Aislo/resources/tester/helper_810_4_sweep.py

"""810-4 전수 도우미 — 원파일 산근 블록 전부(EST 6 · STC 7) × M01 수식 문서 1,518 연결 찾기.

파일마다 프로세스 · 안쪽 산근(형제)은 같은 파일 블록을 우리 단가로 먼저 셈해 묶음 값으로 넣음(저장 층의
형제 SK 읽기와 같은 값) · 확정은 M01 문서를 찾은 옵션(입력값 꼴)으로 다시 셈해 금액 대조.
마지막 줄 = JSON {블록, 판정, 근접항, 확정참고, 확정참고블록, 오류, 확정다시셈다름} · `--rows <pkl>` = 줄 목록 저장(목록 xlsx 용).
"""

from __future__ import annotations

import collections
import json
import os
import pickle
import sys
import time
from multiprocessing import Pool
from pathlib import Path

HERE = Path(__file__).resolve().parent
ROOT = HERE.parents[1]
for p in (ROOT, HERE):
    if str(p) not in sys.path:
        sys.path.insert(0, str(p))


def _ctx():
    import test_810_2_native_doc as t

    from M02_MasterTemplete import M02_Doc_Import as dim
    from M02_MasterTemplete import M02_Doc_Match as dm

    whole = t.whole()
    docs = dm.ml.load_all()
    return t, dim, dm, whole, docs, dm.Index(docs), dm.Memo(dm.ml.Links(whole, docs))


def file_docs(t, dim, dm, whole, kind: str, tag: str) -> dict[int, dict]:
    """파일 블록 → 가져온 식 문서(형제 = 우리 단가 묶음)."""
    saved = t.native(kind, tag)
    nums = t.fn.numbers(saved)
    by_code = {str(b.get("코드")): n for n, b in nums.items()}
    links = dim.Links(whole, "우리")
    made: dict[int, dict] = {}
    ours: dict[int, dict | None] = {}

    def build(no: int, trail: tuple = ()) -> None:
        if no in made:
            return
        out = t.fn.emit(saved, no)
        doc = dim.derive(t.sk_of(out), None, whole)
        for h in out["로직"]["호표"]:
            res = h.get("자원") or {}
            n = by_code.get(str(res.get("코드")))
            if res.get("갈래") != "단가산출" or n is None or n in trail or n == no:
                continue
            build(n, (*trail, no))
            got = ours.get(n)
            line = next(
                (x for x in doc["금액"] if x["이름"] == h["이름"] and x.get("비목") == "묶음"), None
            )  # noqa: E501
            if not line or not got:
                continue
            for v in doc["변수"]:
                if dm.md.var_id(v) == line["단가"] and (v.get("연결") or {}).get("갈래") == "숫자":
                    v["연결"] = {"갈래": "숫자", "값": {k: str(got[k]) for k in dm.md.COSTS}}
        made[no] = doc
        ours[no] = dm._calc(doc, links, {})

    for no in nums:
        build(no)
    return made


PARTS = 10  # 파일마다 블록을 나눠 맡김(큰 파일 한 프로세스가 끝을 붙잡지 않게)


def sweep(arg: tuple[str, str, int]) -> list[dict]:
    kind, tag, part = arg
    t, dim, dm, whole, docs, index, links = _ctx()
    rows = []
    try:
        made = file_docs(t, dim, dm, whole, kind, tag)
    except Exception as e:  # noqa: BLE001 — 원파일 없음(다른 PC) = 빈 줄
        if part:
            return []
        return [{"파일": tag, "오류": repr(e)[:200], "판정": "오류", "없는파일": "없음" in repr(e)}]
    for no, doc in made.items():
        if no % PARTS != part:
            continue
        start = time.time()
        try:
            res = dm.match(doc, dim.Links(whole, "우리"), index, links)
        except Exception as e:  # noqa: BLE001
            res = {"판정": "오류", "까닭": repr(e)[:200]}
        row = {"파일": tag, "꼴": kind, "번": no, "블록": doc["이름"], "단위": doc["결과단위"]}
        row |= {"규격": (doc.get("가져온") or {}).get("규격"), **res}
        if res.get("판정") == dm.SURE:
            again = dm._calc(docs[res["키"]], dm.ml.Links(whole, docs), dict(res["옵션"]))
            money = res["금액"]  # 다시 셈 = 우리 · 참고 없으면 가져온 = 우리(812-2)
            row["다시셈같음"] = (
                again is not None
                and dm._plain(dm._totals(again)) == money["우리"]
                and (bool(res.get("참고")) or money["우리"] == money["가져온"])
            )
        row["초"] = round(time.time() - start, 2)
        rows.append(row)
    return rows


def main() -> None:
    import test_810_1_estx as e
    import test_810_1_stc as c

    tags = [("EST", k) for k in e.FILES] + [("STC", k) for k in c.FILES]
    tags = [(kind, tag, part) for kind, tag in tags for part in range(PARTS)]
    with Pool(min(len(tags), max(2, (os.cpu_count() or 4) - 2))) as pool:
        rows = [r for part in pool.map(sweep, tags) for r in part]
    if "--rows" in sys.argv:
        pickle.dump(rows, open(sys.argv[sys.argv.index("--rows") + 1], "wb"))
    real = [r for r in rows if not r.get("없는파일")]
    out = {
        "블록": sum(1 for r in real if r.get("판정") != "오류" or "번" in r),
        "판정": dict(collections.Counter(r["판정"] for r in real)),
        "근접항": dict(
            collections.Counter(d["항"] for r in real if r["판정"] == "근접" for d in r["다른"])
        ),  # noqa: E501
        "확정참고": dict(
            collections.Counter(
                d["항"] for r in real if r["판정"] == "확정" for d in r.get("참고") or []
            )
        ),
        "확정참고블록": sum(1 for r in real if r["판정"] == "확정" and r.get("참고")),
        "오류": sum(1 for r in real if r["판정"] == "오류"),
        "확정다시셈다름": sum(1 for r in real if r.get("다시셈같음") is False),
    }
    print(json.dumps(out, ensure_ascii=False))


if __name__ == "__main__":
    main()