Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from langchain.document_loaders import PyPDFLoader
|
3 |
+
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
4 |
+
from langchain.embeddings import OpenAIEmbeddings
|
5 |
+
from langchain.vectorstores import FAISS
|
6 |
+
from langchain.chains import RetrievalQA
|
7 |
+
from langchain.llms import OpenAI
|
8 |
+
|
9 |
+
# Initialize the FAISS vector store
|
10 |
+
vector_store = None
|
11 |
+
|
12 |
+
# Function to handle PDF upload and indexing
|
13 |
+
def index_pdf(pdf):
|
14 |
+
global vector_store
|
15 |
+
|
16 |
+
# Load the PDF
|
17 |
+
loader = PyPDFLoader(pdf.name)
|
18 |
+
documents = loader.load()
|
19 |
+
|
20 |
+
# Split the documents into chunks
|
21 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
|
22 |
+
texts = text_splitter.split_documents(documents)
|
23 |
+
|
24 |
+
# Embed the chunks and store them in the vector store
|
25 |
+
embeddings = OpenAIEmbeddings()
|
26 |
+
vector_store = FAISS.from_documents(texts, embeddings)
|
27 |
+
|
28 |
+
return "PDF indexed successfully!"
|
29 |
+
|
30 |
+
# Function to handle chatbot queries
|
31 |
+
def chatbot_query(query):
|
32 |
+
if vector_store is None:
|
33 |
+
return "Please upload and index a PDF first."
|
34 |
+
|
35 |
+
# Create a retrieval-based QA chain
|
36 |
+
retriever = vector_store.as_retriever()
|
37 |
+
qa_chain = RetrievalQA(llm=OpenAI(), retriever=retriever)
|
38 |
+
|
39 |
+
# Get the response from the QA chain
|
40 |
+
response = qa_chain.run(query)
|
41 |
+
|
42 |
+
return response
|
43 |
+
|
44 |
+
# Create the Gradio interface
|
45 |
+
with gr.Blocks() as demo:
|
46 |
+
with gr.Tab("Indexing"):
|
47 |
+
pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
|
48 |
+
index_button = gr.Button("Index PDF")
|
49 |
+
index_output = gr.Textbox(label="Indexing Status")
|
50 |
+
|
51 |
+
index_button.click(index_pdf, inputs=pdf_input, outputs=index_output)
|
52 |
+
|
53 |
+
with gr.Tab("Chatbot"):
|
54 |
+
query_input = gr.Textbox(label="Enter your question")
|
55 |
+
query_button = gr.Button("Submit")
|
56 |
+
query_output = gr.Textbox(label="Response")
|
57 |
+
|
58 |
+
query_button.click(chatbot_query, inputs=query_input, outputs=query_output)
|
59 |
+
|
60 |
+
# Launch the Gradio app
|
61 |
+
demo.launch()
|