feature: scripts/export_run.py 用来导出运行结果

scripts/export_run.py

@@ -65,8 +65,12 @@ def render_input(model_input: dict[str, Any]) -> list[str]:
     return sections or ["无", ""]
 
 
-def is_completed_pass(result: dict[str, Any]) -> bool:
-    return result.get("execution_status") == "completed" and result.get("verdict") == "pass"
+def matches_result(
+    result: dict[str, Any], execution_status: str | None, verdict: str | None
+) -> bool:
+    return (execution_status is None or result.get("execution_status") == execution_status) and (
+        verdict is None or result.get("verdict") == verdict
+    )
 
 
 def render_markdown(run: dict[str, Any], results: list[dict[str, Any]]) -> str:
@@ -83,9 +87,9 @@ def render_markdown(run: dict[str, Any], results: list[dict[str, Any]]) -> str:
         f"- 开始时间：{run.get('started_at', '')}",
         f"- 完成时间：{run.get('finished_at', '')}",
         "",
-        f"- 本报告 Pass 样例：{len(results)}",
+        f"- 本报告导出样例：{len(results)}",
         "",
-        "## Pass 样例详情",
+        "## 测试样例详情",
         "",
     ]
     for index, result in enumerate(results, 1):
@@ -133,6 +137,16 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument("--run-id", type=int, default=7, help="要导出的 run ID（默认：7）")
     parser.add_argument("--base-url", default="http://127.0.0.1:8000", help="API 服务地址")
     parser.add_argument("--username", help="登录用户名；省略时交互输入")
+    parser.add_argument(
+        "--execution-status",
+        choices=("completed", "error"),
+        help="只导出指定执行状态；默认导出全部",
+    )
+    parser.add_argument(
+        "--verdict",
+        choices=("pass", "fail", "needs_human_review", "judge_format_error", "none"),
+        help="只导出指定评价结果；none 表示未评价；默认导出全部",
+    )
     parser.add_argument("--output", type=Path, help="输出文件路径")
     return parser.parse_args()
 
@@ -157,8 +171,18 @@ def main() -> int:
         results = request_json(
             args.base_url, f"/api/v1/runs/{args.run_id}/results", token=token
         )
-        results = [result for result in results if is_completed_pass(result)]
-        output = args.output or Path("outputs") / f"run_{args.run_id}_pass_results.md"
+        verdict = None if args.verdict is None else args.verdict
+        results = [
+            result
+            for result in results
+            if matches_result(
+                result,
+                args.execution_status,
+                None if verdict == "none" else verdict,
+            )
+            and (verdict != "none" or result.get("verdict") is None)
+        ]
+        output = args.output or Path("outputs") / f"run_{args.run_id}_results.md"
         output.parent.mkdir(parents=True, exist_ok=True)
         output.write_text(render_markdown(run, results), encoding="utf-8")
     except (RuntimeError, KeyError, OSError) as exc: