const errors = require('@tryghost/errors'); const logging = require('@tryghost/logging'); module.exports = class MentionSendingService { #discoveryService; #externalRequest; #getSiteUrl; #getPostUrl; #isEnabled; #jobService; constructor({discoveryService, externalRequest, getSiteUrl, getPostUrl, isEnabled, jobService}) { this.#discoveryService = discoveryService; this.#externalRequest = externalRequest; this.#getSiteUrl = getSiteUrl; this.#getPostUrl = getPostUrl; this.#isEnabled = isEnabled; this.#jobService = jobService; } get siteUrl() { try { return new URL(this.#getSiteUrl()); } catch (e) { return null; } } /** * Listen for new and edited published posts and automatically send webmentions. Unpublished posts should send mentions * so the receiver can discover a 404 response and remove the mentions. * @param {*} events */ listen(events) { events.on('post.published', this.sendForPost.bind(this)); events.on('post.published.edited', this.sendForPost.bind(this)); events.on('post.unpublished', this.sendForPost.bind(this)); } async sendForPost(post, options) { // NOTE: this is not ideal and shouldn't really be handled within the package... // for now we don't want to evaluate mentions when importing data (at least needs queueing set up) // we do not want to evaluate mentions with fixture (internal) data, e.g. generating posts // TODO: real solution is likely suppressing event emission when building fixture data if (options && (options.importing || options.context.internal)) { return; } try { if (!this.#isEnabled()) { return; } // TODO: we need to check old url and send webmentions in case the url changed of a post if (post.get('status') === post.previous('status') && post.get('html') === post.previous('html')) { // Not changed return; } if (post.get('status') !== 'published' && post.previous('status') !== 'published') { // Post should be or should have been published return; } // make sure we have something to parse before we create a job let html = post.get('html'); let previousHtml = post.previous('status') === 'published' ? post.previous('html') : null; if (html || previousHtml) { await this.#jobService.addJob('sendWebmentions', async () => { await this.sendAll({ url: new URL(this.#getPostUrl(post)), html: html, previousHtml: previousHtml }); }); } } catch (e) { logging.error('Error in webmention sending service post update event handler:'); logging.error(e); } } async send({source, target, endpoint}) { logging.info('[Webmention] Sending webmention from ' + source.href + ' to ' + target.href + ' via ' + endpoint.href); // default content type is application/x-www-form-encoded which is what we need for the webmentions spec const response = await this.#externalRequest.post(endpoint.href, { form: { source: source.href, target: target.href, source_is_ghost: true }, throwHttpErrors: false, maxRedirects: 10, followRedirect: true, timeout: 10000 }); if (response.statusCode >= 200 && response.statusCode < 300) { return; } throw new errors.BadRequestError({ message: 'Webmention sending failed with status code ' + response.statusCode, statusCode: response.statusCode }); } /** * Send a webmention call for the links in a resource. * @param {object} resource * @param {URL} resource.url * @param {string} resource.html * @param {string|null} [resource.previousHtml] */ async sendAll(resource) { const links = resource.html ? this.getLinks(resource.html) : []; if (resource.previousHtml) { // We also need to send webmentions for removed links const oldLinks = this.getLinks(resource.previousHtml); for (const link of oldLinks) { if (!links.find(l => l.href === link.href)) { links.push(link); } } } if (links.length) { logging.info('[Webmention] Sending all webmentions for ' + resource.url.href); } for (const target of links) { const endpoint = await this.#discoveryService.getEndpoint(target); if (endpoint) { // Send webmention call try { await this.send({source: resource.url, target, endpoint}); } catch (e) { logging.error('[Webmention] Failed sending via ' + endpoint.href + ': ' + e.message); } } } } /** * @private * Get all external links in a HTML document. * Excludes the site's own domain. * @param {string} html * @returns {URL[]} */ getLinks(html) { const cheerio = require('cheerio'); const $ = cheerio.load(html); const urls = []; const siteUrl = this.siteUrl; for (const el of $('a').toArray()) { const href = $(el).attr('href'); if (href) { let url; try { url = new URL(href); if (siteUrl && url.hostname === siteUrl.hostname) { // Ignore links to the site's own domain continue; } if (['http:', 'https:'].includes(url.protocol) && !urls.find(u => u.href === url.href)) { // Ignore duplicate URLs urls.push(url); } } catch (e) { // Ignore invalid URLs } } } return urls; } };