pdftotext.py

def pdfparser(fn):
    import subprocess
    '''Requires pdftotext to be installed on the system. Please change directories
    accordingly.'''
    f = "upload/" + fn # fn is the filename. upload/ is a directory, pls change.
    cmd = 'python3 upload/pdf2txt.py -o %s.txt %s' % (f.split('.')[0], f)
    run = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    out, err = run.communicate()
    # display errors if they occur
    #if err:
        #print(err)
    fileastextfile = f.split('.')[0] + ".txt"
    textfile = open(fileastextfile, 'r', encoding="utf-8")
    extractedtext = textfile.read()
    return(extractedtext)