包详细信息

fenceparser

frencojobs29.7kMIT2.2.0

A well-tested parser for parsing metadata out of fenced code blocks in Markdown

markdown

自述文件

A tiny, well-tested parser for parsing metadata out of fenced code blocks in Markdown.


Overview ・ npm bundle size Codecov

Assuming you have this code fence in your Markdown,

 ```ts twoslash {1-3, 5} title="Hello, World"

Using remark will yield two information about that code block, lang and meta like this.

{
  "lang": "ts",
  "meta": "twoslash {1-3, 5} title=\"Hello, World\""
}

Use fenceparser to parse the meta string out to a useful object.

import parse from 'fenceparser'

console.log(parse(meta))

// {
//   twoslash: true,
//   highlight: { '1-3': true, '5': true },
//   title: 'Hello, World'
// }

The parser won't intentionally handle parsing the language part since it is usually handled by the Markdown parsers.

But if you want to allow loose syntax grammars such as ts{1-3, 5} as well as ts {1-3, 5} which is used by gatsby-remark-vscode as an example, remark won't parse the language correctly.

{
  "lang": "ts{1-3,", // because remark uses space to split
  "meta": "5}"
}

In these cases, you can use the the library's lex function to get a properly tokenized array. You may then take out the first element as lang. For example,

import {lex, parse} from 'fenceparser'
// Notice this ^ parse is not the same the default export function

const full = [node.lang, node.meta].join(' ') // Join them back

const tokens = lex(full)
const lang = tokens.shift() // ts
const meta = parse(tokens) // { highlight: {'1-3': true, '5': true} }

Syntax

The syntax grammar is loosely based on techniques used by various syntax-highlighters. Rules are such that

  • Valid HTML attributes can be used, attribute, data-attribute, etc.
  • Just like in HTML, top-level attribute names are case insensitive.
  • Attributes without values are assigned as true
  • Attribute values can be single or double quoted strings, int/float numbers, booleans, objects or arrays
  • Non-quoted strings are valid as long as they are not separated by a whitespace or a line-break, attr=--theme-color
  • Objects can accept valid attributes as children, or valid attributes with value assigned by : keyword, {1-3, 5, ids: {7}}
  • Arrays are just like JavaScript's arrays
  • Objects without attribute keys {1-3} {7} are merged and assigned to the highlight object
  • No trailing commas

Acknowledgements

  1. This project is made initially to use with Twoslash.
  2. The initial implementations of lexer and parser are based on the examples from the book Crafting Interpreters.

更新日志

fenceparser

2.2.0

Minor Changes

  • 4c3a51d: Optimize the library heavily reducing size from 5kb to less than 1kb, while still keeping all the tests passed. Minor refactors are also added, with a new error class FenceparserError exported.

2.1.1

Patch Changes

  • 2955219: Add build step to ci fixing missing dist directory error.

2.1.0

Minor Changes

  • 28d431d: Add support for CJS and ESM. We now use changeset to manage CHANGELOG and publish.

2.0.0

  • Remove TSDX
  • Opt-in to ES Modules

1.1.1

  • Add minimalistic as default prettier config and format code

1.1.0

  • Return Record<string, VALUE> instead of Map<string, VALUE>

1.0.2

  • Add package information such as description and keywords

1.0.1

  • Export lex and parse functions individually

1.0.0

  • Initial release

0.1.0

  • Prelease