From e6ae88b159dc149c8b0ce6b3d2c5a4fcac6e3a11 Mon Sep 17 00:00:00 2001
From: filesite <release@filesite.io>
Date: Mon, 11 Sep 2023 18:24:59 +0800
Subject: [PATCH] bot for bilibili done

---
 bot/Bilibili.mjs    | 68 +++++++++++++++++++++++++++++++++++++++++++++
 test/scrap_test.mjs | 22 +++++++++++++++
 2 files changed, 90 insertions(+)
 create mode 100644 bot/Bilibili.mjs

diff --git a/bot/Bilibili.mjs b/bot/Bilibili.mjs
new file mode 100644
index 0000000..fa5a6b1
--- /dev/null
+++ b/bot/Bilibili.mjs
@@ -0,0 +1,68 @@
+import Hero from '@ulixee/hero';
+import configs from '../config.mjs';
+import HeroBot from './HeroBot.mjs';
+
+class Bilibili extends HeroBot {
+
+  async scrap(url) {
+    let data = {};
+
+    try {
+        let options = {
+            userAgent: configs.userAgent,
+            viewport: configs.viewport
+        };
+
+        if (this.heroServer) {
+            options.connectionToCore = this.heroServer;
+        }
+
+        const profile = await this.init('bilibili');
+        if (profile) {
+            options.userProfile = profile;
+        }
+
+        const hero = new Hero(options);
+        await hero.goto(url, configs.heroBotOptions);
+
+        //等待所有内容加载完成
+        const tab = await hero.activeTab;
+        await tab.waitForLoad('AllContentLoaded', {timeoutMs: configs.heroTabOptions.timeoutMs});
+        await hero.waitForState({
+            all(assert) {
+                assert(
+                  hero.document.title,
+                  text => text != '',
+                );
+            }
+        });
+
+        //解析网页HTML数据
+        data.title = await hero.document.title;
+        //data.url = await hero.url;
+
+        const elems = await hero.document.querySelectorAll('meta');
+        let meta_name = '';
+        for (const elem of elems) {
+          meta_name = await elem.getAttribute('property');
+          if (!meta_name) {continue;}
+          meta_name = meta_name.toLowerCase();
+          //console.log('meta', meta_name);
+          if (meta_name.indexOf('og:image') > -1) {
+            data.cover = await elem.getAttribute('content');
+          }else if (meta_name.indexOf('og:title') > -1) {
+            data.title = await elem.getAttribute('content');
+          }
+        }
+
+        await hero.close();
+    }catch(error) {
+        console.error("Error got when request %s via hero: %s", url, error);
+    }
+
+    return data;
+  }
+
+}
+
+export default Bilibili;
diff --git a/test/scrap_test.mjs b/test/scrap_test.mjs
index 3d8f29c..c011ed1 100644
--- a/test/scrap_test.mjs
+++ b/test/scrap_test.mjs
@@ -1,6 +1,7 @@
 import Douyin from '../bot/Douyin.mjs';
 import Kuaishou from '../bot/Kuaishou.mjs';
 import Xigua from '../bot/Xigua.mjs';
+import Bilibili from '../bot/Bilibili.mjs';
 import configs from '../config.mjs';
 
 (async () => {
@@ -72,6 +73,27 @@ import configs from '../config.mjs';
             console.log("解析结果:\n%s", JSON.stringify(data));
 
             break;
+
+
+        case 'bilibili':
+            //抖音测试
+            url = 'https://www.bilibili.com/video/BV1ep4y1J75y/?share_source=copy_web&vd_source=00bead49a4c2df31bbf3e609d7d95899'; //pc
+            url = 'https://b23.tv/Lo0jIEt';       //mob
+
+            configs.heroTabOptions.timeoutMs = 20000;   //所有内容加载完成超时
+
+            configs.userAgent = configs.userAgents.mac_chrome;
+            configs.viewport = configs.viewports.pc;
+
+            console.log('Hero配置', configs);
+
+            const bilibili = new Bilibili(heroCloudServer);
+            console.log('请求中: %s ...', url);
+            data = await bilibili.scrap(url);
+            console.log("解析结果:\n%s", JSON.stringify(data));
+
+            break;
+
     }
 
     process.exit(0);