I want to create a simple IO object that represents a pipe opened to another program to that I can periodically write to another program's STDIN as my app runs. I want it to be bullet-proof (in that it catches all errors) and cross-platform. The best options I can find are:
open
sub io_read {
local $SIG{__WARN__} = sub { }; # Silence warning.
open my $pipe, '|-', #_ or die "Cannot exec $_[0]: $!\n";
return $pipe;
}
Advantages:
Cross-platform
Simple
Disadvantages
No $SIG{PIPE} to catch errors from the piped program
Are other errors caught?
IO::Pipe
sub io_read {
IO::Pipe->reader(#_);
}
Advantages:
Simple
Returns an IO::Handle object for OO interface
Supported by the Perl core.
Disadvantages
Still No $SIG{PIPE} to catch errors from the piped program
Not supported on Win32 (or, at least, its tests are skipped)
IPC::Run
There is no interface for writing to a file handle in IPC::Run, only appending to a scalar. This seems…weird.
IPC::Run3
No file handle interface here, either. I could use a code reference, which would be called repeatedly to spool to the child, but looking at the source code, it appears that it actually writes to a temporary file, and then opens it and spools its contents to the pipe'd command's STDIN. Wha?
IPC::Cmd
Still no file handle interface.
What am I missing here? It seems as if this should be a solved problem, and I'm kind of stunned that it's not. IO::Pipe comes closest to what I want, but the lack of $SIG{PIPE} error handling and the lack of support for Windows is distressing. Where is the piping module that will JDWIM?
Thanks to guidance from #ikegami, I have found that the best choice for interactively reading from and writing to another process in Perl is IPC::Run. However, it requires that the program you are reading from and writing to have a known output when it is done writing to its STDOUT, such as a prompt. Here's an example that executes bash, has it run ls -l, and then prints that output:
use v5.14;
use IPC::Run qw(start timeout new_appender new_chunker);
my #command = qw(bash);
# Connect to the other program.
my ($in, #out);
my $ipc = start \#command,
'<' => new_appender("echo __END__\n"), \$in,
'>' => new_chunker, sub { push #out, #_ },
timeout(10) or die "Error: $?\n";
# Send it a command and wait until it has received it.
$in .= "ls -l\n";
$ipc->pump while length $in;
# Wait until our end-of-output string appears.
$ipc->pump until #out && #out[-1] =~ /__END__\n/m;
pop #out;
say #out;
Because it is running as an IPC (I assume), bash does not emit a prompt when it is done writing to its STDOUT. So I use the new_appender() function to have it emit something I can match to find the end of the output (by calling echo __END__). I've also used an anonymous subroutine after a call to new_chunker to collect the output into an array, rather than a scalar (just pass a reference to a scalar to '>' if you want that).
So this works, but it sucks for a whole host of reasons, in my opinion:
There is no generally useful way to know that an IPC-controlled program is done printing to its STDOUT. Instead, you have to use a regular expression on its output to search for a string that usually means it's done.
If it doesn't emit one, you have to trick it into emitting one (as I have done here—god forbid if I should have a file named __END__, though). If I was controlling a database client, I might have to send something like SELECT 'IM OUTTA HERE';. Different applications would require different new_appender hacks.
The writing to the magic $in and $out scalars feels weird and action-at-a-distance-y. I dislike it.
One cannot do line-oriented processing on the scalars as one could if they were file handles. They are therefore less efficient.
The ability to use new_chunker to get line-oriented output is nice, if still a bit weird. That regains a bit of the efficiency on reading output from a program, though, assuming it is buffered efficiently by IPC::Run.
I now realize that, although the interface for IPC::Run could potentially be a bit nicer, overall the weaknesses of the IPC model in particular makes it tricky to deal with at all. There is no generally-useful IPC interface, because one has to know too much about the specifics of the particular program being run to get it to work. This is okay, maybe, if you know exactly how it will react to inputs, and can reliably recognize when it is done emitting output, and don't need to worry much about cross-platform compatibility. But that was far from sufficient for my need for a generally useful way to interact with various database command-line clients in a CPAN module that could be distributed to a whole host of operating systems.
In the end, thanks to packaging suggestions in comments on a blog post, I decided to abandon the use of IPC for controlling those clients, and to use the DBI, instead. It provides an excellent API, robust, stable, and mature, and suffers none of the drawbacks of IPC.
My recommendation for those who come after me is this:
If you just need to execute another program and wait for it to finish, or collect its output when it is done running, use IPC::System::Simple. Otherwise, if what you need to do is to interactively interface with something else, use an API whenever possible. And if it's not possible, then use something like IPC::Run and try to make the best of it—and be prepared to give up quite a bit of your time to get it "just right."
I've done something similar to this. Although it depends on the parent program and what you are trying to pipe. My solution was to spawn a child process (leaving $SIG{PIPE} to function) and writing that to the log, or handling the error in what way you see fit. I use POSIX to handle my child process and am able to utilize all the functionality of the parent. However if you're trying to have the child communicate back to the parent - then things get difficult. Do you have an example of the main program and what you're trying to PIPE?
Related
Assume that I have following code to open filehandle:
open my $fh, "command1 | command2 |";
I found command1 may output that command2 can not handle well, so I'm trying to insert a perl filter between command1 and command2 to deal with them:
use Config;
open my $fh, "command1 | $Config{perlpath} -ple 'blah blah' | command2 |";
My questions are:
Is it OK to use $Config{perlpath} in system call directly?
Calling own perl binary seems nuts. Are there any better solutions?
Thanks
Is it OK to use $Config{perlpath} in system call directly?
Relatively. It's about as portable as the rest of your code (which already depends on running on something unix-ish). There's some security worry, but I'd say not a very large one because someone who can affect the value of that variable already has scope to cause havoc. $^X is probably safer in that regard. You might want to try quoting it using String::ShellQuote just for safety, since you can't bypass the shell in the midst of a pipeline.
Calling own perl binary seems nuts. Are there any better solutions?
Depends on your definition of "better". There's definitely another way around, which is to run both command1 and command2 separately, process command1's output in your original perl process, hand it to command2, and read command2's output. However, you have to be careful how you do it. There are two safe ways.
The first way is easier, but takes more time and memory: run command1 first, read and process all of its output into a string, then run command2 providing the buffered output as input. The best way to do this is to use IPC::Run to handle command2's input and output (and maybe both commands, just for consistency); if you try to just print all the data to command2's input handle and then read all the output you can deadlock, but IPC::Run will interleave reads and writes if necessary behind the scenes, and give you a nice easy interface.
The second way is more complicated but closer in behavior to the original: you need some kind of async framework like IO::Async or POE, using its classes for process construction, and set up handlers to communicate between them, do your filtering, and gather the output.
Here's a tested toy example of that (gist because it's a couple screenfuls of code) that does the equivalent of ls | perl -pe '$_ = uc' | rev, except with the middle part of the pipeline running in the parent perl process. You may never use it, but I thought it was worth illustrating.
I have a Perl script that received input piped from another program. It's buffering with an 8k (Ubuntu default) input buffer, which is causing problems. I'd like to use line buffering or disable buffering completely. It doesn't look like there's a good way to do this. Any suggestions?
use IO::Handle;
use IO::Poll qw[ POLLIN POLLHUP POLLERR ];
use Text::CSV;
my $stdin = new IO::Handle;
$stdin->fdopen(fileno(STDIN), 'r');
$stdin->setbuf(undef);
my $poll = IO::Poll->new() or die "cannot create IO::Poll object";
$poll->mask($stdin => POLLIN);
STDIN->blocking(0);
my $halt = 0;
for(;;) {
$poll->poll($config{poll_timout});
for my $handle ($poll->handles(POLLIN | POLLHUP | POLLERR)) {
next unless($handle eq $stdin);
if(eof) {
$halt = 1;
last;
}
my #row = $csv->getline($stdin);
# Do more stuff here
}
last if($halt);
}
Polling STDIN kind of throws a wrench into things since IO::Poll uses buffering and direct calls like sysread do not (and they can't mix). I don't want to infinitely call sysread without no blocking. I require the use of select or poll since I don't want to hammer the CPU.
PLEASE NOTE: I'm talking about STDIN, NOT STDOUT. $|++ is not the solution.
[EDIT]
Updating my question to clarify based on the comments and other answers.
The program that is writing to STDOUT (on the other side of the pipe) is line buffered and flushed after every write. Every write contains a newline, so in effect, buffering is not an issue for STDOUT of the first program.
To verify this is true, I wrote a small C program that reads piped input from the same program with STDIN buffering disabled (setvbuf with _IONBF). The input appears in STDIN of the test program immediately. Sadly, it does not appear to be an issue with the output from the first program.
[/EDIT]
Thanks for any insight!
PS. I have done a fair amount of Googling. This link is the closest I've found to an answer, but it certainly doesn't satisfy all my needs.
Say there are two short lines in the pipe's buffer.
IO::Poll notifies you there's data to read, which you proceed to read (indirectly) using readline.
Reading one character at a time from a file handle is very inefficient. As such, readline (aka <>) reads a block of data from the file handle at a time. The two lines ends up in a buffer and the first of the two lines is returned.
Then you wait for IO::Poll to notify you that there is more data. It doesn't know about Perl's buffer; it just knows the pipe is empty. As such, it blocks.
This post demonstrates the problem. It uses IO::Select, but the principle (and solution) is the same.
You're actually talking about the other program's STDOUT. The solution is $|=1; (or equivalent) in the other program.
If you can't, you might be able to convince the other program use line-buffering instead of block buffering by connecting its STDOUT to a pseudo-tty instead of a pipe (like Expect.pm does, for example).
The unix program expect has a tool called unbuffer which does that exactly that. (It's part of the expect-dev package on Ubuntu.) Just prefix the command name with unbuffer.
In Perl, to run another Perl script from my script, or to run any system commands like mv, cp, pkgadd, pkgrm, pkginfo, rpm etc, we can use the following:
system()
exec()
`` (Backticks)
Are all the three the same, or are they different? Do all the three give the same result in every case? Are they used in different scenarios, like to call a Perl program we have to use system() and for others we have to use ``(backticks).
Please advise, as I am currently using system() for all the calls.
They're all different, and the docs explain how they're different. Backticks capture and return output; system returns an exit status, and exec never returns at all if it's successful.
IPC::System::Simple is probably what you want.
It provides safe, portable alternatives to backticks, system() and other IPC commands.
It also allows you to avoid the shell for most of said commands, which can be helpful in some circumstances.
The best option is to use some module, either in the standard library or from CPAN, that does the job for you. It's going to be more portable, and possibly even faster for quick tasks (no forking to the system).
However, if that's not good enough for you, you can use one of those three, and no, they are not the same. Read the perldoc pages on system(), exec(), and backticks to see the difference.
Calling system is generally a mistake. For instance, instead of saying
system "mv $foo /tmp" == 0
or die "could not move $foo to /tmp";
system "cp $foo /tmp" == 0
or die "could not copy $foo to /tmp";
you should say
use File::Copy;
move $foo, "/tmp"
or die "could not move $foo to /tmp: $!";
copy $foo, "/tmp"
or die "could not copy $foo to /tmp: $!";
Look on CPAN for modules that handle other commands for you. If you find yourself writing a lot of calls to system, it may be time to drop back into a shell script instead.
Well, the more people the more answers.
My answer is to generally avoid external commands execution. If you can - use modules. There is no point executing "cp", "mv" and a lot of another command - there exist modules that do it. And the benefit of using modules is that they usually work cross-platform. While your system("mv") might not.
When put in situation that I have no other way, but to call external command, I prefer to use IPC::Run. The idea is that all simplistic methods (backticks, qx, system, open with pipe) are inherently insecure, and require attention to parameters.
With IPC::Run, I run commands like I would do with system( #array ), which is much more secure, and I can bind to stdin, stdout and stderr separately, using variables, or callbacks, which is very cool when you'll be in situation that you have to pass data to external program from long-running code.
Also, it has built-in handling of timeouts, which come handy more than once :)
If you don't want the shell getting involved (usually you don't) and if waiting for the system command is acceptable, I recommend using IPC::Run3. It is simple, flexible, and does the common task of executing a program, feeding it input and capturing its output and error streams right.
I've got a Perl script that needs to execute another Perl script. This second script can be executed directly on the command line, but I need to execute it from within my first program. I'll need to pass it a few parameters that would normally be passed in when it's run standalone (the first script runs periodically, and executes the second script under a certain set of system conditions).
Preliminary Google searches suggest using backticks or a system() call. Are there any other ways to run it? (I'm guessing yes, since it's Perl we're talking about :P ) Which method is preferred if I need to capture output from the invoked program (and, if possible, pipe that output as it executes to stdout as though the second program were invoked directly)?
(Edit: oh, now SO suggests some related questions. This one is close, but not exactly the same as what I'm asking. The second program will likely take an hour or more to run (lots of I/O), so I'm not sure a one-off invocation is the right fit for this.)
You can just do it.
{
local #ARGV = qw<param1 param2 param3>;
do '/home/buddy/myscript.pl';
}
Prevents the overhead of loading in another copy of perl.
The location of your current perl interpreter can be found in the special variable $^X. This is important if perl is not in your path, or if you have multiple perl versions available but which to make sure you're using the same one across the board.
When executing external commands, including other Perl programs, determining if they actually ran can be quite difficult. Inspecting $? can leave lasting mental scars, so I prefer to use IPC::System::Simple (available from the CPAN):
use strict;
use warnings;
use IPC::System::Simple qw(system capture);
# Run a command, wait until it finishes, and make sure it works.
# Output from this program goes directly to STDOUT, and it can take input
# from your STDIN if required.
system($^X, "yourscript.pl", #ARGS);
# Run a command, wait until it finishes, and make sure it works.
# The output of this command is captured into $results.
my $results = capture($^X, "yourscript.pl", #ARGS);
In both of the above examples any arguments you wish to pass to your external program go into #ARGS. The shell is also avoided in both of the above examples, which gives you a small speed advantage, and avoids any unwanted interactions involving shell meta-characters. The above code also expects your second program to return a zero exit value to indicate success; if that's not the case, you can specify an additional first argument of allowable exit values:
# Both of these commands allow an exit value of 0, 1 or 2 to be considered
# a successful execution of the command.
system( [0,1,2], $^X, "yourscript.pl", #ARGS );
# OR
capture( [0,1,2, $^X, "yourscript.pl", #ARGS );
If you have a long-running process and you want to process its data while it's being generated, then you're probably going to need a piped open, or one of the more heavyweight IPC modules from the CPAN.
Having said all that, any time you need to be calling another Perl program from Perl, you may wish to consider if using a module would be a better choice. Starting another program carries quite a few overheads, both in terms of start-up costs, and I/O costs for moving data between processes. It also significantly increases the difficulty of error handling. If you can turn your external program into a module, you may find it simplifies your overall design.
All the best,
Paul
I can think of a few ways to do this. You already mentioned the first two, so I won't go into detail on them.
backticks: $retVal = `perl somePerlScript.pl`;
system() call
eval
The eval can be accomplished by slurping the other file into a string (or a list of strings), then 'eval'ing the strings. Heres a sample:
#!/usr/bin/perl
open PERLFILE, "<somePerlScript.pl";
undef $/; # this allows me to slurp the file, ignoring newlines
my $program = <PERLFILE>;
eval $program;
4 . do: do 'somePerlScript.pl'
You already got good answers to your question, but there's always the posibility to take a different point of view: maybe you should consider refactoring the script that you want to run from the first script. Turn the functionality into a module. Use the module from the first and from the second script.
If you need to asynchronously call your external script -you just want to launch it and not wait for it to finish-, then :
# On Unix systems, either of these will execute and just carry-on
# You can't collect output that way
`myscript.pl &`;
system ('myscript.pl &');
# On Windows systems the equivalent would be
`start myscript.pl`;
system ('start myscript.pl');
# If you just want to execute another script and terminate the current one
exec ('myscript.pl');
Use backticks if you need to capture the output of the command.
Use system if you do not need to capture the output of the command.
TMTOWTDI: so there are other ways too, but those are the two easiest and most likely.
See the perlipc documentation for several options for interprocess communication.
If your first script merely sets up the environment for the second script, you may be looking for exec.
#!/usr/bin/perl
use strict;
open(OUTPUT, "date|") or die "Failed to create process: $!\n";
while (<OUTPUT>)
{
print;
}
close(OUTPUT);
print "Process exited with value " . ($? >> 8) . "\n";
This will start the process date and pipe the output of the command to the OUTPUT filehandle which you can process a line at a time. When the command is finished you can close the output filehandle and retrieve the return value of the process. Replace date with whatever you want.
I wanted to do something like this to offload non-subroutines into an external file to make editing easier. I actually made this into a subroutine. The advantage of this way is that those "my" variables in the external file get declared in the main namespace. If you use 'do' they apparently don't migrate to the main namespace. Note the presentation below doesn't include error handling
sub getcode($) {
my #list;
my $filename = shift;
open (INFILE, "< $filename");
#list = <INFILE>;
close (INFILE);
return \#list;
}
# and to use it:
my $codelist = [];
$codelist = getcode('sourcefile.pl');
eval join ("", #$codelist);
Many beginning programmers write code like this:
sub copy_file ($$) {
my $from = shift;
my $to = shift;
`cp $from $to`;
}
Is this bad, and why? Should backticks ever be used? If so, how?
A few people have already mentioned that you should only use backticks when:
You need to capture (or supress) the output.
There exists no built-in function or Perl module to do the same task, or you have a good reason not to use the module or built-in.
You sanitise your input.
You check the return value.
Unfortunately, things like checking the return value properly can be quite challenging. Did it die to a signal? Did it run to completion, but return a funny exit status? The standard ways of trying to interpret $? are just awful.
I'd recommend using the IPC::System::Simple module's capture() and system() functions rather than backticks. The capture() function works just like backticks, except that:
It provides detailed diagnostics if the command doesn't start, is killed by a signal, or returns an unexpected exit value.
It provides detailed diagnostics if passed tainted data.
It provides an easy mechanism for specifying acceptable exit values.
It allows you to call backticks without the shell, if you want to.
It provides reliable mechanisms for avoiding the shell, even if you use a single argument.
The commands also work consistently across operating systems and Perl versions, unlike Perl's built-in system() which may not check for tainted data when called with multiple arguments on older versions of Perl (eg, 5.6.0 with multiple arguments), or which may call the shell anyway under Windows.
As an example, the following code snippet will save the results of a call to perldoc into a scalar, avoids the shell, and throws an exception if the page cannot be found (since perldoc returns 1).
#!/usr/bin/perl -w
use strict;
use IPC::System::Simple qw(capture);
# Make sure we're called with command-line arguments.
#ARGV or die "Usage: $0 arguments\n";
my $documentation = capture('perldoc', #ARGV);
IPC::System::Simple is pure Perl, works on 5.6.0 and above, and doesn't have any dependencies that wouldn't normally come with your Perl distribution. (On Windows it depends upon a Win32:: module that comes with both ActiveState and Strawberry Perl).
Disclaimer: I'm the author of IPC::System::Simple, so I may show some bias.
The rule is simple: never use backticks if you can find a built-in to do the same job, or if their is a robust module on the CPAN which will do it for you. Backticks often rely on unportable code and even if you untaint the variables, you can still open yourself up to a lot of security holes.
Never use backticks with user data unless you have very tightly specified what is allowed (not what is disallowed -- you'll miss things)! This is very, very dangerous.
Backticks should be used if and only if you need to capture the output of a command. Otherwise, system() should be used. And, of course, if there's a Perl function or CPAN module that does the job, this should be used instead of either.
In either case, two things are strongly encouraged:
First, sanitize all inputs: Use Taint mode (-T) if the code is exposed to possible untrusted input. Even if it's not, make sure to handle (or prevent) funky characters like space or the three kinds of quote.
Second, check the return code to make sure the command succeeded. Here is an example of how to do so:
my $cmd = "./do_something.sh foo bar";
my $output = `$cmd`;
if ($?) {
die "Error running [$cmd]";
}
Another way to capture stdout(in addition to pid and exit code) is to use IPC::Open3 possibily negating the use of both system and backticks.
Use backticks when you want to collect the output from the command.
Otherwise system() is a better choice, especially if you don't need to invoke a shell to handle metacharacters or command parsing. You can avoid that by passing a list to system(), eg system('cp', 'foo', 'bar') (however you'd probably do better to use a module for that particular example :))
In Perl, there's always more than one way to do anything you want. The primary point of backticks is to get the standard output of the shell command into a Perl variable. (In your example, anything that the cp command prints will be returned to the caller.) The downside of using backticks in your example is you don't check the shell command's return value; cp could fail and you wouldn't notice. You can use this with the special Perl variable $?. When I want to execute a shell command, I tend to use system:
system("cp $from $to") == 0
or die "Unable to copy $from to $to!";
(Also observe that this will fail on filenames with embedded spaces, but I presume that's not the point of the question.)
Here's a contrived example of where backticks might be useful:
my $user = `whoami`;
chomp $user;
print "Hello, $user!\n";
For more complicated cases, you can also use open as a pipe:
open WHO, "who|"
or die "who failed";
while(<WHO>) {
# Do something with each line
}
close WHO;
From the "perlop" manpage:
That doesn't mean you should go out of
your way to avoid backticks when
they're the right way to get something
done. Perl was made to be a glue
language, and one of the things it
glues together is commands. Just
understand what you're getting
yourself into.
For the case you are showing using the File::Copy module is probably best. However, to answer your question, whenever I need to run a system command I typically rely on IPC::Run3. It provides a lot of functionality such as collecting the return code and the standard and error output.
Whatever you do, as well as sanitising input and checking the return value of your code, make sure you call any external programs with their explicit, full path. e.g. say
my $user = `/bin/whoami`;
or
my $result = `/bin/cp $from $to`;
Saying just "whoami" or "cp" runs the risk of accidentally running a command other than what you intended, if the user's path changes - which is a security vulnerability that a malicious attacker could attempt to exploit.
Your example's bad because there are perl builtins to do that which are portable and usually more efficient than the backtick alternative.
They should be used only when there's no Perl builtin (or module) alternative. This is both for backticks and system() calls. Backticks are intended for capturing output of the executed command.
Backticks are only supposed to be used when you want to capture output. Using them here "looks silly." It's going to clue anyone looking at your code into the fact that you aren't very familiar with Perl.
Use backticks if you want to capture output.
Use system if you want to run a command. One advantage you'll gain is the ability to check the return status.
Use modules where possible for portability. In this case, File::Copy fits the bill.
In general, it's best to use system instead of backticks because:
system encourages the caller to check the return code of the command.
system allows "indirect object" notation, which is more secure and adds flexibility.
Backticks are culturally tied to shell scripting, which might not be common among readers of the code.
Backticks use minimal syntax for what can be a heavy command.
One reason users might be temped to use backticks instead of system is to hide STDOUT from the user. This is more easily and flexibly accomplished by redirecting the STDOUT stream:
my $cmd = 'command > /dev/null';
system($cmd) == 0 or die "system $cmd failed: $?"
Further, getting rid of STDERR is easily accomplished:
my $cmd = 'command 2> error_file.txt > /dev/null';
In situations where it makes sense to use backticks, I prefer to use the qx{} in order to emphasize that there is a heavy-weight command occurring.
On the other hand, having Another Way to Do It can really help. Sometimes you just need to see what a command prints to STDOUT. Backticks, when used as in shell scripts are just the right tool for the job.
Perl has a split personality. On the one hand it is a great scripting language that can replace the use of a shell. In this kind of one-off I-watching-the-outcome use, backticks are convenient.
When used a programming language, backticks are to be avoided. This is a lack of error
checking and, if the separate program backticks execute can be avoided, efficiency is
gained.
Aside from the above, the system function should be used when the command's output is not being used.
Backticks are for amateurs. The bullet-proof solution is a "Safe Pipe Open" (see "man perlipc"). You exec your command in another process, which allows you to first futz with STDERR, setuid, etc. Advantages: it does not rely on the shell to parse #ARGV, unlike open("$cmd $args|"), which is unreliable. You can redirect STDERR and change user priviliges without changing the behavior of your main program. This is more verbose than backticks but you can wrap it in your own function like run_cmd($cmd,#args);
sub run_cmd {
my $cmd = shift #_;
my #args = #_;
my $fh; # file handle
my $pid = open($fh, '-|');
defined($pid) or die "Could not fork";
if ($pid == 0) {
open STDERR, '>/dev/null';
# setuid() if necessary
exec ($cmd, #args) or exit 1;
}
wait; # may want to time out here?
if ($? >> 8) { die "Error running $cmd: [$?]"; }
while (<$fh>) {
# Have fun with the output of $cmd
}
close $fh;
}