tt-help-cli-ycl 1.3.48 → 1.3.50

package/src/lib/tiktok-scraper.mjs

@@ -1,4 +1,3 @@
-import fs from "fs";
 import os from "os";
 import path from "path";
 import { chromium } from "playwright";
@@ -26,11 +25,6 @@ function getFallbackProfileDir() {
   );
 }
 
-function fallbackProfileExists() {
-  const dir = getFallbackProfileDir();
-  return fs.existsSync(dir);
-}
-
 class PageSlot {
   constructor(page) {
     this.page = page;
@@ -87,7 +81,10 @@ async function initContext(executablePath, poolSize, userDataDir) {
   const slots = [];
 
   if (userDataDir) {
-    context = await chromium.launchPersistentContext(userDataDir, createLaunchOptions(executablePath));
+    context = await chromium.launchPersistentContext(
+      userDataDir,
+      createLaunchOptions(executablePath),
+    );
     const existing = context.pages();
     if (existing.length > 0) {
       slots.push(new PageSlot(existing[0]));
@@ -123,10 +120,11 @@ export class TikTokScraper {
     this.slotIdx = 0;
     this.lastWarmTime = 0;
     this.warmPromise = null;
-    this._fallbackBrowser = null;
-    this._fallbackContext = null;
-    this._fallbackSlots = [];
-    this._fallbackSlotIdx = 0;
+    // 登录态 pool（init 时直接启动）
+    this.authBrowser = null;
+    this.authContext = null;
+    this.authSlots = [];
+    this.authSlotIdx = 0;
   }
 
   async init() {
@@ -136,25 +134,27 @@ 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",
       );
     }
-    const { browser, context, slots } = await initContext(executablePath, this.poolSize, null);
+    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;
-    if (!fallbackProfileExists()) {
-      return false;
-    }
+    // 启动登录态 pool（1 个 slot）
     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;
-    return true;
+    const {
+      browser: authBrowser,
+      context: authContext,
+      slots: authSlots,
+    } = await initContext(executablePath, 1, fallbackDir);
+    this.authBrowser = authBrowser;
+    this.authContext = authContext;
+    this.authSlots = authSlots;
+
+    await this.warmWaf();
   }
 
   async close() {
@@ -168,7 +168,9 @@ export class TikTokScraper {
         });
         await Promise.race([
           closePromise,
-          delay(BROWSER_CLOSE_TIMEOUT).then(() => { closeTimedOut = true; }),
+          delay(BROWSER_CLOSE_TIMEOUT).then(() => {
+            closeTimedOut = true;
+          }),
         ]);
         if (closeTimedOut) {
           console.error(
@@ -178,18 +180,20 @@ export class TikTokScraper {
       }
     };
 
-    // 无登录态的 browser 是 chromium.launch() 返回的，直接 close
+    // 无登录态的 browser
     await closeAll(this.browser);
-    // fallback 的 context 是 launchPersistentContext 返回的，直接 close
-    if (this._fallbackContext) {
-      try { await this._fallbackContext.close(); } catch {}
+    // 登录态的 context（launchPersistentContext 返回的是 context 当 browser）
+    if (this.authContext) {
+      try {
+        await this.authContext.close();
+      } catch {}
     }
     this.browser = null;
     this.context = null;
     this.slots = [];
-    this._fallbackBrowser = null;
-    this._fallbackContext = null;
-    this._fallbackSlots = [];
+    this.authBrowser = null;
+    this.authContext = null;
+    this.authSlots = [];
   }
 
   async restart() {
@@ -230,14 +234,12 @@ export class TikTokScraper {
     return Date.now() - this.lastWarmTime > this.wafTtl;
   }
 
-  _pickSlot(slots, idx, setIdx) {
-    const slot = slots[idx % slots.length];
-    return 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 {}
@@ -248,23 +250,49 @@ export class TikTokScraper {
   }
 
   async _fetchViewSource(url, ctx) {
-    const slots = ctx === this._fallbackContext ? this._fallbackSlots : this.slots;
-    const slot = slots[this.slotIdx % slots.length];
-    this.slotIdx++;
+    const slots = ctx === this.authContext ? this.authSlots : this.slots;
+    const slotIdx = ctx === this.authContext ? this.authSlotIdx : this.slotIdx;
+    const slot = slots[slotIdx % slots.length];
+    if (ctx === this.authContext) {
+      this.authSlotIdx++;
+    } else {
+      this.slotIdx++;
+    }
 
     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, {
+
+        await page.goto("view-source:" + url, {
           waitUntil: "domcontentloaded",
           timeout: 15000,
         });
-        await delay(1500);
-        return await page.content();
+
+        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";
+          });
+          return content;
+        });
+
+        // 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(() => {});
+
+        slot.requestCount += 1;
+
+        return content;
       });
     } catch (e) {
       console.error(`[TikTokScraper] _fetchViewSource failed: ${e.message}`);
@@ -277,38 +305,47 @@ export class TikTokScraper {
       ? username.slice(1)
       : username;
     const url = `https://www.tiktok.com/@${normalizedUsername}`;
-    const executablePath = detectBrowser();
 
     // 第一趟：无登录态
     let rawHtml = await this._fetchViewSource(url, this.context);
     let result = rawHtml ? parseUserInfo(rawHtml) : null;
 
-    // 解析失败：fallback 到登录态
+    // 解析失败：先尝试 warmWaf 后重试
     if (!result) {
-      const ok = await this._ensureFallback(executablePath);
-      if (ok) {
-        rawHtml = await this._fetchViewSource(url, this._fallbackContext);
-        result = rawHtml ? parseUserInfo(rawHtml) : null;
-      }
+      try {
+        await this.warmWaf();
+      } catch {}
+      rawHtml = await this._fetchViewSource(url, this.context);
+      result = rawHtml ? parseUserInfo(rawHtml) : null;
+    }
+
+    // 仍然失败：使用登录态 pool
+    if (!result) {
+      rawHtml = await this._fetchViewSource(url, this.authContext);
+      result = rawHtml ? parseUserInfo(rawHtml) : null;
     }
 
     return result || null;
   }
 
   async getVideoInfo(videoUrl) {
-    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
+    // 解析失败：先尝试 warmWaf 后重试
     if (!result) {
-      const ok = await this._ensureFallback(executablePath);
-      if (ok) {
-        rawHtml = await this._fetchViewSource(videoUrl, this._fallbackContext);
-        result = rawHtml ? parseVideoInfo(rawHtml) : null;
-      }
+      try {
+        await this.warmWaf();
+      } catch {}
+      rawHtml = await this._fetchViewSource(videoUrl, this.context);
+      result = rawHtml ? parseVideoInfo(rawHtml) : null;
+    }
+
+    // 仍然失败：使用登录态 pool
+    if (!result) {
+      rawHtml = await this._fetchViewSource(videoUrl, this.authContext);
+      result = rawHtml ? parseVideoInfo(rawHtml) : null;
     }
 
     return result || null;

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

@@ -1,73 +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";
-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);
-});
+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);
+});

package/src/npm-main.js

@@ -1,70 +1,70 @@
-import { parseArgs } from "./lib/args.js";
-import { PUBLIC_HELP_TEXT } from "./lib/constants.js";
-import { handleInfo } from "./cli/info.js";
-import { handleExplore } from "./cli/explore.js";
-import { handleAttach } from "./cli/attach.js";
-import { handleConfig, showConfig, showUsage, version } from "./cli/config.js";
-import { handleOpen } from "./cli/open.js";
-import { handleComments } from "./cli/comments.js";
-
-function exitUnsupportedCommand(command) {
-  console.error(
-    `[${command}] 当前 npm 发布包不包含该命令；如需使用，请在仓库源码环境中运行 node src/main.js ${command} ...`,
-  );
-  process.exit(1);
-}
-
-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":
-    case "webserver":
-    case "videostats":
-    case "db-import":
-      return exitUnsupportedCommand(parsed.subcommand);
-    case "open":
-      return handleOpen(parsed);
-    case "comments":
-      return handleComments(parsed);
-  }
-
-  const {
-    urls,
-    outputFile,
-    exploreCount,
-    showConfig: showCfg,
-    showHelp,
-    showVersion,
-    configAction,
-    configKey,
-    configValue,
-  } = parsed;
-
-  if (showVersion) {
-    console.log(version);
-    process.exit(0);
-  }
-  if (showHelp) return showUsage(PUBLIC_HELP_TEXT);
-  if (configAction) return handleConfig(configAction, configKey, configValue);
-  if (showCfg) return showConfig(urls, outputFile);
-  if (urls.length === 0 && exploreCount === 0)
-    return showUsage(PUBLIC_HELP_TEXT);
-
-  if (exploreCount > 0) {
-    return handleExplore({ ...parsed, subcommand: "explore" });
-  }
-
-  return handleInfo(parsed);
-}
-
-main().catch((err) => {
-  console.error(`错误: ${err.message}`);
-  process.exit(1);
-});
+import { parseArgs } from "./lib/args.js";
+import { PUBLIC_HELP_TEXT } from "./lib/constants.js";
+import { handleInfo } from "./cli/info.js";
+import { handleExplore } from "./cli/explore.js";
+import { handleAttach } from "./cli/attach.js";
+import { handleConfig, showConfig, showUsage, version } from "./cli/config.js";
+import { handleOpen } from "./cli/open.js";
+import { handleComments } from "./cli/comments.js";
+
+function exitUnsupportedCommand(command) {
+  console.error(
+    `[${command}] 当前 npm 发布包不包含该命令；如需使用，请在仓库源码环境中运行 node src/main.js ${command} ...`,
+  );
+  process.exit(1);
+}
+
+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":
+    case "webserver":
+    case "videostats":
+    case "db-import":
+      return exitUnsupportedCommand(parsed.subcommand);
+    case "open":
+      return handleOpen(parsed);
+    case "comments":
+      return handleComments(parsed);
+  }
+
+  const {
+    urls,
+    outputFile,
+    exploreCount,
+    showConfig: showCfg,
+    showHelp,
+    showVersion,
+    configAction,
+    configKey,
+    configValue,
+  } = parsed;
+
+  if (showVersion) {
+    console.log(version);
+    process.exit(0);
+  }
+  if (showHelp) return showUsage(PUBLIC_HELP_TEXT);
+  if (configAction) return handleConfig(configAction, configKey, configValue);
+  if (showCfg) return showConfig(urls, outputFile);
+  if (urls.length === 0 && exploreCount === 0)
+    return showUsage(PUBLIC_HELP_TEXT);
+
+  if (exploreCount > 0) {
+    return handleExplore({ ...parsed, subcommand: "explore" });
+  }
+
+  return handleInfo(parsed);
+}
+
+main().catch((err) => {
+  console.error(`错误: ${err.message}`);
+  process.exit(1);
+});