using subprocess.popen in python with os.tmp file while passing in optional parameters

Question

I am writing a python program in linux and in part of it running the pdftotext executable to convert a pdf text. The code I am currently using is given below.

pdfData = currentPDF.read()

tf = os.tmpfile()
tf.write(pdfData)
tf.seek(0)

out, err = subprocess.Popen(["pdftotext", "-", "-"], stdin = tf, stdout=subprocess.PIPE ).communicate()

This works fine, but now I want to run the pdftotext executable with the -layout option (preserves layout of document). I tried replacing the "-" with layout, replacing "pdftotext" with "pdftotext -layout" etc. None of it works. They all give me an empty text. Since the input is being piped in via the temp file, I am having trouble figureing out the argument list. Most of the documentation on Popen assumes all the parameters are being passed in through the argument list, but in my case the input is being passed in through the temp file.

Any help would be greatly appreciated.

Answer 1

This works for me:

out, err = subprocess.Popen(
    ["pdftotext", '-layout', "-", "-"], stdin = tf, stdout=subprocess.PIPE ).communicate()

Although I couldn't find explicit confirmation in the man page, I believe the first - tells pdftotext to expect PDF-file to come from stdin, and the second - tells pdftotext to expect text-file to be sent to stdout.

Answer 2

You can pass the full command in string with shell=True:

out, err = subprocess.Popen('pdftotext -layout - -', shell=True, stdin=tf, stdout=subprocess.PIPE).communicate()