Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,39 +3,36 @@
|
|
| 3 |
|
| 4 |
import time
|
| 5 |
import traceback
|
| 6 |
-
import requests
|
| 7 |
import gradio as gr
|
| 8 |
|
| 9 |
-
# New Imports for Streaming
|
|
|
|
| 10 |
from google import genai
|
| 11 |
from groq import Groq
|
| 12 |
from groq.types.chat import ChatCompletionChunk
|
| 13 |
|
|
|
|
| 14 |
# ---------------------------
|
| 15 |
# HARDCODED KEYS (TESTING)
|
| 16 |
# ---------------------------
|
| 17 |
GEMINI_KEY = "AIzaSyAPfDiu2V_aD6un00qHt5bkISm6C0Pkx7o"
|
| 18 |
-
# NOTE: Gemini Python SDK handles the URL internally for streaming
|
| 19 |
-
# GEMINI_URL = "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent"
|
| 20 |
-
|
| 21 |
GROQ_KEY = "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"
|
| 22 |
-
# NOTE: Groq Python SDK handles the URL internally
|
| 23 |
-
GROQ_URL = "https://api.groq.com/openai/v1/chat/completions"
|
| 24 |
GROQ_MODEL = "meta-llama/llama-4-scout-17b-16e-instruct"
|
| 25 |
|
| 26 |
# Initialize Clients
|
|
|
|
|
|
|
| 27 |
try:
|
| 28 |
-
# Use the hardcoded key directly in the client for testing
|
| 29 |
GEMINI_CLIENT = genai.Client(api_key=GEMINI_KEY)
|
| 30 |
GROQ_CLIENT = Groq(api_key=GROQ_KEY)
|
| 31 |
except Exception as e:
|
| 32 |
-
# This prevents the app from crashing if keys are completely invalid/missing
|
| 33 |
print(f"WARNING: Failed to initialize one or both API clients: {e}")
|
| 34 |
GEMINI_CLIENT = None
|
| 35 |
GROQ_CLIENT = None
|
| 36 |
|
|
|
|
| 37 |
# ---------------------------
|
| 38 |
-
# Model callers (
|
| 39 |
# ---------------------------
|
| 40 |
def call_gemini_stream(message, history):
|
| 41 |
# History format conversion for the SDK
|
|
@@ -51,13 +48,12 @@ def call_gemini_stream(message, history):
|
|
| 51 |
|
| 52 |
# Use the streaming method and yield chunks
|
| 53 |
response_stream = GEMINI_CLIENT.models.generate_content_stream(
|
| 54 |
-
model="gemini-2.0-flash",
|
| 55 |
contents=contents
|
| 56 |
)
|
| 57 |
|
| 58 |
full_response = ""
|
| 59 |
for chunk in response_stream:
|
| 60 |
-
# Check if text content exists and is not empty
|
| 61 |
if chunk.text:
|
| 62 |
full_response += chunk.text
|
| 63 |
yield full_response
|
|
@@ -78,12 +74,11 @@ def call_llama_via_groq_stream(message, history):
|
|
| 78 |
response_stream = GROQ_CLIENT.chat.completions.create(
|
| 79 |
model=GROQ_MODEL,
|
| 80 |
messages=msgs,
|
| 81 |
-
stream=True
|
| 82 |
)
|
| 83 |
|
| 84 |
full_response = ""
|
| 85 |
for chunk in response_stream:
|
| 86 |
-
# Check if the chunk contains content delta
|
| 87 |
if isinstance(chunk, ChatCompletionChunk) and chunk.choices:
|
| 88 |
content = chunk.choices[0].delta.content
|
| 89 |
if content:
|
|
@@ -91,107 +86,134 @@ def call_llama_via_groq_stream(message, history):
|
|
| 91 |
yield full_response
|
| 92 |
|
| 93 |
# ---------------------------
|
| 94 |
-
# Chat function (
|
| 95 |
# ---------------------------
|
| 96 |
def chat_fn(message, history, model_choice):
|
| 97 |
try:
|
| 98 |
if model_choice == "Google Gemini 2.0 Flash":
|
| 99 |
-
# Delegate to the streaming function
|
| 100 |
yield from call_gemini_stream(message, history)
|
| 101 |
elif model_choice == "Meta LLaMA 4":
|
| 102 |
-
# Delegate to the streaming function
|
| 103 |
yield from call_llama_via_groq_stream(message, history)
|
| 104 |
else:
|
| 105 |
yield f"Unknown model: {model_choice}"
|
| 106 |
except Exception as e:
|
| 107 |
-
yield f"Error: {e}
|
| 108 |
|
| 109 |
# ---------------------------
|
| 110 |
-
# CSS (
|
| 111 |
# ---------------------------
|
| 112 |
css = """
|
| 113 |
/* Topbar layout */
|
| 114 |
-
#topbar {
|
| 115 |
-
|
| 116 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 117 |
|
| 118 |
/* ------------------------------------- */
|
| 119 |
-
/* ** DROPDOWN
|
| 120 |
/* ------------------------------------- */
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 125 |
}
|
| 126 |
|
|
|
|
| 127 |
#model_dropdown .gr-dropdown-wrap > button {
|
| 128 |
-
/* Main button styling (the collapsed box) */
|
| 129 |
background:#1a1a1a !important;
|
| 130 |
-
border:1px solid #
|
| 131 |
color:#ddd !important;
|
| 132 |
-
padding:10px
|
| 133 |
-
border-radius:
|
| 134 |
box-shadow:none !important;
|
| 135 |
-
/* Smooth transition for hover effect */
|
| 136 |
transition: all 0.2s ease-in-out;
|
|
|
|
|
|
|
| 137 |
}
|
| 138 |
#model_dropdown .gr-dropdown-wrap > button:hover {
|
| 139 |
border-color: #ffcc33 !important;
|
| 140 |
background: #252525 !important;
|
| 141 |
}
|
| 142 |
|
| 143 |
-
/* Style
|
| 144 |
.gradio-container .gr-dropdown-options {
|
| 145 |
background: #1a1a1a !important;
|
| 146 |
-
border: 1px solid #ffcc33 !important;
|
| 147 |
border-radius: 10px !important;
|
| 148 |
box-shadow: 0 4px 10px rgba(0, 0, 0, 0.5);
|
| 149 |
padding: 5px 0;
|
| 150 |
}
|
| 151 |
-
.gradio-container .gr-dropdown-options button {
|
| 152 |
-
/* Style for individual options */
|
| 153 |
-
color: #ddd !important;
|
| 154 |
-
padding: 10px 15px !important;
|
| 155 |
-
background: transparent !important;
|
| 156 |
-
transition: background 0.1s ease;
|
| 157 |
-
}
|
| 158 |
.gradio-container .gr-dropdown-options button:hover {
|
| 159 |
-
|
| 160 |
-
background: #ffcc3320 !important; /* Light yellow background on hover */
|
| 161 |
color: #fff !important;
|
| 162 |
}
|
|
|
|
| 163 |
/* ------------------------------------- */
|
| 164 |
|
|
|
|
| 165 |
/* make ChatInterface chat area taller */
|
| 166 |
.gradio-container .chat-interface .chatbot { min-height: calc(100vh - 220px); background:#111; color:#eee; }
|
| 167 |
|
| 168 |
-
/* style send button
|
| 169 |
.gr-button { border-radius:10px !important; background:#2c2c3f !important; color:#fff !important; }
|
| 170 |
"""
|
| 171 |
|
| 172 |
# ---------------------------
|
| 173 |
-
# Build UI
|
| 174 |
# ---------------------------
|
| 175 |
with gr.Blocks(css=css, title="⚡ YellowFlash.ai") as app:
|
| 176 |
with gr.Row(elem_id="topbar"):
|
| 177 |
-
|
| 178 |
-
|
| 179 |
-
|
| 180 |
-
|
| 181 |
-
elem_id="
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 186 |
gr.ChatInterface(
|
| 187 |
fn=chat_fn,
|
| 188 |
-
title="
|
| 189 |
-
description="
|
| 190 |
additional_inputs=[model_dropdown],
|
| 191 |
-
# When streaming, the Submit button automatically becomes a Stop button
|
| 192 |
)
|
| 193 |
|
| 194 |
if __name__ == "__main__":
|
| 195 |
-
# Remove share=True for Hugging Face deployment
|
| 196 |
-
# The launch() call is kept simple for the HF environment
|
| 197 |
app.launch()
|
|
|
|
| 3 |
|
| 4 |
import time
|
| 5 |
import traceback
|
|
|
|
| 6 |
import gradio as gr
|
| 7 |
|
| 8 |
+
# New Imports for Streaming (must be in your requirements.txt)
|
| 9 |
+
# pip install google-genai groq
|
| 10 |
from google import genai
|
| 11 |
from groq import Groq
|
| 12 |
from groq.types.chat import ChatCompletionChunk
|
| 13 |
|
| 14 |
+
|
| 15 |
# ---------------------------
|
| 16 |
# HARDCODED KEYS (TESTING)
|
| 17 |
# ---------------------------
|
| 18 |
GEMINI_KEY = "AIzaSyAPfDiu2V_aD6un00qHt5bkISm6C0Pkx7o"
|
|
|
|
|
|
|
|
|
|
| 19 |
GROQ_KEY = "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"
|
|
|
|
|
|
|
| 20 |
GROQ_MODEL = "meta-llama/llama-4-scout-17b-16e-instruct"
|
| 21 |
|
| 22 |
# Initialize Clients
|
| 23 |
+
# NOTE: Initialization inside the file is fine for testing, but in a real app,
|
| 24 |
+
# this should be wrapped in error handling or moved to environment variables.
|
| 25 |
try:
|
|
|
|
| 26 |
GEMINI_CLIENT = genai.Client(api_key=GEMINI_KEY)
|
| 27 |
GROQ_CLIENT = Groq(api_key=GROQ_KEY)
|
| 28 |
except Exception as e:
|
|
|
|
| 29 |
print(f"WARNING: Failed to initialize one or both API clients: {e}")
|
| 30 |
GEMINI_CLIENT = None
|
| 31 |
GROQ_CLIENT = None
|
| 32 |
|
| 33 |
+
|
| 34 |
# ---------------------------
|
| 35 |
+
# Model callers (STREAMING GENERATORS)
|
| 36 |
# ---------------------------
|
| 37 |
def call_gemini_stream(message, history):
|
| 38 |
# History format conversion for the SDK
|
|
|
|
| 48 |
|
| 49 |
# Use the streaming method and yield chunks
|
| 50 |
response_stream = GEMINI_CLIENT.models.generate_content_stream(
|
| 51 |
+
model="gemini-2.0-flash",
|
| 52 |
contents=contents
|
| 53 |
)
|
| 54 |
|
| 55 |
full_response = ""
|
| 56 |
for chunk in response_stream:
|
|
|
|
| 57 |
if chunk.text:
|
| 58 |
full_response += chunk.text
|
| 59 |
yield full_response
|
|
|
|
| 74 |
response_stream = GROQ_CLIENT.chat.completions.create(
|
| 75 |
model=GROQ_MODEL,
|
| 76 |
messages=msgs,
|
| 77 |
+
stream=True
|
| 78 |
)
|
| 79 |
|
| 80 |
full_response = ""
|
| 81 |
for chunk in response_stream:
|
|
|
|
| 82 |
if isinstance(chunk, ChatCompletionChunk) and chunk.choices:
|
| 83 |
content = chunk.choices[0].delta.content
|
| 84 |
if content:
|
|
|
|
| 86 |
yield full_response
|
| 87 |
|
| 88 |
# ---------------------------
|
| 89 |
+
# Chat function (Main generator)
|
| 90 |
# ---------------------------
|
| 91 |
def chat_fn(message, history, model_choice):
|
| 92 |
try:
|
| 93 |
if model_choice == "Google Gemini 2.0 Flash":
|
|
|
|
| 94 |
yield from call_gemini_stream(message, history)
|
| 95 |
elif model_choice == "Meta LLaMA 4":
|
|
|
|
| 96 |
yield from call_llama_via_groq_stream(message, history)
|
| 97 |
else:
|
| 98 |
yield f"Unknown model: {model_choice}"
|
| 99 |
except Exception as e:
|
| 100 |
+
yield f"Error: An API error occurred: {type(e).__name__}. Check server logs."
|
| 101 |
|
| 102 |
# ---------------------------
|
| 103 |
+
# CSS (Combined for Layout and Style)
|
| 104 |
# ---------------------------
|
| 105 |
css = """
|
| 106 |
/* Topbar layout */
|
| 107 |
+
#topbar {
|
| 108 |
+
display:flex; justify-content:space-between; align-items:center;
|
| 109 |
+
padding:18px 28px; background:#0f0f0f; border-bottom:1px solid #1f1f1f;
|
| 110 |
+
min-height: 80px;
|
| 111 |
+
}
|
| 112 |
+
|
| 113 |
+
/* Main App Title */
|
| 114 |
+
#main-app-title {
|
| 115 |
+
font-weight:800; color:#ffcc33; font-size:20px;
|
| 116 |
+
margin: 0;
|
| 117 |
+
}
|
| 118 |
|
| 119 |
/* ------------------------------------- */
|
| 120 |
+
/* ** COMPACT DROPDOWN AND TITLE AREA ** */
|
| 121 |
/* ------------------------------------- */
|
| 122 |
+
|
| 123 |
+
/* Style the static title above the dropdown (e.g., "Google") */
|
| 124 |
+
#model-title {
|
| 125 |
+
font-size: 14px;
|
| 126 |
+
font-weight: 500;
|
| 127 |
+
color: #bbb;
|
| 128 |
+
margin-bottom: 2px !important;
|
| 129 |
+
padding: 0 !important;
|
| 130 |
+
line-height: 1.2;
|
| 131 |
+
}
|
| 132 |
+
|
| 133 |
+
/* Force the entire dropdown component to a small width */
|
| 134 |
+
#model_dropdown {
|
| 135 |
+
max-width: 140px;
|
| 136 |
+
min-width: 100px;
|
| 137 |
+
}
|
| 138 |
+
|
| 139 |
+
/* Hide Gradio's default label/placeholder text if any */
|
| 140 |
+
#model_dropdown .gr-dropdown-label {
|
| 141 |
+
display: none !important;
|
| 142 |
}
|
| 143 |
|
| 144 |
+
/* Style the actual dropdown button to match the dark aesthetic */
|
| 145 |
#model_dropdown .gr-dropdown-wrap > button {
|
|
|
|
| 146 |
background:#1a1a1a !important;
|
| 147 |
+
border:1px solid #333 !important;
|
| 148 |
color:#ddd !important;
|
| 149 |
+
padding:8px 10px !important; /* Smaller padding */
|
| 150 |
+
border-radius:8px !important;
|
| 151 |
box-shadow:none !important;
|
|
|
|
| 152 |
transition: all 0.2s ease-in-out;
|
| 153 |
+
font-size: 14px;
|
| 154 |
+
height: auto !important;
|
| 155 |
}
|
| 156 |
#model_dropdown .gr-dropdown-wrap > button:hover {
|
| 157 |
border-color: #ffcc33 !important;
|
| 158 |
background: #252525 !important;
|
| 159 |
}
|
| 160 |
|
| 161 |
+
/* Style the options list when opened */
|
| 162 |
.gradio-container .gr-dropdown-options {
|
| 163 |
background: #1a1a1a !important;
|
| 164 |
+
border: 1px solid #ffcc33 !important;
|
| 165 |
border-radius: 10px !important;
|
| 166 |
box-shadow: 0 4px 10px rgba(0, 0, 0, 0.5);
|
| 167 |
padding: 5px 0;
|
| 168 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
.gradio-container .gr-dropdown-options button:hover {
|
| 170 |
+
background: #ffcc3320 !important;
|
|
|
|
| 171 |
color: #fff !important;
|
| 172 |
}
|
| 173 |
+
|
| 174 |
/* ------------------------------------- */
|
| 175 |
|
| 176 |
+
|
| 177 |
/* make ChatInterface chat area taller */
|
| 178 |
.gradio-container .chat-interface .chatbot { min-height: calc(100vh - 220px); background:#111; color:#eee; }
|
| 179 |
|
| 180 |
+
/* style send button */
|
| 181 |
.gr-button { border-radius:10px !important; background:#2c2c3f !important; color:#fff !important; }
|
| 182 |
"""
|
| 183 |
|
| 184 |
# ---------------------------
|
| 185 |
+
# Build UI (Final Layout)
|
| 186 |
# ---------------------------
|
| 187 |
with gr.Blocks(css=css, title="⚡ YellowFlash.ai") as app:
|
| 188 |
with gr.Row(elem_id="topbar"):
|
| 189 |
+
|
| 190 |
+
# 1. Left Column: Model Title and Dropdown (Compact Look)
|
| 191 |
+
with gr.Column(scale=0, min_width=150):
|
| 192 |
+
# Static title for the model vendor/family
|
| 193 |
+
gr.Markdown("**Google**", elem_id="model-title")
|
| 194 |
+
|
| 195 |
+
model_dropdown = gr.Dropdown(
|
| 196 |
+
choices=["Google Gemini 2.0 Flash", "Meta LLaMA 4"],
|
| 197 |
+
value="Google Gemini 2.0 Flash",
|
| 198 |
+
show_label=False,
|
| 199 |
+
elem_id="model_dropdown",
|
| 200 |
+
scale=0
|
| 201 |
+
)
|
| 202 |
+
|
| 203 |
+
# 2. Right Column: The main app title (YellowFlash.ai)
|
| 204 |
+
with gr.Column(scale=1):
|
| 205 |
+
gr.Markdown(f'<span id="main-app-title">⚡ YellowFlash.ai</span>', elem_id="title_md")
|
| 206 |
+
|
| 207 |
+
|
| 208 |
+
# Keep the original "under development" text
|
| 209 |
+
gr.Markdown("under development")
|
| 210 |
+
|
| 211 |
gr.ChatInterface(
|
| 212 |
fn=chat_fn,
|
| 213 |
+
title="", # Title is now handled by the custom header row
|
| 214 |
+
description="",
|
| 215 |
additional_inputs=[model_dropdown],
|
|
|
|
| 216 |
)
|
| 217 |
|
| 218 |
if __name__ == "__main__":
|
|
|
|
|
|
|
| 219 |
app.launch()
|