metascraper-x 5.49.8 → 5.49.10

package/README.md

@@ -20,17 +20,11 @@ $ npm install metascraper-x --save
 
 #### options
 
-##### resolveUrls
-
-Type: `boolean`
-
-Set to `true` if you want to resolve `t.co` URLs into the final URL.
-
 ##### resolveUrl
 
 Type: `function`
 
-The function implementation that resolves `t.co` into a the final URL.
+The function implementation that resolves URLs (e.g., `t.co`) into the final URL.
 
 ## License
 

package/package.json

@@ -2,7 +2,7 @@
   "name": "metascraper-x",
   "description": "Metascraper integration with x.com",
   "homepage": "https://github.com/microlinkhq/metascraper/packages/metascraper-x",
-  "version": "5.49.8",
+  "version": "5.49.10",
   "types": "src/index.d.ts",
   "main": "src/index.js",
   "author": {
@@ -39,5 +39,5 @@
     "test": "NODE_PATH=.. TZ=UTC ava --timeout 15s"
   },
   "license": "MIT",
-  "gitHead": "327e130cb3825bf1e05220f19d3ffecb2e7178ed"
+  "gitHead": "49b73e47537a43054208b9c6d80b906fa3890395"
 }

package/src/index.js

@@ -25,31 +25,51 @@ const test = memoizeOne(url =>
   ['twitter.com', 'x.com'].includes(parseUrl(url).domain)
 )
 
-module.exports = ({ resolveUrls = false, resolveUrl = url => url } = {}) => {
+const getAuthorName = memoizeOne(ogTitle =>
+  ogTitle?.split(' on X:')[0].split(' (@')[0].trim()
+)
+
+module.exports = ({ resolveUrl = url => url } = {}) => {
   const rules = {
     author: [
-      toAuthor($ => {
-        const author = $('meta[property="og:title"]').attr('content')
-        return author?.includes(' on X') ? author.split(' on X')[0] : author
-      })
-    ],
-    title: [
-      toTitle(
-        ($, url) => `@${new URL(url).pathname.toString().split('/')[1]} on X`
+      toAuthor($ =>
+        getAuthorName($('meta[property="og:title"]').attr('content'))
       )
     ],
-    url: [
-      toUrl($ =>
-        $('link[rel="canonical"]').attr('href')?.replace('twitter.com', 'x.com')
-      )
+    title: [
+      toTitle(($, url) => {
+        const username = new URL(url).pathname.split('/')[1]
+        const authorName = getAuthorName(
+          $('meta[property="og:title"]').attr('content')
+        )
+        return authorName
+          ? `${authorName} (@${username}) on X`
+          : `@${username} on X`
+      })
     ],
+    url: [toUrl($ => $('link[rel="canonical"]').attr('href'))],
     description: [
       toDescription(async $ => {
         let description =
           $jsonld('mainEntity.description')($) ||
           $('meta[property="og:description"]').attr('content')
 
-        if (!resolveUrls) return description
+        if (!description) {
+          const ogTitle = $('meta[property="og:title"]').attr('content')
+          if (ogTitle?.includes(' on X: "')) {
+            description = ogTitle.split(' on X: "')[1].split('" / X')[0]
+            const urls = getUrls(description)
+            if (urls.length > 1) {
+              const lastUrl = urls[urls.length - 1]
+              if (description.endsWith(lastUrl)) {
+                description = description.replace(lastUrl, '').trim()
+              }
+            }
+          }
+        }
+
+        if (!description) return
+
         const urls = getUrls(description)
         const resolvedUrls = await Promise.all(urls.map(resolveUrl))