From 47d15c2592c1127b0cc25c3017061d9b5cfd2453 Mon Sep 17 00:00:00 2001
From: vectflow <vectflow@gmail.com>
Date: Sat, 29 Jun 2024 00:47:33 -0400
Subject: [PATCH] feat(linkedin): service to scrape linkedin video post for cdn
 urls serving the video

---
 src/modules/processing/services/linkedin.js | 62 +++++++++++++++++++++
 1 file changed, 62 insertions(+)
 create mode 100644 src/modules/processing/services/linkedin.js

diff --git a/src/modules/processing/services/linkedin.js b/src/modules/processing/services/linkedin.js
new file mode 100644
index 00000000..71acefd8
--- /dev/null
+++ b/src/modules/processing/services/linkedin.js
@@ -0,0 +1,62 @@
+import { genericUserAgent } from "../../config.js";
+
+const qualityMatch = {
+    "mp4-640p-30fp-crf28": 640,
+    "mp4-720p-30fp-crf28": 720
+};
+
+export default async function (obj) {
+    const html = await fetch(
+        `https://www.linkedin.com/feed/update/urn:li:activity:${obj.postId}`,
+        { headers: { "user-agent": genericUserAgent } }
+    )
+        .then((res) => res.text())
+        .catch(() => {});
+
+    if (!html) {
+        return { error: "ErrorCouldntFetch" };
+    }
+
+    let data;
+    try {
+        const json = html
+            .split('data-sources="')[1]
+            .split('" data-poster-url="')[0]
+            .replaceAll("&quot;", '"')
+            .replaceAll("&amp;", "&");
+        data = JSON.parse(json);
+    } catch (error) {
+        return { error: "ErrorCouldntFetch" };
+    }
+
+    let fallbackUrl;
+    const quality = obj.quality === "max" || obj.quality >= 720 ? 720 : 640;
+    const filenameBase = `linkedin_${obj.postId}`;
+
+    for (const source of data) {
+        const videoQuality = qualityMatch[source.src.split("/")[6]];
+
+        if (videoQuality === quality) {
+            return {
+                urls: source.src,
+                filename: `${filenameBase}.mp4`,
+                audioFilename: `${filenameBase}_audio`
+            };
+            // will prioritize using known quality over unknown quality if no matching quality
+        } else if (!videoQuality && !fallbackUrl) {
+            fallbackUrl = source.src;
+        } else {
+            fallbackUrl = source.src;
+        }
+    }
+
+    if (fallbackUrl) {
+        return {
+            urls: fallbackUrl,
+            filename: `${filenameBase}.mp4`,
+            audioFilename: `${filenameBase}_audio`
+        };
+    }
+
+    return { error: "ErrorEmptyDownload" };
+}