micro-mdx-parser/src/utils/find-code-new.js

A tiny parser to convert markdown or html into JSON

// https://regex101.com/r/oPKKoC/1
const REMOVE_CODE_BLOCK_RE = /^(?:[A-Za-z \t]*)?(```(?:[A-Za-z]*)?\n(?:[\s\S]*?)```)([A-Za-z \t]*)*$/gm
// remove inline `code` blocks
const REMOVE_INLINE_CODE_BLOCK = /`[^`\n]*`/g
// https://regex101.com/r/ydaz8U/8
const CODE_BLOCK_RE = /^(>(?: >)*)?([A-Za-z \t]*)([`~]{3,4})([A-Za-z]*)?(.*)?\n\s*([`~]{3,4})?\s*([\s\S]*?)(\6)?[\s]*?(\3)([A-Za-z \t]*)*$/gm
// https://regex101.com/r/ydaz8U/14
const CODE_BLOCK_HTML_RE = /^(>(?: >)*)?([A-Za-z \t]*)([`~]{3,4})([A-Za-z]*)?(.*)?\n\s*([`~]{3,4})?\s*([\s\S]*?)(\6)?[\s]*?(\3)([A-Za-z \t]*)*$|[^`](?:<(pre)\b([^>]*)>*(?:>\s*(<code\b([^>]*)>)?([\s\S]*?)(<\/code>\s*)?<\/pre>))/gm
// https://regex101.com/r/ZEMXNE/3
const MATCH_JS_COMMENT = /^(\/\*[\s\S]*?\*\/|^(\/\/ +[^\n]+(?:\n\/\/*[^\n]+)*)+)\n?/

function findCodeBlocks(block) {
  let matches
  let blocks = []

  while ((matches = CODE_BLOCK_HTML_RE.exec(block)) !== null) {
    if (matches.index === CODE_BLOCK_HTML_RE.lastIndex) {
      CODE_BLOCK_HTML_RE.lastIndex++ // avoid infinite loops with zero-width matches
    }
    const [
      match,
      insideBlockQuote, // 1
      prefix = '', // 2
      openTicks, // 3
      syntax, // 4
      props, // 5
      innerTicksOpen = '', // 6
      _content, // 7
      innerTicksClose = '', // 8
      closeTicks, // 9
      postFix, // 10
      preTag, //  11
      preTagAttrs, // 12
      codeTag, // 13
      codeTagAttrs, // 14
      preCodeContent, // 15
    ] = matches

    //* // debug
    // console.log('match')
    // console.log(match)
    // console.log('preCodeContent', preCodeContent)
    // console.log('insideBlockQuote', insideBlockQuote)
    // console.log('prefix', prefix)
    // console.log('openTicks', openTicks)
    // console.log('syntax', syntax)
    // console.log('props', props)
    // console.log('innerTicksOpen', innerTicksOpen)
    // console.log('_content>>', _content)
    // console.log('innerTicksClose', innerTicksClose)
    // console.log('closeTicks', closeTicks)
    // console.log('postFix', postFix)
    // console.log('insideBlockQuote', insideBlockQuote)
    // console.log('prefix', `"${prefix}"`)
    /** */

    // let codeContent = (preCodeContent) ? preCodeContent.trim() : _content
    // if (insideBlockQuote) {
    //   // replace leading block quote > > arrows
    //   codeContent = _content.replace(new RegExp(`^${insideBlockQuote}`, 'gm'), '')
    //   if (prefix) {
    //     // trim leading spaces from codeContent if prefix exists
    //     codeContent = codeContent.replace(new RegExp(`^${prefix}`, ''), '')
    //   }
    // }

    // let finalCode = prefix + innerTicksOpen + codeContent + innerTicksClose

    // const codeBlock = {}
    // let hasError = false

    // if (includePositions) {
    //   codeBlock.index = matches.index
    // }

    // if (props) {
    //   codeBlock.propsRaw = props
    //   codeBlock.props = parse(props)
    // }


    blocks.push(match)
  }

  return blocks
}

function removeCode(text = '') {
  return text
    .replace(CODE_BLOCK_HTML_RE, '')
    .replace(REMOVE_INLINE_CODE_BLOCK, '')
}

module.exports = {
  findCodeBlocks,
  removeCode,
  CODE_BLOCK_RE,
  CODE_BLOCK_HTML_RE
  // REMOVE_CODE_BLOCK_RE
}