Files
notes/code/misc/python/scripts/pdfparser.py
2024-12-20 21:50:09 +00:00

63 lines
1.2 KiB
Python

import sys,os
import streamlit as st
import ptvsd
import pandas as pd
st.header('sdfsdfs')
st.write(sys.executable)
st.write(sys.version)
st.write(sys.builtin_module_names)
st.header('sdfsdfs')
#ptvsd.enable_attach()#, redirect_output=True)
#ptvsd.wait_for_attach()
print("hello World")
#####
import PyPDF2
pwd=r'n7E;~||ZOs+sDuptXAu'
filename='/home/dl92/Downloads/PGPMessage.pdf'
filename='/home/dl92/Downloads/Transaction.pdf'
pdfFileObj = open(filename,'rb') #'rb' for read binary mode
pdfFile = PyPDF2.PdfFileReader(pdfFileObj)
if pdfFile.isEncrypted:
try:
pdfFile.decrypt(pwd)
print('File Decrypted (PyPDF2)')
except:
command = ("cp "+ filename +
" temp.pdf; qpdf --password=n7E;~||ZOs+sDuptXAu --decrypt temp.pdf " + filename
+ "; rm temp.pdf")
print (command)
os.system(command)
print('File Decrypted (qpdf)')
fp = open(filename)
pdfFile = PyPDF2.PdfFileReader(fp)
print(pdfFile.numPages)
pageObj = pdfFile.getPage(2) #'9' is the page number
pageObj.extractText()
st.header('sdfsdfs')
for i in range(1, 40):
st.write(i)
import pandas as pd
print (pd.__name__)
import requests