import requests import json import pandas as pd import random from itertools import product template_list = [ { # 时间、客户、公司 "keys": ['time', 'customer', 'company'], "sql": """ SELECT `合同名称`, `签订日期`, `客户名称`, `所属分公司` FROM contracts WHERE `所属分公司` LIKE '%{company}%' AND `客户名称` LIKE '%{customer}%' OR `客商类型` LIKE '%{customer}%' OR `合同名称` LIKE '%{customer}%' （AND 根据{time}设置`签订时间`的筛选范围）""", "question": "{time}{company}承接的“客户名称”或“客商类型”或“合同名称”包含{customer}的项目有哪些？参考sql语句：[sql]", }, { # 时间、行业（包括专业） "keys": ['time', 'field'], "sql": """ SELECT `合同名称`, `签订日期`, `聚焦行业` FROM contracts WHERE `合同名称` LIKE '%{field}%' OR `聚焦行业` LIKE '%{field}%' OR `专业` LIKE '%{field}%' （AND 根据{time}设置`签订时间`的筛选范围）""", "question": "{time}承接的“聚焦行业”或“行业”包含{field}的项目有哪些？参考sql语句：[sql]" }, { # 时间、公司、专业 "keys": ['time', 'company', 'major'], "sql": """ SELECT `合同名称`, `签订日期`, `专业`, `所属分公司` FROM contracts WHERE `合同名称` LIKE '%{major}%' OR `专业` LIKE '%{major}%' AND `所属分公司` LIKE '%{company}%' （AND 根据{time}设置`签订时间`的筛选范围）""", "question": "{time}{company}承接的“专业”包含{major}的项目有哪些？参考sql语句：[sql]" }, { # 时间、地点 "keys": ['time', 'area'], "sql": """ SELECT `合同名称`, `地点` FROM contracts WHERE `地点` LIKE '%{area}%' （AND 根据{time}设置`签订时间`的筛选范围）""", "question": "{time}承接的地点在“{area}”的项目有哪些？参考sql语句：[sql]", }, { # 关键字查询（专业、行业、公司、客户、地点） "keys": ['keyword'], "sql": """ SELECT `合同名称`, `所属分公司`, `专业`, `聚焦行业`, `客商类型`, `地点` FROM contracts WHERE `合同名称` LIKE '%{keyword}%' OR `聚焦行业` LIKE '%{keyword}%' OR `所属分公司` LIKE '%{keyword}%' OR `客商类型` LIKE '%{keyword}%' OR `地点` LIKE '%{keyword}%' OR `客户名称` LIKE '%{keyword}%' OR `专业` LIKE '%{keyword}%'""", "question": "“地点”“专业”“聚焦行业”“所属分公司”“客商类型”“客户名称”或“合同名称”包含{keyword}的项目有哪些？参考sql语句：[sql]", }, { # 公司、金额 "keys": ['company', 'money'], "sql": """ SELECT `合同名称`, `所属分公司`, `合同签订金额（人民币）` FROM contracts WHERE `公司` LIKE '%{company}%' （AND 根据{money}筛选`合同签订金额（人民币）`的筛选范围）""", "question": "{company}的合同金额{money}的项目有哪些？参考sql语句：[sql]" }] raw_data = { 'time': list(pd.read_excel("./valueSets.xlsx", sheet_name="时间取值")["时间取值"]), 'major': list(pd.read_excel("./valueSets.xlsx", sheet_name="专业取值")["专业取值"]), 'company': list(pd.read_excel("./valueSets.xlsx", sheet_name="公司取值")["公司取值"]), 'field': list(pd.read_excel("./valueSets.xlsx", sheet_name="行业取值")["行业取值"]), 'customer': list(pd.read_excel("./valueSets.xlsx", sheet_name="客户取值")["客户取值"]), 'area': list(pd.read_excel("./valueSets.xlsx", sheet_name="地区取值")["地区取值"]), 'money': [ "100万以上的", "50万以上的", "2000万以上", "400万以上", "10万以上", "100万到1000万",], # 金额范围取值 } raw_data['keyword'] = [item for key in ['area', 'company', 'customer', 'major', 'field'] for item in raw_data[key]] def combine_key_val(keys: list, k=4) -> dict: """ 计算不同字段的排列组合 Args: keys (_type_): 字段名列表 k (int, optional): 每个字段的取值数量. Defaults to 4. Returns: list: [{key1: value1}]字典列表 """ tmp = {} for key in keys: value = raw_data.get(key, []) # if key == 'company': # tmp[key] = ['一分公司'] # else: # k = len(value) if k > len(value) else k # tmp[key] = random.sample(value, k=k) k = len(value) if k > len(value) else k tmp[key] = random.sample(value, k=k) print(tmp) # 计算笛卡尔积 keys = list(tmp.keys()) values = list(tmp.values()) combinations = list(product(*values)) # 将每个组合转换为字典 return [{key: value for key, value in zip(keys, combination)} for combination in combinations] def form_question(keys, k) -> list: """根据选填字段生成问题 Args: keys (_type_): _description_ Returns: list: _description_ """ combination_dict = combine_key_val(keys, k=k) print(len(combination_dict)) template_idx = 0 for index, item in enumerate(template_list): if set(item['keys']) == set(keys): template_idx = index break template = template_list[template_idx] question_template = template['question'].replace('[sql]', template['sql']) question_list = [] tmp = '' for combination in combination_dict: tmp = question_template tmp = tmp.format_map(combination) question_list.append(tmp) return question_list def test(keys, k=3): question_list = form_question(keys, k=k) try: res = [["问题", "sql", "回答", "thought"]] for question in question_list: if question == "": break payload = {"return_type": "text", "question": question} response = requests.post(url, headers=headers, data=json.dumps(payload)) response_data = response.json() print("data_res", response_data) res.append( [ question, response_data["sql"], response_data["result"], response_data["thought"], ] ) data = pd.DataFrame(res) data.to_excel("./output/{}.xlsx".format('-'.join(keys)), index=False) except Exception as error: print("Error:", error) return "Error" url = "http://localhost:8001/qwen/contracts?apikey=YUVietLgiGmtqzYUVIIGjrNoLMsGM0FI" headers = { "Content-Type": "application/json", # Uncomment the following line if needed and provide the appropriate value for data['apiKey'] # 'Authorization': f'Bearer {data["apiKey"]}' } # # def some_function(): # # print(pdata["问题"]) # ls = pdata["问题"] # qs = [] # tmp = [] # res_q = [] # # for ll in ["数据库字段中的“合同名称”或“聚焦行业”或“专业”包含“key”的项目有哪些?"]: # # for ll in ["time公司承接的money的项目有哪些/多少？今年是2024年。"]: # # for ll in ["time公司承接的customer的项目有哪些/多少？其中，值为“customer”是数据库中“客商类型”字段的部分内容。今年是2024年。"]: # for ll in [ # "company time承接的field的项目有哪些？其中，值为“company”是数据库中“所属分公司”字段的部分内容，“field”是数据库中“合同名称”或“聚焦行业”字段的部分内容。今年是2024年。" # ]: # # for ll in ["time公司承接的field的项目有哪些/多少？其中，值为“field”是数据库中“合同名称”或“聚焦行业”字段的部分内容。"]: # qs = [] # tmp = [] # tmp.append(ll) # if "time" in ll: # for ts in time: # t0 = ll.replace("time", ts) # qs.append(t0) # tmp = qs # qs = [] # for t1 in tmp: # if "major" in t1: # for sp in major: # t0 = t1.replace("major", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # for t1 in tmp: # if "company" in t1: # for sp in company: # t0 = t1.replace("company", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # for t1 in tmp: # if "field" in t1: # for sp in field: # t0 = t1.replace("field", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # for t1 in tmp: # if "key" in t1: # for sp in key: # t0 = t1.replace("key", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # # money # for t1 in tmp: # if "money" in t1: # for sp in money: # t0 = t1.replace("money", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # for t1 in tmp: # if "customer" in t1: # for sp in customer: # t0 = t1.replace("customer", sp) # qs.append(t0) # else: # qs.append(t1) # tmp = qs # qs = [] # for t1 in tmp: # if "area" in t1: # for sp in area: # t0 = t1.replace("area", sp) # qs.append(t0) # else: # qs.append(t1) # # tmp = qs # # qs = [] # # print(len(qs)) # res_q = [*res_q, *qs] # print(len(res_q)) # # print(res_q) # # return # try: # res = [["问题", "sql", "回答", "thought"]] # for ll in res_q: # if ll == "": # break # payload = {"return_type": "text", "question": ll} # response = requests.post(url, headers=headers, data=json.dumps(payload)) # response_data = response.json() # print("data_res", response_data) # res.append( # [ # ll, # response_data["sql"], # response_data["result"], # response_data["thought"], # ] # ) # data = pd.DataFrame(res) # data.to_excel("./行业-时间.xlsx", index=False) # except Exception as error: # print("Error:", error) # return "Error" if __name__ == "__main__": keys_combination_list = [ ['time', 'customer', 'company'], ['time', 'field'], ['time', 'company', 'major'], ['time', 'area'], ['keyword'], ['company', 'money'] ] # print(form_question(keys_combination_list[0])) # form_question(keys_combination_list[2],k=3) # test(keys_combination_list[0], k=3) # test(keys_combination_list[1], k=3) # test(keys_combination_list[2], k=3) # test(keys_combination_list[3], k=3) # test(keys_combination_list[4], k=20) test(keys_combination_list[5], k=3)