feat(data-process): 返回任务文档数量

backend/tests/test_data_process_store.py

@@ -316,6 +316,54 @@ class _TaskDetailStore(DataProcessStore):
         yield self._conn
 
 
+class _TaskListConnection:
+    def __init__(self) -> None:
+        self.task = {
+            **_regeneration_task(
+                status="pending",
+                output_dataset_id=None,
+                output_count=17,
+            ),
+            "created_at": "2026-07-25T18:00:00Z",
+            "deleted_at": None,
+        }
+        self.sources = [
+            {"task_id": "task-1", "deleted_at": None},
+            {"task_id": "task-1", "deleted_at": None},
+            {"task_id": "task-1", "deleted_at": "2026-07-26T00:00:00Z"},
+        ]
+
+    def execute(self, sql: str, params: Any = None) -> _Result:
+        normalized = " ".join(sql.split())
+        if normalized.startswith("SELECT COUNT(*) AS count FROM data_process_tasks task"):
+            assert "task.deleted_at IS NULL" in normalized
+            return _Result(row={"count": 1})
+        assert normalized.startswith("SELECT task.*")
+        assert "source_file.task_id=task.id" in normalized
+        assert "source_file.deleted_at IS NULL" in normalized
+        source_file_count = sum(
+            item["task_id"] == self.task["id"] and item["deleted_at"] is None
+            for item in self.sources
+        )
+        return _Result(
+            rows=[
+                {
+                    **self.task,
+                    "source_file_count": source_file_count,
+                }
+            ]
+        )
+
+
+class _TaskListStore(DataProcessStore):
+    def __init__(self, conn: _TaskListConnection) -> None:
+        self._conn = conn
+
+    @contextmanager
+    def connect(self) -> Iterator[_TaskListConnection]:
+        yield self._conn
+
+
 def test_decode_row_serializes_postgres_numeric_values_as_json_numbers() -> None:
     decoded = _decode_row(
         {
@@ -461,6 +509,19 @@ def _regeneration_task(**updates: Any) -> dict[str, Any]:
     return task
 
 
+def test_list_tasks_exposes_source_and_generation_counts() -> None:
+    page = _TaskListStore(_TaskListConnection()).list_tasks(page=1, page_size=20)
+
+    assert page["total"] == 1
+    assert page["page"] == 1
+    assert page["page_size"] == 20
+    item = page["items"][0]
+    assert item["status"] == "pending"
+    assert item["source_file_count"] == 2
+    assert item["output_count"] == 17
+    assert item["output_dataset_id"] is None
+
+
 def _legacy_published_datasets(task_id: str = "task-1") -> list[dict[str, Any]]:
     specs = (
         ("dataset_train", "制度问答-训练集", "train", "train", 22),