Preparse the newsletter content

packages/api/src/routers/svc/newsletters.ts

@@ -73,8 +73,6 @@ export function newsletterServiceRouter() {
     console.log('create')
 
     const { message, expired } = readPushSubscription(req)
-    console.log('pubsub message:', message, 'expired:', expired)
-
     if (!message) {
       res.status(400).send('Bad Request')
       return

packages/api/src/utils/parser.ts

@@ -466,6 +466,8 @@ export const parseEmailAddress = (from: string): addressparser.EmailAddress => {
 export const fetchFavicon = async (
   url: string
 ): Promise<string | undefined> => {
+  // don't fetch favicon for fake urls
+  if (url.startsWith(FAKE_URL_PREFIX)) return undefined
   try {
     // get the correct url if it's a redirect
     const response = await axios.head(url, { timeout: 5000 })

packages/content-handler/src/index.ts

@@ -72,6 +72,7 @@ const contentHandlers: ContentHandler[] = [
   new BloombergNewsletterHandler(),
   new SubstackHandler(),
   new StackOverflowHandler(),
+  new EnergyWorldHandler(),
 ]
 
 const newsletterHandlers: ContentHandler[] = [

packages/content-handler/src/newsletters/energy-world.ts

@@ -16,4 +16,29 @@ export class EnergyWorldHandler extends ContentHandler {
       input.from === 'ETEnergyworld Latest News<newsletter@etenergyworld.com>'
     )
   }
+
+  shouldPreParse(url: string, dom: Document): boolean {
+    return dom.querySelectorAll('img[src*="etenergyworld.png"]').length > 0
+  }
+
+  async preParse(url: string, dom: Document): Promise<Document> {
+    // get the main content
+    const main = dom.querySelector('table[class="nletter-wrap"]')
+    if (!main) {
+      return Promise.resolve(dom)
+    }
+
+    // create a new dom
+    const newDom = dom.createDocumentFragment()
+
+    // add the content to the new dom
+    main.querySelectorAll('table[class="multi-cols"] tr').forEach((tr) => {
+      const p = dom.createElement('p')
+      p.innerHTML = tr.innerHTML
+      newDom.appendChild(p)
+    })
+    dom.body.replaceChildren(newDom)
+
+    return Promise.resolve(dom)
+  }
 }