"""Event-level feature calculations from parser output."""
from __future__ import annotations

from typing import Any

from libs.schemas.types import ConfidenceOutput, GuidanceOutput, RiskFlagsOutput, SignalsOutput


def guidance_direction_score(guidance: GuidanceOutput) -> float:
    """raised=1.0, inline_or_maintained=0.5, lowered=0.0, else=0.25."""
    mapping = {
        "raised": 1.0,
        "inline_or_maintained": 0.5,
        "lowered": 0.0,
        "withdrawn": 0.0,
        "not_provided": 0.25,
        "unclear": 0.25,
    }
    return mapping.get(guidance.status, 0.25)


def oneoff_penalty(risk_flags: RiskFlagsOutput) -> float:
    """Sum of active risk flags / total flags. Higher = more risk."""
    flags = [
        risk_flags.oneoff_item,
        risk_flags.tax_benefit,
        risk_flags.valuation_gain,
        risk_flags.non_gaap_heavy,
        risk_flags.financing_related,
        risk_flags.legal_or_regulatory_overhang,
    ]
    total = len(flags)
    active = sum(flags)
    return active / total if total > 0 else 0.0


def signal_strength_score(signals: SignalsOutput) -> float:
    """Composite score [0..1] of positive business signals."""
    score = 0.0
    weights = {
        "demand_strength": {"strong": 1.0, "stable": 0.5, "weakening": 0.0, "unknown": 0.0},
        "pricing_power": {"present": 1.0, "mixed": 0.5, "absent": 0.0, "unknown": 0.0},
        "backlog_or_bookings": {"present": 1.0, "mixed": 0.5, "absent": 0.0, "unknown": 0.0},
        "customer_expansion": {"present": 1.0, "mixed": 0.5, "absent": 0.0, "unknown": 0.0},
        "margin_quality": {"improving": 1.0, "stable": 0.5, "deteriorating": 0.0, "unknown": 0.0},
    }
    total_weight = len(weights)
    for field, mapping in weights.items():
        val = getattr(signals, field)
        score += mapping.get(val, 0.0)
    return score / total_weight if total_weight > 0 else 0.0


def document_quality_score(confidence: ConfidenceOutput) -> float:
    """Weighted average of confidence dimensions."""
    return (
        confidence.overall * 0.4
        + confidence.event_type * 0.2
        + confidence.event_direction * 0.2
        + confidence.guidance * 0.1
        + confidence.risk_flags * 0.1
    )


def compute_event_features(parser_output: dict[str, Any]) -> dict[str, Any]:
    """Compute all event features from raw parser output dict."""
    from libs.schemas.types import (
        ConfidenceOutput,
        GuidanceOutput,
        RiskFlagsOutput,
        SignalsOutput,
    )

    guidance = GuidanceOutput.model_validate(parser_output["guidance"])
    signals = SignalsOutput.model_validate(parser_output["signals"])
    risk_flags = RiskFlagsOutput.model_validate(parser_output["risk_flags"])
    confidence = ConfidenceOutput.model_validate(parser_output["confidence"])

    return {
        "guidance_direction_score": guidance_direction_score(guidance),
        "guidance_status": guidance.status,
        "oneoff_penalty": oneoff_penalty(risk_flags),
        "signal_strength_score": signal_strength_score(signals),
        "document_quality_score": document_quality_score(confidence),
        "event_type": parser_output.get("event_type", "unknown"),
        "event_direction": parser_output.get("event_direction", "unknown"),
        "parse_confidence_overall": confidence.overall,
        "parse_confidence_event_direction": confidence.event_direction,
        "parse_confidence_guidance": confidence.guidance,
        "filing_time_bucket": parser_output.get("filing_time_bucket", "unknown"),
        "event_date": parser_output.get("event_date", ""),
    }