scafe.check.dependencies

#!/usr/bin/env perl

# This chunk of stuff was generated by App::FatPacker. To find the original
# file's code, look for the end of this BEGIN block or the string 'FATPACK'
BEGIN {
my %fatpacked;

s/^  //mg for values %fatpacked;

my $class = 'FatPacked::'.(0+\%fatpacked);
no strict 'refs';
*{"${class}::files"} = sub { keys %{$_[0]} };

if ($] < 5.008) {
  *{"${class}::INC"} = sub {
    if (my $fat = $_[0]{$_[1]}) {
      my $pos = 0;
      my $last = length $fat;
      return (sub {
        return 0 if $pos == $last;
        my $next = (1 + index $fat, "\n", $pos) || $last;
        $_ .= substr $fat, $pos, $next - $pos;
        $pos = $next;
        return 1;
      });
    }
  };
}

else {
  *{"${class}::INC"} = sub {
    if (my $fat = $_[0]{$_[1]}) {
      open my $fh, '<', \$fat
        or die "FatPacker error loading $_[1] (could be a perl installation issue?)";
      return $fh;
    }
    return;
  };
}

unshift @INC, bless \%fatpacked, $class;
  } # END OF FATPACK CODE


#====================================================================================================================================================#
#<use>
$|++; #---turn on the auto flush for the progress bar
no warnings 'utf8';
use warnings;
use strict;
use Getopt::Long 'HelpMessage';
use Time::HiRes qw( time );
use File::Spec::Functions qw(rel2abs abs2rel);
use Cwd 'abs_path';
#<\use>
#====================================================================================================================================================#

#====================================================================================================================================================#
#<doc>
=head1 SYNOPSIS

           5'-O~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~AAA-3'
                        O~~~AA      O~~         O~       O~~~~~~~AO~~~~~~~~A
                      O~~    O~~ O~~   O~~     O~O~~     O~~      O~~       
                       O~~      O~~           O~  O~~    O~~      O~~       
                         O~~    O~~          O~~   O~~   O~~~~~AA O~~~~~~A  
                            O~~ O~~         O~~~~~A O~~  O~~      O~~       
                      O~~    O~~ O~~   O~~ O~~       O~~ O~~      O~~       
                        O~~~~A     O~~~   O~~         O~~O~~      O~~~~~~~AA
       ┌─ᐅ 5'-O~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-3'
 ...===┴========================================================================================...

                      Single Cell Analysis of Five-prime End (SCAFE) Tool Suite 
                                  ---> check.dependencies <---
                                  <--- check dependencies --->

 Description:
   This scripts check the integrity of tools and workflow scripts, 3rd executable dependencies and R packages.

 Usage:
   check.dependencies

 Dependencies:
   wget
   tar
   Rscript

 For demo, cd to SCAFE dir and run,
   scafe.check.dependencies

=head1 VERSION

v0.9.0-beta [March 20, 2021]
	-Initial pre-release

v1.0.0 [June 6, 2022]
	-will check optimize_tangent.R
	-will check tabix and bgzip
	-will not check cicero related function

=cut
#<\doc>
#====================================================================================================================================================#

#====================================================================================================================================================#
#<lastCmdCalled>
#
#	notCalledBefore
#
#	notCalledBefore
#
#<\lastCmdCalled>
#====================================================================================================================================================#

#====================================================================================================================================================#
#<global>
#<\global>
#====================================================================================================================================================#

#====================================================================================================================================================#
{	#Main sections lexical scope starts
#====================================================================================================================================================#

#====================================================================================================================================================#
#	section 0_startTasks
#
#<section ID="startTasks" num="0">
my $check_hsh_ref = {};
&readParameters();#->433
my ($scafe_dir, $script_dir, $check_dir) = &prepDir();#->407
&checkPerlScript($script_dir, $check_hsh_ref);#->156
&checkDownload($check_dir, $check_hsh_ref);#->123
&checkThirdParty($scafe_dir, $check_hsh_ref);#->291
&checkRVersion($check_hsh_ref);#->252
&checkRPackages($check_hsh_ref);#->197
&checkRScript($scafe_dir, $check_hsh_ref);#->224
&summary($check_hsh_ref);#->453
#<\section>
#====================================================================================================================================================#

#====================================================================================================================================================#
#	section 1_run
#
#<section ID="run" num="1">
#<\section>
#====================================================================================================================================================#

#====================================================================================================================================================#
}	#Main sections lexical scope ends
#====================================================================================================================================================#

#====================================================================================================================================================#
#List of subroutines by category
#
#	general [n=1]:
#		readParameters
#
#	unassigned [n=8]:
#		checkDownload, checkPerlScript, checkRPackages
#		checkRScript, checkRVersion, checkThirdParty
#		prepDir, summary
#
#====================================================================================================================================================#

sub checkDownload {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_dir, $check_hsh_ref
#	output: 
#	toCall: &checkDownload($check_dir, $check_hsh_ref);
#	calledInLine: 89
#....................................................................................................................................................#
	my ($check_dir, $check_hsh_ref) = @_;
	
	my $status = "checking";
	my $check_name = "wget tar data";
	my $type = "dropbox access";
	printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
	my $check_data_url = "https://www.dropbox.com/s/yx1306zuvtc7li3/demo.tar.gz";
	my $tar_gz_path = "$check_dir/demo.tar.gz";
	system ("rm -f $tar_gz_path");
	system ("rm -rf $check_dir/demo/");
	my $stdout = `wget $check_data_url -P $check_dir 2>&1`;
	system ("tar -xzvf $tar_gz_path -C $check_dir 2>&1 >/dev/null");
	
	$status = "failed";
	$status = 'successful' if (-s "$check_dir/demo/demo.txt");
	printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
	system ("rm -rf $check_dir");
	$check_hsh_ref->{$status}{$type}{$check_name}++;

	return ();
}
sub checkPerlScript {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref, $script_dir
#	output: 
#	toCall: &checkPerlScript($script_dir, $check_hsh_ref);
#	calledInLine: 88
#....................................................................................................................................................#
	my ($script_dir, $check_hsh_ref) = @_;
	print "===============================\n";
	print "start checking\n";
	print "===============================\n";

	printf ("%-20s %-35s %-10s\n", 'Check Type', 'Check Item', 'Check Status');
	printf ("%-20s %-35s %-10s\n", '===============', '===============', '===============');
	my $type = "perl executables";
	my $readme_path = "$script_dir/README.md";
	open README, "<", $readme_path;
	while (<README>) {
		chomp;
		if ($_ =~ m/^\* \[\*\*(.+)\*\*\]/) {
			my $script_name = $1;
			$script_name =~ s/\\//g;
			my $status = "checking";
			printf ("%-20s %-35s %-10s\r", $type, $script_name, $status);
			my $script_path = "$script_dir/$script_name";
			my @help_message = `$script_path --help`;
			$status = "failed";
			$status = 'successful' if ($help_message[0] =~ m/Usage:/);
			printf ("%-20s %-35s %-10s\n", $type, $script_name, $status);
			$check_hsh_ref->{$status}{$type}{$script_name}++;
		}
	}
	close README;

	return ();
}
sub checkRPackages {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref
#	output: 
#	toCall: &checkRPackages($check_hsh_ref);
#	calledInLine: 92
#....................................................................................................................................................#
	my ($check_hsh_ref) = @_;
		
	my $type = "R packages";
	foreach my $package (qw/ROCR PRROC caret e1071 ggplot2 scales reshape2 R.utils/) {
		my $status = "checking";
		printf ("%-20s %-35s %-10s\r", $type, $package, $status);
		my $stdOut = `Rscript -e "packageVersion('$package')" 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/[1]/);
		printf ("%-20s %-35s %-10s\n", $type, $package, $status);
		$check_hsh_ref->{$status}{$type}{$package}++;
	}
	
	return ();
}
sub checkRScript {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref, $scafe_dir
#	output: 
#	toCall: &checkRScript($scafe_dir, $check_hsh_ref);
#	calledInLine: 93
#....................................................................................................................................................#
	my ($scafe_dir, $check_hsh_ref) = @_;
	
	my $type = "R scripts";
	my $R_dir = "$scafe_dir/resources/R/";
	foreach my $R_script (qw/benchmark_roc.R build_glm.R predict_prob.R optimize_tangent.R/) {
		my $status = "checking";
		printf ("%-20s %-35s %-10s\r", $type, $R_script, $status);
		my $stdOut = `Rscript $R_dir/$R_script 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/Error: $R_script has no input/);
		printf ("%-20s %-35s %-10s\n", $type, $R_script, $status);
		$check_hsh_ref->{$status}{$type}{$R_script}++;
	}

	return ();
}
sub checkRVersion {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref
#	output: 
#	toCall: &checkRVersion($check_hsh_ref);
#	calledInLine: 91
#....................................................................................................................................................#
	my ($check_hsh_ref) = @_;
	
	my $type = "R version";
	my $name = "v3.5 or later";
	my $status = "checking";
	printf ("%-20s %-35s %-10s\r", $type, $name, $status);

	my $stdOut = `Rscript --version 2>&1`;
	$status = 'failed';
	if ($stdOut =~ m/version (.+) /) {
		my ($version, $sub_version) = split /\./, $1;
		#print "$version $sub_version\n";
		if ($version >= 3) {
			if ($version > 3) {
				$status = 'successful';
			} else {
				if ($sub_version >= 5) {
					$status = 'successful';
				}
			}
		}
	}
	printf ("%-20s %-35s %-10s\n", $type, $name, $status);
	$check_hsh_ref->{$status}{$type}{$name}++;

	return ();
}
sub checkThirdParty {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref, $scafe_dir
#	output: 
#	toCall: &checkThirdParty($scafe_dir, $check_hsh_ref);
#	calledInLine: 90
#....................................................................................................................................................#
	
	my ($scafe_dir, $check_hsh_ref) = @_;

	my $bedtools_bin = "$scafe_dir/resources/bin/bedtools/bedtools";
	my $samtools_bin = "$scafe_dir/resources/bin/samtools/samtools";
	my $paraclu_bin = "$scafe_dir/resources/bin/paraclu/paraclu";
	my $cut_sh_path = "$scafe_dir/resources/bin/paraclu/paraclu-cut.sh";
	my $bedGraphToBigWig_bin = "$scafe_dir/resources/bin/bedGraphToBigWig/bedGraphToBigWig";
	my $bigWigAverageOverBed_bin = "$scafe_dir/resources/bin/bigWigAverageOverBed/bigWigAverageOverBed";
	my $tabix_bin = "$scafe_dir/resources/bin/tabix/tabix";
	my $bgzip_bin = "$scafe_dir/resources/bin/bgzip/bgzip";
	
	my $type = "3rd-party apps";
	
	{
		my $status = "checking";
		my $check_name = "bgzip";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$bgzip_bin --version 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/bgzip \(htslib\) (\S+)/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	{
		my $status = "checking";
		my $check_name = "tabix";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$tabix_bin --version 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/tabix \(htslib\) (\S+)/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	{
		my $status = "checking";
		my $check_name = "bedtools";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$bedtools_bin --version 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/bedtools v(\S+)/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	{
		my $status = "checking";
		my $check_name = "samtools";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$samtools_bin 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/\s+(Version: \S+)\s+/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}
	
	{
		my $status = "checking";
		my $check_name = "paraclu";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$paraclu_bin 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/paraclu: I need a minValue and a fileName/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}
	
	{
		my $status = "checking";
		my $check_name = "paraclu-cut.sh";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$cut_sh_path 55C7128A 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/awk:/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	{
		my $status = "checking";
		my $check_name = "bedGraphToBigWig";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$bedGraphToBigWig_bin 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/bedGraphToBigWig v (\S+)/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	{
		my $status = "checking";
		my $check_name = "bigWigAverageOverBed";
		printf ("%-20s %-35s %-10s\r", $type, $check_name, $status);
		my $stdOut = `$bigWigAverageOverBed_bin 2>&1`;
		$status = "failed";
		$status = 'successful' if ($stdOut =~ m/bigWigAverageOverBed v(\S+)/);
		printf ("%-20s %-35s %-10s\n", $type, $check_name, $status);
		$check_hsh_ref->{$status}{$type}{$check_name}++;
	}

	return ();
}
sub prepDir {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: none
#	output: $check_dir, $scafe_dir, $script_dir
#	toCall: my ($scafe_dir, $script_dir, $check_dir) = &prepDir();
#	calledInLine: 87
#....................................................................................................................................................#

	my $scriptAbsPath = abs_path($0);
	my @path_split = split /\//, $scriptAbsPath;
	my $scafe_dir_index = $#path_split-2;
	my $script_dir_index = $#path_split-1;
	my $scafe_dir = join "/", @path_split[0..$scafe_dir_index];
	my $script_dir = join "/", @path_split[0..$script_dir_index];

	my @mkDirAry = ();
	my $check_dir = $scafe_dir."/check_log/"; push @mkDirAry, $check_dir;
	foreach my $dir (@mkDirAry) {system ("mkdir -pm 755 $dir");}

	return ($scafe_dir, $script_dir, $check_dir);
}
sub readParameters {
#....................................................................................................................................................#
#	subroutineCategory: general
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: none
#	output: 
#	toCall: &readParameters();
#	calledInLine: 86
#....................................................................................................................................................#
	
	GetOptions 	(
		'help'				=>	sub { HelpMessage(0) },
	);

	return();

}
sub summary {
#....................................................................................................................................................#
#	subroutineCategory: unassigned
#	dependOnSub: >none
#	appearInSub: >none
#	primaryAppearInSection: 0_startTasks|82
#	secondaryAppearInSection: >none
#	input: $check_hsh_ref
#	output: 
#	toCall: &summary($check_hsh_ref);
#	calledInLine: 94
#....................................................................................................................................................#
	my ($check_hsh_ref) = @_;

	print "===============================\n";
	print "Finished checking\n";
	print "===============================\n";
	print "\n";

	if (not exists $check_hsh_ref->{'failed'}) {
		print "Successful for all checks. SCAFE should run well.\n";
	} else {
		print "A few checks failed. Please address.\n";
	}
	

	return ();
}

exit;