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

sub
outputPset
{
  my $workingDir = 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";
  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;

  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";
  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 CONFIG DIRECTORY NJOBS\n";
  print "  or:  $exeName [OPTION]... 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 first argument must be either a DATASET registered in the Tier 3 ntuple\n";
  print "database or a LOCATION which exists on disk. CONFIG must be a valid CMSSW\n";
  print "python configuration which can be used with cmsRun. DIRECTORY is a working\n";
  print "directory that is created and in which all output, both from the CMSSW jobs and\n";
  print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
  print "be created.\n";

  exit;
}
Revision:	1.11
Committed:	Thu Jan 17 20:51:46 2013 UTC (12 years, 3 months ago) by ahart
Branch:	MAIN
Changes since 1.10:	+33 -14 lines
Log Message:	Allow for a location to be given instead of a dataset name.
#	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			printHelp () if $argc != 4;
24			if (!(-e $ARGV[1]))
25			{
26			print "\"$ARGV[1]\" does not exist!\n";
27			exit;
28			}
29			if (-e $ARGV[2])
30			{
31			print "Directory \"$ARGV[2]\" already exists!\n";
32			print "Please delete it or specify another working directory.\n";
33			exit;
34			}
35			mkdir $ARGV[2];
36	ahart	1.5	my $nFiles = outputRunList ($ARGV[0], $ARGV[2]);
37			my $nJobs = $ARGV[3];
38			my $realNJobs = ceil ($nFiles / ceil ($nFiles / $nJobs));
39	ahart	1.1	outputPset ($ARGV[2]);
40	ahart	1.8	outputCondor ($ARGV[2], $realNJobs, $ARGV[0]);
41	ahart	1.1	copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
42			chdir $ARGV[2];
43	ahart	1.5	print "Submitting $realNJobs jobs to run on $nFiles files.\n";
44	ahart	1.1	system ("condor_submit condor.sub");
45
46			sub
47			outputPset
48			{
49			my $workingDir = shift;
50
51			open (PSET, ">$workingDir/config_cfg.py");
52
53			print PSET "import FWCore.ParameterSet.Config as cms\n";
54			print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
55			print PSET "import re\n";
56			print PSET "import userConfig_cfg as pset\n";
57			print PSET "\n";
58			print PSET "fileName = pset.process.TFileService.fileName\n";
59			print PSET "fileName = fileName.pythonValue ()\n";
60			print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
61			print PSET "fileName = re.sub (r'^(.)\\.([^\\.])\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
62			print PSET "pset.process.TFileService.fileName = fileName\n";
63			print PSET "\n";
64			print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
65			print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
66			print PSET "process = pset.process\n";
67
68			close (PSET);
69			}
70
71			sub
72			outputRunList
73			{
74			my $dataset = shift;
75			my $workingDir = shift;
76
77	ahart	1.11	my $location;
78			my $nFiles;
79			my $status;
80			my $crossSection;
81			my $isLocation = 0;
82			$isLocation = 1 if -e $dataset;
83			if (!$isLocation)
84			{
85			($location, $nFiles, $status, $crossSection) = getLocation ($dataset);
86			if ($status ne "present")
87			{
88			print "This dataset is not marked as present on the Tier 3!\n";
89			print "Continue anyway? (y/N): ";
90			my $response = <STDIN>;
91			$response =~ s/\n//g;
92			exit if !$response \|\| lc ($response) ne "y";
93			}
94			if (!(-e $location))
95			{
96			print "The database does not know where this dataset is!\n";
97			exit;
98			}
99	ahart	1.1	}
100	ahart	1.11	else
101	ahart	1.1	{
102	ahart	1.11	$location = $dataset;
103	ahart	1.1	}
104			opendir (LOCATION, $location);
105			my @files = readdir (LOCATION);
106			closedir (LOCATION);
107	ahart	1.11	if (!$isLocation && @files - 2 != $nFiles)
108	ahart	1.1	{
109			print "Number of files does not match database entry!\n";
110	ahart	1.9	print "Continue anyway? (y/N): ";
111			my $response = <STDIN>;
112			$response =~ s/\n//g;
113			exit if !$response \|\| lc ($response) ne "y";
114	ahart	1.1	}
115	ahart	1.11	elsif ($isLocation)
116			{
117			$nFiles = @files - 2;
118			}
119	ahart	1.1	open (RUNLIST, ">$workingDir/runList.py");
120			print RUNLIST "runList = [\n";
121			for (my $i = 0; $i < @files; $i++)
122			{
123			next if $files[$i] eq ".";
124			next if $files[$i] eq "..";
125			print RUNLIST "'file:$location/$files[$i]'";
126			print RUNLIST "," if $i + 1 != @files;
127			print RUNLIST "\n";
128			}
129			print RUNLIST "]";
130			close (RUNLIST);
131	ahart	1.6	if ($crossSection && $crossSection >= 0.0)
132			{
133			open (CROSS_SECTION, ">$workingDir/crossSectionInPicobarn.txt");
134			print CROSS_SECTION "$crossSection\n";
135			close (CROSS_SECTION);
136			}
137	ahart	1.5
138			return $nFiles;
139	ahart	1.1	}
140
141			sub
142			getLocation
143			{
144			my $dataset = shift;
145
146			my $results;
147	ahart	1.2	my $queryDataset = $dataset;
148			$queryDataset =~ s/\*/%/g;
149			$queryDataset =~ s/(.*)/%$1%/g;
150	ahart	1.9	my $query = "select dataset,user,creationTime,location,nFiles,status,crossSectionInPicobarn from ntuple where dataset like '$queryDataset' order by creationTime";
151	ahart	1.1	$db->selectdb ("ntuple");
152			$results = $db->query ($query);
153			if ($results->numrows () == 1)
154			{
155			my @row = $results->fetchrow ();
156	ahart	1.6	return ($row[3], $row[4], $row[5], $row[6]);
157	ahart	1.1	}
158			if ($results->numrows () == 0)
159			{
160			print "Dataset does not exist on the Tier 3!\n";
161			exit;
162			}
163			my %map;
164			print "Found multiple datasets matching\n";
165			print "\"$dataset\":\n";
166			for (my $i = 1; $i <= $results->numrows (); $i++)
167			{
168			my @row = $results->fetchrow ();
169	ahart	1.6	$map{"$i"} = [$row[3], $row[4], $row[5], $row[6]];
170	ahart	1.2	printf "(%2d) $row[0]\n", $i;
171			print " created by $row[1] on $row[2]\n";
172	ahart	1.1	}
173			print "\nWhich dataset would you like to use?: ";
174			my $response = <STDIN>;
175			$response =~ s/[ \t\n]//g;
176			if (!(exists $map{$response}))
177			{
178			print "Your selection \"$response\" was not a valid option! Quitting.\n";
179			exit;
180			}
181
182	ahart	1.6	return ($map{$response}[0], $map{$response}[1], $map{$response}[2], $map{$response}[3]);
183	ahart	1.1	}
184
185			sub
186			outputCondor
187			{
188			my $workingDir = shift;
189			my $nJobs = shift;
190	ahart	1.8	my $dataset = shift;
191	ahart	1.1
192			my $cmsRun = `which cmsRun`;
193			open (SUB, ">$workingDir/condor.sub");
194
195			print SUB "Executable = $cmsRun\n";
196			print SUB "Universe = vanilla\n";
197			print SUB "Getenv = True\n";
198	ahart	1.10	print SUB "Arguments = config_cfg.py True $nJobs \$(Process) $dataset\n";
199	ahart	1.1	print SUB "\n";
200			print SUB "Output = condor_\$(Process).out\n";
201			print SUB "Error = condor_\$(Process).err\n";
202			print SUB "Log = condor_\$(Process).log\n";
203			print SUB "\n";
204	ahart	1.7	print SUB "+IsLocalJob = true\n";
205			print SUB "Rank = TARGET.IsLocalSlot\n";
206			print SUB "\n";
207	ahart	1.1	print SUB "Queue $nJobs\n";
208
209			close (SUB);
210			}
211
212			sub
213			printHelp
214			{
215			my $exeName = $0;
216			$exeName =~ s/^.\/([^\/])$/$1/;
217
218	ahart	1.3	print "Usage: $exeName [OPTION]... DATASET CONFIG DIRECTORY NJOBS\n";
219	ahart	1.11	print " or: $exeName [OPTION]... LOCATION CONFIG DIRECTORY NJOBS\n";
220	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
221			print "\n";
222			printf "%-29s%s\n", " -h, --help", "print this help message";
223			print "\n";
224	ahart	1.11	print "The first argument must be either a DATASET registered in the Tier 3 ntuple\n";
225			print "database or a LOCATION which exists on disk. CONFIG must be a valid CMSSW\n";
226			print "python configuration which can be used with cmsRun. DIRECTORY is a working\n";
227	ahart	1.1	print "directory that is created and in which all output, both from the CMSSW jobs and\n";
228			print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
229			print "be created.\n";
230
231			exit;
232			}