cppcxy/emmyluacodestyle

fast, powerful, and feature-rich Lua formatting and checking tool.

EmmyLuaCodeStyle

English doc

编辑器支持

vscode: vscode-EmmyLuaCodeStyle vscode-Lua

intellij: intellij-EmmyLuaCodeStyle

neovim: Lua Language Server

项目介绍

该项目不是lua代码格式的规范, 而是基于C++实现的lua代码解析\格式化\代码诊断算法库和语言服务实例

该项目主要追求的是合理的格式化，多样的代码风格

该项目除了提供语言服务实例，也提供独立的命令行工具CodeFormat，他能用于批量的代码格式化和代码风格检查。

Feature

• 支持lua5.1, lua5.2, lua5.3, lua5.4, luajit
• 文档格式化
• 范围格式化
• 命名风格检查
• 代码格式检查
• 代码拼写检查
• 可以作为lua库/cli工具/C++解析lua库/语言服务使用

RoadMap

• plugin[0%]

文档

• 格式化行为介绍
• 如何配置格式化
• 如何配置命名风格检查

Contribute

任何pr或者issue都是欢迎的

Build

如果希望能自己编译该项目，确保自己的编译器能基本支持C++20：

• VS2019 16.10以上
• gcc 10以上
• clang 10以上

mkdir build && cd build
cmake ..
cmake --build . 

Developed By

@CppCXY

Contributors

@obszczymucha

@Rainer Poisel

@AndreaWalchshoferSCCH

@一醉青衫

License

MIT

emmyluacodestyle's Issues

将格式化与风格诊断分离

实践中我们通常希望格式化总是输出出稳定的格式，但是手写代码允许有多种可用的格式，例如：

if 1
or 2 then
elseif 3 then
end

if     1
or     2 then
elseif 3 then
end

格式化的时候我希望总是可以输出成下面的格式，但是我手写时上面的格式也是允许的。
因此希望可以增加诊断专用的 allow 系列设置，这些设置仅仅是用于抑制风格诊断。

• 格式化时：根据设置进行格式化，不管 allow 系列设置
• 诊断时：先根据设置进行诊断，然后根据 allow 系列设置排除掉部分诊断

诊断结果中添加类别字段

我可以在上层根据类别来进行消息本地化，或是实现单独禁用某个诊断等功能

`local x <close>` 格式化后会丢失空格

会被格式化为 local x<close> 。
我觉得大部分的习惯（以及lua官方的测试代码中）都是保留空格的，因此默认保留空格比较好。

表达式运算符允许不需要包裹括号

例如：

t[#t+1] = true

log('traceback:'..traceback)

Add option for maximum line separation

As discussed in #33, remove line separation between blocks of code.

For example sake, lets say the variable is set as max_line_seperation = 1 and is the only option that is set.
The formatter would modify the file like so:

Unformatted file:

local M = {}




--- Utilities ---


-- Convert string to title case
function M.firstToUpper(str)
  return (str:gsub('^%l', string.upper))
end


-- Math operations
-- Check if number is prime
function M.is_prime(n)
  local max_divider_checked = math.sqrt(n) + 1
  for i = 2, max_divider_checked do
    if n % i == 0 then
      return false
    end
  end
  return true
end



return M

File diff:

+local M = {}
+
-
-
-
+--- Utilities ---
+
-
+-- Convert string to title case
+function M.firstToUpper(str)
+  return (str:gsub('^%l', string.upper))
+end
+
-
+-- Math operations
+-- Check if number is prime
+function M.is_prime(n)
+  local max_divider_checked = math.sqrt(n) + 1
+  for i = 2, max_divider_checked do
+    if n % i == 0 then
+      return false
+    end
+  end
+  return true
+end
+
-
-
+return M

Final Formatted file:

local M = {}

--- Utilities ---

-- Convert string to title case
function M.firstToUpper(str)
  return (str:gsub('^%l', string.upper))
end

-- Math operations
-- Check if number is prime
function M.is_prime(n)
  local max_divider_checked = math.sqrt(n) + 1
  for i = 2, max_divider_checked do
    if n % i == 0 then
      return false
    end
  end
  return true
end

return M

当函数仅有一个参数且为字面量字符串或表时，移除括号

require('xxx')

格式化为

require 'xxx'

Default editorconfig

Hi, I use neovim and the sumneko LSP server, and I started using the new code formatting feature. BTW, thanks!!! great job :)
Even though it works great if I have a .editorconfig file in my project root, I wondered if it's possible to set a default configuration, so I don't need to set it in every project, but only when I want to change something.
Thanks!

格式化时等号对齐应该使用最小长度

local aaaaaa = 1
local b  = 1
local c        = 1

现在会被格式化为

local aaaaaa   = 1
local b        = 1
local c        = 1

但用户的期望一般是

local aaaaaa = 1
local b      = 1
local c      = 1

对于指定作用域内的变量对齐。

-- line base
local function lineunpack(msg, sz)
  local str = skynet.tostring(msg, sz)
  local t = {}
  for s in string.gmatch(str, "[%a%w]+") do
    table.insert(t, s)
  end
  return tunpack(t)
end

比如这个， str 和 t 就不会对齐，不能使用和模块内变量一样的逻辑嘛。

链式调用时，允许在冒号后面带个空格

Use `insertSpaces` in `formattingOptions` instead of `indent_style` in `.editorconfig`

See https://microsoft.github.io/language-server-protocol/specifications/specification-3-17/#formattingOptions

Since LSP specifies that this field will be included in the formatted request, you should use the LSP field first.
You should only use the settings in .editorconfig if formatting is not requested via LSP.

Clear postemptive spaces on comments

before:

after:

why doesn't the formatter clear all postemptive spaces? Seems like it ignores the comments

the lua extension has a function that would clear this as well:

[Feature Request] Remove last comma for parameters and argumens

function foo(a1, ) end

foo(1,)

become

function foo(a1) end

foo(1)

This behavior is like gofmt.

diagnostic task

• Reimplementing the diagnostic algorithm
• Use .lualint.json instead of .editorconfig for diagnostic configuration

Wrong indent warnings for neovim

I already posted the issue at lua-language-server repo (please see here), but I am not sure whether I should bring it here as well.

Describe the bug

I am using the formatter and code style check. Here is my config.

nvim_lsp.sumneko_lua.setup {
  cmd = { "lua-language-server", "--preview" },
  settings = {
    Lua = {
      runtime = {
        -- Tell the language server which version of Lua you're using (most likely LuaJIT in the case of Neovim)
        version = 'LuaJIT',
        -- Setup your lua path
        path = runtime_path,
      },

      diagnostics = {

        neededFileStatus = {
          ["codestyle-check"] = "Any",
        },

      },

      format = {
        enable = true,
        defaultConfig = {
          indent_style = 'space',
          indent_size = 2,
        },
      },

      workspace = {
        -- Make the server aware of Neovim runtime files
        library = vim.api.nvim_get_runtime_file("", true),
      },
      -- Do not send telemetry data containing a randomized but unique identifier
      telemetry = {
        enable = false,
      },
    },
  },
}

As the config above, my indentation size is 2 spaces. However, it gives me wrong warnings for the indentations in functions.

Environment (please complete the following information):

• OS: macOS (the latest version)
• Client: neovim

Thank you so much.

命名检查支持过滤掉开头的下划线

一般用下划线标记私有字段或元方法，之后的字符还是要符合命名检查

对于函数调用的格式化，是否还能再调整一下。

比如此行代码：

  return net.packString("s2c.game.Scene", desc = desc, map  = 1001, npcs = npcs })

格式化为：

  return net.packString("s2c.game.Scene",
                        { desc = desc, map  = 1001, npcs = npcs })

会比当前的：

    return net.packString("s2c.game.Scene",
        { desc = desc, map = 1001, npcs = npcs })

更符合直觉一些。

格式化选定内容/粘贴时格式化的问题

TEST [[
---@class Class

---@param k Class
for <?k?> in pairs(t) do
end
]]
[[
parameter k: Class
]]

选中倒数第二行的 parameter，选择 格式化选定内容 后，会导致倒数第三行格式改变，变为：

TEST [[
---@class Class

---@param k Class
for <?k?> in pairs(t) do
end
]]
    [[
parameter k: Class
]]

函数定义参数风格检查不支持

函数的定义参数，无论如何设置都不会提示警告

Disable formatting temporarily

Thanks a lot for your great work! I really like the formatter and use it frequently. I have a question regarding the functionalities:
Is there a possibility to disable formatting just like the EmmyLua-Annotation diagnostic?
Especially when it comes to automatic line breaks, such a thing would be really useful at certain points.

[Feature Request] Remove empty header and footer lines in function

local function foo()


    local bar = 1
    return 2

end

become:

local function foo()
    local bar = 1
    return 2
end

Code format not running.

I get a 'formatter not installed' error when I try to format code. I have format enabled and I am trying to use the default .editorconfig in my root directory.

有计划添加自动折行功能吗？

Design more reasonable rules to remove call parentheses

1. call_arg_parentheses = string_smart_remove_exclude_after_match(".:({")
2. call_arg_parenthese = smart_remove and setting by json

any other idea.

"max_line_length": "120"，超出每行最大字符数时，部分情况下格式化后的折行会很离谱

格式化前：
格式化后：
第一次格式化：
第二次格式化：

tablefield支持分块

通过空行分割后，上面一块的对齐规则应该与下面一行无关

当函数调用，以函数定义作为参数时的行为。

问题一：
对于配置了对齐函数调用参数的代码：

skynet.start(function()
               local a = 1
               local b = 2
             end)

实际上这种情况下，配置为不对齐更好看，因为如果函数层级多的话，会出现大量的空白：

skynet.start(function()
  local a = 1
  local b = 2
end)

需求：能否在即时函数定义作为参数的时候，对齐到外层函数调用，而不是对齐到即时函数的定义位置。这种情况下，

• 如果即时函数作为第一个参数，那应该表现很好。
• 如果即时函数作为第二、第三个函数，在即时函数前参数和 function 在一行。
• 如果即时函数后还有参数，那就跟在 end 后面，与它一行。
• 如果多个即时函数作为参数，那就把后面个即时函数的 function 和 前一个 end 放在一行。
• 如果参数有表的话又该怎么搞？

问题二：

local sid = skynet.call(".world", "lua", "userScene", { map = 1001, uid = 2 })
local desc, npcs = skynet.call(sid, "lua", "enter", { map = 1001, uid = 2 }) -- return desc

此断代码，不会对齐，诊断报缩进错误。

允许if的条件有2个空格，用于与之后的and对齐

[Enhancement] comments on `elseif` statements

It would be nice if this could be improved to not do this:

风格检查支持连续行手动对齐

当发现空格大于1个时，检查到后面的符号 and 在连续行上一致，因此过滤掉该空格的检查

[enhancement] can we support the following functions?

Hello, Can we add support for the following functions? I use them for C/C++ on vscode. It will be pretty handy for me if they are available in Lua also.

• SpaceBeforeFunctionOpenParenthesis
• SpaceBeforeOpenSquareBracket
• SpaceBeforeLambdaOpenParenthesis
• SpaceWithinExpression

SpaceBeforeFunctionOpenParenthesis

Before Formatting:

function ff()
end

After Formatting:

function ff ()
end

SpaceBeforeOpenSquareBracket

Before Formatting:

details["names"] = "ashok"

After Formatting:

details ["names"] = "ashok"

SpaceBeforeLambdaOpenParenthesis

Before Formatting:

local details ={}

After Formatting:

local details = {}

SpaceWithinExpression

Before Formatting:

i?"true":"false";
i>0

After Formatting:

i ? "true" : "false";
i > 0

Thank You,
ashok.

[BUG] simple block with comment

if a then return a --end  
end

会被格式化为

if a then return a --end   end

How to disable max_line_length check

Is it possible to disable the check of max_line_length? Thank you.

[BUG] if condition expression with comment

if a --b 
then

end

will be format as :

if a --b then

end

Count of spaces before inline comment

Currently the count of spaces before inline comment is 1. Is it possible to set it to a value we want? Thanks a lot.

test empty issue

保存格式化时，对于函数内的 local 变量无法格式化到。

比如，那个等号，就会显示还应该对齐。
进行二次格式化就可以。

多行字符串不对齐，看起来很难受

local sql = fmt("select a, b, c " ..
    "from table " ..
    "where a > 10 and b < 10")

期待的格式：

local sql = fmt("select a, b, c " ..
                "from table " ..
                "where a > 10 and b < 10")

Ubuntu nvim format 会在行尾出现^M

格式化后每行末尾都会出现^M（windows的换行符）

原issue：LuaLS/vscode-lua#55

Formatting table with index starting with bracket string results in invalid lua code

The fix (ffd2205) for LuaLS/lua-language-server#1205 doesn't account for indexes with operators.

local correctly_formats = {
    [ [[mystr]] ]       = 123,
    [ -[[mystr]] ]      = 123,
    [ #[[mystr]] ]      = 123,
    [ not [[mystr]] ]   = 123,
    [ ~[[mystr]] ]      = 123,
}
local incorrectly_formats = {
    [ [[mystr]] .. 123 ]    = 123,
    [ [[mystr]] + 123 ]     = 123,
    [ [[mystr]] - 123 ]     = 123,
    [ [[mystr]] * 123 ]     = 123,
    [ [[mystr]] / 123 ]     = 123,
    [ [[mystr]] // 123 ]    = 123,
    [ [[mystr]] % 123 ]     = 123,
    [ [[mystr]] ^ 123 ]     = 123,
    [ [[mystr]] == 123 ]    = 123,
    [ [[mystr]] ~= 123 ]    = 123,
    [ [[mystr]] >  123 ]    = 123,
    [ [[mystr]] <  123 ]    = 123,
    [ [[mystr]] >= 123 ]    = 123,
    [ [[mystr]] <= 123 ]    = 123,
    [ [[mystr]] and 123 ]   = 123,
    [ [[mystr]] or 123 ]    = 123,
    [ [[mystr]] & 123 ]     = 123,
    [ [[mystr]] | 123 ]     = 123,
    [ [[mystr]] ~ 123 ]     = 123,
    [ [[mystr]] >> 123 ]    = 123,
    [ [[mystr]] << 123 ]    = 123,
}

表字段无法对齐

return {
    {
        behaviors = {
            ["ranvier-wander"] = { areaRestricted = true, interval = 30 },
        },
        description = "A furry little squirrel",
        id = "squirrel",
        keywords = { "squirrel" },
        level = 2,
        name = "A Squirrel",
    },
}

这段代码无法对齐字段。

CLI should respect gitignore under workspace mode

BTW, I think an extra option like --ignore is necessary.

Help removing empty continuous lines?

I'm using sumneko_lua through neovim and testing the new formatting capabilities out using .editorconfig to see if I can't replace stylua as my default formatter but still keep a similar style.

In my config file I have the line max_continuous_line_distance = 1; My assumption is that this would remove empty lines between statements and would format the examples below, but it doesn't appear to do anything...

How can I achieve this result?
It successfully fixes indentation and providing diagnostics so the rest of the config is clearly working.

print('hello, world')


--- Modules ---




require('lsp.handlers')
require('lsp.null-ls')

require('lsp.installer').setup({


  capabilities = capabilities,
        language_servers = language_servers,

  on_attach = on_attach,
})


local example1 = 'foo'
local example2 = 'bar'

print('hello, world')

--- Modules ---

require('lsp.handlers')
require('lsp.null-ls')
require('lsp.installer').setup({
  capabilities = capabilities,
  language_servers = language_servers,
  on_attach = on_attach,
})

local example1 = 'foo'
local example2 = 'bar'

align_call_args = false should keep the indent

local function describe(tag, f) end

describe('long' ..
    'tag', function()
end)

After formating:

• align_call_args = true

local function describe(tag, f) end

describe('long' ..
    'tag', function()
    end)

end) have been indented.

• align_call_args = false

local function describe(tag, f) end

describe('long' ..
'tag', function()
end)

tag has been deindented.

Expected behavior:

Keep the raw format if align_call_args = false .

关于括号嵌套的格式化有些问题

使用环境是cocoslua
应该是这样的

使用格式化后的样式

Binary in Release zip can't be executed

unzip -Z linux-x64.zip output:

Archive:  linux-x64.zip
Zip file size: 2812943 bytes, number of entries: 4
drwxr-xr-x  3.0 unx        0 bx stor 22-Apr-12 17:23 linux-x64/
drwxr-xr-x  3.0 unx        0 bx stor 22-Apr-12 17:23 linux-x64/bin/
-rw-r--r--  3.0 unx  6992816 bx defN 22-Apr-12 17:23 linux-x64/bin/CodeFormatServer
-rw-r--r--  3.0 unx  3900568 bx defN 22-Apr-12 17:23 linux-x64/bin/CodeFormat
4 files, 10893384 bytes uncompressed, 2812253 bytes compressed:  74.2%

I want to use this tool in GitHub action. tar.gz format is better.

Region Folding indent

Befor Formating

After:

Befor is correct

支持 if 的对齐

指定字符串使用单引号还是双引号

在指定使用单引号时：

print('xxx', "xxx")

格式化为

print('xxx', 'xxx')

例外规则：字符串内包含指定的引号类型时不进行格式化，例如：

print('xxx', "xxx", "xx'xx")

格式化为

print('xxx', 'xxx', "xx'xx")

Truncate the lines in lua-language-server if texts have error

if true then
end
end

local function foo()
end

->

if true then
end

But in CLI, CodeFormat format -f hint Exist lua syntax error and return error code 255.