from __future__ import annotations

from conftest import fixture_text

from companyatlas.sdk import normalize as N


def test_noise_normalisation_replaces_dates_times_counters_tokens() -> None:
    s = "Updated 3 minutes ago on 2024-05-01 at 10:34 PM · March 5, 2024 · © 2024 Acme · 1,234 views · token=abcdef0123456789abcdef0123456789 ?v=123&x=1"
    n = N.normalized_text(s)
    assert "<rel>" in n and "<date>" in n and "<time>" in n and "<count>" in n and "<copyright>" in n and "<hex>" in n and "?v=<q>" in n
    assert "2024" not in n.replace("<date>", "")


def test_text_hash_ignores_noise_but_not_prices() -> None:
    assert N.text_hash("Price $29 per month, updated 2 days ago") == N.text_hash("Price $29 per month, updated 5 hours ago")
    assert N.text_hash("Price $29 per month") != N.text_hash("Price $39 per month")


def test_simhash_near_duplicates_and_hamming() -> None:
    a = N.simhash("The quick brown fox jumps over the lazy dog near the river bank today")
    b = N.simhash("The quick brown fox jumps over the lazy dog near the river bank tonight")
    c = N.simhash("Completely different content about pricing plans and enterprise contracts")
    assert N.hamming(a, b) < N.hamming(a, c)
    assert N.simhash("") == 0


def test_parse_pricing_page_blocks_meta_links() -> None:
    page = N.parse(fixture_text("generic_html", "pricing.html"), url="https://www.acme-cloud.example/pricing")
    assert page.title == "Pricing — Acme Cloud"
    assert page.lang == "en"
    assert page.meta["description"].startswith("Simple")
    assert page.meta["canonical"] == "https://www.acme-cloud.example/pricing"
    assert page.feeds == ["https://www.acme-cloud.example/blog/feed.xml"]
    kinds = [b.kind for b in page.blocks]
    assert kinds.count("pricing_plan") == 3
    assert "hero" in kinds and "nav" in kinds and "footer" in kinds and "faq" in kinds
    # script/style/cookie banner/hidden content dropped
    assert "dataLayer" not in page.text and "We use cookies" not in page.text and "Last updated" not in page.text
    # nav/footer are low weight, pricing plans high
    assert max(b.weight for b in page.blocks if b.kind == "pricing_plan") > max(b.weight for b in page.blocks if b.kind in ("nav", "footer"))
    regions = {ln.region for ln in page.links}
    assert {"nav", "footer", "main"} <= regions
    assert page.main_selector == "main"


def test_block_keys_are_stable_under_reordering() -> None:
    html = fixture_text("generic_html", "pricing.html")
    a = N.parse(html, url="https://x.example/pricing")
    # swap the Starter and Pro cards
    swapped = html.replace('<h3>Starter</h3>', '<h3>TMP</h3>').replace('<h3>Pro</h3>', '<h3>Starter</h3>').replace('<h3>TMP</h3>', '<h3>Pro</h3>')
    b = N.parse(swapped, url="https://x.example/pricing")
    keys_a = {blk.key for blk in a.blocks if blk.kind == "pricing_plan"}
    keys_b = {blk.key for blk in b.blocks if blk.kind == "pricing_plan"}
    # names moved but content per card changed, so at least the Enterprise card key is identical
    assert keys_a & keys_b
    assert N.structural_hash(a.blocks) != N.structural_hash(b.blocks) or [x.key for x in a.blocks] == [x.key for x in b.blocks]


def test_jsonld_and_microdata_extraction() -> None:
    page = N.parse(fixture_text("generic_html", "leadership.html"), url="https://www.acme-cloud.example/about/leadership")
    assert "organizations" in page.jsonld
    assert page.meta["same_as"][0].startswith("https://www.linkedin.com/")
    assert page.jsonld["persons"][0]["name"] == "Jane Doe"
    micro = N.extract_microdata(N.LexborHTMLParser('<div itemscope itemtype="https://schema.org/Person"><span itemprop="name">Ada Lovelace</span><span itemprop="jobTitle">CTO</span></div>'))
    assert micro["persons"][0]["name"] == "Ada Lovelace"


def test_language_guess() -> None:
    assert N.language_guess("whatever", "fr-CA") == "fr"
    assert N.language_guess("Nous sommes une entreprise qui offre des services pour les clients et les partenaires dans le monde") == "fr"
    assert N.language_guess("We are a company that offers services for customers and partners in the world with our team") == "en"
    assert N.language_guess("これは日本語のテキストです") == "ja"


def test_card_containers_are_split_into_children() -> None:
    page = N.parse(fixture_text("generic_html", "newsroom.html"), url="https://www.acme-cloud.example/news")
    assert [b.kind for b in page.blocks].count("news_item") == 4
    home = N.parse(fixture_text("generic_html", "homepage.html"), url="https://www.acme-cloud.example/")
    assert [b.kind for b in home.blocks].count("product_card") == 3