Is there a way in Linux taking a terminal/SSH stream as an input to another program.

I can get the following to work without issue:

#!/bin/bash
FIFO="$1"
ps auwx | grep "[m]rLogger 192.168.10.10"
if [ $? -eq 0 ]; then
    exit
fi
while [ 1 ]
do
  /usr/bin/sshpass -p MrLoggertemp /usr/bin/ssh -t -t -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -l mrlogger 192.168.10.10 >> "$FIFO"
  echo "sshpass failed. restarting"
  sleep 1
done

.. and it will stream to a file

But I want to stream this to a perl script that I want to do some processing and then load into a dataset.

Is there a way to do the following in the perl script like:

my $fifo_fh;
my $fifo_file = <The output of the stream above>
open($fifo_fh, " < $fifo_file") or die

Thanks..

CodePudding user response：

I am not sure exactly what your command "stream"s or what exactly you need with it.

But yes a Perl script can act as a filter, in a number of ways.

If you want to pipe a stream into the scirpt then read it from STDIN (perl's handle for fd 0)

use warnings;
use strict;
use feature 'say';

while (my $line_in = <STDIN>) { 
    chomp $line_in;   # remove newline
    # ...             # process
    say $line_in;     # prints for demo
}

Now with the command

ls -l . | script.pl

the lines of output from ls -l . are processed in the script.pl (only printed above).

So with your example instead of cmd >> "$FIFO" have cmd | script.pl and in the script write (append) to the file, after suitable processing.

If you'd rather do all this from inside a script, a basic way is to use a "pipe-open"

use warnings;
use strict;
use feature 'say';

my @cmd = qw(ls -l .);

open(my $in, '-|', @cmd) or die "Can't pipe-open @cmd: $!";

while (my $line_in = <$in>) { 
    chomp $line_in;
    # ...
    say $line_in;
}

With the command passed as a list (@cmd) it is assumed that the first element of @cmd is the program to run, which is directly invoked and the rest of @cmd is passed to it as its arguments. This way a shell is altogether avoided, even if there are shell metacharacters in the command.

But if the command is meant to use a shell then write it as a string, or pass it as "@cmd", where quoting interpolates @cmd into a string with spaces between elements. See linked docs.

Then, there are libraries that facilitate and improve this. I'd first recommend IPC::Run; it is the most complex one but also by far the most powerful, allowing your program to run a mini-shell almost.^†

Let me know how this meshes (or not) with your uses so that we can adjust it, if needed.

---

When a loop condition is written as while (<$fh>) then a line that it reads, via <> operator from a filehandle $fh, gets assigned to an omnipresent default, $_ variable. (If we explicitly assign to a variable, like I do above for clarity, then the deal is off and $_ is undefined.)

That variable, $_, is a default for many other operators, including chomp and say.

Then examples above can be written as

while (<$in>) { 
    chomp;
    ...      # process $_, which has the line of input
    say;
}

This can lead to very lean and readable code, if used correctly. But if we end up needing a lot of explicit uses of $_ in the loop body (not everything takes it by default!), or end up with cryptic code, then please by all means introduce and use a nice lexical variable.

---

^† Basic uses are simple.

CodePudding user response：

Another skeleton approach use Linux names pipes.
It is real time communication via remote server.

https://en.wikipedia.org/wiki/Named_pipe

mkfifo streamFile    # make pipe file.
exec 3<> streamFile  # change descriptor to not block r/w

# /* Here is if any line appears on streamFile it will pipe to Perl.
#    Whatever content appears on remote /var/log/messages pass to $line

cat streamFile - | ssh user@server  'tail -f /var/log/messages' |  while read line ; do 
    echo $line | perl;   # /* of course it fail, becasue perl expect differnt syntax. \
                         # /* You can change another file which tail file for perl lang
done