Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,76 +1,250 @@
|
|
1 |
import os
|
2 |
import warnings
|
3 |
-
import
|
4 |
from dotenv import load_dotenv
|
5 |
-
|
6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
from langchain_community.vectorstores import FAISS
|
8 |
from langchain_community.embeddings import AzureOpenAIEmbeddings
|
9 |
-
from
|
10 |
-
|
11 |
-
|
12 |
-
import gradio_client.utils
|
13 |
-
gradio_client.utils.json_schema_to_python_type = lambda schema, defs=None: "string"
|
14 |
-
|
15 |
# Load environment variables
|
16 |
load_dotenv()
|
17 |
AZURE_OPENAI_API_KEY = os.getenv("AZURE_OPENAI_API_KEY")
|
18 |
AZURE_OPENAI_ENDPOINT = os.getenv("AZURE_OPENAI_ENDPOINT")
|
19 |
-
#AZURE_END_POINT_O3 = os.getenv("AZURE_END_POINT_O3")
|
20 |
AZURE_OPENAI_LLM_DEPLOYMENT = os.getenv("AZURE_OPENAI_LLM_DEPLOYMENT")
|
21 |
AZURE_OPENAI_EMBEDDING_DEPLOYMENT = os.getenv("AZURE_OPENAI_EMBEDDING_DEPLOYMENT")
|
22 |
-
|
23 |
-
|
24 |
-
if not all([AZURE_OPENAI_API_KEY,
|
25 |
-
AZURE_OPENAI_ENDPOINT,
|
26 |
-
#AZURE_END_POINT_O3,
|
27 |
-
AZURE_OPENAI_LLM_DEPLOYMENT,
|
28 |
-
AZURE_OPENAI_EMBEDDING_DEPLOYMENT]):
|
29 |
raise ValueError("Missing one or more Azure OpenAI environment variables.")
|
30 |
-
|
31 |
-
# Suppress warnings
|
32 |
warnings.filterwarnings("ignore")
|
33 |
-
|
34 |
-
#
|
35 |
embeddings = AzureOpenAIEmbeddings(
|
36 |
azure_deployment=AZURE_OPENAI_EMBEDDING_DEPLOYMENT,
|
37 |
azure_endpoint=AZURE_OPENAI_ENDPOINT,
|
38 |
-
#azure_endpoint=AZURE_END_POINT_O3,
|
39 |
openai_api_key=AZURE_OPENAI_API_KEY,
|
40 |
-
openai_api_version="2025-01-01-preview",
|
41 |
chunk_size=1000
|
42 |
)
|
43 |
-
|
44 |
-
#
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
#
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
)
|
58 |
-
|
59 |
-
# Build conversational RAG chain
|
60 |
-
qa = ConversationalRetrievalChain.from_llm(
|
61 |
-
llm=llm,
|
62 |
-
retriever=vectorstore.as_retriever(),
|
63 |
-
return_source_documents=False
|
64 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
|
66 |
-
|
67 |
-
|
68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
def sysml_chatbot(message, history):
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
# Gradio UI
|
76 |
with gr.Blocks() as demo:
|
|
|
1 |
import os
|
2 |
import warnings
|
3 |
+
import json
|
4 |
from dotenv import load_dotenv
|
5 |
+
from typing import Dict, Any, List, Optional
|
6 |
+
import time
|
7 |
+
from functools import lru_cache
|
8 |
+
import logging
|
9 |
+
|
10 |
+
|
11 |
+
|
12 |
+
from langchain.agents import Tool, AgentExecutor
|
13 |
+
from langchain.tools.retriever import create_retriever_tool
|
14 |
+
from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
|
15 |
from langchain_community.vectorstores import FAISS
|
16 |
from langchain_community.embeddings import AzureOpenAIEmbeddings
|
17 |
+
from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
|
18 |
+
from openai import AzureOpenAI
|
19 |
+
|
|
|
|
|
|
|
20 |
# Load environment variables
|
21 |
load_dotenv()
|
22 |
AZURE_OPENAI_API_KEY = os.getenv("AZURE_OPENAI_API_KEY")
|
23 |
AZURE_OPENAI_ENDPOINT = os.getenv("AZURE_OPENAI_ENDPOINT")
|
|
|
24 |
AZURE_OPENAI_LLM_DEPLOYMENT = os.getenv("AZURE_OPENAI_LLM_DEPLOYMENT")
|
25 |
AZURE_OPENAI_EMBEDDING_DEPLOYMENT = os.getenv("AZURE_OPENAI_EMBEDDING_DEPLOYMENT")
|
26 |
+
|
27 |
+
if not all([AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_LLM_DEPLOYMENT, AZURE_OPENAI_EMBEDDING_DEPLOYMENT]):
|
|
|
|
|
|
|
|
|
|
|
28 |
raise ValueError("Missing one or more Azure OpenAI environment variables.")
|
29 |
+
|
|
|
30 |
warnings.filterwarnings("ignore")
|
31 |
+
|
32 |
+
# Embeddings for retriever
|
33 |
embeddings = AzureOpenAIEmbeddings(
|
34 |
azure_deployment=AZURE_OPENAI_EMBEDDING_DEPLOYMENT,
|
35 |
azure_endpoint=AZURE_OPENAI_ENDPOINT,
|
|
|
36 |
openai_api_key=AZURE_OPENAI_API_KEY,
|
37 |
+
openai_api_version="2025-01-01-preview",
|
38 |
chunk_size=1000
|
39 |
)
|
40 |
+
|
41 |
+
# Get the directory where this script is located
|
42 |
+
SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
|
43 |
+
|
44 |
+
# Build the absolute path to the faiss_index_sysml directory relative to this script
|
45 |
+
FAISS_INDEX_PATH = os.path.join(SCRIPT_DIR, "faiss_index_sysml")
|
46 |
+
# Load FAISS vectorstore
|
47 |
+
vectorstore = FAISS.load_local(FAISS_INDEX_PATH, embeddings, allow_dangerous_deserialization=True)
|
48 |
+
|
49 |
+
# Initialize Azure OpenAI client directly
|
50 |
+
client = AzureOpenAI(
|
51 |
+
api_key=AZURE_OPENAI_API_KEY,
|
52 |
+
api_version="2025-01-01-preview",
|
53 |
+
azure_endpoint=AZURE_OPENAI_ENDPOINT
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
)
|
55 |
+
|
56 |
+
|
57 |
+
logger = logging.getLogger(__name__)
|
58 |
+
# SysML retriever function
|
59 |
+
@lru_cache(maxsize=100)
|
60 |
+
def sysml_retriever(query: str) -> str:
|
61 |
+
start_time = time.time()
|
62 |
+
try:
|
63 |
+
results = vectorstore.similarity_search(query, k=100)
|
64 |
+
contexts = [doc.page_content for doc in results]
|
65 |
+
response = "\n\n".join(contexts)
|
66 |
+
|
67 |
+
# Log performance metrics
|
68 |
+
duration = time.time() - start_time
|
69 |
+
print(f"Retrieval completed in {duration:.2f}s for query: {query[:50]}...")
|
70 |
+
|
71 |
+
return response
|
72 |
+
except Exception as e:
|
73 |
+
logger.error(f"Retrieval error: {str(e)}")
|
74 |
+
return "Unable to retrieve information at this time."
|
75 |
+
|
76 |
+
|
77 |
+
# sysml_retriever = create_retriever_tool(
|
78 |
+
# retriever=vectorstore.as_retriever(),
|
79 |
+
# name="SysMLRetriever",
|
80 |
+
# description="Use this to answer questions about SysML diagrams and modeling."
|
81 |
+
# )
|
82 |
|
83 |
+
# Dummy functions
|
84 |
+
def dummy_weather_lookup(location: str = "London") -> str:
|
85 |
+
return f"The weather in {location} is sunny and 25°C."
|
86 |
+
|
87 |
+
def dummy_time_lookup(timezone: str = "UTC") -> str:
|
88 |
+
return f"The current time in {timezone} is 3:00 PM."
|
89 |
+
|
90 |
+
# Tools definition for OpenAI function calling
|
91 |
+
tools_definition = [
|
92 |
+
{
|
93 |
+
"type": "function",
|
94 |
+
"function": {
|
95 |
+
"name": "SysMLRetriever",
|
96 |
+
"description": "Use this to answer questions about SysML diagrams and modeling.",
|
97 |
+
"parameters": {
|
98 |
+
"type": "object",
|
99 |
+
"properties": {
|
100 |
+
"query": {
|
101 |
+
"type": "string",
|
102 |
+
"description": "The search query to find information about SysML"
|
103 |
+
}
|
104 |
+
},
|
105 |
+
"required": ["query"]
|
106 |
+
}
|
107 |
+
}
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"type": "function",
|
111 |
+
"function": {
|
112 |
+
"name": "WeatherLookup",
|
113 |
+
"description": "Use this to look up the current weather in a specified location.",
|
114 |
+
"parameters": {
|
115 |
+
"type": "object",
|
116 |
+
"properties": {
|
117 |
+
"location": {
|
118 |
+
"type": "string",
|
119 |
+
"description": "The location to look up the weather for"
|
120 |
+
}
|
121 |
+
},
|
122 |
+
"required": ["location"]
|
123 |
+
}
|
124 |
+
},
|
125 |
+
},
|
126 |
+
{
|
127 |
+
"type": "function",
|
128 |
+
"function": {
|
129 |
+
"name": "TimeLookup",
|
130 |
+
"description": "Use this to look up the current time in a specified timezone.",
|
131 |
+
"parameters": {
|
132 |
+
"type": "object",
|
133 |
+
"properties": {
|
134 |
+
"timezone": {
|
135 |
+
"type": "string",
|
136 |
+
"description": "The timezone to look up the current time for"
|
137 |
+
}
|
138 |
+
},
|
139 |
+
"required": ["timezone"]
|
140 |
+
}
|
141 |
+
}
|
142 |
+
}
|
143 |
+
]
|
144 |
+
|
145 |
+
# Tool execution mapping
|
146 |
+
tool_mapping = {
|
147 |
+
"SysMLRetriever": sysml_retriever,
|
148 |
+
"WeatherLookup": dummy_weather_lookup,
|
149 |
+
"TimeLookup": dummy_time_lookup
|
150 |
+
}
|
151 |
+
|
152 |
+
# Convert chat history
|
153 |
+
def convert_history_to_messages(history):
|
154 |
+
messages = []
|
155 |
+
for user, bot in history:
|
156 |
+
messages.append({"role": "user", "content": user})
|
157 |
+
messages.append({"role": "assistant", "content": bot})
|
158 |
+
return messages
|
159 |
+
|
160 |
+
# Main chatbot function with direct function calling
|
161 |
def sysml_chatbot(message, history):
|
162 |
+
# Convert history to messages format
|
163 |
+
chat_messages = convert_history_to_messages(history)
|
164 |
+
|
165 |
+
# Add system message at beginning
|
166 |
+
full_messages = [
|
167 |
+
{"role": "system", "content": "You are a helpful SysML modeling assistant and also a capable smart Assistant "}
|
168 |
+
]
|
169 |
+
full_messages.extend(chat_messages)
|
170 |
+
|
171 |
+
# Add current user message
|
172 |
+
full_messages.append({"role": "user", "content": message})
|
173 |
+
|
174 |
+
try:
|
175 |
+
# First call to get either a direct answer or a function call
|
176 |
+
response = client.chat.completions.create(
|
177 |
+
model=AZURE_OPENAI_LLM_DEPLOYMENT,
|
178 |
+
messages=full_messages,
|
179 |
+
tools=tools_definition,
|
180 |
+
tool_choice={"type": "function", "function": {"name": "SysMLRetriever"}}
|
181 |
+
)
|
182 |
+
|
183 |
+
assistant_message = response.choices[0].message
|
184 |
+
|
185 |
+
# Check if the model wants to call a function
|
186 |
+
if assistant_message.tool_calls:
|
187 |
+
# Get the function call details
|
188 |
+
tool_call = assistant_message.tool_calls[0]
|
189 |
+
function_name = tool_call.function.name
|
190 |
+
function_args = json.loads(tool_call.function.arguments)
|
191 |
+
print("Attempting function calling...")
|
192 |
+
# Execute the function
|
193 |
+
if function_name in tool_mapping:
|
194 |
+
function_response = tool_mapping[function_name](**function_args)
|
195 |
+
|
196 |
+
# Append the assistant's request and the function response to messages
|
197 |
+
full_messages.append({"role": "assistant", "content": None, "tool_calls": [
|
198 |
+
{"id": tool_call.id, "type": "function", "function": {"name": function_name, "arguments": tool_call.function.arguments}}
|
199 |
+
]})
|
200 |
+
|
201 |
+
full_messages.append({
|
202 |
+
"role": "tool",
|
203 |
+
"tool_call_id": tool_call.id,
|
204 |
+
"content": function_response
|
205 |
+
})
|
206 |
+
|
207 |
+
# Second call to get the final answer based on the function result
|
208 |
+
second_response = client.chat.completions.create(
|
209 |
+
model=AZURE_OPENAI_LLM_DEPLOYMENT,
|
210 |
+
messages=full_messages
|
211 |
+
)
|
212 |
+
|
213 |
+
answer = second_response.choices[0].message.content
|
214 |
+
print("Getting final response after function execution...")
|
215 |
+
print(f"Function '{function_name}' executed successfully. Response: {answer}")
|
216 |
+
else:
|
217 |
+
answer = f"I tried to use a function '{function_name}' that's not available. Let me try again with general knowledge: SysML is a modeling language for systems engineering that helps visualize and analyze complex systems."
|
218 |
+
else:
|
219 |
+
# Model provided a direct answer
|
220 |
+
answer = assistant_message.content
|
221 |
+
|
222 |
+
history.append((message, answer))
|
223 |
+
return answer, history
|
224 |
+
|
225 |
+
except Exception as e:
|
226 |
+
print(f"Error in function calling: {str(e)}")
|
227 |
+
|
228 |
+
# Fallback to a direct response without function calling
|
229 |
+
try:
|
230 |
+
simple_messages = [
|
231 |
+
{"role": "system", "content": "You are a helpful SysML modeling assistant."}
|
232 |
+
]
|
233 |
+
simple_messages.extend(chat_messages)
|
234 |
+
simple_messages.append({"role": "user", "content": message})
|
235 |
+
|
236 |
+
fallback_response = client.chat.completions.create(
|
237 |
+
model=AZURE_OPENAI_LLM_DEPLOYMENT,
|
238 |
+
messages=simple_messages
|
239 |
+
)
|
240 |
+
|
241 |
+
answer = fallback_response.choices[0].message.content
|
242 |
+
except Exception as fallback_error:
|
243 |
+
print(f"Error in fallback: {str(fallback_error)}")
|
244 |
+
answer = "I'm having trouble accessing my tools right now. SysML is a modeling language used in systems engineering to visualize and analyze complex systems through various diagram types."
|
245 |
+
|
246 |
+
history.append((message, answer))
|
247 |
+
return answer, history
|
248 |
|
249 |
# Gradio UI
|
250 |
with gr.Blocks() as demo:
|