tt-help-cli-ycl 1.3.100 → 1.4.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.3.100",
+  "version": "1.4.1",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/tag.js

@@ -5,7 +5,6 @@ import { fetchTagData, enrichVideosWithLocation } from "../lib/tag-fetcher.js";
 import { killEdgeProcesses, ensureBrowserReady } from "../lib/browser/cdp.js";
 import { getOrCreatePage } from "../lib/browser/page.js";
 import { TikTokScraper } from "../lib/tiktok-scraper.mjs";
-import { CDNBlockedError } from "../lib/parse-ssr.mjs";
 import {
   DEFAULT_TARGET_LOCATIONS,
   isLocationInList,
@@ -43,18 +42,6 @@ function formatMemoryUsage(mem = process.memoryUsage()) {
   return `rss:${(mem.rss / 1024 / 1024).toFixed(0)}MB heap:${(mem.heapUsed / 1024 / 1024).toFixed(0)}MB ext:${(mem.external / 1024 / 1024).toFixed(0)}MB ab:${(mem.arrayBuffers / 1024 / 1024).toFixed(0)}MB`;
 }
 
-function getCdnCooldownSeconds(blockedCount, totalCount, isTooManyRequests = false) {
-  if (isTooManyRequests) return 120;
-  const ratio = blockedCount / Math.max(totalCount, 1);
-  return ratio > 0.3 ? 120 : 60;
-}
-
-async function cooldownAndRecycle(cooldownSec, recyclePage, maybeRecycleForMemory) {
-  await new Promise((r) => setTimeout(r, cooldownSec * 1000));
-  await recyclePage();
-  await maybeRecycleForMemory();
-}
-
 // 构建带客户端追踪 header 的 fetch 封装
 function buildClientHeaders(clientId, meta, extra = {}) {
   return {
@@ -642,10 +629,8 @@ export async function handleScoreAll(parsed) {
 
   await setupPageRequestBlocking(page);
 
-  const portPoolStart = Math.max(
-    1,
-    parseInt(process.env.TAG_SCOREALL_PORT_POOL_START || "7222", 10) || 7222,
-  );
+  // 端口池起始以 CLI --port / 环境变量 / 默认 7222 为准
+  const portPoolStart = cdpPort;
   const portPoolSize = Math.max(
     2,
     parseInt(process.env.TAG_SCOREALL_PORT_POOL_SIZE || "10", 10) || 10,
@@ -751,7 +736,9 @@ export async function handleScoreAll(parsed) {
 
   let totalScored = 0;
   let emptyRounds = 0; // 连续无任务的轮数
+  let consecutive403 = 0; // 连续 HTTP 403 次数，达 3 次认为被封
   const DISCOVER_AFTER_EMPTY = 3; // 连续 3 轮无任务时触发 discover
+  const CONSECUTIVE_403_BLOCKED = 3; // 连续 3 次 403 认为账号被封
 
   // 生成客户端 ID，用于服务端追踪
   const clientId = randomUUID();
@@ -919,11 +906,12 @@ export async function handleScoreAll(parsed) {
         // CDN 限流检测：有拦截则冷却 + 重启 scraper
         const cdnBlocked = enriched.cdnBlockedCount || 0;
         if (cdnBlocked > 0) {
-          const coolSec = getCdnCooldownSeconds(cdnBlocked, videos.length);
+          const cdnRatio = cdnBlocked / (videos.length || 1);
+          const coolSec = cdnRatio > 0.3 ? 120 : 60;
           log(
-            `  ⚠️ CDN 限流: ${cdnBlocked}/${videos.length} (${((cdnBlocked / Math.max(videos.length, 1)) * 100).toFixed(0)}%)，冷却 ${coolSec} 秒后重启 scraper`,
+            `  ⚠️ CDN 限流: ${cdnBlocked}/${videos.length} (${(cdnRatio * 100).toFixed(0)}%)，冷却 ${coolSec} 秒后重启 scraper`,
           );
-          await cooldownAndRecycle(coolSec, recyclePage, maybeRecycleForMemory);
+          await new Promise((r) => setTimeout(r, coolSec * 1000));
           log(`  正在重启 TikTokScraper...`);
           await enrichScraper.restart();
           log(`  ✅ TikTokScraper 已重启`);
@@ -969,21 +957,11 @@ export async function handleScoreAll(parsed) {
           `  ${icon} ${result.status} score=${result.score} authors=${result.authorCount} matched=${result.matchedAuthors} (${elapsed}s)${mc ? "  " + mc : ""}${memStr}`,
         );
         log("");
+        consecutive403 = 0; // 成功完成，重置 403 计数器
         await recyclePage();
         await maybeRecycleForMemory();
         await randomDelay(3000, 7000);
       } catch (e) {
-        if (e instanceof CDNBlockedError || /HTTP\s+(403|429)/.test(e.message)) {
-          log(`  ⚠️ CDN 被封: ${e.message}`);
-          result.status = "dead";
-          result.error = "cdn_blocked";
-          await reportToServer(baseUrl, result, clientId, clientMeta);
-          totalScored++;
-          const cooldownSec = getCdnCooldownSeconds(1, 1, /429/.test(e.message));
-          log(`  冷却 ${cooldownSec} 秒后再继续...`);
-          await cooldownAndRecycle(cooldownSec, recyclePage, maybeRecycleForMemory);
-          continue;
-        }
         // 区分网络错误和业务错误
         const isNetworkError =
           e.code === "ECONNREFUSED" ||
@@ -998,12 +976,30 @@ export async function handleScoreAll(parsed) {
           await new Promise((r) => setTimeout(r, 15000));
           continue;
         }
+        // 检测 HTTP 403（TikTok 拒绝请求，可能是账号/会话被封）
+        const is403 =
+          e.message &&
+          (e.message.includes("HTTP 403") ||
+            e.message.includes("status 403") ||
+            e.message.includes("status code 403"));
+
+        if (is403) {
+          consecutive403++;
+          log(
+            `  🚫 HTTP 403 — 连续 ${consecutive403}/${CONSECUTIVE_403_BLOCKED} 次`,
+          );
+        } else {
+          consecutive403 = 0;
+        }
+
         log(`  ❌ 失败: ${e.message}`);
         try {
+          // 使用当前正在处理的 tag（tag 变量在 try 块开头已赋值，catch 中仍可访问）
+          const failedTag = typeof tag === "string" && tag ? tag : "unknown";
           await reportToServer(
             baseUrl,
             {
-              tag: "",
+              tag: failedTag,
               status: "error",
               score: 0,
               error: e.message,
@@ -1013,6 +1009,18 @@ export async function handleScoreAll(parsed) {
           );
         } catch {}
         totalScored++;
+
+        // 连续 3 次 403：认为账号被封，等 2 分钟后切换端口
+        if (consecutive403 >= CONSECUTIVE_403_BLOCKED) {
+          log(
+            `  🔐 连续 ${consecutive403} 次 403，账号可能被封！等待 2 分钟后切换端口...`,
+          );
+          await new Promise((r) => setTimeout(r, 120000));
+          await recycleCdpSession("连续 403 封禁");
+          consecutive403 = 0;
+          continue;
+        }
+
         await recyclePage();
         await maybeRecycleForMemory();
       }

package/src/lib/constants.js

@@ -260,8 +260,17 @@ const HELP_TEXT = [
   "    选项:",
   "      --countries <CSV>         目标国家，逗号分隔（默认 13 个欧洲国家）",
   "      -s, --server <URL>        服务端地址（默认 http://127.0.0.1:3000）",
+  "      --port <N>                起始 CDP 端口，端口池默认为 N~N+9（默认 7222）",
+  "      --discover                开启自动发现：连续 3 轮无新标签时自动生成",
+  "    行为:",
+  "      连续 3 次 HTTP 403 → 判断账号被封 → 等 2 分钟后自动切换到池中另一端口",
+  "    环境变量:",
+  "      TAG_SCOREALL_PORT_POOL_SIZE=N  端口池大小（默认 10）",
+  "      TAG_SCOREALL_RECYCLE_RSS_MB=N  内存回收阈值 RSS MB（默认 900）",
+  "      TAG_SCOREALL_RECYCLE_HEAP_MB=N 内存回收阈值 Heap MB（默认 320）",
   "    示例: tt-help tag score-all",
   "          tt-help tag score-all --countries ES -s http://127.0.0.1:3001",
+  "          tt-help tag score-all --port 9222 -s http://127.0.0.1:3001 --discover",
   "",
   "  config  [show|set|unset|reset]",
   "    config                    查看当前配置",

package/src/lib/tag-fetcher.js

@@ -107,9 +107,6 @@ export async function fetchTagData(tag, options = {}) {
       timeout: 30000,
     });
 
-    if (resp.status() === 403 || resp.status() === 429) {
-      throw new CDNBlockedError(`标签页返回 HTTP ${resp.status()}`);
-    }
     if (resp.status() !== 200) {
       throw new Error(`标签页返回 HTTP ${resp.status()}`);
     }