Upload 2 files
Browse files- app.py +118 -0
- requirements.txt +6 -0
app.py
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
from PyPDF2 import PdfReader
|
| 3 |
+
from sentence_transformers import SentenceTransformer
|
| 4 |
+
from transformers import pipeline
|
| 5 |
+
import faiss
|
| 6 |
+
import numpy as np
|
| 7 |
+
|
| 8 |
+
# -------------------- CACHED MODELS --------------------
|
| 9 |
+
@st.cache_resource
|
| 10 |
+
def load_models():
|
| 11 |
+
embedder = SentenceTransformer("all-MiniLM-L6-v2")
|
| 12 |
+
generator = pipeline("text2text-generation", model="google/flan-t5-base")
|
| 13 |
+
return embedder, generator
|
| 14 |
+
|
| 15 |
+
embedder, generator = load_models()
|
| 16 |
+
|
| 17 |
+
# -------------------- PDF PARSING --------------------
|
| 18 |
+
def extract_pdf_text(file):
|
| 19 |
+
reader = PdfReader(file)
|
| 20 |
+
text = ""
|
| 21 |
+
for page in reader.pages:
|
| 22 |
+
content = page.extract_text()
|
| 23 |
+
if content:
|
| 24 |
+
text += content + "\n"
|
| 25 |
+
return text.strip()
|
| 26 |
+
|
| 27 |
+
# -------------------- CHUNKING --------------------
|
| 28 |
+
def chunk_text(text, chunk_size=1000, overlap=200):
|
| 29 |
+
chunks = []
|
| 30 |
+
start = 0
|
| 31 |
+
while start < len(text):
|
| 32 |
+
end = start + chunk_size
|
| 33 |
+
chunks.append(text[start:end])
|
| 34 |
+
start += chunk_size - overlap
|
| 35 |
+
return chunks
|
| 36 |
+
|
| 37 |
+
# -------------------- EMBEDDING + INDEX --------------------
|
| 38 |
+
def embed_and_index(chunks):
|
| 39 |
+
embeddings = embedder.encode(chunks, show_progress_bar=True)
|
| 40 |
+
index = faiss.IndexFlatL2(embeddings.shape[1])
|
| 41 |
+
index.add(np.array(embeddings).astype("float32"))
|
| 42 |
+
return index, embeddings
|
| 43 |
+
|
| 44 |
+
def search(query, chunks, index, top_k=3):
|
| 45 |
+
q_emb = embedder.encode([query])
|
| 46 |
+
D, I = index.search(np.array(q_emb).astype("float32"), k=top_k)
|
| 47 |
+
return [chunks[i] for i in I[0]]
|
| 48 |
+
|
| 49 |
+
# -------------------- SUMMARIZATION --------------------
|
| 50 |
+
def summarize_text(text):
|
| 51 |
+
"""Summarize large text by chunking it for the generator model"""
|
| 52 |
+
summaries = []
|
| 53 |
+
chunks = chunk_text(text, chunk_size=800)
|
| 54 |
+
for chunk in chunks:
|
| 55 |
+
try:
|
| 56 |
+
summary = generator(
|
| 57 |
+
f"Summarize this text:\n\n{chunk}",
|
| 58 |
+
max_length=150,
|
| 59 |
+
min_length=30,
|
| 60 |
+
do_sample=False
|
| 61 |
+
)[0]["generated_text"]
|
| 62 |
+
summaries.append(summary)
|
| 63 |
+
except Exception as e:
|
| 64 |
+
summaries.append(f"[Error: {e}]")
|
| 65 |
+
return " ".join(summaries)
|
| 66 |
+
|
| 67 |
+
# -------------------- STREAMLIT UI --------------------
|
| 68 |
+
st.set_page_config(page_title="PDF Summarizer + RAG QA", page_icon="π", layout="wide")
|
| 69 |
+
|
| 70 |
+
st.title("π Studymate β PDF Summarizer + AI Q&A")
|
| 71 |
+
st.write("Upload a PDF, then choose to summarize it or ask questions using AI (RAG pipeline).")
|
| 72 |
+
|
| 73 |
+
pdf_file = st.file_uploader("Upload a PDF file", type=["pdf"])
|
| 74 |
+
|
| 75 |
+
if pdf_file:
|
| 76 |
+
with st.spinner("Extracting text from PDF..."):
|
| 77 |
+
text = extract_pdf_text(pdf_file)
|
| 78 |
+
|
| 79 |
+
if text:
|
| 80 |
+
st.success(f"β
Extracted {len(text)} characters from PDF.")
|
| 81 |
+
|
| 82 |
+
# Chunk + embed
|
| 83 |
+
with st.spinner("Chunking and embedding document..."):
|
| 84 |
+
chunks = chunk_text(text)
|
| 85 |
+
index, embeddings = embed_and_index(chunks)
|
| 86 |
+
st.info("Document indexed and ready for queries!")
|
| 87 |
+
|
| 88 |
+
# Tabs for Summarization and QA
|
| 89 |
+
tab1, tab2 = st.tabs(["π Summarization", "π¬ Q&A Chat"])
|
| 90 |
+
|
| 91 |
+
with tab1:
|
| 92 |
+
st.subheader("π Summarize PDF")
|
| 93 |
+
if st.button("Generate Summary"):
|
| 94 |
+
with st.spinner("Summarizing document..."):
|
| 95 |
+
summary = summarize_text(text)
|
| 96 |
+
st.subheader("π§ Summary:")
|
| 97 |
+
st.write(summary)
|
| 98 |
+
|
| 99 |
+
with tab2:
|
| 100 |
+
st.subheader("π¬ Ask Questions")
|
| 101 |
+
query = st.text_input("Enter your question about the PDF:")
|
| 102 |
+
if query:
|
| 103 |
+
with st.spinner("Retrieving relevant information..."):
|
| 104 |
+
relevant_chunks = search(query, chunks, index)
|
| 105 |
+
context = " ".join(relevant_chunks)
|
| 106 |
+
with st.spinner("Generating answer..."):
|
| 107 |
+
answer = generator(
|
| 108 |
+
f"Answer the question based on the following context:\n\n{context}\n\nQuestion: {query}",
|
| 109 |
+
max_length=200,
|
| 110 |
+
min_length=50,
|
| 111 |
+
do_sample=False
|
| 112 |
+
)[0]["generated_text"]
|
| 113 |
+
st.subheader("π€ Answer:")
|
| 114 |
+
st.write(answer)
|
| 115 |
+
else:
|
| 116 |
+
st.error("Could not extract text from this PDF.")
|
| 117 |
+
else:
|
| 118 |
+
st.info("Please upload a PDF file to start.")
|
requirements.txt
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit
|
| 2 |
+
PyPDF2
|
| 3 |
+
sentence-transformers
|
| 4 |
+
transformers
|
| 5 |
+
faiss-cpu
|
| 6 |
+
torch
|