Let future sessions wait on archive completion without manual polling

Constraint: The real FMA archive still needs time, but once it finishes the workflow should transition into extraction and readiness with minimal operator attention
Rejected: Keep completion detection as an entirely manual loop | Wastes attention and slows handoff at the exact moment the archive becomes useful
Confidence: high
Scope-risk: narrow
Directive: Use wait_for_fma_and_prepare.py as the passive bridge from long-running download to active dataset onboarding whenever unattended waiting is acceptable
Tested: /usr/local/miniconda3/bin/python -m py_compile acr-engine/scripts/wait_for_fma_and_prepare.py; /usr/local/miniconda3/bin/python acr-engine/scripts/wait_for_fma_and_prepare.py --interval 2 --max-cycles 2
Not-tested: The completed-path handoff into extraction remains pending full archive completion

+#!/usr/bin/env python3
+"""Wait for the FMA archive to finish, then run post-download readiness."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import subprocess
+import time
+
+PYTHON = "/usr/local/miniconda3/bin/python"
+INSPECT = [PYTHON, "scripts/prepare_fma_archive.py", "inspect"]
+POST = [PYTHON, "scripts/fma_postdownload_ready.py"]
+
+
+def inspect() -> dict:
+    return json.loads(subprocess.check_output(INSPECT, text=True))
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--interval", type=float, default=30.0)
+    parser.add_argument("--max-cycles", type=int, default=3)
+    args = parser.parse_args()
+
+    snapshots = []
+    for _ in range(args.max_cycles):
+        snap = inspect()
+        snapshots.append(snap)
+        if snap.get("archive_size", 0) >= snap.get("archive_bytes_expected", 0):
+            result = json.loads(subprocess.check_output(POST, text=True))
+            print(json.dumps({"status": "completed", "snapshots": snapshots, "postdownload": result}, indent=2, ensure_ascii=False))
+            return
+        time.sleep(args.interval)
+
+    print(json.dumps({"status": "waiting", "snapshots": snapshots}, indent=2, ensure_ascii=False))
+
+
+if __name__ == "__main__":
+    main()

@@ -237,6 +237,30 @@
 
 
 
+
+### Stage: FMA 完成前等待并自动切换
+
+完成项：
+- 新增 [acr-engine/scripts/wait_for_fma_and_prepare.py](../acr-engine/scripts/wait_for_fma_and_prepare.py)
+- 支持：
+  - 周期性检查 FMA archive 是否已完整
+  - 完整后自动调用 `fma_postdownload_ready.py`
+  - 未完成时返回结构化 `waiting` 快照
+- 将脚本接入 [docs/open-dataset-workflow.md](./open-dataset-workflow.md) 与 [docs/session-handoff.md](./session-handoff.md)
+
+验证结果：
+- `/usr/local/miniconda3/bin/python -m py_compile scripts/wait_for_fma_and_prepare.py` 成功
+- `/usr/local/miniconda3/bin/python scripts/wait_for_fma_and_prepare.py --interval 2 --max-cycles 2` 成功
+- 当前结果：
+  - 第 1 次快照 `archive_size=2110291968`
+  - 第 2 次快照 `archive_size=2115256320`
+  - `status=waiting`
+  - 最新进度 `27.5439%`
+
+结论：
+- 现在仓库已经具备“等待完成 -> 自动切入解压/就绪检查”的衔接能力
+- 后续 session 可以用单命令挂起等待，而不是反复手工轮询
+
 ### Stage: FMA 下载完成后自动就绪
 
 完成项：

@@ -147,6 +147,20 @@ cd acr-engine
 
 如果归档还没下完，会返回结构化 `archive_not_complete`。
 
+
+### FMA 完成前等待并自动切换
+
+```bash
+cd acr-engine
+/usr/local/miniconda3/bin/python scripts/wait_for_fma_and_prepare.py --interval 30 --max-cycles 120
+```
+
+作用：
+
+- 周期性检查 `fma_small.zip` 是否完成
+- 一旦完成，自动进入 [scripts/fma_postdownload_ready.py](../acr-engine/scripts/fma_postdownload_ready.py)
+- 如果还没完成，则返回 `waiting` 和最近的进度快照
+
 ## Sources
 - See [dataset-spec.md](./dataset-spec.md)
 - See [dataset-sources-and-licensing.md](./dataset-sources-and-licensing.md)

@@ -319,3 +319,5 @@
 - [CHANGELOG.md](./CHANGELOG.md)
 
 - FMA 下载完成后可直接执行：[acr-engine/scripts/fma_postdownload_ready.py](../acr-engine/scripts/fma_postdownload_ready.py)
+
+- 若需要等待下载完成并自动切到解压/就绪检查，可直接执行：[acr-engine/scripts/wait_for_fma_and_prepare.py](../acr-engine/scripts/wait_for_fma_and_prepare.py)