feat(regulatory): 支持按附件4章节核查

review_agent/regulatory_review/workflow.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import json
 import logging
+import re
 from pathlib import Path
 from threading import Thread
 from uuid import uuid4
@@ -48,6 +49,16 @@ NODE_DEFINITIONS = [
 logger = logging.getLogger("review_agent.regulatory_review.workflow")
 
 
+ATTACHMENT4_CHAPTER_LABELS = {
+    "1": "第1章 监管信息",
+    "2": "第2章 综述资料",
+    "3": "第3章 非临床资料",
+    "4": "第4章 临床评价资料",
+    "5": "第5章 产品说明书和标签样稿",
+    "6": "第6章 质量管理体系文件",
+}
+
+
 class WorkflowPausedForUser(Exception):
     pass
 
@@ -89,6 +100,7 @@ def create_regulatory_review_batch(
         source_summary_batch=source_summary_batch,
         batch_no=batch_no,
         work_dir=str(work_dir),
+        condition_json=_initial_condition_json(trigger_message),
     )
     for code, name, group in NODE_DEFINITIONS:
         WorkflowNodeRun.objects.create(
@@ -173,7 +185,7 @@ class RegulatoryWorkflowExecutor:
             self._pause_for_condition_confirmation()
             return
         if node_code == "rule_scope":
-            self.rule_set = load_rule_file()
+            self.rule_set = apply_rule_scope(load_rule_file(), self.batch.condition_json.get("rule_scope") or {})
             return
         if node_code == "completeness_check":
             self.findings.extend(run_completeness_check(self.batch.source_summary_batch, self._rules()))
@@ -258,7 +270,7 @@ class RegulatoryWorkflowExecutor:
 
     def _rules(self) -> dict:
         if self.rule_set is None:
-            self.rule_set = load_rule_file()
+            self.rule_set = apply_rule_scope(load_rule_file(), self.batch.condition_json.get("rule_scope") or {})
         return self.rule_set
 
     def _extract_source_texts(self) -> dict[str, str]:
@@ -298,3 +310,52 @@ def start_regulatory_review_workflow(batch: RegulatoryReviewBatch, *, async_run:
         executor.run()
         return
     Thread(target=executor.run, daemon=True).start()
+
+
+def _initial_condition_json(trigger_message: Message | None) -> dict:
+    scope = detect_attachment4_chapter_scope(trigger_message.content if trigger_message else "")
+    return {"rule_scope": scope} if scope else {}
+
+
+def detect_attachment4_chapter_scope(content: str) -> dict[str, str] | None:
+    normalized = (content or "").strip()
+    if not normalized:
+        return None
+    chapter = _extract_chapter_number(normalized)
+    if chapter not in ATTACHMENT4_CHAPTER_LABELS:
+        return None
+    return {"attachment4_chapter": chapter, "label": ATTACHMENT4_CHAPTER_LABELS[chapter]}
+
+
+def apply_rule_scope(rule_set: dict, rule_scope: dict) -> dict:
+    chapter = str(rule_scope.get("attachment4_chapter") or "")
+    if chapter not in ATTACHMENT4_CHAPTER_LABELS:
+        return rule_set
+    scoped = {**rule_set}
+    scoped["requirements"] = [
+        requirement
+        for requirement in rule_set.get("requirements", [])
+        if _requirement_in_chapter(requirement, chapter)
+    ]
+    scoped["active_rule_scope"] = rule_scope
+    return scoped
+
+
+def _requirement_in_chapter(requirement: dict, chapter: str) -> bool:
+    attachment4_code = str(requirement.get("attachment4_code") or "")
+    return attachment4_code == chapter or attachment4_code.startswith(f"{chapter}.")
+
+
+def _extract_chapter_number(content: str) -> str:
+    match = re.search(r"第\s*([一二三四五六1-6])\s*[章节张]", content)
+    if match:
+        return _normalize_chapter_number(match.group(1))
+    match = re.search(r"(^|[^\d])([1-6])\s*[章节张]", content)
+    if match:
+        return match.group(2)
+    return ""
+
+
+def _normalize_chapter_number(value: str) -> str:
+    chinese = {"一": "1", "二": "2", "三": "3", "四": "4", "五": "5", "六": "6"}
+    return chinese.get(value, value)