[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, "maxEvents|m=s", "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 = $nJobs;
$realNJobs = ceil ($nFiles / ceil ($nFiles / $realNJobs)) if $nFiles;
my $eventsPerJob = -1;
$eventsPerJob = ceil ($opt{"maxEvents"} / $realNJobs) if $opt{"maxEvents"} && $opt{"maxEvents"} >= 0;
my $realMaxEvents = $eventsPerJob * $realNJobs;
outputPset ($directory, $dataset, $opt{"maxEvents"}, $eventsPerJob);
outputCondor ($directory, $realNJobs, $dataset);
copy ($config, "$directory/userConfig_cfg.py");
chdir $directory;
print "Submitting $realNJobs jobs to run on $realMaxEvents events in $nFiles files.\n" if $realMaxEvents >= 0;
print "Submitting $realNJobs jobs to run on all events in $nFiles files.\n" if $realMaxEvents < 0;
system ("condor_submit condor.sub");

sub
outputPset
{
  my $workingDir = shift;
  my $dataset = shift;
  my $nEvents = shift;
  my $eventsPerJob = 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 ($dataset)
    {
      print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
      print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob < 0;
    }
  print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob >= 0;
  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 $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 $nJobs \$(Process) $dataset\n" if $dataset;
  print SUB "Arguments               = config_cfg.py True $nJobs \$(Process) NULL\n" if !$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";
  printf "%-29s%s\n", "  -m, --maxEvents N", "only run over N events in the dataset; default is";
  printf "%-29s%s\n", " ", "to run over all events";
  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:	1.13
Committed:	Fri Feb 8 09:17:45 2013 UTC (12 years, 2 months ago) by ahart
Branch:	MAIN
Changes since 1.12:	+18 -10 lines
Log Message:	Added an option for the maximum number of events to process for a dataset. Actual number processed may differ from the target since job splitting is based on the number of files.
#	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	ahart	1.13	GetOptions (\%opt, "maxEvents\|m=s", "help\|h");
20	ahart	1.1	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	ahart	1.13	my $realNJobs = $nJobs;
56			$realNJobs = ceil ($nFiles / ceil ($nFiles / $realNJobs)) if $nFiles;
57			my $eventsPerJob = -1;
58			$eventsPerJob = ceil ($opt{"maxEvents"} / $realNJobs) if $opt{"maxEvents"} && $opt{"maxEvents"} >= 0;
59			my $realMaxEvents = $eventsPerJob * $realNJobs;
60			outputPset ($directory, $dataset, $opt{"maxEvents"}, $eventsPerJob);
61			outputCondor ($directory, $realNJobs, $dataset);
62	ahart	1.12	copy ($config, "$directory/userConfig_cfg.py");
63			chdir $directory;
64	ahart	1.13	print "Submitting $realNJobs jobs to run on $realMaxEvents events in $nFiles files.\n" if $realMaxEvents >= 0;
65			print "Submitting $realNJobs jobs to run on all events in $nFiles files.\n" if $realMaxEvents < 0;
66	ahart	1.1	system ("condor_submit condor.sub");
67
68			sub
69			outputPset
70			{
71			my $workingDir = shift;
72	ahart	1.12	my $dataset = shift;
73	ahart	1.13	my $nEvents = shift;
74			my $eventsPerJob = shift;
75	ahart	1.1
76			open (PSET, ">$workingDir/config_cfg.py");
77
78			print PSET "import FWCore.ParameterSet.Config as cms\n";
79			print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
80			print PSET "import re\n";
81			print PSET "import userConfig_cfg as pset\n";
82			print PSET "\n";
83			print PSET "fileName = pset.process.TFileService.fileName\n";
84			print PSET "fileName = fileName.pythonValue ()\n";
85			print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
86			print PSET "fileName = re.sub (r'^(.)\\.([^\\.])\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
87			print PSET "pset.process.TFileService.fileName = fileName\n";
88			print PSET "\n";
89	ahart	1.12	if ($dataset)
90			{
91			print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
92	ahart	1.13	print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob < 0;
93	ahart	1.12	}
94	ahart	1.13	print PSET "pset.process.maxEvents.input = cms.untracked.int32 ($eventsPerJob)\n" if $eventsPerJob >= 0;
95	ahart	1.1	print PSET "process = pset.process\n";
96
97			close (PSET);
98			}
99
100			sub
101			outputRunList
102			{
103			my $dataset = shift;
104			my $workingDir = shift;
105
106	ahart	1.12	return 0 if !$dataset;
107	ahart	1.11	my $location;
108			my $nFiles;
109			my $status;
110			my $crossSection;
111			my $isLocation = 0;
112			$isLocation = 1 if -e $dataset;
113			if (!$isLocation)
114			{
115			($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
116			if ($status ne "present")
117			{
118			print "This dataset is not marked as present on the Tier 3!\n";
119			print "Continue anyway? (y/N): ";
120			my $response = <STDIN>;
121			$response =~ s/\n//g;
122			exit if !$response \|\| lc ($response) ne "y";
123			}
124			if (!(-e $location))
125			{
126			print "The database does not know where this dataset is!\n";
127			exit;
128			}
129	ahart	1.1	}
130	ahart	1.11	else
131	ahart	1.1	{
132	ahart	1.11	$location = $dataset;
133	ahart	1.1	}
134			opendir (LOCATION, $location);
135			my @files = readdir (LOCATION);
136			closedir (LOCATION);
137	ahart	1.11	if (!$isLocation && @files - 2 != $nFiles)
138	ahart	1.1	{
139			print "Number of files does not match database entry!\n";
140	ahart	1.9	print "Continue anyway? (y/N): ";
141			my $response = <STDIN>;
142			$response =~ s/\n//g;
143			exit if !$response \|\| lc ($response) ne "y";
144	ahart	1.1	}
145	ahart	1.11	elsif ($isLocation)
146			{
147			$nFiles = @files - 2;
148			}
149	ahart	1.1	open (RUNLIST, ">$workingDir/runList.py");
150			print RUNLIST "runList = [\n";
151			for (my $i = 0; $i < @files; $i++)
152			{
153			next if $files[$i] eq ".";
154			next if $files[$i] eq "..";
155			print RUNLIST "'file:$location/$files[$i]'";
156			print RUNLIST "," if $i + 1 != @files;
157			print RUNLIST "\n";
158			}
159			print RUNLIST "]";
160			close (RUNLIST);
161	ahart	1.6	if ($crossSection && $crossSection >= 0.0)
162			{
163			open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
164			print CROSS_SECTION "$crossSection\n";
165			close (CROSS_SECTION);
166			}
167	ahart	1.5
168			return $nFiles;
169	ahart	1.1	}
170
171			sub
172			getLocation
173			{
174			my $dataset = shift;
175
176			my $results;
177	ahart	1.2	my $queryDataset = $dataset;
178			$queryDataset =~ s/\*/%/g;
179			$queryDataset =~ s/(.*)/%$1%/g;
180	ahart	1.9	my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
181	ahart	1.1	$db->selectdb ("ntuple");
182			$results = $db->query ($query);
183			if ($results->numrows () == 1)
184			{
185			my @row = $results->fetchrow ();
186	ahart	1.6	return ($row[3], $row[4], $row[5], $row[6]);
187	ahart	1.1	}
188			if ($results->numrows () == 0)
189			{
190			print "Dataset does not exist on the Tier 3!\n";
191			exit;
192			}
193			my %map;
194			print "Found multiple datasets matching\n";
195			print "\"$dataset\":\n";
196			for (my $i = 1; $i <= $results->numrows (); $i++)
197			{
198			my @row = $results->fetchrow ();
199	ahart	1.6	$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]];
200	ahart	1.2	printf "(%2d) $row[0]\n", $i;
201			print " created by $row[1] on $row[2]\n";
202	ahart	1.1	}
203			print "\nWhich dataset would you like to use?: ";
204			my $response = <STDIN>;
205			$response =~ s/[ \t\n]//g;
206			if (!(exists $map{$response}))
207			{
208			print "Your selection \"$response\" was not a valid option! Quitting.\n";
209			exit;
210			}
211
212	ahart	1.6	return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
213	ahart	1.1	}
214
215			sub
216			outputCondor
217			{
218			my $workingDir = shift;
219			my $nJobs = shift;
220	ahart	1.8	my $dataset = shift;
221	ahart	1.1
222			my $cmsRun = `which cmsRun`;
223			open (SUB, ">$workingDir/condor.sub");
224
225			print SUB "Executable = $cmsRun\n";
226			print SUB "Universe = vanilla\n";
227			print SUB "Getenv = True\n";
228	ahart	1.13	print SUB "Arguments = config_cfg.py True $nJobs \$(Process) $dataset\n" if $dataset;
229			print SUB "Arguments = config_cfg.py True $nJobs \$(Process) NULL\n" if !$dataset;
230	ahart	1.1	print SUB "\n";
231			print SUB "Output = condor_\$(Process).out\n";
232			print SUB "Error = condor_\$(Process).err\n";
233			print SUB "Log = condor_\$(Process).log\n";
234			print SUB "\n";
235	ahart	1.7	print SUB "+IsLocalJob = true\n";
236			print SUB "Rank = TARGET.IsLocalSlot\n";
237			print SUB "\n";
238	ahart	1.1	print SUB "Queue $nJobs\n";
239
240			close (SUB);
241			}
242
243			sub
244			printHelp
245			{
246			my $exeName = $0;
247			$exeName =~ s/^.\/([^\/])$/$1/;
248
249	ahart	1.12	print "Usage: $exeName [OPTION]... [DATASET \| LOCATION] CONFIG DIRECTORY NJOBS\n";
250	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
251			print "\n";
252			printf "%-29s%s\n", " -h, --help", "print this help message";
253	ahart	1.13	printf "%-29s%s\n", " -m, --maxEvents N", "only run over N events in the dataset; default is";
254			printf "%-29s%s\n", " ", "to run over all events";
255	ahart	1.1	print "\n";
256	ahart	1.12	print "The optional first argument must be either a DATASET registered in the Tier 3\n";
257			print "ntuple database or a LOCATION which exists on disk. CONFIG must be a valid\n";
258			print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a\n";
259			print "working directory that is created and in which all output, both from the CMSSW\n";
260			print "jobs and from Condor, is placed. Finally, NJOBS is the number of Condor jobs\n";
261			print "that will be created.\n";
262	ahart	1.1
263			exit;
264			}