""" 通过本体层文件构建知识图谱 """ from py2neo import Graph, Node, Relationship, NodeMatcher import json import os import logging import re import configparser import glob # 设置日志 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s") logger = logging.getLogger(__name__) # 全局变量 graph = None def read_config(config_file="config.ini"): """ 读取配置文件 Args: config_file: 配置文件路径 Returns: config: 配置对象 """ config = configparser.ConfigParser() config.read(config_file, encoding="utf-8") return config def connect_to_neo4j(uri, user, password): """ 连接到Neo4j数据库 Args: uri: 数据库URI user: 用户名 password: 密码 Returns: graph: 数据库连接对象 """ global graph try: graph = Graph(uri, auth=(user, password)) logger.info("成功连接到Neo4j数据库") return True except Exception as e: logger.error(f"连接Neo4j数据库失败: {e}") return False def clear_database(): """ 清空数据库 """ try: graph.run("MATCH (n) DETACH DELETE n") logger.info("已清空数据库") # 删除所有约束 try: # 获取所有约束 constraints = graph.run("SHOW CONSTRAINTS").data() for constraint in constraints: constraint_name = constraint.get("name") if constraint_name: graph.run(f"DROP CONSTRAINT {constraint_name}") logger.info(f"已删除约束: {constraint_name}") except Exception as e: logger.warning(f"删除约束失败: {e}") return True except Exception as e: logger.error(f"清空数据库失败: {e}") return False # 解析本体层文件，获取实体类型、属性和关系定义 def parse_ontology_file(file_path="Ontology_Layer.txt"): """ 解析本体层文件，获取实体类型、属性和关系定义 Args: file_path: 本体层文件路径 Returns: entity_types: 实体类型和属性的字典 {实体类型名称: {属性名称: 属性类型}} entity_relationships: 实体间关系的列表 [(源实体类型, 关系类型, 目标实体类型)] """ try: with open(file_path, "r", encoding="utf-8") as f: content = f.read() # 分割实体类型和关系部分 parts = content.split("2. 实体间的关系") if len(parts) != 2: logger.error("本体层文件格式错误，无法找到实体类型和关系部分") return {}, [] entity_types_content = parts[0] relationships_content = parts[1] # 解析实体类型和属性 entity_types = {} current_entity = None # 移除 "1.实体类型" 标题行 entity_types_content = entity_types_content.replace("1.实体类型", "").strip() # 按行分割 lines = entity_types_content.strip().split("\n") for line in lines: line = line.strip() if not line: continue # 如果行不包含冒号，视为实体类型 if ":" not in line: current_entity = line entity_types[current_entity] = {} # 否则视为属性定义 else: if current_entity is None: continue parts = line.split(":", 1) if len(parts) == 2: attr_name = parts[0].strip() attr_type = parts[1].strip() entity_types[current_entity][attr_name] = attr_type # 解析实体间的关系 entity_relationships = [] # 匹配关系定义行: (:EntityType)-[:RELATIONSHIP]->(:EntityType) relationship_pattern = r"\(:(\w+)\)-\[:(\w+)\]->\(:(\w+)\)" for line in relationships_content.strip().split("\n"): line = line.strip() if not line: continue match = re.match(relationship_pattern, line) if match: source_entity, relationship_type, target_entity = match.groups() entity_relationships.append((source_entity, relationship_type, target_entity)) logger.info(f"从本体层文件中解析出 {len(entity_types)} 个实体类型和 {len(entity_relationships)} 个关系定义") return entity_types, entity_relationships except Exception as e: logger.error(f"解析本体层文件失败: {e}") return {}, [] # 创建约束和索引以提高性能 - 现在不创建任何约束 def create_constraints_and_indexes(): # 不创建任何约束 logger.info("跳过创建约束") pass # 获取实体之间的关系类型 def get_relationship_type(source_entity_type, target_entity_type, entity_relationships): """ 根据源实体类型和目标实体类型获取关系类型 Args: source_entity_type: 源实体类型 target_entity_type: 目标实体类型 entity_relationships: 实体间关系的列表 Returns: relationship_type: 关系类型，如果没找到则返回None """ for source, rel_type, target in entity_relationships: if source == source_entity_type and target == target_entity_type: return rel_type # 如果没有找到匹配的关系，返回None logger.warning(f"未找到从 {source_entity_type} 到 {target_entity_type} 的关系定义，不创建关系") return None # 创建根节点 def create_root_node(): root = Node("EngineeringData", name="工程") graph.create(root) logger.info("创建根节点: 工程") return root # 处理ProjectDivisionSet def process_project_division_set(data, parent_node, entity_relationships): # 根据您提供的JSON结构，正确访问projectDivision数据 if "projectData" in data and "projectDivision" in data["projectData"]: project_division = data["projectData"]["projectDivision"] elif "projectDivision" in data: project_division = data["projectDivision"] else: logger.warning("JSON中未找到projectDivision数据") logger.info(f"JSON顶层键: {list(data.keys())}") return logger.info(f"开始处理projectDivision，包含 {len(project_division)} 个顶级项目") # 创建新的ProjectDivisionSet节点 - 项目划分集 division_set = Node("ProjectDivisionSet", name="项目划分集") graph.create(division_set) # 从本体层获取关系类型 relationship_type = get_relationship_type("EngineeringData", "ProjectDivisionSet", entity_relationships) if relationship_type: # 只有当关系类型不为None时才创建关系 graph.create(Relationship(parent_node, relationship_type, division_set)) # 处理ProjectDivisionTree for first_level_name, first_level_content in project_division.items(): # 处理第一层下的内容，直接创建合并后的ProjectDivisionTree节点 if isinstance(first_level_content, dict): # 处理一级名称，去掉"工程"字样 processed_first_level = first_level_name.replace("工程", "") for second_level_name, second_level_content in first_level_content.items(): # 确定最终节点名称 if second_level_name == processed_first_level: # 如果二级名称与处理后的一级名称相同，直接使用二级名称 final_name = second_level_name else: # 否则组合二级名称和处理后的一级名称 final_name = f"{second_level_name}{processed_first_level}" # 创建ProjectDivisionTree节点 division_tree = Node("ProjectDivisionTree", name=final_name) # 保存原始名称作为属性 division_tree["original_first_level"] = first_level_name division_tree["original_second_level"] = second_level_name # 如果有GUID，添加到节点属性 guid = None if isinstance(first_level_content, dict) and "GUID" in first_level_content: division_tree["first_level_GUID"] = first_level_content["GUID"] guid = first_level_content["GUID"] graph.create(division_tree) # 从本体层获取关系类型 relationship_type = get_relationship_type( "ProjectDivisionSet", "ProjectDivisionTree", entity_relationships ) if relationship_type: graph.create(Relationship(division_set, relationship_type, division_tree)) # 移除GUID关系建立代码，避免重复创建 # 关系将在establish_relationships函数中批量创建 # 处理第二层下的ProjectDivisionItem列表 if isinstance(second_level_content, list): for item in second_level_content: process_project_division_item(item, division_tree, entity_relationships) else: logger.warning(f"ProjectDivisionTree {final_name} 的内容类型未知: {type(second_level_content)}") else: logger.warning(f"第一层 {first_level_name} 的内容类型未知: {type(first_level_content)}") # 处理ProjectDivisionItem def process_project_division_item(item, parent_node, entity_relationships): # 提取必要属性 guid = item.get("GUID", "") name = item.get("项目名称", "") if not guid and not name: logger.warning("ProjectDivisionItem缺少GUID和项目名称") return # 创建ProjectDivisionItem节点 item_node = Node("ProjectDivisionItem", GUID=guid, name=name) # 添加path属性，表示从ProjectDivisionItem到ProjectDivisionTree的路径，不包含节点类型 if isinstance(parent_node, Node) and "ProjectDivisionTree" in parent_node.labels: # 如果父节点是ProjectDivisionTree，使用"父节点名称/当前节点名称"作为路径 item_node["path"] = f"{parent_node['name']}/{name}" # logger.info(f"为ProjectDivisionItem {name} 设置path: {item_node['path']}") else: # 如果父节点是ProjectDivisionItem，使用"父节点path/当前节点名称"作为路径 parent_path = parent_node.get("path", "") if parent_path: item_node["path"] = f"{parent_path}/{name}" else: # 如果父节点没有path属性（不应该发生，但为了健壮性） item_node["path"] = name # logger.info(f"为ProjectDivisionItem {name} 设置path: {item_node['path']}") # 添加其他属性 for key, value in item.items(): if key not in ["GUID", "项目名称", "children"] and value is not None: # 检查是否为资源库列表 if key == "资源库列表" and isinstance(value, list): # 将资源库列表转换为分号分隔的字符串 resource_names = [] for resource in value: if isinstance(resource, dict) and "资源库名称" in resource: resource_names.append(resource["资源库名称"]) item_node["资源库名称"] = "；".join(resource_names) # logger.info(f"将资源库列表转换为字符串: {item_node['资源库名称']}") # 检查值是否为基本类型 elif isinstance(value, (str, int, float, bool)): item_node[key] = value # 如果是列表，尝试转换为分号分隔的字符串 elif isinstance(value, list): try: if all(isinstance(x, (str, int, float, bool)) for x in value): item_node[key] = "；".join(str(x) for x in value) else: # 对于包含复杂对象的列表，尝试提取关键信息 extracted_values = [] for item_in_list in value: if isinstance(item_in_list, dict): # 尝试提取字典中的名称或标识符 for name_key in ["名称", "name", "标识", "id", "ID"]: if name_key in item_in_list: extracted_values.append(str(item_in_list[name_key])) break else: # 如果没有找到名称键，使用第一个键值对 if item_in_list: first_key = next(iter(item_in_list)) extracted_values.append(f"{first_key}:{item_in_list[first_key]}") else: extracted_values.append(str(item_in_list)) item_node[key] = "；".join(extracted_values) except Exception as e: logger.warning(f"无法将列表属性 {key} 转换为字符串: {e}") # 如果是字典，尝试转换为字符串 elif isinstance(value, dict): try: # 提取字典中的关键信息 extracted_info = [] for dict_key, dict_value in value.items(): if isinstance(dict_value, (str, int, float, bool)): extracted_info.append(f"{dict_key}:{dict_value}") item_node[key] = "；".join(extracted_info) except Exception as e: logger.warning(f"无法将字典属性 {key} 转换为字符串: {e}") graph.create(item_node) # logger.info(f"创建ProjectDivisionItem节点: {name} (GUID: {guid})") # 创建与父节点的关系 if isinstance(parent_node, Node) and "ProjectDivisionTree" in parent_node.labels: # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectDivisionTree", "ProjectDivisionItem", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, item_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {item_node['name']}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {item_node['name']}") else: # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectDivisionItem", "ProjectDivisionItem", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, item_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {item_node['name']}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {item_node['name']}") # 移除GUID关系建立代码，避免重复创建 # 关系将在establish_relationships函数中批量创建 # 处理子项 if "children" in item and item["children"]: children = item["children"] # logger.info(f"ProjectDivisionItem {name} 有 {len(children)} 个子项") for child in children: child_type = child.get("type", child.get("类型", "")) if child_type == "项目划分": # 递归处理子ProjectDivisionItem process_project_division_item(child, item_node, entity_relationships) elif child_type == "8" or child_type == "清单": # 处理List类型节点 process_list_item(child, item_node, entity_relationships) else: # 处理ProjectQuantity及其子类 process_project_quantity(child, item_node, entity_relationships) # 处理List及其子类 def process_list_item(list_item, parent_node, entity_relationships): """处理清单类型的节点""" # 提取必要属性 guid = list_item.get("GUID", "") list_name = list_item.get("清单名称") list_type = list_item.get("类型", "") # 创建List节点 list_node = Node("List", guid=guid, name=list_name, type=list_type) # 添加path属性，包含节点类型 parent_path = parent_node.get("path", "") if parent_path: list_node["path"] = f"{parent_path}/{list_name}(清单)" else: # 如果父节点没有path属性（不应该发生，但为了健壮性） parent_name = parent_node.get("name", "") list_node["path"] = f"{parent_name}/{list_name}(清单)" # logger.info(f"为List节点 {list_name} 设置path: {list_node['path']}") # 添加其他属性 for key, value in list_item.items(): if key not in ["清单名称", "类型", "guid", "children"] and value is not None: # 检查是否为资源库列表 if key == "资源库列表" and isinstance(value, list): # 将资源库列表转换为分号分隔的字符串 resource_names = [] for resource in value: if isinstance(resource, dict) and "资源库名称" in resource: resource_names.append(resource["资源库名称"]) list_node["资源库名称"] = "；".join(resource_names) # logger.info(f"将资源库列表转换为字符串: {list_node['资源库名称']}") # 检查值是否为基本类型 elif isinstance(value, (str, int, float, bool)): list_node[key] = value # 如果是列表，尝试转换为分号分隔的字符串 elif isinstance(value, list): try: if all(isinstance(x, (str, int, float, bool)) for x in value): list_node[key] = "；".join(str(x) for x in value) else: # 对于包含复杂对象的列表，尝试提取关键信息 extracted_values = [] for item_in_list in value: if isinstance(item_in_list, dict): # 尝试提取字典中的名称或标识符 for name_key in ["名称", "name", "标识", "id", "ID"]: if name_key in item_in_list: extracted_values.append(str(item_in_list[name_key])) break else: # 如果没有找到名称键，使用第一个键值对 if item_in_list: first_key = next(iter(item_in_list)) extracted_values.append(f"{first_key}:{item_in_list[first_key]}") else: extracted_values.append(str(item_in_list)) list_node[key] = "；".join(extracted_values) except Exception as e: logger.warning(f"无法将列表属性 {key} 转换为字符串: {e}") # 如果是字典，尝试转换为字符串 elif isinstance(value, dict): try: # 提取字典中的关键信息 extracted_info = [] for dict_key, dict_value in value.items(): if isinstance(dict_value, (str, int, float, bool)): extracted_info.append(f"{dict_key}:{dict_value}") list_node[key] = "；".join(extracted_info) except Exception as e: logger.warning(f"无法将字典属性 {key} 转换为字符串: {e}") graph.create(list_node) # logger.info(f"创建List节点: {list_name} (类型: {list_type})") # 创建与父节点的关系 # 从本体层获取关系类型 relationship_type = get_relationship_type(list(parent_node.labels)[0], "List", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, list_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {list_name}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {list_name}") # 如果有GUID，尝试建立与CostSet的关系 if guid: # 查找对应的CostSet节点 cost_set_query = f""" MATCH (c:CostSet) WHERE c.GUID = '{guid}' RETURN c """ cost_set_nodes = list(graph.run(cost_set_query)) if cost_set_nodes: cost_set_node = cost_set_nodes[0]["c"] relationship_type = get_relationship_type("List", "CostSet", entity_relationships) if relationship_type: graph.create(Relationship(list_node, relationship_type, cost_set_node)) # logger.info(f"创建关系: {list_name} {relationship_type} CostSet (GUID: {guid})") # else: # logger.info(f"不创建关系: {list_name} 到 CostSet (GUID: {guid})") # 处理子项 if "children" in list_item and list_item["children"]: children = list_item["children"] # logger.info(f"List {list_name} 有 {len(children)} 个子项") for child in children: # 确定子项类型 child_type = child.get("type", child.get("类型", "")) if child_type == "项目划分": # 递归处理子ProjectDivisionItem process_project_division_item(child, list_node, entity_relationships) elif child_type == "8" or child_type == "清单": # 递归处理子List process_list_item(child, list_node, entity_relationships) else: # 处理ProjectQuantity及其子类 process_project_quantity(child, list_node, entity_relationships) # 处理ProjectQuantity及其子类 def process_project_quantity(quantity, parent_node, entity_relationships): # 确定具体类型 quantity_type = quantity.get("类型", "") labels = ["ProjectQuantity"] type_name = "ProjectQuantity" # 支持数字和文本类型 if quantity_type == "0" or quantity_type == "定额": labels.append("Quota") type_name = "定额" elif quantity_type == "1" or quantity_type == "主材": labels.append("MainMaterial") type_name = "主材" elif quantity_type == "5" or quantity_type == "设备": labels.append("Equipment") type_name = "设备" # 创建节点 quantity_id = quantity.get("id", "") quantity_name = quantity.get("项目名称", quantity.get("名称", "")) quantity_node = Node(*labels, id=quantity_id, name=quantity_name) # 添加path属性，包含节点类型 parent_path = parent_node.get("path", "") if parent_path: quantity_node["path"] = f"{parent_path}/{quantity_name}({type_name})" else: # 如果父节点没有path属性（不应该发生，但为了健壮性） parent_name = parent_node.get("name", "") quantity_node["path"] = f"{parent_name}/{quantity_name}({type_name})" # logger.info(f"为ProjectQuantity节点 {quantity_name} 设置path: {quantity_node['path']}") # 添加其他属性 for key, value in quantity.items(): if key not in ["id", "名称", "项目名称", "材机列表", "children"] and value is not None: # 检查是否为资源库列表 if key == "资源库列表" and isinstance(value, list): # 将资源库列表转换为分号分隔的字符串 resource_names = [] for resource in value: if isinstance(resource, dict) and "资源库名称" in resource: resource_names.append(resource["资源库名称"]) quantity_node["资源库名称"] = "；".join(resource_names) # logger.info(f"将资源库列表转换为字符串: {quantity_node['资源库名称']}") # 检查值是否为基本类型 elif isinstance(value, (str, int, float, bool)): quantity_node[key] = value # 如果是列表，尝试转换为分号分隔的字符串 elif isinstance(value, list): try: if all(isinstance(x, (str, int, float, bool)) for x in value): quantity_node[key] = "；".join(str(x) for x in value) else: # 对于包含复杂对象的列表，尝试提取关键信息 extracted_values = [] for item_in_list in value: if isinstance(item_in_list, dict): # 尝试提取字典中的名称或标识符 for name_key in ["名称", "name", "标识", "id", "ID"]: if name_key in item_in_list: extracted_values.append(str(item_in_list[name_key])) break else: # 如果没有找到名称键，使用第一个键值对 if item_in_list: first_key = next(iter(item_in_list)) extracted_values.append(f"{first_key}:{item_in_list[first_key]}") else: extracted_values.append(str(item_in_list)) quantity_node[key] = "；".join(extracted_values) except Exception as e: logger.warning(f"无法将列表属性 {key} 转换为字符串: {e}") # 如果是字典，尝试转换为字符串 elif isinstance(value, dict): try: # 提取字典中的关键信息 extracted_info = [] for dict_key, dict_value in value.items(): if isinstance(dict_value, (str, int, float, bool)): extracted_info.append(f"{dict_key}:{dict_value}") quantity_node[key] = "；".join(extracted_info) except Exception as e: logger.warning(f"无法将字典属性 {key} 转换为字符串: {e}") graph.create(quantity_node) # logger.info(f"创建ProjectQuantity节点: {quantity_name} (id: {quantity_id}, 类型: {quantity_type})") # 创建与父节点的关系 # 从本体层获取关系类型 parent_label = list(parent_node.labels)[0] # 获取父节点的第一个标签 relationship_type = get_relationship_type(parent_label, "ProjectQuantity", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, quantity_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {quantity_name}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {quantity_name}") # 移除GUID关系建立代码，避免重复创建 # 关系将在establish_relationships函数中批量创建 # 处理材机列表或children materials = None # 先检查是否有材机列表 if "材机列表" in quantity and quantity["材机列表"]: materials = quantity["材机列表"] # logger.info(f"ProjectQuantity {quantity_name} 有 {len(materials)} 个材机项") for material in materials: process_material_or_equipment(material, quantity_node, entity_relationships) # 如果没有材机列表，则检查是否有children elif "children" in quantity and quantity["children"]: children = quantity["children"] # logger.info(f"ProjectQuantity {quantity_name} 有 {len(children)} 个子项") for child in children: child_type = child.get("类型", child.get("type", "")) # 如果子项类型为人工、材料或机械，则视为MaterialOrEquipment if child_type in ["人工", "材料", "机械", "2", "3", "4"]: process_material_or_equipment(child, quantity_node, entity_relationships) # 如果子项类型为主材、设备或定额，则递归处理为ProjectQuantity elif child_type in ["1", "主材", "5", "设备", "0", "定额"]: process_project_quantity(child, quantity_node, entity_relationships) # 处理MaterialOrEquipment def process_material_or_equipment(material, parent_node, entity_relationships): material_id = material.get("id", material.get("ID", "")) material_name = material.get("名称", "") material_type = material.get("类型", material.get("type", "")) if not material_id and not material_name: logger.warning("MaterialOrEquipment缺少id和名称") return # 创建唯一标识，结合父节点的ID和当前项的id parent_id = parent_node.get("id", parent_node.get("GUID", "")) unique_id = f"{parent_id}_{material_id}" if parent_id else material_id # 直接创建新节点，不检查是否已存在 material_node = Node( "MaterialOrEquipment", id=material_id, unique_id=unique_id, name=material_name, type=material_type ) # 添加其他属性 for key, value in material.items(): if key not in ["id", "ID", "名称", "类型", "type"] and value is not None: # 检查是否为资源库列表 if key == "资源库列表" and isinstance(value, list): # 将资源库列表转换为分号分隔的字符串 resource_names = [] for resource in value: if isinstance(resource, dict) and "资源库名称" in resource: resource_names.append(resource["资源库名称"]) material_node["资源库名称"] = "；".join(resource_names) # logger.info(f"将资源库列表转换为字符串: {material_node['资源库名称']}") # 检查值是否为基本类型 elif isinstance(value, (str, int, float, bool)): material_node[key] = value # 如果是列表，尝试转换为分号分隔的字符串 elif isinstance(value, list): try: if all(isinstance(x, (str, int, float, bool)) for x in value): material_node[key] = "；".join(str(x) for x in value) else: # 对于包含复杂对象的列表，尝试提取关键信息 extracted_values = [] for item_in_list in value: if isinstance(item_in_list, dict): # 尝试提取字典中的名称或标识符 for name_key in ["名称", "name", "标识", "id", "ID"]: if name_key in item_in_list: extracted_values.append(str(item_in_list[name_key])) break else: # 如果没有找到名称键，使用第一个键值对 if item_in_list: first_key = next(iter(item_in_list)) extracted_values.append(f"{first_key}:{item_in_list[first_key]}") else: extracted_values.append(str(item_in_list)) material_node[key] = "；".join(extracted_values) except Exception as e: logger.warning(f"无法将列表属性 {key} 转换为字符串: {e}") # 如果是字典，尝试转换为字符串 elif isinstance(value, dict): try: # 提取字典中的关键信息 extracted_info = [] for dict_key, dict_value in value.items(): if isinstance(dict_value, (str, int, float, bool)): extracted_info.append(f"{dict_key}:{dict_value}") material_node[key] = "；".join(extracted_info) except Exception as e: logger.warning(f"无法将字典属性 {key} 转换为字符串: {e}") graph.create(material_node) # logger.info(f"创建MaterialOrEquipment节点: {material_name} (id: {material_id}, 类型: {material_type})") # 创建与父节点的关系 # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectQuantity", "MaterialOrEquipment", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, material_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {material_name}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {material_name}") # 处理CostSet def process_cost_set(data, root_node, entity_relationships): """处理费用预览，为每个具有GUID的节点创建CostSet节点""" # 根据JSON结构，访问expensePreview数据 if "projectData" in data and "expensePreview" in data["projectData"]: expense_preview = data["projectData"]["expensePreview"] elif "expensePreview" in data: expense_preview = data["expensePreview"] else: logger.warning("JSON中未找到expensePreview数据") logger.info(f"JSON顶层键: {list(data.keys())}") return # logger.info("开始处理expensePreview") # 用于跟踪已处理的GUID，避免重复处理 processed_guids = set() # 递归函数，用于遍历expensePreview中的所有节点 def traverse_expense_preview(node, path=""): if isinstance(node, dict): # 如果节点包含GUID，则创建CostSet节点 if "GUID" in node: guid = node["GUID"] # 标准化GUID格式 guid = guid.strip("{}") guid = "{" + guid.upper() + "}" if guid not in processed_guids: processed_guids.add(guid) # 创建CostSet节点 cost_set_node = Node("CostSet", GUID=guid, name="费用预览集") try: graph.create(cost_set_node) # logger.info(f"创建CostSet节点: 费用预览集 (GUID: {guid}, 路径: {path})") except Exception as e: logger.error(f"创建CostSet节点失败: {e}") return # 创建与根节点的关系 relationship_type = get_relationship_type("EngineeringData", "CostSet", entity_relationships) if relationship_type: try: graph.create(Relationship(root_node, relationship_type, cost_set_node)) # logger.info(f"创建关系: {root_node['name']} {relationship_type} 费用预览集") except Exception as e: logger.error(f"创建关系失败: {e}") # 处理sum数组中的CostItem if "sum" in node and isinstance(node["sum"], list) and node["sum"]: for item in node["sum"]: try: process_cost_item(item, cost_set_node, entity_relationships) except Exception as e: logger.error(f"处理CostItem时出错: {e}") # 处理rcj数组中的MaterialandmachineCostItem if "rcj" in node and isinstance(node["rcj"], list) and node["rcj"]: for item in node["rcj"]: try: process_material_machine_cost_item(item, cost_set_node, entity_relationships) except Exception as e: logger.error(f"处理MaterialandmachineCostItem时出错: {e}") # 递归处理子节点 for key, value in node.items(): if key == "children" and isinstance(value, list): for i, child in enumerate(value): traverse_expense_preview(child, f"{path}.children[{i}]") elif isinstance(node, list): for i, item in enumerate(node): traverse_expense_preview(item, f"{path}[{i}]") # 遍历expensePreview中的所有类别 for category_name, category in expense_preview.items(): for subcategory_name, subcategory in category.items(): if isinstance(subcategory, list): for i, item in enumerate(subcategory): traverse_expense_preview(item, f"expensePreview.{category_name}.{subcategory_name}[{i}]") # logger.info(f"共处理了 {len(processed_guids)} 个CostSet节点") # 处理CostItem - 简化处理逻辑，确保正确创建节点 def process_cost_item(item, parent_node, entity_relationships): # 确保item是字典 if not isinstance(item, dict): logger.error(f"CostItem不是字典类型: {item}") return item_id = item.get("id", "") cost = item.get("cost", "") if not item_id: logger.warning(f"跳过没有id的CostItem: {item}") return # 使用id作为名称 name = item_id # 创建唯一标识，结合父节点的GUID和当前项的id parent_guid = parent_node.get("GUID", "") unique_id = f"{parent_guid}_{item_id}" if parent_guid else item_id # 创建CostItem节点 - 简化属性，只保留关键属性 item_node = Node("CostItem", id=item_id, unique_id=unique_id, cost=cost, name=name) # 添加其他属性 - 简化属性处理 for key, value in item.items(): if key not in ["id", "cost"] and value is not None: if isinstance(value, (str, int, float, bool)): item_node[key] = value graph.create(item_node) # logger.info(f"创建CostItem节点: {name} (id: {item_id}, cost: {cost})") # 创建与父节点的关系 relationship_type = get_relationship_type("CostSet", "CostItem", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, item_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {name}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {name}") # 处理人材机合价项(MaterialandmachineCostItem) def process_material_machine_cost_item(item, parent_node, entity_relationships): """处理人材机合价项""" # 提取必要属性 item_type = item.get("type", "") name = item.get("名称", "") code = item.get("编码", "") unit = item.get("单位", "") supplier = item.get("供货方", "") budget_price = item.get("预算价不含税", "") market_price = item.get("市场价不含税", "") budget_total = item.get("预算价合价", "") market_total = item.get("市场价合价", "") price_diff = item.get("价差", "") quantity = item.get("数量", "") if not name: logger.warning("MaterialandmachineCostItem缺少名称") return # 创建唯一标识，结合父节点的GUID和当前项的编码 parent_guid = parent_node.get("GUID", "") unique_id = f"{parent_guid}_{code}" if parent_guid and code else (parent_guid or code) # 创建人材机合价项节点，确保所有属性值都是原始类型 properties = {"type": str(item_type), "name": str(name), "unique_id": str(unique_id)} # 添加其他属性，确保都是字符串类型 if supplier: properties["供货方"] = str(supplier) if code: properties["编码"] = str(code) if unit: properties["单位"] = str(unit) if budget_price: properties["预算价不含税"] = str(budget_price) if market_price: properties["市场价不含税"] = str(market_price) if budget_total: properties["预算价合价"] = str(budget_total) if market_total: properties["市场价合价"] = str(market_total) if price_diff: properties["价差"] = str(price_diff) if quantity: properties["数量"] = str(quantity) # 创建节点 item_node = Node("MaterialandmachineCostItem", **properties) try: graph.create(item_node) # logger.info(f"创建MaterialandmachineCostItem节点: {name} (类型: {item_type}, 编码: {code})") except Exception as e: logger.error(f"创建MaterialandmachineCostItem节点失败: {e}") # 打印详细的节点属性，帮助调试 for key, value in properties.items(): logger.error(f"属性 {key}: {type(value)} = {value}") return # 创建与父节点的关系 relationship_type = get_relationship_type("CostSet", "MaterialandmachineCostItem", entity_relationships) if relationship_type: try: graph.create(Relationship(parent_node, relationship_type, item_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {name}") except Exception as e: logger.error(f"创建关系失败: {e}") else: logger.info(f"不创建关系: {parent_node['name']} 到 {name}") # 修改establish_relationships函数，添加项目划分与人材机合价集合的关系 def establish_relationships(entity_relationships): # 首先检查数据库中的节点情况 logger.info("检查数据库中的节点情况...") # 检查ProjectDivisionItem节点 pdi_query = """ MATCH (pdi:ProjectDivisionItem) RETURN count(pdi) as count, collect(distinct pdi.GUID)[..10] as sample_guids """ pdi_result = graph.run(pdi_query).data()[0] logger.info(f"数据库中有 {pdi_result['count']} 个ProjectDivisionItem节点") logger.info(f"ProjectDivisionItem节点GUID样本: {pdi_result['sample_guids']}") # 检查ProjectQuantity节点 pq_query = """ MATCH (pq:ProjectQuantity) RETURN count(pq) as count, collect(distinct pq.id)[..10] as sample_ids """ pq_result = graph.run(pq_query).data()[0] logger.info(f"数据库中有 {pq_result['count']} 个ProjectQuantity节点") logger.info(f"ProjectQuantity节点ID样本: {pq_result['sample_ids']}") # 检查CostSet节点 cs_query = """ MATCH (cs:CostSet) RETURN count(cs) as count, collect(distinct cs.GUID)[..10] as sample_guids """ cs_result = graph.run(cs_query).data()[0] logger.info(f"数据库中有 {cs_result['count']} 个CostSet节点") logger.info(f"CostSet节点GUID样本: {cs_result['sample_guids']}") # 获取从ProjectDivisionItem到CostSet的关系类型 relationship_type = get_relationship_type("ProjectDivisionItem", "CostSet", entity_relationships) if relationship_type: # 只有当关系类型不为None时才创建关系 # 建立ProjectDivisionItem与CostSet的关系，使用更宽松的匹配条件，并避免重复创建 query_division_item = f""" MATCH (pdi:ProjectDivisionItem), (cs:CostSet) WHERE toUpper(replace(replace(pdi.GUID, '{{', ''), '}}', '')) = toUpper(replace(replace(cs.GUID, '{{', ''), '}}', '')) AND pdi.GUID <> "" AND NOT EXISTS((pdi)-[:{relationship_type}]->(cs)) CREATE (pdi)-[:{relationship_type}]->(cs) RETURN count(*) as count """ try: result = graph.run(query_division_item) count = result.data()[0]["count"] logger.info(f"创建了 {count} 个 ProjectDivisionItem {relationship_type} CostSet 关系") except Exception as e: logger.error(f"创建ProjectDivisionItem与CostSet关系失败: {e}") else: logger.info("本体层中未定义ProjectDivisionItem到CostSet的关系，跳过创建") # 获取从ProjectQuantity到CostSet的关系类型 relationship_type = get_relationship_type("ProjectQuantity", "CostSet", entity_relationships) if relationship_type: # 只有当关系类型不为None时才创建关系 # 建立ProjectQuantity与CostSet的关系，使用更宽松的匹配条件，并避免重复创建 query_quantity = f""" MATCH (pq:ProjectQuantity), (cs:CostSet) WHERE toUpper(replace(replace(pq.GUID, '{{', ''), '}}', '')) = toUpper(replace(replace(cs.GUID, '{{', ''), '}}', '')) AND pq.GUID <> "" AND NOT EXISTS((pq)-[:{relationship_type}]->(cs)) CREATE (pq)-[:{relationship_type}]->(cs) RETURN count(*) as count """ try: result = graph.run(query_quantity) count = result.data()[0]["count"] logger.info(f"创建了 {count} 个 ProjectQuantity {relationship_type} CostSet 关系") except Exception as e: logger.error(f"创建ProjectQuantity与CostSet关系失败: {e}") else: logger.info("本体层中未定义ProjectQuantity到CostSet的关系，跳过创建") # 检查最终的关系数量 # 检查ProjectDivisionItem到CostSet的关系 pdi_cs_query = """ MATCH (pdi:ProjectDivisionItem)-[r]->(cs:CostSet) RETURN count(r) as count """ pdi_cs_result = graph.run(pdi_cs_query).data()[0] logger.info(f"数据库中最终有 {pdi_cs_result['count']} 个ProjectDivisionItem到CostSet的关系") # 检查ProjectQuantity到CostSet的关系 pq_cs_query = """ MATCH (pq:ProjectQuantity)-[r]->(cs:CostSet) RETURN count(r) as count """ pq_cs_result = graph.run(pq_cs_query).data()[0] logger.info(f"数据库中最终有 {pq_cs_result['count']} 个ProjectQuantity到CostSet的关系") # 处理取费表模板集(FeeTableTemplateSet) def process_fee_table_template_set(data, parent_node, entity_relationships): # 根据JSON结构，访问costSetting数据 if "projectData" in data and "costSetting" in data["projectData"]: cost_setting = data["projectData"]["costSetting"] elif "costSetting" in data: cost_setting = data["costSetting"] else: logger.warning("JSON中未找到costSetting数据") logger.info(f"JSON顶层键: {list(data.keys())}") return # logger.info(f"开始处理costSetting，包含 {len(cost_setting)} 个取费表模板集") # 只创建一个取费表模板集节点 fee_template_set_node = Node("FeeTableTemplateSet", name="取费表模板集") graph.create(fee_template_set_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("EngineeringData", "FeeTableTemplateSet", entity_relationships) graph.create(Relationship(parent_node, relationship_type, fee_template_set_node)) # logger.info("创建FeeTableTemplateSet节点: 取费表模板集") # 处理每个取费表模板集 for template_set_name, template_set_content in cost_setting.items(): # 直接处理取费表模板项 if "tables" in template_set_content and isinstance(template_set_content["tables"], list): for template_item in template_set_content["tables"]: process_fee_table_template_item(template_item, fee_template_set_node, entity_relationships) # 处理取费表模板项(FeeTableTemplateItem) def process_fee_table_template_item(template_item, parent_node, entity_relationships): # 提取必要属性 name = template_item.get("name", "") outlay_id = template_item.get("OutlayID", "") type_name = template_item.get("类型", "") profession = template_item.get("专业", "") if not name: logger.warning("FeeTableTemplateItem缺少name") return # 创建取费表模板项节点 template_item_node = Node( "FeeTableTemplateItem", name=name, outlayID=outlay_id, type=type_name, profession=profession ) graph.create(template_item_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("FeeTableTemplateSet", "FeeTableTemplateItem", entity_relationships) graph.create(Relationship(parent_node, relationship_type, template_item_node)) # logger.info(f"创建FeeTableTemplateItem节点: {name} (OutlayID: {outlay_id})") # 处理取费项 if "children" in template_item and isinstance(template_item["children"], list): for fee_item in template_item["children"]: process_fee(fee_item, template_item_node, entity_relationships) # 处理取费(FeeCollection) def process_fee(fee_item, parent_node, entity_relationships): # 提取必要属性 serial_number = fee_item.get("序号", "") fee_name = fee_item.get("费用名称", "") code = fee_item.get("代码", "") rate = fee_item.get("费率(%)", "") base = fee_item.get("取费基数", "") remark = fee_item.get("备注", "") if not fee_name: logger.warning("Fee缺少费用名称") return # 创建取费节点 fee_node = Node("FeeCollection", serialNumber=serial_number, name=fee_name, code=code) # 添加path属性 if "FeeCollection" in parent_node.labels: # 如果父节点是FeeCollection节点，使用父节点的path加上当前节点名称 parent_path = parent_node.get("path", "") fee_node["path"] = f"{parent_path}/{fee_name}" else: # 如果父节点是FeeTableTemplateItem，直接使用父节点名称作为路径的开始 parent_name = parent_node.get("name", "") fee_node["path"] = f"{parent_name}/{fee_name}" # logger.info(f"为FeeCollection节点 {fee_name} 设置path: {fee_node['path']}") # 添加其他属性 if rate: fee_node["rate"] = rate if base: fee_node["base"] = base if remark: fee_node["remark"] = remark graph.create(fee_node) # 从本体层获取关系类型 parent_label = list(parent_node.labels)[0] relationship_type = get_relationship_type(parent_label, "FeeCollection", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, fee_node)) # logger.info(f"创建Fee节点: {fee_name} (序号: {serial_number}, 代码: {code})") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {fee_name}") # 处理子费用项 if "children" in fee_item and isinstance(fee_item["children"], list): for child_fee in fee_item["children"]: process_fee(child_fee, fee_node, entity_relationships) # 处理费用表集(FeeScheduleSet) def process_fee_schedule_set(data, parent_node, entity_relationships): """处理费用表集、费用表项和费用""" # 检查projectCost是否存在 if "projectData" in data and "projectCost" in data["projectData"]: project_cost = data["projectData"]["projectCost"] elif "projectCost" in data: project_cost = data["projectCost"] else: logger.warning("JSON中未找到projectCost数据") logger.info(f"JSON顶层键: {list(data.keys())}") return # logger.info(f"开始处理projectCost，包含 {len(project_cost)} 个费用表项") # 创建FeeScheduleSet节点 - 工程费用 fee_schedule_set = Node("FeeScheduleSet", name="工程费用") graph.create(fee_schedule_set) # 从本体层获取关系类型 relationship_type = get_relationship_type("EngineeringData", "FeeScheduleSet", entity_relationships) graph.create(Relationship(parent_node, relationship_type, fee_schedule_set)) # logger.info(f"创建FeeScheduleSet节点: 工程费用") # 处理费用表集下的费用表项 for fee_table_name, fee_table_content in project_cost.items(): # 创建FeeScheduleItem节点 fee_schedule_item = Node("FeeScheduleItem", name=fee_table_name) graph.create(fee_schedule_item) # 从本体层获取关系类型 relationship_type = get_relationship_type("FeeScheduleSet", "FeeScheduleItem", entity_relationships) graph.create(Relationship(fee_schedule_set, relationship_type, fee_schedule_item)) # logger.info(f"创建FeeScheduleItem节点: {fee_table_name}") # 处理费用表项下的费用列表 if isinstance(fee_table_content, list): # logger.info(f"FeeScheduleItem {fee_table_name} 包含 {len(fee_table_content)} 个费用项") for fee_item in fee_table_content: process_fee_item(fee_item, fee_schedule_item, entity_relationships) else: logger.warning(f"FeeScheduleItem {fee_table_name} 的内容类型未知: {type(fee_table_content)}") # 处理费用项(Fee) def process_fee_item(fee, parent_node, entity_relationships): """处理费用项""" # 提取必要属性 serial_number = fee.get("序号", "") name = fee.get("费用名称", "") code = fee.get("代码", "") rate = fee.get("费率(%)", "") amount = fee.get("金额", "") if not name: logger.warning("Fee缺少费用名称") return # 创建Fee节点 fee_node = Node("Fee", serialNumber=serial_number, name=name, code=code) # 添加path属性 if "Fee" in parent_node.labels: # 如果父节点是Fee节点，使用父节点的path加上当前节点名称 parent_path = parent_node.get("path", "") fee_node["path"] = f"{parent_path}/{name}" else: # 如果父节点是FeeScheduleItem，直接使用父节点名称作为路径的开始 parent_name = parent_node.get("name", "") fee_node["path"] = f"{parent_name}/{name}" # logger.info(f"为Fee节点 {name} 设置path: {fee_node['path']}") # 添加其他属性 if rate: fee_node["rate"] = rate if amount: fee_node["amount"] = amount # 添加其他属性 for key, value in fee.items(): if key not in ["序号", "费用名称", "代码", "费率(%)", "金额", "children"] and value is not None: if isinstance(value, (str, int, float, bool)): fee_node[key] = value graph.create(fee_node) # logger.info(f"创建Fee节点: {name} (序号: {serial_number})") # 创建与父节点的关系 # 从本体层获取关系类型 parent_label = list(parent_node.labels)[0] relationship_type = get_relationship_type(parent_label, "Fee", entity_relationships) if relationship_type: graph.create(Relationship(parent_node, relationship_type, fee_node)) # logger.info(f"创建关系: {parent_node['name']} {relationship_type} {name}") # else: # logger.info(f"不创建关系: {parent_node['name']} 到 {name}") # 处理子费用项 if "children" in fee and fee["children"]: children = fee["children"] # logger.info(f"Fee {name} 有 {len(children)} 个子费用项") for child in children: process_fee_item(child, fee_node, entity_relationships) # 处理工程属性集(ProjectPropertySet)和工程属性(ProjectProperty) def process_project_property_set(data, parent_node, entity_relationships): # 检查projectInfo是否存在 if "projectData" in data and "projectInfo" in data["projectData"]: project_info = data["projectData"]["projectInfo"] elif "projectInfo" in data: project_info = data["projectInfo"] else: logger.warning("JSON中未找到projectInfo数据") logger.info(f"JSON顶层键: {list(data.keys())}") return # logger.info("开始处理projectInfo") # 创建工程属性集节点 property_set_node = Node("ProjectPropertySet", name="工程属性") graph.create(property_set_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("EngineeringData", "ProjectPropertySet", entity_relationships) graph.create(Relationship(parent_node, relationship_type, property_set_node)) # logger.info("创建ProjectPropertySet节点: 工程属性") # 为每个属性创建单独的节点 for key, value in project_info.items(): if value is not None: # 处理复杂类型的值 if isinstance(value, list): # 对于列表类型，创建一个包含列表摘要的节点 property_node = Node("ProjectProperty", name=key, value=f"列表({len(value)}项)") graph.create(property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectPropertySet", "ProjectProperty", entity_relationships) graph.create(Relationship(property_set_node, relationship_type, property_node)) # logger.info(f"创建ProjectProperty节点: {key} = 列表({len(value)}项)") # 为列表中的每个项创建子节点 for i, item in enumerate(value): if isinstance(item, dict): # 对于字典类型的列表项，创建包含键值对的节点 for sub_key, sub_value in item.items(): if sub_value is not None: sub_property_node = Node( "ProjectProperty", name=f"{key}[{i}].{sub_key}", value=str(sub_value) ) graph.create(sub_property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type( "ProjectProperty", "ProjectProperty", entity_relationships ) graph.create(Relationship(property_node, relationship_type, sub_property_node)) # logger.info(f"创建ProjectProperty子节点: {key}[{i}].{sub_key} = {sub_value}") else: # 对于基本类型的列表项，创建简单节点 sub_property_node = Node("ProjectProperty", name=f"{key}[{i}]", value=str(item)) graph.create(sub_property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type( "ProjectProperty", "ProjectProperty", entity_relationships ) graph.create(Relationship(property_node, relationship_type, sub_property_node)) # logger.info(f"创建ProjectProperty子节点: {key}[{i}] = {item}") elif isinstance(value, dict): # 对于字典类型，创建一个包含字典摘要的节点 property_node = Node("ProjectProperty", name=key, value=f"字典({len(value)}项)") graph.create(property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectPropertySet", "ProjectProperty", entity_relationships) graph.create(Relationship(property_set_node, relationship_type, property_node)) # logger.info(f"创建ProjectProperty节点: {key} = 字典({len(value)}项)") # 为字典中的每个键值对创建子节点 for sub_key, sub_value in value.items(): if sub_value is not None: sub_property_node = Node("ProjectProperty", name=f"{key}.{sub_key}", value=str(sub_value)) graph.create(sub_property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type( "ProjectProperty", "ProjectProperty", entity_relationships ) graph.create(Relationship(property_node, relationship_type, sub_property_node)) # logger.info(f"创建ProjectProperty子节点: {key}.{sub_key} = {sub_value}") else: # 对于基本类型，直接创建节点 property_node = Node("ProjectProperty", name=key, value=str(value)) graph.create(property_node) # 从本体层获取关系类型 relationship_type = get_relationship_type("ProjectPropertySet", "ProjectProperty", entity_relationships) graph.create(Relationship(property_set_node, relationship_type, property_node)) # logger.info(f"创建ProjectProperty节点: {key} = {value}") # 将创建知识图谱的功能封装为函数 def create_KG(json_file_path, ontology_file_path="Ontology_Layer.txt"): """ 创建知识图谱 Args: json_file_path: JSON文件路径 ontology_file_path: 本体层文件路径 Returns: success: 是否成功创建知识图谱 """ try: # 解析本体层文件 entity_types, entity_relationships = parse_ontology_file(ontology_file_path) if not entity_types or not entity_relationships: logger.error("解析本体层文件失败") return False # 读取JSON文件 try: with open(json_file_path, "r", encoding="utf-8") as f: data = json.load(f) except Exception as e: logger.error(f"读取JSON文件失败: {e}") return False # 获取文件名作为工程名称 file_name = os.path.basename(json_file_path) project_name = os.path.splitext(file_name)[0] # 为每个JSON文件创建一个新的EngineeringData节点 root_node = Node("EngineeringData", name=project_name) graph.create(root_node) logger.info(f"创建根节点: {project_name}") # 处理费用预览，创建CostSet节点及其子节点 process_cost_set(data, root_node, entity_relationships) # 处理项目划分，创建ProjectDivisionSet和ProjectDivisionItem节点 process_project_division_set(data, root_node, entity_relationships) # 处理取费表模板集 process_fee_table_template_set(data, root_node, entity_relationships) # 处理费用表集 process_fee_schedule_set(data, root_node, entity_relationships) # 处理工程属性集 process_project_property_set(data, root_node, entity_relationships) # 建立实体间的关系 establish_relationships(entity_relationships) logger.info(f"成功创建知识图谱: {json_file_path}") return True except Exception as e: logger.error(f"创建知识图谱失败: {e}") import traceback traceback.print_exc() return False # 添加一个新的函数，用于处理文件夹中的多个JSON文件 def create_KGs_from_folder(input_folder, ontology_file_path="Ontology_Layer.txt"): """ 从文件夹中创建多个知识图谱 Args: input_folder: 输入文件夹路径，包含多个JSON文件 ontology_file_path: 本体层文件路径 Returns: success_count: 成功处理的文件数量 total_count: 总文件数量 """ # 获取输入文件夹中的所有JSON文件 json_files = [] # 如果输入是文件夹，则获取文件夹中的所有JSON文件 if os.path.isdir(input_folder): json_files = glob.glob(os.path.join(input_folder, "*.json")) # 如果输入是文件，则直接使用该文件 elif os.path.isfile(input_folder) and input_folder.endswith(".json"): json_files = [input_folder] else: logger.error(f"输入路径无效: {input_folder}") return 0, 0 if not json_files: logger.error(f"未找到JSON文件: {input_folder}") return 0, 0 # 连接到Neo4j数据库 config = read_config() if not connect_to_neo4j(config["neo4j"]["uri"], config["neo4j"]["user"], config["neo4j"]["password"]): return 0, 0 # 处理每个JSON文件 success_count = 0 for json_file in json_files: logger.info(f"处理文件: {json_file}") if create_KG(json_file, ontology_file_path): success_count += 1 logger.info(f"知识图谱构建完成，成功处理 {success_count}/{len(json_files)} 个文件") return success_count, len(json_files) # 修改main函数，使用命令行参数 def main(): import argparse parser = argparse.ArgumentParser(description="构建知识图谱") parser.add_argument("--input", "-i", default="final_outputs", help="输入文件夹或文件路径") parser.add_argument("--ontology", "-o", default="Ontology_Layer.txt", help="本体层文件路径") parser.add_argument("--clear", "-c", action="store_true", help="是否清空数据库") args = parser.parse_args() # 读取配置文件 config = read_config() # 连接到Neo4j数据库 if not connect_to_neo4j(config["neo4j"]["uri"], config["neo4j"]["user"], config["neo4j"]["password"]): return # 如果指定了清空数据库，则清空 if args.clear: if not clear_database(): return # 处理输入文件夹或文件 success_count, total_count = create_KGs_from_folder(args.input, args.ontology) logger.info(f"知识图谱构建完成，成功处理 {success_count}/{total_count} 个文件") if __name__ == "__main__": main()