markdown-it/lib/rules_core/linkify.mjs

// Replace link-like texts with link nodes.
//
// Currently restricted by `md.validateLink()` to http/https/ftp
//

import { arrayReplaceAt } from '../common/utils.mjs';


function isLinkOpen(str) {
  return /^<a[>\s]/i.test(str);
}
function isLinkClose(str) {
  return /^<\/a\s*>/i.test(str);
}


export default function linkify(state) {
  var i, j, l, tokens, token, currentToken, nodes, ln, text, pos, lastPos,
      level, htmlLinkLevel, url, fullUrl, urlText,
      blockTokens = state.tokens,
      links;

  if (!state.md.options.linkify) { return; }

  for (j = 0, l = blockTokens.length; j < l; j++) {
    if (blockTokens[j].type !== 'inline' ||
        !state.md.linkify.pretest(blockTokens[j].content)) {
      continue;
    }

    tokens = blockTokens[j].children;

    htmlLinkLevel = 0;

    // We scan from the end, to keep position when new tags added.
    // Use reversed logic in links start/end match
    for (i = tokens.length - 1; i >= 0; i--) {
      currentToken = tokens[i];

      // Skip content of markdown links
      if (currentToken.type === 'link_close') {
        i--;
        while (tokens[i].level !== currentToken.level && tokens[i].type !== 'link_open') {
          i--;
        }
        continue;
      }

      // Skip content of html tag links
      if (currentToken.type === 'html_inline') {
        if (isLinkOpen(currentToken.content) && htmlLinkLevel > 0) {
          htmlLinkLevel--;
        }
        if (isLinkClose(currentToken.content)) {
          htmlLinkLevel++;
        }
      }
      if (htmlLinkLevel > 0) { continue; }

      if (currentToken.type === 'text' && state.md.linkify.test(currentToken.content)) {

        text = currentToken.content;
        links = state.md.linkify.match(text);

        // Now split string to nodes
        nodes = [];
        level = currentToken.level;
        lastPos = 0;

        // forbid escape sequence at the start of the string,
        // this avoids http\://example.com/ from being linkified as
        // http:<a href="//example.com/">//example.com/</a>
        if (links.length > 0 &&
            links[0].index === 0 &&
            i > 0 &&
            tokens[i - 1].type === 'text_special') {
          links = links.slice(1);
        }

        for (ln = 0; ln < links.length; ln++) {
          url = links[ln].url;
          fullUrl = state.md.normalizeLink(url);
          if (!state.md.validateLink(fullUrl)) { continue; }

          urlText = links[ln].text;

          // Linkifier might send raw hostnames like "example.com", where url
          // starts with domain name. So we prepend http:// in those cases,
          // and remove it afterwards.
          //
          if (!links[ln].schema) {
            urlText = state.md.normalizeLinkText('http://' + urlText).replace(/^http:\/\//, '');
          } else if (links[ln].schema === 'mailto:' && !/^mailto:/i.test(urlText)) {
            urlText = state.md.normalizeLinkText('mailto:' + urlText).replace(/^mailto:/, '');
          } else {
            urlText = state.md.normalizeLinkText(urlText);
          }

          pos = links[ln].index;

          if (pos > lastPos) {
            token         = new state.Token('text', '', 0);
            token.content = text.slice(lastPos, pos);
            token.level   = level;
            nodes.push(token);
          }

          token         = new state.Token('link_open', 'a', 1);
          token.attrs   = [ [ 'href', fullUrl ] ];
          token.level   = level++;
          token.markup  = 'linkify';
          token.info    = 'auto';
          nodes.push(token);

          token         = new state.Token('text', '', 0);
          token.content = urlText;
          token.level   = level;
          nodes.push(token);

          token         = new state.Token('link_close', 'a', -1);
          token.level   = --level;
          token.markup  = 'linkify';
          token.info    = 'auto';
          nodes.push(token);

          lastPos = links[ln].lastIndex;
        }
        if (lastPos < text.length) {
          token         = new state.Token('text', '', 0);
          token.content = text.slice(lastPos);
          token.level   = level;
          nodes.push(token);
        }

        // replace current node
        blockTokens[j].children = tokens = arrayReplaceAt(tokens, i, nodes);
      }
    }
  }
}