AutoCorrect

🎯 AutoCorrect 的愿景是提供一套标准化的文案较正方案。以便于在各类场景（例如：撰写书籍、文档、内容发布、项目源代码...）里面应用，让使用者轻松实现标准化、专业化的文案输出 / 校正。

AutoCorrect is a linter and formatter to help you to improve copywriting, correct spaces, words, punctuations between CJK (Chinese, Japanese, Korean).

Like Eslint, Rubocop, Gofmt ..., AutoCorrect allows us to check source code, and output as colorized diff with corrected suggest. You can integrate to CI (GitLab CI, GitHub Action, Travis CI....) for use to checking the contents in source code. Recognize the file name, and find out the strings and the comment part.

基于 Rust 编写的 CLI 工具，用于「自动纠正」或「检查并建议」文案，给 CJK（中文、日语、韩语）与英文混写的场景，补充正确的空格，纠正单词，同时尝试以安全的方式自动纠正标点符号等等。

类似 ESlint、Rubocop、Gofmt 等工具，AutoCorrect 可以用于 CI 环境，它提供 Lint 功能，能便捷的检测出项目中有问题的文案，起到统一规范的作用。

支持各种类型源代码文件，能自动识别文件名，并准确找到字符串、注释做自动纠正。

此方案最早于 2013 年 出现于 Ruby China 的项目，并逐步完善规则细节，当前准确率较高（极少数异常情况），你可以放心用来辅助你完成自动纠正动作。

Features

• Auto add spacing between CJK (Chinese, Japanese, Korean) and English words.
• Support more than 27 programming languages (Markdown, JSON, YAML, JavaScript, HTML ...), use AST parser to only check for strings, comments.
• Correct punctuations into full-width near the CJK.
• Correct punctuations into half-width in english contents.
• (Experimental) Spellcheck and correct words by your own dictionary.
• Lint checking and output diff or JSON result, so you can integrate to everywhere (GitLab CI, GitHub Action, VS Code, Vim, Emacs...)
• Allows using .gitignore or .autocorrectignore to ignore files that you want to ignore.
• Desktop app for macOS, (Windows, Linux WIP).

Installation

$ brew install autocorrect

Or you can just install via this:

$ curl -sSL https://git.io/JcGER | bash

After that, you will get autocorrect command.

AutoCorrect 1.11.0
Jason Lee <[email protected]
A linter and formatter to help you to improve copywriting, correct spaces, words, punctuations between CJK (Chinese, Japanese, Korean).

USAGE:
    autocorrect [FLAGS] [OPTIONS] [file]... [SUBCOMMAND]

FLAGS:
        --debug      Print debug message.
        --type       Directly use set file type.
        --fix        Automatically fix problems and rewrite file.
    -h, --help       Prints help information
        --lint       Lint and output problems.
    -V, --version    Prints version information

OPTIONS:
    -c, --config <config>       Special config file. [default: .autocorrectrc]
        --format <formatter>    Choose an output formatter. [default: diff]  [possible values: json, diff]
        --threads <threads>     Number of threads, 0 - use number of CPU. [default: 0]

ARGS:
    <file>...    Target filepath or dir for format. [default: .]

SUBCOMMANDS:
    help       Prints this message or the help of the given subcommand(s)
    init       Init AutoCorrect config file.
    update     Update AutoCorrect to latest version.

Upgrade

Since: 1.9.0

AutoCorrect allows you to upgrade itself by autocorrect update command.

$ autocorrect update

NOTE: This command need you input your password, because it will install bin into /usr/local/bin directory.

Usage

• Use in CLI
• Configuration
• VS Code Extension
• GitHub Action
• GitLab CI
• Use for programming

Use in CLI

$ autocorrect text.txt
你好 Hello 世界

$ echo "hello世界" | autocorrect --stdin
hello 世界

$ autocorrect --fix text.txt
$ autocorrect --fix zh-CN.yml
$ autocorrect --fix

Lint

$ autocorrect --lint --format json text.txt

$ autocorrect --lint text.txt

Error: 1, Warning: 0

text.txt:1:3
-你好Hello世界
+你好 Hello 世界

You also can lint multiple files:

$ autocorrect --lint

Configuration

$ autocorrect init
Fetching https://github.com/huacnlee/autocorrect/raw/main/.autocorrectrc.template
AutoCorrect init config: .autocorrectrc

NOTE: If you download fail, try to use autocorrect init --local command again.

Now the .autocorrectrc file has created.

.autocorrectrc is allows use YAML, JSON format.

Config file example:

# Config rules
rules:
  # Auto add spacing between CJK (Chinese, Japanese, Korean) and English words.
  # 0 - off, 1 - error, 2 - warning
  space-word: 1
  # Add space between some punctuations.
  space-punctuation: 1
  # Add space between brackets (), [] when near the CJK.
  space-bracket: 1
  # Convert to fullwidth.
  fullwidth: 1
  # To remove space near the fullwidth.
  no-space-fullwidth: 1
  # Fullwidth alphanumeric characters to halfwidth.
  halfwidth-word: 1
  # Fullwidth punctuations to halfwidth in english.
  halfwidth-punctuation: 1
  # Spellcheck
  spellcheck: 2
textRules:
  # Config special rules for some texts
  # For example, if we wants to let "Hello你好" just warning, and "Hi你好" to ignore
  # "Hello你好": 2
  # "Hi你好": 0
spellcheck:
  # Correct Words (Case insensitive) for by Spellcheck
  words:
    - GitHub
    - App Store
    # This means "appstore" into "App Store"
    - AppStore = App Store
    - Git
    - Node.js
    - nodejs = Node.js
    - VIM
    - DNS
    - HTTP
    - SSL

Ignore option

Since: 2.2.0

When you wants to config some special words or texts to ignore on format or lint.

The textRules config may help you.

For example we wants:

• Hello世界 - To just give warning.
• Hi你好 - To ignore.

Use can config:

textRules:
  Hello世界: 2
  Hi你好: 0

After that, the AutoCorrect will follow your textRules to process.

Ignore files

Use .autocorrectignore to ignore files

Sometimes, you may want to ignore some special files that not wants to check.

By default, the file matched .gitignore rule will be ignored.

You can also use .autocorrectignore to ignore other files, format like .gitignore.

Disable with inline comment

If you just want to disable some special lines in file, you can write a comment autocorrect-disable, when AutoCorrect matched comment include that, it will disable temporary.

And then, you can use autocorrect-enable to reopen it again.

For example, in JavaScript:

function hello() {
  // autocorrect-disable
  console.log('现在这行开始autocorrect会暂时禁用');
  console.log('这行也是disable的状态');
  // autocorrect-enable
  let a = '现在起autocorrect回到了启用的状态';
}

The output will:

function hello() {
  // autocorrect-disable
  console.log('现在这行开始autocorrect会暂时禁用');
  console.log('这行也是disable的状态');
  // autocorrect-enable
  let a = '现在起 autocorrect 回到了启用的状态';
}

Disable some rules

Since: 2.0

You can use autocorrect-disable <rule> in comment to disable some rules.

Rule names please see: Configuration

function hello() {
  // autocorrect-disable space-word
  console.log('现在这行开始autocorrect会暂时禁用.');
  // autocorrect-disable fullwidth
  console.log('这行也是disable的状态.');
  // autocorrect-enable
  let a = '现在起autocorrect回到了启用的状态.';
}

Will get:

function hello() {
  // autocorrect-disable space-word
  console.log('现在这行开始autocorrect会暂时禁用。');
  // autocorrect-disable fullwidth, space-word
  console.log('这行也是disable的状态.');
  // autocorrect-enable
  let a = '现在起 autocorrect 回到了启用的状态。';
}

VS Code Extension

https://marketplace.visualstudio.com/items?itemName=huacnlee.auto-correct

内置 Visual Studio Code 插件，安装后会将 AutoCorrect 和 Visual Studio Code 完整集成，可以达到「保存自动格式化」或「纠正提示」。

如下图：

GitHub Action

https://github.com/marketplace/actions/huacnlee-autocorrect

Add to your .github/workflows/ci.yml

steps:
  - name: Check source code
    uses: actions/checkout@main

  - name: AutoCorrect
    uses: huacnlee/autocorrect-action@main

GitLab CI

Add to your .gitlab-ci.yml, to use huacnlee/autocorrect Docker image to check.

autocorrect:
  stage: build
  image: huacnlee/autocorrect:latest
  script:
    - autocorrect --lint
  # Enable allow_failure if you wants.
  # allow_failure: true

Use for programming

AutoCorrect make for supports use in many programming languages.

• Rust - autocorrect
• Ruby - autocorrect-rb
• Go - autocorrect-go
• Python - autocorrect-py
• Node.js - autocorrect-node
• JavaScript (Browser) - autocorrect-wasm
• Java - autocorrect-java

Benchmark

MacBook Pro (13-inch, M1, 2020)

Use make bench to run benchmark tests.

See autocorrect/src/benches/example.rs for details.

test bench_format_050                 ... bench:       9,132 ns/iter (+/- 88)
test bench_format_100                 ... bench:      16,892 ns/iter (+/- 319)
test bench_format_400                 ... bench:      55,126 ns/iter (+/- 1,086)
test bench_format_html                ... bench:     198,822 ns/iter (+/- 2,228)
test bench_format_javascript          ... bench:      88,774 ns/iter (+/- 1,333)
test bench_format_json                ... bench:      42,868 ns/iter (+/- 325)
test bench_format_json_with_2k_lines  ... bench:   9,664,245 ns/iter (+/- 494,651)
test bench_halfwidth_full_english_100 ... bench:      11,242 ns/iter (+/- 550)
test bench_markdown                   ... bench:     998,470 ns/iter (+/- 18,294)
test bench_spellcheck_100             ... bench:      54,168 ns/iter (+/- 451)
test bench_spellcheck_400             ... bench:     189,885 ns/iter (+/- 4,172)
test bench_spellcheck_50              ... bench:      34,920 ns/iter (+/- 2,111)

Real world benchmark

With MDN Translated Content project, it's has about 30K files.

~/work/translated-content $ autocorrect --fix
AutoCorrect spend time: 8402.538ms

License

This project under MIT license.