"""Input/output models for the v3→v2 compatibility adapter.

V3IntelligenceRecord represents the full v3 pipeline output.
V2IntelligenceRecord / V2ImpactRecord match the current document_intelligence
and document_impact_records database schemas.
AdapterLineage captures version and stage provenance.
"""

from __future__ import annotations

import uuid
from datetime import datetime, timezone
from typing import Literal

from pydantic import BaseModel, Field

# ---------------------------------------------------------------------------
# V3 Pipeline Output (input to adapter)
# ---------------------------------------------------------------------------


class V3SentimentDistribution(BaseModel):
    """Per-company calibrated sentiment probabilities."""

    positive: float = Field(ge=0.0, le=1.0)
    negative: float = Field(ge=0.0, le=1.0)
    neutral: float = Field(ge=0.0, le=1.0)


class V3HorizonProbabilities(BaseModel):
    """Probability distribution over impact horizons."""

    intraday: float = Field(ge=0.0, le=1.0, default=0.0)
    one_day: float = Field(ge=0.0, le=1.0, default=0.0)
    seven_day: float = Field(ge=0.0, le=1.0, default=0.0)
    thirty_day: float = Field(ge=0.0, le=1.0, default=0.0)
    ninety_day: float = Field(ge=0.0, le=1.0, default=0.0)


class V3DirectionProbabilities(BaseModel):
    """Probability distribution over market direction."""

    positive: float = Field(ge=0.0, le=1.0, default=0.0)
    negative: float = Field(ge=0.0, le=1.0, default=0.0)
    neutral: float = Field(ge=0.0, le=1.0, default=0.0)


class V3CompanySignal(BaseModel):
    """A single company's signal from the v3 pipeline."""

    company_id: str
    ticker: str
    relevance_probability: float = Field(ge=0.0, le=1.0)
    event_classes: list[str] = Field(default_factory=list)
    sentiment: V3SentimentDistribution
    direction_probabilities: V3DirectionProbabilities
    horizon_probabilities: V3HorizonProbabilities
    expected_magnitude: float | None = None
    evidence_spans: list[str] = Field(default_factory=list)
    adjudicated: bool = False


class V3StageRun(BaseModel):
    """Lineage for a single pipeline stage execution."""

    stage: str
    endpoint_id: str | None = None
    deployment_id: str | None = None
    model_version: str | None = None
    schema_version: str = "1.0.0"
    calibration_version: str | None = None
    started_at: datetime = Field(default_factory=lambda: datetime.now(tz=timezone.utc))
    duration_ms: int = 0
    status: str = "completed"


class V3IntelligenceRecord(BaseModel):
    """Complete v3 pipeline output for a single document.

    This is the adapter's input — the full v3 record with probabilities,
    evidence, and stage lineage.
    """

    document_id: str = Field(default_factory=lambda: str(uuid.uuid4()))
    document_type: str = "article"
    summary: str = ""
    macro_themes: list[str] = Field(default_factory=list)
    novelty_score: float = Field(ge=0.0, le=1.0, default=0.5)
    confidence: float = Field(ge=0.0, le=1.0, default=0.5)
    company_signals: list[V3CompanySignal] = Field(default_factory=list)
    stage_runs: list[V3StageRun] = Field(default_factory=list)
    pipeline_version: str = "3.0.0"
    created_at: datetime = Field(default_factory=lambda: datetime.now(tz=timezone.utc))


# ---------------------------------------------------------------------------
# V2 Output (adapter output — matches current DB schema)
# ---------------------------------------------------------------------------


class V2ImpactRecord(BaseModel):
    """Maps to document_impact_records table.

    Fields match the columns: relevance, sentiment (enum string),
    impact_score (float), impact_horizon (string), catalyst_type,
    key_facts, risks, evidence_spans.
    """

    id: str = Field(default_factory=lambda: str(uuid.uuid4()))
    company_id: str
    ticker: str
    relevance: float = Field(ge=0.0, le=1.0)
    sentiment: Literal["positive", "negative", "neutral", "mixed"]
    impact_score: float = Field(ge=-1.0, le=1.0)
    impact_horizon: Literal["intraday", "1d", "7d", "30d", "90d"]
    catalyst_type: str = "other"
    key_facts: list[str] = Field(default_factory=list)
    risks: list[str] = Field(default_factory=list)
    evidence_spans: list[str] = Field(default_factory=list)


class V2IntelligenceRecord(BaseModel):
    """Maps to document_intelligence table.

    Fields match columns: summary, macro_themes, novelty_score,
    source_credibility, confidence, model_provider, model_name,
    prompt_version, schema_version, plus associated impact records.
    """

    id: str = Field(default_factory=lambda: str(uuid.uuid4()))
    document_id: str
    summary: str = ""
    macro_themes: list[str] = Field(default_factory=list)
    novelty_score: float = Field(ge=0.0, le=1.0)
    source_credibility: float = Field(ge=0.0, le=1.0, default=0.5)
    confidence: float = Field(ge=0.0, le=1.0)
    model_provider: str = "hybrid"
    model_name: str = "intelligence-pipeline-v3"
    prompt_version: str = ""
    schema_version: str = "3.0.0"
    impact_records: list[V2ImpactRecord] = Field(default_factory=list)
    created_at: datetime = Field(default_factory=lambda: datetime.now(tz=timezone.utc))


# ---------------------------------------------------------------------------
# Adapter Lineage
# ---------------------------------------------------------------------------


class AdapterLineage(BaseModel):
    """Records which adapter version produced the v2 record and from what v3 data.

    Stored separately so v3 provenance is never lost.
    """

    adapter_version: str = "1.0.0"
    pipeline_version: str = "3.0.0"
    v3_document_id: str
    v2_intelligence_id: str
    stage_runs: list[V3StageRun] = Field(default_factory=list)
    mapped_at: datetime = Field(default_factory=lambda: datetime.now(tz=timezone.utc))
    mapping_notes: list[str] = Field(default_factory=list)