[ViewVC] Annotation of: cvsroot/UserCode/OSUT3Analysis/DBTools/scripts/osusub

#!/usr/bin/env perl

use strict;
use Mysql;
use File::Copy;
use Getopt::Long;
use POSIX;

sub printHelp;
sub outputPset;
sub outputRunList;
sub getLocation;
sub outputCondor;

our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");

my %opt;
Getopt::Long::Configure ("bundling");
GetOptions (\%opt, "help|h");
my $argc = @ARGV;

printHelp () if $opt{"help"};
printHelp () if $argc != 3 && $argc != 4;
my $dataset;
my $config;
my $directory;
my $nJobs;
if ($argc == 3)
  {
    $dataset = "";
    $config = $ARGV[0];
    $directory = $ARGV[1];
    $nJobs = $ARGV[2];
  }
if ($argc == 4)
  {
    $dataset = $ARGV[0];
    $config = $ARGV[1];
    $directory = $ARGV[2];
    $nJobs = $ARGV[3];
  }
if (!(-e $config))
  {
    print "\"$config\" does not exist!\n";
    exit;
  }
if (-e $directory)
  {
    print "Directory \"$directory\" already exists!\n";
    print "Please delete it or specify another working directory.\n";
    exit;
  }
mkdir $directory;
my $nFiles = outputRunList ($dataset, $directory);
my $realNJobs = 0;
$realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs)) if $nFiles;
outputPset ($directory, $dataset);
outputCondor ($directory, $realNJobs, $nJobs, $dataset);
copy ($config, "$directory/userConfig_cfg.py");
chdir $directory;
print "Submitting $realNJobs jobs to run on $nFiles files.\n";
system ("condor_submit condor.sub");

sub
outputPset
{
  my $workingDir = shift;
  my $dataset = shift;

  open (PSET, ">$workingDir/config_cfg.py");

  print PSET "import FWCore.ParameterSet.Config as cms\n";
  print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
  print PSET "import re\n";
  print PSET "import userConfig_cfg as pset\n";
  print PSET "\n";
  print PSET "fileName = pset.process.TFileService.fileName\n";
  print PSET "fileName = fileName.pythonValue ()\n";
  print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
  print PSET "fileName = re.sub (r'^(.*)\\.([^\\.]*)\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
  print PSET "pset.process.TFileService.fileName = fileName\n";
  print PSET "\n";
  if ($dataset)
    {
      print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
      print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
    }
  print PSET "process = pset.process\n";

  close (PSET);
}

sub
outputRunList
{
  my $dataset = shift;
  my $workingDir = shift;

  return 0 if !$dataset;
  my $location;
  my $nFiles;
  my $status;
  my $crossSection;
  my $isLocation = 0;
  $isLocation = 1 if -e $dataset;
  if (!$isLocation)
    {
      ($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
      if ($status ne "present")
        {
          print "This dataset is not marked as present on the Tier 3!\n";
          print "Continue anyway? (y/N): ";
          my $response = <STDIN>;
          $response =~ s/\n//g;
          exit if !$response || lc ($response) ne "y";
        }
      if (!(-e $location))
        {
          print "The database does not know where this dataset is!\n";
          exit;
        }
    }
  else
    {
      $location = $dataset;
    }
  opendir (LOCATION, $location);
  my @files = readdir (LOCATION);
  closedir (LOCATION);
  if (!$isLocation && @files - 2 != $nFiles)
    {
      print "Number of files does not match database entry!\n";
      print "Continue anyway? (y/N): ";
      my $response = <STDIN>;
      $response =~ s/\n//g;
      exit if !$response || lc ($response) ne "y";
    }
  elsif ($isLocation)
    {
      $nFiles = @files - 2;
    }
  open (RUNLIST, ">$workingDir/runList.py");
  print RUNLIST "runList = [\n";
  for (my $i = 0; $i < @files; $i++)
    {
      next if $files[$i] eq ".";
      next if $files[$i] eq "..";
      print RUNLIST "'file:$location/$files[$i]'";
      print RUNLIST "," if $i + 1 != @files;
      print RUNLIST "\n";
    }
  print RUNLIST "]";
  close (RUNLIST);
  if ($crossSection && $crossSection >= 0.0)
    {
      open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
      print CROSS_SECTION "$crossSection\n";
      close (CROSS_SECTION);
    }

  return $nFiles;
}

sub
getLocation
{
  my $dataset = shift;

  my $results;
  my $queryDataset = $dataset;
  $queryDataset =~ s/\*/%/g;
  $queryDataset =~ s/(.*)/%$1%/g;
  my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
  $db->selectdb ("ntuple");
  $results = $db->query ($query);
  if ($results->numrows () == 1)
    {
      my @row = $results->fetchrow ();
      return ($row[3], $row[4], $row[5], $row[6]);
    }
  if ($results->numrows () == 0)
    {
      print "Dataset does not exist on the Tier 3!\n";
      exit;
    }
  my %map;
  print "Found multiple datasets matching\n";
  print "\"$dataset\":\n";
  for (my $i = 1; $i <= $results->numrows (); $i++)
    {
      my @row = $results->fetchrow ();
      $map{"$i"} = [$row[3], $row[4], $row[5], $row[6]];
      printf "(%2d) $row[0]\n", $i;
      print "     created by $row[1] on $row[2]\n";
    }
  print "\nWhich dataset would you like to use?: ";
  my $response = <STDIN>;
  $response =~ s/[ \t\n]//g;
  if (!(exists $map{$response}))
    {
      print "Your selection \"$response\" was not a valid option! Quitting.\n";
      exit;
    }

  return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
}

sub
outputCondor
{
  my $workingDir = shift;
  my $realNJobs = shift;
  my $nJobs = shift;
  my $dataset = shift;

  my $cmsRun = `which cmsRun`;
  open (SUB, ">$workingDir/condor.sub");

  print SUB "Executable              = $cmsRun\n";
  print SUB "Universe                = vanilla\n";
  print SUB "Getenv                  = True\n";
  print SUB "Arguments               = config_cfg.py True $realNJobs \$(Process) $dataset\n" if $dataset;
  print SUB "Arguments               = config_cfg.py True $realNJobs \$(Process) NULL\n" if !$dataset;
  print SUB "\n";
  print SUB "Output                  = condor_\$(Process).out\n";
  print SUB "Error                   = condor_\$(Process).err\n";
  print SUB "Log                     = condor_\$(Process).log\n";
  print SUB "\n";
  print SUB "+IsLocalJob             = true\n";
  print SUB "Rank                    = TARGET.IsLocalSlot\n";
  print SUB "\n";
  print SUB "Queue $nJobs\n";

  close (SUB);
}

sub
printHelp
{
  my $exeName = $0;
  $exeName =~ s/^.*\/([^\/]*)$/$1/;

  print "Usage: $exeName [OPTION]... [DATASET | LOCATION] CONFIG DIRECTORY NJOBS\n";
  print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
  print "\n";
  printf "%-29s%s\n", "  -h, --help", "print this help message";
  print "\n";
  print "The optional first argument must be either a DATASET registered in the Tier 3\n";
  print "ntuple database or a LOCATION which exists on disk. CONFIG must be a valid\n";
  print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a\n";
  print "working directory that is created and in which all output, both from the CMSSW\n";
  print "jobs and from Condor, is placed. Finally, NJOBS is the number of Condor jobs\n";
  print "that will be created.\n";

  exit;
}
Revision:	1.12
Committed:	Mon Feb 4 04:14:43 2013 UTC (12 years, 3 months ago) by ahart
Branch:	MAIN
Changes since 1.11:	+48 -24 lines
Log Message:	Allow the dataset name to be omitted as in the case of MC generation.
#	User	Rev	Content
1	ahart	1.1	#!/usr/bin/env perl
2
3			use strict;
4			use Mysql;
5			use File::Copy;
6			use Getopt::Long;
7	ahart	1.5	use POSIX;
8	ahart	1.2
9	ahart	1.1	sub printHelp;
10			sub outputPset;
11			sub outputRunList;
12			sub getLocation;
13			sub outputCondor;
14
15	ahart	1.3	our $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
16	ahart	1.1
17			my %opt;
18			Getopt::Long::Configure ("bundling");
19			GetOptions (\%opt, "help\|h");
20			my $argc = @ARGV;
21
22			printHelp () if $opt{"help"};
23	ahart	1.12	printHelp () if $argc != 3 && $argc != 4;
24			my $dataset;
25			my $config;
26			my $directory;
27			my $nJobs;
28			if ($argc == 3)
29	ahart	1.1	{
30	ahart	1.12	$dataset = "";
31			$config = $ARGV[0];
32			$directory = $ARGV[1];
33			$nJobs = $ARGV[2];
34			}
35			if ($argc == 4)
36			{
37			$dataset = $ARGV[0];
38			$config = $ARGV[1];
39			$directory = $ARGV[2];
40			$nJobs = $ARGV[3];
41			}
42			if (!(-e $config))
43			{
44			print "\"$config\" does not exist!\n";
45	ahart	1.1	exit;
46			}
47	ahart	1.12	if (-e $directory)
48	ahart	1.1	{
49	ahart	1.12	print "Directory \"$directory\" already exists!\n";
50	ahart	1.1	print "Please delete it or specify another working directory.\n";
51			exit;
52			}
53	ahart	1.12	mkdir $directory;
54			my $nFiles = outputRunList ($dataset, $directory);
55			my $realNJobs = 0;
56			$realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs)) if $nFiles;
57			outputPset ($directory, $dataset);
58			outputCondor ($directory, $realNJobs, $nJobs, $dataset);
59			copy ($config, "$directory/userConfig_cfg.py");
60			chdir $directory;
61	ahart	1.5	print "Submitting $realNJobs jobs to run on $nFiles files.\n";
62	ahart	1.1	system ("condor_submit condor.sub");
63
64			sub
65			outputPset
66			{
67			my $workingDir = shift;
68	ahart	1.12	my $dataset = shift;
69	ahart	1.1
70			open (PSET, ">$workingDir/config_cfg.py");
71
72			print PSET "import FWCore.ParameterSet.Config as cms\n";
73			print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
74			print PSET "import re\n";
75			print PSET "import userConfig_cfg as pset\n";
76			print PSET "\n";
77			print PSET "fileName = pset.process.TFileService.fileName\n";
78			print PSET "fileName = fileName.pythonValue ()\n";
79			print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
80			print PSET "fileName = re.sub (r'^(.)\\.([^\\.])\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
81			print PSET "pset.process.TFileService.fileName = fileName\n";
82			print PSET "\n";
83	ahart	1.12	if ($dataset)
84			{
85			print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
86			print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
87			}
88	ahart	1.1	print PSET "process = pset.process\n";
89
90			close (PSET);
91			}
92
93			sub
94			outputRunList
95			{
96			my $dataset = shift;
97			my $workingDir = shift;
98
99	ahart	1.12	return 0 if !$dataset;
100	ahart	1.11	my $location;
101			my $nFiles;
102			my $status;
103			my $crossSection;
104			my $isLocation = 0;
105			$isLocation = 1 if -e $dataset;
106			if (!$isLocation)
107			{
108			($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
109			if ($status ne "present")
110			{
111			print "This dataset is not marked as present on the Tier 3!\n";
112			print "Continue anyway? (y/N): ";
113			my $response = <STDIN>;
114			$response =~ s/\n//g;
115			exit if !$response \|\| lc ($response) ne "y";
116			}
117			if (!(-e $location))
118			{
119			print "The database does not know where this dataset is!\n";
120			exit;
121			}
122	ahart	1.1	}
123	ahart	1.11	else
124	ahart	1.1	{
125	ahart	1.11	$location = $dataset;
126	ahart	1.1	}
127			opendir (LOCATION, $location);
128			my @files = readdir (LOCATION);
129			closedir (LOCATION);
130	ahart	1.11	if (!$isLocation && @files - 2 != $nFiles)
131	ahart	1.1	{
132			print "Number of files does not match database entry!\n";
133	ahart	1.9	print "Continue anyway? (y/N): ";
134			my $response = <STDIN>;
135			$response =~ s/\n//g;
136			exit if !$response \|\| lc ($response) ne "y";
137	ahart	1.1	}
138	ahart	1.11	elsif ($isLocation)
139			{
140			$nFiles = @files - 2;
141			}
142	ahart	1.1	open (RUNLIST, ">$workingDir/runList.py");
143			print RUNLIST "runList = [\n";
144			for (my $i = 0; $i < @files; $i++)
145			{
146			next if $files[$i] eq ".";
147			next if $files[$i] eq "..";
148			print RUNLIST "'file:$location/$files[$i]'";
149			print RUNLIST "," if $i + 1 != @files;
150			print RUNLIST "\n";
151			}
152			print RUNLIST "]";
153			close (RUNLIST);
154	ahart	1.6	if ($crossSection && $crossSection >= 0.0)
155			{
156			open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
157			print CROSS_SECTION "$crossSection\n";
158			close (CROSS_SECTION);
159			}
160	ahart	1.5
161			return $nFiles;
162	ahart	1.1	}
163
164			sub
165			getLocation
166			{
167			my $dataset = shift;
168
169			my $results;
170	ahart	1.2	my $queryDataset = $dataset;
171			$queryDataset =~ s/\*/%/g;
172			$queryDataset =~ s/(.*)/%$1%/g;
173	ahart	1.9	my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
174	ahart	1.1	$db->selectdb ("ntuple");
175			$results = $db->query ($query);
176			if ($results->numrows () == 1)
177			{
178			my @row = $results->fetchrow ();
179	ahart	1.6	return ($row[3], $row[4], $row[5], $row[6]);
180	ahart	1.1	}
181			if ($results->numrows () == 0)
182			{
183			print "Dataset does not exist on the Tier 3!\n";
184			exit;
185			}
186			my %map;
187			print "Found multiple datasets matching\n";
188			print "\"$dataset\":\n";
189			for (my $i = 1; $i <= $results->numrows (); $i++)
190			{
191			my @row = $results->fetchrow ();
192	ahart	1.6	$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]];
193	ahart	1.2	printf "(%2d) $row[0]\n", $i;
194			print " created by $row[1] on $row[2]\n";
195	ahart	1.1	}
196			print "\nWhich dataset would you like to use?: ";
197			my $response = <STDIN>;
198			$response =~ s/[ \t\n]//g;
199			if (!(exists $map{$response}))
200			{
201			print "Your selection \"$response\" was not a valid option! Quitting.\n";
202			exit;
203			}
204
205	ahart	1.6	return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
206	ahart	1.1	}
207
208			sub
209			outputCondor
210			{
211			my $workingDir = shift;
212	ahart	1.12	my $realNJobs = shift;
213	ahart	1.1	my $nJobs = shift;
214	ahart	1.8	my $dataset = shift;
215	ahart	1.1
216			my $cmsRun = `which cmsRun`;
217			open (SUB, ">$workingDir/condor.sub");
218
219			print SUB "Executable = $cmsRun\n";
220			print SUB "Universe = vanilla\n";
221			print SUB "Getenv = True\n";
222	ahart	1.12	print SUB "Arguments = config_cfg.py True $realNJobs \$(Process) $dataset\n" if $dataset;
223			print SUB "Arguments = config_cfg.py True $realNJobs \$(Process) NULL\n" if !$dataset;
224	ahart	1.1	print SUB "\n";
225			print SUB "Output = condor_\$(Process).out\n";
226			print SUB "Error = condor_\$(Process).err\n";
227			print SUB "Log = condor_\$(Process).log\n";
228			print SUB "\n";
229	ahart	1.7	print SUB "+IsLocalJob = true\n";
230			print SUB "Rank = TARGET.IsLocalSlot\n";
231			print SUB "\n";
232	ahart	1.1	print SUB "Queue $nJobs\n";
233
234			close (SUB);
235			}
236
237			sub
238			printHelp
239			{
240			my $exeName = $0;
241			$exeName =~ s/^.\/([^\/])$/$1/;
242
243	ahart	1.12	print "Usage: $exeName [OPTION]... [DATASET \| LOCATION] CONFIG DIRECTORY NJOBS\n";
244	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
245			print "\n";
246			printf "%-29s%s\n", " -h, --help", "print this help message";
247			print "\n";
248	ahart	1.12	print "The optional first argument must be either a DATASET registered in the Tier 3\n";
249			print "ntuple database or a LOCATION which exists on disk. CONFIG must be a valid\n";
250			print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a\n";
251			print "working directory that is created and in which all output, both from the CMSSW\n";
252			print "jobs and from Condor, is placed. Finally, NJOBS is the number of Condor jobs\n";
253			print "that will be created.\n";
254	ahart	1.1
255			exit;
256			}