tt-help-cli-ycl 1.4.45 → 1.4.47

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.45",
+  "version": "1.4.47",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/lib/shop-scraper.mjs

@@ -83,58 +83,135 @@ function extractJsonByKey(html, keyPattern) {
 
 /**
  * Windows: 后台服务的 Playwright 浏览器在 Session 0（隔离桌面），
- * RDP 看不到。改为用 start 命令推到交互桌面，然后轮询 undici 等 IP 恢复信用。
+ * RDP 看不到。找到交互桌面 Session，用计划任务推到那个 Session。
+ * 配置环境变量（可选）：
+ *   CAPTCHA_RECOVERY_USER - schtasks /ru 用户名（必需，否则无法跨 Session）
+ *   CAPTCHA_RECOVERY_PASS - schtasks /rp 密码
  */
+let winRecoveryRunning = false;
+
 async function resolveCaptchaWindows(url, proxyUrl) {
-  // 用 Windows start 命令在交互桌面打开 Edge
-  const edgeUrl = url.replace("shop.tiktok.com", "www.tiktok.com/shop/pdp");
-  try {
-    execSync(`start "" "microsoft-edge:${edgeUrl}"`, {
-      shell: "cmd.exe",
-      timeout: 5000,
-    });
-    console.error(
-      "[shop-scraper] Windows: 已用 start 命令在交互桌面打开 Edge，请在 RDP 中完成验证码",
-    );
-  } catch (e) {
-    console.error("[shop-scraper] Windows: start 命令失败:", e.message);
+  if (winRecoveryRunning) {
+    console.error("[shop-scraper] Windows: 验证码恢复已在运行，跳过");
+    return false;
   }
+  winRecoveryRunning = true;
 
-  // 轮询 undici 等待 IP 信用恢复（用户解完验证码后代理 IP 就有信用了）
-  const proxy = proxyUrl || DEFAULT_PROXY;
-  const agent = proxy ? new ProxyAgent(proxy) : null;
-  const POLL_MS = 5000;
-  const MAX_WAIT_MS = 3600_000; // 最长等 1 小时
-  const start = Date.now();
+  try {
+    const edgeUrl = url.replace("shop.tiktok.com", "www.tiktok.com/shop/pdp");
+    const taskName = "TikTokCaptchaRecovery";
 
-  while (Date.now() - start < MAX_WAIT_MS) {
-    await new Promise((r) => setTimeout(r, POLL_MS));
+    // 1. 找到 RDP 交互桌面 Session（Active 且非 Session 0）
+    let sessionId = null;
     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;
+      const sessions = execSync(
+        `powershell -NoProfile -Command "query session 2>$null | findstr Active"`,
+        { shell: "cmd.exe", timeout: 5000, encoding: "utf-8" },
+      );
+      const lines = sessions.trim().split("\n");
+      for (const line of lines) {
+        // 匹配 RDP 会话: ">rdp-tcp#1  username  2  Active"
+        const match = line.match(/\s(\d+)\s+Active/);
+        if (match && match[1] !== "0") {
+          sessionId = parseInt(match[1], 10);
+          break;
+        }
       }
     } catch {
-      // 继续轮询
+      console.error("[shop-scraper] Windows: 无法获取交互 Session");
     }
-    // 每 60 秒打印一次状态
-    if (Math.round((Date.now() - start) / 60000) % 5 === 0) {
+
+    if (sessionId) {
       console.error(
-        `[shop-scraper] 等待验证码解决中... 已等待 ${Math.round((Date.now() - start) / 60000)} 分钟`,
+        `[shop-scraper] Windows: Session ${sessionId}，用计划任务启动 Edge`,
       );
+      // 清理旧任务
+      try {
+        execSync(`schtasks /delete /tn "${taskName}" /f 2>$null`, {
+          shell: "cmd.exe",
+        });
+      } catch {}
+
+      const taskUser = process.env.CAPTCHA_RECOVERY_USER || "";
+      const taskPass = process.env.CAPTCHA_RECOVERY_PASS || "";
+
+      if (taskUser && taskPass) {
+        // 跨 Session：指定用户 + 交互模式
+        const today = new Date().toISOString().slice(0, 10).replace(/-/g, "/");
+        execSync(
+          `schtasks /create /tn "${taskName}" /tr "msedge \\"${edgeUrl}\\"" /sc once /sd ${today} /st 00:00 /ru "${taskUser}" /rp "${taskPass}" /it /f`,
+          { shell: "cmd.exe", timeout: 10000 },
+        );
+      } else {
+        // 回退：无凭证时只能当前 Session（如果 webserver 从 RDP 启动才能看到）
+        console.error(
+          "[shop-scraper] Windows: 未设置 CAPTCHA_RECOVERY_USER/PASS，计划任务仅限当前 Session",
+        );
+        execSync(
+          `schtasks /create /tn "${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 {}
     }
-  }
 
-  console.error("[shop-scraper] Windows: 等待验证码超时");
-  return false;
+    // 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;
+  }
 }
 
 /**