feat(regulatory): 对齐附件4目录核查规则

tests/test_regulatory_rag.py

@@ -6,6 +6,7 @@ from review_agent.regulatory_review.services.rag_citation import (
 )
 from review_agent.regulatory_review.services.rag_embedding import SiliconFlowEmbeddingProvider
 from review_agent.regulatory_review.services.rag_index import chunk_text
+from review_agent.regulatory_review.services.rag_index import collect_source_chunks
 
 
 def test_siliconflow_embedding_provider_posts_expected_payload(monkeypatch):
@@ -70,3 +71,18 @@ def test_retrieve_citations_raises_when_index_missing(settings, tmp_path):
 
     with pytest.raises(RagIndexUnavailable):
         retrieve_citations("注册检验报告", embedding_provider=lambda texts: [[0.1]])
+
+
+def test_collect_source_chunks_requires_attachment4_extraction(monkeypatch, tmp_path):
+    source_dir = tmp_path / "sources"
+    source_dir.mkdir()
+    attachment4 = source_dir / "附件 4 体外诊断试剂注册申报资料要求及说明.doc"
+    attachment4.write_bytes(b"legacy-doc")
+
+    def fail_extract(path):
+        raise RuntimeError("无法通过 LibreOffice 转换法规 .doc 材料")
+
+    monkeypatch.setattr("review_agent.regulatory_review.services.rag_index.extract_text_from_path", fail_extract)
+
+    with pytest.raises(RuntimeError, match="附件 4"):
+        collect_source_chunks(source_dir)