"""Tests for content_loader.py: schema validation (fail loud on a malformed
content/ file) and that the real glossary.yaml/pages.yaml load correctly.
Ported from backend/tests/web/test_content_loader.py (repo-split Stage 4) --
content_loader.py itself moved here in Stage 3."""
import pytest

import content_loader


def _write(tmp_path, name, text):
    path = tmp_path / name
    path.write_text(text, encoding="utf-8")
    return str(path)


# --- load_glossary: schema validation --------------------------------------

def test_load_glossary_rejects_non_mapping_top_level(tmp_path):
    path = _write(tmp_path, "g.yaml", "- just\n- a\n- list\n")
    with pytest.raises(ValueError, match="top-level mapping"):
        content_loader.load_glossary(path)


def test_load_glossary_rejects_missing_terms_key(tmp_path):
    path = _write(tmp_path, "g.yaml", "not_terms: []\n")
    with pytest.raises(ValueError, match="non-empty list"):
        content_loader.load_glossary(path)


def test_load_glossary_rejects_empty_terms(tmp_path):
    path = _write(tmp_path, "g.yaml", "terms: []\n")
    with pytest.raises(ValueError, match="non-empty list"):
        content_loader.load_glossary(path)


def test_load_glossary_rejects_missing_required_field(tmp_path):
    path = _write(tmp_path, "g.yaml", """
terms:
  - slug: foo
    term: Foo
    short: A short definition.
""")
    with pytest.raises(ValueError, match="foo.*body"):
        content_loader.load_glossary(path)


def test_load_glossary_rejects_blank_required_field(tmp_path):
    path = _write(tmp_path, "g.yaml", """
terms:
  - slug: foo
    term: Foo
    short: ""
    body: Full body.
""")
    with pytest.raises(ValueError, match="foo.*short"):
        content_loader.load_glossary(path)


def test_load_glossary_rejects_duplicate_slug(tmp_path):
    path = _write(tmp_path, "g.yaml", """
terms:
  - slug: foo
    term: Foo
    short: s1
    body: b1
  - slug: foo
    term: Foo Again
    short: s2
    body: b2
""")
    with pytest.raises(ValueError, match="duplicate slug"):
        content_loader.load_glossary(path)


def test_load_glossary_accepts_well_formed_content(tmp_path):
    path = _write(tmp_path, "g.yaml", """
terms:
  - slug: foo
    term: Foo
    short: A short definition.
    body: A full body with <b>markup</b>.
  - slug: bar
    term: Bar
    short: Another short definition.
    body: Another full body.
""")
    glossary = content_loader.load_glossary(path)
    assert list(glossary) == ["foo", "bar"]   # file order preserved
    assert glossary["foo"] == {
        "term": "Foo", "short": "A short definition.", "body": "A full body with <b>markup</b>.",
    }


# --- load_pages: schema validation ------------------------------------------

def test_load_pages_rejects_missing_pages_key(tmp_path):
    path = _write(tmp_path, "p.yaml", "not_pages: {}\n")
    with pytest.raises(ValueError, match="non-empty mapping"):
        content_loader.load_pages(path)


def test_load_pages_rejects_missing_required_field(tmp_path):
    path = _write(tmp_path, "p.yaml", """
pages:
  about:
    title: About
""")
    with pytest.raises(ValueError, match="about.*description"):
        content_loader.load_pages(path)


def test_load_pages_accepts_well_formed_content(tmp_path):
    path = _write(tmp_path, "p.yaml", """
pages:
  about:
    title: About Us
    description: A description.
""")
    pages = content_loader.load_pages(path)
    assert pages == {"about": {"title": "About Us", "description": "A description."}}


# --- the real content/ files ------------------------------------------------

def test_real_glossary_loads_and_matches_the_rendered_site():
    """The module-level GLOSSARY content.py loads at import IS content_loader's
    output — this just re-asserts the real file is well-formed and non-trivial,
    since a broken content/glossary.yaml would already have failed every other
    test at collection time (content.py loads it at import)."""
    glossary = content_loader.load_glossary()
    assert len(glossary) >= 8
    assert "percentile" in glossary
    for slug, entry in glossary.items():
        assert entry["term"] and entry["short"] and entry["body"], slug


def test_real_pages_loads_the_four_static_pages():
    pages = content_loader.load_pages()
    assert set(pages) == {"about", "privacy", "hub", "glossary_index"}
    for key, entry in pages.items():
        assert entry["title"] and entry["description"], key