tt-help-cli-ycl 1.4.44 → 1.4.46

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.44",
+  "version": "1.4.46",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/lib/shop-scraper.mjs

@@ -17,6 +17,7 @@
  */
 
 import { ProxyAgent, request } from "undici";
+import { execSync } from "child_process";
 import { DEFAULT_PROXY } from "./constants.js";
 
 const PDP_HEADERS = {
@@ -80,6 +81,116 @@ function extractJsonByKey(html, keyPattern) {
   return null;
 }
 
+/**
+ * Windows: 后台服务的 Playwright 浏览器在 Session 0（隔离桌面），
+ * RDP 看不到。找到交互桌面 Session，用计划任务推到那个 Session。
+ */
+let winRecoveryRunning = false;
+
+async function resolveCaptchaWindows(url, proxyUrl) {
+  if (winRecoveryRunning) {
+    console.error("[shop-scraper] Windows: 验证码恢复已在运行，跳过");
+    return false;
+  }
+  winRecoveryRunning = true;
+
+  try {
+    const edgeUrl = url.replace("shop.tiktok.com", "www.tiktok.com/shop/pdp");
+    const taskName = "TikTokCaptchaRecovery";
+
+    // 1. 找到交互桌面 Session
+    let sessionId = null;
+    try {
+      const sessions = execSync(
+        `powershell -NoProfile -Command "query session 2>$null | Select-String 'Active' | ForEach-Object { \\$_.Line }"`,
+        { shell: "cmd.exe", timeout: 5000, encoding: "utf-8" },
+      );
+      const lines = sessions.trim().split("\n");
+      for (const line of lines) {
+        const match = line.match(/>\s*(\d+)\s/);
+        if (match && match[1] !== "0") {
+          sessionId = parseInt(match[1], 10);
+          break;
+        }
+      }
+    } catch {
+      console.error("[shop-scraper] Windows: 无法获取交互 Session");
+    }
+
+    if (sessionId) {
+      console.error(
+        `[shop-scraper] Windows: Session ${sessionId}，用计划任务启动 Edge`,
+      );
+      try {
+        execSync(`schtasks /delete /tn "${taskName}" /f 2>$null`, {
+          shell: "cmd.exe",
+        });
+      } catch {}
+      execSync(
+        `schtasks /create /tn "${taskName}" /tr "msedge ${edgeUrl}" /sc once /st 00:00 /it /f`,
+        { shell: "cmd.exe", timeout: 10000 },
+      );
+      execSync(`schtasks /run /tn "${taskName}"`, {
+        shell: "cmd.exe",
+        timeout: 10000,
+      });
+      setTimeout(() => {
+        try {
+          execSync(`schtasks /delete /tn "${taskName}" /f 2>$null`, {
+            shell: "cmd.exe",
+          });
+        } catch {}
+      }, 10000);
+    } else {
+      console.error(
+        "[shop-scraper] Windows: 未找到交互 Session，尝试直接 start",
+      );
+      try {
+        execSync(`start "" "microsoft-edge:${edgeUrl}"`, {
+          shell: "cmd.exe",
+          timeout: 5000,
+        });
+      } catch {}
+    }
+
+    // 2. 轮询 undici 等待 IP 信用恢复
+    const proxy = proxyUrl || DEFAULT_PROXY;
+    const agent = proxy ? new ProxyAgent(proxy) : null;
+    const POLL_MS = 5000;
+    const MAX_WAIT_MS = 3600_000;
+    const startTime = Date.now();
+
+    while (Date.now() - startTime < MAX_WAIT_MS) {
+      await new Promise((r) => setTimeout(r, POLL_MS));
+      try {
+        const { body } = await request(url, {
+          headers: PDP_HEADERS,
+          ...(agent ? { dispatcher: agent } : {}),
+        });
+        const html = await body.text();
+        if (
+          html.includes('"product_model"') ||
+          html.includes('"review_ratings"')
+        ) {
+          console.error("[shop-scraper] Captcha solved, IP credit restored");
+          return true;
+        }
+      } catch {}
+      const waited = Math.round((Date.now() - startTime) / 60000);
+      if (waited > 0 && waited % 5 === 0) {
+        console.error(
+          `[shop-scraper] 等待验证码解决中... 已等待 ${waited} 分钟`,
+        );
+      }
+    }
+
+    console.error("[shop-scraper] Windows: 等待验证码超时");
+    return false;
+  } finally {
+    winRecoveryRunning = false;
+  }
+}
+
 /**
  * Open a headed browser for manual captcha solving, extract cookies
  * so undici can reuse them for a while.
@@ -90,6 +201,12 @@ function extractJsonByKey(html, keyPattern) {
 export async function resolveCaptchaWithBrowser(url, proxyUrl) {
   const { chromium } = await import("playwright");
 
+  // Windows 后台服务启动的浏览器在 Session 0，RDP 看不到。
+  // 用 start 命令推到当前交互桌面。
+  if (process.platform === "win32") {
+    return resolveCaptchaWindows(url, proxyUrl);
+  }
+
   const args = ["--no-sandbox"];
   if (proxyUrl) {
     args.push(`--proxy-server=${proxyUrl.replace("http://", "http://")}`);