# -*- coding: utf-8 -*-
"""成交回报回执:②解析 [trade-receipt] 指纹行 + 回执↔账本闭环比对。

背景(2026-09-04 实证):柜台有成交、台账无行、引擎会话成交视图也无行且
时间字段退化——「推送到达→回调处理→落库」三跳全静默,故障点无法定位。
观测契约:
  ① 引擎侧(策略 session 实现):成交回调任何处理之前打一行指纹,永不抛:
     [trade-receipt] tid={柜台成交号} order={订单号} acc={账号} sym={代码}
     vol={量} px={价} t_raw={回报自带时间,无则NONE} recv={本地时刻}
  ② 本模块:解析该行,与 live_trades 当日行闭环——
     有回执无账行=引擎处理丢(②③之间);柜台有成交且无回据=柜台没推(由
     EOD 恒等式/守恒缺口负责,不在此层)。
  ③ 账本行带柜台成交号(硬键),判重/补插不再依赖会退化的时间字段。

设计原则:宁缺勿错——tid=NONE 的行不做模糊匹配硬凑,如实分列。
"""

import re
from typing import Any, Dict, List

# 回执行锚:tag 全文匹配,字段名与①契约严格一致(勿改拼写)
_RECEIPT_RE = re.compile(
    r"\[trade-receipt\]\s+tid=(\S+)\s+order=(\S+)\s+acc=(\S+)\s+sym=(\S+)\s+"
    r"vol=(\S+)\s+px=(\S+)\s+t_raw=(\S+)\s+recv=(\S+)"
)

_NONE = {"NONE", "None", "none", ""}


def _or_none(v: str) -> Any:
    return None if v in _NONE else v


def _to_float(v: str) -> Any:
    try:
        return float(v)
    except (TypeError, ValueError):
        return None


def parse_trade_receipts(log_text: str) -> List[Dict[str, Any]]:
    """从引擎日志文本解析全部回执行;同 tid 重复回执去重并计 dup_count。"""
    seen: Dict[str, Dict[str, Any]] = {}
    order_keys: List[str] = []
    for m in _RECEIPT_RE.finditer(log_text):
        tid = _or_none(m.group(1))
        key = tid if tid is not None else "_orphan_%d" % len(order_keys)
        if key in seen:
            seen[key]["dup_count"] += 1
            continue
        row = {
            "tid": tid,
            "order": _or_none(m.group(2)),
            "acc": _or_none(m.group(3)),
            "sym": _or_none(m.group(4)),
            "vol": _to_float(m.group(5)),
            "px": _to_float(m.group(6)),
            "t_raw": _or_none(m.group(7)),
            "recv": _or_none(m.group(8)),
            "dup_count": 1,
        }
        seen[key] = row
        order_keys.append(key)
    return [seen[k] for k in order_keys]


def compare_receipts_with_ledger(receipts: List[Dict[str, Any]],
                                 ledger: List[Dict[str, Any]]) -> Dict[str, Any]:
    """回执(①的产物)与账本行(live_trades 映射,带 tid)闭环比对。

    ledger 行键约定:tid/symbol/volume/price(由调用方从 DB 行映射)。
    matched 只认 tid 硬键;tid 缺失双方不猜,各自入 receipt_only/ledger_only。
    ledger_only 为信息项(回执上线前的旧数据/repair 行),不判脏;
    receipt_only 才是警报——「到了没记」即引擎处理跳丢失的直接证据。
    """
    ledger_by_tid: Dict[str, Dict[str, Any]] = {
        r["tid"]: r for r in ledger if r.get("tid") is not None
    }
    receipt_only: List[Dict[str, Any]] = []
    matched = 0
    for r in receipts:
        tid = r.get("tid")
        if tid is not None and tid in ledger_by_tid:
            matched += 1
        else:
            receipt_only.append(r)
    matched_tids = set()
    for r in receipts:
        if r.get("tid") is not None and r.get("tid") in ledger_by_tid:
            matched_tids.add(r["tid"])
    ledger_only = [r for r in ledger
                   if r.get("tid") is not None and r["tid"] not in matched_tids]
    ledger_only += [r for r in ledger if r.get("tid") is None]
    return {
        "receipts_n": len(receipts),
        "ledger_n": len(ledger),
        "matched_n": matched,
        "receipt_only": receipt_only,
        "ledger_only": ledger_only,
        "clean": not receipt_only,
    }