Update app.py
Browse files
app.py
CHANGED
|
@@ -1,114 +1,273 @@
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import requests
|
|
|
|
| 3 |
import time
|
|
|
|
| 4 |
|
| 5 |
-
|
| 6 |
-
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
self.base = base.rstrip("/")
|
| 11 |
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
-
def upload(self, file_obj):
|
| 17 |
-
r = requests.post(
|
| 18 |
-
f"{self.base}/files/upload",
|
| 19 |
-
files={"file": (file_obj.name, file_obj, "application/pdf")},
|
| 20 |
-
timeout=120
|
| 21 |
-
)
|
| 22 |
-
r.raise_for_status()
|
| 23 |
-
return r.json()
|
| 24 |
-
|
| 25 |
-
def status(self, file_id):
|
| 26 |
-
r = requests.get(f"{self.base}/files/{file_id}", timeout=20)
|
| 27 |
-
r.raise_for_status()
|
| 28 |
-
return r.json()
|
| 29 |
-
|
| 30 |
-
def chat(self, file_ids, messages):
|
| 31 |
-
r = requests.post(
|
| 32 |
-
f"{self.base}/chat-with-document",
|
| 33 |
-
json={"file_ids": file_ids, "messages": messages},
|
| 34 |
-
timeout=180
|
| 35 |
-
)
|
| 36 |
-
r.raise_for_status()
|
| 37 |
-
return r.json()
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 42 |
|
| 43 |
-
|
|
|
|
|
|
|
| 44 |
if not files:
|
| 45 |
-
return "No files", [],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
-
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
status.append(f"β
{name}")
|
| 63 |
-
break
|
| 64 |
-
time.sleep(2)
|
| 65 |
-
|
| 66 |
-
return "\n".join(status), choices, store
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
def chat_fn(msg, history, selected, chat_state):
|
| 70 |
-
if not selected:
|
| 71 |
-
history.append((msg, "Select a document first"))
|
| 72 |
return history, ""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
answer = res["answer"]
|
| 79 |
-
|
| 80 |
-
chat_state.append({"role": "assistant", "content": answer})
|
| 81 |
-
history[-1] = (msg, answer)
|
| 82 |
-
|
| 83 |
-
return history, ""
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
with gr.Blocks() as demo:
|
| 87 |
-
gr.Markdown("# π RAG Document Chat")
|
| 88 |
-
|
| 89 |
-
files_state = gr.State({})
|
| 90 |
-
chat_state = gr.State([])
|
| 91 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 92 |
with gr.Row():
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
upload_btn.click(
|
| 103 |
-
|
| 104 |
-
inputs=
|
| 105 |
-
outputs=[
|
| 106 |
)
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
)
|
| 113 |
|
| 114 |
-
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
import gradio as gr
|
| 3 |
import requests
|
| 4 |
+
from typing import List, Dict
|
| 5 |
import time
|
| 6 |
+
import os
|
| 7 |
|
| 8 |
+
# Disable Gradio analytics for HF Spaces
|
| 9 |
+
os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
|
| 10 |
|
| 11 |
+
# === CONFIG ===
|
| 12 |
+
API_BASE = "https://sahana31-rag-backend.hf.space"
|
|
|
|
| 13 |
|
| 14 |
+
# === CLIENT ===
|
| 15 |
+
class DwaniClient:
|
| 16 |
+
def __init__(self, base_url=API_BASE):
|
| 17 |
+
self.base_url = base_url.rstrip('/')
|
| 18 |
+
|
| 19 |
+
def upload_file(self, file) -> dict:
|
| 20 |
+
"""Upload a file (HF temp file compatible)"""
|
| 21 |
+
files = {
|
| 22 |
+
"file": (file.name.split("/")[-1], open(file.name, "rb"), "application/pdf")
|
| 23 |
+
}
|
| 24 |
+
resp = requests.post(f"{self.base_url}/files/upload", files=files, timeout=300)
|
| 25 |
+
resp.raise_for_status()
|
| 26 |
+
return resp.json()
|
| 27 |
+
|
| 28 |
+
def get_file_status(self, file_id: str) -> dict:
|
| 29 |
+
resp = requests.get(f"{self.base_url}/files/{file_id}", timeout=60)
|
| 30 |
+
resp.raise_for_status()
|
| 31 |
+
return resp.json()
|
| 32 |
+
|
| 33 |
+
def list_files(self) -> List[dict]:
|
| 34 |
+
resp = requests.get(f"{self.base_url}/files/", timeout=60)
|
| 35 |
+
resp.raise_for_status()
|
| 36 |
+
return resp.json()
|
| 37 |
+
|
| 38 |
+
def chat(self, file_ids: List[str], messages: List[Dict]) -> dict:
|
| 39 |
+
payload = {"file_ids": file_ids, "messages": messages}
|
| 40 |
+
resp = requests.post(f"{self.base_url}/chat-with-document", json=payload, timeout=300)
|
| 41 |
+
resp.raise_for_status()
|
| 42 |
+
return resp.json()
|
| 43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
+
# === GLOBALS ===
|
| 46 |
+
client = DwaniClient()
|
| 47 |
+
uploaded_files = {}
|
| 48 |
+
chat_history: List[Dict] = []
|
| 49 |
+
selected_files = []
|
| 50 |
|
| 51 |
+
# === HELPERS ===
|
| 52 |
+
def poll_file_status(file_id: str, max_wait=60):
|
| 53 |
+
"""Wait for file processing"""
|
| 54 |
+
for _ in range(max_wait):
|
| 55 |
+
try:
|
| 56 |
+
status = client.get_file_status(file_id)
|
| 57 |
+
if status['status'] == 'completed':
|
| 58 |
+
return status, True
|
| 59 |
+
if status['status'] == 'failed':
|
| 60 |
+
return status, False
|
| 61 |
+
time.sleep(2)
|
| 62 |
+
except:
|
| 63 |
+
time.sleep(2)
|
| 64 |
+
return {'status': 'timeout'}, False
|
| 65 |
|
| 66 |
+
def create_file_list():
|
| 67 |
+
"""Display all files with status"""
|
| 68 |
+
if not uploaded_files:
|
| 69 |
+
return "No files uploaded"
|
| 70 |
+
|
| 71 |
+
lines = ["**π Your Files:**"]
|
| 72 |
+
for info in uploaded_files.values():
|
| 73 |
+
emoji = {
|
| 74 |
+
'completed': 'β
',
|
| 75 |
+
'processing': 'π',
|
| 76 |
+
'pending': 'β³',
|
| 77 |
+
'failed': 'β'
|
| 78 |
+
}.get(info['status'], 'β')
|
| 79 |
+
lines.append(f"{emoji} {info['filename']} ({info['status']})")
|
| 80 |
+
return "\n".join(lines)
|
| 81 |
|
| 82 |
+
# === UI FUNCTIONS ===
|
| 83 |
+
def upload_multiple(files):
|
| 84 |
+
"""Handle multiple PDF uploads"""
|
| 85 |
if not files:
|
| 86 |
+
return "No files selected", gr.update(choices=[]), "No files uploaded"
|
| 87 |
+
|
| 88 |
+
global uploaded_files
|
| 89 |
+
status_msgs = []
|
| 90 |
+
|
| 91 |
+
for file in files:
|
| 92 |
+
try:
|
| 93 |
+
# Upload each file
|
| 94 |
+
result = client.upload_file(file)
|
| 95 |
+
file_id = result['file_id']
|
| 96 |
+
filename = result['filename']
|
| 97 |
+
|
| 98 |
+
uploaded_files[file_id] = {
|
| 99 |
+
'filename': filename,
|
| 100 |
+
'status': 'pending',
|
| 101 |
+
'file_id': file_id
|
| 102 |
+
}
|
| 103 |
+
|
| 104 |
+
# Poll for completion
|
| 105 |
+
status, success = poll_file_status(file_id)
|
| 106 |
+
|
| 107 |
+
if success:
|
| 108 |
+
uploaded_files[file_id]['status'] = 'completed'
|
| 109 |
+
status_msgs.append(f"β
{filename} - READY")
|
| 110 |
+
else:
|
| 111 |
+
uploaded_files[file_id]['status'] = 'failed'
|
| 112 |
+
status_msgs.append(f"β {filename} - FAILED")
|
| 113 |
+
|
| 114 |
+
except Exception as e:
|
| 115 |
+
status_msgs.append(f"β {file.name} - ERROR: {str(e)}")
|
| 116 |
+
|
| 117 |
+
# Update choices for only completed files
|
| 118 |
+
choices = [(info['filename'], info['file_id']) for info in uploaded_files.values()
|
| 119 |
+
if info['status'] == 'completed']
|
| 120 |
+
|
| 121 |
+
return "\n".join(status_msgs), gr.update(choices=choices), create_file_list()
|
| 122 |
|
| 123 |
+
def refresh_files():
|
| 124 |
+
"""Refresh file list from backend"""
|
| 125 |
+
try:
|
| 126 |
+
files = client.list_files()
|
| 127 |
+
global uploaded_files
|
| 128 |
+
uploaded_files.clear()
|
| 129 |
+
|
| 130 |
+
for f in files:
|
| 131 |
+
uploaded_files[f['file_id']] = f
|
| 132 |
+
|
| 133 |
+
choices = [(f['filename'], f['file_id']) for f in files if f['status'] == 'completed']
|
| 134 |
+
return create_file_list(), gr.update(choices=choices)
|
| 135 |
+
except:
|
| 136 |
+
return "Refresh failed", gr.update()
|
| 137 |
|
| 138 |
+
def update_selected_files(files):
|
| 139 |
+
"""Update selected files"""
|
| 140 |
+
global selected_files
|
| 141 |
+
selected_files = files or []
|
| 142 |
+
return len(selected_files)
|
| 143 |
|
| 144 |
+
def format_chat_response(result):
|
| 145 |
+
"""Format response with sources"""
|
| 146 |
+
answer = result['answer']
|
| 147 |
+
if result.get('sources'):
|
| 148 |
+
sources = "\n\n**π Sources:**\n"
|
| 149 |
+
for i, src in enumerate(result['sources'][:5], 1):
|
| 150 |
+
sources += f"{i}. **{src['filename']}** (Page {src['page']})\n"
|
| 151 |
+
sources += f" > {src['excerpt'][:120]}...\n\n"
|
| 152 |
+
return answer + sources
|
| 153 |
+
return answer
|
| 154 |
|
| 155 |
+
def send_message(message, history):
|
| 156 |
+
"""Send chat message"""
|
| 157 |
+
global chat_history, selected_files
|
| 158 |
+
|
| 159 |
+
if not message.strip():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 160 |
return history, ""
|
| 161 |
+
|
| 162 |
+
if not selected_files:
|
| 163 |
+
return history, "β οΈ Please select files first!"
|
| 164 |
+
|
| 165 |
+
user_message = {"role": "user", "content": message}
|
| 166 |
+
assistant_message = {"role": "assistant", "content": "Thinking..."}
|
| 167 |
+
new_history = history + [user_message, assistant_message]
|
| 168 |
+
|
| 169 |
+
try:
|
| 170 |
+
api_messages = chat_history + [user_message]
|
| 171 |
+
result = client.chat(selected_files, api_messages)
|
| 172 |
+
|
| 173 |
+
chat_history.append(user_message)
|
| 174 |
+
chat_history.append({"role": "assistant", "content": result['answer']})
|
| 175 |
+
|
| 176 |
+
full_response = format_chat_response(result)
|
| 177 |
+
final_history = history + [user_message, {"role": "assistant", "content": full_response}]
|
| 178 |
+
return final_history, ""
|
| 179 |
+
|
| 180 |
+
except Exception as e:
|
| 181 |
+
error_response = {"role": "assistant", "content": f"β Error: {str(e)}"}
|
| 182 |
+
return new_history[:-1] + [error_response], f"Error: {str(e)}"
|
| 183 |
|
| 184 |
+
def clear_chat():
|
| 185 |
+
global chat_history
|
| 186 |
+
chat_history = []
|
| 187 |
+
return []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 188 |
|
| 189 |
+
# === GRADIO UI ===
|
| 190 |
+
with gr.Blocks(title="Dwani.ai", theme=gr.themes.Soft()) as demo:
|
| 191 |
+
gr.Markdown("# π Dwani.ai - Document Chat")
|
| 192 |
+
gr.Markdown("**Upload multiple PDFs β Chat with page-accurate citations**")
|
| 193 |
+
|
| 194 |
with gr.Row():
|
| 195 |
+
with gr.Column(scale=1):
|
| 196 |
+
gr.Markdown("## π€ Upload Multiple PDFs")
|
| 197 |
+
file_input = gr.File(
|
| 198 |
+
label="Select PDFs (Ctrl+Click for multiple)",
|
| 199 |
+
file_types=[".pdf"],
|
| 200 |
+
file_count="multiple"
|
| 201 |
+
)
|
| 202 |
+
upload_btn = gr.Button("π Upload & Process All", variant="primary")
|
| 203 |
+
status_output = gr.Markdown("Ready to upload...")
|
| 204 |
+
refresh_btn = gr.Button("π Refresh List")
|
| 205 |
+
files_display = gr.Markdown("No files uploaded")
|
| 206 |
+
|
| 207 |
+
with gr.Column(scale=2):
|
| 208 |
+
gr.Markdown("## π File Manager")
|
| 209 |
+
file_checkboxes = gr.CheckboxGroup(
|
| 210 |
+
label="Select documents to chat with:",
|
| 211 |
+
choices=[],
|
| 212 |
+
value=[],
|
| 213 |
+
info="Only completed files appear here"
|
| 214 |
+
)
|
| 215 |
+
file_count = gr.Number(label="Files selected", value=0, interactive=False)
|
| 216 |
+
|
| 217 |
+
with gr.Row():
|
| 218 |
+
gr.Markdown("## π¬ Chat with Documents")
|
| 219 |
+
chatbot = gr.Chatbot(
|
| 220 |
+
label="Ask questions about your documents",
|
| 221 |
+
height=500,
|
| 222 |
+
avatar_images=("user.png", "assistant.png")
|
| 223 |
+
)
|
| 224 |
+
|
| 225 |
+
with gr.Row():
|
| 226 |
+
msg_input = gr.Textbox(
|
| 227 |
+
label="Your question",
|
| 228 |
+
placeholder="e.g., What are the payment terms? When does the contract expire?",
|
| 229 |
+
scale=4
|
| 230 |
+
)
|
| 231 |
+
send_btn = gr.Button("Send", variant="primary", scale=1)
|
| 232 |
+
|
| 233 |
+
with gr.Row():
|
| 234 |
+
clear_btn = gr.Button("ποΈ New Chat", variant="secondary")
|
| 235 |
+
|
| 236 |
+
# === EVENTS ===
|
| 237 |
upload_btn.click(
|
| 238 |
+
upload_multiple,
|
| 239 |
+
inputs=file_input,
|
| 240 |
+
outputs=[status_output, file_checkboxes, files_display]
|
| 241 |
)
|
| 242 |
+
|
| 243 |
+
refresh_btn.click(
|
| 244 |
+
refresh_files,
|
| 245 |
+
outputs=[files_display, file_checkboxes]
|
| 246 |
+
)
|
| 247 |
+
|
| 248 |
+
file_checkboxes.change(
|
| 249 |
+
update_selected_files,
|
| 250 |
+
inputs=file_checkboxes,
|
| 251 |
+
outputs=file_count
|
| 252 |
+
)
|
| 253 |
+
|
| 254 |
+
send_btn.click(
|
| 255 |
+
send_message,
|
| 256 |
+
inputs=[msg_input, chatbot],
|
| 257 |
+
outputs=[chatbot, msg_input]
|
| 258 |
+
)
|
| 259 |
+
|
| 260 |
+
msg_input.submit(
|
| 261 |
+
send_message,
|
| 262 |
+
inputs=[msg_input, chatbot],
|
| 263 |
+
outputs=[chatbot, msg_input]
|
| 264 |
+
)
|
| 265 |
+
|
| 266 |
+
clear_btn.click(
|
| 267 |
+
clear_chat,
|
| 268 |
+
outputs=chatbot
|
| 269 |
)
|
| 270 |
|
| 271 |
+
# === LAUNCH FOR HF SPACES ===
|
| 272 |
+
if __name__ == "__main__":
|
| 273 |
+
demo.launch(show_error=True)
|