"""Shared data models. Every module talks to the others through these types only."""

from __future__ import annotations

from datetime import date, datetime
from typing import Literal

from pydantic import BaseModel, Field


class Quote(BaseModel):
    symbol: str
    name: str
    group: str
    close: float | None = None
    prev_close: float | None = None
    change_pct: float | None = None  # (close/prev_close - 1) * 100
    change_abs: float | None = None  # close - prev_close (used for rates: yield points)
    day_high: float | None = None
    day_low: float | None = None
    volume: float | None = None
    avg_volume_10d: float | None = None
    after_hours_price: float | None = None
    after_hours_change_pct: float | None = None  # vs regular close
    as_of: datetime | None = None
    source: str = "yfinance"


class NewsItem(BaseModel):
    title: str
    summary: str = ""
    link: str
    source: str
    published: datetime


class EconEvent(BaseModel):
    name: str
    time: datetime | None = None
    country: str = "US"
    actual: str | None = None
    forecast: str | None = None
    previous: str | None = None
    importance: Literal["high", "medium", "low"] = "medium"
    kind: Literal["released", "upcoming"] = "released"


class EarningsItem(BaseModel):
    symbol: str
    name: str
    report_date: date
    timing: Literal["bmo", "amc", "unknown"] = "unknown"
    eps_actual: float | None = None
    eps_estimate: float | None = None
    revenue_actual: float | None = None
    revenue_estimate: float | None = None


class MarketPayload(BaseModel):
    """Everything collected for one briefing, before LLM summarisation."""

    session_date: date  # US trading date that just closed
    us_close_time: datetime  # tz-aware (America/New_York)
    generated_at: datetime  # tz-aware (Asia/Seoul)
    krx_date: date  # the Korean trading date this briefing is for
    krx_open_today: bool
    quotes: list[Quote] = Field(default_factory=list)
    news: list[NewsItem] = Field(default_factory=list)
    events: list[EconEvent] = Field(default_factory=list)
    earnings: list[EarningsItem] = Field(default_factory=list)
    history_notes: list[str] = Field(default_factory=list)  # e.g. "S&P 500 3일 연속 하락"
    data_gaps: list[str] = Field(default_factory=list)  # collectors that failed / partial

    def quotes_by_group(self) -> dict[str, list[Quote]]:
        out: dict[str, list[Quote]] = {}
        for q in self.quotes:
            out.setdefault(q.group, []).append(q)
        return out


class Report(BaseModel):
    """LLM output schema. Used verbatim as the JSON Schema for structured output."""

    headline: str = Field(description="한 줄 총평 (40자 이내)")
    market_overview: str = Field(description="지수·섹터 흐름 3~5문장")
    key_drivers: list[str] = Field(description="상승/하락 요인 3~6개, 각 한 문장")
    rates_fx_commodities: str = Field(description="금리·달러·원화·유가·금 흐름 2~4문장")
    korea_implications: list[str] = Field(description="한국 장 시사점 3~6개 (반도체/SOX, 환율, ADR, EWY 등)")
    watchlist_today: list[str] = Field(description="오늘 한국 장에서 확인할 것 2~5개")
    risk_flags: list[str] = Field(description="주의사항 1~4개")
    data_gaps: list[str] = Field(default_factory=list, description="수집 실패로 반영하지 못한 항목")


class BriefingResult(BaseModel):
    payload: MarketPayload
    report: Report | None
    markdown: str
    model_used: str | None = None
    degraded: bool = False  # True when LLM failed and only the numeric table was produced
    error: str | None = None