Checkpoint frontend UI/UX overhaul and ingestion embed benchmark work

apps/ai-service/tests/test_retrieval_service.py

@@ -56,6 +56,82 @@ def table_service(*, visual: bool = False) -> RetrievalService:
     )
 
 
+class _OverviewRetriever:
+    """A fake with `find_by_drug`/`find_by_section` (the Qdrant adapter's
+    shape) — `InMemoryLexicalRetriever` doesn't implement either, so
+    `retrieve_framed`'s overview path is otherwise untestable."""
+
+    def __init__(self, documents: list[RetrievalDocument]) -> None:
+        self._documents = documents
+
+    def find_by_drug(self, drug_id: str) -> list[SearchHit]:
+        return [
+            SearchHit(document=d, score=1.0)
+            for d in self._documents if d.drug_id == drug_id
+        ]
+
+    def find_by_section(self, drug_id: str, section_key: str) -> list[SearchHit]:
+        return [
+            SearchHit(document=d, score=1.0)
+            for d in self._documents
+            if d.drug_id == drug_id and d.section_key == section_key
+        ]
+
+    def search(self, query: str, drug_id: str, limit: int) -> list[SearchHit]:
+        return []
+
+
+_MONOGRAPH_SECTIONS = (
+    "ten_chung_quoc_te", "ma_atc", "loai_thuoc", "dang_thuoc_va_ham_luong",
+    "duoc_ly_va_co_che_tac_dung", "chi_dinh", "chong_chi_dinh", "than_trong",
+    "tac_dung_khong_mong_muon", "lieu_luong_va_cach_dung", "tuong_tac_thuoc",
+    "qua_lieu_va_xu_tri", "do_on_dinh_va_bao_quan", "thong_tin_quy_che",
+)
+
+
+def _monograph_service() -> RetrievalService:
+    documents = [
+        RetrievalDocument(
+            doc_id=f"paracetamol::{section}::0", drug_id="paracetamol",
+            kind="prose", section_key=section,
+            text=f"Nội dung mục {section}.", source_refs=(SOURCE,),
+        )
+        for section in _MONOGRAPH_SECTIONS
+    ]
+    return RetrievalService(
+        _OverviewRetriever(documents), InMemoryParentStore([]),
+        EvidencePolicy(evidence_limit=3),
+    )
+
+
+def test_retrieve_framed_overview_answers_from_intro_sections_only():
+    """Found live 2026-08-06: a bare drug name sent all 14+ sections of the
+    monograph as evidence, producing an answer long enough to intermittently
+    fail generation/entailment. `is_overview=True` must narrow this the same
+    way `retrieve()`'s bare-name branch always has."""
+    result = _monograph_service().retrieve_framed(
+        "paracetamol", None, "paracetamol", is_overview=True
+    )
+    assert result.decision == EvidenceDecision.ANSWERABLE
+    assert result.is_drug_overview is True
+    returned_sections = {e.matched_doc_id.split("::")[1] for e in result.evidence}
+    assert returned_sections <= {
+        "ten_chung_quoc_te", "loai_thuoc", "chi_dinh", "duoc_ly_va_co_che_tac_dung",
+    }
+    assert len(result.evidence) < len(_MONOGRAPH_SECTIONS)
+
+
+def test_retrieve_framed_question_without_section_is_capped_even_without_rerank():
+    # No reranker configured: `_rerank` fails open and returns everything
+    # unfiltered. Hydration must still bound it — an ordering aid failing
+    # open must not also remove the size cap.
+    result = _monograph_service().retrieve_framed(
+        "paracetamol", None, "thuốc này có tác dụng phụ gì", is_overview=False
+    )
+    assert result.decision == EvidenceDecision.ANSWERABLE
+    assert len(result.evidence) <= 3
+
+
 def test_row_hit_hydrates_complete_parent_and_keeps_citation():
     result = table_service().retrieve("acetylcystein 45 kg bao nhiêu ml", "acetylcystein")
     assert result.decision == EvidenceDecision.ANSWERABLE