tt-help-cli-ycl 1.3.44 → 1.3.46

package/src/lib/tiktok-scraper.mjs

@@ -1,3 +1,5 @@
+import os from "os";
+import path from "path";
 import { chromium } from "playwright";
 import { detectBrowser } from "./browser/launch.js";
 import { parseUserInfo, parseVideoInfo } from "./parse-ssr.mjs";
@@ -7,11 +9,22 @@ const DEFAULT_WAF_TTL = 120000;
 const DEFAULT_WARM_URL = "https://www.tiktok.com/@nike";
 const BROWSER_CLOSE_TIMEOUT = 5000;
 const DEFAULT_MAX_REQUESTS_PER_PAGE = 50;
+const FALLBACK_PROFILE_PORT = 9222;
 
 function delay(ms) {
   return new Promise((r) => setTimeout(r, ms));
 }
 
+function getFallbackProfileDir() {
+  const profile = `p${FALLBACK_PROFILE_PORT}`;
+  return path.join(
+    os.homedir(),
+    "Library",
+    "Application Support",
+    `Microsoft Edge For Testing_${profile}`,
+  );
+}
+
 class PageSlot {
   constructor(page) {
     this.page = page;
@@ -47,6 +60,46 @@ class PromiseQueue {
   }
 }
 
+function createLaunchOptions(executablePath) {
+  return {
+    headless: true,
+    executablePath,
+    handleSIGINT: false,
+    handleSIGTERM: false,
+    handleSIGHUP: false,
+    args: [
+      "--no-sandbox",
+      "--disable-setuid-sandbox",
+      "--disable-dev-shm-usage",
+    ],
+  };
+}
+
+async function initContext(executablePath, poolSize, userDataDir) {
+  let context;
+  let browser = null;
+  const slots = [];
+
+  if (userDataDir) {
+    context = await chromium.launchPersistentContext(userDataDir, createLaunchOptions(executablePath));
+    const existing = context.pages();
+    if (existing.length > 0) {
+      slots.push(new PageSlot(existing[0]));
+    }
+    for (let i = slots.length; i < poolSize; i++) {
+      slots.push(new PageSlot(await context.newPage()));
+    }
+  } else {
+    browser = await chromium.launch(createLaunchOptions(executablePath));
+    context = await browser.newContext();
+    for (let i = 0; i < poolSize; i++) {
+      slots.push(new PageSlot(await context.newPage()));
+    }
+  }
+
+  return { browser, context, slots };
+}
+
 export class TikTokScraper {
   constructor({
     poolSize = DEFAULT_POOL_SIZE,
@@ -64,6 +117,10 @@ export class TikTokScraper {
     this.slotIdx = 0;
     this.lastWarmTime = 0;
     this.warmPromise = null;
+    this._fallbackBrowser = null;
+    this._fallbackContext = null;
+    this._fallbackSlots = [];
+    this._fallbackSlotIdx = 0;
   }
 
   async init() {
@@ -73,49 +130,56 @@ export class TikTokScraper {
         "未找到本地浏览器（Chrome/Edge），请先安装浏览器或执行 npx playwright install",
       );
     }
-    this.browser = await chromium.launch({
-      headless: true,
-      executablePath,
-      handleSIGINT: false,
-      handleSIGTERM: false,
-      handleSIGHUP: false,
-      args: [
-        "--no-sandbox",
-        "--disable-setuid-sandbox",
-        "--disable-dev-shm-usage",
-      ],
-    });
-    this.context = await this.browser.newContext();
-    for (let i = 0; i < this.poolSize; i++) {
-      this.slots.push(new PageSlot(await this.context.newPage()));
-    }
+    const { browser, context, slots } = await initContext(executablePath, this.poolSize, null);
+    this.browser = browser;
+    this.context = context;
+    this.slots = slots;
     await this.warmWaf();
   }
 
+  async _ensureFallback(executablePath) {
+    if (this._fallbackContext) return;
+    const fallbackDir = getFallbackProfileDir();
+    console.error(`[TikTokScraper] 无登录态获取失败，fallback 到登录态 profile: ${fallbackDir}`);
+    const { browser, context, slots } = await initContext(executablePath, 1, fallbackDir);
+    this._fallbackBrowser = browser;
+    this._fallbackContext = context;
+    this._fallbackSlots = slots;
+  }
+
   async close() {
-    if (this.browser) {
-      const browser = this.browser;
-      let closeTimedOut = false;
-      const closePromise = browser.close().catch((error) => {
-        console.error(
-          `[TikTokScraper] browser.close() failed: ${error.message}`,
-        );
-      });
-      await Promise.race([
-        closePromise,
-        delay(BROWSER_CLOSE_TIMEOUT).then(() => {
-          closeTimedOut = true;
-        }),
-      ]);
-      if (closeTimedOut) {
-        console.error(
-          `[TikTokScraper] browser.close() 超时 ${BROWSER_CLOSE_TIMEOUT}ms，跳过等待并继续退出`,
-        );
+    const closeAll = async (browser) => {
+      if (browser) {
+        let closeTimedOut = false;
+        const closePromise = browser.close().catch((error) => {
+          console.error(
+            `[TikTokScraper] browser.close() failed: ${error.message}`,
+          );
+        });
+        await Promise.race([
+          closePromise,
+          delay(BROWSER_CLOSE_TIMEOUT).then(() => { closeTimedOut = true; }),
+        ]);
+        if (closeTimedOut) {
+          console.error(
+            `[TikTokScraper] browser.close() 超时 ${BROWSER_CLOSE_TIMEOUT}ms，跳过等待并继续退出`,
+          );
+        }
       }
-      this.browser = null;
-      this.context = null;
-      this.slots = [];
+    };
+
+    // 无登录态的 browser 是 chromium.launch() 返回的，直接 close
+    await closeAll(this.browser);
+    // fallback 的 context 是 launchPersistentContext 返回的，直接 close
+    if (this._fallbackContext) {
+      try { await this._fallbackContext.close(); } catch {}
     }
+    this.browser = null;
+    this.context = null;
+    this.slots = [];
+    this._fallbackBrowser = null;
+    this._fallbackContext = null;
+    this._fallbackSlots = [];
   }
 
   async restart() {
@@ -125,7 +189,7 @@ export class TikTokScraper {
 
   isAlive() {
     try {
-      return !!(this.browser && this.browser.isConnected());
+      return !!(this.context && (!this.browser || this.browser.isConnected()));
     } catch {
       return false;
     }
@@ -152,98 +216,88 @@ export class TikTokScraper {
     return this.warmPromise;
   }
 
-  _needWarm() {
+  isWarmExpired() {
     return Date.now() - this.lastWarmTime > this.wafTtl;
   }
 
-  _pickSlot() {
-    const slot = this.slots[this.slotIdx % this.poolSize];
-    this.slotIdx++;
+  _pickSlot(slots, idx, setIdx) {
+    const slot = slots[idx % slots.length];
     return slot;
   }
 
-  async _ensurePage(slot) {
+  async _ensurePage(slot, context) {
     try {
-      if (
-        !slot.page.isClosed() &&
-        slot.requestCount < this.maxRequestsPerPage
-      ) {
+      if (!slot.page.isClosed() && slot.requestCount < this.maxRequestsPerPage) {
         return slot.page;
       }
     } catch {}
     await slot.page?.close().catch(() => {});
-    slot.page = await this.context.newPage();
+    slot.page = await context.newPage();
     slot.requestCount = 0;
     return slot.page;
   }
 
-  async _fetchViewSource(url, slot) {
-    const page = await this._ensurePage(slot);
-
-    await page.goto("view-source:" + url, {
-      waitUntil: "domcontentloaded",
-      timeout: 15000,
-    });
+  async _fetchViewSource(url, ctx) {
+    const slots = ctx === this._fallbackContext ? this._fallbackSlots : this.slots;
+    const slot = slots[this.slotIdx % slots.length];
+    this.slotIdx++;
 
-    const content = await page.evaluate(() => {
-      const rows = document.querySelectorAll("tr");
-      let content = "";
-      rows.forEach((r) => {
-        const lc = r.querySelector(".line-content");
-        if (lc) content += lc.textContent + "\n";
+    try {
+      return await slot.lock.run(async () => {
+        const page = await this._ensurePage(slot, ctx);
+        slot.requestCount++;
+        if (ctx === this.context && this.isWarmExpired()) {
+          await this.warmWaf();
+        }
+        await page.goto(url, {
+          waitUntil: "domcontentloaded",
+          timeout: 15000,
+        });
+        await delay(1500);
+        return await page.content();
       });
-      return content;
-    });
+    } catch (e) {
+      console.error(`[TikTokScraper] _fetchViewSource failed: ${e.message}`);
+      return null;
+    }
+  }
 
-    // 导航到 about:blank 释放当前页面的 DOM 和 JS 堆
-    await page
-      .goto("about:blank", {
-        waitUntil: "domcontentloaded",
-        timeout: 5000,
-      })
-      .catch(() => {});
+  async getUserInfo(username) {
+    const normalizedUsername = username.startsWith("@")
+      ? username.slice(1)
+      : username;
+    const url = `https://www.tiktok.com/@${normalizedUsername}`;
+    const executablePath = detectBrowser();
 
-    slot.requestCount += 1;
+    // 第一趟：无登录态
+    let rawHtml = await this._fetchViewSource(url, this.context);
+    let result = rawHtml ? parseUserInfo(rawHtml) : null;
 
-    return content;
-  }
+    // 解析失败：fallback 到登录态
+    if (!result) {
+      await this._ensureFallback(executablePath);
+      rawHtml = await this._fetchViewSource(url, this._fallbackContext);
+      result = rawHtml ? parseUserInfo(rawHtml) : null;
+    }
 
-  async getUserInfo(uniqueId) {
-    const slot = this._pickSlot();
-    return slot.lock.run(async () => {
-      let rawHtml = await this._fetchViewSource(
-        `https://www.tiktok.com/@${uniqueId}`,
-        slot,
-      );
-      let result = parseUserInfo(rawHtml);
-      if (!result) {
-        try {
-          await this.warmWaf();
-        } catch {}
-        rawHtml = await this._fetchViewSource(
-          `https://www.tiktok.com/@${uniqueId}`,
-          slot,
-        );
-        result = parseUserInfo(rawHtml);
-      }
-      return result || null;
-    });
+    return result || null;
   }
 
   async getVideoInfo(videoUrl) {
-    const slot = this._pickSlot();
-    return slot.lock.run(async () => {
-      let rawHtml = await this._fetchViewSource(videoUrl, slot);
-      let result = parseVideoInfo(rawHtml);
-      if (!result) {
-        try {
-          await this.warmWaf();
-        } catch {}
-        rawHtml = await this._fetchViewSource(videoUrl, slot);
-        result = parseVideoInfo(rawHtml);
-      }
-      return result || null;
-    });
+    const executablePath = detectBrowser();
+
+    // 第一趟：无登录态
+    let rawHtml = await this._fetchViewSource(videoUrl, this.context);
+    let result = rawHtml ? parseVideoInfo(rawHtml) : null;
+
+    // 解析失败：fallback 到登录态
+    if (!result) {
+      await this._ensureFallback(executablePath);
+      rawHtml = await this._fetchViewSource(videoUrl, this._fallbackContext);
+      result = rawHtml ? parseVideoInfo(rawHtml) : null;
+    }
+
+    return result || null;
   }
 
   async getUserAndVideo(videoUrl) {

package/src/lib/url.js

@@ -1,52 +1,52 @@
-const BASE_URL = 'https://www.tiktok.com';
-
-export function extractUniqueId(url) {
-  const m = url.match(/\/@([\w.-]+)/);
-  return m ? m[1] : null;
-}
-
-export function extractVideoId(url) {
-  const m = url.match(/\/video\/(\d+)/);
-  return m ? m[1] : null;
-}
-
-export function normalizeUsername(input) {
-  return (input || '').replace(/^@/, '');
-}
-
-export function toProfileUrl(handle) {
-  const clean = normalizeUsername(handle);
-  return `${BASE_URL}/@${clean}`;
-}
-
-export function toVideoUrl(handle, videoId) {
-  const clean = normalizeUsername(handle);
-  return `${BASE_URL}/@${clean}/video/${videoId}`;
-}
-
-export function ensureAbsoluteUrl(href) {
-  if (href.startsWith('http')) return href;
-  return `${BASE_URL}${href}`;
-}
-
-export function isProfileUrl(url) {
-  return /\/@[\w.-]+(?:$|[?#])/.test(url);
-}
-
-export function isVideoUrl(url) {
-  return /\/video\/\d+/.test(url);
-}
-
-export function extractDisplayPath(url) {
-  try {
-    const parts = new URL(url).pathname.split('/').filter(Boolean);
-    return parts.slice(-2).join('/');
-  } catch {
-    return url;
-  }
-}
-
-export function extractAuthorFromVideoUrl(url) {
-  const m = url.match(/@([^/]+)\/video/);
-  return m ? '@' + m[1] : null;
-}
+const BASE_URL = 'https://www.tiktok.com';
+
+export function extractUniqueId(url) {
+  const m = url.match(/\/@([\w.-]+)/);
+  return m ? m[1] : null;
+}
+
+export function extractVideoId(url) {
+  const m = url.match(/\/video\/(\d+)/);
+  return m ? m[1] : null;
+}
+
+export function normalizeUsername(input) {
+  return (input || '').replace(/^@/, '');
+}
+
+export function toProfileUrl(handle) {
+  const clean = normalizeUsername(handle);
+  return `${BASE_URL}/@${clean}`;
+}
+
+export function toVideoUrl(handle, videoId) {
+  const clean = normalizeUsername(handle);
+  return `${BASE_URL}/@${clean}/video/${videoId}`;
+}
+
+export function ensureAbsoluteUrl(href) {
+  if (href.startsWith('http')) return href;
+  return `${BASE_URL}${href}`;
+}
+
+export function isProfileUrl(url) {
+  return /\/@[\w.-]+(?:$|[?#])/.test(url);
+}
+
+export function isVideoUrl(url) {
+  return /\/video\/\d+/.test(url);
+}
+
+export function extractDisplayPath(url) {
+  try {
+    const parts = new URL(url).pathname.split('/').filter(Boolean);
+    return parts.slice(-2).join('/');
+  } catch {
+    return url;
+  }
+}
+
+export function extractAuthorFromVideoUrl(url) {
+  const m = url.match(/@([^/]+)\/video/);
+  return m ? '@' + m[1] : null;
+}

package/src/main.js

@@ -1,70 +1,73 @@
-import { parseArgs } from "./lib/args.js";
-import { proxy, HELP_TEXT, getConfigText } from "./lib/constants.js";
-import { handleInfo } from "./cli/info.js";
-import { handleExplore } from "./cli/explore.js";
-import { handleAttach } from "./cli/attach.js";
-import { handleWatch } from "./cli/watch.js";
-import { handleConfig, showConfig, showUsage, version } from "./cli/config.js";
-import { handleOpen } from "./cli/open.js";
-import { handleComments } from "./cli/comments.js";
-import { handleVideoStats } from "./cli/videostats.js";
-import { handleDbImport } from "./cli/db-import.js";
-
-async function main() {
-  const parsed = parseArgs();
-
-  switch (parsed.subcommand) {
-    case "explore":
-      return handleExplore(parsed);
-    case "info":
-      return handleInfo(parsed);
-    case "attach":
-      return handleAttach(parsed);
-    case "watch":
-      return handleWatch(parsed);
-    case "open":
-      return handleOpen(parsed);
-    case "comments":
-      return handleComments(parsed);
-    case "videostats":
-      return handleVideoStats(parsed);
-    case "db-import":
-      return handleDbImport(parsed);
-  }
-
-  const {
-    urls,
-    outputFile,
-    outputFormat,
-    exploreCount,
-    showConfig: showCfg,
-    showHelp,
-    showVersion,
-    customProxy,
-    configAction,
-    configKey,
-    configValue,
-  } = parsed;
-
-  if (showVersion) {
-    console.log(version);
-    process.exit(0);
-  }
-  if (showHelp) return showUsage();
-  if (configAction) return handleConfig(configAction, configKey, configValue);
-  if (showCfg) return showConfig(urls, outputFile);
-  if (urls.length === 0 && exploreCount === 0) return showUsage();
-
-  // 默认行为：URL 走 info，--explore 走 explore
-  if (exploreCount > 0) {
-    return handleExplore({ ...parsed, subcommand: "explore" });
-  }
-
-  // 有 URL 默认走 info
-  return handleInfo(parsed);
-}
-
-main().catch((err) => {
-  console.error(`错误: ${err.message}`);
-  process.exit(1);
-});
+import { parseArgs } from "./lib/args.js";
+import { proxy, HELP_TEXT, getConfigText } from "./lib/constants.js";
+import { handleInfo } from "./cli/info.js";
+import { handleExplore } from "./cli/explore.js";
+import { handleAttach } from "./cli/attach.js";
+import { handleWatch } from "./cli/watch.js";
+import { handleConfig, showConfig, showUsage, version } from "./cli/config.js";
+import { handleOpen } from "./cli/open.js";
+import { handleComments } from "./cli/comments.js";
+import { handleVideoStats } from "./cli/videostats.js";
+import { handleDbImport } from "./cli/db-import.js";
+import { handleWebserver } from "./cli/webserver.js";
+
+async function main() {
+  const parsed = parseArgs();
+
+  switch (parsed.subcommand) {
+    case "explore":
+      return handleExplore(parsed);
+    case "info":
+      return handleInfo(parsed);
+    case "attach":
+      return handleAttach(parsed);
+    case "watch":
+      return handleWatch(parsed);
+    case "webserver":
+      return handleWebserver(parsed);
+    case "open":
+      return handleOpen(parsed);
+    case "comments":
+      return handleComments(parsed);
+    case "videostats":
+      return handleVideoStats(parsed);
+    case "db-import":
+      return handleDbImport(parsed);
+  }
+
+  const {
+    urls,
+    outputFile,
+    outputFormat,
+    exploreCount,
+    showConfig: showCfg,
+    showHelp,
+    showVersion,
+    customProxy,
+    configAction,
+    configKey,
+    configValue,
+  } = parsed;
+
+  if (showVersion) {
+    console.log(version);
+    process.exit(0);
+  }
+  if (showHelp) return showUsage();
+  if (configAction) return handleConfig(configAction, configKey, configValue);
+  if (showCfg) return showConfig(urls, outputFile);
+  if (urls.length === 0 && exploreCount === 0) return showUsage();
+
+  // 默认行为：URL 走 info，--explore 走 explore
+  if (exploreCount > 0) {
+    return handleExplore({ ...parsed, subcommand: "explore" });
+  }
+
+  // 有 URL 默认走 info
+  return handleInfo(parsed);
+}
+
+main().catch((err) => {
+  console.error(`错误: ${err.message}`);
+  process.exit(1);
+});