"""Invariants of the cleaned panel (run after `make clean`)."""
import json, pandas as pd, pytest
from src.common import ROOT, P, REGIONS

@pytest.fixture(scope="module")
def li():
    return pd.read_parquet(P["interim"] / "listings_clean.parquet")

@pytest.fixture(scope="module")
def sm():
    return pd.read_parquet(P["interim"] / "stores_master.parquet")

def test_flow_monotone():
    flow = json.load(open(ROOT / "experiments" / "sample_flow.json"))["flow"]
    vals = [v for _, v in flow]
    assert all(a >= b for a, b in zip(vals, vals[1:]))

def test_price_sample_subset(li):
    assert not (li.f_analysis_price & ~li.f_analysis_assort).any()
    s = li[li.f_analysis_price]
    assert (s.price > 0).all() and (s.price <= 50_000).all() and (s.currency == "CAD").all()

def test_one_rep_per_store_title(li):
    a = li[li.f_analysis_assort]
    assert not a.duplicated(["store_id", "title_norm"]).any()

def test_no_sibling_nonrep_in_assortment(li):
    assert not li.loc[li.f_analysis_assort, "f_sibling_dup"].any()

def test_regions_canonical(sm):
    assert set(sm.region_final.unique()) <= set(REGIONS) | {"Inconnue"}
    assert sm.id.is_unique

def test_store_listing_counts_consistent(li, sm):
    a = li[li.f_analysis_assort].groupby("store_id").size()
    m = sm.set_index("id").n_listings
    assert (a.reindex(m.index).fillna(0).astype(int) == m).all()

def test_match_pairs_cross_store():
    pr = pd.read_parquet(P["interim"] / "match_pairs.parquet")
    assert (pr.store_a != pr.store_b).all() and pr.tier.isin(list("ABCD")).all()
    sib = pd.read_parquet(P["interim"] / "sibling_stores.parquet")
    rep = sib.set_index("id").sibling_rep
    assert not (pr.store_a.map(rep).fillna(pr.store_a) == pr.store_b.map(rep).fillna(pr.store_b)).any()