FabledCurator/tests/test_provenance_service.py

from datetime import UTC, datetime

import pytest

from backend.app.models import (
    Artist,
    ImageProvenance,
    ImageRecord,
    Post,
    PostAttachment,
    Source,
)
from backend.app.services.provenance_service import ProvenanceService

pytestmark = pytest.mark.integration


async def _seed_image(db, sha="a" + "0" * 63) -> ImageRecord:
    rec = ImageRecord(
        path=f"/images/test/{sha}.jpg",
        sha256=sha,
        size_bytes=1, mime="image/jpeg", width=1, height=1,
        origin="imported_filesystem", integrity_status="unknown",
    )
    db.add(rec)
    await db.flush()
    return rec


async def _seed_post(db, *, artist_name, slug, platform, ext_id,
                     title=None, desc=None, count=None) -> tuple:
    artist = Artist(name=artist_name, slug=slug)
    db.add(artist)
    await db.flush()
    source = Source(artist_id=artist.id, platform=platform,
                    url=f"https://{platform}.test/{slug}")
    db.add(source)
    await db.flush()
    post = Post(
        source_id=source.id, artist_id=artist.id, external_post_id=ext_id,
        post_url=f"https://{platform}.test/p/{ext_id}",
        post_title=title, post_date=datetime(2023, 8, 1, tzinfo=UTC),
        description=desc, attachment_count=count,
    )
    db.add(post)
    await db.flush()
    return artist, source, post


@pytest.mark.asyncio
async def test_for_image_missing_returns_none(db):
    svc = ProvenanceService(db)
    assert await svc.for_image(999999) is None


@pytest.mark.asyncio
async def test_for_image_no_provenance_returns_empty_list(db):
    rec = await _seed_image(db)
    svc = ProvenanceService(db)
    payload = await svc.for_image(rec.id)
    assert payload == {
        "image_id": rec.id, "provenance": [], "attachments": []
    }


@pytest.mark.asyncio
async def test_for_image_single_provenance_full_shape(db):
    rec = await _seed_image(db)
    artist, source, post = await _seed_post(
        db, artist_name="Alice", slug="alice", platform="patreon",
        ext_id="555", title="Set 1", desc="<p>hi</p><script>x</script>",
        count=2,
    )
    db.add(ImageProvenance(image_record_id=rec.id, post_id=post.id,
                           source_id=source.id))
    await db.flush()

    svc = ProvenanceService(db)
    payload = await svc.for_image(rec.id)
    assert payload["image_id"] == rec.id
    assert len(payload["provenance"]) == 1
    e = payload["provenance"][0]
    assert e["post"]["id"] == post.id
    assert e["post"]["external_post_id"] == "555"
    assert e["post"]["title"] == "Set 1"
    assert e["post"]["attachment_count"] == 2
    assert e["post"]["description_html"] == "<p>hi</p>"  # script removed
    assert e["post"]["url"] == "https://patreon.test/p/555"
    assert e["post"]["date"].startswith("2023-08-01")
    assert e["source"] == {"id": source.id, "platform": "patreon",
                           "url": source.url}
    assert e["artist"] == {"id": artist.id, "name": "Alice",
                           "slug": "alice"}
    assert e["provenance_id"] is not None
    assert e["captured_at"] is not None


@pytest.mark.asyncio
async def test_for_image_multiple_provenance_peer_ordering(db):
    rec = await _seed_image(db)
    _, s1, p1 = await _seed_post(db, artist_name="A1", slug="a1",
                                 platform="patreon", ext_id="1")
    _, s2, p2 = await _seed_post(db, artist_name="A2", slug="a2",
                                 platform="fanbox", ext_id="2")
    ip1 = ImageProvenance(image_record_id=rec.id, post_id=p1.id,
                          source_id=s1.id)
    ip2 = ImageProvenance(image_record_id=rec.id, post_id=p2.id,
                          source_id=s2.id)
    db.add(ip1)
    await db.flush()
    db.add(ip2)
    await db.flush()

    svc = ProvenanceService(db)
    payload = await svc.for_image(rec.id)
    ids = [e["provenance_id"] for e in payload["provenance"]]
    assert ids == sorted(ids)  # captured_at, id ascending → insertion order
    assert len(payload["provenance"]) == 2


@pytest.mark.asyncio
async def test_for_image_null_post_fields_serialize_null(db):
    rec = await _seed_image(db)
    _, source, post = await _seed_post(
        db, artist_name="Bob", slug="bob", platform="x", ext_id="9",
    )  # title/desc/count/post_url default-ish
    db.add(ImageProvenance(image_record_id=rec.id, post_id=post.id,
                           source_id=source.id))
    await db.flush()
    svc = ProvenanceService(db)
    e = (await svc.for_image(rec.id))["provenance"][0]
    assert e["post"]["title"] is None
    assert e["post"]["description_html"] is None
    assert e["post"]["attachment_count"] is None


@pytest.mark.asyncio
async def test_for_post_missing_returns_none(db):
    svc = ProvenanceService(db)
    assert await svc.for_post(999999) is None


@pytest.mark.asyncio
async def test_for_post_returns_post_source_artist(db):
    artist, source, post = await _seed_post(
        db, artist_name="Carol", slug="carol", platform="patreon",
        ext_id="77", title="T", desc="<p>d</p>", count=3,
    )
    svc = ProvenanceService(db)
    payload = await svc.for_post(post.id)
    assert payload["post"]["id"] == post.id
    assert payload["post"]["title"] == "T"
    assert payload["post"]["description_html"] == "<p>d</p>"
    assert payload["post"]["attachment_count"] == 3
    assert payload["source"] == {"id": source.id, "platform": "patreon",
                                 "url": source.url}
    assert payload["artist"] == {"id": artist.id, "name": "Carol",
                                 "slug": "carol"}


@pytest.mark.asyncio
async def test_for_post_includes_attachments(db):
    artist, source, post = await _seed_post(
        db, artist_name="Att", slug="att", platform="patreon",
        ext_id="55",
    )
    db.add(PostAttachment(
        post_id=post.id, artist_id=artist.id, sha256="t" + "0" * 63,
        path="/images/attachments/t00/t.zip", original_filename="t.zip",
        ext=".zip", mime="application/zip", size_bytes=9,
    ))
    await db.flush()
    svc = ProvenanceService(db)
    payload = await svc.for_post(post.id)
    assert len(payload["attachments"]) == 1
    att = payload["attachments"][0]
    assert att["original_filename"] == "t.zip"
    assert att["download_url"].endswith(
        f"/api/attachments/{att['id']}/download"
    )