增加openai服务支持，销售分析可以使用openai或zhipu
This commit is contained in:
parent
81018ff378
commit
21db91ee71
|
|
@ -48,4 +48,41 @@ curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/retrive/stream \
|
|||
-d '{
|
||||
"message":"视睿电子教学课件系统续费项目更新日志：**项目进展描述**：了解到客户需求和降本要求后，与项目经理杨建线下沟通。客户同意在剩余2套系统上增加模块，但要求降价，具体数 量待内部讨论。",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
||||
|
||||
curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/analysis/stream \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"message":"分析商机广汽汇理汽车金融有限公司的商机建议",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
||||
curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/analysis/stream \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"message":"2. 更新我负责的广汽汇理汽车金融有限公司项目的最新动作:1，今日和客户有做了一次技术交流，他们最近和Ocean Base和Gold DB也做了交流，以及内部也做了沟通，接下来他们希望能够拿出一个业务场景做测试，已确定哪个产品更适合他们。",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
||||
|
||||
curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/analysis/stream \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"message":"更新我负责的广汽汇理汽车金融有限公司项目的最新动作:已经完成了POC，客户对POC效果表示满意",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
||||
curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/analysis/stream \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"message":"openai",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
||||
curl -N -X POST http://127.0.0.1:5002/api/v2/zhipu/analysis/stream \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"message":"zhipu",
|
||||
"knowledge_id":"1843318172036575232"
|
||||
}'
|
||||
|
|
@ -19,7 +19,7 @@ def stream_sse():
|
|||
message = data.get('message', '')
|
||||
|
||||
def event_stream():
|
||||
for chunk in zhipu_service.talk_to_zhipu_sse(message):
|
||||
for chunk in zhipu_service.generate_response_sse(message):
|
||||
if chunk:
|
||||
yield chunk
|
||||
|
||||
|
|
@ -30,7 +30,7 @@ def non_stream():
|
|||
data = request.json
|
||||
message = data.get('message', '')
|
||||
|
||||
response = zhipu_service.talk_to_zhipu(message)
|
||||
response = zhipu_service.generate_response(message)
|
||||
print(f'response: {response}')
|
||||
return response
|
||||
|
||||
|
|
@ -188,7 +188,7 @@ def analysis_stream():
|
|||
"""
|
||||
|
||||
def event_stream():
|
||||
for chunk in zhipu_service.talk_to_zhipu_sse(prompt_analysis):
|
||||
for chunk in zhipu_service.generate_response_sse(prompt_analysis):
|
||||
if chunk:
|
||||
yield chunk
|
||||
|
||||
|
|
|
|||
|
|
@ -4,12 +4,18 @@ from flask import Blueprint, request, Response,session
|
|||
from app.services.zhipu_service import ZhipuService
|
||||
from app.services.zhipu_alltool_service import ZhipuAlltoolService
|
||||
from app.services.zhipu_file_service import ZhipuFileService
|
||||
from app.services.zhipu_kb_service import ZhipuKbService
|
||||
from app.services.openai_service import OpenaiService
|
||||
from app.utils.prompt_repository import PromptRepository # Add this import
|
||||
from app.utils.sessions import init_session
|
||||
|
||||
zhipu_controller_v2 = Blueprint('zhipu_controller_v2', __name__)
|
||||
|
||||
zhipu_service = ZhipuService()
|
||||
openai_service = OpenaiService()
|
||||
zhipu_alltool_service = ZhipuAlltoolService()
|
||||
zhipu_file_service = ZhipuFileService()
|
||||
zhipu_kb_service = ZhipuKbService()
|
||||
|
||||
response_headers = {'Content-Type': 'text/event-stream',
|
||||
'Cache-Control': 'no-cache',
|
||||
|
|
@ -66,7 +72,7 @@ def retrive_stream():
|
|||
|
||||
def event_stream_retrive():
|
||||
accumulated_result = ""
|
||||
for chunk in zhipu_service.retrive_sse(message, knowledge_id, system_prompt="你是一个销售助理，语言对话请以第一人称你我进行"):
|
||||
for chunk in zhipu_kb_service.retrive_sse(message, knowledge_id, system_prompt="你是一个销售助理，语言对话请以第一人称你我进行"):
|
||||
if chunk:
|
||||
accumulated_result += chunk
|
||||
chunk_out = format_chunk(chunk, None, None)
|
||||
|
|
@ -92,7 +98,7 @@ def retrive_stream():
|
|||
prompt_report_template = PromptRepository().get_prompt("report_template")
|
||||
prompt_report_title = f"根据用户提问中\"\"\" {message} \"\"\" 中提到的项目信息 在知识库中查找该项目的销售日志。如果销售日志中缺乏模板中的要点（时间，参与人，事件，获得信息，信息来源，项目进展描述）信息，则该要点内容留空，不要填充信息 日报模板： \"\"\" {prompt_report_template} \"\"\"。输出: 日志报告"
|
||||
generated_report = ""
|
||||
for chunk in zhipu_service.retrive_sse(prompt_report_title + message, knowledge_id, None):
|
||||
for chunk in zhipu_kb_service.retrive_sse(prompt_report_title + message, knowledge_id, None):
|
||||
if chunk:
|
||||
print(chunk)
|
||||
generated_report += chunk
|
||||
|
|
@ -123,7 +129,7 @@ def retrive_stream():
|
|||
prompt_report_template = PromptRepository().get_prompt("report_template")
|
||||
prompt_report_title = f"根据用户提问中\"\"\" {message} \"\"\" 中提到的项目信息 在知识库中查找该项目的销售日志并结合用户提供的新的日志信息 \"\"\"{message} \"\"\"生成日报。如果销售日志中缺乏模板中的要点（时间，参与人，事件，获得信息，信息来源，项目进展描述）信息，则该要点内容留空，不要填充信息 日报模板： \"\"\" {prompt_report_template} \"\"\"。输出: 日志报告"
|
||||
generated_report = ""
|
||||
for chunk in zhipu_service.retrive_sse(prompt_report_title + message, knowledge_id, None):
|
||||
for chunk in zhipu_kb_service.retrive_sse(prompt_report_title + message, knowledge_id, None):
|
||||
if chunk:
|
||||
print(chunk)
|
||||
generated_report += chunk
|
||||
|
|
@ -142,11 +148,32 @@ def retrive_stream():
|
|||
|
||||
@zhipu_controller_v2.route('/zhipu/analysis/stream', methods=['POST'])
|
||||
def analysis_stream():
|
||||
init_session()
|
||||
data = request.json
|
||||
message = data.get('message', '')
|
||||
knowledge_id = data.get('knowledge_id', '')
|
||||
message = message.replace("我", "我（徐春峰）")
|
||||
|
||||
if 'zhipu' in message.lower() or '智谱' in message:
|
||||
logger.info(f'switch to zhipu service, save to session')
|
||||
session['llm_service'] = 'zhipu'
|
||||
return format_chunk("切换到智谱AI服务", None, None)
|
||||
if 'openai' in message.lower() or 'openai' in message:
|
||||
logger.info(f'switch to openai service, save to session')
|
||||
session['llm_service'] = 'openai'
|
||||
return format_chunk("切换到openai服务", None, None)
|
||||
# 默认使用智谱AI服务
|
||||
llm_service = zhipu_service
|
||||
logger.info(f'llm_service: {session["llm_service"]}')
|
||||
current_service = session.get('llm_service', 'zhipu') # Default to 'zhipu' if not set
|
||||
|
||||
if current_service == 'openai':
|
||||
logger.info('Using OpenAI service')
|
||||
llm_service = openai_service
|
||||
else:
|
||||
logger.info('Using Zhipu service')
|
||||
llm_service = zhipu_service
|
||||
|
||||
logger.info(f'/zhipu/analysis/stream v2: {message}')
|
||||
|
||||
intent_categories =["analyze_sales","provide_sales_update_info"]
|
||||
|
|
@ -158,6 +185,11 @@ def analysis_stream():
|
|||
if '更新' in message and '我负责' in message and '项目' in message and '最新' in message:
|
||||
classification_result = {"category":"provide_sales_update_info"}
|
||||
|
||||
# if 'openai' in message.lower():
|
||||
# logger.info(f'switch to openai service')
|
||||
# llm_service = openai_service
|
||||
# message = message.replace("openai", "")
|
||||
|
||||
|
||||
additional_business_info = ""
|
||||
if classification_result.get('category') == 'analyze_sales':
|
||||
|
|
@ -182,7 +214,7 @@ def analysis_stream():
|
|||
请根据用户提供的如下信息，查找相关的 '当前详细状态及已完成工作','Sales stage' 信息，并返回给用户:
|
||||
{message}
|
||||
"""
|
||||
business_info = zhipu_service.retrive(prompt_get_business_info, knowledge_id, None)
|
||||
business_info = zhipu_kb_service.retrive(prompt_get_business_info, knowledge_id, None)
|
||||
logger.info(f'business_info: {business_info}')
|
||||
|
||||
analysis_rule = PromptRepository().get_prompt('sales_analysis')
|
||||
|
|
@ -220,7 +252,7 @@ def analysis_stream():
|
|||
|
||||
def event_stream():
|
||||
accumulated_result = ""
|
||||
for chunk in zhipu_service.talk_to_zhipu_sse(prompt_analysis):
|
||||
for chunk in llm_service.generate_response_sse(prompt_analysis):
|
||||
if chunk:
|
||||
accumulated_result += chunk
|
||||
chunk_out = format_chunk(chunk, None, None)
|
||||
|
|
|
|||
|
|
@ -0,0 +1,11 @@
|
|||
from abc import ABC, abstractmethod
|
||||
|
||||
class AIServiceInterface(ABC):
|
||||
@abstractmethod
|
||||
def generate_response(self, prompt):
|
||||
pass
|
||||
|
||||
@abstractmethod
|
||||
def generate_response_sse(self, prompt):
|
||||
pass
|
||||
|
||||
|
|
@ -0,0 +1,61 @@
|
|||
import logging
|
||||
import time
|
||||
from openai import OpenAI
|
||||
from app.services.ai_service_interface import AIServiceInterface
|
||||
# from app.utils.prompt_repository import PromptRepository
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
class OpenaiService(AIServiceInterface):
|
||||
def __init__(self):
|
||||
self.client = OpenAI(base_url="https://ai.xorbit.link:8443/e5b2a5e5-b41d-4715-9d50-d4a3b0c1a85f/v1", api_key="sk-proj-e5b2a5e5b41d47159d50d4a3b0c1a85f")
|
||||
|
||||
def generate_response(self, prompt):
|
||||
try:
|
||||
response = self.client.chat.completions.create(
|
||||
model="gpt-4o-mini",
|
||||
messages=[{"role": "system", "content": prompt}],
|
||||
temperature=0.7,
|
||||
)
|
||||
return response.choices[0].message.content
|
||||
except Exception as e:
|
||||
logger.error(f"Error generating response: {e}")
|
||||
return "An error occurred while generating the response."
|
||||
|
||||
def generate_response_sse(self, prompt):
|
||||
try:
|
||||
response = self.client.chat.completions.create(
|
||||
model="gpt-4o-mini",
|
||||
messages=[{"role": "system", "content": prompt}],
|
||||
temperature=0.7,
|
||||
stream=True
|
||||
)
|
||||
for chunk in response:
|
||||
if chunk.choices[0].delta.content is not None:
|
||||
yield chunk.choices[0].delta.content
|
||||
yield "answer provided by openai"
|
||||
except Exception as e:
|
||||
logger.error(f"Error generating SSE response: {e}")
|
||||
yield "An error occurred while generating the SSE response."
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
# Set up logging
|
||||
logging.basicConfig(level=logging.INFO)
|
||||
|
||||
# Create an instance of OpenaiService
|
||||
openai_service = OpenaiService()
|
||||
|
||||
# Test the generate_response method
|
||||
test_prompt = "What is the capital of France?"
|
||||
response = openai_service.generate_response(test_prompt)
|
||||
print(f"Response to '{test_prompt}': {response}")
|
||||
|
||||
# Test the generate_response_sse method
|
||||
print("\nTesting generate_response_sse:")
|
||||
sse_prompt = "Count from 1 to 5 slowly."
|
||||
for chunk in openai_service.generate_response_sse(sse_prompt):
|
||||
print(chunk, end='', flush=True)
|
||||
time.sleep(0.1) # Add a small delay to simulate streaming
|
||||
print("\nSSE response complete.")
|
||||
|
||||
|
|
@ -0,0 +1,83 @@
|
|||
import logging
|
||||
import time
|
||||
from zhipuai import ZhipuAI
|
||||
from app.utils.prompt_repository import PromptRepository
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
class ZhipuKbService:
|
||||
def __init__(self):
|
||||
self.model_name = "glm-4"
|
||||
self.app_secret_key = "d54f764a1d67c17d857bd3983b772016.GRjowY0fyiMNurLc"
|
||||
logger.info("ZhipuKbService initialized with model: %s", self.model_name)
|
||||
|
||||
|
||||
def retrive(self, message, knowledge_id, prompt_template):
|
||||
logger.info("Starting retrive call with knowledge_id: %s", knowledge_id)
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
default_prompt = "从文档\n\"\"\"\n{{knowledge}}\n\"\"\"\n中找问题\n\"\"\"\n{{question}}\n\"\"\"\n的答案，找到答案就仅使用文档语句回答问题，找不到答案就用自身知识回答并且告诉用户该信息不是来自文档。\n不要复述问题，直接开始回答。"
|
||||
|
||||
if prompt_template is None or prompt_template == "":
|
||||
prompt_template = default_prompt
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
model="glm-4",
|
||||
messages=[
|
||||
{"role": "user", "content": message},
|
||||
],
|
||||
tools=[
|
||||
{
|
||||
"type": "retrieval",
|
||||
"retrieval": {
|
||||
"knowledge_id": knowledge_id,
|
||||
"prompt_template": prompt_template
|
||||
}
|
||||
}
|
||||
],
|
||||
stream=False,
|
||||
temperature=0.01,
|
||||
top_p=0.1,
|
||||
)
|
||||
result = response.choices[0].message.content
|
||||
end_time = time.time()
|
||||
logger.info("retrive call completed in %.2f seconds", end_time - start_time)
|
||||
return result
|
||||
except Exception as e:
|
||||
logger.error("Error in retrive: %s", str(e))
|
||||
raise
|
||||
|
||||
def retrive_sse(self, message, knowledge_id, prompt_template=None,system_prompt=None):
|
||||
logger.info("Starting retrive_sse call with knowledge_id: %s, message:%s", knowledge_id, message)
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
default_prompt = "从文档\n\"\"\"\n{{knowledge}}\n\"\"\"\n中找问题\n\"\"\"\n{{question}}\n\"\"\"\n的答案，找到答案就仅使用文档语句回答问题，找不到答案就告诉用户知识库中没有该信息。\n不要复述问题，直接开始回答。"
|
||||
messages = [{"role": "user", "content": message}]
|
||||
# if system_prompt != None:
|
||||
# messages.append({"role": "system", "content": system_prompt})
|
||||
if prompt_template is None or prompt_template == "":
|
||||
prompt_template = default_prompt
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
model="glm-4",
|
||||
messages=messages,
|
||||
tools=[
|
||||
{
|
||||
"type": "retrieval",
|
||||
"retrieval": {
|
||||
"knowledge_id": knowledge_id,
|
||||
"prompt_template": prompt_template
|
||||
}
|
||||
}
|
||||
],
|
||||
stream=True,
|
||||
temperature=0.01,
|
||||
top_p=0.1,
|
||||
)
|
||||
for chunk in response:
|
||||
yield chunk.choices[0].delta.content
|
||||
end_time = time.time()
|
||||
logger.info("retrive_sse call completed in %.2f seconds", end_time - start_time)
|
||||
except Exception as e:
|
||||
logger.error("Error in retrive_sse: %s", str(e))
|
||||
raise
|
||||
|
|
@ -1,47 +1,47 @@
|
|||
import logging
|
||||
import time
|
||||
from zhipuai import ZhipuAI
|
||||
from app.services.ai_service_interface import AIServiceInterface
|
||||
from app.utils.prompt_repository import PromptRepository
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
class ZhipuService:
|
||||
class ZhipuService(AIServiceInterface):
|
||||
def __init__(self):
|
||||
self.model_name = "glm-4"
|
||||
self.app_secret_key = "d54f764a1d67c17d857bd3983b772016.GRjowY0fyiMNurLc"
|
||||
self.client = ZhipuAI(api_key=self.app_secret_key)
|
||||
logger.info("ZhipuService initialized with model: %s", self.model_name)
|
||||
|
||||
def talk_to_zhipu(self, message):
|
||||
logger.info("Starting talk_to_zhipu call")
|
||||
def generate_response(self, prompt):
|
||||
logger.info("Starting generate_response call")
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
response = self.client.chat.completions.create(
|
||||
model=self.model_name,
|
||||
messages=[
|
||||
{"role": "user", "content": message},
|
||||
{"role": "user", "content": prompt},
|
||||
],
|
||||
stream=False,
|
||||
temperature=0.01,
|
||||
top_p=0.1,
|
||||
)
|
||||
accum_resp = response.choices[0].message.content
|
||||
result = response.choices[0].message.content
|
||||
end_time = time.time()
|
||||
logger.info("talk_to_zhipu call completed in %.2f seconds", end_time - start_time)
|
||||
return accum_resp
|
||||
logger.info("generate_response call completed in %.2f seconds", end_time - start_time)
|
||||
return result
|
||||
except Exception as e:
|
||||
logger.error("Error in talk_to_zhipu: %s", str(e))
|
||||
logger.error("Error in generate_response: %s", str(e))
|
||||
raise
|
||||
|
||||
def talk_to_zhipu_sse(self, message):
|
||||
logger.info("Starting talk_to_zhipu_sse call")
|
||||
def generate_response_sse(self, prompt):
|
||||
logger.info("Starting generate_response_sse call")
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
response = self.client.chat.completions.create(
|
||||
model=self.model_name,
|
||||
messages=[
|
||||
{"role": "user", "content": message},
|
||||
{"role": "user", "content": prompt},
|
||||
],
|
||||
stream=True,
|
||||
temperature=0.01,
|
||||
|
|
@ -50,21 +50,20 @@ class ZhipuService:
|
|||
for chunk in response:
|
||||
yield chunk.choices[0].delta.content
|
||||
end_time = time.time()
|
||||
logger.info("talk_to_zhipu_sse call completed in %.2f seconds", end_time - start_time)
|
||||
logger.info("generate_response_sse call completed in %.2f seconds", end_time - start_time)
|
||||
except Exception as e:
|
||||
logger.error("Error in talk_to_zhipu_sse: %s", str(e))
|
||||
logger.error("Error in generate_response_sse: %s", str(e))
|
||||
raise
|
||||
|
||||
def retrive(self, message, knowledge_id, prompt_template):
|
||||
logger.info("Starting retrive call with knowledge_id: %s", knowledge_id)
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
default_prompt = "从文档\n\"\"\"\n{{knowledge}}\n\"\"\"\n中找问题\n\"\"\"\n{{question}}\n\"\"\"\n的答案，找到答案就仅使用文档语句回答问题，找不到答案就用自身知识回答并且告诉用户该信息不是来自文档。\n不要复述问题，直接开始回答。"
|
||||
|
||||
if prompt_template is None or prompt_template == "":
|
||||
prompt_template = default_prompt
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
response = self.client.chat.completions.create(
|
||||
model="glm-4",
|
||||
messages=[
|
||||
{"role": "user", "content": message},
|
||||
|
|
@ -93,7 +92,6 @@ class ZhipuService:
|
|||
def retrive_sse(self, message, knowledge_id, prompt_template=None,system_prompt=None):
|
||||
logger.info("Starting retrive_sse call with knowledge_id: %s, message:%s", knowledge_id, message)
|
||||
start_time = time.time()
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
default_prompt = "从文档\n\"\"\"\n{{knowledge}}\n\"\"\"\n中找问题\n\"\"\"\n{{question}}\n\"\"\"\n的答案，找到答案就仅使用文档语句回答问题，找不到答案就告诉用户知识库中没有该信息。\n不要复述问题，直接开始回答。"
|
||||
messages = [{"role": "user", "content": message}]
|
||||
# if system_prompt != None:
|
||||
|
|
@ -101,7 +99,7 @@ class ZhipuService:
|
|||
if prompt_template is None or prompt_template == "":
|
||||
prompt_template = default_prompt
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
response = self.client.chat.completions.create(
|
||||
model="glm-4",
|
||||
messages=messages,
|
||||
tools=[
|
||||
|
|
@ -132,11 +130,10 @@ class ZhipuService:
|
|||
prompt_report_template = PromptRepository().get_prompt("report_template")
|
||||
|
||||
prompt_report_missing_check = f"""{prompt_report_template}
|
||||
请检查以下日志信息是否完整，如果信息缺失则提示要求用户需要补充的信息要点，如果信息完整请直接返回“上述日志信息为全部信息”。日志信息如下：\n\"\"\"\n{message}\n\"\"\"\n"""
|
||||
请检查以下日志信息是否完整，如果信息缺失则提示要求用户需要补充的信息要点，如果信息完整请直接返回"上述日志信息为全部信息"。日志信息如下：\n\"\"\"\n{message}\n\"\"\"\n"""
|
||||
|
||||
client = ZhipuAI(api_key=self.app_secret_key)
|
||||
try:
|
||||
response = client.chat.completions.create(
|
||||
response = self.client.chat.completions.create(
|
||||
model="glm-4-flash",
|
||||
messages=[
|
||||
{"role": "user", "content": prompt_report_missing_check},
|
||||
|
|
|
|||
|
|
@ -9,3 +9,5 @@ def init_session():
|
|||
if 'history' not in session:
|
||||
session['history'] = [] # 初始化会话历史
|
||||
session['session_info'] = {}
|
||||
if 'llm_service' not in session:
|
||||
session['llm_service'] = 'zhipu'
|
||||
|
|
|
|||
|
|
@ -17,4 +17,5 @@ Werkzeug==3.0.4
|
|||
zipp==3.20.2
|
||||
zhipuai==2.1.5.20230904
|
||||
pytz==2024.2
|
||||
flask-debug==0.4.3
|
||||
flask-debug==0.4.3
|
||||
openai==1.52.2
|
||||
Loading…
Reference in New Issue