scrap/tests/stores/test_cdiscount.py

"""
Tests pour pricewatch.app.stores.cdiscount.store

Vérifie match(), canonicalize(), extract_reference() et parse()
pour le store Cdiscount.
"""

import pytest

from pricewatch.app.stores.cdiscount.store import CdiscountStore
from pricewatch.app.core.schema import DebugStatus, StockStatus


class TestCdiscountMatch:
    """Tests de la méthode match() pour Cdiscount."""

    @pytest.fixture
    def store(self) -> CdiscountStore:
        """Fixture: CdiscountStore instance."""
        return CdiscountStore()

    def test_match_cdiscount_com(self, store):
        """cdiscount.com doit retourner 0.9."""
        score = store.match("https://www.cdiscount.com/informatique/example/f-123-sku.html")
        assert score == 0.9

    def test_match_non_cdiscount(self, store):
        """URL non-Cdiscount doit retourner 0.0."""
        score = store.match("https://www.amazon.fr/dp/B08N5WRWNW")
        assert score == 0.0

    def test_match_empty_url(self, store):
        """URL vide doit retourner 0.0."""
        score = store.match("")
        assert score == 0.0

    def test_match_case_insensitive(self, store):
        """Match doit être insensible à la casse."""
        score = store.match("https://www.CDISCOUNT.COM/product/f-123-sku.html")
        assert score == 0.9


class TestCdiscountCanonicalize:
    """Tests de la méthode canonicalize() pour Cdiscount."""

    @pytest.fixture
    def store(self) -> CdiscountStore:
        """Fixture: CdiscountStore instance."""
        return CdiscountStore()

    def test_canonicalize_with_query_params(self, store):
        """URL avec query params doit être normalisée."""
        url = "https://www.cdiscount.com/informatique/pc/product/f-10709-sku.html?idOffre=123&sw=abc"
        canonical = store.canonicalize(url)
        assert canonical == "https://www.cdiscount.com/informatique/pc/product/f-10709-sku.html"
        assert "?" not in canonical

    def test_canonicalize_already_canonical(self, store):
        """URL déjà canonique ne change pas."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        canonical = store.canonicalize(url)
        assert canonical == url

    def test_canonicalize_with_fragment(self, store):
        """URL avec fragment doit être normalisée."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html#mpos=2"
        canonical = store.canonicalize(url)
        assert canonical == "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        assert "#" not in canonical

    def test_canonicalize_empty_url(self, store):
        """URL vide retourne URL vide."""
        canonical = store.canonicalize("")
        assert canonical == ""


class TestCdiscountExtractReference:
    """Tests de la méthode extract_reference() pour Cdiscount."""

    @pytest.fixture
    def store(self) -> CdiscountStore:
        """Fixture: CdiscountStore instance."""
        return CdiscountStore()

    def test_extract_reference_standard_format(self, store):
        """Extraction du SKU depuis format standard /f-{ID}-{SKU}.html."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-tuf608umrv004.html"
        ref = store.extract_reference(url)
        assert ref == "10709-tuf608umrv004"

    def test_extract_reference_long_url(self, store):
        """Extraction du SKU depuis URL longue avec chemin complet."""
        url = "https://www.cdiscount.com/informatique/ordinateurs-pc-portables/pc-portable-gamer-asus/f-10709-tuf608umrv004.html"
        ref = store.extract_reference(url)
        assert ref == "10709-tuf608umrv004"

    def test_extract_reference_with_query_params(self, store):
        """Extraction du SKU depuis URL avec query params."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku123.html?idOffre=456"
        ref = store.extract_reference(url)
        assert ref == "10709-sku123"

    def test_extract_reference_invalid_url(self, store):
        """URL sans SKU retourne None."""
        url = "https://www.cdiscount.com/informatique/"
        ref = store.extract_reference(url)
        assert ref is None

    def test_extract_reference_empty_url(self, store):
        """URL vide retourne None."""
        ref = store.extract_reference("")
        assert ref is None


class TestCdiscountParse:
    """Tests de la méthode parse() pour Cdiscount."""

    @pytest.fixture
    def store(self) -> CdiscountStore:
        """Fixture: CdiscountStore instance."""
        return CdiscountStore()

    @pytest.fixture
    def minimal_html(self) -> str:
        """Fixture: HTML Cdiscount minimal avec titre et prix."""
        return """
        <html>
        <head><title>Test Product</title></head>
        <body>
            <h1 data-e2e="title">PC Portable Test</h1>
            <div class="sc-83lijy-0 kwssIa SecondaryPrice-price">899,99 €</div>
        </body>
        </html>
        """

    @pytest.fixture
    def complete_html(self) -> str:
        """Fixture: HTML Cdiscount plus complet."""
        return """
        <html>
        <head><title>Test Product</title></head>
        <body>
            <h1 data-e2e="title">PC Portable Gamer ASUS</h1>
            <div class="SecondaryPrice-price">1299,99 €</div>
            <img alt="PC Portable Gamer ASUS" src="https://www.cdiscount.com/pdt2/0/0/4/1/700x700/sku123/image1.jpg" />
            <img alt="PC Portable Gamer ASUS" src="https://www.cdiscount.com/pdt2/0/0/4/2/700x700/sku123/image2.jpg" />
        </body>
        </html>
        """

    @pytest.fixture
    def empty_html(self) -> str:
        """Fixture: HTML vide."""
        return "<html><body></body></html>"

    def test_parse_minimal_html(self, store, minimal_html):
        """Parse un HTML minimal avec titre et prix."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku123.html"
        snapshot = store.parse(minimal_html, url)

        assert snapshot.source == "cdiscount"
        assert snapshot.url == "https://www.cdiscount.com/informatique/pc/f-10709-sku123.html"
        assert snapshot.title == "PC Portable Test"
        assert snapshot.price == 899.99
        assert snapshot.currency == "EUR"
        assert snapshot.is_complete() is True

    def test_parse_complete_html(self, store, complete_html):
        """Parse un HTML plus complet avec images."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-asus123.html"
        snapshot = store.parse(complete_html, url)

        assert snapshot.title == "PC Portable Gamer ASUS"
        assert snapshot.price == 1299.99
        assert snapshot.reference == "10709-asus123"
        assert len(snapshot.images) >= 2
        assert snapshot.is_complete() is True
        assert snapshot.debug.status == DebugStatus.SUCCESS

    def test_parse_empty_html(self, store, empty_html):
        """Parse un HTML vide doit retourner un snapshot partiel."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(empty_html, url)

        assert snapshot.source == "cdiscount"
        assert snapshot.title is None
        assert snapshot.price is None
        assert snapshot.is_complete() is False
        assert snapshot.debug.status == DebugStatus.PARTIAL

    def test_parse_canonicalizes_url(self, store, minimal_html):
        """Parse doit canonicaliser l'URL."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html?idOffre=123#mpos=2"
        snapshot = store.parse(minimal_html, url)

        assert snapshot.url == "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        assert "?" not in snapshot.url
        assert "#" not in snapshot.url

    def test_parse_extracts_reference_from_url(self, store, minimal_html):
        """Parse doit extraire le SKU depuis l'URL."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-tuf608umrv004.html"
        snapshot = store.parse(minimal_html, url)

        assert snapshot.reference == "10709-tuf608umrv004"

    def test_parse_sets_fetched_at(self, store, minimal_html):
        """Parse doit définir fetched_at."""
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(minimal_html, url)

        assert snapshot.fetched_at is not None

    def test_parse_partial_status_without_title(self, store):
        """Parse sans titre doit avoir status PARTIAL."""
        html = """
        <html><body>
            <div class="SecondaryPrice-price">299,99 €</div>
        </body></html>
        """
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(html, url)

        assert snapshot.debug.status == DebugStatus.PARTIAL
        assert snapshot.title is None
        assert snapshot.price == 299.99

    def test_parse_partial_status_without_price(self, store):
        """Parse sans prix doit avoir status PARTIAL."""
        html = """
        <html><body>
            <h1 data-e2e="title">Test Product</h1>
        </body></html>
        """
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(html, url)

        assert snapshot.debug.status == DebugStatus.PARTIAL
        assert snapshot.title == "Test Product"
        assert snapshot.price is None

    def test_parse_price_with_comma_separator(self, store):
        """Parse doit gérer les prix avec virgule (format français)."""
        html = """
        <html><body>
            <h1 data-e2e="title">Test</h1>
            <div class="price">1199,99 €</div>
        </body></html>
        """
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(html, url)

        assert snapshot.price == 1199.99

    def test_parse_price_with_dot_separator(self, store):
        """Parse doit gérer les prix avec point (format international)."""
        html = """
        <html><body>
            <h1 data-e2e="title">Test</h1>
            <div class="price">1199.99 €</div>
        </body></html>
        """
        url = "https://www.cdiscount.com/informatique/pc/f-10709-sku.html"
        snapshot = store.parse(html, url)

        assert snapshot.price == 1199.99


class TestCdiscountStoreInit:
    """Tests de l'initialisation du store Cdiscount."""

    def test_store_id(self):
        """Le store_id doit être 'cdiscount'."""
        store = CdiscountStore()
        assert store.store_id == "cdiscount"

    def test_selectors_loaded(self):
        """Les sélecteurs doivent être chargés depuis selectors.yml."""
        store = CdiscountStore()
        assert isinstance(store.selectors, dict)
        assert len(store.selectors) > 0

    def test_repr(self):
        """Test de la représentation string."""
        store = CdiscountStore()
        repr_str = repr(store)
        assert "CdiscountStore" in repr_str
        assert "cdiscount" in repr_str