Agent/examples/process_pipeline/script/search_eval/evaluation/remap_runs_full.py

# -*- coding: utf-8 -*-
"""把 runs_full/qXXXX 里 idx 错位的 q 目录原地搬到正确 idx 位。

错位成因：旧 batch_3forms.py 默认吃 high_priority_queries.json（468 条），写入 runs_full
时用的是那个文件的 idx。后来换默认到 high_priority_queries_full.json（2808 条），
两套 idx 表是不同的，于是 runs_full/q0002 装的 original_q 在新 full 表里其实属于
另一个 idx。本脚本读每个 q 的 form_A.json original_q，查新 full 表的等值 q，
mv 到正确目录 q{正确 idx:04d}/。

规则：
  - 已在正确位的: 跳过
  - 错位但目标位空闲: shutil.move 改名（顺手更新 _manifest.json）
  - 错位但目标位已占用: 不动，写入 _conflicts 段供人工裁决
  - original_q 在新 full 表里找不到: 标记 unmatched（应该不会发生，前面 build_runs_mapping 已过滤）
"""
import json
import re
import shutil
from pathlib import Path

HERE = Path(__file__).parent
SEARCH_EVAL = HERE.parent
RUNS_FULL = SEARCH_EVAL / "runs_full"
FULL = HERE / "high_priority_queries_full.json"
MANIFEST = RUNS_FULL / "_manifest.json"


def _qnum(name: str) -> int:
    m = re.search(r"\d+", name)
    return int(m.group()) if m else -1


def load_original_q(qd: Path):
    for fk in ("A", "B", "C"):
        f = qd / f"form_{fk}.json"
        if not f.exists():
            continue
        try:
            d = json.load(open(f, encoding="utf-8"))
            oq = d.get("original_q") or d.get("query") or ""
            if oq:
                return oq, fk
        except Exception:
            continue
    return "", ""


def main():
    full = json.load(open(FULL, encoding="utf-8"))["queries"]
    q_to_idx = {item["q"]: i for i, item in enumerate(full)}

    q_dirs = sorted([d for d in RUNS_FULL.iterdir() if d.is_dir() and d.name.startswith("q")],
                    key=lambda p: _qnum(p.name))

    already_ok, moved, conflicts, unmatched = [], [], [], []

    # Plan: 先扫描决策，再执行；避免边扫边动导致后续判断错乱
    plan = []
    for qd in q_dirs:
        cur_idx = _qnum(qd.name)
        oq, src_form = load_original_q(qd)
        if not oq:
            unmatched.append({"name": qd.name, "reason": "form_*.json 缺失或 original_q 空"})
            continue
        correct_idx = q_to_idx.get(oq)
        if correct_idx is None:
            unmatched.append({"name": qd.name, "original_q": oq,
                              "reason": "在新 full 表中找不到等值 q（可能是历史模态）"})
            continue
        if correct_idx == cur_idx:
            already_ok.append((qd.name, oq))
            continue
        plan.append((qd, cur_idx, correct_idx, oq))

    # 两轮执行：先把没冲突的全搬走，再回头看冲突（可能搬走后冲突就消解了）
    pending = list(plan)
    progress = True
    while pending and progress:
        progress = False
        next_pending = []
        for qd, cur_idx, correct_idx, oq in pending:
            dst = RUNS_FULL / f"q{correct_idx:04d}"
            if dst.exists():
                next_pending.append((qd, cur_idx, correct_idx, oq))
                continue
            shutil.move(str(qd), str(dst))
            moved.append({"from": qd.name, "to": dst.name, "original_q": oq})
            progress = True
        pending = next_pending

    # 仍有冲突的：目标位真的被占用且非空（让人工看）
    for qd, cur_idx, correct_idx, oq in pending:
        dst = RUNS_FULL / f"q{correct_idx:04d}"
        dst_oq, _ = load_original_q(dst) if dst.exists() else ("", "")
        conflicts.append({
            "src": qd.name, "src_idx": cur_idx, "src_original_q": oq,
            "dst": dst.name, "dst_idx": correct_idx, "dst_original_q": dst_oq,
        })

    # 更新 manifest：用搬完后的 runs_full 状态重写
    new_manifest_items = []
    for qd in sorted([d for d in RUNS_FULL.iterdir() if d.is_dir() and d.name.startswith("q")],
                     key=lambda p: _qnum(p.name)):
        oq, _ = load_original_q(qd)
        idx = _qnum(qd.name)
        if idx >= 0 and idx < len(full):
            item = full[idx]
            new_manifest_items.append({
                "qdir": qd.name, "full_idx": idx,
                "original_q": oq,
                "cell_idx": idx // 18,
                "lens": item["lens"],
                "constraint_value": (item.get("constraint") or {}).get("value") or "无约束",
            })
    MANIFEST.write_text(json.dumps({
        "_doc": "runs_full/qXXXX 的当前状态快照；XXXX = high_priority_queries_full.json idx 4 位 zero-pad",
        "items": new_manifest_items,
    }, ensure_ascii=False, indent=2), encoding="utf-8")

    print(f"=== runs_full 原地重映射结果 ===")
    print(f"  原本就对的: {len(already_ok)}")
    print(f"  移动到位:   {len(moved)}")
    print(f"  冲突待人工: {len(conflicts)}")
    print(f"  未匹配:     {len(unmatched)}")
    print(f"→ manifest 已重写: {MANIFEST}")
    print()
    if moved[:8]:
        print("=== 已搬动（抽样）===")
        for m in moved[:8]:
            print(f"  {m['from']:<7} → {m['to']:<7}  [{m['original_q']}]")
    if conflicts:
        print()
        print("=== ⚠️ 冲突（目标位被别的 original_q 占住，未搬）===")
        for c in conflicts:
            print(f"  {c['src']} (idx={c['src_idx']}, q={c['src_original_q']!r})")
            print(f"    -> 想去 {c['dst']} (idx={c['dst_idx']}, 但被 {c['dst_original_q']!r} 占)")
    if unmatched:
        print()
        print("=== ⚠️ 在新 full 中找不到 ===")
        for u in unmatched:
            print(f"  {u}")


if __name__ == "__main__":
    main()