add support for bloomberg newsletters

packages/api/src/utils/bloomberg-handler.ts

@@ -0,0 +1,28 @@
+import { DOMWindow } from 'jsdom'
+
+export class SubstackHandler {
+  name = 'bloomberg'
+
+  shouldPrehandle = (url: URL, dom: DOMWindow): boolean => {
+    const host = this.name + '.com'
+    // check if url ends with bloomberg.com
+    return url.hostname.endsWith(host)
+  }
+
+  prehandle = (url: URL, dom: DOMWindow): Promise<DOMWindow> => {
+    const body = dom.document.querySelector('.wrapper')
+
+    // this removes header
+    body?.querySelector('.sailthru-variables')?.remove()
+    body?.querySelector('.preview-text')?.remove()
+    body?.querySelector('.logo-wrapper')?.remove()
+    body?.querySelector('.by-the-number-wrapper')?.remove()
+    // this removes footer
+    body?.querySelector('.quote-box-wrapper')?.remove()
+    body?.querySelector('.header-wrapper')?.remove()
+    body?.querySelector('.component-wrapper')?.remove()
+    body?.querySelector('.footer')?.remove()
+
+    return Promise.resolve(dom)
+  }
+}

packages/inbound-email-handler/src/newsletter.ts

@@ -6,9 +6,11 @@ const EMAIL_CONFIRMATION_CODE_RECEIVED_TOPIC = 'emailConfirmationCodeReceived'
 const EMAIL_FORWARDING_SENDER_ADDRESSES = [
   'Gmail Team <forwarding-noreply@google.com>',
 ]
-const NEWSLETTER_SENDER_REGEX = '<.+@axios.com>'
+const NEWSLETTER_SENDER_REGEX =
+  '<.+@((axios.com)|(mail.bloombergbusiness.com))>'
 const CONFIRMATION_CODE_PATTERN = '^\\(#\\d+\\)'
 const AXIOS_URL_PATTERN = 'View in browser at <.+>'
+const BLOOMBERG_URL_PATTERN = '<a class="view-in-browser__url".+>'
 
 export const handleConfirmation = async (email: string, subject: string) => {
   console.log('confirmation email')
@@ -114,11 +116,19 @@ export const getNewsletterUrl = (
   }
 
   // axios newsletter url from html
-  const re = new RegExp(AXIOS_URL_PATTERN)
-  const matches = html.match(re)
+  let re = new RegExp(AXIOS_URL_PATTERN)
+  let matches = html.match(re)
   if (matches) {
     const match = matches[0]
     return match.slice(match.indexOf('>') + 1, match.lastIndexOf('<'))
   }
+
+  // bloomberg newsletter url from html
+  re = new RegExp(BLOOMBERG_URL_PATTERN)
+  matches = html.match(re)
+  if (matches) {
+    const match = matches[0]
+    return match.slice(match.indexOf('href=') + 1, match.lastIndexOf('style'))
+  }
   return undefined
 }

packages/inbound-email-handler/test/newsletter.test.ts

@@ -28,6 +28,11 @@ describe('Newsletter email test', () => {
 
       expect(isNewsletter('', from)).to.be.true
     })
+
+    it('should return true when email is from bloomberg', () => {
+      const from = 'From: Bloomberg <noreply@mail.bloombergbusiness.com>'
+      expect(isNewsletter('', from)).to.be.true
+    })
   })
 
   describe('#getNewsletterUrl()', () => {