打开方案，规程规范和图纸的功能的训练和标准化

api/constants.py

@@ -13,7 +13,7 @@ USELESS_PROGRAM_DEPARTMENT_WORDS = {"项目管理部", "项目部"}
 
 #公司名标准化时需要过滤掉的词汇
 USELESS_COMPANY_WORDS = ["公司","有限","责任","工程","科技"]
-
+USELESS_DESIGN_WORDS = {"方案", "措施"}
 #提取公司名热词需要过滤掉的词汇
 # USELESS_COMPANY_WORDS = ["公司","有限","责任","工程","科技","安徽省","国网","四川省","安徽","集团","电力","建设","建筑","安装","股份"
 #                          "装饰","结构","能源","发展","装饰","电气","股份"]
@@ -43,8 +43,14 @@ RISK_LEVEL = "riskLevel"
 TEAM_NAME = "teamName"
 
 PAGE = "page"
-PROGRAM_NAVIGATION = "programNavigation"
+# PROGRAM_NAVIGATION = "programNavigation"
+DESIGN_SPECIFICATION = "designSpecificationName"
+DESIGN = "designName"
+SPECIFICATION = "specificationName"
 
+PICTURE = "picName"
+#方案的后缀
+design_suffix = ["措施","方案","规划","三措一案"]
 
 # 意图识别和槽位抽取服务返回的关键提示语列表
 SLOT_KEYWORDS = [

api/globalData.py

@@ -9,6 +9,7 @@ from config import redis_url
 
 logger = setup_logger("GlobalData", level=logging.DEBUG)
 
+
 class GlobalData:
     # 数据字段
     standard_company_program = {}
@@ -32,6 +33,8 @@ class GlobalData:
     simply_to_standard_team_leader_name_map = {}
     pinyin_simply_to_standard_team_leader_name_map = {}
 
+    standard_design_pic = {}
+
     @classmethod
     def update_from_redis(cls):
         import sys
@@ -50,20 +53,22 @@ class GlobalData:
 
         #建管单位
         cls._update_list_data('SBD_QUERY_DATA:CONSTRUCTION_UNIT', './standard_data/construct_unit.txt',
-                             cls.standard_construct_name_list, cls.simply_to_standard_construct_name_map,
+                              cls.standard_construct_name_list, cls.simply_to_standard_construct_name_map,
-                             cls.pinyin_simply_to_standard_construct_name_map, clean_useless_company_name)
+                              cls.pinyin_simply_to_standard_construct_name_map, clean_useless_company_name)
         logger.info(f"建管单位数量：{len(cls.standard_construct_name_list)}")
 
         #分包单位
         cls._update_list_data('SBD_QUERY_DATA:SUBCONTRACTOR', './standard_data/sub_contract.txt',
-                             cls.standard_constractor_name_list, cls.simply_to_standard_constractor_name_map,
+                              cls.standard_constractor_name_list, cls.simply_to_standard_constractor_name_map,
-                             cls.pinyin_simply_to_standard_constractor_name_map, clean_useless_company_name)
+                              cls.pinyin_simply_to_standard_constractor_name_map, clean_useless_company_name)
         logger.info(f"分包单位数量：{len(cls.standard_constractor_name_list)}")
 
         #班组名称
         cls._update_list_data('SBD_QUERY_DATA:TEAM', './standard_data/team_leader.txt',
-                             cls.standard_team_leader_name_list, cls.simply_to_standard_team_leader_name_map,
+                              cls.standard_team_leader_name_list, cls.simply_to_standard_team_leader_name_map,
-                             cls.pinyin_simply_to_standard_team_leader_name_map, clean_useless_team_leader_name)
+                              cls.pinyin_simply_to_standard_team_leader_name_map, clean_useless_team_leader_name)
+
+        cls.update_design_pic_info()
         logger.info(f"班组名称数量：{len(cls.standard_team_leader_name_list)}")
 
     @classmethod
@@ -80,7 +85,7 @@ class GlobalData:
             json_str = r.get('SBD_QUERY_DATA:STANDARD_COMPANY_PROGRAM')
             if json_str:
                 temp_data = json.loads(json_str)
-                save_dict_to_file(temp_data,"./standard_data/standard_company_program.json")
+                save_dict_to_file(temp_data, "./standard_data/standard_company_program.json")
                 logger.info("[Info] Loaded STANDARD_COMPANY_PROGRAM from Redis")
             else:
                 raise ValueError("Redis key not found")
@@ -122,7 +127,7 @@ class GlobalData:
         if json_str:
             try:
                 temp_list = json.loads(json_str)
-                save_standard_name_list_to_file(temp_list,local_path)
+                save_standard_name_list_to_file(temp_list, local_path)
                 logger.info(f"[Info] Loaded {redis_key} from Redis")
             except json.JSONDecodeError as e:
                 logger.info(f"[Warning] JSON decode error on key '{redis_key}': {e}")
@@ -144,3 +149,87 @@ class GlobalData:
             pinyin_map.update({
                 text_to_pinyin(cleaner(kw)): kw for kw in temp_list
             })
+
+    @classmethod
+    def update_design_pic_info(cls):
+
+        from utils import (
+            load_standard_json_data,
+            save_dict_to_file,
+            clean_useless_company_name,
+            text_to_pinyin
+        )
+        # 公司与工程关系数据
+        try:
+            r = redis.from_url(redis_url, decode_responses=True)
+            json_str = r.get('SBD_QUERY_DATA:STANDARD_DESIGN_PIC_INFO')
+            if json_str:
+                temp_data = json.loads(json_str)
+                save_dict_to_file(temp_data, "./standard_data/standard_project_info.json")
+                logger.info("[Info] Loaded STANDARD_DESIGN_PIC_INFO from Redis")
+            else:
+                raise ValueError("Redis key not found")
+        except Exception as e:
+            logger.error(f"[Error] Error loading STANDARD_DESIGN_PIC_INFO: {e}")
+            temp_data = load_standard_json_data("./standard_data/standard_project_info.json")
+
+        print(f"STANDARD_DESIGN_PIC_INFO:{temp_data}")
+        if temp_data != cls.standard_design_pic:
+            cls.standard_design_pic.clear()
+            cls.standard_design_pic.update(temp_data)
+
+    @classmethod
+    def get_all_company_from_design_info(cls):
+        #获取所有分公司信息
+        return list(cls.standard_design_pic.keys())
+
+    @classmethod
+    def get_project_from_design_info(cls):
+        #获取所有工程信息
+        project_list = []
+
+        for company, projects in cls.standard_design_pic.items():
+            project_list.extend(projects.keys())
+        return project_list
+
+
+    @classmethod
+    def get_contents_by_company_proj(cls, company_name, project_name):
+        #根据分公司名和工程名 获取方案，图纸和规范规程
+        global company_projects
+        if company_name and project_name:
+            return cls.standard_design_pic.get(company_name, {}).get(project_name, {})
+        elif not company_name and project_name:
+            for company, projects in cls.standard_design_pic.items():
+                if project_name in projects:
+                    return projects[project_name]  # 返回该工程下的“方案/图纸/规范规程”
+        elif company_name and not project_name:
+            result = {
+                "方案": [],
+                "图纸": [],
+                "规范规程": []
+            }
+            company_projects = cls.standard_design_pic.get(company_name, {})
+            # 忽略 "@type" 字段
+            if "@type" in company_projects:
+                del company_projects["@type"]
+
+            for proj_key, project in company_projects.items():
+                if "@type" in project:
+                    del project["@type"]
+                for key in result:
+                    result[key].extend(project.get(key, []))
+
+            # print(f"最终result：{result}")
+            return result
+        else:
+            return None
+
+
+    @classmethod
+    def get_contents_by_proj(cls, project_name):
+        #根据工程名 获取方案，图纸和规范规程
+        for company, projects in cls.standard_design_pic.items():
+            if project_name in projects:
+                return projects[project_name]  # 返回该工程下的“方案/图纸/规范规程”
+            return None  # 没找到

api/main.py

@@ -16,7 +16,7 @@ from globalData import GlobalData
 from apscheduler.schedulers.background import BackgroundScheduler
 
 MODEL_ERNIE_PATH = R"../ernie/output/checkpoint-14672"
-MODEL_UIE_PATH = R"../uie/output/checkpoint-16380"
+MODEL_UIE_PATH = R"../uie/output_temp/checkpoint-18774"
 
 
 # 类别名称列表
@@ -48,8 +48,8 @@ label_map = {
     15: 'B-personName', 34: 'I-personName',
     16: 'B-personQueryType', 35: 'I-personQueryType',
     17: 'B-projectStatus', 36: 'I-projectStatus',
-    18: 'B-skyNet', 37: 'I-skyNet',
+    18: 'B-picName', 37: 'I-picName',
-    19: 'B-programNavigation', 38: 'I-programNavigation'
+    19: 'B-designSpecificationName', 38: 'I-designSpecificationName'
 }
 
 logger = setup_logger("main", level=logging.DEBUG)
@@ -406,6 +406,7 @@ def extract_multi_chat(messages):
     请你仅输出还原后的完整问题，不要输出任何变量、中间步骤或解释说明，确保结果自然通顺，语义完整。
     '''
 
+
     message = [
         {"role": "user", "content": prompt}
     ]

api/main_temp.py

@@ -16,7 +16,7 @@ from globalData import GlobalData
 from apscheduler.schedulers.background import BackgroundScheduler
 
 MODEL_ERNIE_PATH = R"../ernie/output/checkpoint-14672"
-MODEL_UIE_PATH = R"../uie/output/checkpoint-16380"
+MODEL_UIE_PATH = R"../uie/output_temp/checkpoint-20860"
 
 
 # 类别名称列表
@@ -48,8 +48,8 @@ label_map = {
     15: 'B-personName', 34: 'I-personName',
     16: 'B-personQueryType', 35: 'I-personQueryType',
     17: 'B-projectStatus', 36: 'I-projectStatus',
-    18: 'B-skyNet', 37: 'I-skyNet',
+    18: 'B-picName', 37: 'I-picName',
-    19: 'B-programNavigation', 38: 'I-programNavigation'
+    19: 'B-designSpecificationName', 38: 'I-designSpecificationName'
 }
 
 logger = setup_logger("main", level=logging.DEBUG)
@@ -386,37 +386,16 @@ def extract_multi_chat(messages):
         示例：补全模糊表达("今天送一分公司有多少作业计划", "具体是哪些") 返回 "今天送一分公司具体有哪些作业计划"
         
     函数 是查询新属性(文本, 新问题):
-        如果新问题中没有查询主体仅有查询对象 则返回TRUE
+        如果新问题中提取不到主体 且仅能提取到查询属性
-        如果新问题中仅有查询主体但没有查询对象 则返回TRUE
+        且这个查询属性和文本中提取到的查询属性不同 则返回TRUE
         其他情况均返回FALSE
         示例：是查询新属性("今天送一分公司有多少作业计划", "作业内容") 返回 True
     
-    函数 删除数量词(文本):
+    函数 替换新属性(文本，新查询属性):
-        删除“有多少”、“多少”、“几条”、“几个”等数量问句词
+        先删除文本中的"有多少"等类似的表达数量表达，
-    
+        再将文本里的查询属性替换为新查询属性，并保持其他内容不变并返回 且保持新查询属性的语气
-    函数 替换查询属性(文本, 新查询属性):
+        示例：替换新属性("今天送一分公司有多少作业计划", "作业内容") 返回 "今天送一分公司的作业内容"
-        说明：
-            本函数用于在删除数量词后，将原句中与“新查询属性”同类型的核心查询词替换为“新查询属性”，并确保其他内容保持不变且语义自然。
         
-        处理步骤：
-            1. 删除文本中的数量类词语，例如“有多少”、“多少”、“几个”、“几条”等。
-            2. 识别原句中的核心查询属性词，判断其与“新查询属性”是否属于相同类别（如均为对象、地点、组织等）。
-            3. 将原有核心查询词替换为“新查询属性”，保留句中其余上下文结构不变。
-            4. 保持句子语气自然，避免引入“是什么”、“有多少”等疑问表达。
-        
-        返回：
-            返回替换后的文本，语义清晰、语气自然。
-        
-        示例：
-            替换查询属性("今天送一分公司有多少作业计划", "作业内容")
-            → "今天送一分公司的作业内容"
-        
-            替换查询属性("今天送一分公司的班组详情", "送二分公司")
-            → "今天送二分公司的班组详情"
-            
-            替换查询属性("今天送一分公司的班组详情", "明天呢")
-            → "明天送二分公司的班组详情"
-    
     函数 有完整的句意(新问题):
         如果新问题里有主体同时有操作对象或查询对象则返回TRUE
         其他情况均返回FALSE
@@ -427,6 +406,7 @@ def extract_multi_chat(messages):
     请你仅输出还原后的完整问题，不要输出任何变量、中间步骤或解释说明，确保结果自然通顺，语义完整。
     '''
 
+
     message = [
         {"role": "user", "content": prompt}
     ]

api/slotRecognition.py

@@ -2,8 +2,9 @@ import paddle
 from paddlenlp.transformers import ErnieForTokenClassification, ErnieTokenizer
 
 from globalData import GlobalData
-from utils import standardize_name_only_high_score, clean_useless_company_name
+from utils import standardize_name_only_high_score, clean_useless_company_name, is_design_file
-from constants import SUBCONTRACTOR, CONSTRUCTION_UNIT, IMPLEMENTATION_ORG, PAGE, PROGRAM_NAVIGATION, PROJECT_DEPARTMENT
+from constants import SUBCONTRACTOR, CONSTRUCTION_UNIT, IMPLEMENTATION_ORG, PAGE, \
+    PROJECT_DEPARTMENT, DESIGN_SPECIFICATION, DESIGN, SPECIFICATION
 import paddle.nn.functional as F
 
 
@@ -178,14 +179,13 @@ class SlotRecognition:
                         else:
                             updates[key] = value
                             prob_updates[key] = slot_probabilities[key]
-            elif key == PROGRAM_NAVIGATION or key == PAGE:
+            elif key == DESIGN_SPECIFICATION:
-                if "施" in value:
+                if is_design_file(value):
-                    updates[key] = "施工生产管理平台"
+                    updates[DESIGN] = value
+                    prob_updates[DESIGN] = 1
                 else:
-                    updates[key] = value
+                    updates[SPECIFICATION] = value
-                    prob_updates[key] = slot_probabilities[key]
+                    prob_updates[SPECIFICATION] = 1
-
-            # 先不处理 PROJECT_DEPARTMENT，后续单独处理
             elif key != PROJECT_DEPARTMENT:
                 updates[key] = value
                 prob_updates[key] = slot_probabilities[key]

api/utils.py

@@ -13,7 +13,7 @@ import re
 from globalData import GlobalData
 from constants import USELESS_COMPANY_WORDS, USELESS_PROJECT_WORDS, CONSTRUCTION_UNIT, IMPLEMENTATION_ORG, \
     SUBCONTRACTOR, PROJECT_NAME, PROJECT_DEPARTMENT, RISK_LEVEL, TEAM_NAME, USELESS_PROGRAM_DEPARTMENT_WORDS, \
-    SLOT_KEYWORDS
+    SLOT_KEYWORDS, design_suffix, DESIGN, SPECIFICATION, PICTURE, USELESS_DESIGN_WORDS
 
 from logger_util import setup_logger
 
@@ -64,6 +64,7 @@ def load_standard_json_data(path):
         # print(f"[Error] Failed to load local JSON file: {e}")
         return {}
 
+
 #将字典序列的josn 存入本地文件
 def save_dict_to_file(data: dict, file_path: str):
     """
@@ -83,6 +84,7 @@ def save_dict_to_file(data: dict, file_path: str):
         # print(f"[Error] 写入 JSON 文件失败：{e}")
         logger.error("[Error] 写入 JSON 文件失败：", exc_info=e)
 
+
 #从指定文件中加载标准化的名称列表。
 def load_standard_name_list(file_path: str):
     """
@@ -111,6 +113,7 @@ def load_standard_name_list(file_path: str):
         # print(f"读取文件时发生错误：{e}", flush=True)
         raise Exception(f"错误：文件 {file_path} 不存在")
 
+
 #将标准化名称列表写入指定文件中，每行一个名称。
 def save_standard_name_list_to_file(name_list, file_path):
     """
@@ -130,6 +133,7 @@ def save_standard_name_list_to_file(name_list, file_path):
     except Exception as e:
         logger.error(f"[Error] 写入文件失败：{e}")
 
+
 def extract_number(text):
     """
     提取项目部中的数字（支持阿拉伯数字和中文数字），并转换为统一格式（中文数字）。
@@ -148,6 +152,7 @@ def replace_arabic_with_chinese(text):
     将字符串中所有连续的阿拉伯数字转换为对应的中文数字。
     示例：2024年25号 -> 二千零二十四年二十五号
     """
+
     def convert(match):
         num_str = match.group()
         try:
@@ -187,6 +192,7 @@ def fuzzy_match_and_filter(input_key, match_pool, mapping_dict, lower_score=70,
     else:
         return [mapping_dict[m[0]] for m in high_conf_matches[:top_k]]
 
+
 def fuzzy_match_and_filter_only_high_score(input_key, match_pool, mapping_dict, high_score=90, top_k=3):
     """
     对输入字符串在候选池中执行模糊匹配，并返回匹配程度高的映射原始值。
@@ -212,6 +218,7 @@ def fuzzy_match_and_filter_only_high_score(input_key, match_pool, mapping_dict,
     else:
         return [mapping_dict[m[0]] for m in high_conf_matches[:top_k]]
 
+
 def standardize_name(input_name, clean_func, simply_map, pinyin_map, lower_score=70, high_score=85):
     """
     通用名称标准化函数，按中文 → 清洗 → 简化匹配 → 拼音匹配 的顺序进行处理。
@@ -234,6 +241,7 @@ def standardize_name(input_name, clean_func, simply_map, pinyin_map, lower_score
     result = fuzzy_match_and_filter(pinyin_input, list(pinyin_map.keys()), pinyin_map, lower_score, high_score)
     return result
 
+
 def standardize_name_only_high_score(input_name, clean_func, simply_map, pinyin_map, high_score=90):
     """
     通用名称标准化函数，按中文 → 清洗 → 简化匹配 → 拼音匹配 的顺序进行处理。
@@ -256,6 +264,7 @@ def standardize_name_only_high_score(input_name, clean_func, simply_map, pinyin_
     result = fuzzy_match_and_filter_only_high_score(pinyin_input, list(pinyin_map.keys()), pinyin_map, high_score)
     return result
 
+
 #标准化班组名称
 def standardize_team_name(input_name, simply_map, pinyin_map, lower_score=70, high_score=90):
     """
@@ -283,7 +292,9 @@ def standardize_sub_company(input_name, simply_map, pinyin_map, lower_score=55,
     :return: 匹配的标准公司名列表
     """
     temp_input_name = replace_arabic_with_chinese(input_name)
-    return standardize_name(temp_input_name, clean_useless_company_name, simply_map, pinyin_map, lower_score, high_score)
+    return standardize_name(temp_input_name, clean_useless_company_name, simply_map, pinyin_map, lower_score,
+                            high_score)
+
 
 def standardize_project_name(input_name, simply_map, pinyin_map, lower_score=70, high_score=90):
     """
@@ -365,7 +376,8 @@ def multiple_standardize_single_name(origin_input_name, origin_name_list, pinyin
                                     limit=len(origin_name_list))
     # 找到所有相似度 > 80 的匹配项
     original_high_confidence_matches = [(match[0], match[1]) for match in match_results if match[1] >= lower_score]
-    logger.info(f"standardize_pinyin_single_name 原始名匹配, high_confidence_matches:{original_high_confidence_matches[:3]}")
+    logger.info(
+        f"standardize_pinyin_single_name 原始名匹配, high_confidence_matches:{original_high_confidence_matches[:3]}")
 
     combined_low_confidence_matches = []
     if original_high_confidence_matches:
@@ -382,7 +394,7 @@ def multiple_standardize_single_name(origin_input_name, origin_name_list, pinyin
     return list(dict.fromkeys(combined_low_confidence_matches))
 
 
-def generate_project_prompt_with_key(matched_projects, original_name="", slot_key = IMPLEMENTATION_ORG):
+def generate_project_prompt_with_key(matched_projects, original_name="", slot_key=IMPLEMENTATION_ORG):
     """
     生成提示信息，用于让用户确认匹配的项目名或分公司名或项目名。
 
@@ -393,7 +405,8 @@ def generate_project_prompt_with_key(matched_projects, original_name="", slot_ke
     返回:
         str: 生成的提示信息。如果未找到匹配项，返回提示用户提供更准确信息的字符串。
     """
-    logger.info(f"generate_project_prompt_with_key slot_key:{slot_key},original_name:{original_name},matched_projects:{matched_projects} ")
+    logger.info(
+        f"generate_project_prompt_with_key slot_key:{slot_key},original_name:{original_name},matched_projects:{matched_projects} ")
     type = ""
     if slot_key == CONSTRUCTION_UNIT:
         type = "建管单位名"
@@ -409,7 +422,7 @@ def generate_project_prompt_with_key(matched_projects, original_name="", slot_ke
     # print(f"generate_project_prompt_with_key type:{type} ")
     logger.info(f"generate_project_prompt_with_key type:{type} ")
     if not matched_projects:
-        if slot_key in (CONSTRUCTION_UNIT,IMPLEMENTATION_ORG,SUBCONTRACTOR):
+        if slot_key in (CONSTRUCTION_UNIT, IMPLEMENTATION_ORG, SUBCONTRACTOR):
             return f"<p>未找到匹配的<strong>公司名</strong>：{original_name}，请提供更准确的公司名信息。</p>"
         else:
             return f"<p>未找到匹配的：{original_name}，请提供更准确的信息。</p>"
@@ -424,6 +437,7 @@ def generate_project_prompt_with_key(matched_projects, original_name="", slot_ke
         html_parts.append("<p>请确认您要选择哪一个？</p>")
         return "\n".join(html_parts)
 
+
 def generate_project_prompt(matched_projects, original_name="", type="项目部名"):
     """
     生成提示信息，用于让用户确认匹配的项目名或分公司名或项目名。
@@ -448,6 +462,7 @@ def generate_project_prompt(matched_projects, original_name="", type="项目部
     html_parts.append("<p>请确认您要选择哪一个？</p>")
     return "\n".join(html_parts)
 
+
 def generate_confirm_prompt(matched_projects, original_name="", type="项目部名"):
     """
     生成提示信息，用于让用户确认匹配的项目名或分公司名或项目名。
@@ -495,6 +510,8 @@ company_symbols_pattern = re.compile(r"[\s\W_]+")
 
 useless_team_leader_words_pattern = re.compile("班组")
 
+useless_design_words_pattern = re.compile("|".join(USELESS_DESIGN_WORDS))
+
 
 def clean_useless_project_name(name: str) -> str:
     # 去掉无意义词
@@ -510,11 +527,13 @@ def clean_useless_company_name(name: str) -> str:
     name = company_symbols_pattern.sub("", name)
     return name.strip()
 
+
 def clean_useless_team_leader_name(name: str) -> str:
     # 去掉无意义词
     name = useless_team_leader_words_pattern.sub("", name)
     return name.strip()
 
+
 #去掉项目部里面的不重要词汇
 def clean_useless_program_departement_name(name: str) -> str:
     # 去掉无意义词
@@ -523,6 +542,21 @@ def clean_useless_program_departement_name(name: str) -> str:
     name = project_symbols_pattern.sub("", name)
     return name.strip()
 
+
+def clean_useless_design_name(name: str) -> str:
+    # 去掉无意义词
+    name = useless_design_words_pattern.sub("", name)
+    # 去掉数字、字母、符号
+    name = project_symbols_pattern.sub("", name)
+    return name.strip()
+
+
+def clean_useless_specification_name(name: str) -> str:
+    # 去掉数字、字母、符号
+    name = project_symbols_pattern.sub("", name)
+    return name.strip()
+
+
 #槽位缺失检查
 def check_lost(int_res, slot):
     #labels: ["天气查询","通用对话","页面切换","日计划数量查询","周计划数量查询","日计划作业内容","周计划作业内容","施工人数","作业考勤人数","知识问答"]
@@ -572,7 +606,7 @@ def check_lost(int_res, slot):
     apologize_str = "非常抱歉，"
     # if int_res == 2:
     #     return CheckResult.NEEDS_MORE_ROUNDS, f"{apologize_str}请问你想查询哪个页面？"
-    if int_res in [3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 15,16]:
+    if int_res in [3, 4, 5, 6, 7, 8, 11, 12, 13, 14, 15, 16]:
         return CheckResult.NEEDS_MORE_ROUNDS, f"{apologize_str}请问你想查询什么时间的{intention_mapping[int_res]}？"
 
 
@@ -607,7 +641,8 @@ def check_standard_name_slot_probability(int_res, slot) -> tuple:
             if match_results and len(match_results) == 1:
                 slot[key] = match_results[0]
             else:
-                prompt = generate_project_prompt_with_key(match_results, original_name=slot[IMPLEMENTATION_ORG], slot_key= IMPLEMENTATION_ORG)
+                prompt = generate_project_prompt_with_key(match_results, original_name=slot[IMPLEMENTATION_ORG],
+                                                          slot_key=IMPLEMENTATION_ORG)
                 return CheckResult.NEEDS_MORE_ROUNDS, prompt
 
         if key == CONSTRUCTION_UNIT:
@@ -618,7 +653,8 @@ def check_standard_name_slot_probability(int_res, slot) -> tuple:
             if match_results and len(match_results) == 1:
                 slot[key] = match_results[0]
             else:
-                prompt = generate_project_prompt_with_key(match_results, original_name=slot[CONSTRUCTION_UNIT], slot_key= CONSTRUCTION_UNIT)
+                prompt = generate_project_prompt_with_key(match_results, original_name=slot[CONSTRUCTION_UNIT],
+                                                          slot_key=CONSTRUCTION_UNIT)
                 return CheckResult.NEEDS_MORE_ROUNDS, prompt
 
         if key == SUBCONTRACTOR:
@@ -629,12 +665,14 @@ def check_standard_name_slot_probability(int_res, slot) -> tuple:
             if match_results and len(match_results) == 1:
                 slot[key] = match_results[0]
             else:
-                prompt = generate_project_prompt_with_key(match_results, original_name=slot[SUBCONTRACTOR], slot_key= SUBCONTRACTOR)
+                prompt = generate_project_prompt_with_key(match_results, original_name=slot[SUBCONTRACTOR],
+                                                          slot_key=SUBCONTRACTOR)
                 return CheckResult.NEEDS_MORE_ROUNDS, prompt
 
         if key == PROJECT_DEPARTMENT:
             logger.info(f"check_standard_name_slot 原始项目部名 : {slot[PROJECT_DEPARTMENT]}")
-            match_results = standardize_projectDepartment(slot[IMPLEMENTATION_ORG], value, GlobalData.standard_company_program,
+            match_results = standardize_projectDepartment(slot[IMPLEMENTATION_ORG], value,
+                                                          GlobalData.standard_company_program,
                                                           high_score=95)
             logger.info(f"check_standard_name_slot 匹配后项目部名： result:{match_results}")
             if match_results and len(match_results) == 1:
@@ -657,9 +695,139 @@ def check_standard_name_slot_probability(int_res, slot) -> tuple:
                                                                                                  "五级"]:
                 return CheckResult.NEEDS_MORE_ROUNDS, "您查询的风险等级在系统中未找到，请确认风险等级后再次提问"
 
+    #前提已经做过公司名和工程名的标准化
+    return standardize_specification_design_pic(slot)
+    # return CheckResult.NO_MATCH, ""
+
+
+def standardize_implement_company(slot_item) -> tuple:
+    if IMPLEMENTATION_ORG in slot_item:
+        value = slot_item[IMPLEMENTATION_ORG]
+        logger.info(f"standardize_specification_design_pic 原始分公司名 : {value}")
+        match_results = standardize_sub_company(value, GlobalData.simply_to_standard_company_name_map,
+                                                GlobalData.pinyin_simply_to_standard_company_name_map, 70, 90)
+        logger.info(f"standardize_specification_design_pic 匹配后分公司名： result:{match_results}")
+        if match_results and len(match_results) == 1:
+            slot_item[IMPLEMENTATION_ORG] = match_results[0]
+        else:
+            prompt = generate_project_prompt_with_key(match_results, original_name=slot_item[IMPLEMENTATION_ORG],
+                                                      slot_key=IMPLEMENTATION_ORG)
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
     return CheckResult.NO_MATCH, ""
 
 
+def standardize_project(slot_item) -> tuple:
+    if PROJECT_NAME in slot_item:
+        value = slot_item[PROJECT_NAME]
+        logger.info(f"standardize_specification_design_pic 原始工程名 : {slot_item[PROJECT_NAME]}")
+        match_results = standardize_project_name(value, GlobalData.simply_to_standard_project_name_map,
+                                                 GlobalData.pinyin_simply_to_standard_project_name_map, 70, 90)
+        logger.info(f"standardize_specification_design_pic 匹配后工程名 ：result:{match_results}")
+
+        if match_results and len(match_results) == 1:
+            slot_item[PROJECT_NAME] = match_results[0]
+        else:
+            prompt = generate_project_prompt(match_results, original_name=slot_item[PROJECT_NAME], type="工程名")
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
+    return CheckResult.NO_MATCH, ""
+
+
+def standardize_design(slot_item) -> tuple:
+    if PROJECT_NAME in slot_item:
+        value = slot_item[PROJECT_NAME]
+        logger.info(f"standardize_specification_design_pic 原始工程名 : {slot_item[PROJECT_NAME]}")
+        match_results = standardize_project_name(value, GlobalData.simply_to_standard_project_name_map,
+                                                 GlobalData.pinyin_simply_to_standard_project_name_map, 70, 90)
+        logger.info(f"standardize_specification_design_pic 匹配后工程名 ：result:{match_results}")
+
+        if match_results and len(match_results) == 1:
+            slot_item[PROJECT_NAME] = match_results[0]
+        else:
+            prompt = generate_project_prompt(match_results, original_name=slot_item[PROJECT_NAME], type="工程名")
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
+    return CheckResult.NO_MATCH, ""
+
+
+def standardize_specification_design_pic(slot) -> tuple:
+    # #分公司名标准化
+    # result_type, prompt = standardize_implement_company(slot)
+    # if CheckResult.NEEDS_MORE_ROUNDS == result_type:
+    #     return result_type, prompt
+
+    standard_implement_company = slot[IMPLEMENTATION_ORG] if IMPLEMENTATION_ORG in slot else ""
+
+    # #工程名标准化
+    # result_type, prompt = standardize_project(slot)
+    # if CheckResult.NEEDS_MORE_ROUNDS == result_type:
+    #     return result_type, prompt
+    standard_project = slot[PROJECT_NAME] if PROJECT_NAME in slot else ""
+    simple_design_map = {}
+
+    if DESIGN in slot:
+        value = slot[DESIGN]
+        temp_content = GlobalData.get_contents_by_company_proj(standard_implement_company, standard_project)
+        if temp_content:
+            design_list = temp_content["方案"]
+            simple_design_map.update({
+                clean_useless_design_name(kw): kw for kw in design_list
+            })
+            simply_input = clean_useless_design_name(value)
+            result = fuzzy_match_and_filter(simply_input, list(simple_design_map.keys()), simple_design_map, 70, 90)
+            if result and len(result) == 1:
+                slot[DESIGN] = result[0]
+            else:
+                prompt = generate_project_prompt(result, original_name=slot[DESIGN], type="方案名")
+                print(f"方案名标准化返回：{prompt}")
+                return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        else:
+            prompt = generate_project_prompt([], original_name=slot[DESIGN], type="方案名")
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        return CheckResult.NO_MATCH, ''
+
+    elif SPECIFICATION in slot:
+        value = slot[SPECIFICATION]
+        temp_content = GlobalData.get_contents_by_company_proj(standard_implement_company, standard_project)
+        if temp_content:
+            design_list = temp_content["规范规程"]
+            simple_design_map.update({
+                clean_useless_specification_name(kw): kw for kw in design_list
+            })
+            simply_input = clean_useless_specification_name(value)
+            result = fuzzy_match_and_filter(simply_input, list(simple_design_map.keys()), simple_design_map, 70, 90)
+            if result and len(result) == 1:
+                slot[SPECIFICATION] = result[0]
+            else:
+                prompt = generate_project_prompt(result, original_name=slot[SPECIFICATION], type="规范规程名")
+                print(f"规程规范名标准化返回：{prompt}")
+                return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        else:
+            prompt = generate_project_prompt([], original_name=slot[DESIGN], type="规范规程名")
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        return CheckResult.NO_MATCH, ''
+
+    elif PICTURE in slot:
+        value = slot[PICTURE]
+        temp_content = GlobalData.get_contents_by_company_proj(standard_implement_company, standard_project)
+        if temp_content:
+            design_list = temp_content["图纸"]
+            simple_design_map.update({
+                kw: kw for kw in design_list
+            })
+            result = fuzzy_match_and_filter(value, list(simple_design_map.keys()), simple_design_map, 70, 90)
+            if result and len(result) == 1:
+                slot[PICTURE] = result[0]
+            else:
+                prompt = generate_project_prompt(result, original_name=slot[PICTURE], type="图纸名")
+                print(f"图纸名标准化返回：{prompt}")
+                return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        else:
+            prompt = generate_project_prompt([], original_name=slot[DESIGN], type="规范规程名")
+            return CheckResult.NEEDS_MORE_ROUNDS, prompt
+        return CheckResult.NO_MATCH, ''
+
+    return CheckResult.NO_MATCH, ''
+
+
 def process_msg_content(content):
     if not any(keyword in content for keyword in SLOT_KEYWORDS):
         match = re.search(r"^.*?[。！？.!?：:]", content)
@@ -669,4 +837,10 @@ def process_msg_content(content):
         else:
             return content.strip()
     else:
-        return content
+        return content
+
+
+def is_design_file(file_name):
+    # 清除结尾的引号、空格、标点等
+    text_clean = re.sub(r"[》〉》】】））>）>」』」》））》」』）】\s]+$", "", file_name)
+    return text_clean.endswith(tuple(design_suffix))

generated_data/generated.py

@@ -50,7 +50,8 @@ BASE_DATA = {
         "宿州萧砀线路工程建筑部分",
         "1000kV淮芜Ⅰ线(PROJ-2020-0204-0003)",
         "35kV接地极线路雁淮线",
-        "110kV接地极线路（吉泉线）(PROJ-2020-0204-0002)"
+        "110kV接地极线路（吉泉线）(PROJ-2020-0204-0002)",
+        "国网安徽宣城供电公司500kV河沥变加装固定融冰装置项目工程"
     ],
     # 项目部名称
     "project_departments": ["第一项目部金上","调试一队", "第9项目管理部","第9项目管理部门", "金上第十一项目部门", "第八项目管理部（合肥）", "肥东9号项目部",
@@ -77,10 +78,11 @@ BASE_DATA = {
     "risk_levels": ["1级", "一级", "二级", "5级", "四级"],
     # 8+2工况
     "operatings": ["8+2工况", "8加2工况"],
-    # 页面切换
+    # 页面切换,不能有方案，图纸和规程规范这些数据集出现在pages，否则会冲突
     "pages": ["风险管控", "日计划", "周风险", "日计划统计报表", "日计划推送", "生产管控中心", "考勤统计详情",
               "今日作业计划", "周风险统计报表", "周风险推送", "进度管理", "技术管理", "项目团队", "质量管理",
-              "云上会议", "项目巡航", "施工生产管理平台"],
+              "云上会议", "项目巡航", "施工生产管理平台", "数字化项目部","数字化项目部管理平台","施工生产管理平台",
+              "经营管理", "物资管理", "共享资料", "党建+", "党建加", "摄像头", "视频"],
     # 具体人名
     "person_names": ["何东洋", "李东","王孙强林"],
     # 人名查询目标
@@ -89,10 +91,22 @@ BASE_DATA = {
     # 工程状态
     "project_status_s": ["在建", "在作业", "在施工",""],
 
+    "pic_names": ["四号线施工图", "框架柱详图", "500kvgis室吊车梁布置图", "_站区道路及进站道路详图_A2", "_辅助用房建筑设计说明一_A2","平断面定位图目录",
+                  "基础明细表","杆塔明细表","(500-SJC31151)_1-110 ","接地装置施工图","平断面定位图卷册说明"],
+
+    "design_specification_names": [
+        "《35kV电力电缆交流耐压试验方案》","220kV南蒙2753线拆线、拆塔施工方案","悬索封网实验方案","灌注桩承台基础施工方案"
+        "一般跨越施工措施", "省道专项施工方案","吊车组立角钢塔施工方案","承台基础及接地施工措施","项目管理实施规划","电力电缆方案","线路拆旧跨越110kV线路施工方案",
+        "断面悬浮抱杆组塔施工方案","灌注桩基础及接地施工措施"
+
+        "110kV-750kV架空输电线路铁塔基础施工工艺导则","国网（基建2）112-2022　国家电网有限公司输变电工程建设质量管理规定","1000kV架空输电线路施工质量检验及评定规程","《国家电网有限公司施工项目部标准化管理手册线路工程分册》",
+                            "国家电网有限公司输变电工程标准工艺（电缆工程分册）2022版","架空输电线路螺旋锚基础施工及质量验收规范","国家电网有限公司安全生产反违章工作管理办法"],
+
+
     #皖送天网
-    "sky_nets": ["摄像头", "视频"],
+    # "sky_nets": ["摄像头", "视频"],
     #项目巡航
-    "program_navigations": ["数字化项目部", "数字化项目部管理平台", "施工生产管理平台"],
+    # "program_navigations": ["数字化项目部", "数字化项目部管理平台", "施工生产管理平台"],
 }
 
 # 自然语言模板配置
@@ -748,39 +762,43 @@ TEMPLATE_CONFIG = {
         "date": ["今日", "昨日", "2024年5月24日", "5月24日", "今天", "昨天"],
         "templates": [
             ("打开{page}页面", ["page"]),
-            ("打开{page}", ["page"]),
+            ("打开{page}<页面>", ["page"]),
-            ("切换{page}模块", ["page"]),
-            ("切换到{page}页面", ["page"]),
-            ("跳转到{page}。", ["page"]),
-            ("跳转到{page}模块", ["page"]),
             ("切换到{page}页面", ["page"]),
             ("切换{page}模块", ["page"]),
-            ("请打开{page}模块", ["page"]),
+            ("切换{page}模块", ["page"]),
-            ("请打开{page}。", ["page"]),
+
-            ("请切换到{page}页面", ["page"]),
+            ("打开{implementation_organization}{page}", ["implementation_organization", "page"]),
-            ("切换{page}", ["page"]),
+
-            #施工生产管理平台
+            ("打开{implementation_organization}{project_department}{page}",
-            ("打开{program_navigation}", ["program_navigation"]),
+             ["implementation_organization", "project_department", "page"]),
-            ("打开{program_navigation}。", ["program_navigation"]),
+
-            #项目巡航：分公司
+            ("切换到{implementation_organization}{page}",
-            ("打开{implementation_organization}{program_navigation}", ["implementation_organization", "program_navigation"]),
+             ["implementation_organization", "page"]),
-            ("打开{implementation_organization}{program_navigation}。", ["implementation_organization", "program_navigation"]),
+
-            #项目巡航：分公司、项目部
+            ("打开{project_name}{page}", ["project_name", "page"]),
-            ("打开{implementation_organization}{project_department}{program_navigation}",
+            ("切换到{project_name}{page}", ["project_name", "page"]),
-             ["implementation_organization", "project_department", "program_navigation"]),
+
-            #项目巡航：分公司
+            ("切换到{team_name}{page}", ["team_name", "page"]),
-            ("切换到{implementation_organization}{program_navigation}",
+            ("打开{team_name}{page}", ["team_name", "page"]),
-             ["implementation_organization", "program_navigation"]),
+
-            #项目巡航，工程
+            # design_names,  pic_names,specification_names
-            ("打开{project_name}{program_navigation}", ["project_name", "program_navigation"]),
+            #方案和规程规范
-            #皖智天网，工程名摄像头
+            ("打开{design_specification_name}<方案>", ["project_name", "design_specification_name"]),
-            ("打开{project_name}{sky_net}", ["project_name", "sky_net"]),
+            ("打开{design_specification_name}", ["project_name", "design_specification_name"]),
-            ("切换到{project_name}{sky_net}", ["project_name", "sky_net"]),
+            ("打开{project_name}{design_specification_name}<方案>", ["project_name", "design_specification_name"]),
-            #皖智天网，班组名摄像头
+
-            ("切换到{team_name}{sky_net}", ["team_name", "sky_net"]),
+            ("打开{project_name}的{design_specification_name}", ["project_name", "design_specification_name"]),
-            ("切换{team_name}{sky_net}", ["team_name", "sky_net"]),
+
-            ("打开{team_name}{sky_net}", ["team_name", "sky_net"]),
+            ("打开{implementation_organization}{project_name}的{design_specification_name}",
-            #施工生产管理平台
+             ["implementation_organization", "project_name", "design_specification_name"]),
+
+            #图纸
+            ("打开{pic_name}", ["pic_name"]),
+            ("打开{project_name}{pic_name}", ["project_name", "pic_name"]),
+            ("打开{project_name}{pic_name}<图纸>", ["project_name", "pic_name"]),
+            ("打开{project_name}的{pic_name}", ["project_name", "pic_name"]),
+            ("打开{implementation_organization}{project_name}的{pic_name}",
+             ["implementation_organization", "project_name", "pic_name"]),
         ]
     },
     "作业面查询": {
@@ -1348,8 +1366,11 @@ def generate_natural_samples(config, label):
         "person_name": BASE_DATA["person_names"],
         "person_query_type": BASE_DATA["person_query_types"],
         "project_status": BASE_DATA["project_status_s"],
-        "sky_net": BASE_DATA["sky_nets"],
+        "pic_name": BASE_DATA["pic_names"],
-        "program_navigation": BASE_DATA["program_navigations"],
+        "design_specification_name": BASE_DATA["design_specification_names"],
+
+        # "sky_net": BASE_DATA["sky_nets"],
+        # "program_navigation": BASE_DATA["program_navigations"],
     }
 
     for template, variables in config["templates"]: