Add missing fuller logging launcher

code/scripts/launch_oven_fuller_logging_batch.py

@@ -0,0 +1,336 @@
+from pathlib import Path
+import argparse
+import json
+import os
+import shutil
+import subprocess
+import sys
+import time
+from typing import Dict, List, Optional, Sequence
+
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+if str(PROJECT_ROOT) not in sys.path:
+    sys.path.insert(0, str(PROJECT_ROOT))
+
+
+def _configure_thread_env() -> None:
+    defaults = {
+        "OMP_NUM_THREADS": "1",
+        "OPENBLAS_NUM_THREADS": "1",
+        "MKL_NUM_THREADS": "1",
+        "NUMEXPR_NUM_THREADS": "1",
+        "VECLIB_MAXIMUM_THREADS": "1",
+        "BLIS_NUM_THREADS": "1",
+        "MALLOC_ARENA_MAX": "2",
+    }
+    for key, value in defaults.items():
+        os.environ.setdefault(key, value)
+
+
+def _configure_coppeliasim_env() -> None:
+    coppeliasim_root = os.environ.setdefault("COPPELIASIM_ROOT", "/workspace/coppelia_sim")
+    ld_library_path_parts = [
+        part for part in os.environ.get("LD_LIBRARY_PATH", "").split(":") if part
+    ]
+    if coppeliasim_root not in ld_library_path_parts:
+        ld_library_path_parts.insert(0, coppeliasim_root)
+    os.environ["LD_LIBRARY_PATH"] = ":".join(ld_library_path_parts)
+
+
+_configure_thread_env()
+_configure_coppeliasim_env()
+
+from rr_label_study.oven_study import _aggregate_summary, _episode_dirs
+
+
+def _select_episode_indices(
+    total_episodes: int,
+    episode_offset: int,
+    max_episodes: Optional[int],
+    episode_indices: Optional[Sequence[int]],
+) -> List[int]:
+    if episode_indices is not None:
+        selected: List[int] = []
+        seen = set()
+        for raw_index in episode_indices:
+            episode_index = int(raw_index)
+            if not (0 <= episode_index < total_episodes):
+                raise ValueError(
+                    f"episode index {episode_index} outside available range 0..{total_episodes - 1}"
+                )
+            if episode_index in seen:
+                continue
+            selected.append(episode_index)
+            seen.add(episode_index)
+        return selected
+
+    remaining = max(0, total_episodes - episode_offset)
+    if max_episodes is not None:
+        remaining = min(remaining, max_episodes)
+    if remaining <= 0:
+        return []
+    return list(range(episode_offset, episode_offset + remaining))
+
+
+def _is_complete_episode_dir(output_dir: Path, episode_name: str) -> bool:
+    required = [
+        output_dir.joinpath(f"{episode_name}.dense.csv"),
+        output_dir.joinpath(f"{episode_name}.keyframes.csv"),
+        output_dir.joinpath(f"{episode_name}.debug.jsonl"),
+        output_dir.joinpath(f"{episode_name}.metrics.json"),
+        output_dir.joinpath("summary.json"),
+        output_dir.joinpath("templates.json"),
+        output_dir.joinpath("templates.pkl"),
+    ]
+    return all(path.exists() for path in required)
+
+
+def _load_metrics(result_dir: Path, episode_names: Sequence[str]) -> List[Dict[str, object]]:
+    metrics: List[Dict[str, object]] = []
+    for episode_name in episode_names:
+        metrics_path = result_dir.joinpath(episode_name, f"{episode_name}.metrics.json")
+        if metrics_path.exists():
+            with metrics_path.open("r", encoding="utf-8") as handle:
+                metrics.append(json.load(handle))
+    return metrics
+
+
+def _write_json(path: Path, payload: Dict[str, object]) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with path.open("w", encoding="utf-8") as handle:
+        json.dump(payload, handle, indent=2)
+
+
+def _run_episode(
+    dataset_root: Path,
+    episode_dir: Path,
+    output_dir: Path,
+    checkpoint_stride: int,
+    num_workers: int,
+    base_display: int,
+    templates_json: Path,
+    stagger_seconds: float,
+    thread_count: int,
+    log_path: Path,
+) -> int:
+    env = os.environ.copy()
+    thread_count_str = str(thread_count)
+    env["OMP_NUM_THREADS"] = thread_count_str
+    env["OPENBLAS_NUM_THREADS"] = thread_count_str
+    env["MKL_NUM_THREADS"] = thread_count_str
+    env["NUMEXPR_NUM_THREADS"] = thread_count_str
+    env["VECLIB_MAXIMUM_THREADS"] = thread_count_str
+    env["BLIS_NUM_THREADS"] = thread_count_str
+    env["MALLOC_ARENA_MAX"] = "2"
+    env["PYTHONUNBUFFERED"] = "1"
+    with log_path.open("w", encoding="utf-8") as log_handle:
+        process = subprocess.Popen(
+            [
+                sys.executable,
+                str(PROJECT_ROOT.joinpath("scripts", "recompute_oven_episode_parallel.py")),
+                "--dataset-root",
+                str(dataset_root),
+                "--episode-dir",
+                str(episode_dir),
+                "--output-dir",
+                str(output_dir),
+                "--checkpoint-stride",
+                str(checkpoint_stride),
+                "--num-workers",
+                str(num_workers),
+                "--base-display",
+                str(base_display),
+                "--templates-json",
+                str(templates_json),
+                "--stagger-seconds",
+                str(stagger_seconds),
+            ],
+            stdout=log_handle,
+            stderr=subprocess.STDOUT,
+            cwd=str(PROJECT_ROOT),
+            env=env,
+        )
+        return process.wait()
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--dataset-root",
+        default="/workspace/data/bimanual_take_tray_out_of_oven_train_128",
+    )
+    parser.add_argument("--result-dir", required=True)
+    parser.add_argument("--templates-json", required=True)
+    parser.add_argument("--episode-offset", type=int, default=0)
+    parser.add_argument("--max-episodes", type=int, default=100)
+    parser.add_argument("--episode-indices")
+    parser.add_argument("--checkpoint-stride", type=int, default=16)
+    parser.add_argument("--num-workers", type=int, default=24)
+    parser.add_argument("--base-display", type=int, default=900)
+    parser.add_argument("--stagger-seconds", type=float, default=0.15)
+    parser.add_argument("--thread-count", type=int, default=1)
+    parser.add_argument("--max-retries", type=int, default=2)
+    args = parser.parse_args()
+
+    dataset_root = Path(args.dataset_root)
+    result_dir = Path(args.result_dir)
+    result_dir.mkdir(parents=True, exist_ok=True)
+    templates_json = Path(args.templates_json)
+    if not templates_json.exists():
+        raise FileNotFoundError(f"missing templates json: {templates_json}")
+
+    all_episode_dirs = _episode_dirs(dataset_root)
+    explicit_episode_indices = None
+    if args.episode_indices:
+        explicit_episode_indices = [
+            int(chunk.strip()) for chunk in args.episode_indices.split(",") if chunk.strip()
+        ]
+    selected_episode_indices = _select_episode_indices(
+        total_episodes=len(all_episode_dirs),
+        episode_offset=args.episode_offset,
+        max_episodes=args.max_episodes,
+        episode_indices=explicit_episode_indices,
+    )
+    selected_episode_names = [f"episode{index}" for index in selected_episode_indices]
+
+    manifest = {
+        "dataset_root": str(dataset_root.resolve()),
+        "result_dir": str(result_dir.resolve()),
+        "templates_json": str(templates_json.resolve()),
+        "episode_indices": selected_episode_indices,
+        "checkpoint_stride": args.checkpoint_stride,
+        "num_workers": args.num_workers,
+        "base_display": args.base_display,
+        "stagger_seconds": args.stagger_seconds,
+        "thread_count": args.thread_count,
+        "max_retries": args.max_retries,
+        "started_at_epoch": time.time(),
+    }
+    _write_json(result_dir.joinpath("run_manifest.json"), manifest)
+
+    progress_path = result_dir.joinpath("progress.json")
+    logs_dir = result_dir.joinpath("logs")
+    logs_dir.mkdir(parents=True, exist_ok=True)
+
+    completed: List[int] = []
+    failed: List[Dict[str, object]] = []
+
+    for episode_index in selected_episode_indices:
+        episode_name = f"episode{episode_index}"
+        episode_dir = all_episode_dirs[episode_index]
+        final_output_dir = result_dir.joinpath(episode_name)
+        if _is_complete_episode_dir(final_output_dir, episode_name):
+            completed.append(episode_index)
+            _write_json(
+                progress_path,
+                {
+                    "current_episode": None,
+                    "completed_episode_indices": completed,
+                    "failed": failed,
+                    "total_selected": len(selected_episode_indices),
+                    "updated_at_epoch": time.time(),
+                },
+            )
+            continue
+
+        attempt_success = False
+        current_failure: Optional[Dict[str, object]] = None
+        for attempt_index in range(1, args.max_retries + 2):
+            temp_output_dir = result_dir.joinpath(f".{episode_name}.tmp")
+            if temp_output_dir.exists():
+                shutil.rmtree(temp_output_dir)
+            log_path = logs_dir.joinpath(f"{episode_name}.attempt{attempt_index:02d}.log")
+            _write_json(
+                progress_path,
+                {
+                    "current_episode": episode_name,
+                    "current_attempt": attempt_index,
+                    "completed_episode_indices": completed,
+                    "failed": failed,
+                    "total_selected": len(selected_episode_indices),
+                    "updated_at_epoch": time.time(),
+                },
+            )
+            return_code = _run_episode(
+                dataset_root=dataset_root,
+                episode_dir=episode_dir,
+                output_dir=temp_output_dir,
+                checkpoint_stride=args.checkpoint_stride,
+                num_workers=args.num_workers,
+                base_display=args.base_display,
+                templates_json=templates_json,
+                stagger_seconds=args.stagger_seconds,
+                thread_count=args.thread_count,
+                log_path=log_path,
+            )
+            if return_code == 0 and _is_complete_episode_dir(temp_output_dir, episode_name):
+                if final_output_dir.exists():
+                    shutil.rmtree(final_output_dir)
+                temp_output_dir.rename(final_output_dir)
+                completed.append(episode_index)
+                attempt_success = True
+                current_failure = None
+                break
+
+            current_failure = {
+                "episode_index": episode_index,
+                "episode_name": episode_name,
+                "attempt": attempt_index,
+                "return_code": return_code,
+                "log_path": str(log_path),
+                "updated_at_epoch": time.time(),
+            }
+            if temp_output_dir.exists():
+                failed_dir = result_dir.joinpath("failed_attempts", f"{episode_name}.attempt{attempt_index:02d}")
+                failed_dir.parent.mkdir(parents=True, exist_ok=True)
+                if failed_dir.exists():
+                    shutil.rmtree(failed_dir)
+                temp_output_dir.rename(failed_dir)
+
+        if not attempt_success:
+            failed.append(current_failure or {"episode_index": episode_index, "episode_name": episode_name})
+            _write_json(
+                progress_path,
+                {
+                    "current_episode": None,
+                    "completed_episode_indices": completed,
+                    "failed": failed,
+                    "total_selected": len(selected_episode_indices),
+                    "updated_at_epoch": time.time(),
+                },
+            )
+            raise RuntimeError(f"failed to produce complete result for {episode_name}")
+
+        metrics = _load_metrics(result_dir, selected_episode_names)
+        if metrics:
+            _write_json(result_dir.joinpath("summary.json"), _aggregate_summary(metrics))
+        _write_json(
+            progress_path,
+            {
+                "current_episode": None,
+                "completed_episode_indices": completed,
+                "failed": failed,
+                "total_selected": len(selected_episode_indices),
+                "updated_at_epoch": time.time(),
+            },
+        )
+
+    metrics = _load_metrics(result_dir, selected_episode_names)
+    if metrics:
+        _write_json(result_dir.joinpath("summary.json"), _aggregate_summary(metrics))
+    _write_json(
+        progress_path,
+        {
+            "current_episode": None,
+            "completed_episode_indices": completed,
+            "failed": failed,
+            "total_selected": len(selected_episode_indices),
+            "finished_at_epoch": time.time(),
+        },
+    )
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())