FabledCurator/tests/test_gallery_facets.py

"""Phase-2 faceted refine: new composable filter params + the facets()
aggregate (platform / curation-flags / date-range counts, scoped live to the
current filter with per-group minus-self semantics)."""

from datetime import UTC, datetime

import pytest

from backend.app.models import (
    Artist,
    ImageProvenance,
    ImageRecord,
    Post,
    Source,
    Tag,
    TagKind,
)
from backend.app.models.tag import image_tag
from backend.app.services.gallery_service import (
    UNSOURCED_PLATFORM,
    GalleryService,
)

pytestmark = pytest.mark.integration

_BASE = datetime(2026, 1, 1, 12, 0, tzinfo=UTC)


async def _artist(db, name):
    a = Artist(name=name, slug=name.lower())
    db.add(a)
    await db.flush()
    return a


async def _sourced_image(db, n, platform, artist, eff):
    """A downloaded image with a Source(platform) + Post + provenance row."""
    s = Source(
        artist_id=artist.id, platform=platform,
        url=f"https://{platform}.test/{artist.slug}/{n}",
    )
    db.add(s)
    await db.flush()
    p = Post(source_id=s.id, artist_id=artist.id, external_post_id=f"{platform}-{n}")
    db.add(p)
    await db.flush()
    rec = ImageRecord(
        path=f"/images/s/{platform}-{n}.jpg", sha256=f"a{n:063d}",
        size_bytes=1, mime="image/jpeg", width=1, height=1,
        origin="downloaded", integrity_status="ok",
        artist_id=artist.id, primary_post_id=p.id,
    )
    rec.created_at = eff
    rec.effective_date = eff
    db.add(rec)
    await db.flush()
    db.add(ImageProvenance(image_record_id=rec.id, post_id=p.id, source_id=s.id))
    await db.flush()
    return rec


async def _fs_image(db, n, eff, artist=None):
    """A filesystem-imported image: no provenance/source (unsourced),
    artist optional."""
    rec = ImageRecord(
        path=f"/images/fs/{n}.jpg", sha256=f"b{n:063d}",
        size_bytes=1, mime="image/jpeg", width=1, height=1,
        origin="imported_filesystem", integrity_status="unknown",
        artist_id=artist.id if artist else None,
    )
    rec.created_at = eff
    rec.effective_date = eff
    db.add(rec)
    await db.flush()
    return rec


async def _tag(db, image, name):
    t = Tag(name=name, kind=TagKind.general)
    db.add(t)
    await db.flush()
    await db.execute(image_tag.insert().values(
        image_record_id=image.id, tag_id=t.id, source="manual"))
    await db.flush()
    return t


# --- facets() aggregate ----------------------------------------------------

@pytest.mark.asyncio
async def test_facets_platform_counts(db):
    ar = await _artist(db, "Ar")
    await _sourced_image(db, 1, "patreon", ar, _BASE)
    await _sourced_image(db, 2, "patreon", ar, _BASE)
    await _sourced_image(db, 3, "pixiv", ar, _BASE)
    await _fs_image(db, 4, _BASE)  # unsourced → null bucket
    svc = GalleryService(db)
    f = await svc.facets()
    counts = {p["value"]: p["count"] for p in f.platforms}
    assert counts["patreon"] == 2
    assert counts["pixiv"] == 1
    assert counts[None] == 1
    assert f.total == 4


@pytest.mark.asyncio
async def test_facets_counts_image_under_each_platform(db):
    """A single image cross-posted to two platforms counts under both
    (DISTINCT image per platform), but is one image in total."""
    ar = await _artist(db, "Ar")
    rec = await _sourced_image(db, 1, "patreon", ar, _BASE)
    s2 = Source(artist_id=ar.id, platform="pixiv", url="https://pixiv.test/ar/x")
    db.add(s2)
    await db.flush()
    p2 = Post(source_id=s2.id, artist_id=ar.id, external_post_id="pixiv-x")
    db.add(p2)
    await db.flush()
    db.add(ImageProvenance(image_record_id=rec.id, post_id=p2.id, source_id=s2.id))
    await db.flush()
    svc = GalleryService(db)
    f = await svc.facets()
    counts = {p["value"]: p["count"] for p in f.platforms}
    assert counts["patreon"] == 1
    assert counts["pixiv"] == 1
    assert f.total == 1


@pytest.mark.asyncio
async def test_facets_curation_flags(db):
    ar = await _artist(db, "Ar")
    tagged = await _fs_image(db, 1, _BASE, artist=ar)
    await _tag(db, tagged, "x")
    await _fs_image(db, 2, _BASE, artist=ar)    # untagged, has artist
    await _fs_image(db, 3, _BASE, artist=None)  # untagged, no artist
    svc = GalleryService(db)
    f = await svc.facets()
    assert f.untagged == 2
    assert f.no_artist == 1


@pytest.mark.asyncio
async def test_facets_date_bounds(db):
    ar = await _artist(db, "Ar")
    await _fs_image(db, 1, datetime(2020, 1, 1, tzinfo=UTC), artist=ar)
    await _fs_image(db, 2, datetime(2026, 6, 1, tzinfo=UTC), artist=ar)
    svc = GalleryService(db)
    f = await svc.facets()
    assert f.date_min == datetime(2020, 1, 1, tzinfo=UTC)
    assert f.date_max == datetime(2026, 6, 1, tzinfo=UTC)


@pytest.mark.asyncio
async def test_facets_platform_minus_self(db):
    """The platform group ignores its OWN active selection so siblings stay
    visible/switchable; total still honors the platform filter."""
    ar = await _artist(db, "Ar")
    await _sourced_image(db, 1, "patreon", ar, _BASE)
    await _sourced_image(db, 2, "pixiv", ar, _BASE)
    svc = GalleryService(db)
    f = await svc.facets(platform="patreon")
    counts = {p["value"]: p["count"] for p in f.platforms}
    assert counts.get("pixiv") == 1
    assert counts.get("patreon") == 1
    assert f.total == 1


@pytest.mark.asyncio
async def test_facets_other_filter_scopes_platforms(db):
    """A non-platform filter (media) narrows the platform counts."""
    ar = await _artist(db, "Ar")
    await _sourced_image(db, 1, "patreon", ar, _BASE)
    vid = await _sourced_image(db, 2, "patreon", ar, _BASE)
    vid.mime = "video/mp4"
    await db.flush()
    svc = GalleryService(db)
    f = await svc.facets(media_type="image")
    counts = {p["value"]: p["count"] for p in f.platforms}
    assert counts["patreon"] == 1


# --- new scroll filter params ----------------------------------------------

@pytest.mark.asyncio
async def test_scroll_platform_filter(db):
    ar = await _artist(db, "Ar")
    pat = await _sourced_image(db, 1, "patreon", ar, _BASE)
    await _sourced_image(db, 2, "pixiv", ar, _BASE)
    svc = GalleryService(db)
    page = await svc.scroll(cursor=None, limit=10, platform="patreon")
    assert [i.id for i in page.images] == [pat.id]


@pytest.mark.asyncio
async def test_scroll_platform_unsourced(db):
    ar = await _artist(db, "Ar")
    await _sourced_image(db, 1, "patreon", ar, _BASE)
    fs = await _fs_image(db, 2, _BASE)
    svc = GalleryService(db)
    page = await svc.scroll(cursor=None, limit=10, platform=UNSOURCED_PLATFORM)
    assert [i.id for i in page.images] == [fs.id]


@pytest.mark.asyncio
async def test_scroll_untagged_filter(db):
    ar = await _artist(db, "Ar")
    tagged = await _fs_image(db, 1, _BASE, artist=ar)
    await _tag(db, tagged, "x")
    untag = await _fs_image(db, 2, _BASE, artist=ar)
    svc = GalleryService(db)
    page = await svc.scroll(cursor=None, limit=10, untagged=True)
    assert [i.id for i in page.images] == [untag.id]


@pytest.mark.asyncio
async def test_scroll_no_artist_filter(db):
    ar = await _artist(db, "Ar")
    await _fs_image(db, 1, _BASE, artist=ar)
    none = await _fs_image(db, 2, _BASE, artist=None)
    svc = GalleryService(db)
    page = await svc.scroll(cursor=None, limit=10, no_artist=True)
    assert [i.id for i in page.images] == [none.id]


@pytest.mark.asyncio
async def test_scroll_date_range_filter(db):
    ar = await _artist(db, "Ar")
    old = await _fs_image(db, 1, datetime(2020, 1, 1, tzinfo=UTC), artist=ar)
    new = await _fs_image(db, 2, datetime(2026, 6, 1, tzinfo=UTC), artist=ar)
    svc = GalleryService(db)
    page = await svc.scroll(
        cursor=None, limit=10, date_from=datetime(2026, 1, 1, tzinfo=UTC))
    assert [i.id for i in page.images] == [new.id]
    page2 = await svc.scroll(
        cursor=None, limit=10, date_to=datetime(2021, 1, 1, tzinfo=UTC))
    assert [i.id for i in page2.images] == [old.id]


@pytest.mark.asyncio
async def test_scroll_combines_new_filters_and(db):
    """platform + untagged + no_artist compose with AND."""
    ar = await _artist(db, "Ar")
    # match: unsourced, untagged, no artist
    match = await _fs_image(db, 1, _BASE, artist=None)
    # decoys violating one clause each
    await _fs_image(db, 2, _BASE, artist=ar)            # has artist
    tagged = await _fs_image(db, 3, _BASE, artist=None)
    await _tag(db, tagged, "x")                          # tagged
    svc = GalleryService(db)
    page = await svc.scroll(
        cursor=None, limit=10,
        platform=UNSOURCED_PLATFORM, untagged=True, no_artist=True)
    assert [i.id for i in page.images] == [match.id]