Add the plugin to your .eslintrc and use the processor option in an overrides entry to enable the plugin's markdown/markdown processor on Markdown files.
Each fenced code block inside a Markdown document has a virtual filename appended to the Markdown file's path.
The virtual filename's extension will match the fenced code block's syntax tag, so for example, ```js code blocks in README.md would match README.md/*.js.
overrides glob patterns for these virtual filenames can customize configuration for code blocks without affecting regular code.
For more information on configuring processors, refer to the ESLint documentation.
// .eslintrc.jsmodule.exports={// 1. Add the plugin.plugins: ["markdown"],overrides: [{// 2. Enable the Markdown processor for all .md files.files: ["**/*.md"],processor: "markdown/markdown"},{// 3. Optionally, customize the configuration ESLint uses for ```js// fenced code blocks inside .md files.files: ["**/*.md/*.js"],// ...rules: {// ...}}]};
Frequently-Disabled Rules
Some rules that catch mistakes in regular code are less helpful in documentation.
For example, no-undef would flag variables that are declared outside of a code snippet because they aren't relevant to the example.
The plugin:markdown/recommended config disables these rules in Markdown files:
module.exports={// ...overrides: [// ...{// 1. Target ```js code blocks in .md files.files: ["**/*.md/*.js"],rules: {// 2. Disable other rules."no-console": "off","import/no-unresolved": "off"}}]};
Strict Mode
"use strict" directives in every code block would be annoying.
The plugin:markdown/recommended config enables the impliedStrict parser option and disables the strict rule in Markdown files.
This opts into strict mode parsing without repeated "use strict" directives.
Unsatisfiable Rules
Markdown code blocks are not real files, so ESLint's file-format rules do not apply.
The plugin:markdown/recommended config disables these rules in Markdown files:
eol-last: The Markdown parser trims trailing newlines from code blocks.
unicode-bom: Markdown code blocks do not have Unicode Byte Order Marks.
Migrating from eslint-plugin-markdown v1
eslint-plugin-markdown v1 used an older version of ESLint's processor API.
The Markdown processor automatically ran on .md, .mkdn, .mdown, and .markdown files, and it only extracted fenced code blocks marked with js, javascript, jsx, or node syntax.
Configuration specifically for fenced code blocks went inside an overrides entry with a files pattern matching the containing Markdown document's filename that applied to all fenced code blocks inside the file.
// .eslintrc.js for eslint-plugin-markdown v1module.exports={plugins: ["markdown"],overrides: [{files: ["**/*.md"],// In v1, configuration for fenced code blocks went inside an// `overrides` entry with a .md pattern, for example:parserOptions: {ecmaFeatures: {impliedStrict: true}},rules: {"no-console": "off"}}]};
RFC3 designed a new processor API to remove these limitations, and the new API was implemented as part of ESLint v6.
eslint-plugin-markdown v2 uses this new API.
All of the Markdown file extensions that were previously hard-coded are now fully configurable in .eslintrc.js.
Use the new processor option to apply the markdown/markdown processor on any Markdown documents matching a files pattern.
Each fenced code block inside a Markdown document has a virtual filename appended to the Markdown file's path.
The virtual filename's extension will match the fenced code block's syntax tag, so for example, ```js code blocks in README.md would match README.md/*.js.
// eslintrc.js for eslint-plugin-markdown v2module.exports={plugins: ["markdown"],overrides: [{// In v2, explicitly apply eslint-plugin-markdown's `markdown`// processor on any Markdown files you want to lint.files: ["**/*.md"],processor: "markdown/markdown"},{// In v2, configuration for fenced code blocks is separate from the// containing Markdown file. Each code block has a virtual filename// appended to the Markdown file's path.files: ["**/*.md/*.js"],// Configuration for fenced code blocks goes with the override for// the code block's virtual filename, for example:parserOptions: {ecmaFeatures: {impliedStrict: true}},rules: {"no-console": "off"}}]};
If you need to precisely mimic the behavior of v1 with the hard-coded Markdown extensions and fenced code block syntaxes, you can use those as glob patterns in overrides[].files:
Use the --ext option to include .js and .md extensions in ESLint's file search:
eslint --ext js,md .
Autofixing
With this plugin, ESLint's --fix option can automatically fix some issues in your Markdown fenced code blocks.
To enable this, pass the --fix flag when you run ESLint:
eslint --fix .
What Gets Linted?
With this plugin, ESLint will lint fenced code blocks in your Markdown documents:
```js// This gets lintedvar answer =6*7;console.log(answer);```
Here is some regular Markdown text that will be ignored.
```js// This also gets linted/* eslint quotes: [2, "double"] */functionhello() {console.log("Hello, world!");}hello();``````jsx// This can be linted too if you add `.jsx` files to `overrides` in ESLint v7// or pass `--ext jsx` in ESLint v6.var div = <div className="jsx"></div>;```
Blocks that don't specify a syntax are ignored:
```This is plain text and doesn't get linted.```
Unless a fenced code block's syntax appears as a file extension in overrides[].files in ESLint v7, it will be ignored.
If using ESLint v6, you must also include the extension with the --ext option.
```pythonprint("This doesn't get linted either.")```
Configuration Comments
The processor will convert HTML comments immediately preceding a code block into JavaScript block comments and insert them at the beginning of the source code that it passes to ESLint.
This permits configuring ESLint via configuration comments while keeping the configuration comments themselves hidden when the markdown is rendered.
Comment bodies are passed through unmodified, so the plugin supports any configuration comments supported by ESLint itself.
This example enables the browser environment, disables the no-alert rule, and configures the quotes rule to prefer single quotes:
Each code block in a file is linted separately, so configuration comments apply only to the code block that immediately follows.
Assuming `no-alert` is enabled in `.eslintrc`, the first code block will have no error from `no-alert`:
<!-- eslint-env browser --><!-- eslint-disable no-alert -->```jsalert("Hello, world!");```
But the next code block will have an error from `no-alert`:
<!-- eslint-env browser -->```jsalert("Hello, world!");```
Skipping Blocks
Sometimes it can be useful to have code blocks marked with js even though they don't contain valid JavaScript syntax, such as commented JSON blobs that need js syntax highlighting.
Standard eslint-disable comments only silence rule reporting, but ESLint still reports any syntax errors it finds.
In cases where a code block should not even be parsed, insert a non-standard <!-- eslint-skip --> comment before the block, and this plugin will hide the following block from ESLint.
Neither rule nor syntax errors will be reported.
There are comments in this JSON, so we use `js` syntax for better
highlighting. Skip the block to prevent warnings about invalid syntax.
<!-- eslint-skip -->```js{// This code block is hidden from ESLint."hello":"world"}``````jsconsole.log("This code block is linted normally.");```
Editor Integrations
VSCode
vscode-eslint has built-in support for the Markdown processor.
In order to see eslint-plugin-markdown work its magic within Markdown code blocks in your Atom editor, you can go to linter-eslint's settings and within "List of scopes to run ESLint on...", add the cursor scope "source.gfm".
However, this reports a problem when viewing Markdown which does not have configuration, so you may wish to use the cursor scope "source.embedded.js", but note that eslint-plugin-markdown configuration comments and skip directives won't work in this context.
Contributing
$ git clone https://github.com/eslint/eslint-plugin-markdown.git
$ cd eslint-plugin-markdown
$ npm install
$ npm test
请发表评论