Merge pull request #21633 from gpanders/editorconfig

Builtin EditorConfig support

8 files changed, 568 insertions, 2 deletions

diff --git a/runtime/doc/editorconfig.txt b/runtime/doc/editorconfig.txt
new file mode 100644
index 0000000000..e93713e5ff
--- /dev/null
+++ b/runtime/doc/editorconfig.txt
@@ -0,0 +1,89 @@
+*editorconfig.txt*  Nvim
+
+
+                            NVIM REFERENCE MANUAL
+
+
+EditorConfig integration                                        *editorconfig*
+
+Nvim natively supports EditorConfig. When a file is opened, Nvim searches
+upward through all of the parent directories of that file looking for
+".editorconfig" files. Each of these is parsed and any properties that match
+the opened file are applied.
+
+For more information on EditorConfig, see https://editorconfig.org/.
+
+                                                       *g:editorconfig_enable*
+EditorConfig integration can be disabled by adding >lua
+
+    vim.g.editorconfig_enable = false
+<
+to the user's |init.lua| file (or the Vimscript equivalent to |init.vim|).
+
+                                                              *b:editorconfig*
+When Nvim finds a valid .editorconfig file it will store the applied
+properties in the buffer variable |b:editorconfig|.
+
+                                                     *editorconfig-properties*
+The following properties are supported by default:
+
+                                                           *editorconfig_root*
+root                    If "true", then stop searching for .editorconfig files
+                        in parent directories. This property must be at the
+                        top-level of the .editorconfig file (i.e. it must not
+                        be within a glob section).
+
+                                                        *editorconfig_charset*
+charset                 One of "utf-8", "utf-8-bom", "latin1", "utf-16be", or
+                        "utf-16le". Sets the 'fileencoding' and 'bomb'
+                        options.
+
+                                                    *editorconfig_end_of_line*
+end_of_line             One of "lf", "crlf", or "cr". These correspond to
+                        setting 'fileformat' to "unix", "dos", or "mac",
+                        respectively.
+
+                                                   *editorconfig_indent_style*
+indent_style            One of "tab" or "space". Sets the 'expandtab' option.
+
+                                                    *editorconfig_indent_size*
+indent_size             A number indicating the size of a single indent.
+                        Alternatively, use the value "tab" to use the value of
+                        the tab_width property. Sets the 'shiftwidth' and
+                        'softtabstop'.
+
+                                           *editorconfig_insert_final_newline*
+insert_final_newline    "true" or "false" to ensure the file always has a
+                        trailing newline as its last byte. Sets the
+                        'fixendofline' and 'endofline' options.
+
+                                                *editorconfig_max_line_length*
+max_line_length         A number indicating the maximum length of a single
+                        line. Sets the 'textwidth' option.
+
+                                                      *editorconfig_tab_width*
+tab_width               The display size of a single tab character. Sets the
+                        'tabstop' option.
+
+                                       *editorconfig_trim_trailing_whitespace*
+trim_trailing_whitespace
+                        When "true", trailing whitespace is automatically
+                        removed when the buffer is written.
+
+                                              *editorconfig-custom-properties*
+New properties can be added by adding a new entry to the "properties" table.
+The table key is a property name and the value is a callback function which
+accepts the number of the buffer to be modified, the value of the property
+in the .editorconfig file, and (optionally) a table containing all of the
+other properties and their values (useful for properties which depend on other
+properties). The value is always a string and must be coerced if necessary.
+Example: >lua
+
+    require('editorconfig').properties.foo = function(bufnr, val, opts)
+      if opts.charset and opts.charset ~= "utf-8" then
+        error("foo can only be set when charset is utf-8", 0)
+      end
+      vim.b[bufnr].foo = val
+    end
+<
+ vim:tw=78:ts=8:noet:ft=help:norl:
diff --git a/runtime/doc/news.txt b/runtime/doc/news.txt
index ffc8ce2c88..03bbf1567b 100644
--- a/runtime/doc/news.txt
+++ b/runtime/doc/news.txt
@@ -50,6 +50,13 @@ NEW FEATURES                                                    *news-features*
 
 The following new APIs or features were added.
 
+• EditorConfig support is now builtin. This is enabled by default and happens
+  automatically. To disable it, users should add >lua
+
+    vim.g.editorconfig_enable = false
+<
+  (or the Vimscript equivalent) to their |config| file.
+
 • Added a |vim.lsp.codelens.clear()| function to clear codelenses.
 
 • |vim.inspect_pos()|, |vim.show_pos()| and |:Inspect| allow a user to get or show items
diff --git a/runtime/lua/editorconfig.lua b/runtime/lua/editorconfig.lua
new file mode 100644
index 0000000000..2079006234
--- /dev/null
+++ b/runtime/lua/editorconfig.lua
@@ -0,0 +1,246 @@
+local M = {}
+
+M.properties = {}
+
+--- Modified version of the builtin assert that does not include error position information
+---
+---@param v any Condition
+---@param message string Error message to display if condition is false or nil
+---@return any v if not false or nil, otherwise an error is displayed
+---
+---@private
+local function assert(v, message)
+  return v or error(message, 0)
+end
+
+--- Show a warning message
+---
+---@param msg string Message to show
+---
+---@private
+local function warn(msg, ...)
+  vim.notify(string.format(msg, ...), vim.log.levels.WARN, {
+    title = 'editorconfig',
+  })
+end
+
+function M.properties.charset(bufnr, val)
+  assert(
+    vim.tbl_contains({ 'utf-8', 'utf-8-bom', 'latin1', 'utf-16be', 'utf-16le' }, val),
+    'charset must be one of "utf-8", "utf-8-bom", "latin1", "utf-16be", or "utf-16le"'
+  )
+  if val == 'utf-8' or val == 'utf-8-bom' then
+    vim.bo[bufnr].fileencoding = 'utf-8'
+    vim.bo[bufnr].bomb = val == 'utf-8-bom'
+  elseif val == 'utf-16be' then
+    vim.bo[bufnr].fileencoding = 'utf-16'
+  else
+    vim.bo[bufnr].fileencoding = val
+  end
+end
+
+function M.properties.end_of_line(bufnr, val)
+  vim.bo[bufnr].fileformat = assert(
+    ({ lf = 'unix', crlf = 'dos', cr = 'mac' })[val],
+    'end_of_line must be one of "lf", "crlf", or "cr"'
+  )
+end
+
+function M.properties.indent_style(bufnr, val, opts)
+  assert(val == 'tab' or val == 'space', 'indent_style must be either "tab" or "space"')
+  vim.bo[bufnr].expandtab = val == 'space'
+  if val == 'tab' and not opts.indent_size then
+    vim.bo[bufnr].shiftwidth = 0
+    vim.bo[bufnr].softtabstop = 0
+  end
+end
+
+function M.properties.indent_size(bufnr, val, opts)
+  if val == 'tab' then
+    vim.bo[bufnr].shiftwidth = 0
+    vim.bo[bufnr].softtabstop = 0
+  else
+    local n = assert(tonumber(val), 'indent_size must be a number')
+    vim.bo[bufnr].shiftwidth = n
+    vim.bo[bufnr].softtabstop = -1
+    if not opts.tab_width then
+      vim.bo[bufnr].tabstop = n
+    end
+  end
+end
+
+function M.properties.tab_width(bufnr, val)
+  vim.bo[bufnr].tabstop = assert(tonumber(val), 'tab_width must be a number')
+end
+
+function M.properties.max_line_length(bufnr, val)
+  local n = tonumber(val)
+  if n then
+    vim.bo[bufnr].textwidth = n
+  else
+    assert(val == 'off', 'max_line_length must be a number or "off"')
+    vim.bo[bufnr].textwidth = 0
+  end
+end
+
+function M.properties.trim_trailing_whitespace(bufnr, val)
+  assert(
+    val == 'true' or val == 'false',
+    'trim_trailing_whitespace must be either "true" or "false"'
+  )
+  if val == 'true' then
+    vim.api.nvim_create_autocmd('BufWritePre', {
+      group = 'editorconfig',
+      buffer = bufnr,
+      callback = function()
+        local view = vim.fn.winsaveview()
+        vim.api.nvim_command('silent! undojoin')
+        vim.api.nvim_command('silent keepjumps keeppatterns %s/\\s\\+$//e')
+        vim.fn.winrestview(view)
+      end,
+    })
+  else
+    vim.api.nvim_clear_autocmds({
+      event = 'BufWritePre',
+      group = 'editorconfig',
+      buffer = bufnr,
+    })
+  end
+end
+
+function M.properties.insert_final_newline(bufnr, val)
+  assert(val == 'true' or val == 'false', 'insert_final_newline must be either "true" or "false"')
+  vim.bo[bufnr].fixendofline = val == 'true'
+  vim.bo[bufnr].endofline = val == 'true'
+end
+
+--- Modified version of |glob2regpat()| that does not match path separators on *.
+---
+--- This function replaces single instances of * with the regex pattern [^/]*. However, the star in
+--- the replacement pattern also gets interpreted by glob2regpat, so we insert a placeholder, pass
+--- it through glob2regpat, then replace the placeholder with the actual regex pattern.
+---
+---@param glob string Glob to convert into a regular expression
+---@return string Regular expression
+---
+---@private
+local function glob2regpat(glob)
+  local placeholder = '@@PLACEHOLDER@@'
+  return (
+    string.gsub(
+      vim.fn.glob2regpat(
+        vim.fn.substitute(
+          string.gsub(glob, '{(%d+)%.%.(%d+)}', '[%1-%2]'),
+          '\\*\\@<!\\*\\*\\@!',
+          placeholder,
+          'g'
+        )
+      ),
+      placeholder,
+      '[^/]*'
+    )
+  )
+end
+
+--- Parse a single line in an EditorConfig file
+---
+---@param line string Line
+---@return string|nil If the line contains a pattern, the glob pattern
+---@return string|nil If the line contains a key-value pair, the key
+---@return string|nil If the line contains a key-value pair, the value
+---
+---@private
+local function parse_line(line)
+  if line:find('^%s*[^ #;]') then
+    local glob = (line:match('%b[]') or ''):match('^%s*%[(.*)%]%s*$')
+    if glob then
+      return glob, nil, nil
+    end
+
+    local key, val = line:match('^%s*([^:= ][^:=]-)%s*[:=]%s*(.-)%s*$')
+    if key ~= nil and val ~= nil then
+      return nil, key:lower(), val:lower()
+    end
+  end
+end
+
+--- Parse options from an .editorconfig file
+---
+---@param filepath string File path of the file to apply EditorConfig settings to
+---@param dir string Current directory
+---@return table Table of options to apply to the given file
+---
+---@private
+local function parse(filepath, dir)
+  local pat = nil
+  local opts = {}
+  local f = io.open(dir .. '/.editorconfig')
+  if f then
+    for line in f:lines() do
+      local glob, key, val = parse_line(line)
+      if glob then
+        glob = glob:find('/') and (dir .. '/' .. glob:gsub('^/', '')) or ('**/' .. glob)
+        local ok, regpat = pcall(glob2regpat, glob)
+        if ok then
+          pat = vim.regex(regpat)
+        else
+          pat = nil
+          warn('editorconfig: Error occurred while parsing glob pattern "%s": %s', glob, regpat)
+        end
+      elseif key ~= nil and val ~= nil then
+        if key == 'root' then
+          opts.root = val == 'true'
+        elseif pat and pat:match_str(filepath) then
+          opts[key] = val
+        end
+      end
+    end
+    f:close()
+  end
+  return opts
+end
+
+--- Configure the given buffer with options from an .editorconfig file
+---
+---@param bufnr number Buffer number to configure
+---
+---@private
+function M.config(bufnr)
+  bufnr = bufnr or vim.api.nvim_get_current_buf()
+  local path = vim.fs.normalize(vim.api.nvim_buf_get_name(bufnr))
+  if vim.bo[bufnr].buftype ~= '' or not vim.bo[bufnr].modifiable or path == '' then
+    return
+  end
+
+  local opts = {}
+  for parent in vim.fs.parents(path) do
+    for k, v in pairs(parse(path, parent)) do
+      if opts[k] == nil then
+        opts[k] = v
+      end
+    end
+
+    if opts.root then
+      break
+    end
+  end
+
+  local applied = {}
+  for opt, val in pairs(opts) do
+    if val ~= 'unset' then
+      local func = M.properties[opt]
+      if func then
+        local ok, err = pcall(func, bufnr, val, opts)
+        if ok then
+          applied[opt] = val
+        else
+          warn('editorconfig: invalid value for option %s: %s. %s', opt, val, err)
+        end
+      end
+    end
+  end
+
+  vim.b[bufnr].editorconfig = applied
+end
+
+return M
diff --git a/runtime/lua/vim/filetype.lua b/runtime/lua/vim/filetype.lua
index 7eec567b66..c3ab39a1a3 100644
--- a/runtime/lua/vim/filetype.lua
+++ b/runtime/lua/vim/filetype.lua
@@ -1378,13 +1378,13 @@ local filename = {
   npmrc = 'dosini',
   ['/etc/yum.conf'] = 'dosini',
   ['.npmrc'] = 'dosini',
-  ['.editorconfig'] = 'dosini',
   ['/etc/pacman.conf'] = 'confini',
   ['mpv.conf'] = 'confini',
   dune = 'dune',
   jbuild = 'dune',
   ['dune-workspace'] = 'dune',
   ['dune-project'] = 'dune',
+  ['.editorconfig'] = 'editorconfig',
   ['elinks.conf'] = 'elinks',
   ['mix.lock'] = 'elixir',
   ['filter-rules'] = 'elmfilt',
diff --git a/runtime/plugin/editorconfig.lua b/runtime/plugin/editorconfig.lua
new file mode 100644
index 0000000000..60eb861aaa
--- /dev/null
+++ b/runtime/plugin/editorconfig.lua
@@ -0,0 +1,11 @@
+if vim.g.editorconfig_enable == false or vim.g.editorconfig_enable == 0 then
+  return
+end
+
+local group = vim.api.nvim_create_augroup('editorconfig', {})
+vim.api.nvim_create_autocmd({ 'BufNewFile', 'BufRead', 'BufFilePost' }, {
+  group = group,
+  callback = function(args)
+    require('editorconfig').config(args.buf)
+  end,
+})
diff --git a/runtime/syntax/editorconfig.vim b/runtime/syntax/editorconfig.vim
new file mode 100644
index 0000000000..006b99cec5
--- /dev/null
+++ b/runtime/syntax/editorconfig.vim
@@ -0,0 +1,18 @@
+runtime! syntax/dosini.vim
+unlet! b:current_syntax
+
+syntax match editorconfigInvalidProperty "^\s*\zs\w\+\ze\s*="
+syntax keyword editorconfigProperty root
+
+lua<<
+local props = {}
+for k in pairs(require('editorconfig').properties) do
+  props[#props + 1] = k
+end
+vim.cmd(string.format('syntax keyword editorconfigProperty %s', table.concat(props, ' ')))
+.
+
+hi def link editorconfigInvalidProperty Error
+hi def link editorconfigProperty dosiniLabel
+
+let b:current_syntax = 'editorconfig'
diff --git a/src/nvim/testdir/test_filetype.vim b/src/nvim/testdir/test_filetype.vim
index 912e702631..c17ac16a97 100644
--- a/src/nvim/testdir/test_filetype.vim
+++ b/src/nvim/testdir/test_filetype.vim
@@ -162,7 +162,7 @@ let s:filename_checks = {
     \ 'dnsmasq': ['/etc/dnsmasq.conf', '/etc/dnsmasq.d/file', 'any/etc/dnsmasq.conf', 'any/etc/dnsmasq.d/file'],
     \ 'dockerfile': ['Containerfile', 'Dockerfile', 'dockerfile', 'file.Dockerfile', 'file.dockerfile', 'Dockerfile.debian', 'Containerfile.something'],
     \ 'dosbatch': ['file.bat'],
-    \ 'dosini': ['.editorconfig', '/etc/yum.conf', 'file.ini', 'npmrc', '.npmrc', 'php.ini', 'php.ini-5', 'php.ini-file', '/etc/yum.repos.d/file', 'any/etc/yum.conf', 'any/etc/yum.repos.d/file', 'file.wrap'],
+    \ 'dosini': ['/etc/yum.conf', 'file.ini', 'npmrc', '.npmrc', 'php.ini', 'php.ini-5', 'php.ini-file', '/etc/yum.repos.d/file', 'any/etc/yum.conf', 'any/etc/yum.repos.d/file', 'file.wrap'],
     \ 'dot': ['file.dot', 'file.gv'],
     \ 'dracula': ['file.drac', 'file.drc', 'filelvs', 'filelpe', 'drac.file', 'lpe', 'lvs', 'some-lpe', 'some-lvs'],
     \ 'dtd': ['file.dtd'],
@@ -174,6 +174,7 @@ let s:filename_checks = {
     \ 'dylanlid': ['file.lid'],
     \ 'ecd': ['file.ecd'],
     \ 'edif': ['file.edf', 'file.edif', 'file.edo'],
+    \ 'editorconfig': ['.editorconfig'],
     \ 'eelixir': ['file.eex', 'file.leex'],
     \ 'elinks': ['elinks.conf'],
     \ 'elixir': ['file.ex', 'file.exs', 'mix.lock'],
diff --git a/test/functional/plugin/editorconfig_spec.lua b/test/functional/plugin/editorconfig_spec.lua
new file mode 100644
index 0000000000..f71b8088ed
--- /dev/null
+++ b/test/functional/plugin/editorconfig_spec.lua
@@ -0,0 +1,194 @@
+local helpers = require('test.functional.helpers')(after_each)
+local clear = helpers.clear
+local command = helpers.command
+local eq = helpers.eq
+local pathsep = helpers.get_pathsep()
+local curbufmeths = helpers.curbufmeths
+
+local testdir = 'Xtest-editorconfig'
+
+local function test_case(name, expected)
+  local filename = testdir .. pathsep .. name
+  command('edit ' .. filename)
+  for opt, val in pairs(expected) do
+    eq(val, curbufmeths.get_option(opt), name)
+  end
+end
+
+setup(function()
+  helpers.mkdir_p(testdir)
+  helpers.write_file(
+    testdir .. pathsep .. '.editorconfig',
+    [[
+    root = true
+
+    [3_space.txt]
+    indent_style = space
+    indent_size = 3
+    tab_width = 3
+
+    [4_space.py]
+    indent_style = space
+    indent_size = 4
+    tab_width = 8
+
+    [space.txt]
+    indent_style = space
+    indent_size = tab
+
+    [tab.txt]
+    indent_style = tab
+
+    [4_tab.txt]
+    indent_style = tab
+    indent_size = 4
+    tab_width = 4
+
+    [4_tab_width_of_8.txt]
+    indent_style = tab
+    indent_size = 4
+    tab_width = 8
+
+    [lf.txt]
+    end_of_line = lf
+
+    [crlf.txt]
+    end_of_line = crlf
+
+    [cr.txt]
+    end_of_line = cr
+
+    [utf-8.txt]
+    charset = utf-8
+
+    [utf-8-bom.txt]
+    charset = utf-8-bom
+
+    [utf-16be.txt]
+    charset = utf-16be
+
+    [utf-16le.txt]
+    charset = utf-16le
+
+    [latin1.txt]
+    charset = latin1
+
+    [with_newline.txt]
+    insert_final_newline = true
+
+    [without_newline.txt]
+    insert_final_newline = false
+
+    [trim.txt]
+    trim_trailing_whitespace = true
+
+    [no_trim.txt]
+    trim_trailing_whitespace = false
+
+    [max_line_length.txt]
+    max_line_length = 42
+    ]]
+  )
+end)
+
+teardown(function()
+  helpers.rmdir(testdir)
+end)
+
+describe('editorconfig', function()
+  before_each(function()
+    -- Remove -u NONE so that plugins (i.e. editorconfig.lua) are loaded
+    clear({ args_rm = { '-u' } })
+  end)
+
+  it('sets indent options', function()
+    test_case('3_space.txt', {
+      expandtab = true,
+      shiftwidth = 3,
+      softtabstop = -1,
+      tabstop = 3,
+    })
+
+    test_case('4_space.py', {
+      expandtab = true,
+      shiftwidth = 4,
+      softtabstop = -1,
+      tabstop = 8,
+    })
+
+    test_case('space.txt', {
+      expandtab = true,
+      shiftwidth = 0,
+      softtabstop = 0,
+    })
+
+    test_case('tab.txt', {
+      expandtab = false,
+      shiftwidth = 0,
+      softtabstop = 0,
+    })
+
+    test_case('4_tab.txt', {
+      expandtab = false,
+      shiftwidth = 4,
+      softtabstop = -1,
+      tabstop = 4,
+    })
+
+    test_case('4_tab_width_of_8.txt', {
+      expandtab = false,
+      shiftwidth = 4,
+      softtabstop = -1,
+      tabstop = 8,
+    })
+  end)
+
+  it('sets end-of-line options', function()
+    test_case('lf.txt', { fileformat = 'unix' })
+    test_case('crlf.txt', { fileformat = 'dos' })
+    test_case('cr.txt', { fileformat = 'mac' })
+  end)
+
+  it('sets encoding options', function()
+    test_case('utf-8.txt', { fileencoding = 'utf-8', bomb = false })
+    test_case('utf-8-bom.txt', { fileencoding = 'utf-8', bomb = true })
+    test_case('utf-16be.txt', { fileencoding = 'utf-16', bomb = false })
+    test_case('utf-16le.txt', { fileencoding = 'utf-16le', bomb = false })
+    test_case('latin1.txt', { fileencoding = 'latin1', bomb = false })
+  end)
+
+  it('sets newline options', function()
+    test_case('with_newline.txt', { fixendofline = true, endofline = true })
+    test_case('without_newline.txt', { fixendofline = false, endofline = false })
+  end)
+
+  it('respects trim_trailing_whitespace', function()
+    local filename = testdir .. pathsep .. 'trim.txt'
+    -- luacheck: push ignore 613
+    local untrimmed = [[
+This line ends in whitespace 
+So does this one    
+And this one         
+But not this one
+]]
+    -- luacheck: pop
+    local trimmed = untrimmed:gsub('%s+\n', '\n')
+
+    helpers.write_file(filename, untrimmed)
+    command('edit ' .. filename)
+    command('write')
+    command('bdelete')
+    eq(trimmed, helpers.read_file(filename))
+
+    filename = testdir .. pathsep .. 'no_trim.txt'
+    helpers.write_file(filename, untrimmed)
+    command('edit ' .. filename)
+    command('write')
+    command('bdelete')
+    eq(untrimmed, helpers.read_file(filename))
+  end)
+
+  it('sets textwidth', function()
+    test_case('max_line_length.txt', { textwidth = 42 })
+  end)
+end)