? string
+  --- @field method string
+  --- @field params? any[]
+  --- @field partialResult? any
+  --- @field proposed? boolean
+  --- @field registrationMethod? string
+  --- @field registrationOptions? any
+  --- @field result any
+  --- @field since? string
+
+  ---@type (vim._gen_lsp.Request|vim._gen_lsp.Notification)[]
+  local all = vim.list_extend(protocol.requests, protocol.notifications)
+  table.sort(all, function(a, b)
+    return to_luaname(a.method) < to_luaname(b.method)
+  end)
+
+  local output = { '-- Generated by gen_lsp.lua, keep at end of file.' }
+
+  if gen_methods then
+    output[#output + 1] = '--- @alias vim.lsp.protocol.Method.ClientToServer'
+
+    for _, item in ipairs(all) do
+      if item.method and item.messageDirection == 'clientToServer' then
+        output[#output + 1] = ("--- | '%s',"):format(item.method)
+      end
+    end
+
+    vim.list_extend(output, {
+      '',
+      '--- @alias vim.lsp.protocol.Method.ServerToClient',
+    })
+    for _, item in ipairs(all) do
+      if item.method and item.messageDirection == 'serverToClient' then
+        output[#output + 1] = ("--- | '%s',"):format(item.method)
+      end
+    end
+
+    vim.list_extend(output, {
+      '',
+      '--- @alias vim.lsp.protocol.Method',
+      '--- | vim.lsp.protocol.Method.ClientToServer',
+      '--- | vim.lsp.protocol.Method.ServerToClient',
+      '',
+      '-- Generated by gen_lsp.lua, keep at end of file.',
+      '--- @enum vim.lsp.protocol.Methods',
+      '--- @see https://microsoft.github.io/language-server-protocol/specification/#metaModel',
+      '--- LSP method names.',
+      'protocol.Methods = {',
+    })
+
+    for _, item in ipairs(all) do
+      if item.method then
+        if item.documentation then
+          local document = vim.split(item.documentation, '\n?\n', { trimempty = true })
+          for _, docstring in ipairs(document) do
+            output[#output + 1] = indent .. '--- ' .. docstring
+          end
+        end
+        output[#output + 1] = ("%s%s = '%s',"):format(indent, to_luaname(item.method), item.method)
+      end
+    end
+    output[#output + 1] = '}'
+  end
+
+  if gen_capabilities then
+    vim.list_extend(output, {
+      '',
+      '-- stylua: ignore start',
+      '-- Generated by gen_lsp.lua, keep at end of file.',
+      '--- Maps method names to the required server capability',
+      'protocol._request_name_to_capability = {',
+    })
+
+    for _, item in ipairs(all) do
+      if item.serverCapability then
+        output[#output + 1] = ("%s['%s'] = { %s },"):format(
+          indent,
+          item.method,
+          table.concat(
+            vim
+              .iter(vim.split(item.serverCapability, '.', { plain = true }))
+              :map(function(segment)
+                return "'" .. segment .. "'"
+              end)
+              :totable(),
+            ', '
+          )
+        )
+      end
+    end
+
+    output[#output + 1] = '}'
+    output[#output + 1] = '-- stylua: ignore end'
+  end
+
+  output[#output + 1] = ''
+  output[#output + 1] = 'return protocol'
+
+  local fname = './runtime/lua/vim/lsp/protocol.lua'
+  local bufnr = vim.fn.bufadd(fname)
+  vim.fn.bufload(bufnr)
+  vim.api.nvim_set_current_buf(bufnr)
+  local lines = vim.api.nvim_buf_get_lines(bufnr, 0, -1, false)
+  local index = vim.iter(ipairs(lines)):find(function(key, item)
+    return vim.startswith(item, '-- Generated by') and key or nil
+  end)
+  index = index and index - 1 or vim.api.nvim_buf_line_count(bufnr) - 1
+  vim.api.nvim_buf_set_lines(bufnr, index, -1, true, output)
+  vim.cmd.write()
+end
+
+---@class vim._gen_lsp.opt
+---@field output_file string
+---@field version string
+---@field methods boolean
+---@field capabilities boolean
+
+---@param opt vim._gen_lsp.opt
+function M.gen(opt)
+  --- @type vim._gen_lsp.Protocol
+  local protocol = read_json(opt)
+
+  write_to_protocol(protocol, opt.methods, opt.capabilities)
+
+  local output = {
+    '--' .. '[[',
+    'THIS FILE IS GENERATED by scr/gen/gen_lsp.lua',
+    'DO NOT EDIT MANUALLY',
+    '',
+    'Based on LSP protocol ' .. opt.version,
+    '',
+    'Regenerate:',
+    ([=[nvim -l scr/gen/gen_lsp.lua gen --version %s]=]):format(DEFAULT_LSP_VERSION),
+    '--' .. ']]',
+    '',
+    '---@meta',
+    "error('Cannot require a meta file')",
+    '',
+    '---@alias lsp.null nil',
+    '---@alias uinteger integer',
+    '---@alias decimal number',
+    '---@alias lsp.DocumentUri string',
+    '---@alias lsp.URI string',
+    '',
+  }
+
+  local anonymous_num = 0
+
+  ---@type string[]
+  local anonym_classes = {}
+
+  local simple_types = {
+    'string',
+    'boolean',
+    'integer',
+    'uinteger',
+    'decimal',
+  }
+
+  ---@param documentation string
+  local _process_documentation = function(documentation)
+    documentation = documentation:gsub('\n', '\n---')
+    -- Remove <200b> (zero-width space) unicode characters: e.g., `**/<200b>*`
+    documentation = documentation:gsub('\226\128\139', '')
+    -- Escape annotations that are not recognized by lua-ls
+    documentation = documentation:gsub('%^---@sample', '---\\@sample')
+    return '---' .. documentation
+  end
+
+  --- @class vim._gen_lsp.Type
+  --- @field kind string a common field for all Types.
+  --- @field name? string for ReferenceType, BaseType
+  --- @field element? any for ArrayType
+  --- @field items? vim._gen_lsp.Type[] for OrType, AndType
+  --- @field key? vim._gen_lsp.Type for MapType
+  --- @field value? string|vim._gen_lsp.Type for StringLiteralType, MapType, StructureLiteralType
+
+  ---@param type vim._gen_lsp.Type
+  ---@param prefix? string Optional prefix associated with the this type, made of (nested) field name.
+  ---             Used to generate class name for structure literal types.
+  ---@return string
+  local function parse_type(type, prefix)
+    -- ReferenceType | BaseType
+    if type.kind == 'reference' or type.kind == 'base' then
+      if vim.tbl_contains(simple_types, type.name) then
+        return type.name
+      end
+      return 'lsp.' .. type.name
+
+    -- ArrayType
+    elseif type.kind == 'array' then
+      local parsed_items = parse_type(type.element, prefix)
+      if type.element.items and #type.element.items > 1 then
+        parsed_items = '(' .. parsed_items .. ')'
+      end
+      return parsed_items .. '[]'
+
+    -- OrType
+    elseif type.kind == 'or' then
+      local val = ''
+      for _, item in ipairs(type.items) do
+        val = val .. parse_type(item, prefix) .. '|' --[[ @as string ]]
+      end
+      val = val:sub(0, -2)
+      return val
+
+    -- StringLiteralType
+    elseif type.kind == 'stringLiteral' then
+      return '"' .. type.value .. '"'
+
+    -- MapType
+    elseif type.kind == 'map' then
+      local key = assert(type.key)
+      local value = type.value --[[ @as vim._gen_lsp.Type ]]
+      return 'table<' .. parse_type(key, prefix) .. ', ' .. parse_type(value, prefix) .. '>'
+
+    -- StructureLiteralType
+    elseif type.kind == 'literal' then
+      -- can I use ---@param disabled? {reason: string}
+      -- use | to continue the inline class to be able to add docs
+      -- https://github.com/LuaLS/lua-language-server/issues/2128
+      anonymous_num = anonymous_num + 1
+      local anonymous_classname = 'lsp._anonym' .. anonymous_num
+      if prefix then
+        anonymous_classname = anonymous_classname .. '.' .. prefix
+      end
+      local anonym = vim
+        .iter({
+          (anonymous_num > 1 and { '' } or {}),
+          { '---@class ' .. anonymous_classname },
+        })
+        :flatten()
+        :totable()
+
+      --- @class vim._gen_lsp.StructureLiteral translated to anonymous @class.
+      --- @field deprecated? string
+      --- @field description? string
+      --- @field properties vim._gen_lsp.Property[]
+      --- @field proposed? boolean
+      --- @field since? string
+
+      ---@type vim._gen_lsp.StructureLiteral
+      local structural_literal = assert(type.value) --[[ @as vim._gen_lsp.StructureLiteral ]]
+      for _, field in ipairs(structural_literal.properties) do
+        anonym[#anonym + 1] = '---'
+        if field.documentation then
+          anonym[#anonym + 1] = _process_documentation(field.documentation)
+        end
+        anonym[#anonym + 1] = '---@field '
+          .. field.name
+          .. (field.optional and '?' or '')
+          .. ' '
+          .. parse_type(field.type, prefix .. '.' .. field.name)
+      end
+      -- anonym[#anonym + 1] = ''
+      for _, line in ipairs(anonym) do
+        if line then
+          anonym_classes[#anonym_classes + 1] = line
+        end
+      end
+      return anonymous_classname
+
+    -- TupleType
+    elseif type.kind == 'tuple' then
+      local tuple = '['
+      for _, value in ipairs(type.items) do
+        tuple = tuple .. parse_type(value, prefix) .. ', '
+      end
+      -- remove , at the end
+      tuple = tuple:sub(0, -3)
+      return tuple .. ']'
+    end
+
+    vim.print('WARNING: Unknown type ', type)
+    return ''
+  end
+
+  --- @class vim._gen_lsp.Structure translated to @class
+  --- @field deprecated? string
+  --- @field documentation? string
+  --- @field extends? { kind: string, name: string }[]
+  --- @field mixins? { kind: string, name: string }[]
+  --- @field name string
+  --- @field properties? vim._gen_lsp.Property[]  members, translated to @field
+  --- @field proposed? boolean
+  --- @field since? string
+  for _, structure in ipairs(protocol.structures) do
+    -- output[#output + 1] = ''
+    if structure.documentation then
+      output[#output + 1] = _process_documentation(structure.documentation)
+    end
+    local class_string = ('---@class lsp.%s'):format(structure.name)
+    if structure.extends or structure.mixins then
+      local inherits_from = table.concat(
+        vim.list_extend(
+          vim.tbl_map(parse_type, structure.extends or {}),
+          vim.tbl_map(parse_type, structure.mixins or {})
+        ),
+        ', '
+      )
+      class_string = class_string .. ': ' .. inherits_from
+    end
+    output[#output + 1] = class_string
+
+    --- @class vim._gen_lsp.Property translated to @field
+    --- @field deprecated? string
+    --- @field documentation? string
+    --- @field name string
+    --- @field optional? boolean
+    --- @field proposed? boolean
+    --- @field since? string
+    --- @field type { kind: string, name: string }
+    for _, field in ipairs(structure.properties or {}) do
+      output[#output + 1] = '---' -- Insert a single newline between @fields (and after @class)
+      if field.documentation then
+        output[#output + 1] = _process_documentation(field.documentation)
+      end
+      output[#output + 1] = '---@field '
+        .. field.name
+        .. (field.optional and '?' or '')
+        .. ' '
+        .. parse_type(field.type, field.name)
+    end
+    output[#output + 1] = ''
+  end
+
+  --- @class vim._gen_lsp.Enumeration translated to @enum
+  --- @field deprecated string?
+  --- @field documentation string?
+  --- @field name string?
+  --- @field proposed boolean?
+  --- @field since string?
+  --- @field suportsCustomValues boolean?
+  --- @field values { name: string, value: string, documentation?: string, since?: string }[]
+  for _, enum in ipairs(protocol.enumerations) do
+    if enum.documentation then
+      output[#output + 1] = _process_documentation(enum.documentation)
+    end
+    local enum_type = '---@alias lsp.' .. enum.name
+    for _, value in ipairs(enum.values) do
+      enum_type = enum_type
+        .. '\n---| '
+        .. (type(value.value) == 'string' and '"' .. value.value .. '"' or value.value)
+        .. ' # '
+        .. value.name
+    end
+    output[#output + 1] = enum_type
+    output[#output + 1] = ''
+  end
+
+  --- @class vim._gen_lsp.TypeAlias translated to @alias
+  --- @field deprecated? string?
+  --- @field documentation? string
+  --- @field name string
+  --- @field proposed? boolean
+  --- @field since? string
+  --- @field type vim._gen_lsp.Type
+  for _, alias in ipairs(protocol.typeAliases) do
+    if alias.documentation then
+      output[#output + 1] = _process_documentation(alias.documentation)
+    end
+    if alias.type.kind == 'or' then
+      local alias_type = '---@alias lsp.' .. alias.name .. ' '
+      for _, item in ipairs(alias.type.items) do
+        alias_type = alias_type .. parse_type(item, alias.name) .. '|'
+      end
+      alias_type = alias_type:sub(0, -2)
+      output[#output + 1] = alias_type
+    else
+      output[#output + 1] = '---@alias lsp.'
+        .. alias.name
+        .. ' '
+        .. parse_type(alias.type, alias.name)
+    end
+    output[#output + 1] = ''
+  end
+
+  -- anonymous classes
+  for _, line in ipairs(anonym_classes) do
+    output[#output + 1] = line
+  end
+
+  tofile(opt.output_file, table.concat(output, '\n') .. '\n')
+end
+
+---@type vim._gen_lsp.opt
+local opt = {
+  output_file = 'runtime/lua/vim/lsp/_meta/protocol.lua',
+  version = DEFAULT_LSP_VERSION,
+  methods = false,
+  capabilities = false,
+}
+
+local command = nil
+local i = 1
+while i <= #_G.arg do
+  if _G.arg[i] == '--out' then
+    opt.output_file = assert(_G.arg[i + 1], '--out <outfile> needed')
+    i = i + 1
+  elseif _G.arg[i] == '--version' then
+    opt.version = assert(_G.arg[i + 1], '--version <version> needed')
+    i = i + 1
+  elseif _G.arg[i] == '--methods' then
+    opt.methods = true
+  elseif _G.arg[i] == '--capabilities' then
+    opt.capabilities = true
+  elseif vim.startswith(_G.arg[i], '-') then
+    error('Unrecognized args: ' .. _G.arg[i])
+  else
+    if command then
+      error('More than one command was given: ' .. _G.arg[i])
+    else
+      command = _G.arg[i]
+    end
+  end
+  i = i + 1
+end
+
+if not command then
+  print(USAGE)
+elseif M[command] then
+  M[command](opt) -- see M.gen()
+else
+  error('Unknown command: ' .. command)
+end
+
+return M

src/gen/gen_options.lua

@@ -1,5 +1,5 @@
 --- @module 'nvim.options'
-local options = require('options')
+local options = require('nvim.options')
 local options_meta = options.options
 local cstr = options.cstr
 local valid_scopes = options.valid_scopes
@@ -418,7 +418,7 @@ end
 --- @param option_index table<string,string>
 local function gen_map(output_file, option_index)
   -- Generate option index map.
-  local hashy = require('generators.hashy')
+  local hashy = require('gen.hashy')
 
   local neworder, hashfun = hashy.hashy_hash(
     'find_option',

src/gen/gen_vimvim.lua

@@ -15,9 +15,9 @@ local function w(s)
   end
 end
 
-local options = require('options')
-local auevents = require('auevents')
-local ex_cmds = require('ex_cmds')
+local options = require('nvim.options')
+local auevents = require('nvim.auevents')
+local ex_cmds = require('nvim.ex_cmds')
 
 local function cmd_kw(prev_cmd, cmd)
   if not prev_cmd then

src/gen/luacats_grammar.lua

@@ -0,0 +1,207 @@
+--[[!
+LPEG grammar for LuaCATS
+]]
+
+local lpeg = vim.lpeg
+local P, R, S = lpeg.P, lpeg.R, lpeg.S
+local C, Ct, Cg = lpeg.C, lpeg.Ct, lpeg.Cg
+
+--- @param x vim.lpeg.Pattern
+local function rep(x)
+  return x ^ 0
+end
+
+--- @param x vim.lpeg.Pattern
+local function rep1(x)
+  return x ^ 1
+end
+
+--- @param x vim.lpeg.Pattern
+local function opt(x)
+  return x ^ -1
+end
+
+local ws = rep1(S(' \t'))
+local fill = opt(ws)
+local any = P(1) -- (consume one character)
+local letter = R('az', 'AZ')
+local num = R('09')
+
+--- @param x string | vim.lpeg.Pattern
+local function Pf(x)
+  return fill * P(x) * fill
+end
+
+--- @param x string | vim.lpeg.Pattern
+local function Plf(x)
+  return fill * P(x)
+end
+
+--- @param x string
+local function Sf(x)
+  return fill * S(x) * fill
+end
+
+--- @param x vim.lpeg.Pattern
+local function paren(x)
+  return Pf('(') * x * fill * P(')')
+end
+
+--- @param x vim.lpeg.Pattern
+local function parenOpt(x)
+  return paren(x) + x
+end
+
+--- @param x vim.lpeg.Pattern
+local function comma1(x)
+  return parenOpt(x * rep(Pf(',') * x))
+end
+
+--- @param x vim.lpeg.Pattern
+local function comma(x)
+  return opt(comma1(x))
+end
+
+--- @type table<string,vim.lpeg.Pattern>
+local v = setmetatable({}, {
+  __index = function(_, k)
+    return lpeg.V(k)
+  end,
+})
+
+--- @class nvim.luacats.Param
+--- @field kind 'param'
+--- @field name string
+--- @field type string
+--- @field desc? string
+
+--- @class nvim.luacats.Return
+--- @field kind 'return'
+--- @field [integer] { type: string, name?: string}
+--- @field desc? string
+
+--- @class nvim.luacats.Generic
+--- @field kind 'generic'
+--- @field name string
+--- @field type? string
+
+--- @class nvim.luacats.Class
+--- @field kind 'class'
+--- @field name string
+--- @field parent? string
+--- @field access? 'private'|'protected'|'package'
+
+--- @class nvim.luacats.Field
+--- @field kind 'field'
+--- @field name string
+--- @field type string
+--- @field desc? string
+--- @field access? 'private'|'protected'|'package'
+
+--- @class nvim.luacats.Note
+--- @field desc? string
+
+--- @alias nvim.luacats.grammar.result
+--- | nvim.luacats.Param
+--- | nvim.luacats.Return
+--- | nvim.luacats.Generic
+--- | nvim.luacats.Class
+--- | nvim.luacats.Field
+--- | nvim.luacats.Note
+
+--- @class nvim.luacats.grammar
+--- @field match fun(self, input: string): nvim.luacats.grammar.result?
+
+local function annot(nm, pat)
+  if type(nm) == 'string' then
+    nm = P(nm)
+  end
+  if pat then
+    return Ct(Cg(P(nm), 'kind') * fill * pat)
+  end
+  return Ct(Cg(P(nm), 'kind'))
+end
+
+local colon = Pf(':')
+local ellipsis = P('...')
+local ident_first = P('_') + letter
+local ident = ident_first * rep(ident_first + num)
+local opt_ident = ident * opt(P('?'))
+local ty_ident_sep = S('-._')
+local ty_ident = ident * rep(ty_ident_sep * ident)
+local string_single = P "'" * rep(any - P "'") * P "'"
+local string_double = P('"') * rep(any - P('"')) * P('"')
+local generic = P('`') * ty_ident * P('`')
+local literal = string_single + string_double + (opt(P('-')) * rep1(num)) + P('false') + P('true')
+local ty_prims = ty_ident + literal + generic
+
+local array_postfix = rep1(Plf('[]'))
+local opt_postfix = rep1(Plf('?'))
+local rep_array_opt_postfix = rep(array_postfix + opt_postfix)
+
+local typedef = P({
+  'typedef',
+  typedef = C(v.type),
+
+  type = v.ty * rep_array_opt_postfix * rep(Pf('|') * v.ty * rep_array_opt_postfix),
+  ty = v.composite + paren(v.typedef),
+  composite = (v.types * array_postfix) + (v.types * opt_postfix) + v.types,
+  types = v.generics + v.kv_table + v.tuple + v.dict + v.table_literal + v.fun + ty_prims,
+
+  tuple = Pf('[') * comma1(v.type) * Plf(']'),
+  dict = Pf('{') * comma1(Pf('[') * v.type * Pf(']') * colon * v.type) * Plf('}'),
+  kv_table = Pf('table') * Pf('<') * v.type * Pf(',') * v.type * Plf('>'),
+  table_literal = Pf('{') * comma1(opt_ident * Pf(':') * v.type) * Plf('}'),
+  fun_param = (opt_ident + ellipsis) * opt(colon * v.type),
+  fun_ret = v.type + (ellipsis * opt(colon * v.type)),
+  fun = Pf('fun') * paren(comma(v.fun_param)) * opt(Pf(':') * comma1(v.fun_ret)),
+  generics = P(ty_ident) * Pf('<') * comma1(v.type) * Plf('>'),
+}) / function(match)
+  return vim.trim(match):gsub('^%((.*)%)$', '%1'):gsub('%?+', '?')
+end
+
+local access = P('private') + P('protected') + P('package')
+local caccess = Cg(access, 'access')
+local cattr = Cg(comma(access + P('exact')), 'access')
+local desc_delim = Sf '#:' + ws
+local desc = Cg(rep(any), 'desc')
+local opt_desc = opt(desc_delim * desc)
+local ty_name = Cg(ty_ident, 'name')
+local opt_parent = opt(colon * Cg(ty_ident, 'parent'))
+local lname = (ident + ellipsis) * opt(P('?'))
+
+local grammar = P {
+  rep1(P('@') * (v.ats + v.ext_ats)),
+
+  ats = annot('param', Cg(lname, 'name') * ws * v.ctype * opt_desc)
+    + annot('return', comma1(Ct(v.ctype * opt(ws * (ty_name + Cg(ellipsis, 'name'))))) * opt_desc)
+    + annot('type', comma1(Ct(v.ctype)) * opt_desc)
+    + annot('cast', ty_name * ws * opt(Sf('+-')) * v.ctype)
+    + annot('generic', ty_name * opt(colon * v.ctype))
+    + annot('class', opt(paren(cattr)) * fill * ty_name * opt_parent)
+    + annot('field', opt(caccess * ws) * v.field_name * ws * v.ctype * opt_desc)
+    + annot('operator', ty_name * opt(paren(Cg(v.ctype, 'argtype'))) * colon * v.ctype)
+    + annot(access)
+    + annot('deprecated')
+    + annot('alias', ty_name * opt(ws * v.ctype))
+    + annot('enum', ty_name)
+    + annot('overload', v.ctype)
+    + annot('see', opt(desc_delim) * desc)
+    + annot('diagnostic', opt(desc_delim) * desc)
+    + annot('meta'),
+
+  --- Custom extensions
+  ext_ats = (
+    annot('note', desc)
+    + annot('since', desc)
+    + annot('nodoc')
+    + annot('inlinedoc')
+    + annot('brief', desc)
+  ),
+
+  field_name = Cg(lname + (v.ty_index * opt(P('?'))), 'name'),
+  ty_index = C(Pf('[') * typedef * fill * P(']')),
+  ctype = Cg(typedef, 'type'),
+}
+
+return grammar --[[@as nvim.luacats.grammar]]

src/gen/luacats_parser.lua

@@ -0,0 +1,535 @@
+local luacats_grammar = require('gen.luacats_grammar')
+
+--- @class nvim.luacats.parser.param : nvim.luacats.Param
+
+--- @class nvim.luacats.parser.return
+--- @field name string
+--- @field type string
+--- @field desc string
+
+--- @class nvim.luacats.parser.note
+--- @field desc string
+
+--- @class nvim.luacats.parser.brief
+--- @field kind 'brief'
+--- @field desc string
+
+--- @class nvim.luacats.parser.alias
+--- @field kind 'alias'
+--- @field type string[]
+--- @field desc string
+
+--- @class nvim.luacats.parser.fun
+--- @field name string
+--- @field params nvim.luacats.parser.param[]
+--- @field returns nvim.luacats.parser.return[]
+--- @field desc string
+--- @field access? 'private'|'package'|'protected'
+--- @field class? string
+--- @field module? string
+--- @field modvar? string
+--- @field classvar? string
+--- @field deprecated? true
+--- @field since? string
+--- @field attrs? string[]
+--- @field nodoc? true
+--- @field generics? table<string,string>
+--- @field table? true
+--- @field notes? nvim.luacats.parser.note[]
+--- @field see? nvim.luacats.parser.note[]
+
+--- @class nvim.luacats.parser.field : nvim.luacats.Field
+--- @field classvar? string
+--- @field nodoc? true
+
+--- @class nvim.luacats.parser.class : nvim.luacats.Class
+--- @field desc? string
+--- @field nodoc? true
+--- @field inlinedoc? true
+--- @field fields nvim.luacats.parser.field[]
+--- @field notes? string[]
+
+--- @class nvim.luacats.parser.State
+--- @field doc_lines? string[]
+--- @field cur_obj? nvim.luacats.parser.obj
+--- @field last_doc_item? nvim.luacats.parser.param|nvim.luacats.parser.return|nvim.luacats.parser.note
+--- @field last_doc_item_indent? integer
+
+--- @alias nvim.luacats.parser.obj
+--- | nvim.luacats.parser.class
+--- | nvim.luacats.parser.fun
+--- | nvim.luacats.parser.brief
+--- | nvim.luacats.parser.alias
+
+-- Remove this when we document classes properly
+--- Some doc lines have the form:
+---   param name some.complex.type (table) description
+--- if so then transform the line to remove the complex type:
+---   param name (table) description
+--- @param line string
+local function use_type_alt(line)
+  for _, type in ipairs({ 'table', 'function' }) do
+    line = line:gsub('@param%s+([a-zA-Z_?]+)%s+.*%((' .. type .. ')%)', '@param %1 %2')
+    line = line:gsub('@param%s+([a-zA-Z_?]+)%s+.*%((' .. type .. '|nil)%)', '@param %1 %2')
+    line = line:gsub('@param%s+([a-zA-Z_?]+)%s+.*%((' .. type .. '%?)%)', '@param %1 %2')
+
+    line = line:gsub('@return%s+.*%((' .. type .. ')%)', '@return %1')
+    line = line:gsub('@return%s+.*%((' .. type .. '|nil)%)', '@return %1')
+    line = line:gsub('@return%s+.*%((' .. type .. '%?)%)', '@return %1')
+  end
+  return line
+end
+
+--- If we collected any `---` lines. Add them to the existing (or new) object
+--- Used for function/class descriptions and multiline param descriptions.
+--- @param state nvim.luacats.parser.State
+local function add_doc_lines_to_obj(state)
+  if state.doc_lines then
+    state.cur_obj = state.cur_obj or {}
+    local cur_obj = assert(state.cur_obj)
+    local txt = table.concat(state.doc_lines, '\n')
+    if cur_obj.desc then
+      cur_obj.desc = cur_obj.desc .. '\n' .. txt
+    else
+      cur_obj.desc = txt
+    end
+    state.doc_lines = nil
+  end
+end
+
+--- @param line string
+--- @param state nvim.luacats.parser.State
+local function process_doc_line(line, state)
+  line = line:sub(4):gsub('^%s+@', '@')
+  line = use_type_alt(line)
+
+  local parsed = luacats_grammar:match(line)
+
+  if not parsed then
+    if line:match('^ ') then
+      line = line:sub(2)
+    end
+
+    if state.last_doc_item then
+      if not state.last_doc_item_indent then
+        state.last_doc_item_indent = #line:match('^%s*') + 1
+      end
+      state.last_doc_item.desc = (state.last_doc_item.desc or '')
+        .. '\n'
+        .. line:sub(state.last_doc_item_indent or 1)
+    else
+      state.doc_lines = state.doc_lines or {}
+      table.insert(state.doc_lines, line)
+    end
+    return
+  end
+
+  state.last_doc_item_indent = nil
+  state.last_doc_item = nil
+  state.cur_obj = state.cur_obj or {}
+  local cur_obj = assert(state.cur_obj)
+
+  local kind = parsed.kind
+
+  if kind == 'brief' then
+    state.cur_obj = {
+      kind = 'brief',
+      desc = parsed.desc,
+    }
+  elseif kind == 'class' then
+    --- @cast parsed nvim.luacats.Class
+    cur_obj.kind = 'class'
+    cur_obj.name = parsed.name
+    cur_obj.parent = parsed.parent
+    cur_obj.access = parsed.access
+    cur_obj.desc = state.doc_lines and table.concat(state.doc_lines, '\n') or nil
+    state.doc_lines = nil
+    cur_obj.fields = {}
+  elseif kind == 'field' then
+    --- @cast parsed nvim.luacats.Field
+    parsed.desc = parsed.desc or state.doc_lines and table.concat(state.doc_lines, '\n') or nil
+    if parsed.desc then
+      parsed.desc = vim.trim(parsed.desc)
+    end
+    table.insert(cur_obj.fields, parsed)
+    state.doc_lines = nil
+  elseif kind == 'operator' then
+    parsed.desc = parsed.desc or state.doc_lines and table.concat(state.doc_lines, '\n') or nil
+    if parsed.desc then
+      parsed.desc = vim.trim(parsed.desc)
+    end
+    table.insert(cur_obj.fields, parsed)
+    state.doc_lines = nil
+  elseif kind == 'param' then
+    state.last_doc_item_indent = nil
+    cur_obj.params = cur_obj.params or {}
+    if vim.endswith(parsed.name, '?') then
+      parsed.name = parsed.name:sub(1, -2)
+      parsed.type = parsed.type .. '?'
+    end
+    state.last_doc_item = {
+      name = parsed.name,
+      type = parsed.type,
+      desc = parsed.desc,
+    }
+    table.insert(cur_obj.params, state.last_doc_item)
+  elseif kind == 'return' then
+    cur_obj.returns = cur_obj.returns or {}
+    for _, t in ipairs(parsed) do
+      table.insert(cur_obj.returns, {
+        name = t.name,
+        type = t.type,
+        desc = parsed.desc,
+      })
+    end
+    state.last_doc_item_indent = nil
+    state.last_doc_item = cur_obj.returns[#cur_obj.returns]
+  elseif kind == 'private' then
+    cur_obj.access = 'private'
+  elseif kind == 'package' then
+    cur_obj.access = 'package'
+  elseif kind == 'protected' then
+    cur_obj.access = 'protected'
+  elseif kind == 'deprecated' then
+    cur_obj.deprecated = true
+  elseif kind == 'inlinedoc' then
+    cur_obj.inlinedoc = true
+  elseif kind == 'nodoc' then
+    cur_obj.nodoc = true
+  elseif kind == 'since' then
+    cur_obj.since = parsed.desc
+  elseif kind == 'see' then
+    cur_obj.see = cur_obj.see or {}
+    table.insert(cur_obj.see, { desc = parsed.desc })
+  elseif kind == 'note' then
+    state.last_doc_item_indent = nil
+    state.last_doc_item = {
+      desc = parsed.desc,
+    }
+    cur_obj.notes = cur_obj.notes or {}
+    table.insert(cur_obj.notes, state.last_doc_item)
+  elseif kind == 'type' then
+    cur_obj.desc = parsed.desc
+    parsed.desc = nil
+    parsed.kind = nil
+    cur_obj.type = parsed
+  elseif kind == 'alias' then
+    state.cur_obj = {
+      kind = 'alias',
+      desc = parsed.desc,
+    }
+  elseif kind == 'enum' then
+    -- TODO
+    state.doc_lines = nil
+  elseif
+    vim.tbl_contains({
+      'diagnostic',
+      'cast',
+      'overload',
+      'meta',
+    }, kind)
+  then
+    -- Ignore
+    return
+  elseif kind == 'generic' then
+    cur_obj.generics = cur_obj.generics or {}
+    cur_obj.generics[parsed.name] = parsed.type or 'any'
+  else
+    error('Unhandled' .. vim.inspect(parsed))
+  end
+end
+
+--- @param fun nvim.luacats.parser.fun
+--- @return nvim.luacats.parser.field
+local function fun2field(fun)
+  local parts = { 'fun(' }
+
+  local params = {} ---@type string[]
+  for _, p in ipairs(fun.params or {}) do
+    params[#params + 1] = string.format('%s: %s', p.name, p.type)
+  end
+  parts[#parts + 1] = table.concat(params, ', ')
+  parts[#parts + 1] = ')'
+  if fun.returns then
+    parts[#parts + 1] = ': '
+    local tys = {} --- @type string[]
+    for _, p in ipairs(fun.returns) do
+      tys[#tys + 1] = p.type
+    end
+    parts[#parts + 1] = table.concat(tys, ', ')
+  end
+
+  return {
+    name = fun.name,
+    type = table.concat(parts, ''),
+    access = fun.access,
+    desc = fun.desc,
+    nodoc = fun.nodoc,
+  }
+end
+
+--- Function to normalize known form for declaring functions and normalize into a more standard
+--- form.
+--- @param line string
+--- @return string
+local function filter_decl(line)
+  -- M.fun = vim._memoize(function(...)
+  --   ->
+  -- function M.fun(...)
+  line = line:gsub('^local (.+) = memoize%([^,]+, function%((.*)%)$', 'local function %1(%2)')
+  line = line:gsub('^(.+) = memoize%([^,]+, function%((.*)%)$', 'function %1(%2)')
+  return line
+end
+
+--- @param line string
+--- @param state nvim.luacats.parser.State
+--- @param classes table<string,nvim.luacats.parser.class>
+--- @param classvars table<string,string>
+--- @param has_indent boolean
+local function process_lua_line(line, state, classes, classvars, has_indent)
+  line = filter_decl(line)
+
+  if state.cur_obj and state.cur_obj.kind == 'class' then
+    local nm = line:match('^local%s+([a-zA-Z0-9_]+)%s*=')
+    if nm then
+      classvars[nm] = state.cur_obj.name
+    end
+    return
+  end
+
+  do
+    local parent_tbl, sep, fun_or_meth_nm =
+      line:match('^function%s+([a-zA-Z0-9_]+)([.:])([a-zA-Z0-9_]+)%s*%(')
+    if parent_tbl then
+      -- Have a decl. Ensure cur_obj
+      state.cur_obj = state.cur_obj or {}
+      local cur_obj = assert(state.cur_obj)
+
+      -- Match `Class:foo` methods for defined classes
+      local class = classvars[parent_tbl]
+      if class then
+        --- @cast cur_obj nvim.luacats.parser.fun
+        cur_obj.name = fun_or_meth_nm
+        cur_obj.class = class
+        cur_obj.classvar = parent_tbl
+        -- Add self param to methods
+        if sep == ':' then
+          cur_obj.params = cur_obj.params or {}
+          table.insert(cur_obj.params, 1, {
+            name = 'self',
+            type = class,
+          })
+        end
+
+        -- Add method as the field to the class
+        local cls = classes[class]
+        local field = fun2field(cur_obj)
+        field.classvar = cur_obj.classvar
+        table.insert(cls.fields, field)
+        return
+      end
+
+      -- Match `M.foo`
+      if cur_obj and parent_tbl == cur_obj.modvar then
+        cur_obj.name = fun_or_meth_nm
+        return
+      end
+    end
+  end
+
+  do
+    -- Handle: `function A.B.C.foo(...)`
+    local fn_nm = line:match('^function%s+([.a-zA-Z0-9_]+)%s*%(')
+    if fn_nm then
+      state.cur_obj = state.cur_obj or {}
+      state.cur_obj.name = fn_nm
+      return
+    end
+  end
+
+  do
+    -- Handle: `M.foo = {...}` where `M` is the modvar
+    local parent_tbl, tbl_nm = line:match('([a-zA-Z_]+)%.([a-zA-Z0-9_]+)%s*=')
+    if state.cur_obj and parent_tbl and parent_tbl == state.cur_obj.modvar then
+      state.cur_obj.name = tbl_nm
+      state.cur_obj.table = true
+      return
+    end
+  end
+
+  do
+    -- Handle: `foo = {...}`
+    local tbl_nm = line:match('^([a-zA-Z0-9_]+)%s*=')
+    if tbl_nm and not has_indent then
+      state.cur_obj = state.cur_obj or {}
+      state.cur_obj.name = tbl_nm
+      state.cur_obj.table = true
+      return
+    end
+  end
+
+  do
+    -- Handle: `vim.foo = {...}`
+    local tbl_nm = line:match('^(vim%.[a-zA-Z0-9_]+)%s*=')
+    if state.cur_obj and tbl_nm and not has_indent then
+      state.cur_obj.name = tbl_nm
+      state.cur_obj.table = true
+      return
+    end
+  end
+
+  if state.cur_obj then
+    if line:find('^%s*%-%- luacheck:') then
+      state.cur_obj = nil
+    elseif line:find('^%s*local%s+') then
+      state.cur_obj = nil
+    elseif line:find('^%s*return%s+') then
+      state.cur_obj = nil
+    elseif line:find('^%s*[a-zA-Z_.]+%(%s+') then
+      state.cur_obj = nil
+    end
+  end
+end
+
+--- Determine the table name used to export functions of a module
+--- Usually this is `M`.
+--- @param str string
+--- @return string?
+local function determine_modvar(str)
+  local modvar --- @type string?
+  for line in vim.gsplit(str, '\n') do
+    do
+      --- @type string?
+      local m = line:match('^return%s+([a-zA-Z_]+)')
+      if m then
+        modvar = m
+      end
+    end
+    do
+      --- @type string?
+      local m = line:match('^return%s+setmetatable%(([a-zA-Z_]+),')
+      if m then
+        modvar = m
+      end
+    end
+  end
+  return modvar
+end
+
+--- @param obj nvim.luacats.parser.obj
+--- @param funs nvim.luacats.parser.fun[]
+--- @param classes table<string,nvim.luacats.parser.class>
+--- @param briefs string[]
+--- @param uncommitted nvim.luacats.parser.obj[]
+local function commit_obj(obj, classes, funs, briefs, uncommitted)
+  local commit = false
+  if obj.kind == 'class' then
+    --- @cast obj nvim.luacats.parser.class
+    if not classes[obj.name] then
+      classes[obj.name] = obj
+      commit = true
+    end
+  elseif obj.kind == 'alias' then
+    -- Just pretend
+    commit = true
+  elseif obj.kind == 'brief' then
+    --- @cast obj nvim.luacats.parser.brief`
+    briefs[#briefs + 1] = obj.desc
+    commit = true
+  else
+    --- @cast obj nvim.luacats.parser.fun`
+    if obj.name then
+      funs[#funs + 1] = obj
+      commit = true
+    end
+  end
+  if not commit then
+    table.insert(uncommitted, obj)
+  end
+  return commit
+end
+
+--- @param filename string
+--- @param uncommitted nvim.luacats.parser.obj[]
+-- luacheck: no unused
+local function dump_uncommitted(filename, uncommitted)
+  local out_path = 'luacats-uncommited/' .. filename:gsub('/', '%%') .. '.txt'
+  if #uncommitted > 0 then
+    print(string.format('Could not commit %d objects in %s', #uncommitted, filename))
+    vim.fn.mkdir(vim.fs.dirname(out_path), 'p')
+    local f = assert(io.open(out_path, 'w'))
+    for i, x in ipairs(uncommitted) do
+      f:write(i)
+      f:write(': ')
+      f:write(vim.inspect(x))
+      f:write('\n')
+    end
+    f:close()
+  else
+    vim.fn.delete(out_path)
+  end
+end
+
+local M = {}
+
+function M.parse_str(str, filename)
+  local funs = {} --- @type nvim.luacats.parser.fun[]
+  local classes = {} --- @type table<string,nvim.luacats.parser.class>
+  local briefs = {} --- @type string[]
+
+  local mod_return = determine_modvar(str)
+
+  --- @type string
+  local module = filename:match('.*/lua/([a-z_][a-z0-9_/]+)%.lua') or filename
+  module = module:gsub('/', '.')
+
+  local classvars = {} --- @type table<string,string>
+
+  local state = {} --- @type nvim.luacats.parser.State
+
+  -- Keep track of any partial objects we don't commit
+  local uncommitted = {} --- @type nvim.luacats.parser.obj[]
+
+  for line in vim.gsplit(str, '\n') do
+    local has_indent = line:match('^%s+') ~= nil
+    line = vim.trim(line)
+    if vim.startswith(line, '---') then
+      process_doc_line(line, state)
+    else
+      add_doc_lines_to_obj(state)
+
+      if state.cur_obj then
+        state.cur_obj.modvar = mod_return
+        state.cur_obj.module = module
+      end
+
+      process_lua_line(line, state, classes, classvars, has_indent)
+
+      -- Commit the object
+      local cur_obj = state.cur_obj
+      if cur_obj then
+        if not commit_obj(cur_obj, classes, funs, briefs, uncommitted) then
+          --- @diagnostic disable-next-line:inject-field
+          cur_obj.line = line
+        end
+      end
+
+      state = {}
+    end
+  end
+
+  -- dump_uncommitted(filename, uncommitted)
+
+  return classes, funs, briefs, uncommitted
+end
+
+--- @param filename string
+function M.parse(filename)
+  local f = assert(io.open(filename, 'r'))
+  local txt = f:read('*all')
+  f:close()
+
+  return M.parse_str(txt, filename)
+end
+
+return M

src/gen/preload.lua

@@ -0,0 +1,6 @@
+local srcdir = table.remove(arg, 1)
+
+package.path = (srcdir .. '/src/?.lua;') .. (srcdir .. '/runtime/lua/?.lua;') .. package.path
+
+arg[0] = table.remove(arg, 1)
+return loadfile(arg[0])()

src/gen/preload_nlua.lua

@@ -1,8 +1,12 @@
 local srcdir = table.remove(arg, 1)
 local nlualib = table.remove(arg, 1)
 local gendir = table.remove(arg, 1)
-package.path = srcdir .. '/src/nvim/?.lua;' .. srcdir .. '/runtime/lua/?.lua;' .. package.path
-package.path = gendir .. '/?.lua;' .. package.path
+
+package.path = (srcdir .. '/src/?.lua;')
+  .. (srcdir .. '/runtime/lua/?.lua;')
+  .. (gendir .. '/?.lua;')
+  .. package.path
+
 _G.vim = require 'vim.shared'
 _G.vim.inspect = require 'vim.inspect'
 package.cpath = package.cpath .. ';' .. nlualib

src/gen/util.lua

@@ -0,0 +1,399 @@
+-- TODO(justinmk): move most of this to `vim.text`.
+
+local fmt = string.format
+
+--- @class nvim.util.MDNode
+--- @field [integer] nvim.util.MDNode
+--- @field type string
+--- @field text? string
+
+local INDENTATION = 4
+
+local NBSP = string.char(160)
+
+local M = {}
+
+local function contains(t, xs)
+  return vim.tbl_contains(xs, t)
+end
+
+-- Map of api_level:version, by inspection of:
+--    :lua= vim.mpack.decode(vim.fn.readfile('test/functional/fixtures/api_level_9.mpack','B')).version
+M.version_level = {
+  [13] = '0.11.0',
+  [12] = '0.10.0',
+  [11] = '0.9.0',
+  [10] = '0.8.0',
+  [9] = '0.7.0',
+  [8] = '0.6.0',
+  [7] = '0.5.0',
+  [6] = '0.4.0',
+  [5] = '0.3.2',
+  [4] = '0.3.0',
+  [3] = '0.2.1',
+  [2] = '0.2.0',
+  [1] = '0.1.0',
+}
+
+--- @param txt string
+--- @param srow integer
+--- @param scol integer
+--- @param erow? integer
+--- @param ecol? integer
+--- @return string
+local function slice_text(txt, srow, scol, erow, ecol)
+  local lines = vim.split(txt, '\n')
+
+  if srow == erow then
+    return lines[srow + 1]:sub(scol + 1, ecol)
+  end
+
+  if erow then
+    -- Trim the end
+    for _ = erow + 2, #lines do
+      table.remove(lines, #lines)
+    end
+  end
+
+  -- Trim the start
+  for _ = 1, srow do
+    table.remove(lines, 1)
+  end
+
+  lines[1] = lines[1]:sub(scol + 1)
+  lines[#lines] = lines[#lines]:sub(1, ecol)
+
+  return table.concat(lines, '\n')
+end
+
+--- @param text string
+--- @return nvim.util.MDNode
+local function parse_md_inline(text)
+  local parser = vim.treesitter.languagetree.new(text, 'markdown_inline')
+  local root = parser:parse(true)[1]:root()
+
+  --- @param node TSNode
+  --- @return nvim.util.MDNode?
+  local function extract(node)
+    local ntype = node:type()
+
+    if ntype:match('^%p$') then
+      return
+    end
+
+    --- @type table<any,any>
+    local ret = { type = ntype }
+    ret.text = vim.treesitter.get_node_text(node, text)
+
+    local row, col = 0, 0
+
+    for child, child_field in node:iter_children() do
+      local e = extract(child)
+      if e and ntype == 'inline' then
+        local srow, scol = child:start()
+        if (srow == row and scol > col) or srow > row then
+          local t = slice_text(ret.text, row, col, srow, scol)
+          if t and t ~= '' then
+            table.insert(ret, { type = 'text', j = true, text = t })
+          end
+        end
+        row, col = child:end_()
+      end
+
+      if child_field then
+        ret[child_field] = e
+      else
+        table.insert(ret, e)
+      end
+    end
+
+    if ntype == 'inline' and (row > 0 or col > 0) then
+      local t = slice_text(ret.text, row, col)
+      if t and t ~= '' then
+        table.insert(ret, { type = 'text', text = t })
+      end
+    end
+
+    return ret
+  end
+
+  return extract(root) or {}
+end
+
+--- @param text string
+--- @return nvim.util.MDNode
+local function parse_md(text)
+  local parser = vim.treesitter.languagetree.new(text, 'markdown', {
+    injections = { markdown = '' },
+  })
+
+  local root = parser:parse(true)[1]:root()
+
+  local EXCLUDE_TEXT_TYPE = {
+    list = true,
+    list_item = true,
+    section = true,
+    document = true,
+    fenced_code_block = true,
+    fenced_code_block_delimiter = true,
+  }
+
+  --- @param node TSNode
+  --- @return nvim.util.MDNode?
+  local function extract(node)
+    local ntype = node:type()
+
+    if ntype:match('^%p$') or contains(ntype, { 'block_continuation' }) then
+      return
+    end
+
+    --- @type table<any,any>
+    local ret = { type = ntype }
+
+    if not EXCLUDE_TEXT_TYPE[ntype] then
+      ret.text = vim.treesitter.get_node_text(node, text)
+    end
+
+    if ntype == 'inline' then
+      ret = parse_md_inline(ret.text)
+    end
+
+    for child, child_field in node:iter_children() do
+      local e = extract(child)
+      if child_field then
+        ret[child_field] = e
+      else
+        table.insert(ret, e)
+      end
+    end
+
+    return ret
+  end
+
+  return extract(root) or {}
+end
+
+--- Prefixes each line in `text`.
+---
+--- Does not wrap, not important for "meta" files? (You probably want md_to_vimdoc instead.)
+---
+--- @param text string
+--- @param prefix_ string
+function M.prefix_lines(prefix_, text)
+  local r = ''
+  for _, l in ipairs(vim.split(text, '\n', { plain = true })) do
+    r = r .. vim.trim(prefix_ .. l) .. '\n'
+  end
+  return r
+end
+
+--- @param x string
+--- @param start_indent integer
+--- @param indent integer
+--- @param text_width integer
+--- @return string
+function M.wrap(x, start_indent, indent, text_width)
+  local words = vim.split(vim.trim(x), '%s+')
+  local parts = { string.rep(' ', start_indent) } --- @type string[]
+  local count = indent
+
+  for i, w in ipairs(words) do
+    if count > indent and count + #w > text_width - 1 then
+      parts[#parts + 1] = '\n'
+      parts[#parts + 1] = string.rep(' ', indent)
+      count = indent
+    elseif i ~= 1 then
+      parts[#parts + 1] = ' '
+      count = count + 1
+    end
+    count = count + #w
+    parts[#parts + 1] = w
+  end
+
+  return (table.concat(parts):gsub('%s+\n', '\n'):gsub('\n+$', ''))
+end
+
+--- @param node nvim.util.MDNode
+--- @param start_indent integer
+--- @param indent integer
+--- @param text_width integer
+--- @param level integer
+--- @return string[]
+local function render_md(node, start_indent, indent, text_width, level, is_list)
+  local parts = {} --- @type string[]
+
+  -- For debugging
+  local add_tag = false
+  -- local add_tag = true
+
+  local ntype = node.type
+
+  if add_tag then
+    parts[#parts + 1] = '<' .. ntype .. '>'
+  end
+
+  if ntype == 'text' then
+    parts[#parts + 1] = node.text
+  elseif ntype == 'html_tag' then
+    error('html_tag: ' .. node.text)
+  elseif ntype == 'inline_link' then
+    vim.list_extend(parts, { '*', node[1].text, '*' })
+  elseif ntype == 'shortcut_link' then
+    if node[1].text:find('^<.*>$') then
+      parts[#parts + 1] = node[1].text
+    elseif node[1].text:find('^%d+$') then
+      vim.list_extend(parts, { '[', node[1].text, ']' })
+    else
+      vim.list_extend(parts, { '|', node[1].text, '|' })
+    end
+  elseif ntype == 'backslash_escape' then
+    parts[#parts + 1] = node.text
+  elseif ntype == 'emphasis' then
+    parts[#parts + 1] = node.text:sub(2, -2)
+  elseif ntype == 'code_span' then
+    vim.list_extend(parts, { '`', node.text:sub(2, -2):gsub(' ', NBSP), '`' })
+  elseif ntype == 'inline' then
+    if #node == 0 then
+      local text = assert(node.text)
+      parts[#parts + 1] = M.wrap(text, start_indent, indent, text_width)
+    else
+      for _, child in ipairs(node) do
+        vim.list_extend(parts, render_md(child, start_indent, indent, text_width, level + 1))
+      end
+    end
+  elseif ntype == 'paragraph' then
+    local pparts = {}
+    for _, child in ipairs(node) do
+      vim.list_extend(pparts, render_md(child, start_indent, indent, text_width, level + 1))
+    end
+    parts[#parts + 1] = M.wrap(table.concat(pparts), start_indent, indent, text_width)
+    parts[#parts + 1] = '\n'
+  elseif ntype == 'code_fence_content' then
+    local lines = vim.split(node.text:gsub('\n%s*$', ''), '\n')
+
+    local cindent = indent + INDENTATION
+    if level > 3 then
+      -- The tree-sitter markdown parser doesn't parse the code blocks indents
+      -- correctly in lists. Fudge it!
+      lines[1] = '    ' .. lines[1] -- ¯\_(ツ)_/¯
+      cindent = indent - level
+      local _, initial_indent = lines[1]:find('^%s*')
+      initial_indent = initial_indent + cindent
+      if initial_indent < indent then
+        cindent = indent - INDENTATION
+      end
+    end
+
+    for _, l in ipairs(lines) do
+      if #l > 0 then
+        parts[#parts + 1] = string.rep(' ', cindent)
+        parts[#parts + 1] = l
+      end
+      parts[#parts + 1] = '\n'
+    end
+  elseif ntype == 'fenced_code_block' then
+    parts[#parts + 1] = '>'
+    for _, child in ipairs(node) do
+      if child.type == 'info_string' then
+        parts[#parts + 1] = child.text
+        break
+      end
+    end
+    parts[#parts + 1] = '\n'
+    for _, child in ipairs(node) do
+      if child.type ~= 'info_string' then
+        vim.list_extend(parts, render_md(child, start_indent, indent, text_width, level + 1))
+      end
+    end
+    parts[#parts + 1] = '<\n'
+  elseif ntype == 'html_block' then
+    local text = node.text:gsub('^<pre>help', '')
+    text = text:gsub('</pre>%s*$', '')
+    parts[#parts + 1] = text
+  elseif ntype == 'list_marker_dot' then
+    parts[#parts + 1] = node.text
+  elseif contains(ntype, { 'list_marker_minus', 'list_marker_star' }) then
+    parts[#parts + 1] = '• '
+  elseif ntype == 'list_item' then
+    parts[#parts + 1] = string.rep(' ', indent)
+    local offset = node[1].type == 'list_marker_dot' and 3 or 2
+    for i, child in ipairs(node) do
+      local sindent = i <= 2 and 0 or (indent + offset)
+      vim.list_extend(
+        parts,
+        render_md(child, sindent, indent + offset, text_width, level + 1, true)
+      )
+    end
+  else
+    if node.text then
+      error(fmt('cannot render:\n%s', vim.inspect(node)))
+    end
+    for i, child in ipairs(node) do
+      local start_indent0 = i == 1 and start_indent or indent
+      vim.list_extend(
+        parts,
+        render_md(child, start_indent0, indent, text_width, level + 1, is_list)
+      )
+      if ntype ~= 'list' and i ~= #node then
+        if (node[i + 1] or {}).type ~= 'list' then
+          parts[#parts + 1] = '\n'
+        end
+      end
+    end
+  end
+
+  if add_tag then
+    parts[#parts + 1] = '</' .. ntype .. '>'
+  end
+
+  return parts
+end
+
+--- @param text_width integer
+local function align_tags(text_width)
+  --- @param line string
+  --- @return string
+  return function(line)
+    local tag_pat = '%s*(%*.+%*)%s*$'
+    local tags = {}
+    for m in line:gmatch(tag_pat) do
+      table.insert(tags, m)
+    end
+
+    if #tags > 0 then
+      line = line:gsub(tag_pat, '')
+      local tags_str = ' ' .. table.concat(tags, ' ')
+      --- @type integer
+      local conceal_offset = select(2, tags_str:gsub('%*', '')) - 2
+      local pad = string.rep(' ', text_width - #line - #tags_str + conceal_offset)
+      return line .. pad .. tags_str
+    end
+
+    return line
+  end
+end
+
+--- @param text string
+--- @param start_indent integer
+--- @param indent integer
+--- @param is_list? boolean
+--- @return string
+function M.md_to_vimdoc(text, start_indent, indent, text_width, is_list)
+  -- Add an extra newline so the parser can properly capture ending ```
+  local parsed = parse_md(text .. '\n')
+  local ret = render_md(parsed, start_indent, indent, text_width, 0, is_list)
+
+  local lines = vim.split(table.concat(ret):gsub(NBSP, ' '), '\n')
+
+  lines = vim.tbl_map(align_tags(text_width), lines)
+
+  local s = table.concat(lines, '\n')
+
+  -- Reduce whitespace in code-blocks
+  s = s:gsub('\n+%s*>([a-z]+)\n', ' >%1\n')
+  s = s:gsub('\n+%s*>\n?\n', ' >\n')
+
+  return s
+end
+
+return M

src/nvim/CMakeLists.txt

@@ -292,7 +292,7 @@ set(UI_METADATA ${PROJECT_BINARY_DIR}/ui_metadata.mpack)
 set(BINARY_LIB_DIR ${PROJECT_BINARY_DIR}/lib/nvim)
 set(GENERATED_DIR ${PROJECT_BINARY_DIR}/src/nvim/auto)
 set(GENERATED_INCLUDES_DIR ${PROJECT_BINARY_DIR}/include)
-set(GENERATOR_DIR ${CMAKE_CURRENT_LIST_DIR}/generators)
+set(GENERATOR_DIR ${PROJECT_SOURCE_DIR}/src/gen)
 set(GEN_EVAL_TOUCH ${TOUCHES_DIR}/gen_doc_eval)
 set(LUAJIT_RUNTIME_DIR ${DEPS_PREFIX}/share/luajit-2.1/jit)
 set(NVIM_RUNTIME_DIR ${PROJECT_SOURCE_DIR}/runtime)
@@ -306,7 +306,8 @@ set(EX_CMDS_GENERATOR ${GENERATOR_DIR}/gen_ex_cmds.lua)
 set(FUNCS_GENERATOR ${GENERATOR_DIR}/gen_eval.lua)
 set(GENERATOR_C_GRAMMAR ${GENERATOR_DIR}/c_grammar.lua)
 set(GENERATOR_HASHY ${GENERATOR_DIR}/hashy.lua)
-set(GENERATOR_PRELOAD ${GENERATOR_DIR}/preload.lua)
+set(GENERATOR_PRELOAD ${GENERATOR_DIR}/preload_nlua.lua)
+set(NVIM_LUA_PRELOAD ${GENERATOR_DIR}/preload.lua)
 set(HEADER_GENERATOR ${GENERATOR_DIR}/gen_declarations.lua)
 set(OPTIONS_GENERATOR ${GENERATOR_DIR}/gen_options.lua)
 
@@ -514,6 +515,9 @@ add_custom_command(
 set(LUA_GEN ${LUA_GEN_PRG} ${GENERATOR_PRELOAD} ${PROJECT_SOURCE_DIR} $<TARGET_FILE:nlua0> ${PROJECT_BINARY_DIR})
 set(LUA_GEN_DEPS ${GENERATOR_PRELOAD} $<TARGET_FILE:nlua0>)
 
+# Like LUA_GEN but includes also vim.fn, vim.api, vim.uv, etc
+set(NVIM_LUA $<TARGET_FILE:nvim_bin> -u NONE -l ${NVIM_LUA_PRELOAD} ${PROJECT_SOURCE_DIR})
+
 # NVIM_GENERATED_FOR_HEADERS: generated headers to be included in headers
 # NVIM_GENERATED_FOR_SOURCES: generated headers to be included in sources
 # These lists must be mutually exclusive.
@@ -937,12 +941,12 @@ file(GLOB LUA_SOURCES CONFIGURE_DEPENDS
 )
 
 add_target(doc-vim
-  COMMAND $<TARGET_FILE:nvim_bin> -u NONE -l scripts/gen_vimdoc.lua
+  COMMAND ${NVIM_LUA} src/gen/gen_vimdoc.lua
   DEPENDS
     nvim
     ${API_SOURCES}
     ${LUA_SOURCES}
-    ${PROJECT_SOURCE_DIR}/scripts/gen_vimdoc.lua
+    ${PROJECT_SOURCE_DIR}/src/gen/gen_vimdoc.lua
     ${NVIM_RUNTIME_DIR}/doc/api.txt
     ${NVIM_RUNTIME_DIR}/doc/diagnostic.txt
     ${NVIM_RUNTIME_DIR}/doc/lsp.txt
@@ -951,11 +955,11 @@ add_target(doc-vim
   )
 
 add_target(doc-eval
-  COMMAND $<TARGET_FILE:nvim_bin> -u NONE -l ${PROJECT_SOURCE_DIR}/scripts/gen_eval_files.lua
+  COMMAND ${NVIM_LUA} ${PROJECT_SOURCE_DIR}/src/gen/gen_eval_files.lua
   DEPENDS
     nvim
     ${FUNCS_METADATA}
-    ${PROJECT_SOURCE_DIR}/scripts/gen_eval_files.lua
+    ${PROJECT_SOURCE_DIR}/src/gen/gen_eval_files.lua
     ${PROJECT_SOURCE_DIR}/src/nvim/eval.lua
     ${PROJECT_SOURCE_DIR}/src/nvim/options.lua
     ${PROJECT_SOURCE_DIR}/src/nvim/vvars.lua
@@ -966,7 +970,7 @@ add_custom_target(doc)
 add_dependencies(doc doc-vim doc-eval)
 
 add_target(lintdoc
-  COMMAND $<TARGET_FILE:nvim_bin> -u NONE -l scripts/lintdoc.lua
+  COMMAND ${NVIM_LUA} scripts/lintdoc.lua
   DEPENDS ${DOCFILES}
   CUSTOM_COMMAND_ARGS USES_TERMINAL)
 add_dependencies(lintdoc nvim)

src/nvim/func_attr.h

@@ -1,6 +1,6 @@
 // Undefined DEFINE_FUNC_ATTRIBUTES and undefined DEFINE_EMPTY_ATTRIBUTES
 // leaves file with untouched FUNC_ATTR_* macros. This variant is used for
-// scripts/gen_declarations.lua.
+// src/gen/gen_declarations.lua.
 //
 // Empty macros are used for *.c files.
 // (undefined DEFINE_FUNC_ATTRIBUTES and defined DEFINE_EMPTY_ATTRIBUTES)