Spaces:
Sleeping
Sleeping
Chandranshu Jain
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -10,6 +10,7 @@ from langchain.prompts import PromptTemplate
|
|
10 |
from langchain_community.document_loaders import PyPDFLoader
|
11 |
from langchain_chroma import Chroma
|
12 |
import io
|
|
|
13 |
|
14 |
st.set_page_config(page_title="Document Genie", layout="wide")
|
15 |
|
@@ -34,14 +35,8 @@ Follow these simple steps to interact with the chatbot:
|
|
34 |
|
35 |
def get_pdf(pdf_file):
|
36 |
# Extract file path from UploadedFile object
|
37 |
-
|
38 |
-
|
39 |
-
pdf_content = pdf_file.read()
|
40 |
-
# Close the file object
|
41 |
-
pdf_file.close()
|
42 |
-
# Load the PDF content using PyPDFLoader
|
43 |
-
loader = PyPDFLoader(io.BytesIO(pdf_content))
|
44 |
-
docs = loader.load()
|
45 |
return docs
|
46 |
|
47 |
def text_splitter(text):
|
|
|
10 |
from langchain_community.document_loaders import PyPDFLoader
|
11 |
from langchain_chroma import Chroma
|
12 |
import io
|
13 |
+
import fitz
|
14 |
|
15 |
st.set_page_config(page_title="Document Genie", layout="wide")
|
16 |
|
|
|
35 |
|
36 |
def get_pdf(pdf_file):
|
37 |
# Extract file path from UploadedFile object
|
38 |
+
file_pdf = fitz.open(stream=pdf_file.read(), filetype="pdf")
|
39 |
+
docs = file_pdf.load()
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
return docs
|
41 |
|
42 |
def text_splitter(text):
|