piping with subprocess

Rick Dooling rpdooling at gmail.com
Sat Feb 1 16:35:01 CET 2014


On Saturday, February 1, 2014 8:00:59 AM UTC-6, Rick Dooling wrote:
> On Saturday, February 1, 2014 7:54:34 AM UTC-6, Rick Dooling wrote:
> 
> > On Saturday, February 1, 2014 6:54:09 AM UTC-6, Peter Otten wrote:
> 
> > > Rick Dooling wrote:
> 
> > > > I spent half a day trying to convert this bash script (on Mac)
> > > > textutil -convert html $1 -stdout | pandoc -f html -t markdown -o $2
> > > > into Python using subprocess pipes.
> 
> > > > It works if I save the above into a shell script called convert.sh and
> > > > then do
> > 
> > > > subprocess.check_call(["convert.sh", file, markdown_file])
> 
> > > > where file and markdown_file are variables.
> 
> > > > But otherwise my piping attempts fail.
> 
> > > It is always a good idea to post your "best effort" failed attempt, if only to give us an idea of your level of expertise.
> > > > Could someone show me how to pipe in subprocess. Yes, I've read the doc,
> > > > especially http://docs.python.org/2/library/subprocess.html#replacing-shell-pipeline
> 
> > > > But I'm a feeble hobbyist, not a computer scientist.
> 
> > > Try to convert the example from the above page
> 
> > > """
> 
> > > output=`dmesg | grep hda`
> 
> > > # becomes
> 
> > > p1 = Popen(["dmesg"], stdout=PIPE)
> 
> > > p2 = Popen(["grep", "hda"], stdin=p1.stdout, stdout=PIPE)
> 
> > > p1.stdout.close()  # Allow p1 to receive a SIGPIPE if p2 exits.
> 
> > > output = p2.communicate()[0]
> 
> > > """
> 
> > > to your usecase. Namely, replace
> 
> > > ["dmesg"] --> ["textutil", "-convert", "html", infile, "-stdout"]
> 
> > > ["grep", "hda"] --> ["pandoc", "-f", "html", "-t", "marktown", "-o"    outfile]
> 
> > > Don't forget to set
> 
> > > infile = ... 
> 
> > > outfile = ... 
> 
> > > to filenames (with absolute paths, to avoid one source of error).
> 
> > > If that doesn't work post the code you wrote along with the error messages.
> 
> > p1 = subprocess.Popen(["textutil", "-convert", "html", file], stdout=subprocess.PIPE)
> 
> > p2 = subprocess.check_call(["pandoc", "-f", "html", "-t", "markdown", "-o", markdown_file], stdin=p1.stdout, stdout=subprocess.PIPE)
> 
> > p1.stdout.close()  # Allow p1 to receive a SIGPIPE if p2 exits.

> > output = p2.communicate()[0]
> > Errors
> > Traceback (most recent call last):
> >   File "/Users/me/Python/any2pandoc.py", line 70, in <module>
> >     convert_word_file(file, markdown_file)
> >   File "/Users/me/Python/any2pandoc.py", line 59, in convert_word_file
> >     output = p2.communicate()[0]
> > AttributeError: 'int' object has no attribute 'communicate'
> > I get a markdown_file created but it's empty.
> > Thanks,
> > RD
> > ps - Daniel's works fine but I still don't learn to pipe :)
> Okay, sorry. I fixed that obvious goof

>     p1 = subprocess.Popen(["textutil", "-convert", "html", file], stdout=subprocess.PIPE)
> 
>     p2 = subprocess.Popen(["pandoc", "-f", "html", "-t", "markdown", "-o", markdown_file], stdin=p1.stdout, stdout=subprocess.PIPE)
> 
>     p1.stdout.close()  # Allow p1 to receive a SIGPIPE if p2 exits.
> 
>     output = p2.communicate()[0]
> 
> Now I get no errors, but I still get a blank markdown file.

Okay, blank lines removed. Apologies. I didn't know Google inserted them.

RD



More information about the Python-list mailing list