How to use 'pdftotext python' in Python

4
def pdftotext(filename: str) -> str:
5
    """Convert a PDF file to a text equivalent.
6
7
    Args:
8
      filename: A string path, the filename to convert.
9
    Returns:
10
      A string, the text contents of the filename.
11
    """
12
13
    executable = ['pdftotext', '-layout', filename, '-']
14
    pipe = Popen(executable, stdout=PIPE, stderr=PIPE)
15
    stdout, stderr = pipe.communicate()
16
17
    if stderr:
18
        raise ValueError(stderr.decode('utf-8'))
19
    else:
20
        return stdout.decode()

946
def writetextdoctopdf(text, **kwargs):
947
948
    output = {}
949
    try:
950
        import reportlab
951
    except ImportError:
952
        output['message'] += "You do not have reportlab installed. You need to do that. "
953
        output['status'] = 1
954
        return output
955
956
    from reportlab.pdfgen import canvas
957
    from reportlab.lib.pagesizes import letter, landscape
958
    from reportlab.lib.testutils import setOutDir,makeSuiteForClasses, outputfile, printLocation
959
    from reportlab.platypus import Paragraph, SimpleDocTemplate, XBox, Indenter, XPreformatted
960
    from reportlab.lib.styles import ParagraphStyle
961
    from reportlab.lib.units import inch
962
    from reportlab.lib.colors import red, black, navy, white, green
963
    from reportlab.lib.randomtext import randomText
964
    from reportlab.rl_config import defaultPageSize
965
966
    styNormal = ParagraphStyle('normal')
967
968
    styIndent1 = ParagraphStyle('normal', leftIndent=10)
969
970
    body = []
971
972
    for index,item in enumerate(text.split('\n')):
973
        tabs = item.count('\t')
974
        thisStyle = ParagraphStyle('normal', leftIndent=10*tabs)
975
        body.append(Paragraph(item, thisStyle))
976
977
    body.append(Paragraph("""<para>\n\tThis has newlines and tabs on the front but inside the para tag</para>""", styNormal))
978
    doc = SimpleDocTemplate(kwargs['outputfile'])
979
    # doc.build(body, onFirstPage=formatted_page, onLaterPages=formatted_page)
980
    doc.build(body)
981
982
    return output