parse bloomberg newsletter

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

@@ -6,7 +6,13 @@ export class SubstackHandler {
   shouldPrehandle = (url: URL, dom: DOMWindow): boolean => {
     const host = this.name + '.com'
     // check if url ends with bloomberg.com
-    return url.hostname.endsWith(host)
+    return (
+      url.hostname.endsWith(host) ||
+      dom.document
+        .querySelector('.logo-image')
+        ?.getAttribute('alt')
+        ?.toLowerCase() === this.name
+    )
   }
 
   prehandle = (url: URL, dom: DOMWindow): Promise<DOMWindow> => {

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

@@ -7,10 +7,11 @@ const EMAIL_FORWARDING_SENDER_ADDRESSES = [
   'Gmail Team <forwarding-noreply@google.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".+>'
+  /<.+@((axios.com)|(mail.bloombergbusiness.com))>/
+const CONFIRMATION_CODE_PATTERN = /^\\(#\\d+\\)/
+const AXIOS_URL_PATTERN = /View in browser at <a.*>(.*)<\/a>/
+const BLOOMBERG_URL_PATTERN =
+  /<a class="view-in-browser__url" href=["']([^"']*)["']/
 
 export const handleConfirmation = async (email: string, subject: string) => {
   console.log('confirmation email')
@@ -116,19 +117,15 @@ export const getNewsletterUrl = (
   }
 
   // axios newsletter url from html
-  let re = new RegExp(AXIOS_URL_PATTERN)
-  let matches = html.match(re)
+  let matches = html.match(AXIOS_URL_PATTERN)
   if (matches) {
-    const match = matches[0]
-    return match.slice(match.indexOf('>') + 1, match.lastIndexOf('<'))
+    return matches[1]
   }
 
   // bloomberg newsletter url from html
-  re = new RegExp(BLOOMBERG_URL_PATTERN)
-  matches = html.match(re)
+  matches = html.match(BLOOMBERG_URL_PATTERN)
   if (matches) {
-    const match = matches[0]
-    return match.slice(match.indexOf('href=') + 1, match.lastIndexOf('style'))
+    return matches[1]
   }
   return undefined
 }

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

@@ -46,10 +46,23 @@ describe('Newsletter email test', () => {
 
     it('returns url when email is from Axios', () => {
       const rawUrl = ''
-      const html = `View in browser at <a>https://axios.com/blog/2019/02/28/the-best-way-to-build-a-web-app</a>`
+      const html = `View in browser at <a>https://axios.com/blog/the-best-way-to-build-a-web-app</a>`
 
       expect(getNewsletterUrl(rawUrl, html)).to.equal(
-        'https://axios.com/blog/2019/02/28/the-best-way-to-build-a-web-app'
+        'https://axios.com/blog/the-best-way-to-build-a-web-app'
+      )
+    })
+
+    it('returns url when email is from Bloomberg', () => {
+      const rawUrl = ''
+      const html = `
+        <a class="view-in-browser__url" href="https://www.bloomberg.com/news/google-is-now-a-partner">
+        View in browser
+        </a>
+      `
+
+      expect(getNewsletterUrl(rawUrl, html)).to.equal(
+        'https://www.bloomberg.com/news/google-is-now-a-partner'
       )
     })
   })