fix(application-form-fill): 清理填表说明并收窄按钮话术

review_agent/application_form_fill/services/field_extract.py

@@ -28,6 +28,15 @@ FIELD_ALIASES = {
     "storage_condition_and_validity": ["产品储存条件及有效期", "储存条件及有效期", "储存条件", "有效期"],
 }
 
+STATIC_STOP_LABELS = [
+    "申请人",
+    "国家药品监督管理局",
+    "填表说明",
+    "注",
+    "保证书",
+    "应附资料",
+]
+
 
 def collect_document_texts(summary_batch: FileSummaryBatch) -> dict[str, str]:
     texts: dict[str, str] = {}
@@ -180,7 +189,7 @@ def _field_aliases(field: dict[str, str]) -> list[str]:
 
 
 def _all_field_labels(fields: list[dict[str, str]]) -> list[str]:
-    labels: list[str] = []
+    labels: list[str] = list(STATIC_STOP_LABELS)
     for field in fields:
         for label in _field_aliases(field):
             if label not in labels:
@@ -194,7 +203,7 @@ def _extract_label_value(text: str, label: str, labels: list[str]) -> tuple[str,
 
 def _extract_colon_label_value(text: str, label: str, labels: list[str]) -> tuple[str, str]:
     escaped_labels = "|".join(re.escape(item) for item in labels if item != label)
-    stop_pattern = rf"(?=\n\s*(?:{escaped_labels})\s*[:：])" if escaped_labels else r"(?=\Z)"
+    stop_pattern = rf"(?=\n\s*(?:{escaped_labels})(?:\s*[:：]|\s*$))" if escaped_labels else r"(?=\Z)"
     pattern = re.compile(rf"{re.escape(label)}\s*[:：]\s*(.+?)(?:{stop_pattern}|\Z)", re.S)
     match = pattern.search(text or "")
     if not match: