Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,193 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import gradio as gr
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from llama_index.core import SimpleDirectoryReader, VectorStoreIndex
|
| 5 |
+
|
| 6 |
+
# Ensure data directory exists
|
| 7 |
+
Path("data").mkdir(exist_ok=True)
|
| 8 |
+
|
| 9 |
+
# Set OpenAI API key
|
| 10 |
+
os.environ["OPENAI_API_KEY"] = os.getenv("OPENAI_API_KEY")
|
| 11 |
+
|
| 12 |
+
# Global state for engines
|
| 13 |
+
state = {"insurance_engine": None, "paul_engine": None, "custom_engine": None}
|
| 14 |
+
|
| 15 |
+
# Build index from file
|
| 16 |
+
def build_index_from_file(file_path):
|
| 17 |
+
documents = SimpleDirectoryReader(input_files=[file_path]).load_data()
|
| 18 |
+
index = VectorStoreIndex.from_documents(documents)
|
| 19 |
+
return index.as_query_engine()
|
| 20 |
+
|
| 21 |
+
# Preload fixed documents
|
| 22 |
+
def preload_indexes():
|
| 23 |
+
try:
|
| 24 |
+
state["insurance_engine"] = build_index_from_file("data/insurance_FirstRAG.pdf")
|
| 25 |
+
except:
|
| 26 |
+
print("β Failed to load insurance.pdf")
|
| 27 |
+
|
| 28 |
+
try:
|
| 29 |
+
state["paul_engine"] = build_index_from_file("data/paul_graham_FirstRAG.txt")
|
| 30 |
+
except:
|
| 31 |
+
print("β Failed to load paul_graham.txt")
|
| 32 |
+
|
| 33 |
+
# Upload custom file
|
| 34 |
+
def upload_and_refresh(file_path):
|
| 35 |
+
if file_path is None:
|
| 36 |
+
return "β οΈ No file uploaded."
|
| 37 |
+
try:
|
| 38 |
+
state["custom_engine"] = build_index_from_file(file_path)
|
| 39 |
+
return f"β
Uploaded & Indexed: {os.path.basename(file_path)}"
|
| 40 |
+
except Exception as e:
|
| 41 |
+
return f"β Failed: {str(e)}"
|
| 42 |
+
|
| 43 |
+
# Query helpers
|
| 44 |
+
def ask_insurance(query): return _query_engine(state["insurance_engine"], query)
|
| 45 |
+
def ask_paul(query): return _query_engine(state["paul_engine"], query)
|
| 46 |
+
def ask_custom(query): return _query_engine(state["custom_engine"], query)
|
| 47 |
+
|
| 48 |
+
def _query_engine(engine, query):
|
| 49 |
+
if not query:
|
| 50 |
+
return "β Please enter a question."
|
| 51 |
+
if engine is None:
|
| 52 |
+
return "π Document not loaded yet."
|
| 53 |
+
try:
|
| 54 |
+
return str(engine.query(query))
|
| 55 |
+
except Exception as e:
|
| 56 |
+
return f"β Error: {str(e)}"
|
| 57 |
+
|
| 58 |
+
# Summarize
|
| 59 |
+
def summarize_insurance(): return _query_engine(state["insurance_engine"], "Summarize the insurance document.")
|
| 60 |
+
def summarize_paul(): return _query_engine(state["paul_engine"], "Summarize Paul Graham's main ideas.")
|
| 61 |
+
def summarize_custom(): return _query_engine(state["custom_engine"], "Summarize the uploaded documents.")
|
| 62 |
+
|
| 63 |
+
# Clear
|
| 64 |
+
def clear_fields(): return "", ""
|
| 65 |
+
|
| 66 |
+
# Launch app
|
| 67 |
+
def launch():
|
| 68 |
+
preload_indexes()
|
| 69 |
+
|
| 70 |
+
with gr.Blocks(
|
| 71 |
+
title="RAG App with LlamaIndex",
|
| 72 |
+
css="""
|
| 73 |
+
body {
|
| 74 |
+
background-color: #f5f5dc;
|
| 75 |
+
font-family: 'Georgia', 'Merriweather', serif;
|
| 76 |
+
}
|
| 77 |
+
h1 {
|
| 78 |
+
font-size: 2.5em;
|
| 79 |
+
font-weight: bold;
|
| 80 |
+
color: #4e342e;
|
| 81 |
+
margin-bottom: 0.3em;
|
| 82 |
+
}
|
| 83 |
+
.subtitle {
|
| 84 |
+
font-size: 1.2em;
|
| 85 |
+
color: #6d4c41;
|
| 86 |
+
margin-bottom: 1.5em;
|
| 87 |
+
}
|
| 88 |
+
.gr-box, .gr-column, .gr-group {
|
| 89 |
+
border-radius: 15px;
|
| 90 |
+
padding: 20px;
|
| 91 |
+
background-color: #fffaf0;
|
| 92 |
+
box-shadow: 2px 4px 14px rgba(0, 0, 0, 0.1);
|
| 93 |
+
margin-top: 10px;
|
| 94 |
+
}
|
| 95 |
+
textarea, input[type="text"], input[type="file"] {
|
| 96 |
+
background-color: #fffaf0;
|
| 97 |
+
border: 1px solid #d2b48c;
|
| 98 |
+
color: #4e342e;
|
| 99 |
+
border-radius: 8px;
|
| 100 |
+
}
|
| 101 |
+
button {
|
| 102 |
+
background-color: #a1887f;
|
| 103 |
+
color: white;
|
| 104 |
+
font-weight: bold;
|
| 105 |
+
border-radius: 8px;
|
| 106 |
+
transition: background-color 0.3s ease;
|
| 107 |
+
}
|
| 108 |
+
button:hover {
|
| 109 |
+
background-color: #8d6e63;
|
| 110 |
+
}
|
| 111 |
+
.gr-button {
|
| 112 |
+
border-radius: 8px !important;
|
| 113 |
+
}
|
| 114 |
+
.tabitem {
|
| 115 |
+
border-radius: 15px !important;
|
| 116 |
+
}
|
| 117 |
+
"""
|
| 118 |
+
) as app:
|
| 119 |
+
|
| 120 |
+
with gr.Column():
|
| 121 |
+
gr.Markdown("""
|
| 122 |
+
<div style='text-align: center;'>
|
| 123 |
+
<h1>RAG Application with LlamaIndex</h1>
|
| 124 |
+
<div class='subtitle'>π Ask questions from documents using Retrieval-Augmented Generation (RAG)</div>
|
| 125 |
+
</div>
|
| 126 |
+
""")
|
| 127 |
+
|
| 128 |
+
# π Insurance Tab
|
| 129 |
+
with gr.Tab("π Insurance Summary"):
|
| 130 |
+
with gr.Column():
|
| 131 |
+
gr.Markdown("### π‘οΈ Ask about Insurance Document")
|
| 132 |
+
with gr.Group():
|
| 133 |
+
insurance_q = gr.Textbox(label="Your Question", placeholder="e.g., What does this cover?")
|
| 134 |
+
with gr.Row():
|
| 135 |
+
insurance_ask = gr.Button("Submit")
|
| 136 |
+
insurance_clear = gr.Button("Clear")
|
| 137 |
+
insurance_ans = gr.Textbox(label="Response", lines=6)
|
| 138 |
+
insurance_ask.click(fn=ask_insurance, inputs=insurance_q, outputs=insurance_ans)
|
| 139 |
+
insurance_clear.click(fn=clear_fields, outputs=[insurance_q, insurance_ans])
|
| 140 |
+
|
| 141 |
+
with gr.Group():
|
| 142 |
+
summarize_btn = gr.Button("Summarize Insurance Document")
|
| 143 |
+
summarize_output = gr.Textbox(label="Summary", lines=6)
|
| 144 |
+
summarize_btn.click(fn=summarize_insurance, outputs=summarize_output)
|
| 145 |
+
|
| 146 |
+
# π§ Paul Graham Tab
|
| 147 |
+
with gr.Tab("π§ Paul Graham"):
|
| 148 |
+
with gr.Column():
|
| 149 |
+
gr.Markdown("### π§ Ask about Paul Graham's Writings")
|
| 150 |
+
with gr.Group():
|
| 151 |
+
paul_q = gr.Textbox(label="Your Question", placeholder="e.g., What does Paul say about startups?")
|
| 152 |
+
with gr.Row():
|
| 153 |
+
paul_ask = gr.Button("Ask")
|
| 154 |
+
paul_clear = gr.Button("Clear")
|
| 155 |
+
paul_ans = gr.Textbox(label="Response", lines=6)
|
| 156 |
+
paul_ask.click(fn=ask_paul, inputs=paul_q, outputs=paul_ans)
|
| 157 |
+
paul_clear.click(fn=clear_fields, outputs=[paul_q, paul_ans])
|
| 158 |
+
|
| 159 |
+
with gr.Group():
|
| 160 |
+
summarize_btn2 = gr.Button("Summarize Paul Graham's Ideas")
|
| 161 |
+
summarize_output2 = gr.Textbox(label="Summary", lines=6)
|
| 162 |
+
summarize_btn2.click(fn=summarize_paul, outputs=summarize_output2)
|
| 163 |
+
|
| 164 |
+
# π€ Upload Custom
|
| 165 |
+
with gr.Tab("π€ Upload & Ask Your File"):
|
| 166 |
+
with gr.Column():
|
| 167 |
+
gr.Markdown("### π€ Upload Your Own Document")
|
| 168 |
+
with gr.Group():
|
| 169 |
+
file_input = gr.File(label="Upload PDF or TXT", type="filepath")
|
| 170 |
+
status = gr.Textbox(label="Status", interactive=False)
|
| 171 |
+
file_input.change(fn=upload_and_refresh, inputs=file_input, outputs=status)
|
| 172 |
+
|
| 173 |
+
with gr.Group():
|
| 174 |
+
gr.Markdown("#### π¬ Ask a Question")
|
| 175 |
+
custom_q = gr.Textbox(label="Your Query")
|
| 176 |
+
with gr.Row():
|
| 177 |
+
custom_ask = gr.Button("Ask")
|
| 178 |
+
custom_clear = gr.Button("Clear")
|
| 179 |
+
custom_ans = gr.Textbox(label="Response", lines=6)
|
| 180 |
+
custom_ask.click(fn=ask_custom, inputs=custom_q, outputs=custom_ans)
|
| 181 |
+
custom_clear.click(fn=clear_fields, outputs=[custom_q, custom_ans])
|
| 182 |
+
|
| 183 |
+
with gr.Group():
|
| 184 |
+
gr.Markdown("#### π Summarize the File")
|
| 185 |
+
summarize = gr.Button("Summarize Uploaded Document")
|
| 186 |
+
summary_output = gr.Textbox(label="Summary", lines=6)
|
| 187 |
+
summarize.click(fn=summarize_custom, outputs=summary_output)
|
| 188 |
+
|
| 189 |
+
app.launch()
|
| 190 |
+
|
| 191 |
+
|
| 192 |
+
# Run the app
|
| 193 |
+
launch()
|