tests/unit/test_platform_presence.py

"""Unit tests for Platform Presence Scanner."""

from unittest.mock import AsyncMock, MagicMock, patch

import pytest

from app.services.platform_presence import (
    SCORED_PLATFORMS,
    CatalogContext,
    PlatformListing,
    build_recommendations,
    compute_score,
    confidence_to_status,
    parse_goodreads_book_links,
    platform_presence_from_dict,
    platform_presence_to_dict,
    scan_platform_presence,
    search_title,
    title_match_score,
)
from app.repositories.platform_listing_repo import PlatformListingRepository


class TestTitleMatchScore:
    def test_exact_title_match(self):
        score = title_match_score("Regime Change", "Regime Change", "Patrick Deneen", "Patrick Deneen")
        assert score >= 0.85

    def test_series_suffix_stripped_for_match(self):
        score = title_match_score(
            "Her Deadly Homecoming (Carolina McKay, #1)",
            "Her Deadly Homecoming",
        )
        assert score >= 0.85

    def test_partial_title_no_match(self):
        score = title_match_score("Unique Obscure Title XYZ", "Completely Different Book")
        assert score < 0.5

    def test_empty_title_returns_zero(self):
        assert title_match_score("", "Some Title") == 0.0


class TestSearchTitle:
    def test_strips_series_parenthetical(self):
        assert search_title("Her Deadly Homecoming (Carolina McKay, #1)") == "Her Deadly Homecoming"

    def test_strips_bracket_series(self):
        assert search_title("Dark Matter [The Trilogy #2]") == "Dark Matter"

    def test_strips_a_novel_suffix(self):
        assert search_title("Winter Counts (A Novel)") == "Winter Counts"

    def test_plain_title_unchanged(self):
        assert search_title("Regime Change") == "Regime Change"

    def test_strips_amazon_marketing_subtitle(self):
        long_title = (
            "The Girl Behind the Gates: The gripping, heartbreaking historical "
            "bestseller based on a true story"
        )
        assert search_title(long_title) == "The Girl Behind the Gates"

    def test_search_query_uses_core_title_without_quotes(self):
        from app.services.platform_presence import _search_query
        q = _search_query("Her Deadly Homecoming (Carolina McKay Crime Thriller Book 1)", "Jane Doe", None)
        assert q == "Her Deadly Homecoming Jane Doe"
        assert '"' not in q


class TestGoodreadsBookLinks:
    SAMPLE = (
        '{"__typename":"BookLink","name":"Barnes \\u0026 Noble",'
        '"url":"https://www.barnesandnoble.com/s/Her+Deadly+Homecoming","ref":null},'
        '{"__typename":"BookLink","name":"Kobo","url":"https://www.kobo.com/us/en/search?query=x","ref":null},'
        '{"__typename":"BookLink","name":"Thriftbooks","url":"https://www.thriftbooks.com/w/x","ref":null}'
    )

    def test_parses_retailer_links(self):
        links = parse_goodreads_book_links(self.SAMPLE)
        assert "barnes_noble" in links
        assert "kobo" in links
        assert "thriftbooks" in links
        assert "barnesandnoble.com" in links["barnes_noble"]

    def test_empty_html_returns_empty(self):
        assert parse_goodreads_book_links("") == {}


class TestNormalizeRetailerUrl:
    def test_unwraps_synergy_redirect(self):
        from app.services.isbn_catalog import normalize_retailer_url
        raw = (
            "https://click.linksynergy.com/fs-bin/click?id=x"
            "&RD_PARM1=https%3A%2F%2Fwww.kobo.com%2Fus%2Fen%2Fsearch%3FQuery%3Dtest"
        )
        assert "kobo.com" in normalize_retailer_url(raw)

    def test_unwraps_partnerize_destination(self):
        from app.services.isbn_catalog import normalize_retailer_url
        raw = (
            "https://prf.hn/click/camref:1101ljNE7/pubref:x/"
            "destination:https://www.thriftbooks.com/browse/?b.search=Test"
        )
        assert normalize_retailer_url(raw) == "https://www.thriftbooks.com/browse/?b.search=Test"

    def test_bookshop_from_isbn(self):
        from app.services.isbn_catalog import bookshop_url_from_isbn
        url = bookshop_url_from_isbn("9780306406157", "Test")
        assert url == "https://bookshop.org/search?keywords=9780306406157"


class TestRetailerUrlClassification:
    def test_amazon_product_url(self):
        from app.services.isbn_catalog import classify_retailer_url, sanitize_platform_url
        url = "http://www.amazon.com/gp/product/B0DHWS3P6T/ref=affiliate"
        clean = sanitize_platform_url(url, "amazon")
        assert clean == "https://www.amazon.com/dp/B0DHWS3P6T"
        assert classify_retailer_url(clean, "amazon") == "product"

    def test_search_url_classified(self):
        from app.services.isbn_catalog import classify_retailer_url
        url = "https://www.barnesandnoble.com/s/Lethal+Lines"
        assert classify_retailer_url(url, "barnes_noble") == "search"

    def test_goodreads_product_url(self):
        from app.services.isbn_catalog import classify_retailer_url
        url = "https://www.goodreads.com/book/show/219584318-lethal-lines"
        assert classify_retailer_url(url, "goodreads") == "product"

    def test_broken_url_invalid(self):
        from app.services.isbn_catalog import sanitize_platform_url
        assert sanitize_platform_url("https:", "thriftbooks") is None


class TestScoring:
    def _listing(self, pid: str, status: str, weight: float) -> PlatformListing:
        return PlatformListing(
            platform_id=pid,
            display_name=pid,
            status=status,
            confidence=0.9 if status == "verified" else 0.75,
            listing_url="https://example.com",
            benefit="test",
            weight=weight,
        )

    def test_compute_score_counts_verified_platforms(self):
        platforms = [
            self._listing("amazon", "verified", 2.0),
            self._listing("kobo", "likely", 1.0),
            self._listing("goodreads", "verified", 1.0),
            self._listing("walmart", "not_found", 0.8),
        ]
        assert compute_score(platforms) == 2.0

    def test_confidence_to_status_binary(self):
        assert confidence_to_status(0.90) == "verified"
        assert confidence_to_status(0.75) == "not_found"
        assert confidence_to_status(0.50) == "not_found"
        assert confidence_to_status(0.80, isbn_match=True) == "verified"

    def test_recommendations_sorted_by_weight(self):
        platforms = [
            self._listing("amazon", "not_found", 2.0),
            self._listing("abebooks", "not_found", 0.5),
            self._listing("apple_books", "verified", 1.2),
        ]
        recs = build_recommendations(platforms)
        assert recs[0]["platform_id"] == "amazon"
        assert recs[1]["platform_id"] == "abebooks"


class TestSerialization:
    def test_round_trip_dict(self):
        listing = PlatformListing(
            platform_id="amazon",
            display_name="Amazon",
            status="verified",
            confidence=1.0,
            listing_url="https://amazon.com/dp/123",
            benefit="Largest channel",
            weight=2.0,
            icon="🛍",
        )
        from app.services.platform_presence import PlatformPresenceResult
        result = PlatformPresenceResult(
            score=2.0,
            listed_count=1,
            scanned_at="2026-01-01T00:00:00+00:00",
            source_isbn="9781234567890",
            platforms=[listing],
            recommendations=[],
        )
        restored = platform_presence_from_dict(platform_presence_to_dict(result))
        assert restored.score == 2.0
        assert restored.platforms[0].platform_id == "amazon"


@pytest.mark.asyncio
async def test_scan_source_platform_marked_verified():
    """Source URL platform should be verified without HTTP calls."""
    from app.services.catalog_hub import CatalogSnapshot

    snap = CatalogSnapshot(title="Test Book", author="Jane Author", isbn13="9781234567890")
    with patch(
        "app.services.platform_presence._load_goodreads_hub",
        new=AsyncMock(return_value=("Jane Author", None, {})),
    ), patch(
        "app.services.catalog_hub.build_catalog_snapshot",
        new=AsyncMock(return_value=snap),
    ), patch(
        "app.services.platform_presence._enrich_listing_prices",
        new=AsyncMock(),
    ):
        scan = await scan_platform_presence(
            title="Test Book",
            author="Jane Author",
            isbn="9781234567890",
            source_platform="amazon",
            source_url="https://www.amazon.com/dp/B01234567",
            buy_url="https://www.amazon.com/dp/B01234567",
        )
    amazon = next(p for p in scan.platforms if p.platform_id == "amazon")
    assert amazon.status == "verified"
    assert amazon.confidence == 1.0
    assert amazon.listing_url == "https://www.amazon.com/dp/B01234567"
    assert scan.listed_count >= 1


@pytest.mark.asyncio
async def test_scan_uses_goodreads_catalog_links():
    """Goodreads product buy links verify retailers without HTML scraping."""
    from app.services.catalog_hub import CatalogSnapshot

    gr_links = {
        "goodreads": "https://www.goodreads.com/book/show/55046753",
        "thriftbooks": "https://www.thriftbooks.com/w/x",
    }
    snap = CatalogSnapshot(
        title="Her Deadly Homecoming",
        author="Tony Urban",
        goodreads_url=gr_links["goodreads"],
        goodreads_confidence=0.95,
        retailer_links=gr_links,
    )
    with patch(
        "app.services.platform_presence._load_goodreads_hub",
        new=AsyncMock(return_value=("Tony Urban", gr_links["goodreads"], gr_links)),
    ), patch(
        "app.services.catalog_hub.build_catalog_snapshot",
        new=AsyncMock(return_value=snap),
    ):
        scan = await scan_platform_presence(
            title="Her Deadly Homecoming (Carolina McKay, #1)",
            author="",
        )
    verified = {p.platform_id for p in scan.platforms if p.status == "verified"}
    assert {"goodreads", "thriftbooks"}.issubset(verified)


@pytest.mark.asyncio
async def test_build_catalog_uses_slug_fetch_and_bookshop_isbn():
    """Catalog should prefer slug page fetch and add bookshop when ISBN known."""
    from app.services.platform_presence import _build_catalog_context

    html = (
        '{"__typename":"BookLink","name":"Amazon","url":"https://www.amazon.com/dp/B08GFXGTJZ","ref":null}'
    )
    mock_client = MagicMock()
    with patch(
        "app.services.platform_presence._goodreads_autocomplete_best",
        new=AsyncMock(return_value={
            "book_id": "55046753",
            "book_path": "/book/show/55046753-her-deadly-homecoming",
            "author_name": "Tony Urban",
        }),
    ), patch(
        "app.services.platform_presence._fetch_goodreads_book_page",
        new=AsyncMock(return_value=html),
    ) as fetch_mock:
        ctx = await _build_catalog_context(
            mock_client,
            "Her Deadly Homecoming",
            "",
            isbn="9780306406157",
        )
    fetch_mock.assert_called_once_with(
        mock_client, "55046753", "/book/show/55046753-her-deadly-homecoming",
    )
    assert ctx.resolved_author == "Tony Urban"
    assert "amazon" in ctx.retailer_links
    assert "bookshop" in ctx.retailer_links
    assert "9780306406157" in ctx.retailer_links["bookshop"]


@pytest.mark.asyncio
async def test_scan_cache_key_busted_for_v10():
    from app.services.platform_presence import scan_cache_key
    assert scan_cache_key("https://example.com/book").startswith("platform_scan:v10:")


@pytest.mark.asyncio
async def test_scan_empty_title_skips():
    scan = await scan_platform_presence(title="", author="")
    assert scan.score == 0.0
    assert all(p.status == "skipped" for p in scan.platforms)


@pytest.mark.asyncio
async def test_scan_mocks_catalog_hub():
    """Scan uses catalog hub APIs — mock snapshot for deterministic output."""
    from app.services.catalog_hub import CatalogSnapshot

    snap = CatalogSnapshot(
        title="Test Book",
        author="Jane Author",
        isbn13="9780306406157",
        gb_volume_id="vol1",
        gb_confidence=0.92,
        ol_confidence=0.90,
        ol_key="/works/OL1W",
        goodreads_url="https://www.goodreads.com/book/show/1",
        goodreads_confidence=0.95,
    )
    with patch(
        "app.services.platform_presence._load_goodreads_hub",
        new=AsyncMock(return_value=("Jane Author", snap.goodreads_url, {})),
    ), patch(
        "app.services.catalog_hub.build_catalog_snapshot",
        new=AsyncMock(return_value=snap),
    ), patch(
        "app.services.platform_presence._enrich_listing_prices",
        new=AsyncMock(),
    ):
        scan = await scan_platform_presence(
            title="Test Book",
            author="Jane Author",
            isbn="9780306406157",
        )
    assert len(SCORED_PLATFORMS) == 10
    verified_ids = {p.platform_id for p in scan.platforms if p.status == "verified"}
    assert "google_books" in verified_ids
    assert "goodreads" in verified_ids
    assert "open_library" in verified_ids


def test_replace_from_scan_defaults_visible_in_widget_false():
    """Repo must never expose listings to widget by default."""
    repo = PlatformListingRepository.__new__(PlatformListingRepository)
    from app.services.platform_presence import PlatformPresenceResult
    scan = PlatformPresenceResult(
        score=2.0,
        listed_count=1,
        scanned_at="2026-01-01T00:00:00+00:00",
        source_isbn=None,
        platforms=[
            PlatformListing(
                platform_id="amazon",
                display_name="Amazon",
                status="verified",
                confidence=1.0,
                listing_url="https://amazon.com/x",
                benefit="",
                weight=2.0,
            ),
        ],
        recommendations=[],
    )
    # Verify create payload would set visible_in_widget=False
    for p in scan.platforms:
        assert p.listing_url  # saved internally


def test_widget_helpers_ignore_platform_listings():
    """get_book_links must not import platform listing repo."""
    import inspect
    from app.services.pipeline import helpers
    source = inspect.getsource(helpers.get_book_links)
    assert "platform_listing" not in source.lower()
    assert "BookPlatformListing" not in source


def test_infer_distribution_note_kindle_only():
    from app.services.platform_presence import infer_distribution_note

    platforms = [
        PlatformListing(
            platform_id="amazon", display_name="Amazon", status="verified",
            confidence=0.98, listing_url="https://amazon.com/dp/B0X", benefit="", weight=2.0,
        ),
        PlatformListing(
            platform_id="goodreads", display_name="Goodreads", status="verified",
            confidence=0.95, listing_url="https://goodreads.com/book/show/1", benefit="", weight=1.0,
        ),
        PlatformListing(
            platform_id="kobo", display_name="Kobo", status="not_found",
            confidence=0.0, listing_url=None, benefit="", weight=1.0,
        ),
    ]
    note = infer_distribution_note(platforms)
    assert note is not None
    assert "Kindle" in note


def test_infer_distribution_note_suppressed_with_isbn():
    from app.services.platform_presence import infer_distribution_note

    platforms = [
        PlatformListing(
            platform_id="amazon", display_name="Amazon", status="verified",
            confidence=0.98, listing_url="https://amazon.com/dp/B0X", benefit="", weight=2.0,
        ),
    ]
    assert infer_distribution_note(platforms, source_isbn="9781529374544") is None


def test_parse_goodreads_book_ref():
    from app.services.platform_presence import parse_goodreads_book_ref

    ref = parse_goodreads_book_ref(
        "https://www.goodreads.com/book/show/219584318-lethal-lines?q=1"
    )
    assert ref == ("219584318", "/book/show/219584318-lethal-lines")


@pytest.mark.integration
@pytest.mark.asyncio
async def test_lethal_lines_kindle_only_scan():
    """Lethal Lines is Amazon Kindle + Goodreads only — not wide retail."""
    scan = await scan_platform_presence(
        title="Lethal Lines",
        author="Paige Black",
        source_url="https://www.goodreads.com/book/show/219584318-lethal-lines",
        source_platform="goodreads",
    )
    verified = [p for p in scan.platforms if p.status == "verified"]
    verified_ids = {p.platform_id for p in verified}
    assert verified_ids == {"amazon", "goodreads"}
    assert scan.listed_count == 2
    assert scan.distribution_note
    amazon = next(p for p in verified if p.platform_id == "amazon")
    assert "B0DHWS3P6T" in (amazon.listing_url or "")


@pytest.mark.integration
@pytest.mark.asyncio
async def test_girl_behind_gates_wide_distribution_scan():
    """Trade-published title should resolve Goodreads + multi-retailer coverage."""
    scan = await scan_platform_presence(
        title=(
            "The Girl Behind the Gates: The gripping, heartbreaking historical "
            "bestseller based on a true story"
        ),
        author="Brenda Davies",
        isbn="9781529374551",
        source_platform="amazon",
        source_url="https://www.amazon.com/dp/B0847KDV8W",
        buy_url="https://www.amazon.com/dp/B0847KDV8W",
    )
    verified = {p.platform_id for p in scan.platforms if p.status == "verified"}
    assert "amazon" in verified
    assert "goodreads" in verified
    assert scan.listed_count >= 5
    assert scan.distribution_note is None