| 3 |
|
use strict; |
| 4 |
|
use Getopt::Long; |
| 5 |
|
use Mysql; |
| 6 |
+ |
use Term::ANSIColor; |
| 7 |
|
|
| 8 |
|
sub printHelp; |
| 9 |
|
sub parseCrabDir; |
| 12 |
|
|
| 13 |
|
my %opt; |
| 14 |
|
Getopt::Long::Configure ("bundling"); |
| 15 |
< |
GetOptions (\%opt, "dumb|d", "force|f", "dataset|s", "help|h"); |
| 15 |
> |
GetOptions (\%opt, "dumb|d", "force|f", "destination|o=s", "print|p", "dataset|s", "help|h"); |
| 16 |
|
|
| 17 |
< |
printHelp () if $opt{"help"} || !$ARGV[0] || (!$opt{"dumb"} && !$opt{"dataset"} && !$ARGV[1]); |
| 17 |
> |
printHelp () if $opt{"help"} || !$ARGV[0] || (!$opt{"dumb"} && !$opt{"dataset"} && !$opt{"print"} && !$ARGV[1]); |
| 18 |
|
my $crabDir = $ARGV[0]; |
| 19 |
|
my $outputDir = $ARGV[0]; |
| 20 |
|
my $dataset = $ARGV[0]; |
| 22 |
|
|
| 23 |
|
if ($opt{"dataset"}) |
| 24 |
|
{ |
| 25 |
< |
my $query = "select workingDirectory,location,dataset,user,creationTime from ntuple where dataset like '$dataset' order by creationTime"; |
| 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) |
| 34 |
|
$crabDir = "$row[0]/ntuple"; |
| 35 |
|
$outputDir = $row[1]; |
| 36 |
|
} |
| 37 |
< |
if ($results->numrows () == 0) |
| 37 |
> |
die "Database entry does not exist, stopped" if $results->numrows () == 0; |
| 38 |
> |
if ($results->numrows () > 1) |
| 39 |
|
{ |
| 40 |
< |
print "Database entry does not exist.\n"; |
| 41 |
< |
exit; |
| 42 |
< |
} |
| 43 |
< |
my %workingDir; |
| 44 |
< |
my %location; |
| 45 |
< |
print "Found multiple database entries matching\n"; |
| 46 |
< |
print "\"$dataset\":\n"; |
| 47 |
< |
for (my $i = 1; $i <= $results->numrows (); $i++) |
| 48 |
< |
{ |
| 49 |
< |
my @row = $results->fetchrow (); |
| 50 |
< |
$workingDir{"$i"} = $row[0]; |
| 51 |
< |
$location{"$i"} = $row[1]; |
| 52 |
< |
printf "(%2d) $row[2]\n", $i; |
| 53 |
< |
print " created by $row[3] on $row[4]\n"; |
| 54 |
< |
} |
| 55 |
< |
print "\nWhich entry would you like to use?: "; |
| 56 |
< |
my $response = <STDIN>; |
| 57 |
< |
$response =~ s/[ \t\n]//g; |
| 58 |
< |
if (!(exists $workingDir{$response})) |
| 59 |
< |
{ |
| 60 |
< |
print "Your selection \"$response\" was not a valid option! Quitting.\n"; |
| 61 |
< |
exit; |
| 40 |
> |
my %workingDir; |
| 41 |
> |
my %location; |
| 42 |
> |
print "Found multiple database entries matching\n"; |
| 43 |
> |
print "\"$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 |
|
} |
| 58 |
– |
$crabDir = "$workingDir{$response}/ntuple"; |
| 59 |
– |
$outputDir = $location{$response}; |
| 66 |
|
} |
| 67 |
|
|
| 68 |
< |
if (!(-e $outputDir) || !(-d $outputDir)) |
| 68 |
> |
die "Output directory does not exist, stopped" if !$opt{"print"} && (!(-e $outputDir) || !(-d $outputDir)); |
| 69 |
> |
my @crabOutput; |
| 70 |
> |
if (!$opt{"print"}) |
| 71 |
|
{ |
| 72 |
< |
print "Output directory $outputDir does not exist!\n"; |
| 73 |
< |
exit; |
| 72 |
> |
opendir (CRAB_OUTPUT, "$outputDir"); |
| 73 |
> |
@crabOutput = readdir (CRAB_OUTPUT); |
| 74 |
> |
closedir (CRAB_OUTPUT); |
| 75 |
|
} |
| 67 |
– |
opendir (CRAB_OUTPUT, "$outputDir"); |
| 68 |
– |
my @crabOutput = readdir (CRAB_OUTPUT); |
| 69 |
– |
closedir (CRAB_OUTPUT); |
| 76 |
|
my %filesToKeep; |
| 77 |
|
if (!$opt{"dumb"}) |
| 78 |
|
{ |
| 79 |
< |
if (!(-e $crabDir) || !(-d $crabDir)) |
| 74 |
< |
{ |
| 75 |
< |
print "CRAB directory $crabDir does not exist!\n"; |
| 76 |
< |
exit; |
| 77 |
< |
} |
| 79 |
> |
die "CRAB directory does not exist, stopped" if !(-e $crabDir) || !(-d $crabDir); |
| 80 |
|
parseCrabDir ($crabDir, \%filesToKeep); |
| 81 |
|
} |
| 82 |
|
else |
| 96 |
|
} |
| 97 |
|
} |
| 98 |
|
} |
| 99 |
+ |
if ($opt{"destination"}) |
| 100 |
+ |
{ |
| 101 |
+ |
mkdir $opt{"destination"} or die "Failed to create destination directory, stopped" if !(-e $opt{"destination"}); |
| 102 |
+ |
die "Destination is not a directory, stopped" if !(-d $opt{"destination"}); |
| 103 |
+ |
open (TEST_FILE, ">$opt{'destination'}/.crabCleanTest") or die "Unable to write to destination directory, stopped"; |
| 104 |
+ |
close (TEST_FILE); |
| 105 |
+ |
unlink "$opt{'destination'}/.crabCleanTest"; |
| 106 |
+ |
} |
| 107 |
+ |
if (!$opt{"print"}) |
| 108 |
+ |
{ |
| 109 |
+ |
opendir (CRAB_OUTPUT, "$outputDir"); |
| 110 |
+ |
my @crabOutput = readdir (CRAB_OUTPUT); |
| 111 |
+ |
closedir (CRAB_OUTPUT); |
| 112 |
+ |
my $nFiles = @crabOutput - 2; |
| 113 |
+ |
print "Before cleaning: $nFiles files\n"; |
| 114 |
+ |
} |
| 115 |
|
foreach my $file (@crabOutput) |
| 116 |
|
{ |
| 117 |
|
next if ($file eq "." || $file eq ".."); |
| 124 |
|
{ |
| 125 |
|
print "No information found for job $jobNumber!\n"; |
| 126 |
|
} |
| 127 |
< |
elsif (($opt{"dumb"} && $filesToKeep{$jobNumber} != $submissionNumber) |
| 127 |
> |
elsif ((($opt{"dumb"} && $filesToKeep{$jobNumber} != $submissionNumber) |
| 128 |
|
|| (!$opt{"dumb"} && $filesToKeep{$jobNumber} ne $file)) |
| 129 |
+ |
&& !$opt{"destination"} && !$opt{"print"}) |
| 130 |
|
{ |
| 131 |
|
system ("rm -f $outputDir/$file") if $opt{"force"}; |
| 132 |
|
system ("rm -i $outputDir/$file") if !$opt{"force"}; |
| 133 |
|
} |
| 134 |
+ |
elsif ((($opt{"dumb"} && $filesToKeep{$jobNumber} == $submissionNumber) |
| 135 |
+ |
|| (!$opt{"dumb"} && $filesToKeep{$jobNumber} eq $file)) |
| 136 |
+ |
&& $opt{"destination"} && !$opt{"print"}) |
| 137 |
+ |
{ |
| 138 |
+ |
system ("mv -f $outputDir/$file $opt{'destination'}/$file") if $opt{"force"}; |
| 139 |
+ |
system ("mv -i $outputDir/$file $opt{'destination'}/$file") if !$opt{"force"}; |
| 140 |
+ |
} |
| 141 |
+ |
} |
| 142 |
+ |
if (!$opt{"print"}) |
| 143 |
+ |
{ |
| 144 |
+ |
opendir (CRAB_OUTPUT, "$outputDir"); |
| 145 |
+ |
my @crabOutput = readdir (CRAB_OUTPUT); |
| 146 |
+ |
closedir (CRAB_OUTPUT); |
| 147 |
+ |
opendir (CRAB_OUTPUT, "$opt{'destination'}"); |
| 148 |
+ |
my @crabOutputInDestination = readdir (CRAB_OUTPUT); |
| 149 |
+ |
closedir (CRAB_OUTPUT); |
| 150 |
+ |
my $nFiles = @crabOutput - 2; |
| 151 |
+ |
$nFiles = @crabOutputInDestination - 2 if $opt{"destination"}; |
| 152 |
+ |
print "After cleaning: $nFiles files\n"; |
| 153 |
+ |
} |
| 154 |
+ |
else |
| 155 |
+ |
{ |
| 156 |
+ |
foreach my $file (keys %filesToKeep) |
| 157 |
+ |
{ |
| 158 |
+ |
print $filesToKeep{$file} . "\n"; |
| 159 |
+ |
} |
| 160 |
|
} |
| 161 |
|
|
| 162 |
|
sub |
| 167 |
|
|
| 168 |
|
print "Usage: $exeName [OPTION]... CRAB_DIR OUTPUT_DIR\n"; |
| 169 |
|
print " or: $exeName -d [OPTION]... OUTPUT_DIR\n"; |
| 170 |
+ |
print " or: $exeName -p [OPTION]... CRAB_DIR\n"; |
| 171 |
|
print " or: $exeName -s [OPTION]... DATASET\n"; |
| 172 |
|
print "Deletes excess EDM output caused by multiple CRAB submissions using the results\n"; |
| 173 |
|
print "contained in the CRAB working directory provided.\n"; |
| 179 |
|
printf "%-29s%s\n", " -f, --force", "never prompt (default is to prompt before any"; |
| 180 |
|
printf "%-29s%s\n", " ", "removal)"; |
| 181 |
|
printf "%-29s%s\n", " -h, --help", "print this help message"; |
| 182 |
+ |
printf "%-29s%s\n", " -o, --destination DIR", "do not remove excess output and move the good files"; |
| 183 |
+ |
printf "%-29s%s\n", " ", "to DIR"; |
| 184 |
+ |
printf "%-29s%s\n", " -p, --print", "do not remove any files, just print the names of"; |
| 185 |
+ |
printf "%-29s%s\n", " ", "the good files to the screen"; |
| 186 |
|
printf "%-29s%s\n", " -s, --dataset", "tells the script to expect a dataset name instead"; |
| 187 |
|
printf "%-29s%s\n", " ", "of a directory"; |
| 188 |
|
|
| 212 |
|
print "Job $jobNumber failed (return value $wrapperStatus). Skipping.\n"; |
| 213 |
|
next; |
| 214 |
|
} |
| 215 |
< |
my @surlForGrid = `grep -A 1 '<SurlForGrid>' $crabDir/res/$file`; |
| 215 |
> |
my @surlForGrid = `grep -A 1 '<SurlForGrid' $crabDir/res/$file`; |
| 216 |
|
my $fileName = $surlForGrid[1]; |
| 217 |
< |
$fileName =~ s/^.*\/([^\/]*)\n/$1/; |
| 217 |
> |
$fileName = $surlForGrid[0] if $surlForGrid[0] =~ m/Value=/; |
| 218 |
> |
$fileName =~ s/^.*\/([^\/]*)\n/$1/ if !($fileName =~ m/Value=/); |
| 219 |
> |
$fileName =~ s/^.*\/([^\/]*)"\/>\n/$1/ if $fileName =~ m/Value=/; |
| 220 |
|
$filesToKeep->{$jobNumber} = $fileName; |
| 221 |
|
} |
| 222 |
|
} |
