Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import tempfile
|
| 3 |
+
from utils import extract_pdf_by_page, split_by_chapter, build_faiss_index, retrieve_text, generate_notes_questions
|
| 4 |
+
|
| 5 |
+
chunks = []
|
| 6 |
+
faiss_index = None
|
| 7 |
+
chunk_texts = []
|
| 8 |
+
groq_api_key = ""
|
| 9 |
+
|
| 10 |
+
def upload_pdf(file, use_chapter_split):
|
| 11 |
+
global chunks, faiss_index, chunk_texts
|
| 12 |
+
with tempfile.NamedTemporaryFile(delete=False) as tmp:
|
| 13 |
+
tmp.write(file.read())
|
| 14 |
+
tmp_path = tmp.name
|
| 15 |
+
|
| 16 |
+
pages = extract_pdf_by_page(tmp_path)
|
| 17 |
+
chunks = split_by_chapter(pages) if use_chapter_split else pages
|
| 18 |
+
faiss_index, chunk_texts, chunks = build_faiss_index(chunks)
|
| 19 |
+
return f"β
Uploaded and indexed {len(chunks)} chunks."
|
| 20 |
+
|
| 21 |
+
def set_api_key(api_key):
|
| 22 |
+
global groq_api_key
|
| 23 |
+
groq_api_key = api_key
|
| 24 |
+
return "β
API key set!"
|
| 25 |
+
|
| 26 |
+
def process_query(query):
|
| 27 |
+
if not groq_api_key:
|
| 28 |
+
return "β Please provide a valid Groq API key."
|
| 29 |
+
results = retrieve_text(query, faiss_index, chunk_texts, chunks)
|
| 30 |
+
output = generate_notes_questions(results[0]['text'], groq_api_key)
|
| 31 |
+
return f"π **{results[0].get('title', 'Page')}** (pages: {results[0].get('pages')})\n\n{output}"
|
| 32 |
+
|
| 33 |
+
with gr.Blocks(title="RAG PDF Notes & Questions Generator") as demo:
|
| 34 |
+
gr.Markdown("# π PDF Chapter Notes & Question Generator using Groq LLM")
|
| 35 |
+
|
| 36 |
+
with gr.Row():
|
| 37 |
+
pdf = gr.File(label="Upload PDF", file_types=[".pdf"])
|
| 38 |
+
use_chapter = gr.Checkbox(label="Split by Chapter", value=True)
|
| 39 |
+
upload_btn = gr.Button("Process PDF")
|
| 40 |
+
|
| 41 |
+
output_text = gr.Textbox(label="Upload Status")
|
| 42 |
+
upload_btn.click(fn=upload_pdf, inputs=[pdf, use_chapter], outputs=output_text)
|
| 43 |
+
|
| 44 |
+
api_key_input = gr.Textbox(label="π Groq API Key", type="password")
|
| 45 |
+
api_btn = gr.Button("Set API Key")
|
| 46 |
+
api_output = gr.Textbox(label="API Key Status")
|
| 47 |
+
api_btn.click(fn=set_api_key, inputs=[api_key_input], outputs=api_output)
|
| 48 |
+
|
| 49 |
+
query_input = gr.Textbox(label="Enter chapter title or page keyword")
|
| 50 |
+
query_btn = gr.Button("Generate Notes & Questions")
|
| 51 |
+
result_output = gr.Markdown()
|
| 52 |
+
query_btn.click(fn=process_query, inputs=[query_input], outputs=[result_output])
|
| 53 |
+
|
| 54 |
+
demo.launch()
|