diff --git a/pdf_info_extract.py b/pdf_info_extract.py index b4ca91d..67cfcf7 100644 --- a/pdf_info_extract.py +++ b/pdf_info_extract.py @@ -154,17 +154,17 @@ def extract_content_by_keyword(text, field_name): """ # 定义关键词映射 keyword_mapping = { - "质量保证方案": ["质量保证", "质量管理", "质量控制", "质量方案", "质量措施"], - "安全保障": ["安全保障", "安全措施", "安全管理", "安全防护", "安全方案"], - "进度管理方案": ["进度管理", "进度控制", "进度安排", "时间安排", "工期管理"], + "质量保证方案": ["质量保证","质量目标","质量管理","质量控制"], + "安全保障": ["安全保障", "安全措施","安全生产责任制","安全教育培训"], + "进度管理方案": ["进度管理", "进度控制", "进度计划","工期保证"], "应急预案": ["应急预案", "应急处理", "应急措施", "突发事件", "应急响应"], - "服务承诺": ["服务承诺", "服务保证", "服务质量", "服务标准"], - "施工标准": ["施工标准", "施工规范", "施工要求", "技术标准", "作业标准"], - "巡查考核方案": ["巡查", "考核", "检查", "监督", "评估"], - "沟通配合措施": ["沟通配合", "协调", "配合", "沟通机制"], - "安全文明建设": ["安全文明", "文明施工", "现场管理"], - "制度建设": ["制度建设", "管理制度", "规章制度", "制度完善"], - "资料整编方案": ["资料整编", "资料管理", "档案管理", "文档整理"] + "服务承诺": ["服务承诺", "服务保证","售后服务","保修承诺","满意"], + "施工标准": ["施工标准", "施工规范" , "技术标准","GB/T","ISO","标准化","BIM"], + "巡查考核方案": ["组织巡查", "监督考核","定期检查","第三方抽检"], + "沟通配合措施": ["沟通配合", "沟通机制","联络人","信息通报","问题反馈","争议解决","业主","利益相关"], + "安全文明建设": ["安全文明", "文明施工","场容场貌","环境保护","绿色","节能","环保"], + "制度建设": ["制度建设","管理流程","内部审核","信息化管理","内部评审"], + "资料整编方案": [ "资料管理","电子化管理","审核流程","资料规范","资料移交"] } keywords = keyword_mapping.get(field_name, [field_name]) @@ -258,7 +258,7 @@ def determine_file_type(file_name, text_preview=""): # 基于文件名判断 if any(keyword in file_name_lower for keyword in ["招标", "采购", "公告"]): return "招标文件" - elif any(keyword in file_name_lower for keyword in ["投标", "响应", "报价"]): + elif any(keyword in file_name_lower for keyword in ["投标", "响应", "报价" , "技术部分"]): return "投标文件"