feat(pull_swe_bench_images): support comma-separated --dataset

- Allow --dataset swe_bench_verified,swe_bench_lite to pre-pull
  multiple SWE-bench variants in one run.
- Validate dataset keys before loading.
- Update myread.md with multi-dataset example.

bash/pull_swe_bench_images.py

@@ -111,6 +111,25 @@ def load_swe_samples(dataset_id: str):
     return loader.load()
 
 
+def load_multiple_datasets(dataset_keys: list[str]) -> list:
+    """Load samples from one or more dataset keys and tag each sample with its source."""
+    all_samples = []
+    for key in dataset_keys:
+        dataset_id = DATASET_IDS[key]
+        samples = load_swe_samples(dataset_id)
+        # Tag each sample so get_image_name can still work and state stays per-dataset.
+        for sample in samples:
+            if isinstance(sample, dict):
+                sample.setdefault('dataset_key', key)
+            else:
+                metadata = getattr(sample, 'metadata', None)
+                if metadata is not None:
+                    metadata.setdefault('dataset_key', key)
+        all_samples.extend(samples)
+        print(f'   {key}: {len(samples)} 个样本')
+    return all_samples
+
+
 def get_image_name(instance, namespace: str, arch: str) -> str:
     """根据 swebench 规则生成远程镜像名称。"""
     if isinstance(instance, dict):
@@ -217,8 +236,8 @@ def main():
     parser.add_argument(
         '--dataset',
         default='swe_bench_verified',
-        choices=list(DATASET_IDS.keys()),
-        help='SWE-bench dataset variant',
+        type=lambda s: [x.strip() for x in s.split(',') if x.strip()],
+        help='SWE-bench dataset variant(s), comma-separated, e.g. swe_bench_verified or swe_bench_verified,swe_bench_lite',
     )
     parser.add_argument(
         '--max-workers',
@@ -264,14 +283,19 @@ def main():
         state['failed'] = []
         print(f'🔄 重试 {len(failed)} 个之前失败的镜像')
 
-    dataset_id = DATASET_IDS[args.dataset]
-    samples = load_swe_samples(dataset_id)
+    # Validate dataset keys
+    for key in args.dataset:
+        if key not in DATASET_IDS:
+            print(f'ERROR: unknown dataset "{key}". Supported: {list(DATASET_IDS.keys())}')
+            sys.exit(1)
+
+    samples = load_multiple_datasets(args.dataset)
 
     if args.dry_run > 0:
         samples = samples[:args.dry_run]
         print(f'⚠️  测试模式：仅处理前 {args.dry_run} 个样本以验证镜像名和网络...')
     else:
-        print(f'✅ 成功加载 {len(samples)} 个样本。')
+        print(f'✅ 成功加载 {len(samples)} 个样本（数据集: {args.dataset}）。')
 
     already_done = len(state['done']) + len(state['skipped'])
     print(f'📂 进度文件: {get_state_path(output_dir)}')

myread.md

@@ -243,6 +243,11 @@ export PYTHONUNBUFFERED=1
 python bash/pull_swe_bench_images.py \
   --dataset swe_bench_verified \
   --max-workers 1
+
+# 也可以同时预拉多个数据集（逗号分隔）
+python bash/pull_swe_bench_images.py \
+  --dataset swe_bench_verified,swe_bench_lite \
+  --max-workers 1
 ```
 
 脚本特性：
@@ -250,6 +255,7 @@ python bash/pull_swe_bench_images.py \
 - 断点续传：状态文件 `output/.swe_bench_pull_state.json`，中断后再次运行会自动跳过已成功的镜像。
 - 本地已有镜像自动跳过。
 - 单 worker 模式稳定，减少限流概率；若网络非常好，可适当提高 `--max-workers`。
+- `--dataset` 支持逗号分隔的多个数据集。
 
 镜像会存到 Docker `data-root` 配置的路径：