Fixes embed links for twitter and tiktok (#74)

src/lib/extractHtmlMeta.ts

@@ -1,15 +1,18 @@
+import {extractTwitterMeta} from './extractTwitterMeta'
 import {extractYoutubeMeta} from './extractYoutubeMeta'
 
 interface ExtractHtmlMetaInput {
   html: string
   hostname?: string
+  pathname?: string
 }
 
 export const extractHtmlMeta = ({
   html,
   hostname,
+  pathname,
 }: ExtractHtmlMetaInput): Record<string, string> => {
-  const htmlTitleRegex = /<title>([^<]+)<\/title>/i
+  const htmlTitleRegex = /<title.*>([^<]+)<\/title>/i
 
   let res: Record<string, string> = {}
 
@@ -56,9 +59,12 @@ export const extractHtmlMeta = ({
 
   const isYoutubeUrl =
     hostname?.includes('youtube.') || hostname?.includes('youtu.be')
+  const isTwitterUrl = hostname?.includes('twitter.')
+  // Workaround for some websites not having a title or description in the meta tags in the initial serve
   if (isYoutubeUrl) {
-    // Workaround for Youtube not having a title in the meta tags
     res = {...res, ...extractYoutubeMeta(html)}
+  } else if (isTwitterUrl) {
+    res = {...extractTwitterMeta({pathname})}
   }
 
   return res

src/lib/extractTwitterMeta.ts

@@ -0,0 +1,20 @@
+export const extractTwitterMeta = ({
+  pathname,
+}: {
+  pathname: string
+}): Record<string, string> => {
+  const res = {title: 'Twitter'}
+  const parsedPathname = pathname.split('/')
+  if (parsedPathname.length <= 1 || parsedPathname[1].length <= 1) {
+    // Excluding one letter usernames as they're reserved by twitter for things like cases like twitter.com/i/articles/follows/-1675653703
+    return res
+  }
+  const username = parsedPathname?.[1]
+  const isUserProfile = parsedPathname?.length === 2
+
+  res.title = isUserProfile
+    ? `@${username} on Twitter`
+    : `Tweet by @${username}`
+
+  return res
+}

src/lib/link-meta.ts

@@ -63,6 +63,7 @@ export async function getLinkMeta(
     const httpResMeta = extractHtmlMeta({
       html: httpResBody,
       hostname: urlp?.hostname,
+      pathname: urlp?.pathname,
     })
     meta.title = httpResMeta.title ? he.decode(httpResMeta.title) : undefined
     meta.description = httpResMeta.description