fix: Actualiser le scraper de TikTok.

regseb · Jan 29, 2024 · 80b0556 · 80b0556
1 parent 9bee8b7
commit 80b0556
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 25 deletions.
diff --git a/src/core/scraper/tiktok.js b/src/core/scraper/tiktok.js
@@ -20,14 +20,15 @@ import { matchPattern } from "../tools/matchpattern.js";
  */
 const action = async function (_url, metadata) {
     const doc = await metadata.html();
-    const script = doc.querySelector("script#SIGI_STATE");
+    const script = doc.querySelector(
+        "script#__UNIVERSAL_DATA_FOR_REHYDRATION__",
+    );
     if (null === script) {
         return undefined;
     }
 
     const json = JSON.parse(script.text);
-    return "ItemModule" in json
-        ? Object.values(json.ItemModule)[0].video.playAddr
-        : undefined;
+    return json["__DEFAULT_SCOPE__"]["webapp.video-detail"]?.itemInfo.itemStruct
+        .video.playAddr;
 };
 export const extract = matchPattern(action, "*://www.tiktok.com/*");
diff --git a/test/unit/core/scraper/lemonde.js b/test/unit/core/scraper/lemonde.js
@@ -207,13 +207,20 @@ describe("core/scraper/lemonde.js", function () {
             const stub = sinon.stub(globalThis, "fetch").resolves(
                 new Response(
                     `<html><body>
-                       <script id="SIGI_STATE">${JSON.stringify({
-                           AppContext: {},
-                           ItemModule: [
-                               {
-                                   video: { playAddr: "http://foo.io/bar.mp4" },
+                       <script id="__UNIVERSAL_DATA_FOR_REHYDRATION__"
+                       >${JSON.stringify({
+                           __DEFAULT_SCOPE__: {
+                               "webapp.video-detail": {
+                                   itemInfo: {
+                                       itemStruct: {
+                                           video: {
+                                               playAddr:
+                                                   "http://foo.io/bar.mp4",
+                                           },
+                                       },
+                                   },
                                },
-                           ],
+                           },
                        })}</script>
                      </body></html>`,
                     { headers: { "Content-Type": "text/html" } },

diff --git a/test/unit/core/scraper/tiktok.js b/test/unit/core/scraper/tiktok.js
@@ -39,8 +39,9 @@ describe("core/scraper/tiktok.js", function () {
                     Promise.resolve(
                         new DOMParser().parseFromString(
                             `<html><body>
-                               <script id="SIGI_STATE">${JSON.stringify({
-                                   AppContext: {},
+                               <script id="__UNIVERSAL_DATA_FOR_REHYDRATION__"
+                               >${JSON.stringify({
+                                   __DEFAULT_SCOPE__: {},
                                })}</script>
                              </body></html>`,
                             "text/html",
@@ -58,19 +59,24 @@ describe("core/scraper/tiktok.js", function () {
                 html: () =>
                     Promise.resolve(
                         new DOMParser().parseFromString(
-                            `
-                    <html><body>
-                      <script id="SIGI_STATE">${JSON.stringify({
-                          AppContext: {},
-                          ItemModule: [
-                              {
-                                  video: {
-                                      playAddr: "https://bar.com/baz.mp4",
-                                  },
-                              },
-                          ],
-                      })}</script>
-                    </body></html>`,
+                            `<html><body>
+                               <script id="__UNIVERSAL_DATA_FOR_REHYDRATION__"
+                               >${JSON.stringify({
+                                   __DEFAULT_SCOPE__: {
+                                       "webapp.video-detail": {
+                                           itemInfo: {
+                                               itemStruct: {
+                                                   video: {
+                                                       playAddr:
+                                                           "https://bar.com" +
+                                                           "/baz.mp4",
+                                                   },
+                                               },
+                                           },
+                                       },
+                                   },
+                               })}</script>
+                             </body></html>`,
                             "text/html",
                         ),
                     ),