highlight.js/lib/languages/markdown.js

/*
Language: Markdown
Requires: xml.js
Author: John Crepezzi <john.crepezzi@gmail.com>
Website: https://daringfireball.net/projects/markdown/
Category: common, markup
*/

function markdown(hljs) {
  const regex = hljs.regex;
  const INLINE_HTML = {
    begin: /<\/?[A-Za-z_]/,
    end: '>',
    subLanguage: 'xml',
    relevance: 0
  };
  const HORIZONTAL_RULE = {
    begin: '^[-\\*]{3,}',
    end: '$'
  };
  const CODE = {
    className: 'code',
    variants: [
      // TODO: fix to allow these to work with sublanguage also
      { begin: '(`{3,})[^`](.|\\n)*?\\1`*[ ]*' },
      { begin: '(~{3,})[^~](.|\\n)*?\\1~*[ ]*' },
      // needed to allow markdown as a sublanguage to work
      {
        begin: '```',
        end: '```+[ ]*$'
      },
      {
        begin: '~~~',
        end: '~~~+[ ]*$'
      },
      { begin: '`.+?`' },
      {
        begin: '(?=^( {4}|\\t))',
        // use contains to gobble up multiple lines to allow the block to be whatever size
        // but only have a single open/close tag vs one per line
        contains: [
          {
            begin: '^( {4}|\\t)',
            end: '(\\n)$'
          }
        ],
        relevance: 0
      }
    ]
  };
  const LIST = {
    className: 'bullet',
    begin: '^[ \t]*([*+-]|(\\d+\\.))(?=\\s+)',
    end: '\\s+',
    excludeEnd: true
  };
  const LINK_REFERENCE = {
    begin: /^\[[^\n]+\]:/,
    returnBegin: true,
    contains: [
      {
        className: 'symbol',
        begin: /\[/,
        end: /\]/,
        excludeBegin: true,
        excludeEnd: true
      },
      {
        className: 'link',
        begin: /:\s*/,
        end: /$/,
        excludeBegin: true
      }
    ]
  };
  const URL_SCHEME = /[A-Za-z][A-Za-z0-9+.-]*/;
  const LINK = {
    variants: [
      // too much like nested array access in so many languages
      // to have any real relevance
      {
        begin: /\[.+?\]\[.*?\]/,
        relevance: 0
      },
      // popular internet URLs
      {
        begin: /\[.+?\]\(((data|javascript|mailto):|(?:http|ftp)s?:\/\/).*?\)/,
        relevance: 2
      },
      {
        begin: regex.concat(/\[.+?\]\(/, URL_SCHEME, /:\/\/.*?\)/),
        relevance: 2
      },
      // relative urls
      {
        begin: /\[.+?\]\([./?&#].*?\)/,
        relevance: 1
      },
      // whatever else, lower relevance (might not be a link at all)
      {
        begin: /\[.*?\]\(.*?\)/,
        relevance: 0
      }
    ],
    returnBegin: true,
    contains: [
      {
        // empty strings for alt or link text
        match: /\[(?=\])/ },
      {
        className: 'string',
        relevance: 0,
        begin: '\\[',
        end: '\\]',
        excludeBegin: true,
        returnEnd: true
      },
      {
        className: 'link',
        relevance: 0,
        begin: '\\]\\(',
        end: '\\)',
        excludeBegin: true,
        excludeEnd: true
      },
      {
        className: 'symbol',
        relevance: 0,
        begin: '\\]\\[',
        end: '\\]',
        excludeBegin: true,
        excludeEnd: true
      }
    ]
  };
  const BOLD = {
    className: 'strong',
    contains: [], // defined later
    variants: [
      {
        begin: /_{2}/,
        end: /_{2}/
      },
      {
        begin: /\*{2}/,
        end: /\*{2}/
      }
    ]
  };
  const ITALIC = {
    className: 'emphasis',
    contains: [], // defined later
    variants: [
      {
        begin: /\*(?!\*)/,
        end: /\*/
      },
      {
        begin: /_(?!_)/,
        end: /_/,
        relevance: 0
      }
    ]
  };

  // 3 level deep nesting is not allowed because it would create confusion
  // in cases like `***testing***` because where we don't know if the last
  // `***` is starting a new bold/italic or finishing the last one
  const BOLD_WITHOUT_ITALIC = hljs.inherit(BOLD, { contains: [] });
  const ITALIC_WITHOUT_BOLD = hljs.inherit(ITALIC, { contains: [] });
  BOLD.contains.push(ITALIC_WITHOUT_BOLD);
  ITALIC.contains.push(BOLD_WITHOUT_ITALIC);

  let CONTAINABLE = [
    INLINE_HTML,
    LINK
  ];

  [
    BOLD,
    ITALIC,
    BOLD_WITHOUT_ITALIC,
    ITALIC_WITHOUT_BOLD
  ].forEach(m => {
    m.contains = m.contains.concat(CONTAINABLE);
  });

  CONTAINABLE = CONTAINABLE.concat(BOLD, ITALIC);

  const HEADER = {
    className: 'section',
    variants: [
      {
        begin: '^#{1,6}',
        end: '$',
        contains: CONTAINABLE
      },
      {
        begin: '(?=^.+?\\n[=-]{2,}$)',
        contains: [
          { begin: '^[=-]*$' },
          {
            begin: '^',
            end: "\\n",
            contains: CONTAINABLE
          }
        ]
      }
    ]
  };

  const BLOCKQUOTE = {
    className: 'quote',
    begin: '^>\\s+',
    contains: CONTAINABLE,
    end: '$'
  };

  return {
    name: 'Markdown',
    aliases: [
      'md',
      'mkdown',
      'mkd'
    ],
    contains: [
      HEADER,
      INLINE_HTML,
      LIST,
      BOLD,
      ITALIC,
      BLOCKQUOTE,
      CODE,
      HORIZONTAL_RULE,
      LINK,
      LINK_REFERENCE
    ]
  };
}

module.exports = markdown;