[ViewVC] Annotation of: cvsroot/UserCode/OSUT3Analysis/AnaTools/scripts/mergeHists

#!/usr/bin/env perl

use strict;
use Getopt::Long;
use POSIX;

sub processArgs;
sub printHelp;
sub getRunList;
sub countEvents;

my %opt;
Getopt::Long::Configure ("bundling");
GetOptions (\%opt, "cutflow|c=s", "luminosity|l=s", "prefix|p=s", "weight|w=s", "xsection|x=s", "help|h");

printHelp () if $opt{"help"} || !$opt{"prefix"};
my $files = processArgs (\@ARGV);
my %rootFiles;
my %weights;
my @weights;
my $nGoodJobs = 0;
my $nBadJobs = 0;
my $counting = 0;
my %exitCodes;
my %signals;
my %crossSections;
my $integratedLuminosity = 10000;
$integratedLuminosity = $opt{"luminosity"} if $opt{"luminosity"};
my $cutFlow = "cutFlow";
$cutFlow = $opt{"cutflow"} if $opt{"cutflow"};
foreach my $file (@$files)
  {
    next if $file eq ".";
    next if $file eq "..";
    my $dir = $file;
    $dir =~ s/^(.*)\/[^\/]*$/$1/;
    if ($file =~ m/^.*\/condor_[^_]*\.log$/)
      {
        my $jobNumber = $file;
        $jobNumber =~ s/^.*\/condor_([^_]*)\.log$/$1/;
        open (FILE, "<$file");
        my @fileContents = <FILE>;
        close (FILE);
        my $fileContents = join ("", @fileContents);
        $fileContents =~ s/\n/ /g;
        if ($fileContents =~ m/return value/)
          {
            $fileContents =~ s/.*\(return value ([^)]*)\).*/$1/g;
            $nGoodJobs++;
            print "WARNING: Nonzero exit code for job $jobNumber. (return value $fileContents)\n" if $fileContents != 0;
            $exitCodes{$dir}{$jobNumber} = $fileContents;
            $counting = 1;
          }
        if ($fileContents =~ m/signal/)
          {
            $fileContents =~ s/.*\(signal ([^)]*)\).*/$1/g;
            $nBadJobs++;
            print "WARNING: Skipping job $jobNumber. (signal $fileContents)\n";
            $signals{$dir}{$jobNumber} = $fileContents;
            $counting = 1;
          }
      }
    if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
      {
        open (CROSS_SECTION, "<$file");
        my $crossSection = <CROSS_SECTION>;
        close (CROSS_SECTION);
        $crossSections{$dir} = $crossSection;
      }
  }
foreach my $file (@$files)
  {
    next if $file eq ".";
    next if $file eq "..";
    my $dir = $file;
    $dir =~ s/^(.*)\/[^\/]*$/$1/;
    my $badJob = 0;
    if ($file =~ m/^.*_[^_]*\.root$/)
      {
        my $jobNumber = $file;
        $jobNumber =~ s/^.*_([^_]*)\.root$/$1/;
        $badJob = defined $signals{$dir} && defined $signals{$dir}{$jobNumber};
      }
    next if $badJob;
    if ($file =~ m/^.*\.root$/)
      {
        foreach my $arg (@ARGV)
          {
            if (substr ($file, 0, length ($arg)) eq $arg)
              {
                push (@{$rootFiles{$arg}}, $file);
                push (@weights, $opt{"weight"}) if $opt{"weight"};
                push (@weights, $opt{"xsection"} * $integratedLuminosity) if !$opt{"weight"} && $opt{"xsection"};
                push (@weights, $crossSections{$dir} * $integratedLuminosity) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
                push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
                $weights{$arg} = $weights[-1];
              }
          }
      }
  }
if (!%rootFiles)
  {
    print "Found no ROOT files to merge!\n";
    exit;
  }
my %nTotalEvents;
my $nTotalEvents = 0;
my @mergedFiles;
my @mergedWeights;
foreach my $arg (@ARGV)
  {
    my $rootFiles = join (" ", @{$rootFiles{$arg}});
    my $tmpName = $arg . "_" . "$opt{'prefix'}.root";
    $tmpName =~ s/\//_/g;
    system ("mergeTFileServiceHistograms -i $rootFiles -o $tmpName");
    my $count = countEvents ($tmpName, $cutFlow);
    system ("cutFlowLimits $tmpName");
    $nTotalEvents{$arg} = $count;
    $nTotalEvents += $count;
    push (@mergedFiles, "$tmpName");
    push (@mergedWeights, $weights{$arg} / $count);
  }
my $mergedFiles = join (" ", @mergedFiles);
my $mergedWeights = join (",", @mergedWeights);
system ("mergeTFileServiceHistograms -i $mergedFiles -o $opt{'prefix'}.root -w $mergedWeights");
foreach my $mergedFile (@mergedFiles)
  {
    unlink ("$mergedFile");
  }
print "Weights:\n";
foreach my $arg (keys %weights)
  {
    my $shortArg = $arg;
    $shortArg =~ s/^.*\/([^\/]*)$/$1/;
    printf "  $shortArg: %.5g (%.5g pb)\n", $weights{$arg}, $weights{$arg} / $integratedLuminosity;
  }
my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
print "$nGoodJobs jobs ran successfully over $nTotalEvents ($goodEvents weighted) events.\n" if $counting;
print "$nBadJobs jobs failed to run.\n" if $counting;

sub
processArgs
{
  my $argv = shift;

  my @files;
  foreach my $arg (@$argv)
    {
      $arg =~ s/\/*$//;
      if (!(-e $arg))
        {
          print "$arg does not exist!\n";
          exit;
        }
      next if ($arg =~ m/\/\.$/ || $arg =~ m/\/\.\.$/);
      if (-d $arg)
        {
          opendir (DIR, $arg);
          my @dirContents = readdir (DIR);
          closedir (DIR);
          for (my $i = 0; $i < @dirContents; $i++)
            {
              $dirContents[$i] = "$arg/$dirContents[$i]";
            }
          my $newFiles = processArgs (\@dirContents);
          push (@files, @$newFiles);
        }
      else
        {
          push (@files, $arg);
        }
    }

  return \@files;
}

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

  print "Usage: $exeName [OPTION]... -p PREFIX DIRECTORIES_AND_FILES\n";
  print "Merges ROOT files containing histograms. If there are Condor logs in the\n";
  print "specified directories, checks for nonzero return values. If the directories\n";
  print "where created by \"osusub\", uses the cross section from the database to weight\n";
  print "all histograms.\n";
  print "\n";
  print "Mandatory arguments to long options are mandatory for short options too.\n";
  printf "%-29s%s\n", "  -c, --cutflow HISTOGRAM", "name of histogram to use for the cutflow (default:";
  printf "%-29s%s\n", "                  ", "cutFlow)";
  printf "%-29s%s\n", "  -h, --help", "print this help message";
  printf "%-29s%s\n", "  -l, --luminosity", "integrated luminosity to which the histograms are";
  printf "%-29s%s\n", "                  ", "weighted (default: 10000/pb)";
  printf "%-29s%s\n", "  -p, --prefix PREFIX", "output is  named PREFIX.root";
  printf "%-29s%s\n", "  -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
  printf "%-29s%s\n", "                     ", "automatic weighting using the cross section from";
  printf "%-29s%s\n", "                     ", "the database";
  printf "%-29s%s\n", "  -x, --xsection XSECTION", "use XSECTION to weight the histograms instead of";
  printf "%-29s%s\n", "                     ", "the value in the database";

  exit;
}

sub
getRunList
{
  my $runListFile = shift;

  open (RUN_LIST, "<$runListFile");
  my @runList0 = <RUN_LIST>;
  close (RUN_LIST);
  my @runList;
  foreach my $file (@runList0)
    {
      next if !($file =~ m/^.*file:.*\.root.*/);
      $file =~ s/.*file:(.*)\.root.*/$1.root/;
      push (@runList, $file);
    }

  return \@runList;
}

sub
countEvents
{
  my $file = shift;
  my $cutFlow = shift;

  my $output = `getEventsFromCutFlow $file $cutFlow`;
  if ($output =~ m/Did not find a histogram named/)
    {
      print $output;
      return -1;
    }
  $output =~ s/^.*: (.*)$/$1/;
  $output =~ s/\n//g;

  return $output;
}
Revision:	1.13
Committed:	Thu Jan 3 21:43:31 2013 UTC (12 years, 4 months ago) by ahart
Branch:	MAIN
Changes since 1.12:	+20 -39 lines
Log Message:	Do not discard the unweighted merged files, but instead calculate upper limits for their cutflow histograms and then weight them.
#	User	Rev	Content
1	ahart	1.1	#!/usr/bin/env perl
2
3			use strict;
4			use Getopt::Long;
5	ahart	1.3	use POSIX;
6
7			sub processArgs;
8			sub printHelp;
9			sub getRunList;
10			sub countEvents;
11	ahart	1.1
12			my %opt;
13			Getopt::Long::Configure ("bundling");
14	ahart	1.6	GetOptions (\%opt, "cutflow\|c=s", "luminosity\|l=s", "prefix\|p=s", "weight\|w=s", "xsection\|x=s", "help\|h");
15	ahart	1.1
16			printHelp () if $opt{"help"} \|\| !$opt{"prefix"};
17			my $files = processArgs (\@ARGV);
18	ahart	1.11	my %rootFiles;
19			my %weights;
20	ahart	1.13	my @weights;
21	ahart	1.5	my $nGoodJobs = 0;
22			my $nBadJobs = 0;
23	ahart	1.3	my $counting = 0;
24	ahart	1.4	my %exitCodes;
25	ahart	1.8	my %signals;
26	ahart	1.5	my %crossSections;
27			my $integratedLuminosity = 10000;
28			$integratedLuminosity = $opt{"luminosity"} if $opt{"luminosity"};
29	ahart	1.6	my $cutFlow = "cutFlow";
30			$cutFlow = $opt{"cutflow"} if $opt{"cutflow"};
31	ahart	1.3	foreach my $file (@$files)
32			{
33			next if $file eq ".";
34			next if $file eq "..";
35	ahart	1.5	my $dir = $file;
36			$dir =~ s/^(.)\/[^\/]$/$1/;
37	ahart	1.4	if ($file =~ m/^.\/condor_[^_]\.log$/)
38	ahart	1.3	{
39			my $jobNumber = $file;
40			$jobNumber =~ s/^.\/condor_([^_])\.log$/$1/;
41			open (FILE, "<$file");
42			my @fileContents = <FILE>;
43			close (FILE);
44			my $fileContents = join ("", @fileContents);
45			$fileContents =~ s/\n/ /g;
46	ahart	1.8	if ($fileContents =~ m/return value/)
47	ahart	1.3	{
48	ahart	1.8	$fileContents =~ s/.\(return value ([^)])\).*/$1/g;
49	ahart	1.9	$nGoodJobs++;
50			print "WARNING: Nonzero exit code for job $jobNumber. (return value $fileContents)\n" if $fileContents != 0;
51	ahart	1.8	$exitCodes{$dir}{$jobNumber} = $fileContents;
52			$counting = 1;
53			}
54			if ($fileContents =~ m/signal/)
55			{
56			$fileContents =~ s/.\(signal ([^)])\).*/$1/g;
57	ahart	1.5	$nBadJobs++;
58	ahart	1.8	print "WARNING: Skipping job $jobNumber. (signal $fileContents)\n";
59			$signals{$dir}{$jobNumber} = $fileContents;
60			$counting = 1;
61	ahart	1.3	}
62			}
63	ahart	1.5	if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
64			{
65			open (CROSS_SECTION, "<$file");
66			my $crossSection = <CROSS_SECTION>;
67			close (CROSS_SECTION);
68			$crossSections{$dir} = $crossSection;
69			}
70	ahart	1.3	}
71	ahart	1.1	foreach my $file (@$files)
72			{
73			next if $file eq ".";
74			next if $file eq "..";
75	ahart	1.5	my $dir = $file;
76			$dir =~ s/^(.)\/[^\/]$/$1/;
77	ahart	1.3	my $badJob = 0;
78			if ($file =~ m/^._[^_]\.root$/)
79	ahart	1.1	{
80	ahart	1.3	my $jobNumber = $file;
81			$jobNumber =~ s/^._([^_])\.root$/$1/;
82	ahart	1.8	$badJob = defined $signals{$dir} && defined $signals{$dir}{$jobNumber};
83	ahart	1.5	}
84			next if $badJob;
85			if ($file =~ m/^.*\.root$/)
86			{
87	ahart	1.11	foreach my $arg (@ARGV)
88			{
89			if (substr ($file, 0, length ($arg)) eq $arg)
90			{
91			push (@{$rootFiles{$arg}}, $file);
92	ahart	1.13	push (@weights, $opt{"weight"}) if $opt{"weight"};
93			push (@weights, $opt{"xsection"} * $integratedLuminosity) if !$opt{"weight"} && $opt{"xsection"};
94			push (@weights, $crossSections{$dir} * $integratedLuminosity) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
95			push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
96			$weights{$arg} = $weights[-1];
97	ahart	1.11	}
98			}
99	ahart	1.5	}
100			}
101	ahart	1.11	if (!%rootFiles)
102	ahart	1.6	{
103			print "Found no ROOT files to merge!\n";
104			exit;
105			}
106	ahart	1.11	my %nTotalEvents;
107			my $nTotalEvents = 0;
108	ahart	1.13	my @mergedFiles;
109			my @mergedWeights;
110	ahart	1.11	foreach my $arg (@ARGV)
111			{
112			my $rootFiles = join (" ", @{$rootFiles{$arg}});
113	ahart	1.13	my $tmpName = $arg . "_" . "$opt{'prefix'}.root";
114			$tmpName =~ s/\//_/g;
115			system ("mergeTFileServiceHistograms -i $rootFiles -o $tmpName");
116			my $count = countEvents ($tmpName, $cutFlow);
117			system ("cutFlowLimits $tmpName");
118	ahart	1.11	$nTotalEvents{$arg} = $count;
119			$nTotalEvents += $count;
120	ahart	1.13	push (@mergedFiles, "$tmpName");
121			push (@mergedWeights, $weights{$arg} / $count);
122	ahart	1.11	}
123	ahart	1.13	my $mergedFiles = join (" ", @mergedFiles);
124			my $mergedWeights = join (",", @mergedWeights);
125			system ("mergeTFileServiceHistograms -i $mergedFiles -o $opt{'prefix'}.root -w $mergedWeights");
126			foreach my $mergedFile (@mergedFiles)
127	ahart	1.5	{
128	ahart	1.13	unlink ("$mergedFile");
129	ahart	1.1	}
130	ahart	1.10	print "Weights:\n";
131			foreach my $arg (keys %weights)
132			{
133	ahart	1.12	my $shortArg = $arg;
134			$shortArg =~ s/^.\/([^\/])$/$1/;
135			printf " $shortArg: %.5g (%.5g pb)\n", $weights{$arg}, $weights{$arg} / $integratedLuminosity;
136	ahart	1.10	}
137	ahart	1.6	my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
138	ahart	1.5	print "$nGoodJobs jobs ran successfully over $nTotalEvents ($goodEvents weighted) events.\n" if $counting;
139	ahart	1.4	print "$nBadJobs jobs failed to run.\n" if $counting;
140	ahart	1.1
141			sub
142			processArgs
143			{
144			my $argv = shift;
145
146			my @files;
147			foreach my $arg (@$argv)
148			{
149			$arg =~ s/\/*$//;
150			if (!(-e $arg))
151			{
152			print "$arg does not exist!\n";
153			exit;
154			}
155			next if ($arg =~ m/\/\.$/ \|\| $arg =~ m/\/\.\.$/);
156			if (-d $arg)
157			{
158			opendir (DIR, $arg);
159			my @dirContents = readdir (DIR);
160			closedir (DIR);
161			for (my $i = 0; $i < @dirContents; $i++)
162			{
163			$dirContents[$i] = "$arg/$dirContents[$i]";
164			}
165			my $newFiles = processArgs (\@dirContents);
166			push (@files, @$newFiles);
167			}
168			else
169			{
170			push (@files, $arg);
171			}
172			}
173
174			return \@files;
175			}
176
177			sub
178			printHelp
179			{
180			my $exeName = $0;
181			$exeName =~ s/^.\/([^\/])$/$1/;
182
183	ahart	1.6	print "Usage: $exeName [OPTION]... -p PREFIX DIRECTORIES_AND_FILES\n";
184	ahart	1.7	print "Merges ROOT files containing histograms. If there are Condor logs in the\n";
185			print "specified directories, checks for nonzero return values. If the directories\n";
186			print "where created by \"osusub\", uses the cross section from the database to weight\n";
187			print "all histograms.\n";
188	ahart	1.1	print "\n";
189			print "Mandatory arguments to long options are mandatory for short options too.\n";
190	ahart	1.6	printf "%-29s%s\n", " -c, --cutflow HISTOGRAM", "name of histogram to use for the cutflow (default:";
191			printf "%-29s%s\n", " ", "cutFlow)";
192	ahart	1.1	printf "%-29s%s\n", " -h, --help", "print this help message";
193	ahart	1.5	printf "%-29s%s\n", " -l, --luminosity", "integrated luminosity to which the histograms are";
194			printf "%-29s%s\n", " ", "weighted (default: 10000/pb)";
195	ahart	1.7	printf "%-29s%s\n", " -p, --prefix PREFIX", "output is named PREFIX.root";
196	ahart	1.5	printf "%-29s%s\n", " -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
197			printf "%-29s%s\n", " ", "automatic weighting using the cross section from";
198			printf "%-29s%s\n", " ", "the database";
199	ahart	1.6	printf "%-29s%s\n", " -x, --xsection XSECTION", "use XSECTION to weight the histograms instead of";
200			printf "%-29s%s\n", " ", "the value in the database";
201	ahart	1.1
202			exit;
203			}
204	ahart	1.3
205			sub
206			getRunList
207			{
208			my $runListFile = shift;
209
210			open (RUN_LIST, "<$runListFile");
211			my @runList0 = <RUN_LIST>;
212			close (RUN_LIST);
213			my @runList;
214			foreach my $file (@runList0)
215			{
216			next if !($file =~ m/^.file:.\.root.*/);
217			$file =~ s/.file:(.)\.root.*/$1.root/;
218			push (@runList, $file);
219			}
220
221			return \@runList;
222			}
223
224			sub
225			countEvents
226			{
227	ahart	1.4	my $file = shift;
228	ahart	1.6	my $cutFlow = shift;
229	ahart	1.4
230	ahart	1.6	my $output = `getEventsFromCutFlow $file $cutFlow`;
231	ahart	1.5	if ($output =~ m/Did not find a histogram named/)
232			{
233			print $output;
234			return -1;
235			}
236	ahart	1.4	$output =~ s/^.: (.)$/$1/;
237			$output =~ s/\n//g;
238
239			return $output;
240	ahart	1.3	}