from bs4 import BeautifulSoup from configs.connections import create_or_get_node, relationship_exists, connect_graph, get_node from openai import OpenAI from py2neo import Relationship import ast import logging logger = logging.getLogger(__name__) from flask import current_app api_key = "sk-86d4622141d74e9a8d7c38ee873c4d91" base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1" model_name = "qwen-turbo" # 提取共有标签 def llm_client(content): try: if content is None or content == []: return [] client = OpenAI(api_key=api_key, base_url=base_url, ) response = client.chat.completions.create( model=model_name, messages=[ {"role": "system", "content": "你是一个专业的人力资源经理，根据用户的提示提取标签"}, {"role": "user", "content": f"请提取以下内容的特征，精简概要，能够对所有求职者的特征进行概括的词语，并按顺序返回结果。输出是列表格式，最多返回5个标签。" f"例如，如果输入是 '英语过了六级,两年以上iOS开发经验,沟通协商能力强,熟练使用Python,有项目管理经验'，" f"输出应该是 ['英语六级', '2年iOS开发经验', '熟练使用Python', '项目管理经验']。" f"不要使用宽泛的词语，如 '为企业着想'、'肯干'等类似的词语，只提取具体的、关键的特征能够描述大部分人的特征。" f"根据不同的行业特点，提取最相关的特征。例如，对于技术岗位，关注技术技能；对于销售岗位，关注销售业绩和客户关系管理等。" f"不要使用宽泛的词语，只提取具体的、关键的特征。内容如下：{content}"}, ], max_tokens=1024, temperature=0.1, stream=False ) result = response.choices[0].message.content temp = result.replace("'", '"') result = ast.literal_eval(temp) return result except Exception as e: current_app.logger.error(f'llm_client error: {e}') return [] def bs_data(data): try: soup = BeautifulSoup(data,'html.parser') text = soup.get_text() return text except Exception as e: current_app.logger.error(f'bs_data error: {e}') return "" def create_job_dataList(dataList): try: ''' 拼接招聘要求文本，送给LLM，给出标签，建立关系 :param dataList: :return: ''' for item in dataList: value_list = [ item['type'], item['expType'], item['eduType'], f"{item['payFrom']}到{item['payTo']}每{item['payUnit']}", ] + item["tagList"]+llm_client(bs_data(item['content']+item['requirement'])) # 招聘职位 job = create_or_get_node('job', uniqueId=item['id'], name=item['name']) # 标签 for label in value_list: all_label = create_or_get_node('jobLabel', name=label) if not relationship_exists(job, 'connection', all_label): connection = Relationship(job, 'connection', all_label) connect_graph.create(connection) except Exception as e: current_app.logger.error(f'create_job_dataList error: {e}') return str(e) def create_enterprise_dataList(dataList): try: ''' 拼接招聘要求文本，送给LLM，给出标签，建立关系 :param dataList: :return: ''' for item in dataList: if item['businessResp'] is None: business_scope = [] else: business_scope = item['businessResp']['businessScope'] \ if item['businessResp']['businessScope'] is not None else [] value_list = [ item['financingStatus'], item['scale'], item['workTime'], ] + item["welfareList"] if item['welfareList'] is not None else []\ + item["tagList"] if item['tagList'] is not None else [] \ +llm_client(item['introduce'] if item['introduce'] is not None else [])\ +llm_client(business_scope) # 招聘职位 enterprise = create_or_get_node('enterprise', uniqueId=item['id'], name=item['name'],alias = item['anotherName']) # 标签 for label in value_list: all_label = create_or_get_node('enterpriseLabel', name=label) if not relationship_exists(enterprise, 'connection', all_label): connection = Relationship(enterprise, 'connection', all_label) connect_graph.create(connection) except Exception as e: current_app.logger.error(f'create_enterprise_dataList error: {e}') return str(e) # 全部新增 def create_seeker_dataList(dataList): try: ''' 拼接招聘要求文本，送给LLM，给出标签，建立关系 :param dataList: :return: ''' for item in dataList: if item['person'] is None: person_jobType = '' jobStatus = '' expType = '' eduType = '' advantage = '' sex = '' else: person_jobType = item['person'].get('jobType', '') jobStatus = item['person'].get('jobStatus', '') expType = item['person'].get('expType', '') eduType = item['person'].get('eduType', '') advantage = item['person'].get('advantage', '') sex = item['person'].get('sex', '') # 确保 advantage 是一个字符串 advantage_str = advantage if isinstance(advantage, str) else (', '.join(advantage) if isinstance(advantage, list) else '') interestedList = item.get('interestedList', []) edu_list = item.get('eduList', []) workList = item.get('workList', []) projectList = item.get('projectList', []) trainList = item.get('trainList', []) data = [] content_list = [] for record in edu_list: content_list.append(record.get('content', '')) data.append(record.get('schoolName', '')) data.append(record.get('educationType', '')) data.append(record.get('educationSystemType', '')) for record in interestedList: data.append(record['jobType']) for record in workList: data.append(record.get('positionName', '')) content_list.append(record.get('content', '')) for record in projectList: data.append(record.get('name', '')) content_list.append(record.get('content', '')) for record in trainList: data.append(record.get('orgName', '')) data.append(record.get('course', '')) content_list.append(record.get('content', '')) content_list = [str(content) for content in content_list if content is not None] data.extend([str(label) for label in llm_client("\n".join(content_list)) if label is not None]) advantage_labels = [str(label) for label in llm_client(advantage_str) if label is not None] if advantage_str else [] value_list = [ person_jobType, jobStatus, expType, eduType, sex ]+ advantage_labels + data # 剔除value_list为空的值 value_list = [x for x in value_list if x is not None and x != ""] # 求职者 seeker = create_or_get_node('seeker', uniqueId=item['person']['userId'], name=item['person']['name']) # 标签 for label in value_list: all_label = create_or_get_node('seekerLabel', name=label) if not relationship_exists(seeker, 'connection', all_label): connection = Relationship(seeker, 'connection', all_label) connect_graph.create(connection) except Exception as e: current_app.logger.error(f'create_seeker_dataList error: {e}') return str(e) # 单个新增 def add_seeker_dataList(dataList): try: for record in dataList: seeker = create_or_get_node('seeker', uniqueId=record['userId'], name=record['name']) value_list = [ record['jobType'], record['jobStatus'], record['expType'], record['eduType'], record['sex'] ] + llm_client( record['advantage']) # 标签 for label in value_list: all_label = create_or_get_node('seekerLabel', name=label) if not relationship_exists(seeker, 'connection', all_label): connection = Relationship(seeker, 'connection', all_label) connect_graph.create(connection) except Exception as e: current_app.logger.error(f'add_seeker_dataList error: {e}') return str(e) # 临时新增 def part_seeker_dataList(dataList): try: for record in dataList: seeker = create_or_get_node('seeker', uniqueId=record['person']['id'], name=record['person']['name']) value_list = [ record['person']['jobType'], record['person']['jobStatus'], record['person']['expType'], record['person']['eduType'], record['person']['sex'] ] + llm_client(record['person']['advantage']) # 标签 for label in value_list: if label and label.strip(): # 检查标签是否非空且非空白字符串 all_label = create_or_get_node('seekerLabel', name=label) if not relationship_exists(seeker, 'connection', all_label): connection = Relationship(seeker, 'connection', all_label) connect_graph.create(connection) except Exception as e: current_app.logger.error(f'add_seeker_dataList error: {e}') return str(e) # 单个修改 def update_seeker_dataList(dataList): pass