extract text from https://ar5iv.labs.arxiv.org/

extractTextFromArxiv.user.js

// ==UserScript==
// @name        arxiv论文转markdown拷贝到剪贴板
// @description 方便粘贴到 chatgpt 进行问答
// @namespace   github.com/taowen
// @match       https://ar5iv.labs.arxiv.org/*
// @match       https://browse.arxiv.org/html/*
// @match       https://arxiv.org/html/*
// @version     1.0.1
// @author      taowen
// @license     MIT
// @grant       GM.registerMenuCommand
// @grant       GM_setClipboard
// ==/UserScript==
GM.registerMenuCommand("复制论文到剪贴板", () => {
  const selection = document.getSelection();
  const targetNode = selection.baseNode ? getCommonAncestor(selection.baseNode, selection.extentNode): document.body;
  const text = extractText(targetNode);
  GM_setClipboard (text);
  alert('copied ' + text.length + ' characters');
});
function getCommonAncestor(node1, node2) {
    var method = "contains" in node1 ? "contains" : "compareDocumentPosition",
        test   = method === "contains" ? 1 : 0x10;
    while (node1 = node1.parentNode) {
        if ((node1[method](node2) & test) === test)
            return node1;
    }
    return null;
}
function extractText(node, parts) {
  let returnString = false;
  if (parts === undefined) { parts = []; returnString = true; }
  if (node.wholeText !== undefined) { parts.push(node.wholeText.trim()); return; }
  const tagName = (node.tagName || '').toLowerCase();
  if (tagName === 'button') { return; }
  if (tagName === 'h1') { parts.push('\n# ' + node.innerText + '\n'); return; }
  if (tagName === 'h2') { parts.push('\n## ' + node.innerText + '\n'); return; }
  if (tagName === 'h3') { parts.push('\n### ' + node.innerText + '\n'); return; }
  if (tagName === 'h4') { parts.push('\n#### ' + node.innerText + '\n'); return; }
  if (tagName === 'h5') { parts.push('\n##### ' + node.innerText + '\n'); return; }
  if (tagName === 'h6') { parts.push('\n###### ' + node.innerText + '\n'); return; }
  if (tagName === 'math') { parts.push('$' + node.attributes.alttext.value + '$'); return; }
  const cssClass = node.attributes && node.attributes.class && node.attributes.class.value;
  if (cssClass === 'ltx_note_outer') {
    text = '';
    for(const child of node.childNodes) {
      text += extractText(child)
    }
    parts.push(`\n> ${text.replace('\n', '')}\n\n`);
  } else {
    for(const child of node.childNodes) {
      extractText(child, parts)
    }
  }
  if (tagName === 'p' || tagName === 'li' || cssClass === 'ltx_listingline') {
    parts.push('\n')
  }
  if (returnString) {
    return parts.join(' ')
  }
}