Use capture_both with Feeder

[Jeet Ray]

Hello! Would it be possible to use the feeder with capture_both? I would like to store the result in a variable until I can display or return it manually! My current code is something like this:

from sarge import Feeder, get_stdout, capture_both

feeder = Feeder()
for line in capture_both("cowsay", input=feeder, async_=True).stdout:
    print(line.decode("utf-8").rstrip())
feeder.feed(get_stdout("fortune"))
feeder.close()

Thank you kindly for the help!

[Vinay Sajip]

I’m not sure I quite get what you mean. Can you show a snippet which illustrates your desired usage, even if it doesn’t work at the moment?

[Jeet Ray]

Essentially, all I want to do is store the resultant value for later use, perhaps as a generator or similar, as I can't store the result of run().

[Vinay Sajip]

The resultant value of what? capture_both and run both return Pipeline instances. Why can't you use get_both if you just want the final output streams?

[Jeet Ray]

But does feeder not provide an alternative to piping commands, such as command 1 | command 2? In addition, if I wanted to save memory, wouldn’t capture_both be better? I do apologize; I may be completely misunderstanding the use of feeder.

[Vinay Sajip]

I may be completely misunderstanding the use of feeder.

Perhaps you are - I can't really say. Feeder is I believe sufficiently documented, so I propose to close this issue as Invalid because I can’t see what enhancement you are looking for. Perhaps you can re-frame it so that it makes more sense to me? Or try asking on the forum.

[Jeet Ray]

Alright, so I’ll try one last time before heading to the forum:

So as far as I can tell, you can use Feeder to pass in large amounts of input data to a child process, i.e. through the stdin of the second process; however, you can only print it to the terminal using run, and not store the result of the combined commands, such storing the result of fortune | cowsay, for later formatting. I see Feeder as a more efficient and safer way of piping than the above fortune | cowsay, preventing premature closure of pipes before all the data has been passed to the child / second process. Am I correct in this statement?

[Vinay Sajip]

It's not intended to replace fortune | cowsay - it's where you need to generate data in some other way (e.g. programmatically) to feed to a child process. Otherwise you could just build a command line using e.g. fortune | cowsay | your-script and run that (or capture_both, or get_both). If you want to store results, use a two-step process e.g. get_both with fortune | cowsay to get your intermediate data, and then pass that to your-script, possibly using Feeder. Otherwise (to access the data a bit at a time), you would have to use piecemeal reading of the stdout of the fortune | cowsay pipeline.

[Jeet Ray]

Hmm... I'm currently using sarge to create a module which essentially runs shell programs as functions, similar to http://amoffat.github.io/sh/; to implement piping, I am required to attach pipes to each argument using string manipulation; would Feeder be a suitable alternative, if my situation counts as programmatically?