app/tests/test_content_similar_api.py

"""HTTP tests for POST /content/similar (library + draft overlap)."""

import pytest

from app.db.models import Document, IngestStatus
from app.models.schemas import SearchResult


class _FakeVS:
    """Vector store returning fixed chunks (no FAISS on disk)."""

    def __init__(self, rows: list[SearchResult]) -> None:
        self._rows = rows

    def search(self, query: str, tenant_id: str, k: int = 10, **kwargs: object) -> list[SearchResult]:
        return self._rows


@pytest.mark.asyncio
async def test_content_similar_returns_library_and_filename(
    async_client: object,
    test_db: object,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    doc_id = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
    tenant = "tenant_test"
    doc = Document(
        id=doc_id,
        tenant_id=tenant,
        filename="rics_guidance.pdf",
        file_path="/tmp/rics_guidance.pdf",
        status=IngestStatus.complete,
    )
    test_db.add(doc)
    await test_db.commit()

    hits = [
        SearchResult(
            chunk_id="chunk-1",
            doc_id=doc_id,
            tenant_id=tenant,
            text="Approved Document guidance on damp proof courses and minimum heights",
            score=0.35,
            section_type="paragraph",
        ),
    ]
    monkeypatch.setattr(
        "app.services.content_similarity.get_vectorstore",
        lambda: _FakeVS(hits),
    )

    res = await async_client.post(
        "/content/similar",
        json={
            "text": "damp proof course minimum height regulations above ground level",
            "section_code": "E4",
            "peer_sections": {},
        },
        headers={"X-Tenant-ID": tenant},
    )
    assert res.status_code == 200
    data = res.json()
    assert len(data["library_matches"]) == 1
    assert data["library_matches"][0]["document_id"] == doc_id
    assert data["library_matches"][0]["filename"] == "rics_guidance.pdf"
    assert "damp" in data["library_matches"][0]["snippet"].lower()


@pytest.mark.asyncio
async def test_content_similar_excludes_document_ids(
    async_client: object,
    test_db: object,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    doc_id = "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb"
    tenant = "tenant_test"
    doc = Document(
        id=doc_id,
        tenant_id=tenant,
        filename="survey.docx",
        file_path="/tmp/survey.docx",
        status=IngestStatus.complete,
    )
    test_db.add(doc)
    await test_db.commit()

    hits = [
        SearchResult(
            chunk_id="c1",
            doc_id=doc_id,
            tenant_id=tenant,
            text="Only chunk from excluded survey file",
            score=0.9,
            section_type="paragraph",
        ),
    ]
    monkeypatch.setattr(
        "app.services.content_similarity.get_vectorstore",
        lambda: _FakeVS(hits),
    )

    res = await async_client.post(
        "/content/similar",
        json={
            "text": "matching text long enough for library search twelve",
            "section_code": "D",
            "peer_sections": {},
            "exclude_document_ids": [doc_id],
        },
        headers={"X-Tenant-ID": tenant},
    )
    assert res.status_code == 200
    data = res.json()
    assert data["library_matches"] == []
    assert "excluded" in data["message"].lower() or "filters" in data["message"].lower()


@pytest.mark.asyncio
async def test_content_similar_draft_overlap_without_library(
    async_client: object,
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    monkeypatch.setattr(
        "app.services.content_similarity.get_vectorstore",
        lambda: _FakeVS([]),
    )

    text = (
        "planning permission required for loft conversion and rear extension "
        "under permitted development rules"
    )
    peer = (
        "planning permission required for loft conversion and rear extension "
        "under permitted development rules 2024 update"
    )

    res = await async_client.post(
        "/content/similar",
        json={
            "text": text,
            "section_code": "E2",
            "peer_sections": {"I1": peer},
        },
        headers={"X-Tenant-ID": "tenant_test"},
    )
    assert res.status_code == 200
    data = res.json()
    assert data["draft_overlaps"]
    assert data["draft_overlaps"][0]["other_section_code"] == "I1"