"""End-to-end pipeline test on the clearly labelled synthetic fixture."""
import json
import os
import shutil
from pathlib import Path

import pytest


def test_pipeline_on_synthetic(tmp_path, monkeypatch):
    from ucml import synthetic
    db = synthetic.make(tmp_path / "synthetic" / "autoka-synthetic.db", n_listings=1500)
    monkeypatch.setenv("UCML_ROOT", str(tmp_path))
    monkeypatch.setenv("UCML_SNAPSHOT", str(db))
    import importlib
    from ucml import config
    importlib.reload(config)
    from ucml import ingest, cleaning
    importlib.reload(ingest); importlib.reload(cleaning)
    (tmp_path / "data" / "raw" / "registry").mkdir(parents=True)
    real_reg = Path(__file__).resolve().parents[1] / "data" / "raw" / "registry"
    srcs = {"_comment": "synthetic", "sources": [{"id": f"synthdealer{i:02d}", "name": f"D{i}", "type": "direct", "region": "Montréal"} for i in range(20)]
            + [{"id": "otogo", "name": "Otogo", "type": "portail"}]}
    (tmp_path / "data" / "raw" / "registry" / "sources.json").write_text(json.dumps(srcs))
    (tmp_path / "data" / "raw" / "registry" / "villes_gps.json").write_text(json.dumps({"ville": [45.5, -73.6]}))
    m = ingest.run(snapshot=db, outdir=tmp_path / "data" / "interim", with_hash=False)
    assert m["rows"]["vehicles"] == 1500
    importlib.reload(config)
    stats = cleaning.run(interim=tmp_path / "data" / "interim", out=tmp_path / "data" / "processed", registry=tmp_path / "data" / "raw" / "registry")
    assert stats["S0"] == 1500
    assert stats["S1_hedonic"] > 1000
    assert (tmp_path / "data" / "processed" / "exclusions.csv").exists()
    assert (tmp_path / "data" / "processed" / "price_events.parquet").exists()


def test_first_cut_never_null_in_risk_set():
    """Regression test for E-016: first_cut must be 0 (not NULL) for listing-days of never-cut listings."""
    import polars as pl
    from ucml import config as C
    p = C.PROCESSED / "risk_set.parquet"
    if not p.exists():
        pytest.skip("risk set not built")
    rs = pl.read_parquet(p, columns=["first_cut", "cut", "exit"])
    assert rs["first_cut"].null_count() == 0 and rs["cut"].null_count() == 0 and rs["exit"].null_count() == 0