generate missing filetypes for neovim

filetype_gen_missing.lua

local fmt = string.format
local function write_file(path, txt, flag)
  local data = type(txt) == "string" and txt or vim.inspect(txt)
  local uv = vim.loop
  uv.fs_open(path, flag, 438, function(open_err, fd)
    assert(not open_err, open_err)
    uv.fs_write(fd, data, -1, function(write_err)
      assert(not write_err, write_err)
      uv.fs_close(fd, function(close_err)
        assert(not close_err, close_err)
      end)
    end)
  end)
end

local function read_file(path)
  local uv = vim.loop
  local fd = assert(uv.fs_open(path, "r", 438))
  local fstat = assert(uv.fs_fstat(fd))
  local contents = assert(uv.fs_read(fd, fstat.size, 0))
  assert(uv.fs_close(fd))
  return contents
end

local filetype_lua_url = "https://raw.githubusercontent.com/neovim/neovim/9e1ee9fb1d747facb6fa97a32dc5e22c7dfcb346/runtime/lua/vim/filetype.lua"
local languages_yaml_url = "https://raw.githubusercontent.com/github/linguist/master/lib/linguist/languages.yml"
local language_stats_url = "https://raw.githubusercontent.com/oprogramador/github-languages/master/1000-pushed-after-2020-06-07.json"

local awk_script = [['NR==33, NR==1010; NR==1011 {print; exit}']]

vim.fn.system(fmt("curl -LSs '%s' | yq > %s", languages_yaml_url, "languages.json"))
vim.fn.system(fmt("curl -LSs '%s' | awk '{print tolower($0)}' > %s", language_stats_url, "stats.json"))
vim.fn.system(fmt("curl -LSs '%s' | awk %s | xargs -0 echo 'return {' > %s", filetype_lua_url, awk_script, "lua/scratch/defaults.lua"))

local lang_maps = vim.json.decode(read_file("languages.json"))
local stats = vim.json.decode(read_file("stats.json"))

local collection = {}
for lang, v in pairs(lang_maps) do
  if v.extensions and type(v.extensions) == "table" then
    for _, entry in ipairs(v.extensions) do
      collection[string.gsub(entry, "%.", "")] = lang:lower()
    end
  end
end
table.sort(collection)

write_file("lua/scratch/imported.lua", "return " .. vim.inspect(collection), "w")
write_file("lua/scratch/stats.lua", "return " .. vim.inspect(stats), "w")

local function is_top_language(name)
  -- check that the language exists and has more than 5 stars
  return stats[name] and stats[name] > 5
end

local defaults = require("scratch.defaults")
local missing = {}
for ext, lang in pairs(collection) do
  if is_top_language(lang) and not defaults[ext] then
    missing[ext] = lang
  end
end

table.sort(missing)

write_file("lua/scratch/missing.lua", "return " .. vim.inspect(missing), "w")

missing.lua

return {
  _coffee = "coffeescript",
  _js = "javascript",
  a51 = "assembly",
  appsrc = "erlang",
  aux = "tex",
  aw = "php",
  bats = "shell",
  bbx = "tex",
  bones = "javascript",
  boot = "clojure",
  cake = "c#",
  cats = "c",
  cbx = "tex",
  cgi = "perl",
  cjsx = "coffeescript",
  cl2 = "clojure",
  cljscm = "clojure",
  cljshl = "clojure",
  cmakein = "cmake",
  cmd = "batchfile",
  coffee = "coffeescript",
  command = "shell",
  cp = "c++",
  cr = "crystal",
  dfm = "pascal",
  di = "d",
  dockerfile = "dockerfile",
  eliom = "ocaml",
  eliomi = "ocaml",
  emacs = "emacs lisp",
  emacsdesktop = "emacs lisp",
  es6 = "javascript",
  escript = "erlang",
  eye = "ruby",
  fcgi = "lua",
  fp = "glsl",
  frag = "javascript",
  frg = "glsl",
  fsh = "glsl",
  fshader = "glsl",
  geo = "glsl",
  geom = "glsl",
  glslf = "glsl",
  glslv = "glsl",
  god = "ruby",
  grt = "groovy",
  gshader = "glsl",
  gtpl = "groovy",
  gvy = "groovy",
  gyp = "python",
  gypi = "python",
  ["h++"] = "c++",
  handlebars = "handlebars",
  hic = "clojure",
  hta = "html",
  htmlhl = "html",
  hx = "haxe",
  hxsl = "haxe",
  iced = "coffeescript",
  idc = "c",
  ins = "tex",
  ixx = "c++",
  jade = "pug",
  jake = "javascript",
  jbuilder = "ruby",
  jsb = "javascript",
  jscad = "javascript",
  jsfl = "javascript",
  jslib = "javascript",
  jsm = "javascript",
  jsonnet = "jsonnet",
  jspre = "javascript",
  jss = "javascript",
  kojo = "scala",
  lbx = "tex",
  libsonnet = "jsonnet",
  linq = "c#",
  lmi = "python",
  make = "makefile",
  makefile = "makefile",
  matlab = "matlab",
  mdx = "markdown",
  me = "roff",
  mkdown = "markdown",
  mkfile = "makefile",
  ml4 = "ocaml",
  mspec = "ruby",
  nasm = "assembly",
  njs = "javascript",
  nomad = "hcl",
  ny = "common lisp",
  p8 = "lua",
  pascal = "pascal",
  pd_lua = "lua",
  perl = "perl",
  pgsql = "plpgsql",
  ph = "perl",
  php3 = "php",
  php4 = "php",
  php5 = "php",
  phps = "php",
  pm = "perl",
  podsl = "common lisp",
  podspec = "ruby",
  prawn = "ruby",
  py3 = "python",
  pyde = "python",
  pyp = "python",
  pyt = "python",
  rabl = "ruby",
  rbi = "ruby",
  rbuild = "ruby",
  rbx = "ruby",
  rbxs = "lua",
  rchit = "glsl",
  re = "c++",
  rmiss = "glsl",
  ronn = "markdown",
  rpy = "python",
  rsin = "rust",
  rsx = "r",
  ruby = "ruby",
  shader = "glsl",
  shin = "shell",
  sjs = "javascript",
  sls = "scheme",
  smk = "python",
  sps = "scheme",
  ssjs = "javascript",
  t = "perl",
  tac = "python",
  tesc = "glsl",
  tese = "glsl",
  thor = "ruby",
  tmux = "shell",
  tool = "shell",
  vapi = "vala",
  vert = "glsl",
  vimrc = "vim script",
  vmb = "vim script",
  vrx = "glsl",
  vsh = "glsl",
  vshader = "glsl",
  watchr = "ruby",
  wlua = "lua",
  workbook = "markdown",
  workflow = "hcl",
  wsgi = "python",
  xpy = "python",
  xrl = "erlang",
  xsjs = "javascript",
  xsjslib = "javascript",
  yrl = "erlang",
  ["zsh-theme"] = "shell"
}