Merge pull request #128 from wuaishare/master

修复快手接口

.env.example

@@ -15,6 +15,7 @@ DISALLOW_ROBOT=true
 REDIS_HOST="127.0.0.1"
 REDIS_PORT=6379
 REDIS_PASSWORD=""
+REDIS_DB=0
 
 # 缓存时长（ 秒 ）
 CACHE_TTL=3600
@@ -27,3 +28,6 @@ USE_LOG_FILE=true
 
 # RSS Mode
 RSS_MODE=false
+
+# Weibo
+FILTER_WEIBO_ADVERTISEMENT=false

src/config.ts

@@ -15,7 +15,9 @@ export type Config = {
   REDIS_HOST: string;
   REDIS_PORT: number;
   REDIS_PASSWORD: string;
+  REDIS_DB: number;
   ZHIHU_COOKIE: string;
+  FILTER_WEIBO_ADVERTISEMENT: boolean;
 };
 
 // 验证并提取环境变量
@@ -52,5 +54,7 @@ export const config: Config = {
   REDIS_HOST: getEnvVariable("REDIS_HOST") || "127.0.0.1",
   REDIS_PORT: getNumericEnvVariable("REDIS_PORT", 6379),
   REDIS_PASSWORD: getEnvVariable("REDIS_PASSWORD") || "",
+  REDIS_DB:  getNumericEnvVariable("REDIS_DB", 0),
   ZHIHU_COOKIE: getEnvVariable("ZHIHU_COOKIE") || "",
+  FILTER_WEIBO_ADVERTISEMENT: getBooleanEnvVariable("FILTER_WEIBO_ADVERTISEMENT", false),
 };

src/router.types.d.ts

@@ -89,8 +89,10 @@ export type RouterType = {
     word_scheme: string;
     note: string;
     flag_desc: string;
-    num: number;
+    // num: number;
+    desc_extr: number;
     onboard_time: number;
+    pic: string;
   };
   zhihu: {
     target: {

src/routes/huxiu.ts

@@ -1,7 +1,7 @@
 import type { RouterData } from "../types.js";
 import type { RouterType } from "../router.types.js";
-import { get } from "../utils/getData.js";
 import { getTime } from "../utils/getTime.js";
+import axios from "axios";
 
 export const handleRoute = async (_: undefined, noCache: boolean) => {
   const listData = await getList(noCache);
@@ -16,36 +16,39 @@ export const handleRoute = async (_: undefined, noCache: boolean) => {
   return routeData;
 };
 
-// 标题处理
-const titleProcessing = (text: string) => {
-  const paragraphs = text.split("<br><br>");
-  const title = paragraphs.shift()?.replace(/。$/, "");
-  const intro = paragraphs.join("<br><br>");
-  return { title, intro };
-};
-
 const getList = async (noCache: boolean) => {
-  const url = `https://www.huxiu.com/moment/`;
-  const result = await get({
-    url,
-    noCache,
+  // PC 端接口
+  const url = `https://moment-api.huxiu.com/web-v3/moment/feed?platform=www`;
+  const res = await axios.get(url, {
+    headers: {
+      "User-Agent": "Mozilla/5.0",
+      Referer: "https://www.huxiu.com/moment/",
+    },
+    timeout: 10000,
   });
-  // 正则查找
-  const pattern =
-    /<script>[\s\S]*?window\.__INITIAL_STATE__\s*=\s*(\{[\s\S]*?\});[\s\S]*?<\/script>/;
-  const matchResult = result.data.match(pattern);
-  const jsonObject = JSON.parse(matchResult[1]).moment.momentList.moment_list.datalist;
+  const list: RouterType["huxiu"][] = res.data?.data?.moment_list?.datalist || [];
   return {
-    ...result,
-    data: jsonObject.map((v: RouterType["huxiu"]) => ({
-      id: v.object_id,
-      title: titleProcessing(v.content).title,
-      desc: titleProcessing(v.content).intro,
-      author: v.user_info.username,
+    fromCache: false,
+    updateTime: new Date().toISOString(),
+    data: list.map((v: RouterType["huxiu"]) => {
+      const content = (v.content || "").replace(/<br\s*\/?>/gi, "\n");
+      const [titleLine, ...rest] = content
+        .split("\n")
+        .map((s) => s.trim())
+        .filter(Boolean);
+      const title = titleLine?.replace(/。$/, "") || "";
+      const intro = rest.join("\n");
+      const momentId = v.object_id;
+      return {
+        id: momentId,
+        title,
+        desc: intro,
+        author: v.user_info?.username || "",
         timestamp: getTime(v.publish_time),
-      hot: undefined,
-      url: v.url || `https://www.huxiu.com/moment/${v.object_id}.html`,
-      mobileUrl: v.url || `https://m.huxiu.com/moment/${v.object_id}.html`,
-    })),
+        hot: v.count_info?.agree_num,
+        url: `https://www.huxiu.com/moment/${momentId}.html`,
+        mobileUrl: `https://m.huxiu.com/moment/${momentId}.html`,
+      };
+    }),
   };
 };

src/routes/kuaishou.ts

@@ -4,6 +4,8 @@ import { get } from "../utils/getData.js";
 import { parseChineseNumber } from "../utils/getNum.js";
 import UserAgent from "user-agents";
 
+const APOLLO_STATE_PREFIX = "window.__APOLLO_STATE__=";
+
 export const handleRoute = async (_: undefined, noCache: boolean) => {
   const listData = await getList(noCache);
   const routeData: RouterData = {
@@ -32,21 +34,52 @@ const getList = async (noCache: boolean) => {
   });
   const listData: ListItem[] = [];
   // 获取主要内容
-  const pattern = /window.__APOLLO_STATE__=(.*);\(function\(\)/s;
-  const matchResult = result.data?.match(pattern);
-  const jsonObject = JSON.parse(matchResult[1])["defaultClient"];
+  const html = result.data || "";
+  const start = html.indexOf(APOLLO_STATE_PREFIX);
+  if (start === -1) {
+    throw new Error("快手页面结构变更，未找到 APOLLO_STATE");
+  }
+  const scriptSlice = html.slice(start + APOLLO_STATE_PREFIX.length);
+  const sentinelA = scriptSlice.indexOf(";(function(");
+  const sentinelB = scriptSlice.indexOf("</script>");
+  const cutIndex =
+    sentinelA !== -1 && sentinelB !== -1 ? Math.min(sentinelA, sentinelB) : Math.max(sentinelA, sentinelB);
+  if (cutIndex === -1) {
+    throw new Error("快手页面结构变更，未找到 APOLLO_STATE 结束标记");
+  }
+  const raw = scriptSlice.slice(0, cutIndex).trim().replace(/;$/, "");
+  let jsonObject;
+  try {
+    // 快手返回的 JSON 末尾常带 undefined/null，需要截断到最后一个 '}' 出现
+    const lastBrace = raw.lastIndexOf("}");
+    const cleanRaw = lastBrace !== -1 ? raw.slice(0, lastBrace + 1) : raw;
+    jsonObject = JSON.parse(cleanRaw)["defaultClient"];
+  } catch (err) {
+    const msg =
+      err instanceof Error
+        ? `${err.message} | snippet=${raw.slice(0, 200)}...`
+        : "未知错误";
+    throw new Error(`快手数据解析失败: ${msg}`);
+  }
   // 获取所有分类
-  const allItems = jsonObject['$ROOT_QUERY.visionHotRank({"page":"home"})']["items"];
+  const allItems =
+    jsonObject['$ROOT_QUERY.visionHotRank({"page":"home"})']?.items ||
+    jsonObject['$ROOT_QUERY.visionHotRank({"page":"home","platform":"web"})']
+      ?.items ||
+    [];
   // 获取全部热榜
-  allItems?.forEach((item: { id: string }) => {
+  allItems.forEach((item: { id: string }) => {
     // 基础数据
     const hotItem: RouterType["kuaishou"] = jsonObject[item.id];
+    if (!hotItem) return;
     const id = hotItem.photoIds?.json?.[0];
+    const hotValue = hotItem.hotValue ?? "";
+    const poster = hotItem.poster ? decodeURIComponent(hotItem.poster) : undefined;
     listData.push({
       id: hotItem.id,
       title: hotItem.name,
-      cover: decodeURIComponent(hotItem.poster),
-      hot: parseChineseNumber(hotItem.hotValue),
+      cover: poster,
+      hot: parseChineseNumber(String(hotValue)),
       timestamp: undefined,
       url: `https://www.kuaishou.com/short-video/${id}`,
       mobileUrl: `https://www.kuaishou.com/short-video/${id}`,

src/routes/linuxdo.ts

@@ -1,16 +1,7 @@
 import type { RouterData } from "../types.js";
 import { get } from "../utils/getData.js";
 import { getTime } from "../utils/getTime.js";
-
-interface Topic {
-  id: number;
-  title: string;
-  excerpt: string;
-  last_poster_username: string;
-  created_at: string;
-  views: number;
-  like_count: number;
-}
+import { parseRSS } from "../utils/parseRSS.js";
 
 export const handleRoute = async (_: undefined, noCache: boolean) => {
   const listData = await getList(noCache);
@@ -19,7 +10,7 @@ export const handleRoute = async (_: undefined, noCache: boolean) => {
     title: "Linux.do",
     type: "热门文章",
     description: "Linux 技术社区热搜",
-    link: "https://linux.do/hot",
+    link: "https://linux.do/top/weekly",
     total: listData.data?.length || 0,
     ...listData,
   };
@@ -27,31 +18,34 @@ export const handleRoute = async (_: undefined, noCache: boolean) => {
 };
 
 const getList = async (noCache: boolean) => {
-  const url = "https://linux.do/top/weekly.json";
+  const url = "https://linux.do/top.rss?period=weekly";
   const result = await get({
     url,
     noCache,
     headers: {
-      "Accept": "application/json",
-    }
+      "Accept": "application/rss+xml, application/xml;q=0.9, */*;q=0.8",
+      "User-Agent":
+        "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36",
+    },
   });
 
-  const topics = result.data.topic_list.topics as Topic[];
-  const list = topics.map((topic) => {
+  const items = await parseRSS(result.data);
+  const list = items.map((item, index) => {
+    const link = item.link || "";
     return {
-      id: topic.id,
-      title: topic.title,
-      desc: topic.excerpt,
-      author: topic.last_poster_username,
-      timestamp: getTime(topic.created_at),
-      url: `https://linux.do/t/${topic.id}`,
-      mobileUrl: `https://linux.do/t/${topic.id}`,
-      hot: topic.views || topic.like_count
+      id: item.guid || link || index,
+      title: item.title || "",
+      desc: item.contentSnippet?.trim() || item.content?.trim() || "",
+      author: item.author,
+      timestamp: getTime(item.pubDate || 0),
+      url: link,
+      mobileUrl: link,
+      hot: undefined,
     };
   });
 
   return {
     ...result,
-    data: list
+    data: list,
   };
 };

src/routes/weibo.ts

@@ -2,6 +2,7 @@ import type { RouterData } from "../types.js";
 import type { RouterType } from "../router.types.js";
 import { get } from "../utils/getData.js";
 import { getTime } from "../utils/getTime.js";
+import { config } from "../config";
 
 export const handleRoute = async (_: undefined, noCache: boolean) => {
   const listData = await getList(noCache);
@@ -18,35 +19,35 @@ export const handleRoute = async (_: undefined, noCache: boolean) => {
 };
 
 const getList = async (noCache: boolean) => {
-  const url =
-    "https://m.weibo.cn/api/container/getIndex?containerid=106003type%3D25%26t%3D3%26disable_hot%3D1%26filter_type%3Drealtimehot&title=%E5%BE%AE%E5%8D%9A%E7%83%AD%E6%90%9C&extparam=filter_type%3Drealtimehot%26mi_cid%3D100103%26pos%3D0_0%26c_type%3D30%26display_time%3D1540538388&luicode=10000011&lfid=231583";
+  const url = "https://weibo.com/ajax/side/hotSearch";
 
   const result = await get({
     url,
     noCache,
     ttl: 60,
     headers: {
-      Referer: "https://s.weibo.com/top/summary?cate=realtimehot",
-      "MWeibo-Pwa": "1",
-      "X-Requested-With": "XMLHttpRequest",
+      Referer: "https://weibo.com/",
       "User-Agent":
-        "Mozilla/5.0 (iPhone; CPU iPhone OS 11_0 like Mac OS X) AppleWebKit/604.1.38 (KHTML, like Gecko) Version/11.0 Mobile/15A372 Safari/604.1",
+        "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36",
     },
   });
-  const list = result.data.data.cards?.[0]?.card_group;
+
+  if (!result.data?.data?.realtime) {
+    return { ...result, data: [] };
+  }
+
+  const list = result.data.data.realtime;
   return {
     ...result,
-    data: list.map((v: RouterType["weibo"]) => {
-      const key = v.word_scheme ? v.word_scheme : `#${v.desc}`;
+    data: list.map((v: any, index: number) => {
+      const title = v.word || v.word_scheme || `热搜${index + 1}`;
       return {
-        id: v.itemid,
-        title: v.desc,
-        desc: key,
-        // author: v.flag_desc,
-        timestamp: getTime(v.onboard_time),
-        // hot: v.num,
-        url: `https://s.weibo.com/weibo?q=${encodeURIComponent(key)}&t=31&band_rank=1&Refer=top`,
-        mobileUrl: v?.scheme,
+        id: v.mid || v.word_scheme || `weibo-${index}`,
+        title: title,
+        desc: v.word_scheme || `#${title}#`,
+        timestamp: getTime(v.onboard_time || Date.now()),
+        url: `https://s.weibo.com/weibo?q=${encodeURIComponent(title)}`,
+        mobileUrl: `https://s.weibo.com/weibo?q=${encodeURIComponent(title)}`,
       };
     }),
   };

src/utils/cache.ts

@@ -26,6 +26,7 @@ const redis = new Redis({
   host: config.REDIS_HOST,
   port: config.REDIS_PORT,
   password: config.REDIS_PASSWORD,
+  db: config.REDIS_DB,
   maxRetriesPerRequest: 5,
   // 重试策略：最小延迟 50ms，最大延迟 2s
   retryStrategy: (times) => Math.min(times * 50, 2000),