fix: Actualiser le scraper de Melty.

regseb · Dec 9, 2023 · 075d9f4 · 075d9f4
1 parent 2a4627c
commit 075d9f4
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 25 deletions.
diff --git a/src/core/scraper/melty.js b/src/core/scraper/melty.js
@@ -9,10 +9,18 @@
 import { extract as metaExtract } from "../scrapers.js";
 import { matchPattern } from "../tools/matchpattern.js";
 
+/**
+ * L'expression rationnelle pour extraire les données de la vidéo.
+ *
+ * @type {RegExp}
+ */
+const DATA_REGEXP =
+    /init_nouveau_player_dailymotion\("[^"]+","(?<id>[^"]+)", params\);/u;
+
 /**
  * Extrait les informations nécessaires pour lire une vidéo sur Kodi.
  *
- * @param {URL}      url               L'URL d'un article de Melty.
+ * @param {URL}      _url              L'URL d'un article de Melty.
  * @param {Object}   metadata          Les métadonnées de l'URL.
  * @param {Function} metadata.html     La fonction retournant la promesse
  *                                     contenant le document HTML.
@@ -25,19 +33,27 @@ import { matchPattern } from "../tools/matchpattern.js";
  *                                      <em>fichier</em> ou
  *                                      <code>undefined</code>.
  */
-const action = async function (url, metadata, context) {
+const action = async function (_url, metadata, context) {
     if (context.depth) {
         return undefined;
     }
 
     const doc = await metadata.html();
-    for (const meta of doc.querySelectorAll('meta[itemprop="contentUrl"]')) {
-        const file = await metaExtract(new URL(meta.content, url), {
-            ...context,
-            depth: true,
-        });
-        if (undefined !== file) {
-            return file;
+    for (const script of doc.querySelectorAll("script:not([src])")) {
+        const result = DATA_REGEXP.exec(script.text);
+        if (null !== result) {
+            const file = await metaExtract(
+                new URL(
+                    `https://www.dailymotion.com/video/${result.groups.id}`,
+                ),
+                {
+                    ...context,
+                    depth: true,
+                },
+            );
+            if (undefined !== file) {
+                return file;
+            }
         }
     }
     return undefined;

diff --git a/test/unit/core/scraper/melty.js b/test/unit/core/scraper/melty.js
@@ -24,11 +24,13 @@ describe("core/scraper/melty.js", function () {
                 html: () =>
                     Promise.resolve(
                         new DOMParser().parseFromString(
-                            `<html><head>
-                               <meta itemprop="contentUrl"
-                                     content="http://www.dailymotion.com` +
-                                `/embed/video/bar" />
-                             </head></html>`,
+                            `<html><body>
+                               <script>
+                                 jQuery(document).ready(function (){
+                                     init_nouveau_player_dailymotion(` +
+                                `"player_bar","baz", params);
+                               </script>
+                             </body></html>`,
                             "text/html",
                         ),
                     ),
@@ -45,9 +47,9 @@ describe("core/scraper/melty.js", function () {
                 html: () =>
                     Promise.resolve(
                         new DOMParser().parseFromString(
-                            `<html><head>
-                               <meta name="description" content="bar" />
-                             </head></html>`,
+                            `<html><body>
+                               <script src="bar.js"></script>
+                             </body></html>`,
                             "text/html",
                         ),
                     ),
@@ -66,13 +68,17 @@ describe("core/scraper/melty.js", function () {
                 html: () =>
                     Promise.resolve(
                         new DOMParser().parseFromString(
-                            `<html><head>
-                               <meta itemprop="contentUrl"
-                                     content="http://foo.com/bar.html" />
-                               <meta itemprop="contentUrl"
-                                     content="http://www.dailymotion.com` +
-                                `/embed/video/bar" />
-                             </head></html>`,
+                            `<html><body>
+                               <script>
+                                 jQuery(document).ready(function (){
+                                     console.log("bar");
+                               </script>
+                               <script>
+                                 jQuery(document).ready(function (){
+                                     init_nouveau_player_dailymotion(` +
+                                `"player_baz","qux", params);
+                               </script>
+                             </body></html>`,
                             "text/html",
                         ),
                     ),
@@ -82,7 +88,7 @@ describe("core/scraper/melty.js", function () {
             const file = await scraper.extract(url, metadata, context);
             assert.equal(
                 file,
-                "plugin://plugin.video.dailymotion_com/?mode=playVideo&url=bar",
+                "plugin://plugin.video.dailymotion_com/?mode=playVideo&url=qux",
             );
 
             assert.equal(stub.callCount, 1);