"""Regression: dataset YAML files parse and have the expected shape.

Bug captured: schema_docs.yaml had values that began with a quote character
(e.g. `gender: 'M' or 'F'.`), which YAML treats as a quoted scalar — anything
after the closing quote is a parse error.

The YAML structure is now flat — tables.<name> is a description string,
columns are keyed `table.column`, and relationships are an explicit list.
The DDL structure (column types, nullability) lives in extracted_schema.json
generated by `make recon`.
"""

from pathlib import Path

import pytest
import yaml

DATASETS_DIR = Path(__file__).resolve().parent.parent / "api" / "datasets"
DATASETS = [p.name for p in DATASETS_DIR.iterdir() if p.is_dir() and not p.name.startswith("_")]


@pytest.mark.parametrize("dataset", DATASETS)
def test_schema_docs_parses(dataset):
    docs = yaml.safe_load((DATASETS_DIR / dataset / "schema_docs.yaml").read_text())
    assert docs is not None
    assert "tables" in docs
    assert isinstance(docs["tables"], dict) and docs["tables"]
    # New shape: each table value is a description string, not a dict.
    for name, val in docs["tables"].items():
        assert isinstance(val, str), f"tables.{name} should be a string description, got {type(val)}"


@pytest.mark.parametrize("dataset", DATASETS)
def test_metrics_parses(dataset):
    metrics = yaml.safe_load((DATASETS_DIR / dataset / "metrics.yaml").read_text())
    assert metrics is not None
    assert "metrics" in metrics
    assert isinstance(metrics["metrics"], dict) and metrics["metrics"]


def test_financial_columns_with_quotes_survive_parse():
    """Values containing quoted codes (M, F, OWNER, PRIJEM, ...) must parse
    without truncation. Columns are flat `table.col` keys now."""
    docs = yaml.safe_load(
        (DATASETS_DIR / "financial" / "schema_docs.yaml").read_text()
    )
    cols = docs["columns"]
    assert '"M"' in cols["client.gender"] and '"F"' in cols["client.gender"]
    assert '"OWNER"' in cols["disp.type"] and '"DISPONENT"' in cols["disp.type"]
    assert '"PRIJEM"' in cols["trans.type"] and '"VYDAJ"' in cols["trans.type"]


def test_financial_has_relationships():
    docs = yaml.safe_load(
        (DATASETS_DIR / "financial" / "schema_docs.yaml").read_text()
    )
    rels = docs.get("relationships") or []
    assert len(rels) >= 5  # we have 8; assert ≥5 to allow some pruning
    # all entries have from/to
    for r in rels:
        assert "from" in r and "to" in r
        assert "." in r["from"] and "." in r["to"]