agent-platform/tests/test_knowledge_pgvector_fallback.py

from __future__ import annotations

import sys
import base64
from pathlib import Path


REPO_ROOT = Path(__file__).resolve().parents[1]
for module_name in list(sys.modules):
    if module_name == "app" or module_name.startswith("app."):
        del sys.modules[module_name]
for path in [
    REPO_ROOT / "libs" / "core-domain" / "src",
    REPO_ROOT / "libs" / "core-shared" / "src",
    REPO_ROOT / "libs" / "core-db" / "src",
    REPO_ROOT / "services" / "knowledge-service",
]:
    sys.path.insert(0, str(path))

from core_db import Base

from app.application.services import KnowledgeApplicationService
from app.bootstrap.settings import KnowledgeServiceSettings
from app.db.session import build_session_factory
from app.domain.repositories import (
    KnowledgeBaseRepository,
    KnowledgeChunkRepository,
    KnowledgeDocumentRepository,
)
from app.schemas.knowledge import (
    KnowledgeBaseCreateRequest,
    KnowledgeDocumentCreateRequest,
    KnowledgeSearchRequest,
)


def test_knowledge_search_falls_back_without_pgvector(tmp_path: Path) -> None:
    settings = KnowledgeServiceSettings(
        database_url=f"sqlite:///{tmp_path / 'knowledge_service.db'}",
        embedding_provider="local",
    )
    session_factory = build_session_factory(settings)
    Base.metadata.create_all(bind=session_factory.kw["bind"])

    with session_factory() as db:
        service = KnowledgeApplicationService(
            settings=settings,
            base_repository=KnowledgeBaseRepository(db),
            document_repository=KnowledgeDocumentRepository(db),
            chunk_repository=KnowledgeChunkRepository(db),
        )
        base = service.create_base(
            KnowledgeBaseCreateRequest(tenant_id="t1", code="kb", name="KB")
        )
        _, chunks = service.create_document(
            KnowledgeDocumentCreateRequest(
                tenant_id="t1",
                knowledge_base_id=base.id,
                title="Refund Policy",
                content_text="Refunds are available within seven days for eligible orders.",
                chunk_size=40,
                chunk_overlap=5,
            )
        )
        assert chunks[0].embedding_vector is not None

        results = service.search(
            KnowledgeSearchRequest(
                tenant_id="t1",
                knowledge_base_id=base.id,
                query="refund seven days",
                top_k=3,
            )
        )

        assert results
        assert results[0][3]["retrieval_mode"] == "hybrid"
        assert results[0][3]["rerank_enabled"] is True
        assert "citation" in results[0][3]
        assert results[0][3]["weights"]["rerank"] == settings.retrieval_rerank_weight
    session_factory.kw["bind"].dispose()


def test_create_document_parses_base64_markdown_before_indexing(tmp_path: Path) -> None:
    settings = KnowledgeServiceSettings(
        database_url=f"sqlite:///{tmp_path / 'knowledge_service.db'}",
        embedding_provider="local",
    )
    session_factory = build_session_factory(settings)
    Base.metadata.create_all(bind=session_factory.kw["bind"])

    with session_factory() as db:
        service = KnowledgeApplicationService(
            settings=settings,
            base_repository=KnowledgeBaseRepository(db),
            document_repository=KnowledgeDocumentRepository(db),
            chunk_repository=KnowledgeChunkRepository(db),
        )
        base = service.create_base(
            KnowledgeBaseCreateRequest(tenant_id="t1", code="kb", name="KB")
        )
        encoded = base64.b64encode(
            "# Refund Policy\nRefunds are available within seven days.".encode("utf-8")
        ).decode("ascii")

        document, chunks = service.create_document(
            KnowledgeDocumentCreateRequest(
                tenant_id="t1",
                knowledge_base_id=base.id,
                title="Refund Policy",
                source_type="markdown",
                content_base64=encoded,
                chunk_size=80,
                chunk_overlap=0,
            )
        )

        assert document.source_type == "markdown"
        assert document.content_text.startswith("Refund Policy")
        assert document.metadata_json is not None
        assert document.metadata_json["parser_metadata"]["parser"] == "knowledge-document-parser-v1"
        assert chunks
        assert chunks[0].content_text.startswith("Refund Policy")
    session_factory.kw["bind"].dispose()
    Base.metadata.clear()