evotraders/backend/domains/news.py

# -*- coding: utf-8 -*-
"""News/explain domain helpers shared by app surfaces and gateway fallbacks."""

from __future__ import annotations

from typing import Any

from backend.data.market_store import MarketStore
from backend.data.market_ingest import update_ticker_incremental
from backend.enrich.news_enricher import enrich_news_for_symbol
from backend.explain.range_explainer import build_range_explanation
from backend.explain.similarity_service import find_similar_days
from backend.explain.story_service import get_or_create_stock_story


def news_rows_need_enrichment(rows: list[dict[str, Any]]) -> bool:
    """Return whether news rows are missing explain-oriented analysis fields."""
    if not rows:
        return True
    return all(
        not row.get("sentiment")
        and not row.get("relevance")
        and not row.get("key_discussion")
        for row in rows
    )


def ensure_news_fresh(
    store: MarketStore,
    *,
    ticker: str,
    target_date: str | None = None,
    refresh_if_stale: bool = True,
) -> dict[str, Any]:
    """Refresh raw news incrementally when stored watermarks are stale."""
    normalized_target = str(target_date or "").strip()[:10]
    if not normalized_target:
        return {
            "ticker": ticker,
            "target_date": None,
            "last_news_fetch": None,
            "refreshed": False,
        }

    watermarks = store.get_ticker_watermarks(ticker)
    last_news_fetch = str(watermarks.get("last_news_fetch") or "").strip()[:10]
    refreshed = False
    if refresh_if_stale and (not last_news_fetch or last_news_fetch < normalized_target):
        update_ticker_incremental(
            ticker,
            end_date=normalized_target,
            store=store,
        )
        refreshed = True
        watermarks = store.get_ticker_watermarks(ticker)
        last_news_fetch = str(watermarks.get("last_news_fetch") or "").strip()[:10]

    return {
        "ticker": ticker,
        "target_date": normalized_target,
        "last_news_fetch": last_news_fetch or None,
        "refreshed": refreshed,
    }


def get_enriched_news(
    store: MarketStore,
    *,
    ticker: str,
    start_date: str | None = None,
    end_date: str | None = None,
    limit: int = 100,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=end_date,
        refresh_if_stale=refresh_if_stale,
    )
    rows = store.get_news_items_enriched(
        ticker,
        start_date=start_date,
        end_date=end_date,
        limit=limit,
    )
    if news_rows_need_enrichment(rows):
        enrich_news_for_symbol(
            store,
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=limit,
        )
        rows = store.get_news_items_enriched(
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=limit,
        )
    return {"ticker": ticker, "news": rows, "freshness": freshness}


def get_news_for_date(
    store: MarketStore,
    *,
    ticker: str,
    date: str,
    limit: int = 20,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=date,
        refresh_if_stale=refresh_if_stale,
    )
    rows = store.get_news_items_enriched(
        ticker,
        trade_date=date,
        limit=limit,
    )
    if news_rows_need_enrichment(rows):
        enrich_news_for_symbol(
            store,
            ticker,
            start_date=date,
            end_date=date,
            limit=limit,
        )
        rows = store.get_news_items_enriched(
            ticker,
            trade_date=date,
            limit=limit,
        )
    return {"ticker": ticker, "date": date, "news": rows, "freshness": freshness}


def get_news_timeline(
    store: MarketStore,
    *,
    ticker: str,
    start_date: str,
    end_date: str,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=end_date,
        refresh_if_stale=refresh_if_stale,
    )
    timeline = store.get_news_timeline_enriched(
        ticker,
        start_date=start_date,
        end_date=end_date,
    )
    if not timeline:
        enrich_news_for_symbol(
            store,
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=200,
        )
        timeline = store.get_news_timeline_enriched(
            ticker,
            start_date=start_date,
            end_date=end_date,
        )
    return {
        "ticker": ticker,
        "timeline": timeline,
        "start_date": start_date,
        "end_date": end_date,
        "freshness": freshness,
    }


def get_news_categories(
    store: MarketStore,
    *,
    ticker: str,
    start_date: str | None = None,
    end_date: str | None = None,
    limit: int = 200,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=end_date,
        refresh_if_stale=refresh_if_stale,
    )
    rows = store.get_news_items_enriched(
        ticker,
        start_date=start_date,
        end_date=end_date,
        limit=limit,
    )
    if news_rows_need_enrichment(rows):
        enrich_news_for_symbol(
            store,
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=limit,
        )
    categories = store.get_news_categories_enriched(
        ticker,
        start_date=start_date,
        end_date=end_date,
        limit=limit,
    )
    return {"ticker": ticker, "categories": categories, "freshness": freshness}


def get_similar_days_payload(
    store: MarketStore,
    *,
    ticker: str,
    date: str,
    n_similar: int = 5,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=date,
        refresh_if_stale=refresh_if_stale,
    )
    result = find_similar_days(
        store,
        symbol=ticker,
        target_date=date,
        top_k=n_similar,
    )
    result["freshness"] = freshness
    return result


def get_story_payload(
    store: MarketStore,
    *,
    ticker: str,
    as_of_date: str,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=as_of_date,
        refresh_if_stale=refresh_if_stale,
    )
    enrich_news_for_symbol(
        store,
        ticker,
        end_date=as_of_date,
        limit=80,
    )
    result = get_or_create_stock_story(
        store,
        symbol=ticker,
        as_of_date=as_of_date,
    )
    result["freshness"] = freshness
    return result


def get_range_explain_payload(
    store: MarketStore,
    *,
    ticker: str,
    start_date: str,
    end_date: str,
    article_ids: list[str] | None = None,
    limit: int = 100,
    refresh_if_stale: bool = False,
) -> dict[str, Any]:
    freshness = ensure_news_fresh(
        store,
        ticker=ticker,
        target_date=end_date,
        refresh_if_stale=refresh_if_stale,
    )
    news_rows = []
    if article_ids:
        news_rows = store.get_news_by_ids_enriched(ticker, article_ids)
    if not news_rows:
        news_rows = store.get_news_items_enriched(
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=limit,
        )
    if news_rows_need_enrichment(news_rows):
        enrich_news_for_symbol(
            store,
            ticker,
            start_date=start_date,
            end_date=end_date,
            limit=limit,
        )
        news_rows = (
            store.get_news_by_ids_enriched(ticker, article_ids)
            if article_ids
            else store.get_news_items_enriched(
                ticker,
                start_date=start_date,
                end_date=end_date,
                limit=limit,
            )
        )
    result = build_range_explanation(
        ticker=ticker,
        start_date=start_date,
        end_date=end_date,
        news_rows=news_rows,
    )
    return {"ticker": ticker, "result": result, "freshness": freshness}