Spaces:
Sleeping
Sleeping
Update chatbot.py
Browse files- chatbot.py +63 -69
chatbot.py
CHANGED
@@ -2,23 +2,32 @@ import streamlit as st
|
|
2 |
import pandas as pd
|
3 |
import os
|
4 |
from datetime import datetime
|
5 |
-
import google.generativeai as genai
|
6 |
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
|
|
11 |
|
12 |
-
class
|
13 |
def __init__(self):
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
if 'uploaded_df' not in st.session_state:
|
15 |
st.session_state.uploaded_df = None
|
16 |
if 'chat_history' not in st.session_state:
|
17 |
st.session_state.chat_history = []
|
18 |
|
19 |
-
def
|
20 |
-
|
21 |
-
st.
|
|
|
|
|
|
|
22 |
|
23 |
# File upload section
|
24 |
uploaded_file = st.file_uploader("Choose a CSV file", type="csv")
|
@@ -31,6 +40,7 @@ class GeminiDataChatbot:
|
|
31 |
self._render_chat_window()
|
32 |
|
33 |
def _process_uploaded_file(self, uploaded_file):
|
|
|
34 |
try:
|
35 |
df = pd.read_csv(uploaded_file)
|
36 |
st.session_state.uploaded_df = df
|
@@ -39,16 +49,10 @@ class GeminiDataChatbot:
|
|
39 |
with st.expander("View Data Preview"):
|
40 |
st.dataframe(df.head())
|
41 |
|
42 |
-
# Initial analysis
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
First give a very brief (2-3 sentence) overview of what this data might contain.
|
47 |
-
Then suggest 3 specific questions I could ask about this data.
|
48 |
-
"""
|
49 |
-
|
50 |
-
with st.spinner("Analyzing your data..."):
|
51 |
-
response = self._generate_gemini_response(initial_prompt, df)
|
52 |
st.session_state.chat_history.append({
|
53 |
"role": "assistant",
|
54 |
"content": response
|
@@ -58,6 +62,7 @@ class GeminiDataChatbot:
|
|
58 |
st.error(f"Error processing file: {str(e)}")
|
59 |
|
60 |
def _render_chat_window(self):
|
|
|
61 |
st.subheader("Chat About Your Data")
|
62 |
|
63 |
# Display chat history
|
@@ -67,55 +72,44 @@ class GeminiDataChatbot:
|
|
67 |
|
68 |
# User input
|
69 |
if prompt := st.chat_input("Ask about your data..."):
|
70 |
-
|
71 |
-
st.session_state.chat_history.append({"role": "user", "content": prompt})
|
72 |
-
|
73 |
-
# Display user message
|
74 |
-
with st.chat_message("user"):
|
75 |
-
st.markdown(prompt)
|
76 |
-
|
77 |
-
# Generate and display assistant response
|
78 |
-
with st.chat_message("assistant"):
|
79 |
-
with st.spinner("Thinking..."):
|
80 |
-
response = self._generate_gemini_response(prompt, st.session_state.uploaded_df)
|
81 |
-
st.markdown(response)
|
82 |
-
|
83 |
-
# Add assistant response to chat history
|
84 |
-
st.session_state.chat_history.append({"role": "assistant", "content": response})
|
85 |
|
86 |
-
def
|
87 |
-
"""
|
88 |
-
|
89 |
-
|
90 |
-
data_summary = f"""
|
91 |
-
Data Summary:
|
92 |
-
- Shape: {df.shape}
|
93 |
-
- Columns: {', '.join(df.columns)}
|
94 |
-
- First 5 rows:
|
95 |
-
{df.head().to_markdown()}
|
96 |
-
"""
|
97 |
-
|
98 |
-
# Create prompt with context
|
99 |
-
full_prompt = f"""
|
100 |
-
You are a data analysis assistant. The user has uploaded a dataset with the following characteristics:
|
101 |
-
{data_summary}
|
102 |
-
|
103 |
-
User Question: {prompt}
|
104 |
-
|
105 |
-
Provide a detailed response answering their question about the data. If appropriate, include:
|
106 |
-
- Relevant statistics
|
107 |
-
- Potential visualizations that would help
|
108 |
-
- Any data quality issues to consider
|
109 |
-
- Business insights if applicable
|
110 |
-
"""
|
111 |
-
|
112 |
-
response = model.generate_content(full_prompt)
|
113 |
-
return response.text
|
114 |
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
import pandas as pd
|
3 |
import os
|
4 |
from datetime import datetime
|
|
|
5 |
|
6 |
+
try:
|
7 |
+
import google.generativeai as genai
|
8 |
+
GEMINI_AVAILABLE = True
|
9 |
+
except ImportError:
|
10 |
+
GEMINI_AVAILABLE = False
|
11 |
|
12 |
+
class ChatbotManager:
|
13 |
def __init__(self):
|
14 |
+
if GEMINI_AVAILABLE and 'GEMINI_API_KEY' in os.environ:
|
15 |
+
genai.configure(api_key=os.environ['GEMINI_API_KEY'])
|
16 |
+
self.model = genai.GenerativeModel('gemini-pro')
|
17 |
+
else:
|
18 |
+
self.model = None
|
19 |
+
|
20 |
if 'uploaded_df' not in st.session_state:
|
21 |
st.session_state.uploaded_df = None
|
22 |
if 'chat_history' not in st.session_state:
|
23 |
st.session_state.chat_history = []
|
24 |
|
25 |
+
def render_chat_interface(self):
|
26 |
+
"""Render the main chat interface"""
|
27 |
+
st.header("π Data Analysis Chatbot")
|
28 |
+
|
29 |
+
if not GEMINI_AVAILABLE:
|
30 |
+
st.warning("Gemini API not available - running in limited mode")
|
31 |
|
32 |
# File upload section
|
33 |
uploaded_file = st.file_uploader("Choose a CSV file", type="csv")
|
|
|
40 |
self._render_chat_window()
|
41 |
|
42 |
def _process_uploaded_file(self, uploaded_file):
|
43 |
+
"""Process the uploaded CSV file"""
|
44 |
try:
|
45 |
df = pd.read_csv(uploaded_file)
|
46 |
st.session_state.uploaded_df = df
|
|
|
49 |
with st.expander("View Data Preview"):
|
50 |
st.dataframe(df.head())
|
51 |
|
52 |
+
# Initial analysis
|
53 |
+
if self.model:
|
54 |
+
initial_prompt = f"Briefly describe this dataset with {len(df)} rows and {len(df.columns)} columns."
|
55 |
+
response = self._generate_response(initial_prompt)
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
st.session_state.chat_history.append({
|
57 |
"role": "assistant",
|
58 |
"content": response
|
|
|
62 |
st.error(f"Error processing file: {str(e)}")
|
63 |
|
64 |
def _render_chat_window(self):
|
65 |
+
"""Render the chat conversation window"""
|
66 |
st.subheader("Chat About Your Data")
|
67 |
|
68 |
# Display chat history
|
|
|
72 |
|
73 |
# User input
|
74 |
if prompt := st.chat_input("Ask about your data..."):
|
75 |
+
self._handle_user_input(prompt)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
|
77 |
+
def _handle_user_input(self, prompt):
|
78 |
+
"""Handle user input and generate response"""
|
79 |
+
# Add user message to chat history
|
80 |
+
st.session_state.chat_history.append({"role": "user", "content": prompt})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
|
82 |
+
# Display user message
|
83 |
+
with st.chat_message("user"):
|
84 |
+
st.markdown(prompt)
|
85 |
+
|
86 |
+
# Generate and display assistant response
|
87 |
+
with st.chat_message("assistant"):
|
88 |
+
with st.spinner("Thinking..."):
|
89 |
+
response = self._generate_response(prompt)
|
90 |
+
st.markdown(response)
|
91 |
+
|
92 |
+
# Add assistant response to chat history
|
93 |
+
st.session_state.chat_history.append({"role": "assistant", "content": response})
|
94 |
+
|
95 |
+
def _generate_response(self, prompt: str) -> str:
|
96 |
+
"""Generate response using available backend"""
|
97 |
+
df = st.session_state.uploaded_df
|
98 |
+
|
99 |
+
if self.model:
|
100 |
+
# Use Gemini if available
|
101 |
+
try:
|
102 |
+
data_summary = f"Data: {len(df)} rows, columns: {', '.join(df.columns)}"
|
103 |
+
full_prompt = f"{data_summary}\n\nUser question: {prompt}"
|
104 |
+
response = self.model.generate_content(full_prompt)
|
105 |
+
return response.text
|
106 |
+
except Exception as e:
|
107 |
+
return f"Gemini error: {str(e)}"
|
108 |
+
else:
|
109 |
+
# Fallback basic analysis
|
110 |
+
if "summary" in prompt.lower():
|
111 |
+
return f"Basic summary:\n{df.describe().to_markdown()}"
|
112 |
+
elif "columns" in prompt.lower():
|
113 |
+
return f"Columns: {', '.join(df.columns)}"
|
114 |
+
else:
|
115 |
+
return "I can provide basic info about columns and summary statistics."
|