yttitle.pl: rewrite, using irssi pidwait, input_add, etc, instead of timers and waitpid

1 files changed, 82 insertions, 82 deletions

diff --git a/yttitle.pl b/yttitle.pl
index a1aaf3a..145d5ac 100644
--- a/yttitle.pl
+++ b/yttitle.pl
@@ -1,16 +1,28 @@
 #!/usr/bin/perl
 
+# yttitle.pl
+
+# When someone says something with a youtube video link in it, get the
+# title and say it in the channel.
+
+# Uses yt-dlp to do the title lookups. Works much better than the
+# usual HTML-scraping, but unfortunately is *slow* (like, average 3
+# sec per lookup).
+
+# Since lookups take so long, they're done asynchronously, via
+# fork(). This script owes a great debt to the dns.pl script
+# included with irssi, which shows how to use irssi's pidwait and
+# input_(add|remove).
+
 no strict;
 
-use POSIX ":sys_wait_h";
+use POSIX '_exit';
 
 use Irssi qw/
 	signal_add_first
-	signal_continue
-	signal_register
-	command_bind
-	timeout_add_once
-	get_irssi_dir
+	pidwait_add
+	input_add
+	input_remove
 /;
 
 our $VERSION = "0.1";
@@ -26,24 +38,29 @@ our %IRSSI = (
 # video ID => title
 our %cache;
 
-our $tmpdir = get_irssi_dir . "/yttitle.tmp";
+# keep track of video IDs of in-progress lookups, so we don't spawn
+# a 2nd process for the same ID.
+our %in_progress;
 
-# pid => [server, target, video ID]
-our %jobs;
+# TODO: the rest of these variables should be irssi settings.
 
-# video ID keys (values meaningless; only care about key presence)
-our %job_videos;
+# how many jobs are currently running?
+our $jobcount = 0;
 
-# attempts to spawn more jobs than this are just ignored.
+# attempts to spawn more simultaneous jobs than this are just ignored.
 our $maxjobs = 10;
 
-# milliseconds: how often to check %jobs to see if any jobs are done.
-our $queue_time = 1000;
+# yt-dlp processes that don't respond within this many seconds
+# are terminated.
+our $job_timeout = 15;
+
+# passed to yt-dlp itself.
+our $socket_timeout = 10;
 
-# command to execute, 1st %s replaced with video ID, 2nd with tmp filename.
+# command to execute, %s replaced with video ID.
 # --socket-timeout arg is seconds, should be longer that $timeout.
 our $command_fmt =
-	"yt-dlp -q --print '%%(title)s' --socket-timeout 10 -- %s >%s 2>/dev/null";
+	"yt-dlp -q --print '%%(title)s' --socket-timeout $socket_timeout -- %s 2>/dev/null";
 
 our $DEBUG = 1;
 
@@ -55,35 +72,8 @@ sub debugf {
 	Irssi::print(sprintf(@_)) if $DEBUG;
 }
 
-sub start_timer {
-	timeout_add_once($queue_time, "check_jobs", undef);
-}
-
-sub get_tmp_filename {
-	my $id = shift;
-	return $tmpdir . "/" . $id;
-}
-
-sub read_tmp_file {
-	my $file = shift;
-
-	open my $fh, "<:encoding(UTF-8)", $file;
-
-	if(!$fh) {
-		debug("read_tmp_file() failed to open $file: $!");
-		return;
-	}
-
-	my $result = <$fh>;
-	close $fh;
-
-	chomp $result if defined $result;
-	return length($result) ? $result : undef;
-}
-
 sub spawn_job {
 	my ($server, $target, $video_id) = @_;
-	my $jobcount = keys %jobs;
 
 	debug("spawn_job() called, video_id $video_id");
 
@@ -92,45 +82,76 @@ sub spawn_job {
 		return;
 	}
 
-	if($job_videos{$video_id}) {
+	if($in_progress{$video_id}) {
 		debug("spawn_job(): video_id $video_id already in queue");
 		return;
 	}
 
-	my $cmd = sprintf($command_fmt, $video_id, get_tmp_filename($video_id));
-	debug("spawn_job() command is: $cmd");
+	$in_progress{$video_id} = 1;
+
+	my $cmd = sprintf($command_fmt, $video_id);
+   debug("spawn_job() command is: $cmd");
+
+	my($read, $write);
+	pipe($read, $write);
 
 	my $pid = fork();
+	if(!defined($pid)) {
+		debug("spawn_job(): video_id $video_id: fork() failed!");
+		close $read;
+		close $write;
+		return;
+	}
 
 	if($pid) {
 		# parent
-		debug("spawn_job() forked, kid pid is $pid");
-		start_timer unless keys %jobs;
-		$jobs{$pid} = [ $server, $target, $video_id ];
-		$job_videos{$video_id} = 1;
-		debug("spawn_job(): job pids: " . join(",", keys %jobs));
+		$jobcount++;
+		debug("spawn_job() forked, kid pid is $pid, jobcount now $jobcount");
+		close $write;
+		my $pipe_args = [ $pid, $server, $target, $video_id, $read ];
+		pidwait_add($pid);
+		$pipe_tags{$video_id} = input_add(fileno($read), INPUT_READ, \&job_done, $pipe_args);
+		return;
 	} else {
 		# child, debug() and Irssi::print don't work, here.
-		exec $cmd;
+		eval {
+			local $SIG{ALRM} = sub { die "alarum\n" };
+			alarm $job_timeout;
+			print $write `$cmd`;
+			alarm 0;
+		};
+		# don't bother to die() on error here, we're about to exit anyway.
+		if($@ && ($@ eq "alarum\n")) {
+			# print *something* if $cmd timed out
+			print $write "\n";
+		}
+		# not sure this needs to be wrapped in eval, but...
+		eval {
+			close $write;
+		};
+		_exit(1);
 	}
 }
 
 sub job_done {
-	my $pid = shift;
+	my $pipe_args = shift;
+	my ($pid, $server, $target, $video_id, $read) = @$pipe_args;
 
-	my ($server, $target, $video_id) = @{$jobs{$pid}};
+	debug("job_done() pid $pid, target $target, video_id $video_id, jobcount was $jobcount");
 
-	debug("job_done() pid $pid, target $target, video_id $video_id");
+	my $title = <$read>;
+	chomp $title;
+	close $read;
+	input_remove($pipe_tags{$video_id});
 
-	my $file = get_tmp_filename($video_id);
-	my $title = read_tmp_file($file);
-
-	unlink $file;
+	$jobcount--;
+	delete $in_progress{$video_id};
 
 	$cache{$video_id} = $title;
-	delete $job_videos{$video_id};
 
-	if(!defined($title)) {
+	if(defined($title)) {
+		debug("job_done(): video_id $video_id title is $title");
+	} else {
 		debug("job_done(): video_id $video_id failed to get title");
 		return;
 	}
@@ -138,26 +159,6 @@ sub job_done {
 	say_title($server, $target, $video_id, $title);
 }
 
-sub check_jobs {
-	my @k = keys %jobs;
-
-	debug("check_jobs(): " . @k . " jobs");
-	return unless @k;
-
-	for my $jobpid (@k) {
-		debug("check_jobs(): about to waitpid($jobpid, WNOHANG)");
-		$pid = waitpid($jobpid, WNOHANG);
-
-		debug("check_jobs(): jobpid $jobpid, pid $pid");
-
-		next if $pid == 0; # still running, let it run
-		job_done($jobpid) if $pid > 0; # -1 is "no such pid"
-		delete $jobs{$jobpid};
-	}
-
-	start_timer if keys %jobs;
-}
-
 sub say_title {
 	my ($server, $target, $video_id, $title) = @_;
 	my $tag = "YouTube" . ($DEBUG ? "($video_id)" : "");
@@ -190,5 +191,4 @@ sub on_public_msg {
 }
 
 ### main()
-mkdir $tmpdir;
 signal_add_first("message public", "on_public_msg");