improve bot of douyin and xigua

bot/Bilibili.mjs

@@ -87,6 +87,8 @@ class Bilibili extends HeroBot {
         }catch(error) {
             console.error("Error got when request %s via hero: %s", url, error);
             await hero.close();
+            //删除profile文件后重试
+            await this.deleteProfile();
         }
 
         return data;

bot/Douyin.mjs

@@ -9,8 +9,10 @@ class Douyin extends HeroBot {
         let data = {url: url, done: false};
 
         //use iphone
+        if (this.ua == 'mob') {
             configs.userAgent = configs.userAgents.iphone_safari;
             configs.viewport = configs.viewports.mob;
+        }
 
         let options = {
             userAgent: configs.userAgent,     //default mac os
@@ -42,48 +44,54 @@ class Douyin extends HeroBot {
             const tab = await hero.activeTab;
 
             //for mob
+            if (this.ua == 'mob') {
                 await tab.waitForLoad('DomContentLoaded', {timeoutMs: configs.heroTabOptions.timeoutMs});
-
+            }else {
                 //for pc
-            //await tab.waitForLoad('AllContentLoaded', {timeoutMs: configs.heroTabOptions.timeoutMs});
+                await tab.waitForLoad('DomContentLoaded', {timeoutMs: configs.heroTabOptions.timeoutMs});   //AllContentLoaded, DomContentLoaded
-            //await hero.waitForState({
+                await hero.waitForState({
-            //    all(assert) {
+                    all(assert) {
-            //        assert(
+                        assert(
-            //          hero.document.title,
+                          hero.document.title,
-            //          text => text != '',
+                          text => text != '',
-            //        );
+                        );
-            //    }
+                    }
-            //}, {timeoutMs: configs.heroTabOptions.timeoutMs});
+                }, {timeoutMs: configs.heroTabOptions.timeoutMs});
+            }
+
 
             //解析网页HTML数据
             data.title = await hero.document.title;
 
+            if (this.ua == 'mob') {
+                //手机版网页解析
+                const imgElem = await hero.querySelector('.video-container img.poster');
+                if (!imgElem) {
+                    console.error('HTML解析出错，找不到封面图', data);
+                    await hero.close();
+                    //删除profile文件后重试
+                    await this.deleteProfile();
+                    this.ua = 'pc';     //切换到pc模式重试
+                    return false;
+                }
+
+                data.cover = await imgElem.src;
+            }else {
                 //pc版网页解析
-            /*
                 const elems = await hero.document.querySelectorAll('meta');
                 let meta_name = '';
                 for (const elem of elems) {
                     meta_name = await elem.getAttribute('name');
                     if (!meta_name) {continue;}
                     meta_name = meta_name.toLowerCase();
-                //console.log('meta', meta_name);
                     if (meta_name.indexOf('video_cover_image_url') > -1) {
                         data.cover = await elem.getAttribute('content');
                     }else if (meta_name.indexOf('video_title') > -1) {
                         data.title = await elem.getAttribute('content');
                     }
                 }
-            */
-
-            //手机版网页解析
-            const imgElem = await hero.querySelector('.video-container img.poster');
-            if (!imgElem) {
-                console.error('HTML解析出错，找不到封面图', data);
-                await hero.close();
-                return false;
             }
 
-            data.cover = await imgElem.src;
 
             //get cover image's base64 data
             if (typeof(data.cover) != 'undefined' && data.cover) {
@@ -105,6 +113,11 @@ class Douyin extends HeroBot {
         }catch(error) {
             console.error("Error got when request %s via hero: %s", url, error);
             await hero.close();
+
+            //删除profile文件后重试
+            await this.deleteProfile();
+            //切换模式
+            this.ua = this.ua == 'mob' ? 'pc' : 'mob';
         }
 
         return data;

bot/HeroBot.mjs

@@ -1,5 +1,6 @@
 import Hero from '@ulixee/hero';
 import fs from 'node:fs';
+import {unlink} from 'node:fs/promises';
 import path from 'node:path';
 import { fileURLToPath } from 'url';
 import ClientLogPlugin from '../plugin/ClientLogPlugin.mjs';
@@ -20,6 +21,9 @@ class HeroBot {
 
         const __filename = fileURLToPath(import.meta.url);
         this.root = path.dirname(__filename);
+
+        //记录浏览器模式，便于在需要的时候切换
+        this.ua = 'pc';
     }
 
     //返回profile对象
@@ -94,6 +98,25 @@ class HeroBot {
         return true;
     }
 
+    //删除profile
+    async deleteProfile() {
+        if (this.name == '') {return false;}
+
+        const botName = this.name;
+
+        try {
+            const profilePath = path.resolve(this.root, '../tmp/', `profile_${botName}.json`);
+            if (fs.existsSync(profilePath) != false) {
+                return await unlink(profilePath);
+            }
+        }catch(error) {
+            console.error("Error got when delete profile of %s, error detail:\n%s", botName, error);
+            return false;
+        }
+
+        return true;
+    }
+
     //处理name为空的cookie
     fixCookies(profile) {
         let fixedProfile = profile;

bot/Kuaishou.mjs

@@ -68,6 +68,8 @@ class Kuaishou extends HeroBot {
         }catch(error) {
             console.error("Error got when request %s via hero: %s", url, error);
             await hero.close();
+            //删除profile文件后重试
+            await this.deleteProfile();
         };
 
         return data;

bot/Xigua.mjs

@@ -41,7 +41,6 @@ class Xigua extends HeroBot {
 
             //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 = '';
@@ -57,6 +56,15 @@ class Xigua extends HeroBot {
                 }
             }
 
+            //尝试从 <xg-poster class="xgplayer-poster hide"> 再获取一次
+            if (typeof(data.cover) == 'undefined' || !data.cover)  {
+                const imgTag = await tab.querySelector('xg-poster');
+                let backgroundCss = await imgTag.style.backdgroundImage;
+                if (backgroundCss && /url\(.+\)/i.test(backgroundCss)) {
+                    data.cover = backgroundCss.replace('url(', '').replace(')', '').replace('"', '');
+                }
+            }
+
             //get cover image's base64 data
             if (typeof(data.cover) != 'undefined' && data.cover) {
                 data.cover = common.getAbsoluteUrl(data.cover);
@@ -77,6 +85,8 @@ class Xigua extends HeroBot {
         }catch(error) {
             console.error("Error got when request %s via hero: %s", url, error);
             await hero.close();
+            //删除profile文件后重试
+            await this.deleteProfile();
         };
 
         return data;

test/scrap_test.mjs

@@ -61,7 +61,7 @@ import getConfigs from '../config.mjs';
         case 'xigua':
             //西瓜测试
             url = 'https://v.ixigua.com/ieUaqrFN/';       //mobile
-            url = 'https://www.ixigua.com/7248225527335813636';     //pc
+            url = 'https://www.ixigua.com/7343928492197118518';     //pc
 
             configs.heroBotOptions.referrer = url;
             configs.userAgent = configs.userAgents.mac_chrome;