| 1 |
#!/usr/bin/env perl
|
| 2 |
|
| 3 |
use strict;
|
| 4 |
use Getopt::Long;
|
| 5 |
use Mysql;
|
| 6 |
use Term::ANSIColor;
|
| 7 |
|
| 8 |
sub printHelp;
|
| 9 |
sub parseCrabDir;
|
| 10 |
|
| 11 |
our $db = Mysql->connect ("cmshead.mps.ohio-state.edu", "ntuple", "osuT3User") or die "Failed to connect to Tier 3, stopped";
|
| 12 |
|
| 13 |
my %opt;
|
| 14 |
Getopt::Long::Configure ("bundling");
|
| 15 |
GetOptions (\%opt, "dumb|d", "force|f", "destination|o=s", "dataset|s", "help|h");
|
| 16 |
|
| 17 |
printHelp () if $opt{"help"} || !$ARGV[0] || (!$opt{"dumb"} && !$opt{"dataset"} && !$ARGV[1]);
|
| 18 |
my $crabDir = $ARGV[0];
|
| 19 |
my $outputDir = $ARGV[0];
|
| 20 |
my $dataset = $ARGV[0];
|
| 21 |
$outputDir = $ARGV[1] if !$opt{"dumb"} && !$opt{"dataset"};
|
| 22 |
|
| 23 |
if ($opt{"dataset"})
|
| 24 |
{
|
| 25 |
my $queryDataset = $dataset;
|
| 26 |
$queryDataset =~ s/\*/%/g;
|
| 27 |
$queryDataset =~ s/(.*)/%$1%/g;
|
| 28 |
my $query = "select workingDirectory,location,dataset,user,creationTime,status from ntuple where dataset like '$queryDataset' order by creationTime";
|
| 29 |
$db->selectdb ("ntuple");
|
| 30 |
my $results = $db->query ($query);
|
| 31 |
if ($results->numrows () == 1)
|
| 32 |
{
|
| 33 |
my @row = $results->fetchrow ();
|
| 34 |
$crabDir = "$row[0]/ntuple";
|
| 35 |
$outputDir = $row[1];
|
| 36 |
}
|
| 37 |
die "Database entry does not exist, stopped" if $results->numrows () == 0;
|
| 38 |
if ($results->numrows () > 1)
|
| 39 |
{
|
| 40 |
my %workingDir;
|
| 41 |
my %location;
|
| 42 |
print "Found multiple database entries matching\n";
|
| 43 |
print "\"$dataset\":\n";
|
| 44 |
for (my $i = 1; $i <= $results->numrows (); $i++)
|
| 45 |
{
|
| 46 |
my @row = $results->fetchrow ();
|
| 47 |
$row[4] =~ s/([^ ]*) [^ ]*/$1/g;
|
| 48 |
$workingDir{"$i"} = $row[0];
|
| 49 |
$location{"$i"} = $row[1];
|
| 50 |
printf "(%2d) $row[2]\n", $i;
|
| 51 |
print " (";
|
| 52 |
print color "green" if $row[5] eq "present";
|
| 53 |
print color "bold yellow" if $row[5] eq "submitted";
|
| 54 |
print color "bold red" if $row[5] eq "created" or $row[5] eq "cancelled" or $row[5] eq "deprecated";
|
| 55 |
print $row[5];
|
| 56 |
print color "reset";
|
| 57 |
print ") created by $row[3] on $row[4]\n";
|
| 58 |
}
|
| 59 |
print "\nWhich entry would you like to use?: ";
|
| 60 |
my $response = <STDIN>;
|
| 61 |
$response =~ s/[ \t\n]//g;
|
| 62 |
die "Your selection was not a valid option, stopped" if !(exists $workingDir{$response});
|
| 63 |
$crabDir = "$workingDir{$response}/ntuple";
|
| 64 |
$outputDir = $location{$response};
|
| 65 |
}
|
| 66 |
}
|
| 67 |
|
| 68 |
die "Output directory does not exist, stopped" if !(-e $outputDir) || !(-d $outputDir);
|
| 69 |
opendir (CRAB_OUTPUT, "$outputDir");
|
| 70 |
my @crabOutput = readdir (CRAB_OUTPUT);
|
| 71 |
closedir (CRAB_OUTPUT);
|
| 72 |
my %filesToKeep;
|
| 73 |
if (!$opt{"dumb"})
|
| 74 |
{
|
| 75 |
die "CRAB directory does not exist, stopped" if !(-e $crabDir) || !(-d $crabDir);
|
| 76 |
parseCrabDir ($crabDir, \%filesToKeep);
|
| 77 |
}
|
| 78 |
else
|
| 79 |
{
|
| 80 |
foreach my $file (@crabOutput)
|
| 81 |
{
|
| 82 |
next if ($file eq "." || $file eq "..");
|
| 83 |
next if !($file =~ m/^.*_[^_]*_[^_]*_[^_]*\.root$/);
|
| 84 |
my $jobNumber = $file;
|
| 85 |
my $submissionNumber = $file;
|
| 86 |
$jobNumber =~ s/^.*_([^_]*)_[^_]*_[^_]*\.root$/$1/;
|
| 87 |
$submissionNumber =~ s/^.*_[^_]*_([^_]*)_[^_]*\.root$/$1/;
|
| 88 |
if (!(defined $filesToKeep{$jobNumber})
|
| 89 |
|| $filesToKeep{$jobNumber} < $submissionNumber)
|
| 90 |
{
|
| 91 |
$filesToKeep{$jobNumber} = $submissionNumber;
|
| 92 |
}
|
| 93 |
}
|
| 94 |
}
|
| 95 |
if ($opt{"destination"})
|
| 96 |
{
|
| 97 |
mkdir $opt{"destination"} or die "Failed to create destination directory, stopped" if !(-e $opt{"destination"});
|
| 98 |
die "Destination is not a directory, stopped" if !(-d $opt{"destination"});
|
| 99 |
open (TEST_FILE, ">$opt{'destination'}/.crabCleanTest") or die "Unable to write to destination directory, stopped";
|
| 100 |
close (TEST_FILE);
|
| 101 |
unlink "$opt{'destination'}/.crabCleanTest";
|
| 102 |
}
|
| 103 |
opendir (CRAB_OUTPUT, "$outputDir");
|
| 104 |
my @crabOutput = readdir (CRAB_OUTPUT);
|
| 105 |
closedir (CRAB_OUTPUT);
|
| 106 |
my $nFiles = @crabOutput - 2;
|
| 107 |
print "Before cleaning: $nFiles files\n";
|
| 108 |
foreach my $file (@crabOutput)
|
| 109 |
{
|
| 110 |
next if ($file eq "." || $file eq "..");
|
| 111 |
next if !($file =~ m/^.*_[^_]*_[^_]*_[^_]*\.root$/);
|
| 112 |
my $jobNumber = $file;
|
| 113 |
my $submissionNumber = $file;
|
| 114 |
$jobNumber =~ s/^.*_([^_]*)_[^_]*_[^_]*\.root$/$1/;
|
| 115 |
$submissionNumber =~ s/^.*_[^_]*_([^_]*)_[^_]*\.root$/$1/;
|
| 116 |
if (!(defined $filesToKeep{$jobNumber}))
|
| 117 |
{
|
| 118 |
print "No information found for job $jobNumber!\n";
|
| 119 |
}
|
| 120 |
elsif ((($opt{"dumb"} && $filesToKeep{$jobNumber} != $submissionNumber)
|
| 121 |
|| (!$opt{"dumb"} && $filesToKeep{$jobNumber} ne $file))
|
| 122 |
&& !$opt{"destination"})
|
| 123 |
{
|
| 124 |
system ("rm -f $outputDir/$file") if $opt{"force"};
|
| 125 |
system ("rm -i $outputDir/$file") if !$opt{"force"};
|
| 126 |
}
|
| 127 |
elsif ((($opt{"dumb"} && $filesToKeep{$jobNumber} == $submissionNumber)
|
| 128 |
|| (!$opt{"dumb"} && $filesToKeep{$jobNumber} eq $file))
|
| 129 |
&& $opt{"destination"})
|
| 130 |
{
|
| 131 |
system ("mv -f $outputDir/$file $opt{'destination'}/$file") if $opt{"force"};
|
| 132 |
system ("mv -i $outputDir/$file $opt{'destination'}/$file") if !$opt{"force"};
|
| 133 |
}
|
| 134 |
}
|
| 135 |
opendir (CRAB_OUTPUT, "$outputDir");
|
| 136 |
my @crabOutput = readdir (CRAB_OUTPUT);
|
| 137 |
closedir (CRAB_OUTPUT);
|
| 138 |
opendir (CRAB_OUTPUT, "$opt{'destination'}");
|
| 139 |
my @crabOutputInDestination = readdir (CRAB_OUTPUT);
|
| 140 |
closedir (CRAB_OUTPUT);
|
| 141 |
my $nFiles = @crabOutput - 2;
|
| 142 |
$nFiles = @crabOutputInDestination - 2 if $opt{"destination"};
|
| 143 |
print "After cleaning: $nFiles files\n";
|
| 144 |
|
| 145 |
sub
|
| 146 |
printHelp
|
| 147 |
{
|
| 148 |
my $exeName = $0;
|
| 149 |
$exeName =~ s/^.*\/([^\/]*)$/$1/;
|
| 150 |
|
| 151 |
print "Usage: $exeName [OPTION]... CRAB_DIR OUTPUT_DIR\n";
|
| 152 |
print " or: $exeName -d [OPTION]... OUTPUT_DIR\n";
|
| 153 |
print " or: $exeName -s [OPTION]... DATASET\n";
|
| 154 |
print "Deletes excess EDM output caused by multiple CRAB submissions using the results\n";
|
| 155 |
print "contained in the CRAB working directory provided.\n";
|
| 156 |
print "\n";
|
| 157 |
print "Mandatory arguments to long options are mandatory for short options too.\n";
|
| 158 |
printf "%-29s%s\n", " -d, --dumb", "tries to use the filenames of the output instead of";
|
| 159 |
printf "%-29s%s\n", " ", "the results from the CRAB working directory; useful";
|
| 160 |
printf "%-29s%s\n", " ", "if the working directory has been lost";
|
| 161 |
printf "%-29s%s\n", " -f, --force", "never prompt (default is to prompt before any";
|
| 162 |
printf "%-29s%s\n", " ", "removal)";
|
| 163 |
printf "%-29s%s\n", " -h, --help", "print this help message";
|
| 164 |
printf "%-29s%s\n", " -o, --destination DIR", "do not remove excess output and move the good files";
|
| 165 |
printf "%-29s%s\n", " ", "to DIR";
|
| 166 |
printf "%-29s%s\n", " -s, --dataset", "tells the script to expect a dataset name instead";
|
| 167 |
printf "%-29s%s\n", " ", "of a directory";
|
| 168 |
|
| 169 |
exit;
|
| 170 |
}
|
| 171 |
|
| 172 |
sub
|
| 173 |
parseCrabDir
|
| 174 |
{
|
| 175 |
my $crabDir = shift;
|
| 176 |
my $filesToKeep = shift;
|
| 177 |
|
| 178 |
opendir (CRAB_RESULTS, "$crabDir/res");
|
| 179 |
my @crabResults = readdir (CRAB_RESULTS);
|
| 180 |
closedir (CRAB_RESULTS);
|
| 181 |
for my $file (@crabResults)
|
| 182 |
{
|
| 183 |
next if !($file =~ m/^crab_fjr_[^\.]*\.xml$/);
|
| 184 |
my $jobNumber = $file;
|
| 185 |
$jobNumber =~ s/^crab_fjr_([^\.]*)\.xml$/$1/;
|
| 186 |
my @status = `grep 'ExitStatus' $crabDir/res/$file`;
|
| 187 |
my $wrapperStatus = $status[0];
|
| 188 |
$wrapperStatus =~ s/^.*ExitStatus="([^"]*)".*$/$1/;
|
| 189 |
$wrapperStatus =~ s/\n//g;
|
| 190 |
if ($wrapperStatus != 0)
|
| 191 |
{
|
| 192 |
print "Job $jobNumber failed (return value $wrapperStatus). Skipping.\n";
|
| 193 |
next;
|
| 194 |
}
|
| 195 |
my @surlForGrid = `grep -A 1 '<SurlForGrid' $crabDir/res/$file`;
|
| 196 |
my $fileName = $surlForGrid[1];
|
| 197 |
$fileName = $surlForGrid[0] if $surlForGrid[0] =~ m/Value=/;
|
| 198 |
$fileName =~ s/^.*\/([^\/]*)\n/$1/ if !($fileName =~ m/Value=/);
|
| 199 |
$fileName =~ s/^.*\/([^\/]*)"\/>\n/$1/ if $fileName =~ m/Value=/;
|
| 200 |
$filesToKeep->{$jobNumber} = $fileName;
|
| 201 |
}
|
| 202 |
}
|
