How to upload a pdf file in streamlit

[Gyanaranjan_pathi]

How to upload a .pdf file in streamlit and then process it futher to extract the information .

[andfanilo]

Hello @Gyanaranjan_pathi, welcome to the Streamlit forums

• On the uploading part, you can use Streamlit’s file_uploader to display a file uploader on your app, as such :

import streamlit as st

uploaded_file = st.file_uploader('Choose your .pdf file', type="pdf")
if uploaded_file is not None:
    df = extract_data(uploaded_file)

• Then your PDF upload will be available as a StringIO object in the uploaded_file variable, so now to extract data from the PDF, you will need a Python library that can read your pdf as StringIO or a filelike object.

I used pdfplumber to extract tables from PDFs in one of my Streamlit apps, pdfplumber.load accepts StringIO so you can do :

def extract_data(feed):
    data = []
    with pdfplumber.load(feed) as pdf:
        pages = pdf.pages
        for p in pages:
            data.append(p.extract_tables())
    return None # build more code to return a dataframe 

but there are multiple other librairies like camelot, tabula-py or pdfminersix and I had to test multiple ones for my use case before going with pdfplumber so you may need to test multiple ones too depending on the info you need to extract !

Hope this helps

[Gyanaranjan_pathi]

Thank you @andfanilo

[santosh_boina]

@andfanilo, I came across this discussion while looking for PDF file upload and analysis. I am working on PDF files using ‘pdfminer.six’ . I could not find anything in documentation to load file(st.file_uploader object) like you mentioned for pdfplumber.

Any suggestions on handling pdf files using pdfminer.six library in streamlit app will be very helpful. Thanks:)

[andfanilo]

Don’t have a lot of experience with pdfminer.six but at least the following seems to work with Streamlit’s file uploader:

import pdfminer
from pdfminer.high_level import extract_pages
import streamlit as st

st.write(pdfminer.__version__)  

uploaded_file = st.file_uploader("Choose a file", "pdf")
if uploaded_file is not None:
    for page_layout in extract_pages(uploaded_file):
        for element in page_layout:
            st.write(element)

Hope this can serve as a good starting point.

Fanilo

[ANAND_VERMA]

How can I extract images from a pdf of images

[santosh_boina]

Hi @andfanilo , I am working on a use-case of extracting tabular data from pdf files. For this, I am using camelot as a table extraction library. How to parse the pdf uploaded through st.file_uploader() and pass it to camelot. As per my understanding from camelot documentation, camelot.read_pdf() only accepts file path as input.

[andfanilo]

Hello @santosh_boina

If it absolutely requires a filepath and not a File-related object, you could try to write the uploaded file in a temporary folder and provide camelot with the URl to said file, then destroy the temporary file at the end of the job. You can copy the following bit of code:

Problem in reading a "db" object using file uploader

Indeed, it is probably a permission issue then. Here is another version, it creates a temporary file with a random uuid name in the directory from which your have started streamlit. When the connection object is created, it removes the temporary file. import streamlit as st import sqlite3 import pathlib import uuid conn = None db = st.file_uploader("Upload a SQLite database file.", type="db") if db: fp = pathlib.Path(str(uuid.uuid4())) # fp = pathlib.Path("/path/to/your/tmpfile") …

Hope this gets you started!
Fanilo

[okld]

Hello @santosh_boina, you might have better luck with this bit of code instead, which fixes a bug in the former:

Problem in reading a "db" object using file uploader

Try that, it’s a little bit cleaner and it should work in your case. import sqlite3 import streamlit as st from contextlib import contextmanager from pathlib import Path from uuid import uuid4 @contextmanager def sqlite_connect(db_bytes): fp = Path(str(uuid4())) fp.write_bytes(db_bytes.getvalue()) conn = sqlite3.connect(str(fp)) try: yield conn finally: conn.close() fp.unlink() db = st.file_uploader("Upload a SQLite database file.", type="db") i…

[Tarun_Aswini]

How to render a pdf file in streamlit?

def show_pdf(file_path):
with open(file_path,“rb”) as f:
base64_pdf = base64.b64encode(f.read()).decode(‘utf-8’)
pdf_display = F’’
st.markdown(pdf_display, unsafe_allow_html=True)
print(‘Done’)

show_pdf('C:/Users/Tarun/Downloads/SOFTWARE ENGINEERING NOTES.pdf')

I tried this, but neither pdf displays nor any errror msg. Pls help…

[Vital-Fernandez]

Did you find a solution?

[MagicDash91]

no, it doesn’t work

[dylanxia2017]

well, pdfplumber.load just doesn’t work

[shailesh_sharma]

You can simply upload a pdf file and open it using pdfplumber

import pdfplumber
import streamlit as st

uploaded_file = st.file_uploader("Choose a file")
if uploaded_file is not None:
    st.success("Uploaded the file")
    with pdfplumber.open(uploaded_file) as file:
        all_pages = file.pages
        st.write(all_pages[0].extract_text()) # you can print and check the data from any page in pdf

[imuroo]

Here is an example of how I used the pymupdf4llm library to convert the contents of a PDF file into markdown format. I hope this helps!

# pdf file upload
pdf_file = st.file_uploader('Upload a PDF file', type=['pdf'])

if pdf_file is not None:
    bytes_data = pdf_file.read()
    
    # create a temporary file
    with tempfile.NamedTemporaryFile(delete=False) as tmp_file:
        tmp_file.write(bytes_data)
        temp_file_path = tmp_file.name

        md_text = pymupdf4llm.to_markdown(temp_file_path)
        st.markdown(md_text)