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

#!/usr/bin/env perl

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

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

our $db;

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 $ARGV[2];
outputRunList ($ARGV[0], $ARGV[2]);
outputPset ($ARGV[2]);
outputCondor ($ARGV[2], $ARGV[3]);
copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
chdir $ARGV[2];
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, $nFiles, $status) = getLocation ($dataset);
  if ($status ne "present")
    {
      print "This dataset is not marked as present on the Tier 3!\n";
      exit;
    }
  if (!(-e $location))
    {
      print "The database does not know where this dataset is!\n";
      exit;
    }
  opendir (LOCATION, $location);
  my @files = readdir (LOCATION);
  closedir (LOCATION);
  if (@files - 2 != $nFiles)
    {
      print "Number of files does not match database entry!\n";
      exit;
    }
  open (RUNLIST, ">$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);
}

sub
getLocation
{
  my $dataset = shift;

  my $results;
  my $queryDataset = $dataset;
  $queryDataset =~ s/\*/%/g;
  $queryDataset =~ s/(.*)/%$1%/g;
  $db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
  my $query = "select dataset,user,creationTime,location,nFiles,status from ntuple where dataset like '$queryDataset' order by lastUpdateTime";
  $db->selectdb ("ntuple");
  $results = $db->query ($query);
  if ($results->numrows () == 1)
    {
      my @row = $results->fetchrow ();
      return ($row[3], $row[4], $row[5]);
    }
  if ($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]];
      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]);
}

sub
outputCondor
{
  my $workingDir = shift;
  my $nJobs = 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 $nJobs \$(Process)\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 "Queue $nJobs\n";

  close (SUB);
}

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

  print "Usage: $exeName [OPTIONS] DATASET 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 DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n";
  print "CMSSW 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:	1.2
Committed:	Thu Jun 14 17:42:30 2012 UTC (12 years, 10 months ago) by ahart
Branch:	MAIN
CVS Tags:	V00-00-02
Changes since 1.1:	+10 -6 lines
Log Message:	Allow fuzzy dataset names for osudb and osusub. First commit of AnaTools directory.
#	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	1.2
8	ahart	1.1	sub printHelp;
9			sub outputPset;
10			sub outputRunList;
11			sub getLocation;
12			sub outputCondor;
13
14			our $db;
15
16			my %opt;
17			Getopt::Long::Configure ("bundling");
18			GetOptions (\%opt, "help\|h");
19			my $argc = @ARGV;
20
21			printHelp () if $opt{"help"};
22			printHelp () if $argc != 4;
23			if (!(-e $ARGV[1]))
24			{
25			print "\"$ARGV[1]\" does not exist!\n";
26			exit;
27			}
28			if (-e $ARGV[2])
29			{
30			print "Directory \"$ARGV[2]\" already exists!\n";
31			print "Please delete it or specify another working directory.\n";
32			exit;
33			}
34			mkdir $ARGV[2];
35			outputRunList ($ARGV[0], $ARGV[2]);
36			outputPset ($ARGV[2]);
37			outputCondor ($ARGV[2], $ARGV[3]);
38			copy ($ARGV[1], "$ARGV[2]/userConfig_cfg.py");
39			chdir $ARGV[2];
40			system ("condor_submit condor.sub");
41
42			sub
43			outputPset
44			{
45			my $workingDir = shift;
46
47			open (PSET, ">$workingDir/config_cfg.py");
48
49			print PSET "import FWCore.ParameterSet.Config as cms\n";
50			print PSET "import OSUT3Analysis.DBTools.osusub_cfg as osusub\n";
51			print PSET "import re\n";
52			print PSET "import userConfig_cfg as pset\n";
53			print PSET "\n";
54			print PSET "fileName = pset.process.TFileService.fileName\n";
55			print PSET "fileName = fileName.pythonValue ()\n";
56			print PSET "fileName = fileName[1:(len (fileName) - 1)]\n";
57			print PSET "fileName = re.sub (r'^(.)\\.([^\\.])\$', r'\\1_' + str (osusub.jobNumber) + r'.\\2', fileName)\n";
58			print PSET "pset.process.TFileService.fileName = fileName\n";
59			print PSET "\n";
60			print PSET "pset.process.source.fileNames = cms.untracked.vstring (osusub.runList)\n";
61			print PSET "pset.process.maxEvents.input = cms.untracked.int32 (-1)\n";
62			print PSET "process = pset.process\n";
63
64			close (PSET);
65			}
66
67			sub
68			outputRunList
69			{
70			my $dataset = shift;
71			my $workingDir = shift;
72
73			my ($location, $nFiles, $status) = getLocation ($dataset);
74			if ($status ne "present")
75			{
76			print "This dataset is not marked as present on the Tier 3!\n";
77			exit;
78			}
79			if (!(-e $location))
80			{
81			print "The database does not know where this dataset is!\n";
82			exit;
83			}
84			opendir (LOCATION, $location);
85			my @files = readdir (LOCATION);
86			closedir (LOCATION);
87			if (@files - 2 != $nFiles)
88			{
89			print "Number of files does not match database entry!\n";
90			exit;
91			}
92			open (RUNLIST, ">$workingDir/runList.py");
93			print RUNLIST "runList = [\n";
94			for (my $i = 0; $i < @files; $i++)
95			{
96			next if $files[$i] eq ".";
97			next if $files[$i] eq "..";
98			print RUNLIST "'file:$location/$files[$i]'";
99			print RUNLIST "," if $i + 1 != @files;
100			print RUNLIST "\n";
101			}
102			print RUNLIST "]";
103			close (RUNLIST);
104			}
105
106			sub
107			getLocation
108			{
109			my $dataset = shift;
110
111			my $results;
112	ahart	1.2	my $queryDataset = $dataset;
113			$queryDataset =~ s/\*/%/g;
114			$queryDataset =~ s/(.*)/%$1%/g;
115	ahart	1.1	$db = Mysql->connect ("cmshead", "ntuple", "osuT3User");
116	ahart	1.2	my $query = "select dataset,user,creationTime,location,nFiles,status from ntuple where dataset like '$queryDataset' order by lastUpdateTime";
117	ahart	1.1	$db->selectdb ("ntuple");
118			$results = $db->query ($query);
119			if ($results->numrows () == 1)
120			{
121			my @row = $results->fetchrow ();
122	ahart	1.2	return ($row[3], $row[4], $row[5]);
123	ahart	1.1	}
124			if ($results->numrows () == 0)
125			{
126			print "Dataset does not exist on the Tier 3!\n";
127			exit;
128			}
129			my %map;
130			print "Found multiple datasets matching\n";
131			print "\"$dataset\":\n";
132			for (my $i = 1; $i <= $results->numrows (); $i++)
133			{
134			my @row = $results->fetchrow ();
135	ahart	1.2	$map{"$i"} = [$row[3], $row[4], $row[5]];
136			printf "(%2d) $row[0]\n", $i;
137			print " created by $row[1] on $row[2]\n";
138	ahart	1.1	}
139			print "\nWhich dataset would you like to use?: ";
140			my $response = <STDIN>;
141			$response =~ s/[ \t\n]//g;
142			if (!(exists $map{$response}))
143			{
144			print "Your selection \"$response\" was not a valid option! Quitting.\n";
145			exit;
146			}
147
148			return ($map{$response}[0], $map{$response}[1], $map{$response}[2]);
149			}
150
151			sub
152			outputCondor
153			{
154			my $workingDir = shift;
155			my $nJobs = shift;
156
157			my $cmsRun = `which cmsRun`;
158			open (SUB, ">$workingDir/condor.sub");
159
160			print SUB "Executable = $cmsRun\n";
161			print SUB "Universe = vanilla\n";
162			print SUB "Getenv = True\n";
163			print SUB "Arguments = config_cfg.py $nJobs \$(Process)\n";
164			print SUB "\n";
165			print SUB "Output = condor_\$(Process).out\n";
166			print SUB "Error = condor_\$(Process).err\n";
167			print SUB "Log = condor_\$(Process).log\n";
168			print SUB "\n";
169			print SUB "Queue $nJobs\n";
170
171			close (SUB);
172			}
173
174			sub
175			printHelp
176			{
177			my $exeName = $0;
178			$exeName =~ s/^.\/([^\/])$/$1/;
179
180	ahart	1.2	print "Usage: $exeName [OPTIONS] DATASET CONFIG DIRECTORY NJOBS\n";
181	ahart	1.1	print "Submits CMSSW jobs to the OSU Tier 3 compute nodes using Condor.\n";
182			print "\n";
183			printf "%-29s%s\n", " -h, --help", "print this help message";
184			print "\n";
185			print "The DATASET must exist in the Tier 3 ntuple database, and CONFIG must be a valid\n";
186			print "CMSSW python configuration which can be used with cmsRun. DIRECTORY is a working\n";
187			print "directory that is created and in which all output, both from the CMSSW jobs and\n";
188			print "from Condor, is placed. Finally, NJOBS is the number of Condor jobs that will\n";
189			print "be created.\n";
190
191			exit;
192			}