You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
181 lines
11 KiB
181 lines
11 KiB
#coding:utf8
|
|
import re
|
|
from jsonpath_ng import parse
|
|
import json
|
|
import traceback
|
|
|
|
def get_content(inputdata,logging):
|
|
"""
|
|
重新组装参数
|
|
:param inputdata:原json数据
|
|
:return: 组装的prompt及其他参数
|
|
"""
|
|
res={}
|
|
admin=inputdata["metadata"]["admin"]
|
|
data=inputdata["data"]
|
|
prompt=admin["prompt"]
|
|
if_user=re.findall("{{(.*)}}",prompt)
|
|
if_data=re.findall("@@(.*?)@@",prompt)
|
|
if if_user != []:
|
|
user_data=inputdata["metadata"]["user"]
|
|
if if_user[0] in user_data.keys():
|
|
tmp=user_data[if_user[0]]
|
|
prompt=re.sub("{{(.*)}}",tmp,prompt)
|
|
if if_data!=[] :
|
|
for rule in if_data:
|
|
try:
|
|
if ':' in rule:
|
|
s=rule.split(':')
|
|
rule1=s[0]
|
|
rule2=s[1]
|
|
if rule1 in data.keys():
|
|
tmp1 = data[rule1]
|
|
#按照给定的规则解析字段
|
|
json_obj = json.loads(tmp1)
|
|
jsonpath_expr = parse(rule2)
|
|
result = str([match.value for match in jsonpath_expr.find(json_obj)][0])
|
|
rep="@@{}@@".format(rule)
|
|
#增加转义字符
|
|
rep_escaped = re.escape(rep)
|
|
prompt=re.sub(rep_escaped,result,prompt)
|
|
else:
|
|
if rule in data.keys():
|
|
tmp1=data[rule]
|
|
rep = "@@{}@@".format(rule)
|
|
prompt=re.sub(rep,tmp1,prompt)
|
|
except:
|
|
rep = "@@{}@@".format(rule)
|
|
prompt = prompt.replace(rep,'')
|
|
logging.info("动态字段获取数据失败。{}-{}".format(rule, traceback.format_exc()))
|
|
logging.info("拼接后的问题：{}".format(prompt))
|
|
res["prompt"]=prompt
|
|
res["authorization"]=admin["authorization"]
|
|
res["model"]=admin["model"]
|
|
res["temperature"]=admin["temperature"]
|
|
res["authorization"]=admin["authorization"]
|
|
res["top_p"]=admin["top_p"]
|
|
res["n"]=admin["n"]
|
|
return res
|
|
|
|
# def get_content(inputdata,logging):
|
|
# """
|
|
# 重新组装参数
|
|
# :param inputdata:原json数据
|
|
# :return: 组装的prompt及其他参数
|
|
# """
|
|
# res={}
|
|
# admin=inputdata["metadata"]["admin"]
|
|
# data=inputdata["data"]
|
|
# prompt=admin["prompt"]
|
|
# if_user=re.findall("{{(.*)}}",prompt)
|
|
# if_data=re.findall("@@(.*)@@",prompt)
|
|
# if if_user != []:
|
|
# user_data=inputdata["metadata"]["user"]
|
|
# if if_user[0] in user_data.keys():
|
|
# tmp=user_data[if_user[0]]
|
|
# prompt=re.sub("{{(.*)}}",tmp,prompt)
|
|
# if if_data!=[] and if_data[0] in data.keys():
|
|
# tmp1=data[if_data[0]]
|
|
# prompt=re.sub("@@(.*)@@",tmp1,prompt)
|
|
# res["prompt"]=prompt
|
|
# res["authorization"]=admin["authorization"]
|
|
# res["model"]=admin["model"]
|
|
# res["temperature"]=admin["temperature"]
|
|
# res["authorization"]=admin["authorization"]
|
|
# res["top_p"]=admin["top_p"]
|
|
# res["n"]=admin["n"]
|
|
# return res
|
|
|
|
if __name__=="__main__":
|
|
|
|
prompt = "用@@11_任务拆分:$.lang@@，生成@@11_任务拆分:$.quantity@@条@@11_任务拆分:$.age@@的@@11_任务拆分:$.sex@@发布@@11_任务拆分:$.emotion@@的@@11_任务拆分:$.subject@@的@@11_任务拆分:$.content_type@@。以JSON数组泛型是String类型的格式进行输出，结构外层需要用“resultList”进行接收，不用多余的文字。"
|
|
|
|
if_data = re.findall("@@(.*?)@@", prompt)
|
|
print(if_data)
|
|
# inputdata={
|
|
# "metadata":{
|
|
# "output":{
|
|
# "output_type":"table",
|
|
# "label_col":[
|
|
# "相似内容抽取"
|
|
# ]
|
|
# },
|
|
# "input":{
|
|
# "input_type":"text",
|
|
# "label":[
|
|
# "3_相似内容过滤器"
|
|
# ]
|
|
# },
|
|
# "address":"http://172.18.1.181:9011/chatGpt/",
|
|
# "admin":{
|
|
# "authorization":"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",
|
|
# "top_p":"1",
|
|
# "user_input":[
|
|
# {
|
|
# "keyname":"tag",
|
|
# "keydesc":""
|
|
# }
|
|
# ],
|
|
# "temperature":"1",
|
|
# "model":"gpt-3.5-turbo-16k",
|
|
# "prompt":"以JSON数组泛型是String类型的格式进行输出，不用多余的文字。参考”@@11_任务拆分:$.quantity@@不仅仅是一种工具，更是一种改变世界的力量“生成@@11_任务拆分:$.quantity@@条@@11_任务拆分:$.lang@@的@@11_任务拆分:$.content_type@@",
|
|
# "n":"1"
|
|
# },
|
|
# "index":3,
|
|
# "user":{
|
|
# "tag":""
|
|
# }
|
|
# },
|
|
# "data":{
|
|
# "10_任务提取":"[{\"age\":\"18~24岁\",\"collection_element\":\"0,1\",\"collection_quantity\":1438,\"collection_task\":\"https://twitter.com/MFA_China\",\"collection_type\":0,\"content_type\":\"发帖内容\",\"create_user_id\":\"652468062228768915\",\"del\":0,\"emotion\":\"积极/乐观\",\"id\":178,\"lang\":\"英语\",\"model_status\":1,\"model_type\":1,\"name\":\"TW用户-发言办公室01-04~12-31\",\"quantity\":10,\"sex\":\"男性\",\"site_id\":181,\"status\":0,\"subject\":\"社会问题和时事主题\",\"tenant_id\":237,\"type\":1},{\"age\":\"18~24岁\",\"collection_element\":\"0,1\",\"collection_quantity\":444378,\"collection_task\":\"hongkong\",\"collection_type\":1,\"content_type\":\"发帖内容\",\"create_user_id\":\"652468062228768915\",\"del\":0,\"emotion\":\"积极/乐观\",\"id\":179,\"lang\":\"英语\",\"model_status\":1,\"model_type\":1,\"name\":\"TW关键词-hongkong01-04~12-31\",\"quantity\":10,\"sex\":\"女性\",\"site_id\":181,\"status\":0,\"subject\":\"旅行和探险主题\",\"tenant_id\":237,\"type\":1},{\"age\":\"18~24岁\",\"collection_element\":\"0,1\",\"collection_quantity\":256,\"collection_task\":\"https://www.facebook.com/tsaiingwen\",\"collection_type\":0,\"content_type\":\"发帖内容\",\"create_user_id\":\"652468062228768915\",\"del\":0,\"emotion\":\"积极/乐观\",\"id\":180,\"lang\":\"英语\",\"model_status\":1,\"model_type\":2,\"name\":\"FB用户-蔡英文-01-04~12-31\",\"quantity\":10,\"sex\":\"男性\",\"site_id\":182,\"status\":0,\"subject\":\"科技和创新主题\",\"tenant_id\":237,\"type\":1},{\"age\":\"18~24岁\",\"collection_element\":\"0,1\",\"collection_quantity\":1253,\"collection_task\":\"台湾新闻\",\"collection_type\":1,\"content_type\":\"发帖内容\",\"create_user_id\":\"652468062228768915\",\"del\":0,\"emotion\":\"积极/乐观\",\"id\":183,\"lang\":\"英语\",\"model_status\":1,\"model_type\":2,\"name\":\"FB关键词-台湾新闻-0110~12-13\",\"quantity\":10,\"sex\":\"女性\",\"site_id\":182,\"status\":0,\"subject\":\"健康和生活方式主题\",\"tenant_id\":237,\"type\":1}]",
|
|
# "3_相似内容过滤器":"{\"age\":\"18~24岁\",\"collection_element\":\"0,1\",\"collection_quantity\":1253,\"collection_task\":\"台湾新闻\",\"collection_type\":1,\"content_type\":\"发帖内容\",\"create_user_id\":\"652468062228768915\",\"del\":0,\"emotion\":\"积极/乐观\",\"id\":183,\"lang\":\"英语\",\"model_status\":1,\"model_type\":2,\"name\":\"FB关键词-台湾新闻-0110~12-13\",\"quantity\":10,\"sex\":\"女性\",\"site_id\":182,\"size\":21,\"status\":0,\"subject\":\"健康和生活方式主题\",\"tenant_id\":237,\"type\":1}",
|
|
# "11_任务拆分":"{\"tenant_id\":237,\"create_user_id\":\"652468062228768915\",\"collection_quantity\":1253,\"quantity\":10,\"subject\":\"健康和生活方式主题\",\"sex\":\"女性\",\"model_type\":2,\"del\":0,\"type\":1,\"collection_element\":\"0,1\",\"collection_type\":1,\"model_status\":1,\"collection_task\":\"台湾新闻\",\"emotion\":\"积极/乐观\",\"content_type\":\"发帖内容\",\"size\":21,\"name\":\"FB关键词-台湾新闻-0110~12-13\",\"site_id\":182,\"id\":183,\"lang\":\"英语\",\"age\":\"18~24岁\",\"status\":0}",
|
|
# "1_mysql数据查询":"{\"resultList\": [{\"id\": 178, \"tenant_id\": 237, \"name\": \"TW用户-发言办公室01-04~12-31\", \"site_id\": 181, \"collection_type\": 0, \"collection_element\": \"0,1\", \"collection_task\": \"https://twitter.com/MFA_China\", \"status\": 0, \"collection_quantity\": 1438, \"create_user\": null, \"create_user_id\": \"652468062228768915\", \"update_user\": null, \"update_user_id\": null, \"del\": 0, \"type\": 1, \"model_type\": 1, \"quantity\": 10, \"content_type\": \"发帖内容\", \"model_status\": 1, \"lang\": \"英语\", \"age\": \"18~24岁\", \"sex\": \"男性\", \"emotion\": \"积极/乐观\", \"subject\": \"社会问题和时事主题\", \"similar_content\": null}, {\"id\": 179, \"tenant_id\": 237, \"name\": \"TW关键词-hongkong01-04~12-31\", \"site_id\": 181, \"collection_type\": 1, \"collection_element\": \"0,1\", \"collection_task\": \"hongkong\", \"status\": 0, \"collection_quantity\": 444378, \"create_user\": null, \"create_user_id\": \"652468062228768915\", \"update_user\": null, \"update_user_id\": null, \"del\": 0, \"type\": 1, \"model_type\": 1, \"quantity\": 10, \"content_type\": \"发帖内容\", \"model_status\": 1, \"lang\": \"英语\", \"age\": \"18~24岁\", \"sex\": \"女性\", \"emotion\": \"积极/乐观\", \"subject\": \"旅行和探险主题\", \"similar_content\": null}, {\"id\": 180, \"tenant_id\": 237, \"name\": \"FB用户-蔡英文-01-04~12-31\", \"site_id\": 182, \"collection_type\": 0, \"collection_element\": \"0,1\", \"collection_task\": \"https://www.facebook.com/tsaiingwen\", \"status\": 0, \"collection_quantity\": 256, \"create_user\": null, \"create_user_id\": \"652468062228768915\", \"update_user\": null, \"update_user_id\": null, \"del\": 0, \"type\": 1, \"model_type\": 2, \"quantity\": 10, \"content_type\": \"发帖内容\", \"model_status\": 1, \"lang\": \"英语\", \"age\": \"18~24岁\", \"sex\": \"男性\", \"emotion\": \"积极/乐观\", \"subject\": \"科技和创新主题\", \"similar_content\": null}, {\"id\": 183, \"tenant_id\": 237, \"name\": \"FB关键词-台湾新闻-0110~12-13\", \"site_id\": 182, \"collection_type\": 1, \"collection_element\": \"0,1\", \"collection_task\": \"台湾新闻\", \"status\": 0, \"collection_quantity\": 1253, \"create_user\": null, \"create_user_id\": \"652468062228768915\", \"update_user\": null, \"update_user_id\": null, \"del\": 0, \"type\": 1, \"model_type\": 2, \"quantity\": 10, \"content_type\": \"发帖内容\", \"model_status\": 1, \"lang\": \"英语\", \"age\": \"18~24岁\", \"sex\": \"女性\", \"emotion\": \"积极/乐观\", \"subject\": \"健康和生活方式主题\", \"similar_content\": null}]}"
|
|
# },
|
|
# "created":1691004265000,
|
|
# "module":"ChatGPT",
|
|
# "start_tag":"false",
|
|
# "multi_branch":0,
|
|
# "last_edit":1693932236000,
|
|
# "next_app_id":[
|
|
# {
|
|
# "start_id":188,
|
|
# "edge_id":92,
|
|
# "end_id":190
|
|
# }
|
|
# ],
|
|
# "transfer_id":5,
|
|
# "version":1,
|
|
# "blueprint_id":6,
|
|
# "scenes_id":7,
|
|
# "scenario":{
|
|
# "dataloss":1,
|
|
# "autoCommitTriggerLast":1,
|
|
# "maxErrors":3,
|
|
# "autoCommit":1,
|
|
# "freshVariables":1
|
|
# },
|
|
# "wait_condition":[
|
|
#
|
|
# ],
|
|
# "scheduling":{
|
|
# "interval":-1,
|
|
# "type":"single"
|
|
# },
|
|
# "name":"相似内容抽取",
|
|
# "id":188,
|
|
# "position":[
|
|
# 100,
|
|
# 200
|
|
# ],
|
|
# "describe":"相似内容抽取"
|
|
# }
|
|
# a=get_content(inputdata,"")
|
|
# print(a)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|