Bookmarklet to extract a Mastodon thread into JSON

mastodon-thread-to-json.js

javascript:(function () {
	/* ========================================
	Mastodon thread scraper, by Rick Osborne.
	Copy this entire file, from the opening "javascript:" to the last "();".
	Create a new bookmark in your browser, using the copied text as the URL.
	When you are on a Mastodon thread page, activate this bookmark.
	======================================== */
	function findOne(parent, selector, block) {
		if (parent == null) {
			return undefined;
		}
		var els = parent.querySelectorAll(selector);
		if (els.length === 1) {
			if (block != null) {
				block(els[0]);
			}
			return els[0];
		}
		console.log(`Not found: ${selector}`, els);
		return undefined;
	}

	function numberish(s) {
		return ["number", "string"].includes(typeof s) ? Number(s) : undefined;
	}

	function ancestor(el, className) {
		if (el == null || el.classList.contains(className)) {
			return el;
		}
		return ancestor(el.parentElement, className);
	}

	function author(el) {
		var a = {};
		findOne(el, ".display-name .display-name__html", e => a.name = e.innerText);
		findOne(el, ".display-name .display-name__account", e => a.handle = e.innerText);
		findOne(el, ".detailed-status__display-name, .status__display-name", header => {
			author.href = header.href;
			var avatar = findOne(header, ".account__avatar img");
			author.avatar = avatar?.src;
			author.id = avatar?.alt;
		});
		return a;
	}

	function elChildren(el, tagName, block) {
		const children = [...el.children].filter(e => e.tagName);
		if (block != null) {
			block(children);
		}
		return children;
	}

	function elNum(el, tagName, index, block) {
		const child = elChildren(el, tagName).at(index);
		if (block != null && child != null) {
			block(child);
		}
		return child;
	}

	var top = document.querySelector(".detailed-status");
	if (top == null) {
		alert("No detailed status found.  Use this bookmarklet on a detailed status page.");
		return;
	}

	var post = {
		author: author(top)
	};
	findOne(top, ".status__content .status__content__text", e => post.html = e.innerHTML);
	findOne(top, ".detailed-status__meta", (meta) => {
		findOne(meta, ".detailed-status__datetime", dt => {
			post.url = dt.href;
			findOne(dt, "span", e => post.timestamp = e.innerText);
		});
		findOne(meta, ".dropdown-menu__text-button > span", e => post.edited = e.innerText);
		findOne(meta, "i:first-of-type", e => post.visibility = e.title);
		findOne(meta, ".detailed-status__reblogs span span span", e => post.boosts = numberish(e.innerText));
		findOne(meta, ".detailed-status__link span span span", e => post.favorites = numberish(e.innerText));
	});
	var scrollable = ancestor(top, "scrollable");
	elNum(scrollable, "DIV", 1, replyContainer => {
		var replies = elChildren(replyContainer, "DIV");
		post.replies = replies.map(replyEl => {
			var reply = {};
			findOne(replyEl, ".status", statusEl => {
				reply.id = statusEl.getAttribute("data-id");
				findOne(statusEl, ".status__relative-time", e => reply.href = e.href);
				findOne(statusEl, ".status__visibility-icon i", e => reply.visibility = e.title);
				findOne(statusEl, "time", e => reply.timestamp = e.getAttribute("datetime"));
				reply.author = author(statusEl);
				findOne(statusEl, ".status__content .status__content__text", e => reply.html = e.innerHTML);
			});
			return reply;
		});
	});
	console.log(post);
	if (navigator.clipboard) {
		navigator.clipboard.writeText(JSON.stringify(post, null, 2)).then(() => {
			alert("The thread has been copied to your clipboard in JSON format.")
		});
	}
})();