BeatDebate / src /ui /chat_interface.py
SulmanK's picture
Update competition links in chat interface to reflect new resources - Changed the AgentX submission link to the updated URL and modified the GitHub repository link for better accuracy. This update aims to ensure users have access to the correct and current resources related to the project.
3620bb2
Raw
History Blame Contribute Delete
36.6 kB
"""
Gradio ChatInterface for BeatDebate Music Recommendation System
This module provides a ChatGPT-style interface that showcases the 4-agent
planning system with real-time progress indicators and planning visualization.
"""
import logging
from typing import Dict, List, Optional, Tuple, Any
import gradio as gr
import requests
from .response_formatter import ResponseFormatter
from .planning_display import PlanningDisplay
# Import fallback service components
from ..services.llm_fallback_service import (
LLMFallbackService,
FallbackRequest,
FallbackTrigger
)
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
# Query examples from intent-aware recommendation system design document
QUERY_EXAMPLES = {
"By Artist": [
"Songs by Mk.gee",
"Give me tracks by Radiohead",
"Play some Beatles songs"
],
"Artist Similarity": [
"Songs like Mk.gee",
"Similar artists to BROCKHAMPTON",
"Songs that sound like Radiohead"
],
"Discovery": [
"Find me underground electronic music",
"Something completely new and different",
"Discover underground tracks by Kendrick Lamar"
],
"Genre/Mood": [
"Upbeat electronic music",
"Sad indie songs",
"Chill lo-fi hip hop"
],
"Contextual": [
"Music for studying",
"Workout playlist songs",
"Background music for coding"
],
"Hybrid": [
"Songs like Kendrick Lamar but jazzy",
"Songs by Michael Jackson that are R&B",
"Electronic music similar to Aphex Twin"
],
"Follow-ups": [
"More tracks",
"More like that",
"Similar to these",
"More from this artist",
"More underground like these",
"More for studying like these"
]
}
class BeatDebateChatInterface:
"""
ChatGPT-style interface for BeatDebate music recommendations.
Features:
- Real-time agent progress indicators
- Planning strategy visualization
- Audio preview integration
- Conversation history management
- Last.fm player embeds
- LLM fallback for unknown intents
"""
def __init__(self, backend_url: str = "http://localhost:8000"):
"""
Initialize the chat interface.
Args:
backend_url: URL of the FastAPI backend
"""
self.backend_url = backend_url
self.response_formatter = ResponseFormatter()
self.planning_display = PlanningDisplay()
# βœ… REMOVED: Global session management - now handled per-user via gr.State
# βœ… REMOVED: Global conversation history - now managed by backend session store
# Initialize fallback service
self.fallback_service = None
self._initialize_fallback_service()
logger.info(
f"BeatDebate Chat Interface initialized (multi-user safe), "
f"fallback_available: {self.fallback_service is not None}"
)
def _initialize_fallback_service(self) -> None:
"""Initialize the LLM fallback service."""
try:
# Import Gemini client creation function
from ..services.enhanced_recommendation_service import create_gemini_client
from ..api.rate_limiter import UnifiedRateLimiter
import os
# Get Gemini API key
gemini_api_key = os.getenv('GEMINI_API_KEY', 'demo_gemini_key')
if gemini_api_key and gemini_api_key != 'demo_gemini_key':
# Create Gemini client
gemini_client = create_gemini_client(gemini_api_key)
if gemini_client:
# Create rate limiter for fallback service
rate_limiter = UnifiedRateLimiter.for_gemini(calls_per_minute=8)
# Initialize fallback service
self.fallback_service = LLMFallbackService(
gemini_client=gemini_client,
rate_limiter=rate_limiter
)
logger.info("LLM fallback service initialized successfully")
else:
logger.warning("Failed to create Gemini client for fallback service")
else:
logger.warning("No valid Gemini API key found, fallback service disabled")
except Exception as e:
logger.error(f"Failed to initialize fallback service: {e}")
self.fallback_service = None
async def process_message(
self,
message: str,
history: List[Tuple[str, str]],
session_id: str
) -> Tuple[str, List[Tuple[str, str]], str, str]:
"""
Process user message and return response with track info.
Enhanced with fallback support and per-user session management.
Args:
message: User input message
history: Chat history as list of (user, assistant) tuples
session_id: User-specific session identifier
Returns:
Tuple of (response, updated_history, lastfm_player_html, updated_session_id)
"""
if not message.strip():
return "", history, "", session_id
logger.info(f"Processing message: {message} for session: {session_id}")
try:
# Primary: Get recommendations from 4-agent system with session ID
recommendations_response = await self._get_recommendations(message, session_id)
# Check if fallback is needed
should_fallback, trigger_reason = self._should_use_fallback(
recommendations_response
)
if should_fallback:
logger.info(f"Using LLM fallback for session {session_id} due to: {trigger_reason.value}")
recommendations_response = await self._get_fallback_recommendations(
message, trigger_reason, session_id
)
if recommendations_response:
# Get potentially updated session_id from backend
updated_session_id = recommendations_response.get("session_id", session_id)
# Format the response
formatted_response = (
self.response_formatter.format_recommendations(
recommendations_response
)
)
# Add to history using tuple format
history.append((message, formatted_response))
# Create Last.fm player HTML for latest recommendations
lastfm_player_html = self._create_lastfm_player_html(
recommendations_response.get("recommendations", [])
)
return "", history, lastfm_player_html, updated_session_id
else:
# Final emergency fallback
error_response = self._create_emergency_response(message)
history.append((message, error_response))
return "", history, "", session_id
except Exception as e:
logger.error(f"Error processing message for session {session_id}: {e}")
error_response = f"An error occurred: {str(e)}"
history.append((message, error_response))
return "", history, "", session_id
def _should_use_fallback(
self,
response: Optional[Dict]
) -> Tuple[bool, FallbackTrigger]:
"""
Determine if fallback should be used based on backend response.
Args:
response: Response from backend recommendation system
Returns:
Tuple of (should_fallback, trigger_reason)
"""
if response is None:
return True, FallbackTrigger.API_ERROR
# Check for explicit unknown intent
intent = response.get("intent", "").lower()
if intent in ["unknown", "unsupported", "fallback"]:
return True, FallbackTrigger.UNKNOWN_INTENT
# Check for empty recommendations
recommendations = response.get("recommendations", [])
if not recommendations or len(recommendations) == 0:
return True, FallbackTrigger.NO_RECOMMENDATIONS
# Check for error indicators
if response.get("error") or response.get("detail"):
return True, FallbackTrigger.API_ERROR
return False, None
async def _get_fallback_recommendations(
self,
query: str,
trigger_reason: FallbackTrigger,
session_id: str
) -> Optional[Dict[str, Any]]:
"""
Get fallback recommendations from LLM service.
Args:
query: User query
trigger_reason: Reason fallback was triggered
session_id: User-specific session identifier
Returns:
Fallback recommendations response or None if unavailable
"""
if not self.fallback_service:
logger.warning("Fallback service not available")
return None
try:
# Prepare fallback request
fallback_request = FallbackRequest(
query=query,
session_id=session_id,
chat_context=self._get_chat_context(),
trigger_reason=trigger_reason,
max_recommendations=10
)
# Get fallback recommendations
fallback_response = await self.fallback_service.get_fallback_recommendations(
fallback_request
)
# Add fallback disclaimer to explanation
if fallback_response and fallback_response.get("fallback_used"):
original_explanation = fallback_response.get("explanation", "")
fallback_explanation = (
f"**⚠️ DEFAULTING TO REGULAR LLM** - This query is outside our "
f"specialized 4-agent system's scope.\n\n{original_explanation}"
)
fallback_response["explanation"] = fallback_explanation
return fallback_response
except Exception as e:
logger.error(f"Fallback service failed: {e}")
return None
def _get_chat_context(self) -> Optional[Dict]:
"""Get chat context for fallback requests."""
# βœ… UPDATED: Chat context now managed entirely by backend session store
# The backend will retrieve session history based on session_id
# Frontend no longer maintains global conversation history
return None
def _create_emergency_response(self, query: str) -> str:
"""Create emergency response when all systems fail."""
return (
"**🚨 SYSTEM TEMPORARILY UNAVAILABLE**\n\n"
f"I apologize, but I'm unable to process your request for '{query}' "
"at the moment. Our recommendation systems are experiencing issues.\n\n"
"**Please try:**\n"
"- Waiting a few moments and trying again\n"
"- Simplifying your query (e.g., 'music like [artist name]')\n"
"- Checking your internet connection\n\n"
"We're working to restore full functionality. Thank you for your patience! 🎡"
)
async def _get_planning_strategy(self, query: str) -> Optional[Dict]:
"""Get planning strategy from backend."""
try:
response = requests.post(
f"{self.backend_url}/planning",
json={
"query": query,
"session_id": "planning-session" # Planning doesn't need user session
},
timeout=60
)
if response.status_code == 200:
return response.json()
else:
logger.error(
f"Planning request failed: {response.status_code}"
)
return None
except Exception as e:
logger.error(f"Error getting planning strategy: {e}")
return None
async def _get_recommendations(self, query: str, session_id: str) -> Optional[Dict]:
"""Get recommendations from backend with user-specific session context."""
try:
# Prepare request with user-specific session ID
request_data = {
"query": query,
"session_id": session_id,
"max_recommendations": 10,
"include_previews": True
}
# βœ… UPDATED: Chat context now retrieved by backend from session store
# No need to pass chat_context explicitly - backend handles it
response = requests.post(
f"{self.backend_url}/recommendations",
json=request_data,
timeout=120
)
if response.status_code == 200:
response_data = response.json()
return response_data
else:
logger.error(
f"Recommendations request failed: {response.status_code} for session {session_id}"
)
return None
except Exception as e:
logger.error(f"Error getting recommendations for session {session_id}: {e}")
return None
def _create_lastfm_player_html(self, recommendations: List[Dict]) -> str:
"""Create HTML for track preview links and info."""
if not recommendations:
return """
<div style="
padding: 20px;
text-align: center;
color: #cbd5e1;
background: rgba(30, 41, 59, 0.5);
border-radius: 0 0 12px 12px;
">
<p><em>No tracks yet!</em></p>
<p>Ask for music recommendations to see track info here</p>
</div>
"""
player_html = []
# Container with scrolling for all tracks
player_html.append("""
<div style="
max-height: 400px;
overflow-y: auto;
border-radius: 0 0 12px 12px;
background: rgba(30, 41, 59, 0.5);
">
""")
# Show all tracks with rank numbers
for i, rec in enumerate(recommendations):
rank = i + 1
title = rec.get("title", "Unknown Title")
artist = rec.get("artist", "Unknown Artist")
confidence = rec.get("confidence", 0.0)
confidence_pct = int(confidence * 100)
# Create search queries
search_query = f"{artist} {title}".replace(" ", "+")
lastfm_url = f"https://www.last.fm/search?q={search_query}"
spotify_url = f"https://open.spotify.com/search/{search_query}"
youtube_url = f"https://www.youtube.com/results?search_query={search_query}"
# Dark mode compatible colors based on confidence
if confidence_pct >= 70:
border_color = "#10b981" # emerald
bg_gradient = "linear-gradient(135deg, rgba(16, 185, 129, 0.15) 0%, rgba(5, 150, 105, 0.15) 100%)"
elif confidence_pct >= 50:
border_color = "#f59e0b" # amber
bg_gradient = "linear-gradient(135deg, rgba(245, 158, 11, 0.15) 0%, rgba(217, 119, 6, 0.15) 100%)"
else:
border_color = "#ef4444" # red
bg_gradient = "linear-gradient(135deg, rgba(239, 68, 68, 0.15) 0%, rgba(220, 38, 38, 0.15) 100%)"
player_html.append(f"""
<div style="
margin: 12px 20px;
padding: 15px;
background: {bg_gradient};
border-left: 4px solid {border_color};
border-radius: 8px;
box-shadow: 0 2px 8px rgba(0,0,0,0.3);
transition: all 0.2s ease;
border: 1px solid rgba(75, 85, 99, 0.3);
position: relative;
">
<!-- Rank Number -->
<div style="
position: absolute;
top: -8px;
left: 15px;
background: {border_color};
color: white;
width: 24px;
height: 24px;
border-radius: 50%;
display: flex;
align-items: center;
justify-content: center;
font-size: 12px;
font-weight: bold;
box-shadow: 0 2px 4px rgba(0,0,0,0.3);
">
{rank}
</div>
<div style="
display: flex;
justify-content: space-between;
align-items: center;
margin-bottom: 8px;
margin-top: 8px;
">
<div style="
font-weight: 700;
color: #ffffff;
font-size: 15px;
text-shadow:
0 1px 3px rgba(0,0,0,0.8),
0 2px 6px rgba(0,0,0,0.6),
1px 1px 2px rgba(0,0,0,0.9);
">
{artist}
</div>
<div style="
background: {border_color};
color: white;
padding: 3px 10px;
border-radius: 12px;
font-size: 12px;
font-weight: bold;
text-shadow: 0 1px 2px rgba(0,0,0,0.3);
">
{confidence_pct}%
</div>
</div>
<div style="
color: #ffffff;
margin-bottom: 10px;
font-size: 14px;
font-weight: 600;
text-shadow:
0 1px 3px rgba(0,0,0,0.8),
0 2px 6px rgba(0,0,0,0.6),
1px 1px 2px rgba(0,0,0,0.9);
">
{title}
</div>
<div style="
display: flex;
gap: 8px;
font-size: 12px;
">
<a href="{lastfm_url}" target="_blank" style="
color: #fca5a5;
text-decoration: none;
padding: 4px 8px;
border-radius: 4px;
background: rgba(0, 0, 0, 0.3);
border: 1px solid rgba(239, 68, 68, 0.5);
transition: all 0.2s ease;
font-weight: 500;
">🎡 Last.fm</a>
<a href="{spotify_url}" target="_blank" style="
color: #86efac;
text-decoration: none;
padding: 4px 8px;
border-radius: 4px;
background: rgba(0, 0, 0, 0.3);
border: 1px solid rgba(34, 197, 94, 0.5);
transition: all 0.2s ease;
font-weight: 500;
">🎧 Spotify</a>
<a href="{youtube_url}" target="_blank" style="
color: #fda4af;
text-decoration: none;
padding: 4px 8px;
border-radius: 4px;
background: rgba(0, 0, 0, 0.3);
border: 1px solid rgba(244, 63, 94, 0.5);
transition: all 0.2s ease;
font-weight: 500;
">πŸ“Ί YouTube</a>
</div>
</div>
""")
# Close the scrollable container
player_html.append("</div>")
return ''.join(player_html)
def create_interface(self) -> gr.Blocks:
"""
Create the Gradio interface.
Returns:
Gradio Blocks interface
"""
with gr.Blocks(
title="🎡 BeatDebate",
theme=gr.themes.Soft(
primary_hue="violet",
secondary_hue="blue",
neutral_hue="slate"
),
css="""
.main-container {
max-width: 1400px;
margin: 0 auto;
padding: 20px;
}
.header-section {
text-align: center;
margin-bottom: 30px;
padding: 20px;
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
color: white;
border-radius: 15px;
box-shadow: 0 4px 15px rgba(0,0,0,0.3);
}
.examples-section {
margin: 20px 0;
padding: 20px;
background: rgba(55, 65, 81, 0.8);
border-radius: 12px;
border: 1px solid rgba(75, 85, 99, 0.5);
}
.examples-section h2 {
color: #f8fafc !important;
margin-bottom: 15px;
}
.examples-section h3 {
color: #e2e8f0 !important;
margin-bottom: 10px;
}
.example-chip {
display: inline-block;
margin: 4px;
padding: 10px 16px;
background: rgba(30, 41, 59, 0.9) !important;
border: 1px solid rgba(100, 116, 139, 0.5) !important;
border-radius: 20px;
font-size: 14px;
font-weight: 500;
color: #f1f5f9 !important;
cursor: pointer;
transition: all 0.2s ease;
letter-spacing: 0.025em;
}
.example-chip:hover {
background: rgba(51, 65, 85, 0.9) !important;
border-color: rgba(139, 92, 246, 0.7) !important;
transform: translateY(-1px);
box-shadow: 0 4px 12px rgba(139, 92, 246, 0.2);
font-weight: 600;
}
.chat-container {
background: rgba(55, 65, 81, 0.8) !important;
border: 1px solid rgba(75, 85, 99, 0.5) !important;
border-radius: 15px 15px 0 0 !important;
border-bottom: none !important;
min-height: 400px !important;
}
.chat-header {
padding: 15px 20px;
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
color: white;
text-align: center;
font-weight: 600;
font-size: 16px;
border-radius: 15px 15px 0 0;
margin: 0;
pointer-events: none;
}
.input-section {
margin-top: 0;
border-radius: 0 0 15px 15px;
background: rgba(55, 65, 81, 0.8);
padding: 20px;
border-top: 1px solid rgba(75, 85, 99, 0.5);
}
.info-container {
background: rgba(30, 41, 59, 0.6);
border-radius: 12px;
border: 1px solid rgba(75, 85, 99, 0.3);
margin-left: 20px;
}
.agent-info {
background: rgba(55, 65, 81, 0.8);
padding: 15px;
border-radius: 10px;
margin-bottom: 15px;
border: 1px solid rgba(75, 85, 99, 0.5);
color: #f1f5f9;
}
.agent-info h3 {
color: #f8fafc !important;
}
.agent-info ul li {
color: #e2e8f0 !important;
}
/* Global dark mode overrides */
.gradio-container {
background: #0f172a !important;
color: #f1f5f9 !important;
}
/* Input styling */
.gr-textbox input {
background: rgba(30, 41, 59, 0.9) !important;
border: 1px solid rgba(100, 116, 139, 0.5) !important;
color: #f1f5f9 !important;
}
.gr-textbox input::placeholder {
color: #94a3b8 !important;
}
/* Button styling */
.gr-button {
background: linear-gradient(135deg, #8b5cf6 0%, #7c3aed 100%) !important;
border: none !important;
color: white !important;
}
.gr-button:hover {
background: linear-gradient(135deg, #7c3aed 0%, #6d28d9 100%) !important;
box-shadow: 0 4px 12px rgba(139, 92, 246, 0.3) !important;
}
"""
) as interface:
# βœ… ADD: Per-user session state management
# NOTE: gr.State() without value generates unique session per user
import uuid
session_id_state = gr.State()
with gr.Column(elem_classes=["main-container"]):
# Header with gradient background
with gr.Column(elem_classes=["header-section"]):
gr.Markdown("""
# 🎡 BeatDebate
### AI Music Discovery with Intent-Aware 4-Agent Recommendation System
**πŸ† AgentX Competition Entry** | **πŸš€ Advanced Agentic Planning System**
Discover perfect tracks using our sophisticated 4-agent AI system that understands your musical intent and collaborates intelligently!
**πŸ”— Competition Links:** [AgentX Submission](https://rdi.berkeley.edu/agentx/) β€’ [GitHub Repository](https://github.com/SulmanK/BeatDebate)
""")
# Query examples prominently displayed
with gr.Column(elem_classes=["examples-section"]):
gr.Markdown("## πŸ’‘ **Try These Examples** - Click any to get started!")
gr.Markdown("*Our system recognizes all these intent types and optimizes agent coordination accordingly*")
example_buttons = []
with gr.Row():
with gr.Column(scale=1):
gr.Markdown("**🎯 By Artist**")
for example in QUERY_EXAMPLES["By Artist"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
with gr.Column(scale=1):
gr.Markdown("**🎯 Artist Similarity**")
for example in QUERY_EXAMPLES["Artist Similarity"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
with gr.Column(scale=1):
gr.Markdown("**🎯 Discovery**")
for example in QUERY_EXAMPLES["Discovery"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
with gr.Row():
with gr.Column(scale=1):
gr.Markdown("**🎡 Genre/Mood**")
for example in QUERY_EXAMPLES["Genre/Mood"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
with gr.Column(scale=1):
gr.Markdown("**πŸ“ Contextual**")
for example in QUERY_EXAMPLES["Contextual"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
with gr.Column(scale=1):
gr.Markdown("**🎭 Hybrid Queries**")
for example in QUERY_EXAMPLES["Hybrid"]:
btn = gr.Button(
example,
elem_classes=["example-chip"],
size="sm",
variant="secondary"
)
example_buttons.append((btn, example))
# Main content area
with gr.Row():
with gr.Column(scale=7):
# Fixed chat header
gr.HTML(
"""<div class="chat-header">🎡 Music Recommendations</div>""",
elem_classes=[]
)
# Chat interface
chatbot = gr.Chatbot(
label="",
height=500,
elem_classes=["chat-container"],
show_label=False,
container=False,
render_markdown=True
)
# Input area connected to chat
with gr.Column(elem_classes=["input-section"]):
with gr.Row():
msg_input = gr.Textbox(
placeholder="What music are you in the mood for?",
label="",
scale=4,
lines=1,
show_label=False
)
send_btn = gr.Button(
"Send",
scale=1,
variant="primary",
size="lg"
)
with gr.Column(scale=3):
# Track info with improved styling
with gr.Column(elem_classes=["info-container"]):
gr.HTML("""
<div style="
padding: 20px;
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
color: white;
border-radius: 12px 12px 0 0;
text-align: center;
font-weight: 600;
font-size: 16px;
">
<h3 style="margin: 0; font-size: 18px; font-weight: 600;">🎧 Latest Tracks</h3>
<p style="margin: 8px 0 0 0; opacity: 0.9; font-size: 14px;">Click links to listen!</p>
</div>
""")
player_display = gr.HTML(
label="",
elem_classes=[]
)
# Agent system info with improved styling
with gr.Column(elem_classes=["agent-info"]):
gr.Markdown("""
### πŸ€– **AI Agent System**
- **🧠 Planner**: Analyzes your query intent
- **🎡 GenreMood**: Finds style/vibe matches
- **πŸ” Discovery**: Uncovers hidden gems
- **βš–οΈ Judge**: Ranks & selects best tracks
""")
# Event handlers with session state management
async def handle_message(message, history, session_id):
# Generate session ID if not already set (first interaction)
if session_id is None:
session_id = str(uuid.uuid4())
logger.info(f"πŸ†• Generated new session ID: {session_id}")
return await self.process_message(message, history, session_id)
# Example button handlers
for btn, example_text in example_buttons:
btn.click(
fn=lambda x=example_text: x,
inputs=[],
outputs=[msg_input]
)
# Submit on button click or enter with session state
send_btn.click(
fn=handle_message,
inputs=[msg_input, chatbot, session_id_state],
outputs=[msg_input, chatbot, player_display, session_id_state]
)
msg_input.submit(
fn=handle_message,
inputs=[msg_input, chatbot, session_id_state],
outputs=[msg_input, chatbot, player_display, session_id_state]
)
return interface
def create_chat_interface(backend_url: str = "http://localhost:8000") -> gr.Blocks:
"""
Factory function to create the BeatDebate chat interface.
Args:
backend_url: URL of the FastAPI backend
Returns:
Gradio Blocks interface
"""
chat_interface = BeatDebateChatInterface(backend_url)
return chat_interface.create_interface()
if __name__ == "__main__":
# For testing the interface standalone
interface = create_chat_interface()
interface.launch(
server_name="0.0.0.0",
server_port=7860,
share=False
)