update app.py

app.py

@@ -1,16 +1,39 @@
 import streamlit as st
-from PyPDF2 import PdfFileReader
+from langchain.document_loaders import PyPDFLoader
+
+from langchain.text_splitter import SentenceTransformersTokenTextSplitter
+from langchain.embeddings import HuggingFaceEmbeddings, SentenceTransformerEmbeddings
+from langchain.vectorstores import FAISS
+
+
 
 def read_pdf(file):
-    pdf = PdfFileReader(file)
-    text = ""
-    for page in range(pdf.getNumPages()):
-        text += pdf.getPage(page).extractText()
-    return text
+    loader = PyPDFLoader(file)
+    raw_documents = loader.load()
+    return raw_documents
 
 st.title('PDF Text Extractor')
 
 uploaded_file = st.file_uploader("Choose a PDF file", type="pdf")
 if uploaded_file is not None:
     text = read_pdf(uploaded_file)
-    st.write(text)
+
+
+splitter = SentenceTransformersTokenTextSplitter(model_name='dangvantuan/sentence-camembert-large',
+chunk_overlap=50
+)
+
+documents = splitter.split_documents(raw_documents)
+
+
+embeddings_fun = HuggingFaceEmbeddings(model_name='dangvantuan/sentence-camembert-large')
+
+# embeddings_text = embeddings_fun.embed_documents(documents)
+
+faiss_db = FAISS.from_documents(documents, embeddings_fun)
+
+query = st.text_input("Entrer une question")
+docs = faiss_db.similarity_search(query)
+
+st.text('La reponse à votre question:')
+st.write(docs[0].page_content)