"""
Forex Signal Intelligence Platform - News & Sentiment Engine
=============================================================
Monitors economic calendar events and news sentiment to provide
fundamental context for technical signals.

Key principle: High-impact news events override technical signals.
A perfect technical setup can be destroyed by an unexpected NFP print.
"""

import hashlib
import logging
import re
import threading
import time
import xml.etree.ElementTree as ET
from dataclasses import dataclass, field
from datetime import datetime, timezone
from email.utils import parsedate_to_datetime
from typing import Dict, List, Optional, Tuple

import requests

from config import NewsConfig
from llm_client import get_llm_client

logger = logging.getLogger(__name__)


# ── DeepSeek prompts ─────────────────────────────────────────────────
# STATIC constants on purpose: DeepSeek prices a repeated byte-identical
# prompt prefix at the cache-hit rate (~50x cheaper), so nothing here may
# be interpolated or timestamped. All variable content travels in the
# user payload. LLM output is display/warning-only — it never gates
# signals (see NewsConfig.sentiment_affects_signals).

_LLM_SENTIMENT_SYSTEM = """\
You are a forex/gold news analyst. You receive JSON:
{"headlines": ["...", ...]}
For EACH headline, in the same order, score:
- "s": overall market sentiment of the headline, -1.0 (very bearish)
  to 1.0 (very bullish), 0.0 if irrelevant or neutral.
- "xau": the effect on the GOLD (XAU/USD) price, -1.0 (gold down)
  to 1.0 (gold up), 0.0 if irrelevant.
Gold doctrine: hawkish Fed / rate hikes / rising real yields / strong
dollar are gold-bearish; easing / rate cuts / safe-haven flows /
geopolitical stress / weak dollar are gold-bullish.
Handle negation correctly ("no rate hike" is not hawkish).
Reply with ONLY JSON: {"scores": [[s, xau], ...]} — exactly one
[s, xau] pair per input headline, in the same order, nothing else.

Examples:
"Fed signals faster rate hikes as inflation stays hot" -> [0.4, -0.7]
"Gold surges to record high on safe-haven demand" -> [0.8, 0.9]
"Dollar slips after soft jobs report; Treasury yields fall" -> [-0.4, 0.6]
"""

_LLM_EVENT_SYSTEM = """\
You are a forex/gold analyst. You receive one released economic event as
JSON: {"event", "currency", "actual", "forecast", "previous",
"vs_forecast"}. "vs_forecast" states whether the actual print came in
above or below the forecast; it is computed programmatically — trust it
over your own comparison of the numbers. If it is "unknown", compare
actual and forecast yourself.
Decide the likely effect of this RELEASE on the GOLD (XAU/USD) price.
A stronger-than-forecast USD economy or a hawkish implication is usually
gold-bearish; weaker data or a dovish implication is usually
gold-bullish.
Inverted metrics — for unemployment rate and unemployment/jobless
claims a LOWER number means a STRONGER labor market: below forecast =
strong economy = usually gold-bearish; above forecast = weak economy =
usually gold-bullish.
Example: {"event": "Unemployment Claims", "actual": "202K", "forecast":
"217K", "vs_forecast": "below forecast"} ->
{"gold_direction": "bearish", "reason": "fewer claims show strong labor market"}
Reply with ONLY JSON:
{"gold_direction": "bullish"|"bearish"|"neutral", "reason": "<max 10 words>"}
"""

_LLM_SUMMARY_SYSTEM = """\
You are a forex/gold market-context writer. You receive JSON with recent
"headlines" and today's high-impact "events" (released: true means the
number is out). Write a 1-2 sentence plain-English summary of the
current macro backdrop for a GOLD (XAU/USD) trader: the dominant theme
and its likely pull on gold. No advice, no hedging boilerplate, max 45
words.
Reply with ONLY JSON: {"summary": "..."}
"""


_RELEASE_VALUE_RE = re.compile(r"(-?\d+(?:\.\d+)?)\s*([kmbt])?", re.IGNORECASE)
_RELEASE_MULTIPLIERS = {"k": 1e3, "m": 1e6, "b": 1e9, "t": 1e12}


def _parse_release_value(raw: Optional[str]) -> Optional[float]:
    """Parse a calendar release value like '0.3%', '-1.7%', '228K', '1.1M'."""
    if not raw:
        return None
    match = _RELEASE_VALUE_RE.search(raw.replace(",", ""))
    if not match:
        return None
    multiplier = _RELEASE_MULTIPLIERS.get((match.group(2) or "").lower(), 1.0)
    return float(match.group(1)) * multiplier


@dataclass
class EconomicEvent:
    """Single economic calendar event."""
    time: datetime
    currency: str
    impact: str  # "low", "medium", "high"
    event: str
    forecast: Optional[str] = None
    previous: Optional[str] = None
    actual: Optional[str] = None

    @property
    def surprise(self) -> Optional[str]:
        """Actual vs forecast: "positive"/"negative"/"inline", or None when
        either value is missing or non-numeric. Direction is numeric only —
        it does not know whether higher is market-bullish for this event."""
        actual = _parse_release_value(self.actual)
        forecast = _parse_release_value(self.forecast)
        if actual is None or forecast is None:
            return None
        if actual > forecast:
            return "positive"
        if actual < forecast:
            return "negative"
        return "inline"

    @property
    def is_upcoming(self) -> bool:
        return self.time > datetime.now(timezone.utc)

    @property
    def minutes_until(self) -> float:
        delta = self.time - datetime.now(timezone.utc)
        return delta.total_seconds() / 60

    @property
    def is_high_impact(self) -> bool:
        return self.impact == "high"


@dataclass
class NewsSentiment:
    """News sentiment analysis result."""
    headline: str
    source: str
    timestamp: datetime
    currencies_mentioned: List[str]
    sentiment_score: float  # -1.0 (bearish) to +1.0 (bullish)
    keywords: List[str]
    # Per-currency score: monetary-policy keywords flip sign for XAU
    # (hawkish is dollar-bullish but gold-bearish).
    sentiment_by_currency: Dict[str, float] = field(default_factory=dict)


@dataclass
class FundamentalContext:
    """Complete fundamental/news context for signal decisions."""
    events_today: List[EconomicEvent] = field(default_factory=list)
    events_upcoming_1h: List[EconomicEvent] = field(default_factory=list)
    high_impact_warning: bool = False
    warning_message: str = ""
    news_sentiment: Dict[str, float] = field(default_factory=dict)  # currency -> sentiment
    news_headlines: List[NewsSentiment] = field(default_factory=list)
    event_impact_on_pairs: Dict[str, str] = field(default_factory=dict)
    calendar_available: bool = False  # True if live calendar data was fetched
    llm_summary: str = ""  # DeepSeek market-context blurb (display-only)


class NewsEngine:
    """
    Economic calendar and news sentiment analyzer.

    Provides fundamental context that modifies technical signal confidence.
    """

    # Keywords that indicate directional sentiment
    BULLISH_KEYWORDS = {
        "rate hike", "hawkish", "strong growth", "beat expectations",
        "above forecast", "above estimate", "above expectation",
        "surplus", "expansion", "employment gains",
        "strong jobs", "gdp growth", "inflation rising", "tightening",
        "positive surprise", "record high", "bullish", "rally",
        "stronger than expected", "upbeat",
        "beats", "surges", "soars", "jumps", "climbs", "rises",
        "higher than expected", "better than expected",
        "strong data", "robust", "resilient", "accelerates",
    }

    BEARISH_KEYWORDS = {
        "rate cut", "dovish", "recession", "slowdown", "contraction",
        "below forecast", "below estimate", "below expectation",
        "deficit", "miss expectations", "weak data",
        "unemployment rise", "negative surprise", "bearish", "selloff",
        "weaker than expected", "downbeat", "easing", "stimulus",
        "misses", "slides", "drops", "falls", "plunges", "tumbles",
        "weakens", "slumps", "declines", "retreats",
        "lower than expected", "worse than expected",
        "disappointing", "soft data",
    }

    # Monetary-policy keywords whose direction inverts for gold: tighter
    # policy lifts the dollar and yields, weighing on non-yielding gold;
    # easing does the opposite. Only unambiguous policy words — economic
    # data words (beats/misses) keep their normal direction.
    GOLD_INVERSE_KEYWORDS = {
        "rate hike", "hawkish", "tightening",
        "rate cut", "dovish", "easing", "stimulus",
    }

    # Currency-specific keywords for headline parsing. Matched with word
    # boundaries — bare "us"/"eu"/"uk" are deliberately excluded because as
    # substrings they false-positive constantly ("versus" → USD).
    CURRENCY_KEYWORDS = {
        "USD": ["dollar", "fed", "federal reserve", "fomc", "treasury", "american", "nonfarm", "nfp"],
        "EUR": ["euro", "ecb", "eurozone", "german", "french"],
        "GBP": ["pound", "sterling", "boe", "bank of england", "british"],
        "JPY": ["yen", "boj", "bank of japan", "japanese"],
        "CHF": ["franc", "snb", "swiss"],
        "AUD": ["aussie", "rba", "australian", "australia"],
        "CAD": ["loonie", "boc", "canadian", "canada"],
        "NZD": ["kiwi", "rbnz", "new zealand"],
        "XAU": ["gold", "precious metal", "safe haven", "bullion"],
    }

    # One compiled alternation per currency (code + keywords). (?<![a-z]) /
    # (?![a-z]) instead of \b so "0.3%fed" still matches but "versus",
    # "audit", "cascade" don't.
    _CURRENCY_PATTERNS = {
        cur: re.compile(
            r"(?<![a-z])(?:"
            + "|".join(re.escape(kw) for kw in [cur.lower()] + kws)
            + r")(?![a-z])"
        )
        for cur, kws in CURRENCY_KEYWORDS.items()
    }

    _calendar_backoff_until: float = 0.0

    # Fetched-data cache, class-level: a fresh NewsEngine is built per scan,
    # so instance state would never be reused. Guarded by _cache_lock since
    # scans run concurrently in a thread pool.
    _cache_lock = threading.Lock()
    _calendar_cache: Tuple[float, List[EconomicEvent]] = (0.0, [])
    _headlines_cache: Tuple[float, List[NewsSentiment]] = (0.0, [])
    _EMPTY_RESULT_TTL = 60.0  # retry sooner after a failed/empty fetch

    # DeepSeek result caches. Headlines and releases are immutable, so a
    # scored one never needs re-scoring; bounded by dropping the oldest
    # insertion. _llm_sent_cache is only touched under _cache_lock (its
    # writer runs inside _cached_fetch); the other two take the lock
    # around cache access but keep HTTP calls outside it.
    _llm_sent_cache: Dict[str, Tuple[float, float]] = {}   # sha1 → (s, xau)
    _llm_event_cache: Dict[Tuple[str, str, str], str] = {}
    _llm_summary_cache: Tuple[float, str] = (0.0, "")
    _LLM_SENT_CACHE_MAX = 400
    _LLM_EVENT_CACHE_MAX = 100

    # Per-source fetch health for status displays (telegram /status).
    # name → {"ok": ts|None, "err": ts|None, "detail": str, "extra": str}.
    # All writers run under _cache_lock (fetches happen inside
    # _cached_fetch); readers render without the lock — worst case a
    # display line is one cycle stale. A source is "failing" when its
    # last error is newer than its last success.
    _source_health: Dict[str, dict] = {}

    def __init__(self, config: NewsConfig):
        self.config = config
        self._session = requests.Session()
        self._session.headers.update({
            "User-Agent": "Mozilla/5.0 (compatible; ForexIntel/1.0)"
        })

    def get_fundamental_context(
        self, currencies: Optional[List[str]] = None
    ) -> FundamentalContext:
        """
        Get complete fundamental context for signal decisions.

        Combines economic calendar + news sentiment into actionable context.
        """
        if currencies is None:
            currencies = self.config.tracked_currencies

        context = FundamentalContext()

        # 1. Economic calendar events
        try:
            events = self._fetch_economic_calendar(currencies)
            now = datetime.now(timezone.utc)

            # Compare dates in UTC — feed timestamps carry their own offset
            # (FF sends US-Eastern), so a late-evening ET event would
            # otherwise land on the wrong day.
            context.events_today = [
                e for e in events
                if e.time.astimezone(timezone.utc).date() == now.date()
            ]
            context.events_upcoming_1h = [
                e for e in events
                if 0 < e.minutes_until <= 60
            ]
            context.calendar_available = bool(events)

            # Check for high-impact warnings
            for event in events:
                if event.is_high_impact and 0 < event.minutes_until <= self.config.pre_event_buffer_minutes:
                    context.high_impact_warning = True
                    context.warning_message = (
                        f"HIGH IMPACT: {event.currency} {event.event} "
                        f"in {event.minutes_until:.0f} min — signals may be unreliable"
                    )
                    break

            # Map upcoming high-impact events to affected pairs
            for event in context.events_upcoming_1h:
                if event.is_high_impact:
                    affected = self._get_affected_pairs(event.currency)
                    for pair in affected:
                        context.event_impact_on_pairs[pair] = (
                            f"{event.currency} {event.event} in {event.minutes_until:.0f}min"
                        )

        except Exception as e:
            logger.warning(f"Economic calendar fetch failed: {e}")

        # 2. News sentiment
        try:
            headlines = self._fetch_news_sentiment()
            context.news_headlines = headlines

            # Aggregate sentiment per currency (per-currency score where
            # available, so gold-inverted policy keywords are respected)
            for currency in currencies:
                scores = [
                    h.sentiment_by_currency.get(currency, h.sentiment_score)
                    for h in headlines
                    if currency in h.currencies_mentioned
                ]
                if scores:
                    context.news_sentiment[currency] = sum(scores) / len(scores)

        except Exception as e:
            logger.warning(f"News sentiment fetch failed: {e}")

        # 3. LLM market-context summary (display-only)
        try:
            context.llm_summary = self._llm_market_summary(context)
        except Exception as e:
            logger.debug(f"LLM market summary failed: {e}")

        return context

    @staticmethod
    def _parse_pair(pair_symbol: str) -> Tuple[str, str]:
        if "/" in pair_symbol:
            parts = pair_symbol.split("/")
            return parts[0], parts[1]
        clean = pair_symbol.replace("=X", "")
        return clean[:3], clean[3:]

    def get_event_risk_level(self, pair_symbol: str, context: FundamentalContext) -> Tuple[str, str]:
        """
        Get event risk level for a specific pair.
        Returns (risk_level, description).
        risk_level: "clear", "caution", "danger"
        """
        base, quote = self._parse_pair(pair_symbol)

        # Check upcoming high-impact events. Grade from the SOONEST
        # matching event, not the first listed — an unsorted feed could
        # otherwise report "caution" for a far event while a nearer one
        # warrants "danger".
        soonest = None
        for event in context.events_upcoming_1h:
            if event.is_high_impact and event.currency in (base, quote):
                if soonest is None or event.minutes_until < soonest.minutes_until:
                    soonest = event
        if soonest is not None:
            mins = soonest.minutes_until
            if mins <= self.config.pre_event_buffer_minutes:
                return "danger", f"{soonest.currency} {soonest.event} in {mins:.0f}min"
            return "caution", f"{soonest.currency} {soonest.event} in {mins:.0f}min"

        # Check for recent high-impact releases. A surprise print (actual
        # away from forecast) keeps the market unstable longer than an
        # inline release, so it gets a wider post-event window.
        for event in context.events_today:
            if event.is_high_impact and event.currency in (base, quote) and not event.is_upcoming:
                surprise = event.surprise
                surprised = surprise in ("positive", "negative")
                window = (
                    self.config.post_event_surprise_buffer_minutes
                    if surprised
                    else self.config.post_event_buffer_minutes
                )
                if event.minutes_until > -window:
                    if surprised:
                        desc = (
                            f"{event.currency} {event.event} released — {surprise} surprise "
                            f"(actual {event.actual} vs forecast {event.forecast})"
                        )
                        # LLM enrichment touches the DESCRIPTION only —
                        # `surprise` is numeric-direction and does not know
                        # whether a beat is gold-bullish; the LLM does.
                        gold_dir = self._llm_event_direction(event)
                        if gold_dir:
                            desc += f" — gold {gold_dir}"
                        return "caution", desc
                    return "caution", f"{event.currency} {event.event} just released"

        return "clear", "No significant events"

    def get_pair_sentiment(
        self, pair_symbol: str, context: FundamentalContext
    ) -> Tuple[float, str]:
        """
        Get net sentiment for a pair based on news.
        Returns (score, description) where score is -1 to +1.
        Positive = bullish for the pair, negative = bearish.
        """
        base, quote = self._parse_pair(pair_symbol)

        base_sent = context.news_sentiment.get(base, 0.0)
        quote_sent = context.news_sentiment.get(quote, 0.0)

        # Net sentiment: positive base or negative quote = bullish for pair
        net = base_sent - quote_sent

        if abs(net) < 0.1:
            desc = "Neutral news sentiment"
        elif net > 0.3:
            desc = f"Bullish sentiment ({base} positive / {quote} negative)"
        elif net > 0:
            desc = "Slightly bullish sentiment"
        elif net < -0.3:
            desc = f"Bearish sentiment ({base} negative / {quote} positive)"
        else:
            desc = "Slightly bearish sentiment"

        return net, desc

    def get_week_events(
        self, currencies: Optional[List[str]] = None
    ) -> List[EconomicEvent]:
        """Full cached calendar week (past and upcoming), filtered to
        `currencies` (default: tracked_currencies), sorted by time.

        Read-only accessor for display surfaces (the web dashboard) —
        rides the same class-level cache as signal scans, so it adds no
        network I/O. No production signal path calls this.
        """
        if currencies is None:
            currencies = self.config.tracked_currencies
        events = self._cached_fetch("_calendar_cache", self._fetch_calendar_raw)
        return sorted(
            (e for e in events if e.currency in currencies),
            key=lambda e: e.time,
        )

    # ── Private Methods ────────────────────────────────────────────

    # Maps country names in MyFXBook titles to currency codes
    _COUNTRY_TO_CURRENCY = {
        "united states": "USD", "euro area": "EUR", "united kingdom": "GBP",
        "japan": "JPY", "switzerland": "CHF", "australia": "AUD",
        "canada": "CAD", "new zealand": "NZD",
    }

    @classmethod
    def _record_source(cls, name: str, ok: bool, detail: str = "",
                       extra: str = "") -> None:
        """Track one dependency's fetch outcome for status displays.
        Logs state TRANSITIONS so the log file carries the story without
        a line every 10-min cycle: first failure WARNING (with detail),
        repeat failures DEBUG, recovery INFO."""
        h = cls._source_health.setdefault(
            name, {"ok": None, "err": None, "detail": "", "extra": ""})
        was_failing = h["err"] is not None and (
            h["ok"] is None or h["err"] > h["ok"])
        if ok:
            h["ok"] = time.time()
            h["extra"] = extra
            if was_failing:
                logger.info("Source recovered: %s", name)
        else:
            h["err"] = time.time()
            h["detail"] = detail
            if was_failing:
                logger.debug("Source still failing: %s (%s)", name, detail)
            else:
                logger.warning("Source failing: %s (%s)", name, detail)

    @staticmethod
    def source_is_failing(entry: Optional[dict]) -> bool:
        """True when the entry's last error is newer than its last ok."""
        return bool(entry and entry["err"] is not None and
                    (entry["ok"] is None or entry["err"] > entry["ok"]))

    def _cached_fetch(self, cache_attr: str, fetch_fn) -> list:
        """Return cached data for `cache_attr` or refetch when the TTL
        expired. Successful results live for context_cache_ttl_seconds;
        empty results only _EMPTY_RESULT_TTL so outages retry quickly.
        The lock is held across the fetch so concurrent scans share one
        request instead of stampeding the source."""
        ts, data = getattr(NewsEngine, cache_attr)
        ttl = self.config.context_cache_ttl_seconds if data else self._EMPTY_RESULT_TTL
        if time.time() - ts < ttl:
            return data

        with NewsEngine._cache_lock:
            ts, data = getattr(NewsEngine, cache_attr)  # re-check under lock
            ttl = self.config.context_cache_ttl_seconds if data else self._EMPTY_RESULT_TTL
            if time.time() - ts < ttl:
                return data
            data = fetch_fn()
            setattr(NewsEngine, cache_attr, (time.time(), data))
            return data

    def _fetch_economic_calendar(self, currencies: List[str]) -> List[EconomicEvent]:
        """Cached calendar fetch, filtered to the tracked currencies."""
        events = self._cached_fetch("_calendar_cache", self._fetch_calendar_raw)
        return [e for e in events if e.currency in currencies]

    def _fetch_calendar_raw(self) -> List[EconomicEvent]:
        """
        Fetch economic calendar events from a free API.
        Tries Forex Factory first, falls back to MyFXBook RSS.
        Returns an empty list (with a log warning) if unavailable —
        we never generate fake events since that causes constant false alarms.
        """
        sources = (
            (self._try_fetch_calendar_api, "calendar:forexfactory",
             "ForexFactory"),
            (self._try_fetch_myfxbook_rss, "calendar:myfxbook", "MyFXBook"),
        )
        for fetcher, src, label in sources:
            try:
                events = fetcher()
                if events:
                    self._record_source(src, True)
                    self._record_source(
                        "calendar", True,
                        extra=f"{len(events)} events via {label}")
                    return events
                self._record_source(src, False, "no events returned")
            except Exception as e:
                self._record_source(
                    src, False, f"{type(e).__name__}: {e}")
                logger.debug(f"Calendar fetch exception ({fetcher.__name__}): {e}")

        self._record_source("calendar", False, "all sources failed")
        logger.info(
            "Economic calendar data unavailable. "
            "Event risk checks will be skipped for this cycle."
        )
        return []

    def _try_fetch_calendar_api(self) -> List[EconomicEvent]:
        """Attempt to fetch from the Forex Factory free JSON calendar."""
        now = time.time()
        if now < NewsEngine._calendar_backoff_until:
            wait = NewsEngine._calendar_backoff_until - now
            logger.debug(f"Calendar API in backoff for {wait:.0f}s more")
            return []

        url = "https://nfs.faireconomy.media/ff_calendar_thisweek.json"

        for attempt in range(3):
            try:
                resp = self._session.get(url, timeout=10)
            except requests.RequestException as e:
                logger.debug(f"Calendar request failed (attempt {attempt + 1}): {e}")
                if attempt < 2:
                    time.sleep(2 ** attempt)
                continue

            if resp.status_code == 429:
                retry_after = int(resp.headers.get("Retry-After", 300))
                retry_after = min(retry_after, 600)
                NewsEngine._calendar_backoff_until = time.time() + retry_after
                logger.warning(
                    f"Calendar API rate-limited (429), backing off {retry_after}s"
                )
                return []

            if resp.status_code != 200:
                logger.debug(f"Calendar API returned {resp.status_code}")
                return []

            data = resp.json()
            events = []
            impact_map = {"High": "high", "Medium": "medium", "Low": "low"}

            for item in data:
                try:
                    impact = impact_map.get(item.get("impact", ""), "low")

                    date_str = item.get("date", "")
                    if not date_str:
                        continue
                    event_time = datetime.fromisoformat(date_str.replace("Z", "+00:00"))

                    events.append(EconomicEvent(
                        time=event_time,
                        currency=item.get("country", ""),
                        impact=impact,
                        event=item.get("title", ""),
                        forecast=item.get("forecast") or None,
                        previous=item.get("previous") or None,
                        actual=item.get("actual") or None,
                    ))
                except Exception:
                    continue

            return events

        return []

    def _try_fetch_myfxbook_rss(self) -> List[EconomicEvent]:
        """Fallback calendar source: MyFXBook economic calendar RSS."""
        url = "https://www.myfxbook.com/rss/forex-economic-calendar-events"
        try:
            resp = self._session.get(url, timeout=10)
        except requests.RequestException as e:
            logger.debug(f"MyFXBook calendar request failed: {e}")
            return []

        if resp.status_code != 200:
            return []

        try:
            root = ET.fromstring(resp.content)
        except ET.ParseError:
            return []

        events = []
        for item in root.iter("item"):
            try:
                title_el = item.find("title")
                pub_el = item.find("pubDate")
                desc_el = item.find("description")
                if title_el is None or title_el.text is None:
                    continue

                title = title_el.text.strip()

                currency = ""
                title_lower = title.lower()
                for country, cur in self._COUNTRY_TO_CURRENCY.items():
                    if title_lower.startswith(country):
                        currency = cur
                        title = title[len(country):].strip()
                        break
                if not currency:
                    continue

                event_time = self._parse_rss_date(
                    pub_el.text if pub_el is not None else None
                )
                if event_time is None:
                    continue

                impact = "low"
                desc_text = desc_el.text if desc_el is not None and desc_el.text else ""
                impact_match = re.search(r"sprite-(high|medium|low)-impact", desc_text)
                if impact_match:
                    impact = impact_match.group(1)

                cells = re.findall(r"<td[^>]*>(.*?)</td>", desc_text, re.DOTALL)
                cells = [re.sub(r"<[^>]+>", "", c).strip() for c in cells]
                previous = cells[2] if len(cells) > 2 and cells[2] else None
                forecast = cells[3] if len(cells) > 3 and cells[3] else None
                actual = cells[4] if len(cells) > 4 and cells[4] else None

                events.append(EconomicEvent(
                    time=event_time,
                    currency=currency,
                    impact=impact,
                    event=title,
                    forecast=forecast,
                    previous=previous,
                    actual=actual,
                ))
            except Exception:
                continue

        logger.debug(f"MyFXBook calendar: parsed {len(events)} events")
        return events

    def _fetch_news_sentiment(self) -> List[NewsSentiment]:
        """Cached headline fetch + sentiment analysis."""
        return self._cached_fetch("_headlines_cache", self._fetch_news_sentiment_raw)

    def _fetch_news_sentiment_raw(self) -> List[NewsSentiment]:
        """Fetch and analyze news headlines for sentiment."""
        headlines = []
        seen_titles = set()
        ok_feeds = 0

        for feed_url in self.config.rss_feeds:
            host = feed_url.split("/")[2] if "//" in feed_url else feed_url
            kept_before = len(headlines)
            try:
                resp = self._session.get(feed_url, timeout=10)
                if resp.status_code != 200:
                    self._record_source(
                        f"rss:{host}", False, f"HTTP {resp.status_code}")
                    continue

                root = ET.fromstring(resp.content)

                for item in root.iter("item"):
                    title_el = item.find("title")
                    pub_date_el = item.find("pubDate")

                    if title_el is None:
                        continue

                    title = title_el.text or ""
                    pub_date = self._parse_rss_date(
                        pub_date_el.text if pub_date_el is not None else None
                    )

                    # Only recent headlines (last 24h)
                    if pub_date and (datetime.now(timezone.utc) - pub_date).total_seconds() > 86400:
                        continue

                    # Dedup within and across feeds so a syndicated story
                    # doesn't count twice in the sentiment average. Google
                    # News appends " - Publisher", so compare without it.
                    title_key = title.lower().rsplit(" - ", 1)[0].strip()
                    if title_key in seen_titles:
                        continue
                    seen_titles.add(title_key)

                    sentiment = self._analyze_headline_sentiment(title)
                    currencies = self._extract_currencies(title)

                    if currencies:
                        headlines.append(NewsSentiment(
                            headline=title,
                            source=feed_url,
                            timestamp=pub_date or datetime.now(timezone.utc),
                            currencies_mentioned=currencies,
                            sentiment_score=sentiment,
                            keywords=self._extract_keywords(title),
                            sentiment_by_currency={
                                cur: self._analyze_headline_sentiment(title, cur)
                                for cur in currencies
                            },
                        ))

                ok_feeds += 1
                self._record_source(
                    f"rss:{host}", True,
                    extra=f"{len(headlines) - kept_before} kept")
            except Exception as e:
                self._record_source(
                    f"rss:{host}", False, f"{type(e).__name__}: {e}")
                logger.debug(f"Failed to fetch RSS from {feed_url}: {e}")

        self._record_source(
            "headlines", ok_feeds > 0,
            detail="all feeds failed" if ok_feeds == 0 else "",
            extra=(f"{len(headlines)} headlines from "
                   f"{ok_feeds}/{len(self.config.rss_feeds)} feeds"))

        try:
            self._apply_llm_sentiment(headlines)
        except Exception as e:
            logger.debug(f"LLM sentiment failed, keyword scores stand: {e}")

        return headlines

    def _analyze_headline_sentiment(
        self, headline: str, for_currency: Optional[str] = None
    ) -> float:
        """
        Simple keyword-based sentiment analysis.

        Returns -1.0 (very bearish) to +1.0 (very bullish).

        When for_currency is "XAU", monetary-policy keywords contribute
        with inverted sign (see GOLD_INVERSE_KEYWORDS).

        Note: This is intentionally simple. NLP models like FinBERT
        would be more accurate but require large dependencies. This catches
        the most impactful sentiment signals from headlines.
        """
        headline_lower = headline.lower()
        invert_policy = for_currency == "XAU"

        bullish_score = 0
        bearish_score = 0
        for kw in self.BULLISH_KEYWORDS:
            if kw in headline_lower:
                if invert_policy and kw in self.GOLD_INVERSE_KEYWORDS:
                    bearish_score += 1
                else:
                    bullish_score += 1
        for kw in self.BEARISH_KEYWORDS:
            if kw in headline_lower:
                if invert_policy and kw in self.GOLD_INVERSE_KEYWORDS:
                    bullish_score += 1
                else:
                    bearish_score += 1

        total = bullish_score + bearish_score
        if total == 0:
            return 0.0

        return (bullish_score - bearish_score) / total

    # ── DeepSeek enrichment (display/warning-only, keyword fallback) ──

    def _llm_headline_scores(
        self, texts: List[str]
    ) -> Optional[List[Tuple[float, float]]]:
        """One batched DeepSeek call → [(overall, xau), ...] aligned with
        `texts`, or None on any failure/mismatch (caller keeps keyword
        scores). Tests stub this method directly."""
        result = get_llm_client().chat_json(
            _LLM_SENTIMENT_SYSTEM, {"headlines": texts})
        if not result:
            return None
        scores = result.get("scores")
        if not isinstance(scores, list) or len(scores) != len(texts):
            return None
        out = []
        for item in scores:
            try:
                s = max(-1.0, min(1.0, float(item[0])))
                x = max(-1.0, min(1.0, float(item[1])))
            except (TypeError, KeyError, ValueError, IndexError):
                return None
            out.append((s, x))
        return out

    def _apply_llm_sentiment(self, headlines: List[NewsSentiment]) -> None:
        """Overwrite keyword sentiment on `headlines` with LLM scores.

        Runs inside _cached_fetch, i.e. UNDER _cache_lock — cache access
        is direct and the (rare) HTTP call blocks sibling scans at most
        once per cache window, matching the existing shared-fetch design.
        Only headlines missing from the cache are sent; a failed batch
        changes nothing.
        """
        if not headlines or not get_llm_client().available:
            return
        cache = NewsEngine._llm_sent_cache
        keyed = [(hashlib.sha1(h.headline.encode()).hexdigest(), h)
                 for h in headlines]
        # Cap the batch so the reply fits the client's max_tokens — the
        # overflow stays keyword-scored this cycle and, still uncached,
        # goes out with the next window's batch.
        new = [(k, h) for k, h in keyed if k not in cache][:25]
        if new:
            scores = self._llm_headline_scores([h.headline for _, h in new])
            if scores is not None:
                for (k, _), sc in zip(new, scores):
                    cache[k] = sc
                while len(cache) > self._LLM_SENT_CACHE_MAX:
                    cache.pop(next(iter(cache)))
        for k, h in keyed:
            sc = cache.get(k)
            if sc is None:
                continue
            overall, xau = sc
            h.sentiment_score = overall
            h.sentiment_by_currency = {
                cur: (xau if cur == "XAU" else overall)
                for cur in h.currencies_mentioned
            }

    def _llm_event_direction(self, event: EconomicEvent) -> Optional[str]:
        """Gold direction for a released surprise, e.g. "bearish: strong
        jobs lift yields". None → caller keeps the numeric description.
        Cached per release; failures are NOT cached (retry next scan).

        The actual-vs-forecast comparison is precomputed and sent as
        "vs_forecast": the 2026-08-06 validation found the model reads
        close K/M-suffixed pairs backwards ~11% of the time (e.g. claims
        202K vs 217K called "higher than forecast") when left to compare
        the raw numbers itself."""
        key = (event.event, event.currency, str(event.actual or ""))
        with NewsEngine._cache_lock:
            if key in NewsEngine._llm_event_cache:
                return NewsEngine._llm_event_cache[key] or None
        client = get_llm_client()
        if not client.available:
            return None
        vs_forecast = {
            "positive": "above forecast",
            "negative": "below forecast",
            "inline": "in line with forecast",
        }.get(event.surprise or "", "unknown")
        result = client.chat_json(_LLM_EVENT_SYSTEM, {
            "event": event.event,
            "currency": event.currency,
            "actual": event.actual,
            "forecast": event.forecast,
            "previous": event.previous,
            "vs_forecast": vs_forecast,
        })
        if not result:
            return None
        direction = result.get("gold_direction")
        if direction not in ("bullish", "bearish", "neutral"):
            return None
        reason = str(result.get("reason", "")).strip()[:80]
        text = "" if direction == "neutral" else (
            f"{direction}: {reason}" if reason else direction)
        with NewsEngine._cache_lock:
            NewsEngine._llm_event_cache[key] = text
            while len(NewsEngine._llm_event_cache) > self._LLM_EVENT_CACHE_MAX:
                NewsEngine._llm_event_cache.pop(
                    next(iter(NewsEngine._llm_event_cache)))
        return text or None

    def _llm_market_summary(self, context: FundamentalContext) -> str:
        """1-2 sentence macro blurb for alerts; "" when unavailable.
        Cached (success OR failure) for the context-cache TTL so at most
        one call is made per window."""
        with NewsEngine._cache_lock:
            ts, cached = NewsEngine._llm_summary_cache
            if time.time() - ts < self.config.context_cache_ttl_seconds:
                return cached
        high_impact = [e for e in context.events_today if e.is_high_impact]
        if not context.news_headlines and not high_impact:
            return ""
        client = get_llm_client()
        if not client.available:
            return ""
        result = client.chat_json(_LLM_SUMMARY_SYSTEM, {
            "headlines": [h.headline for h in context.news_headlines[:15]],
            "events": [
                {"event": e.event, "currency": e.currency,
                 "actual": e.actual, "forecast": e.forecast,
                 "released": not e.is_upcoming}
                for e in high_impact[:10]
            ],
        })
        summary = ""
        if result:
            summary = str(result.get("summary", "")).strip()[:300]
        with NewsEngine._cache_lock:
            NewsEngine._llm_summary_cache = (time.time(), summary)
        return summary

    def _extract_currencies(self, headline: str) -> List[str]:
        """Extract mentioned currencies from a headline."""
        headline_lower = headline.lower()
        return [
            currency
            for currency, pattern in self._CURRENCY_PATTERNS.items()
            if pattern.search(headline_lower)
        ]

    def _extract_keywords(self, headline: str) -> List[str]:
        """Extract relevant trading keywords from headline."""
        headline_lower = headline.lower()
        return [kw for kw in self.BULLISH_KEYWORDS | self.BEARISH_KEYWORDS if kw in headline_lower]

    @staticmethod
    def _parse_rss_date(date_str: Optional[str]) -> Optional[datetime]:
        """Parse various RSS date formats into a timezone-aware datetime."""
        if not date_str:
            return None

        # RFC 822/2822 forms, with or without seconds, numeric offsets AND
        # named zones (GMT, EST/EDT, PST…). The old strptime "%Z" path
        # accepted ANY abbreviation but attached no offset, so e.g. an EST
        # pubDate was silently stamped UTC — a 5-hour error.
        try:
            dt = parsedate_to_datetime(date_str.strip())
            if dt is not None:
                if dt.tzinfo is None:   # unknown zone name — assume UTC
                    return dt.replace(tzinfo=timezone.utc)
                return dt.astimezone(timezone.utc)
        except (TypeError, ValueError):
            pass

        formats = [
            "%Y-%m-%dT%H:%M:%S%z",         # ISO 8601 with offset
            "%Y-%m-%d %H:%M:%S",           # ISO 8601 naive (assume UTC)
        ]
        for fmt in formats:
            try:
                dt = datetime.strptime(date_str.strip(), fmt)
                # If strptime already attached tzinfo (from %z), keep it.
                # Otherwise the string had no timezone info — assume UTC.
                if dt.tzinfo is not None:
                    return dt
                return dt.replace(tzinfo=timezone.utc)
            except ValueError:
                continue
        return None

    @staticmethod
    def _get_affected_pairs(currency: str) -> List[str]:
        _MAP = {
            "USD": ["XAU/USD"],
            "XAU": ["XAU/USD"],
        }
        return _MAP.get(currency, [])