tt-help-cli-ycl 1.3.99 → 1.4.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.3.99",
+  "version": "1.4.0",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/tag.js

@@ -738,7 +738,9 @@ 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();
@@ -957,6 +959,7 @@ 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);
@@ -975,12 +978,29 @@ 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,
@@ -990,6 +1010,18 @@ 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/watch/data-store.js

@@ -1160,59 +1160,112 @@ export function createStore(filePath, options = {}) {
       })();
     }
 
-    // 常规移动：INSERT + DELETE 事务
-    const moveTxn = getDb().transaction(() => {
-      getDb()
-        .prepare(
-          `
-        INSERT OR IGNORE INTO jobs (
-          unique_id, nickname, status, sources, pinned,
-          tt_seller, verified, video_count, comment_count,
-          guessed_location, location_created, confirmed_location,
-          follower_count, following_count, heart_count,
-          created_at, updated_at, region, signature, bio_link, sec_uid,
-          status_code, latest_video_time, user_create_time
-        )
-        SELECT
-          unique_id, nickname, 'pending', sources, pinned,
-          tt_seller, verified, video_count, comment_count,
-          guessed_location, location_created, confirmed_location,
-          follower_count, following_count, heart_count,
-          created_at, updated_at, region, signature, bio_link, sec_uid,
-          status_code, latest_video_time, user_create_time
-        FROM raw_jobs
-        WHERE ${whereSql}
+    // 常规移动：多国家时先按国家均衡补充，再用全局兜底补齐剩余额度
+    const insertFromRawSql = `
+      INSERT OR IGNORE INTO jobs (
+        unique_id, nickname, status, sources, pinned,
+        tt_seller, verified, video_count, comment_count,
+        guessed_location, location_created, confirmed_location,
+        follower_count, following_count, heart_count,
+        created_at, updated_at, region, signature, bio_link, sec_uid,
+        status_code, latest_video_time, user_create_time
+      )
+      SELECT
+        unique_id, nickname, 'pending', sources, pinned,
+        tt_seller, verified, video_count, comment_count,
+        guessed_location, location_created, confirmed_location,
+        follower_count, following_count, heart_count,
+        created_at, updated_at, region, signature, bio_link, sec_uid,
+        status_code, latest_video_time, user_create_time
+      FROM raw_jobs
+      WHERE __WHERE__
+      ORDER BY
+        CASE WHEN sources LIKE '%tag%' THEN 0 ELSE 1 END,
+        COALESCE(video_count, 0) DESC, created_at DESC
+      LIMIT ?
+    `;
+
+    const deleteFromRawSql = `
+      DELETE FROM raw_jobs
+      WHERE unique_id IN (
+        SELECT unique_id FROM raw_jobs
+        WHERE __WHERE__
         ORDER BY
           CASE WHEN sources LIKE '%tag%' THEN 0 ELSE 1 END,
           COALESCE(video_count, 0) DESC, created_at DESC
         LIMIT ?
-      `,
-        )
-        .run(...args, safeLimit);
+      )
+    `;
 
-      // 删除已移动的记录：用子查询匹配刚 INSERT 的 unique_id
-      getDb()
-        .prepare(
-          `
-        DELETE FROM raw_jobs
-        WHERE unique_id IN (
-          SELECT unique_id FROM raw_jobs
-          WHERE ${whereSql}
-          ORDER BY
-            CASE WHEN sources LIKE '%tag%' THEN 0 ELSE 1 END,
-            COALESCE(video_count, 0) DESC, created_at DESC
-          LIMIT ?
-        )
-      `,
-        )
-        .run(...args, safeLimit);
+    const uniqueLocations = normalizedLocations
+      ? Array.from(new Set(normalizedLocations))
+      : [];
+    const shouldBalanceByCountry = uniqueLocations.length > 1;
+
+    const moveTxn = getDb().transaction(() => {
+      let moved = 0;
+      const movedByCountry = {};
+
+      if (shouldBalanceByCountry) {
+        const baseQuota = Math.floor(safeLimit / uniqueLocations.length);
+        const remainder = safeLimit % uniqueLocations.length;
+
+        for (let i = 0; i < uniqueLocations.length; i++) {
+          if (moved >= safeLimit) break;
+
+          const location = uniqueLocations[i];
+          const quota = baseQuota + (i < remainder ? 1 : 0);
+          const currentLimit = Math.max(0, Math.min(quota, safeLimit - moved));
+          if (!currentLimit) continue;
+
+          const locationWhere = `${whereSql} AND UPPER(COALESCE(guessed_location, '')) = ?`;
+          const locationArgs = [...args, location];
+
+          getDb()
+            .prepare(insertFromRawSql.replace("__WHERE__", locationWhere))
+            .run(...locationArgs, currentLimit);
+          const del = getDb()
+            .prepare(deleteFromRawSql.replace("__WHERE__", locationWhere))
+            .run(...locationArgs, currentLimit);
+
+          const movedThisCountry = del?.changes || 0;
+          moved += movedThisCountry;
+          movedByCountry[location] = movedThisCountry;
+        }
+
+        // 某些国家库存不足时，用全局查询补齐剩余额度（仍受 whereSql 国家范围约束）
+        const remaining = safeLimit - moved;
+        if (remaining > 0) {
+          getDb()
+            .prepare(insertFromRawSql.replace("__WHERE__", whereSql))
+            .run(...args, remaining);
+          const del = getDb()
+            .prepare(deleteFromRawSql.replace("__WHERE__", whereSql))
+            .run(...args, remaining);
+          moved += del?.changes || 0;
+        }
+
+        console.error(
+          `[data-store] refill 国家均衡: ${uniqueLocations
+            .map((loc) => `${loc}:${movedByCountry[loc] || 0}`)
+            .join(", ")} | total=${moved}`,
+        );
+      } else {
+        getDb()
+          .prepare(insertFromRawSql.replace("__WHERE__", whereSql))
+          .run(...args, safeLimit);
+        const del = getDb()
+          .prepare(deleteFromRawSql.replace("__WHERE__", whereSql))
+          .run(...args, safeLimit);
+        moved = del?.changes || 0;
+      }
+
+      return moved;
     });
 
-    moveTxn();
+    const moved = moveTxn();
     markStatsDirty();
-
-    const actualMoved = Math.min(count, safeLimit);
-    return { moved: actualMoved };
+    return { moved };
   }
 
   async function claimNextJob(
@@ -1512,8 +1565,7 @@ export function createStore(filePath, options = {}) {
         }
         return null;
       }
-      const refillResult = (async () => {
-        refillLock = Promise.resolve(); // 占位
+      const refillPromise = (async () => {
         const result = refillJobsFromRaw(
           normalizedLocations.length ? normalizedLocations : null,
           500,
@@ -1521,30 +1573,17 @@ export function createStore(filePath, options = {}) {
         );
         // refillJobsFromRaw 在 LLM 模式下返回 Promise
         if (result && typeof result.then === "function") {
-          return result.finally(() => {
-            refillLock = null;
-          });
+          return await result;
         }
         return result;
       })();
-      if (refillResult && typeof refillResult.then === "function") {
-        const awaited = await refillResult;
-        if (awaited.moved > 0) {
-          console.error(
-            `[data-store] 从 raw_jobs 补充了 ${awaited.moved} 条任务到 jobs`,
-          );
-          for (const requireVideo of [true, false]) {
-            const pinned = findPinnedPending(requireVideo);
-            if (pinned) {
-              return claimRow(pinned);
-            }
-            const ranked = findPrioritizedPending(requireVideo);
-            if (ranked) {
-              return claimRow(ranked);
-            }
-          }
-        }
-      } else if (refillResult.moved > 0) {
+      // 让并发请求等待同一个 refill，并且无论成功/失败都释放锁
+      refillLock = refillPromise.finally(() => {
+        refillLock = null;
+      });
+
+      const refillResult = await refillLock;
+      if (refillResult.moved > 0) {
         console.error(
           `[data-store] 从 raw_jobs 补充了 ${refillResult.moved} 条任务到 jobs`,
         );