scripts/convert_rl_to_sft_2048_validaction.py

#!/usr/bin/env python3
"""
Convert RL eval trajectories from Game 2048 into LLM SFT-ready chat data.
Matches the specific text format of the RAGEN/Maniskill environment runner.
"""

import argparse
import json
import math
from pathlib import Path
from typing import List, Tuple
import numpy as np
try:
    import yaml # type: ignore
except Exception:
    yaml = None


# 2048 action lookup from config (0..3)
ACTION_LOOKUP_2048 = {0: "Up", 1: "Right", 2: "Down", 3: "Left"}


def load_env_instruction_and_cfg(repo_root: Path) -> Tuple[str, int, str, bool, int]:
    """Load 2048 env instruction and base agent_proxy configs."""
    max_tokens = 8192 
    action_sep = " || " 
    enable_think = True
    max_actions = 700 

    instruction_template = (
        "You are playing the 2048 game on a 4x4 grid. Merge equal tiles by sliding Up, Right, Down, or Left.\n"
        "If a move is invalid (no tiles move), a small penalty is applied. Respond with a single action.\n"
        "Example: <answer>Up</answer>\n\n"
        "Your available actions are:\n"
        "Up, Right, Down, Left\n"
        "You can make up to {max_actions} actions, separated by the action separator \"{action_sep}\""
    )

    if yaml is not None:
        envs_yaml = repo_root / "config" / "envs.yaml"
        if envs_yaml.exists():
            try:
                with open(envs_yaml, "r", encoding="utf-8") as f:
                    envs = yaml.safe_load(f)
                if isinstance(envs, dict) and "custom_envs" in envs and "game_2048" in envs["custom_envs"]:
                    e = envs["custom_envs"]["game_2048"]
                    max_actions = int(e.get("max_actions_per_traj", max_actions))
            except Exception:
                pass

        base_yaml = repo_root / "config" / "base.yaml"
        if base_yaml.exists():
            try:
                with open(base_yaml, "r", encoding="utf-8") as f:
                    base_cfg = yaml.safe_load(f)
                    ap = base_cfg.get("agent_proxy", {}) if isinstance(base_cfg, dict) else {}
                    enable_think = bool(ap.get("enable_think", enable_think))
            except Exception:
                pass

    instruction = instruction_template.format(max_actions=max_actions, action_sep=action_sep)
    return instruction, max_tokens, action_sep, enable_think, max_actions


def get_grid_matrix(state: List[List[List[float]]]) -> List[List[int]]:
    """Decode 2048 CNN one-hot channels (C x 4 x 4) into a 4x4 integer matrix."""
    if not state or not isinstance(state, list):
        return [[0]*4 for _ in range(4)]
    
    C = len(state)
    H = len(state[0]) if C > 0 else 0
    W = len(state[0][0]) if (C > 0 and H > 0) else 0
    
    grid_vals = [[0 for _ in range(W)] for __ in range(H)]
    
    for i in range(H):
        for j in range(W):
            max_c = 0
            max_v = -float("inf")
            for c in range(C):
                v = state[c][i][j]
                if v > max_v:
                    max_v = v
                    max_c = c
            
            if max_c <= 0:
                grid_vals[i][j] = 0
            else:
                try:
                    grid_vals[i][j] = int(2 ** max_c)
                except Exception:
                    grid_vals[i][j] = 0
    return grid_vals


def grid_to_text(grid_vals: List[List[int]]) -> str:
    """Convert 4x4 matrix to text representation."""
    lines = ["Current 2048 Grid:"]
    H = len(grid_vals)
    W = len(grid_vals[0]) if H > 0 else 0
    for r in range(H):
        row_str = ", ".join(str(grid_vals[r][c]) for c in range(W))
        lines.append(f"Row {r+1}: [{row_str}]")
    return "\n".join(lines)


def get_valid_actions(grid: List[List[int]]) -> List[int]:
    """
    Determine valid moves for a 4x4 grid.
    Returns a list of action indices: 0(Up), 1(Right), 2(Down), 3(Left)
    """
    valid_actions = []
    
    def can_move_left_row(row: List[int]) -> bool:
        """Check if a single row can compress or merge to the left."""
        # Check 1: Can merge? (Adjacent equal non-zeros)
        tiles = [x for x in row if x != 0]
        for i in range(len(tiles) - 1):
            if tiles[i] == tiles[i+1]:
                return True
        
        # Check 2: Can slide? (Is there a 0 to the left of a non-0?)
        # Logic: If we see a 0, then subsequently see a non-0, we can move.
        seen_zero = False
        for x in row:
            if x == 0:
                seen_zero = True
            elif seen_zero: # x != 0 and seen_zero is True
                return True
        return False

    # 0: Up (Check columns effectively moving "left" if transposed)
    # Transpose grid to treat columns as rows
    cols = [[grid[r][c] for r in range(4)] for c in range(4)]
    if any(can_move_left_row(col) for col in cols):
        valid_actions.append(0)

    # 1: Right (Check rows reversed)
    if any(can_move_left_row(row[::-1]) for row in grid):
        valid_actions.append(1)

    # 2: Down (Check columns reversed)
    cols_rev = [[grid[r][c] for r in reversed(range(4))] for c in range(4)]
    if any(can_move_left_row(col) for col in cols_rev):
        valid_actions.append(2)

    # 3: Left (Check rows normal)
    if any(can_move_left_row(row) for row in grid):
        valid_actions.append(3)

    return valid_actions


def build_messages_for_episode(
    states: List[List[List[List[float]]]],
    actions: List[int],
    rewards: List[float],
    instruction: str,
    max_tokens: int,
    action_sep: str,
    enable_think: bool,
    max_actions: int,
) -> List[dict]:
    messages = [
        {"role": "system", "content": "You're a helpful assistant. "},
        {"role": "user", "content": instruction},
    ]

    total_actions = len(actions)
    current_score = 0.0

    for t, state in enumerate(states):
        # 1. Decode Matrix
        grid_matrix = get_grid_matrix(state)
        # 2. Convert to Text
        grid_text = grid_to_text(grid_matrix)
        # 3. Calculate Valid Actions
        valid_idxs = get_valid_actions(grid_matrix)
        
        # Fallback: if somehow no actions are valid (game over state), strictly speaking the game ends.
        # But if the log continues, we default to all or keep empty.
        # Usually we format whatever is valid.
        valid_actions_str_parts = []
        for idx in sorted(valid_idxs):
            name = ACTION_LOOKUP_2048.get(idx, str(idx))
            valid_actions_str_parts.append(f"{idx}({name})")
        
        if valid_actions_str_parts:
            valid_actions_str = ", ".join(valid_actions_str_parts) + "."
        else:
            # Should imply Game Over, but for prompting consistency:
            valid_actions_str = "None (Game Over)."

        actions_left = max(0, max_actions - t)
        format_prompt = (
            "<think> [Your thoughts] </think> <answer> [your answer] </answer>"
            if enable_think
            else "<answer> [your answer] </answer>"
        )
        length_prompt = f"Max response length: {max_tokens} words (tokens)."

        turn_content = (
            f"\n\nTurn {t + 1}:\n"
            f"State:\n"
            f"{grid_text}\n\n"
            f"Valid Actions: {valid_actions_str}\n"
            f"Goal: Merge same numbers to reach 2048.\n"
            f"Current Score: {int(current_score)}\n"
            f"What is your next move?\n"
            f"You have {actions_left} actions left. Always output: {format_prompt} "
            f"with no extra text. Strictly follow this format. {length_prompt}"
        )

        messages[-1]["content"] += turn_content

        if t < total_actions:
            a = actions[t]
            a_name = ACTION_LOOKUP_2048.get(int(a), str(a))
            if enable_think:
                assistant_text = f"<think> </think><answer>{a_name}</answer>"
            else:
                assistant_text = f"<answer>{a_name}</answer>"
            messages.append({"role": "assistant", "content": assistant_text})
            
            r = rewards[t] if t < len(rewards) else 0.0
            current_score += r
            messages.append({"role": "user", "content": f"Reward:\n{float(np.log2(r + 1.0)) * 0.1}"})

    return messages[:-1]


def find_latest_step_dir(traj_root: Path) -> Path:
    step_dirs = [p for p in traj_root.iterdir() if p.is_dir() and p.name.startswith("step_")]
    if not step_dirs:
        raise FileNotFoundError(f"No step_* directories under {traj_root}")
    step_dirs.sort(key=lambda p: int(p.name.split("_")[-1]))
    return step_dirs[-1]


def convert_file(step_dir: Path, output_dir: Path, repo_root: Path, include_failed: bool, max_actions_cap: int | None) -> Path:
    traj_path = step_dir / "trajectories.jsonl"
    metrics_path = step_dir / "metrics.json"
    if not traj_path.exists():
        raise FileNotFoundError(f"Missing trajectories.jsonl at {traj_path}")

    instruction, max_tokens, action_sep, enable_think, default_max_actions = load_env_instruction_and_cfg(repo_root)
    max_actions = int(max_actions_cap) if max_actions_cap is not None else int(default_max_actions)
    output_dir.mkdir(parents=True, exist_ok=True)
    out_path = output_dir / f"{step_dir.name}_sft.jsonl"

    global_step = None
    if metrics_path.exists():
        try:
            with open(metrics_path, "r", encoding="utf-8") as f:
                m = json.load(f)
                global_step = m.get("global_step")
        except Exception:
            pass

    written = 0

    with open(traj_path, "r", encoding="utf-8") as fin, open(out_path, "w", encoding="utf-8") as fout:
        for line in fin:
            line = line.strip()
            if not line:
                continue
            traj = json.loads(line)

            ep_success = bool(traj.get("episode_success", False))
            if (not include_failed) and (not ep_success):
                continue

            states = traj.get("states", [])
            actions = traj.get("actions", [])
            rewards = traj.get("rewards", [])
            
            if len(actions) > max_actions:
                continue

            messages = build_messages_for_episode(
                states=states,
                actions=actions,
                rewards=rewards,
                instruction=instruction,
                max_tokens=max_tokens,
                action_sep=action_sep,
                enable_think=enable_think,
                max_actions=max_actions,
            )

            record = {
                "messages": messages,
                "meta": {
                    "episode_return": traj.get("episode_return", None),
                    "episode_success": ep_success,
                    "global_step": global_step,
                },
            }
            # Threshold check
            if traj.get("episode_return", 0) > 7000:
                fout.write(json.dumps(record, ensure_ascii=False) + "\n")
                written += 1
    import pdb;pdb.set_trace()            
    if written == 0:
        with open(out_path, "w", encoding="utf-8"):
            pass
    return out_path


def main():
    parser = argparse.ArgumentParser(description="Convert 2048 RL trajectories to LLM SFT chat JSONL")
    parser.add_argument("run_dir", help="Path to the run directory (contains trajectories/)")
    parser.add_argument("--step", default=None, help="Specific step directory name (e.g., step_993280)")
    parser.add_argument("--include_failed", action="store_true", help="Include failed episodes in SFT data")
    parser.add_argument("--max_actions", type=int, default=700, help="Override max actions cap (default from envs.yaml)")
    args = parser.parse_args()

    repo_root = Path(__file__).resolve().parents[1]
    run_dir = Path(args.run_dir)
    traj_root = run_dir / "trajectories"
    if not traj_root.exists():
        raise FileNotFoundError(f"Not found trajectories directory: {traj_root}")

    step_dir = traj_root / args.step if args.step else find_latest_step_dir(traj_root)
    output_dir = run_dir / "sft"
    out_path = convert_file(step_dir=step_dir, output_dir=output_dir, repo_root=repo_root, include_failed=args.include_failed, max_actions_cap=args.max_actions)
    print(f"SFT data written to: {out_path}")


if __name__ == "__main__":
    main()