feat(regulatory): 对齐附件4目录核查规则

review_agent/regulatory_review/services/rag_index.py

@@ -107,12 +107,19 @@ def collect_source_chunks(source_dir: Path) -> list[TextChunk]:
         try:
             text = extract_text_from_path(path)
         except RuntimeError as exc:
+            if _is_attachment4(path):
+                raise RuntimeError(f"附件 4 核心法规材料抽取失败：{path.name}") from exc
             logger.warning("Regulatory source extraction skipped", extra={"path": str(path), "error": str(exc)})
             continue
         chunks.extend(chunk_text(text, source=str(path.relative_to(source_dir))))
     return chunks
 
 
+def _is_attachment4(path: Path) -> bool:
+    normalized = path.name.replace(" ", "")
+    return "附件4" in normalized and "体外诊断试剂注册申报资料要求及说明" in normalized
+
+
 def build_chroma_index(
     *,
     source_dir: Path,