Wire the guarded conversational RAG answer layer end-to-end

2026-08-05 14:33:13 +07:00
parent 834d9e51b0
commit ef08b4929e
127 changed files with 37921 additions and 169 deletions
@@ -0,0 +1,72 @@
+from ingestion.validation.readiness import evaluate, evaluate_chunks
+
+
+def _count(monographs, gate_name):
+    return next(g.count for g in evaluate(monographs) if g.name == gate_name)
+
+
+def test_readiness_rejects_a_part_without_source_span_provenance():
+    corpus = [{
+        "drug_id": "testdrug",
+        "source_page_range": [100, 100],
+        "sections": {
+            "ma_atc": {
+                "text": "N00AA00",
+                "parts": [{"kind": "prose", "text": "N00AA00", "source_span_ids": []}],
+            },
+        },
+        "tables": [],
+    }]
+    assert _count(corpus, "section_without_provenance") == 0
+    assert _count(corpus, "part_without_source_span_ids") == 1
+
+
+def test_readiness_accepts_part_level_source_span_provenance():
+    corpus = [{
+        "drug_id": "testdrug",
+        "source_page_range": [100, 100],
+        "sections": {
+            "ma_atc": {
+                "text": "N00AA00",
+                "parts": [{
+                    "kind": "prose", "text": "N00AA00",
+                    "source_span_ids": ["p100_b1_l0_s0"],
+                }],
+            },
+        },
+        "tables": [],
+    }]
+    assert _count(corpus, "part_without_source_span_ids") == 0
+
+
+def test_readiness_rejects_duplicate_physical_region_ids():
+    corpus = [{
+        "drug_id": "testdrug",
+        "source_page_range": [100, 100],
+        "sections": {},
+        "tables": [
+            {"table_id": "p100_t0", "quarantined": True},
+            {"table_id": "p100_t0", "quarantined": True},
+        ],
+    }]
+    assert _count(corpus, "duplicate_table_id") == 1
+
+
+def test_chunk_readiness_requires_a_verified_printed_page_range():
+    chunk = {
+        "chunk_id": "drug__dose__0",
+        "drug_id": "drug",
+        "section_key": "dose",
+        "chunk_kind": "prose",
+        "text": "Dose.",
+        "est_tokens": 2,
+        "attachments": [],
+    }
+    gates = evaluate_chunks([], [chunk])
+    missing = next(g for g in gates if g.name == "chunk_without_printed_page_range")
+    assert missing.count == 1
+
+    chunk["printed_page_range"] = [101, 102]
+    gates = evaluate_chunks([], [chunk])
+    present = next(g for g in gates if g.name == "chunk_without_printed_page_range")
+    assert present.count == 0