feat(route): douyin hashtag (#9905)

2025-12-09 23:00:48 +08:00 · 2022-06-07 00:41:20 +11:00
parent 3aa43220fa
commit daaa31771b
4 changed files with 151 additions and 11 deletions
--- a/docs/social-media.md
+++ b/docs/social-media.md
@@ -850,16 +850,6 @@ YouTube 官方亦有提供频道 RSS，形如 <https://www.youtube.com/feeds/vid

 ## 抖音

-### 博主
-
-<Route author="Max-Tortoise Rongronggg9" example="/douyin/user/MS4wLjABAAAARcAHmmF9mAG3JEixq_CdP72APhBlGlLVbN-1eBcPqao" path="/douyin/user/:uid/:routeParams?" :paramsDesc="['uid，可在用户页面 URL 中找到', '额外参数，query string 格式，请参阅下面的表格']" anticrawler="1" radar="1" rssbud="1" puppeteer="1">
-
-| 键        | 含义                                  | 值                      | 默认值     |
-| -------- | ----------------------------------- | ---------------------- | ------- |
-| `embed`  | 是否启用内嵌视频                            | `0`/`1`/`true`/`false` | `false` |
-| `iframe` | 是否启用 iframe 变通解决方案，仅在内嵌视频开启时有效，详见下文 | `0`/`1`/`true`/`false` | `false` |
-| `relay`  | 视频反代服务的 URL，仅在内嵌视频开启时有效，详见下文        |                        |         |
-
 ::: warning 注意

 反爬严格，需要启用 puppeteer。\
@@ -875,7 +865,21 @@ YouTube 官方亦有提供频道 RSS，形如 <https://www.youtube.com/feeds/vid

 :::

-</Route>
+额外参数
+
+| 键        | 含义                                  | 值                      | 默认值     |
+| -------- | ----------------------------------- | ---------------------- | ------- |
+| `embed`  | 是否启用内嵌视频                            | `0`/`1`/`true`/`false` | `false` |
+| `iframe` | 是否启用 iframe 变通解决方案，仅在内嵌视频开启时有效，详见下文 | `0`/`1`/`true`/`false` | `false` |
+| `relay`  | 视频反代服务的 URL，仅在内嵌视频开启时有效，详见下文        |                        |         |
+
+### 博主
+
+<Route author="Max-Tortoise Rongronggg9" example="/douyin/user/MS4wLjABAAAARcAHmmF9mAG3JEixq_CdP72APhBlGlLVbN-1eBcPqao" path="/douyin/user/:uid/:routeParams?" :paramsDesc="['uid，可在用户页面 URL 中找到', '额外参数，query string 格式，请参阅上面的表格']" anticrawler="1" radar="1" rssbud="1" puppeteer="1" />
+
+### 标签
+
+<Route author="TonyRL" example="/douyin/hashtag/1592824105719812" path="/douyin/hashtag/:cid/:routeParams?" :paramsDesc="['标签 ID，可在标签页面 URL 中找到', '额外参数，query string 格式，请参阅上面的表格']" anticrawler="1" radar="1" rssbud="1" puppeteer="1" />

 ## 豆瓣

--- a/lib/v2/douyin/hashtag.js
+++ b/lib/v2/douyin/hashtag.js
@@ -0,0 +1,129 @@
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+const config = require('@/config').value;
+const { fallback, queryToBoolean } = require('@/utils/readable-social');
+
+const templates = {
+    desc: path.join(__dirname, 'templates/desc.art'),
+    cover: path.join(__dirname, 'templates/cover.art'),
+    embed: path.join(__dirname, 'templates/embed.art'),
+    iframe: path.join(__dirname, 'templates/iframe.art'),
+};
+
+const resolveUrl = (url, tls = true, forceResolve = false) => {
+    if (!url) {
+        return '';
+    }
+    if (url.startsWith('//')) {
+        return (tls ? 'https:' : 'http:') + url;
+    }
+    if (forceResolve && !url.match(/^https?:\/\//)) {
+        return (tls ? 'https://' : 'http://') + url;
+    }
+    return url;
+};
+
+const proxyVideo = (url, proxy) => {
+    if (!(url && proxy)) {
+        return url + '';
+    }
+    if (proxy.includes('?')) {
+        if (!proxy.endsWith('=')) {
+            proxy += '=';
+        }
+        return proxy + encodeURIComponent(url);
+    } else {
+        if (!proxy.endsWith('/')) {
+            proxy += '/';
+        }
+        return proxy + url;
+    }
+};
+
+const getOriginAvatar = (url) =>
+    resolveUrl(url)
+        .replace(/^(.*\.douyinpic\.com\/).*(\/aweme-avatar\/)([^?]*)(\?.*)?$/, '$1origin$2$3')
+        .replace(/~\w+_\d+x\d+/g, '');
+
+module.exports = async (ctx) => {
+    const { cid } = ctx.params;
+    if (isNaN(cid)) {
+        throw Error('Invalid tag ID. Tag ID should be a number.');
+    }
+    const routeParams = Object.fromEntries(new URLSearchParams(ctx.params.routeParams));
+    const embed = fallback(undefined, queryToBoolean(routeParams.embed), false); // embed video
+    const iframe = fallback(undefined, queryToBoolean(routeParams.iframe), false); // embed video in iframe
+    const relay = resolveUrl(routeParams.relay, true, true); // embed video behind a reverse proxy
+
+    const tagUrl = `https://www.douyin.com/hashtag/${cid}`;
+
+    const tagData = await ctx.cache.tryGet(
+        `douyin:hashtag:${cid}`,
+        async () => {
+            const browser = await require('@/utils/puppeteer')();
+            const page = await browser.newPage();
+            await page.goto(tagUrl, {
+                waitUntil: 'domcontentloaded',
+            });
+            const html = await page.evaluate(() => document.documentElement.innerHTML);
+            await browser.close();
+
+            const $ = cheerio.load(html);
+            const renderData = JSON.parse(decodeURIComponent($('script#RENDER_DATA').text()));
+            const dataKey = Object.keys(renderData).find((key) => renderData[key].topicDetail && renderData[key].defaultData);
+            return renderData[dataKey];
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    const tagInfo = tagData.topicDetail;
+    const tagName = tagInfo.chaName;
+    const userAvatar = getOriginAvatar(tagInfo.hashtagProfile);
+
+    const posts = tagData.defaultData;
+    const items = posts.map((post) => {
+        // parse video
+        let videoList = post.video && post.video.bitRateList && post.video.bitRateList.map((item) => resolveUrl(item.playApi));
+        if (relay) {
+            videoList = videoList.map((item) => proxyVideo(item, relay));
+        }
+        let duration = post.video && post.video.duration;
+        duration = duration && duration / 1000;
+        let img;
+        // if (!embed) {
+        //     img = post.video && post.video.dynamicCover; // dynamic cover (webp)
+        // }
+        img =
+            img ||
+            (post.video &&
+                ((post.video.coverUrlList && post.video.coverUrlList[post.video.coverUrlList.length - 1]) || // HD
+                    post.video.originCover || // LD
+                    post.video.cover)); // even more LD
+        img = img && resolveUrl(img);
+
+        // render description
+        const desc = post.desc && post.desc.replace(/\n/g, '<br>');
+        let media = art(embed && videoList ? templates.embed : templates.cover, { img, videoList, duration });
+        media = embed && videoList && iframe ? art(templates.iframe, { content: media }) : media; // warp in iframe
+        const description = art(templates.desc, { desc, media });
+
+        return {
+            title: post.desc,
+            description,
+            link: `https://www.douyin.com/video/${post.awemeId}`,
+            pubDate: parseDate(post.createTime * 1000),
+            category: post.textExtra.map((extra) => extra.hashtagName),
+        };
+    });
+
+    ctx.state.data = {
+        title: tagName,
+        description: `${tagInfo.viewCount} 次播放`,
+        image: userAvatar,
+        link: tagUrl,
+        item: items,
+    };
+};
--- a/lib/v2/douyin/radar.js
+++ b/lib/v2/douyin/radar.js
@@ -2,6 +2,12 @@ module.exports = {
    'douyin.com': {
        _name: '抖音',
        '.': [
+            {
+                title: '标签',
+                docs: 'https://docs.rsshub.app/social-media.html#dou-yin',
+                source: '/hashtag/:cid',
+                target: '/douyin/hashtag/:cid',
+            },
            {
                title: '博主',
                docs: 'https://docs.rsshub.app/social-media.html#dou-yin',
--- a/lib/v2/douyin/router.js
+++ b/lib/v2/douyin/router.js
@@ -1,3 +1,4 @@
 module.exports = function (router) {
+    router.get('/hashtag/:cid/:routeParams?', require('./hashtag'));
    router.get('/user/:uid/:routeParams?', require('./user'));
 };