src/densefeed/models.py

"""Shared data models for DenseFeed pipeline."""

from __future__ import annotations

from typing import Literal

from pydantic import BaseModel


class Item(BaseModel):
    """Uniform output from all fetchers."""

    source: str
    title: str
    url: str
    summary: str = ""
    published_ts: float = 0.0
    score: float | None = None
    reason: str | None = None


class Article(BaseModel):
    """Extracted article content."""

    url: str
    title: str
    body: str
    author: str = ""
    published: str = ""


class ScriptLine(BaseModel):
    """A single line of podcast dialogue."""

    speaker: Literal["A", "B"]
    text: str
    emotion: str = "neutral"


class Script(BaseModel):
    """Full podcast script."""

    date: str
    title: str
    lines: list[ScriptLine]


class StageResult(BaseModel):
    """Metadata for a completed pipeline stage."""

    stage: int
    name: str
    success: bool
    duration_s: float = 0.0
    output_path: str = ""
    error: str | None = None
    items_in: int = 0
    items_out: int = 0


class PipelineRun(BaseModel):
    """Full pipeline run state."""

    date: str
    register_dir: str = ""
    run_id: str = ""  # e.g. "2026-06-06_001"
    user_id: str = ""  # HF username (empty for CLI)
    stages: list[StageResult] = []
    started_at: str = ""
    finished_at: str = ""
    config_json: dict = {}  # snapshot of effective config for reproducibility


class UserPreferences(BaseModel):
    """Per-user preferences stored in DuckDB.

    This model is the single source of truth for preference defaults and schema.
    DuckDBBackend.DEFAULT_PREFS is derived from this model's defaults.
    """

    user_id: str = ""
    topics: list[str] = []
    excluded_sources: list[str] = []
    speaker_a_voice: str = "en-US-AndrewMultilingualNeural"
    speaker_b_voice: str = "en-US-AvaMultilingualNeural"
    speaker_a_rate: str = "+10%"
    speaker_b_rate: str = "+10%"
    cross_speaker_pause_ms: int = 550
    same_speaker_pause_ms: int = 250
    min_relevance: int = 6
    top_n_items: int = 12
    hf_dataset_repo: str = ""
    hf_audio_repo: str = ""