feat(collect): choose product status scope

app/gui/tabs/collect.py

@@ -8,6 +8,7 @@ from ...collect_skip import (
     format_skip_reason_summary,
     normalize_skip_reason_counts,
 )
+from ... import product_status
 from ..models import TaskTableModel
 from ..widgets import *
 from ..workers import CollectWorker as _RealCollectWorker, WriteBackWorker as _RealWriteBackWorker
@@ -28,6 +29,7 @@ COLLECT_ACTIVITY_STEP_LABELS = {
     "prepare_task": "准备采集",
     "open_product": "打开商品页",
     "wait_ready": "等待商品页加载",
+    "read_product_status": "读取商品状态",
     "read_title": "读取标题",
     "read_cover": "读取封面",
     "download_cover": "下载封面",
@@ -718,11 +720,16 @@ class CollectTab(QWidget):
         if not tasks:
             self._set_status("没有可采集任务")
             return
+        collect_scope = self._choose_collect_scope()
+        if collect_scope is None:
+            self._set_status("已取消采集")
+            return
         worker = CollectWorker(
             tasks,
             db_path=self.db_path,
             config=self.config,
             diagnostic_log_dir=diagnostics.DEFAULT_LOG_DIR,
+            collect_scope=collect_scope,
         )
         activity_signal = getattr(worker, "activity", None)
         if activity_signal is not None:
@@ -742,6 +749,32 @@ class CollectTab(QWidget):
         self._start_collect_activity()
         thread.start()
 
+    def _choose_collect_scope(self):
+        box = QMessageBox(self)
+        box.setIcon(QMessageBox.Question)
+        box.setWindowTitle("选择采集范围")
+        box.setText("请选择本轮要采集的商品范围。")
+        box.setInformativeText(
+            "程序会逐个打开商品详情页检测状态并保存结果。"
+            "采集所有状态商品包含未上架、审核中和状态未知商品，"
+            "可能增加采集时间，但本步骤不消耗 AI 点数。"
+        )
+        normal_button = box.addButton("只采集状态正常的商品", QMessageBox.AcceptRole)
+        normal_button.setObjectName("collectNormalOnlyButton")
+        all_button = box.addButton("采集所有状态的商品", QMessageBox.DestructiveRole)
+        all_button.setObjectName("collectAllStatusesButton")
+        all_button.setStyleSheet("color: #cf222e; font-weight: 600;")
+        cancel_button = box.addButton("取消", QMessageBox.RejectRole)
+        cancel_button.setObjectName("collectScopeCancelButton")
+        box.setDefaultButton(normal_button)
+        box.setEscapeButton(cancel_button)
+        box.exec()
+        if box.clickedButton() is normal_button:
+            return product_status.COLLECT_SCOPE_NORMAL_ONLY
+        if box.clickedButton() is all_button:
+            return product_status.COLLECT_SCOPE_ALL
+        return None
+
     def stop_collect(self, checked=False):
         if self.collect_worker is not None:
             self.collect_worker.cancel()
@@ -870,6 +903,17 @@ class CollectTab(QWidget):
             skipped=payload.get("skipped", 0),
             failed=payload.get("failed", 0),
         )
+        status_counts = payload.get("product_status_counts") or {}
+        status_skipped = int(payload.get("status_scope_skipped", 0) or 0)
+        if status_counts:
+            message += "；状态正常{normal}，未上架{unlisted}，审核中{reviewing}，状态未知{unknown}".format(
+                normal=status_counts.get("normal", 0),
+                unlisted=status_counts.get("unlisted", 0),
+                reviewing=status_counts.get("reviewing", 0),
+                unknown=status_counts.get("unknown", 0),
+            )
+        if status_skipped:
+            message += f"；按范围略过{status_skipped}"
         self._show_collect_account_summary(payload, message)
         if payload.get("collected", 0) > 0:
             batch_id = self._active_batch_id()
@@ -917,14 +961,25 @@ class CollectTab(QWidget):
         reused = payload.get("reused_accounts") or []
         login_required = payload.get("login_required_accounts") or []
         skipped = max(0, int(payload.get("skipped", 0) or 0))
+        status_scope_skipped = int(payload.get("status_scope_skipped", 0) or 0)
+        account_skipped = max(0, skipped - status_scope_skipped)
         skip_counts = normalize_skip_reason_counts(
             payload.get("skip_reason_counts"),
-            skipped_total=skipped,
+            skipped_total=account_skipped,
         )
-        if not launched and not reused and not login_required and skipped == 0:
+        if (
+            not launched
+            and not reused
+            and not login_required
+            and account_skipped == 0
+            and status_scope_skipped == 0
+        ):
             return
         lines = [message]
-        skip_summary = format_skip_reason_summary(skip_counts, skipped_total=skipped)
+        skip_summary = format_skip_reason_summary(
+            skip_counts,
+            skipped_total=account_skipped,
+        )
         if skip_summary:
             lines.append(skip_summary)
         if skip_counts[ALIAS_UNMATCHED] > 0:
@@ -941,6 +996,8 @@ class CollectTab(QWidget):
             )
         if skip_counts[LOGIN_REQUIRED] > 0:
             lines.append("请到账号管理完成对应账号登录后，再重新采集略过任务。")
+        if status_scope_skipped:
+            lines.append(f"本轮按范围略过{status_scope_skipped}个非正常状态商品，未下载标题和封面。")
         if launched or reused or login_required:
             lines.append("采集结束后不会自动关闭账号 Chrome，请按需自行关闭。")
         text = "\n".join(lines)