Spaces:
Sleeping
Sleeping
File size: 23,739 Bytes
5b7f559 aef0372 bd67e9b aef0372 bd67e9b 5b7f559 bd67e9b 5b7f559 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 4d7cda7 ba5c284 aef0372 da1c657 aef0372 ba5c284 da1c657 ba5c284 aef0372 2d0370f da1c657 2d0370f da1c657 bd67e9b da1c657 ba5c284 da1c657 5b7f559 da1c657 ba5c284 da1c657 ba5c284 bd67e9b ba5c284 bd67e9b da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 bd67e9b ba5c284 da1c657 bd67e9b da1c657 bd67e9b ba5c284 bd67e9b ba5c284 bd67e9b da1c657 bd67e9b da1c657 bd67e9b da1c657 bd67e9b 5b7f559 bd67e9b da1c657 bd67e9b 5b7f559 bd67e9b 5b7f559 bd67e9b da1c657 bd67e9b da1c657 bd67e9b 5b7f559 bd67e9b da1c657 bd67e9b da1c657 bd67e9b 5b7f559 bd67e9b da1c657 bd67e9b 5b7f559 bd67e9b da1c657 bd67e9b 2d0370f da1c657 bd67e9b da1c657 bd67e9b 2d0370f da1c657 bd67e9b 2d0370f da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 ba5c284 da1c657 bd67e9b 8d3bbf0 da1c657 bd67e9b 8d3bbf0 da1c657 ba5c284 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 |
import os
import time
import pandas as pd
import gradio as gr
from langchain_groq import ChatGroq
from langchain_huggingface import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma
from langchain_core.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from PyPDF2 import PdfReader
# Configuration constants
COLLECTION_NAME = "GBVRS"
DATA_FOLDER = "./"
APP_VERSION = "v1.0.0"
APP_NAME = "Ijwi ry'Ubufasha"
MAX_HISTORY_MESSAGES = 8 # Limit history to avoid token limits
# Global variables for application state
llm = None
embed_model = None
vectorstore = None
retriever = None
rag_chain = None
# User session management
class UserSession:
def __init__(self, session_id, llm):
"""Initialize a user session with unique ID and language model."""
self.session_id = session_id
self.user_info = {"Nickname": "Guest"}
self.conversation_history = []
self.llm = llm
self.welcome_message = None
self.last_activity = time.time()
def set_user(self, user_info):
"""Set user information and generate welcome message."""
self.user_info = user_info
self.generate_welcome_message()
# Initialize conversation history with welcome message
welcome = self.get_welcome_message()
self.conversation_history = [
{"role": "assistant", "content": welcome},
]
def get_user(self):
"""Get current user information."""
return self.user_info
def generate_welcome_message(self):
"""Generate a dynamic welcome message using the LLM."""
try:
nickname = self.user_info.get("Nickname", "Guest")
# Use the LLM to generate the message
prompt = (
f"Create a brief and warm welcome message for {nickname} that's about 1-2 sentences. "
f"Emphasize this is a safe space for discussing gender-based violence issues "
f"and that we provide support and resources. Keep it warm and reassuring."
)
response = self.llm.invoke(prompt)
welcome = response.content.strip()
# Format the message with HTML styling
self.welcome_message = (
f"<div style='font-size: 18px; color: #4E6BBF;'>"
f"{welcome}"
f"</div>"
)
except Exception as e:
# Fallback welcome message
nickname = self.user_info.get("Nickname", "Guest")
self.welcome_message = (
f"<div style='font-size: 18px; color: #4E6BBF;'>"
f"Welcome, {nickname}! You're in a safe space. We're here to provide support with "
f"gender-based violence issues and connect you with resources that can help."
f"</div>"
)
def get_welcome_message(self):
"""Get the formatted welcome message."""
if not self.welcome_message:
self.generate_welcome_message()
return self.welcome_message
def add_to_history(self, role, message):
"""Add a message to the conversation history."""
self.conversation_history.append({"role": role, "content": message})
self.last_activity = time.time()
# Trim history if it gets too long
if len(self.conversation_history) > MAX_HISTORY_MESSAGES * 2: # Keep pairs of messages
# Keep the first message (welcome) and the most recent messages
self.conversation_history = [self.conversation_history[0]] + self.conversation_history[-MAX_HISTORY_MESSAGES*2+1:]
def get_conversation_history(self):
"""Get the full conversation history."""
return self.conversation_history
def get_formatted_history(self):
"""Get conversation history formatted as a string for the LLM."""
# Skip the welcome message and only include the last few exchanges
recent_history = self.conversation_history[1:] if len(self.conversation_history) > 1 else []
# Limit to last MAX_HISTORY_MESSAGES exchanges
if len(recent_history) > MAX_HISTORY_MESSAGES * 2:
recent_history = recent_history[-MAX_HISTORY_MESSAGES*2:]
formatted_history = ""
for entry in recent_history:
role = "User" if entry["role"] == "user" else "Assistant"
# Truncate very long messages to avoid token limits
content = entry["content"]
if len(content) > 500: # Limit message length
content = content[:500] + "..."
formatted_history += f"{role}: {content}\n\n"
return formatted_history
def is_expired(self, timeout_seconds=3600):
"""Check if the session has been inactive for too long."""
return (time.time() - self.last_activity) > timeout_seconds
# Session manager to handle multiple users
class SessionManager:
def __init__(self):
"""Initialize the session manager."""
self.sessions = {}
self.session_timeout = 3600 # 1 hour timeout
def get_session(self, session_id):
"""Get an existing session or create a new one."""
# Clean expired sessions first
self._clean_expired_sessions()
# Create new session if needed
if session_id not in self.sessions:
self.sessions[session_id] = UserSession(session_id, llm)
return self.sessions[session_id]
def _clean_expired_sessions(self):
"""Remove expired sessions to free up memory."""
expired_keys = []
for key, session in self.sessions.items():
if session.is_expired(self.session_timeout):
expired_keys.append(key)
for key in expired_keys:
del self.sessions[key]
# Initialize the session manager
session_manager = SessionManager()
def initialize_assistant():
"""Initialize the assistant with necessary components and configurations."""
global llm, embed_model, vectorstore, retriever, rag_chain
# Initialize API key - try both possible key names
groq_api_key = os.environ.get('GBV') or os.environ.get('GBV')
if not groq_api_key:
print("WARNING: No GROQ API key found in userdata.")
# Initialize LLM - Default to Llama model which is more widely available
llm = ChatGroq(
model="llama-3.3-70b-versatile", # More reliable than whisper model
api_key=groq_api_key
)
# Set up embedding model
try:
embed_model = HuggingFaceEmbeddings(model_name="mixedbread-ai/mxbai-embed-large-v1")
except Exception as e:
# Fallback to smaller model
embed_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
# Process data and create vector store
print("Processing data files...")
data = process_data_files()
print("Creating vector store...")
vectorstore = create_vectorstore(data)
retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
# Create RAG chain
print("Setting up RAG chain...")
rag_chain = create_rag_chain()
print(f"β
{APP_NAME} initialized successfully")
def process_data_files():
"""Process all data files from the specified folder."""
context_data = []
try:
if not os.path.exists(DATA_FOLDER):
print(f"WARNING: Data folder does not exist: {DATA_FOLDER}")
return context_data
# Get list of data files
all_files = os.listdir(DATA_FOLDER)
data_files = [f for f in all_files if f.lower().endswith(('.csv', '.xlsx', '.xls'))]
if not data_files:
print(f"WARNING: No data files found in: {DATA_FOLDER}")
return context_data
# Process each file
for index, file_name in enumerate(data_files, 1):
print(f"Processing file {index}/{len(data_files)}: {file_name}")
file_path = os.path.join(DATA_FOLDER, file_name)
try:
# Read file based on extension
if file_name.lower().endswith('.csv'):
df = pd.read_csv(file_path)
else:
df = pd.read_excel(file_path)
# Check if column 3 exists (source data is in third column)
if df.shape[1] > 2:
column_data = df.iloc[:, 2].dropna().astype(str).tolist()
# Each row becomes one chunk with metadata
for i, text in enumerate(column_data):
if text and len(text.strip()) > 0:
context_data.append({
"page_content": text,
"metadata": {
"source": file_name,
"row": i+1
}
})
else:
print(f"WARNING: File {file_name} has fewer than 3 columns.")
except Exception as e:
print(f"ERROR processing file {file_name}: {e}")
print(f"β
Created {len(context_data)} chunks from {len(data_files)} files.")
except Exception as e:
print(f"ERROR accessing data folder: {e}")
return context_data
def create_vectorstore(data):
"""
Creates and returns a Chroma vector store populated with the provided data.
Parameters:
data (list): A list of dictionaries, each containing 'page_content' and 'metadata'.
Returns:
Chroma: The populated Chroma vector store instance.
"""
# Initialize the vector store
vectorstore = Chroma(
collection_name=COLLECTION_NAME,
embedding_function=embed_model,
persist_directory="./"
)
if not data:
print("β οΈ No data provided. Returning an empty vector store.")
return vectorstore
try:
# Extract text and metadata from the data
texts = [doc["page_content"] for doc in data]
# Add the texts and metadata to the vector store
vectorstore.add_texts(texts)
except Exception as e:
print(f"β Failed to add documents to vector store: {e}")
return vs
def create_rag_chain():
"""Create the RAG chain for processing user queries."""
# Define the prompt template
template = """
You are a compassionate and supportive AI assistant specializing in helping individuals affected by Gender-Based Violence (GBV). Your responses must be based EXCLUSIVELY on the information provided in the context. Your primary goal is to provide emotionally intelligent support while maintaining appropriate boundaries.
**Previous conversation:** {conversation_history}
**Context information:** {context}
**User's Question:** {question}
When responding follow these guidelines:
1. **Strict Context Adherence**
- Only use information that appears in the provided {context}
- If the answer is not found in the context, state "I don't have that information in my available resources" rather than generating a response
2. **Personalized Communication**
- Avoid contractions (e.g., use I am instead of I'm)
- Incorporate thoughtful pauses or reflective questions when the conversation involves difficult topics
- Use selective emojis (π, π€, β€οΈ) only when tone-appropriate and not during crisis discussions
- Balance warmth with professionalism
3. **Emotional Intelligence**
- Validate feelings without judgment
- Offer reassurance when appropriate, always centered on empowerment
- Adjust your tone based on the emotional state conveyed
4. **Conversation Management**
- Refer to {conversation_history} to maintain continuity and avoid repetition
- Use clear paragraph breaks for readability
5. **Information Delivery**
- Extract only relevant information from {context} that directly addresses the question
- Present information in accessible, non-technical language
- When information is unavailable, respond with: "I don't have that specific information right now, {first_name}. Would it be helpful if I focus on [alternative support option]?"
6. **Safety and Ethics**
- Do not generate any speculative content or advice not supported by the context
- If the context contains safety information, prioritize sharing that information
Your response must come entirely from the provided context, maintaining the supportive tone while never introducing information from outside the provided materials.
**Context:** {context}
**User's Question:** {question}
**Your Response:**
"""
rag_prompt = PromptTemplate.from_template(template)
def get_context_and_question(query_with_session):
# Extract query and session_id
query = query_with_session["query"]
session_id = query_with_session["session_id"]
# Get the user session
session = session_manager.get_session(session_id)
user_info = session.get_user()
first_name = user_info.get("Nickname", "User")
conversation_hist = session.get_formatted_history()
try:
# Retrieve relevant documents
retrieved_docs = retriever.invoke(query)
context_str = format_context(retrieved_docs)
except Exception as e:
print(f"ERROR retrieving documents: {e}")
context_str = "No relevant information found."
# Return the combined inputs for the prompt
return {
"context": context_str,
"question": query,
"first_name": first_name,
"conversation_history": conversation_hist
}
# Build the chain
try:
chain = (
RunnablePassthrough()
| get_context_and_question
| rag_prompt
| llm
| StrOutputParser()
)
return chain
except Exception as e:
print(f"ERROR creating RAG chain: {e}")
# Return a simple function as fallback
def fallback_chain(query_with_session):
session_id = query_with_session["session_id"]
session = session_manager.get_session(session_id)
nickname = session.get_user().get("Nickname", "there")
return f"I'm here to help you, {nickname}, but I'm experiencing some technical difficulties right now. Please try again shortly."
return fallback_chain
def format_context(retrieved_docs):
"""Format retrieved documents into a string context."""
if not retrieved_docs:
return "No relevant information available."
return "\n\n".join([doc.page_content for doc in retrieved_docs])
def rag_memory_stream(message, history, session_id):
"""Process user message and generate response with memory."""
# Get the user session
session = session_manager.get_session(session_id)
# Add user message to history
session.add_to_history("user", message)
try:
# Get response from RAG chain
print(f"Processing message for session {session_id}: {message[:50]}...")
# Pass both query and session_id to the chain
response = rag_chain.invoke({
"query": message,
"session_id": session_id
})
print(f"Generated response: {response[:50]}...")
# Add assistant response to history
session.add_to_history("assistant", response)
# Yield the response
yield response
except Exception as e:
import traceback
print(f"ERROR in rag_memory_stream: {e}")
print(f"Detailed error: {traceback.format_exc()}")
nickname = session.get_user().get("Nickname", "there")
error_msg = f"I'm sorry, {nickname}. I encountered an error processing your request. Let's try a different question."
session.add_to_history("assistant", error_msg)
yield error_msg
def collect_user_info(nickname, session_id):
"""Store user details and initialize session."""
if not nickname or nickname.strip() == "":
return "Nickname is required to proceed.", gr.update(visible=False), gr.update(visible=True), []
# Store user info for chat session
user_info = {
"Nickname": nickname.strip(),
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S")
}
# Get the session and set user info
session = session_manager.get_session(session_id)
session.set_user(user_info)
# Generate welcome message
welcome_message = session.get_welcome_message()
# Return welcome message and update UI
return welcome_message, gr.update(visible=True), gr.update(visible=False), [(None, welcome_message)]
def get_css():
"""Define CSS for the UI."""
return """
:root {
--primary: #4E6BBF;
--primary-light: #697BBF;
--text-primary: #333333;
--text-secondary: #666666;
--background: #F9FAFC;
--card-bg: #FFFFFF;
--border: #E1E5F0;
--shadow: rgba(0, 0, 0, 0.05);
}
body, .gradio-container {
margin: 0;
padding: 0;
width: 100vw;
height: 100vh;
display: flex;
flex-direction: column;
justify-content: center;
align-items: center;
background: var(--background);
color: var(--text-primary);
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
}
.gradio-container {
max-width: 100%;
max-height: 100%;
}
.gr-box {
background: var(--card-bg);
color: var(--text-primary);
border-radius: 12px;
padding: 2rem;
border: 1px solid var(--border);
box-shadow: 0 4px 12px var(--shadow);
}
.gr-button-primary {
background: var(--primary);
color: white;
padding: 12px 24px;
border-radius: 8px;
transition: all 0.3s ease;
border: none;
font-weight: bold;
}
.gr-button-primary:hover {
transform: translateY(-1px);
box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
background: var(--primary-light);
}
footer {
text-align: center;
color: var(--text-secondary);
padding: 1rem;
font-size: 0.9em;
}
.gr-markdown h2 {
color: var(--primary);
margin-bottom: 0.5rem;
font-size: 1.8em;
}
.gr-markdown h3 {
color: var(--text-secondary);
margin-bottom: 1.5rem;
font-weight: normal;
}
#chatbot_container .chat-title h1,
#chatbot_container .empty-chatbot {
color: var(--primary);
}
#input_nickname {
padding: 12px;
border-radius: 8px;
border: 1px solid var(--border);
background: var(--card-bg);
transition: all 0.3s ease;
}
#input_nickname:focus {
border-color: var(--primary);
box-shadow: 0 0 0 2px rgba(78, 107, 191, 0.2);
outline: none;
}
.chatbot-container .message.user {
background: #E8F0FE;
border-radius: 12px 12px 0 12px;
}
.chatbot-container .message.bot {
background: #F5F7FF;
border-radius: 12px 12px 12px 0;
}
"""
def create_ui():
"""Create and configure the Gradio UI."""
with gr.Blocks(css=get_css(), theme=gr.themes.Soft()) as demo:
# Create a unique session ID for this browser tab
session_id = gr.State(value=f"session_{int(time.time())}_{os.urandom(4).hex()}")
# Registration section
with gr.Column(visible=True, elem_id="registration_container") as registration_container:
gr.Markdown(f"## Welcome to {APP_NAME}")
gr.Markdown("### Your privacy is important to us. Please provide a nickname to continue.")
with gr.Row():
first_name = gr.Textbox(
label="Nickname",
placeholder="Enter your nickname",
scale=1,
elem_id="input_nickname"
)
with gr.Row():
submit_btn = gr.Button("Start Chatting", variant="primary", scale=2)
response_message = gr.Markdown()
# Chatbot section (initially hidden)
with gr.Column(visible=False, elem_id="chatbot_container") as chatbot_container:
# Create a custom chat interface to pass session_id to our function
chatbot = gr.Chatbot(
elem_id="chatbot",
height=500,
show_label=False
)
with gr.Row():
msg = gr.Textbox(
placeholder="Type your message here...",
show_label=False,
container=False,
scale=9
)
submit = gr.Button("Send", scale=1, variant="primary")
examples = gr.Examples(
examples=[
"What resources are available for GBV victims?",
"How can I report an incident?",
"What are my legal rights?",
"I need help, what should I do first?"
],
inputs=msg
)
# Footer with version info
gr.Markdown(f"{APP_NAME} {APP_VERSION} Β© 2025")
# Handle chat message submission
def respond(message, chat_history, session_id):
bot_message = ""
for chunk in rag_memory_stream(message, chat_history, session_id):
bot_message += chunk
chat_history.append((message, bot_message))
return "", chat_history
msg.submit(respond, [msg, chatbot, session_id], [msg, chatbot])
submit.click(respond, [msg, chatbot, session_id], [msg, chatbot])
# Handle user registration
submit_btn.click(
collect_user_info,
inputs=[first_name, session_id],
outputs=[response_message, chatbot_container, registration_container, chatbot]
)
return demo
def launch_app():
"""Launch the Gradio interface."""
ui = create_ui()
ui.launch(share=True)
# Main execution
if __name__ == "__main__":
try:
# Initialize and launch the assistant
initialize_assistant()
launch_app()
except Exception as e:
import traceback
print(f"β Fatal error initializing GBV Assistant: {e}")
print(traceback.format_exc())
# Create a minimal emergency UI to display the error
with gr.Blocks() as error_demo:
gr.Markdown("## System Error")
gr.Markdown(f"An error occurred while initializing the application: {str(e)}")
gr.Markdown("Please check your configuration and try again.")
error_demo.launch(share=True, inbrowser=True, debug=True) |