[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;
sub isEDM;

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 %nTotalEvents;
my $nTotalEvents = 0;
my %skimEventCounts;
my %weights;
my @weights;
my $nGoodJobs = 0;
my $nBadJobs = 0;
my $nIncompleteJobs = 0;
my $counting = 0;
my %exitCodes;
my %signals;
my %partial;
my %crossSections;
my %dirs;
my %channels;
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;
        $counting = 1;
        if ($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} = $fileContents;
          }
        elsif ($fileContents =~ m/signal/)
          {
            $fileContents =~ s/.*\(signal ([^)]*)\).*/$1/g;
            $nBadJobs++;
            print "WARNING: Skipping job $jobNumber. (signal $fileContents)\n";
            $signals{$dir}{$jobNumber} = $fileContents;
          }
        else
          {
            $nIncompleteJobs++;
            $partial{$dir}{$jobNumber} = 1;
          }
      }
    if ($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;
    my $jobNumber;
    my $fileIsEDM = 0;
    $fileIsEDM = isEDM ($file) if $file =~ m/^.*\.root$/;
    if ($fileIsEDM < 0)
      {
        unlink ($file);
        next;
      }
    if ($file =~ m/^.*_[^_]*\.root$/ && !$fileIsEDM)
      {
        $jobNumber = $file;
        $jobNumber =~ s/^.*_([^_]*)\.root$/$1/;
        $badJob = (defined $signals{$dir} && defined $signals{$dir}{$jobNumber}) || (defined $partial{$dir} && defined $partial{$dir}{$jobNumber});
      }
    next if $badJob;
    if ($file =~ m/^.*\.root$/ && !$fileIsEDM)
      {
        foreach my $arg (@ARGV)
          {
            if (substr ($file, 0, length ($arg)) eq $arg)
              {
                $dirs{$arg} = $dir;
                my $count = countEvents ($file, $cutFlow);
                if ($count->{$cutFlow} < 0)
                  {
                    $nGoodJobs--;
                    $nBadJobs++;
                    print "WARNING: Skipping job $jobNumber. (bad ROOT file)\n";
                    last;
                  }
                $nTotalEvents{$arg} += $count->{$cutFlow} if defined $nTotalEvents{$arg};
                $nTotalEvents{$arg} = $count->{$cutFlow} if !(defined $nTotalEvents{$arg});
                $nTotalEvents += $count->{$cutFlow};
                push (@{$rootFiles{$arg}}, $file);
                if ($opt{"weight"})
                  {
                    push (@weights, $opt{"weight"}) if !($opt{"weight"} =~ m/,/);
                    push (@weights, 1.0) if $opt{"weight"} =~ m/,/;
                  }
                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];

                foreach my $channelCutFlow (keys %{$count})
                  {
                    my $channel = substr ($channelCutFlow, 0, length ($channelCutFlow) - length ($cutFlow));
                    next if !length ($channel);
                    $skimEventCounts{$arg} = {} if !(defined $skimEventCounts{$arg});
                    $skimEventCounts{$arg}{$channel} += $count->{$channelCutFlow} if defined $skimEventCounts{$arg};
                    $skimEventCounts{$arg}{$channel} = $count->{$channelCutFlow} if !(defined $skimEventCounts{$arg});
                  }
              }
          }
      }
  }
if (!%rootFiles)
  {
    print "Found no ROOT files to merge!\n";
    exit;
  }
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 = $nTotalEvents{$arg};
    system ("cutFlowLimits $tmpName");
    if ($count)
      {
        $weights{$arg} /= $count if !$opt{"weight"} && $opt{"xsection"};
        $weights{$arg} /= $count if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dirs{$arg}};
      }
    else
      {
        $weights{$arg} *= $count if !$opt{"weight"} && $opt{"xsection"};
        $weights{$arg} *= $count if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dirs{$arg}};
      }
    push (@mergedFiles, $tmpName);
    push (@mergedWeights, $weights{$arg});
    foreach my $rootFile (@{$rootFiles{$arg}})
      {
        my $weight = $weights{$arg};
        system ("weightTrees $rootFile $weight");
      }
    open (EVENT_COUNT, ">$dirs{$arg}/numberOfEvents.txt");
    print EVENT_COUNT sprintf ("%.16g", $count) . "\n";
    close (EVENT_COUNT);
    foreach my $channel (keys %{$skimEventCounts{$arg}})
      {
        open (SKIM_COUNT, ">$dirs{$arg}/$channel/skimNumberOfEvents.txt");
        print SKIM_COUNT sprintf ("%.16g", $skimEventCounts{$arg}{$channel}) . "\n";
        close (SKIM_COUNT);
      }
  }
my $mergedFiles = join (" ", @mergedFiles);
my $mergedWeights = join (",", @mergedWeights);
$mergedWeights = $opt{"weight"} if $opt{"weight"} && $opt{"weight"} =~ m/,/;
system ("mergeTFileServiceHistograms -i $mergedFiles -o $opt{'prefix'}.root -w $mergedWeights");
foreach my $mergedFile (@mergedFiles)
  {
    unlink ("$mergedFile");
  }
print "=============================================\n";
my $output = sprintf "Cross-section of samples:\n";
my $printOutput = 0;
foreach my $arg (keys %weights)
{
    my $shortArg = $arg;
    $shortArg =~ s/^.*\/([^\/]*)$/$1/;
    $output = sprintf "%s  $shortArg: %.5g pb\n", $output, ($crossSections{$dirs{$arg}});
    $printOutput = 1 if $weights{$arg} != 1;
}
print $output if $printOutput;

my $output = sprintf "Effective luminosities of samples:\n";
my $printOutput = 0;
foreach my $arg (keys %weights)
  {
    my $shortArg = $arg;
    $shortArg =~ s/^.*\/([^\/]*)$/$1/;
    $output = sprintf "%s  $shortArg: %.5g/fb\n", $output, ($integratedLuminosity / (1000.0 * $weights{$arg})) if $weights{$arg} != 1 && $weights{$arg} != 0;
    $printOutput = 1 if $weights{$arg} != 1;
  }
print $output if $printOutput;

$output = sprintf "Weights for target luminosity of %g/fb:\n", ($integratedLuminosity / 1000.0);
$printOutput = 0;
foreach my $arg (keys %weights)
  {
    my $shortArg = $arg;
    $shortArg =~ s/^.*\/([^\/]*)$/$1/;
    $output = sprintf "%s  $shortArg: %.5g\n", $output, $weights{$arg} if $weights{$arg} != 1 && $weights{$arg} != 0;
    $printOutput = 1 if $weights{$arg} != 1;
  }
print $output if $printOutput;

my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
printf "$nGoodJobs jobs ran successfully over %.1f (%.1f weighted) events.\n", $nTotalEvents, $goodEvents->{$cutFlow} if $counting;
print "$nBadJobs jobs failed to run.\n" if $counting;
print "$nIncompleteJobs jobs have not finished.\n" if $counting;
print "=============================================\n";

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; WEIGHT may be a single number or a";
  printf "%-29s%s\n", "                     ", "comma-separated list, one for each input file";
  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[0] =~ m/Did not find a histogram named/ || $output[0] =~ m/appears to be empty/ || $output[0] =~ m/Failed to open/)
    {
      print $output[0];
      return {$cutFlow => -1};
    }
  my %weight;
  my %count;
  foreach my $line (@output)
    {
      my $cutFlowName = $line;
      my $weight = $line;
      my $count = $line;
      $cutFlowName =~ s/(.*): .* \/ .*\n*/$1/;
      $weight =~ s/.*: .* \/ (.*)\n*/$1/;
      $count =~ s/.*: (.*) \/ .*\n*/$1/;
      $count{$cutFlowName} = $count + 0;
      $weight{$cutFlowName} = $weight + 0;
    }
  foreach my $cutFlowName (keys %count)
    {
      $count{$cutFlowName} *= $count{$cutFlow} / $weight{$cutFlowName} if $weight{$cutFlowName};
      $count{$cutFlowName} *= $count{$cutFlow} * $weight{$cutFlowName} if !$weight{$cutFlowName};
    }

  return \%count;
}

sub
isEDM
{
  my $file = shift;

  my $output = `edmFileUtil $file 2>&1`;
  return 0 if $output =~ m/appears to be missing/ || $output =~ m/not a ROOT file/;
  $output =~ s/[\f\n\r]//g;
  $output =~ s/^.*, ([^,]*) events,.*$/$1/;
  $output += 0;
  $output-- if $output == 0;
  return $output;
}
Revision:	1.30
Committed:	Tue Jul 9 19:25:00 2013 UTC (11 years, 9 months ago) by ahart
Branch:	MAIN
CVS Tags:	V02-03-02, HEAD
Changes since 1.29:	+1 -1 lines
Log Message:	In the countEvents function, return a hash from the cutflow histogram name to negative one instead of returning negative one. The return value of this function should now always be a hash.
#	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	1.27	sub isEDM;
12	ahart	1.1
13			my %opt;
14			Getopt::Long::Configure ("bundling");
15	ahart	1.6	GetOptions (\%opt, "cutflow\|c=s", "luminosity\|l=s", "prefix\|p=s", "weight\|w=s", "xsection\|x=s", "help\|h");
16	ahart	1.1
17			printHelp () if $opt{"help"} \|\| !$opt{"prefix"};
18			my $files = processArgs (\@ARGV);
19	ahart	1.11	my %rootFiles;
20	ahart	1.29	my %nTotalEvents;
21			my $nTotalEvents = 0;
22	ahart	1.28	my %skimEventCounts;
23	ahart	1.11	my %weights;
24	ahart	1.13	my @weights;
25	ahart	1.5	my $nGoodJobs = 0;
26			my $nBadJobs = 0;
27	ahart	1.17	my $nIncompleteJobs = 0;
28	ahart	1.3	my $counting = 0;
29	ahart	1.4	my %exitCodes;
30	ahart	1.8	my %signals;
31	ahart	1.17	my %partial;
32	ahart	1.5	my %crossSections;
33	ahart	1.14	my %dirs;
34	ahart	1.29	my %channels;
35	ahart	1.5	my $integratedLuminosity = 10000;
36			$integratedLuminosity = $opt{"luminosity"} if $opt{"luminosity"};
37	ahart	1.6	my $cutFlow = "cutFlow";
38			$cutFlow = $opt{"cutflow"} if $opt{"cutflow"};
39	ahart	1.3	foreach my $file (@$files)
40			{
41			next if $file eq ".";
42			next if $file eq "..";
43	ahart	1.5	my $dir = $file;
44			$dir =~ s/^(.)\/[^\/]$/$1/;
45	ahart	1.4	if ($file =~ m/^.\/condor_[^_]\.log$/)
46	ahart	1.3	{
47			my $jobNumber = $file;
48			$jobNumber =~ s/^.\/condor_([^_])\.log$/$1/;
49			open (FILE, "<$file");
50			my @fileContents = <FILE>;
51			close (FILE);
52			my $fileContents = join ("", @fileContents);
53			$fileContents =~ s/\n/ /g;
54	ahart	1.19	$counting = 1;
55	ahart	1.8	if ($fileContents =~ m/return value/)
56	ahart	1.3	{
57	ahart	1.8	$fileContents =~ s/.\(return value ([^)])\).*/$1/g;
58	ahart	1.9	$nGoodJobs++;
59			print "WARNING: Nonzero exit code for job $jobNumber. (return value $fileContents)\n" if $fileContents != 0;
60	ahart	1.8	$exitCodes{$dir}{$jobNumber} = $fileContents;
61			}
62	ahart	1.17	elsif ($fileContents =~ m/signal/)
63	ahart	1.8	{
64			$fileContents =~ s/.\(signal ([^)])\).*/$1/g;
65	ahart	1.5	$nBadJobs++;
66	ahart	1.8	print "WARNING: Skipping job $jobNumber. (signal $fileContents)\n";
67			$signals{$dir}{$jobNumber} = $fileContents;
68	ahart	1.3	}
69	ahart	1.17	else
70			{
71			$nIncompleteJobs++;
72			$partial{$dir}{$jobNumber} = 1;
73			}
74	ahart	1.3	}
75	ahart	1.5	if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
76			{
77			open (CROSS_SECTION, "<$file");
78			my $crossSection = <CROSS_SECTION>;
79			close (CROSS_SECTION);
80			$crossSections{$dir} = $crossSection;
81			}
82	ahart	1.3	}
83	ahart	1.1	foreach my $file (@$files)
84			{
85			next if $file eq ".";
86			next if $file eq "..";
87	ahart	1.5	my $dir = $file;
88			$dir =~ s/^(.)\/[^\/]$/$1/;
89	ahart	1.3	my $badJob = 0;
90	ahart	1.20	my $jobNumber;
91	ahart	1.27	my $fileIsEDM = 0;
92			$fileIsEDM = isEDM ($file) if $file =~ m/^.*\.root$/;
93	ahart	1.29	if ($fileIsEDM < 0)
94			{
95			unlink ($file);
96			next;
97			}
98	ahart	1.27	if ($file =~ m/^._[^_]\.root$/ && !$fileIsEDM)
99	ahart	1.1	{
100	ahart	1.20	$jobNumber = $file;
101	ahart	1.3	$jobNumber =~ s/^._([^_])\.root$/$1/;
102	ahart	1.17	$badJob = (defined $signals{$dir} && defined $signals{$dir}{$jobNumber}) \|\| (defined $partial{$dir} && defined $partial{$dir}{$jobNumber});
103	ahart	1.5	}
104			next if $badJob;
105	ahart	1.27	if ($file =~ m/^.*\.root$/ && !$fileIsEDM)
106	ahart	1.5	{
107	ahart	1.11	foreach my $arg (@ARGV)
108			{
109	ahart	1.19	if (substr ($file, 0, length ($arg)) eq $arg)
110	ahart	1.11	{
111	ahart	1.19	$dirs{$arg} = $dir;
112	ahart	1.29	my $count = countEvents ($file, $cutFlow);
113			if ($count->{$cutFlow} < 0)
114	ahart	1.20	{
115			$nGoodJobs--;
116			$nBadJobs++;
117			print "WARNING: Skipping job $jobNumber. (bad ROOT file)\n";
118			last;
119			}
120	ahart	1.29	$nTotalEvents{$arg} += $count->{$cutFlow} if defined $nTotalEvents{$arg};
121			$nTotalEvents{$arg} = $count->{$cutFlow} if !(defined $nTotalEvents{$arg});
122			$nTotalEvents += $count->{$cutFlow};
123	ahart	1.11	push (@{$rootFiles{$arg}}, $file);
124	ahart	1.24	if ($opt{"weight"})
125			{
126			push (@weights, $opt{"weight"}) if !($opt{"weight"} =~ m/,/);
127			push (@weights, 1.0) if $opt{"weight"} =~ m/,/;
128			}
129	ahart	1.13	push (@weights, $opt{"xsection"} * $integratedLuminosity) if !$opt{"weight"} && $opt{"xsection"};
130			push (@weights, $crossSections{$dir} * $integratedLuminosity) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
131			push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
132			$weights{$arg} = $weights[-1];
133	ahart	1.29
134			foreach my $channelCutFlow (keys %{$count})
135			{
136			my $channel = substr ($channelCutFlow, 0, length ($channelCutFlow) - length ($cutFlow));
137			next if !length ($channel);
138			$skimEventCounts{$arg} = {} if !(defined $skimEventCounts{$arg});
139			$skimEventCounts{$arg}{$channel} += $count->{$channelCutFlow} if defined $skimEventCounts{$arg};
140			$skimEventCounts{$arg}{$channel} = $count->{$channelCutFlow} if !(defined $skimEventCounts{$arg});
141			}
142	ahart	1.11	}
143			}
144	ahart	1.5	}
145			}
146	ahart	1.11	if (!%rootFiles)
147	ahart	1.6	{
148			print "Found no ROOT files to merge!\n";
149			exit;
150			}
151	ahart	1.13	my @mergedFiles;
152			my @mergedWeights;
153	ahart	1.11	foreach my $arg (@ARGV)
154			{
155			my $rootFiles = join (" ", @{$rootFiles{$arg}});
156	ahart	1.28	my $tmpName = "." . $arg . "_" . "$opt{'prefix'}.root";
157	ahart	1.13	$tmpName =~ s/\//_/g;
158			system ("mergeTFileServiceHistograms -i $rootFiles -o $tmpName");
159	ahart	1.29	my $count = $nTotalEvents{$arg};
160	ahart	1.13	system ("cutFlowLimits $tmpName");
161	ahart	1.29	if ($count)
162			{
163			$weights{$arg} /= $count if !$opt{"weight"} && $opt{"xsection"};
164			$weights{$arg} /= $count if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dirs{$arg}};
165			}
166			else
167			{
168			$weights{$arg} *= $count if !$opt{"weight"} && $opt{"xsection"};
169			$weights{$arg} *= $count if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dirs{$arg}};
170			}
171	ahart	1.19	push (@mergedFiles, $tmpName);
172	ahart	1.14	push (@mergedWeights, $weights{$arg});
173	ahart	1.25	foreach my $rootFile (@{$rootFiles{$arg}})
174			{
175			my $weight = $weights{$arg};
176			system ("weightTrees $rootFile $weight");
177			}
178	ahart	1.29	open (EVENT_COUNT, ">$dirs{$arg}/numberOfEvents.txt");
179			print EVENT_COUNT sprintf ("%.16g", $count) . "\n";
180			close (EVENT_COUNT);
181			foreach my $channel (keys %{$skimEventCounts{$arg}})
182	ahart	1.26	{
183	ahart	1.29	open (SKIM_COUNT, ">$dirs{$arg}/$channel/skimNumberOfEvents.txt");
184			print SKIM_COUNT sprintf ("%.16g", $skimEventCounts{$arg}{$channel}) . "\n";
185			close (SKIM_COUNT);
186	ahart	1.26	}
187	ahart	1.11	}
188	ahart	1.13	my $mergedFiles = join (" ", @mergedFiles);
189			my $mergedWeights = join (",", @mergedWeights);
190	ahart	1.24	$mergedWeights = $opt{"weight"} if $opt{"weight"} && $opt{"weight"} =~ m/,/;
191	ahart	1.13	system ("mergeTFileServiceHistograms -i $mergedFiles -o $opt{'prefix'}.root -w $mergedWeights");
192			foreach my $mergedFile (@mergedFiles)
193	ahart	1.5	{
194	ahart	1.13	unlink ("$mergedFile");
195	ahart	1.1	}
196	ahart	1.19	print "=============================================\n";
197	lantonel	1.23	my $output = sprintf "Cross-section of samples:\n";
198	ahart	1.21	my $printOutput = 0;
199	ahart	1.10	foreach my $arg (keys %weights)
200	lantonel	1.23	{
201	ahart	1.12	my $shortArg = $arg;
202			$shortArg =~ s/^.\/([^\/])$/$1/;
203	lantonel	1.23	$output = sprintf "%s $shortArg: %.5g pb\n", $output, ($crossSections{$dirs{$arg}});
204	ahart	1.21	$printOutput = 1 if $weights{$arg} != 1;
205	lantonel	1.23	}
206			print $output if $printOutput;
207	jbrinson	1.22
208	lantonel	1.23	my $output = sprintf "Effective luminosities of samples:\n";
209			my $printOutput = 0;
210	jbrinson	1.22	foreach my $arg (keys %weights)
211	lantonel	1.23	{
212	jbrinson	1.22	my $shortArg = $arg;
213			$shortArg =~ s/^.\/([^\/])$/$1/;
214	ahart	1.29	$output = sprintf "%s $shortArg: %.5g/fb\n", $output, ($integratedLuminosity / (1000.0 * $weights{$arg})) if $weights{$arg} != 1 && $weights{$arg} != 0;
215	jbrinson	1.22	$printOutput = 1 if $weights{$arg} != 1;
216	lantonel	1.23	}
217			print $output if $printOutput;
218	jbrinson	1.22
219	ahart	1.19	$output = sprintf "Weights for target luminosity of %g/fb:\n", ($integratedLuminosity / 1000.0);
220			$printOutput = 0;
221			foreach my $arg (keys %weights)
222			{
223			my $shortArg = $arg;
224			$shortArg =~ s/^.\/([^\/])$/$1/;
225	ahart	1.29	$output = sprintf "%s $shortArg: %.5g\n", $output, $weights{$arg} if $weights{$arg} != 1 && $weights{$arg} != 0;
226	ahart	1.18	$printOutput = 1 if $weights{$arg} != 1;
227	ahart	1.10	}
228	ahart	1.18	print $output if $printOutput;
229	lantonel	1.23
230	ahart	1.6	my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
231	ahart	1.29	printf "$nGoodJobs jobs ran successfully over %.1f (%.1f weighted) events.\n", $nTotalEvents, $goodEvents->{$cutFlow} if $counting;
232	ahart	1.4	print "$nBadJobs jobs failed to run.\n" if $counting;
233	ahart	1.17	print "$nIncompleteJobs jobs have not finished.\n" if $counting;
234	ahart	1.19	print "=============================================\n";
235	ahart	1.1
236			sub
237			processArgs
238			{
239			my $argv = shift;
240
241			my @files;
242			foreach my $arg (@$argv)
243			{
244			$arg =~ s/\/*$//;
245			if (!(-e $arg))
246			{
247			print "$arg does not exist!\n";
248			exit;
249			}
250			next if ($arg =~ m/\/\.$/ \|\| $arg =~ m/\/\.\.$/);
251			if (-d $arg)
252			{
253			opendir (DIR, $arg);
254			my @dirContents = readdir (DIR);
255			closedir (DIR);
256			for (my $i = 0; $i < @dirContents; $i++)
257			{
258			$dirContents[$i] = "$arg/$dirContents[$i]";
259			}
260			my $newFiles = processArgs (\@dirContents);
261			push (@files, @$newFiles);
262			}
263			else
264			{
265			push (@files, $arg);
266			}
267			}
268
269			return \@files;
270			}
271
272			sub
273			printHelp
274			{
275			my $exeName = $0;
276			$exeName =~ s/^.\/([^\/])$/$1/;
277
278	ahart	1.6	print "Usage: $exeName [OPTION]... -p PREFIX DIRECTORIES_AND_FILES\n";
279	ahart	1.7	print "Merges ROOT files containing histograms. If there are Condor logs in the\n";
280			print "specified directories, checks for nonzero return values. If the directories\n";
281			print "where created by \"osusub\", uses the cross section from the database to weight\n";
282			print "all histograms.\n";
283	ahart	1.1	print "\n";
284			print "Mandatory arguments to long options are mandatory for short options too.\n";
285	ahart	1.6	printf "%-29s%s\n", " -c, --cutflow HISTOGRAM", "name of histogram to use for the cutflow (default:";
286			printf "%-29s%s\n", " ", "cutFlow)";
287	ahart	1.1	printf "%-29s%s\n", " -h, --help", "print this help message";
288	ahart	1.24	printf "%-29s%s\n", " -l, --luminosity", "integrated luminosity to which the histograms are";
289			printf "%-29s%s\n", " ", "weighted (default: 10000/pb)";
290	ahart	1.7	printf "%-29s%s\n", " -p, --prefix PREFIX", "output is named PREFIX.root";
291	ahart	1.5	printf "%-29s%s\n", " -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
292			printf "%-29s%s\n", " ", "automatic weighting using the cross section from";
293	ahart	1.24	printf "%-29s%s\n", " ", "the database; WEIGHT may be a single number or a";
294			printf "%-29s%s\n", " ", "comma-separated list, one for each input file";
295	ahart	1.6	printf "%-29s%s\n", " -x, --xsection XSECTION", "use XSECTION to weight the histograms instead of";
296			printf "%-29s%s\n", " ", "the value in the database";
297	ahart	1.1
298			exit;
299			}
300	ahart	1.3
301			sub
302			getRunList
303			{
304			my $runListFile = shift;
305
306			open (RUN_LIST, "<$runListFile");
307			my @runList0 = <RUN_LIST>;
308			close (RUN_LIST);
309			my @runList;
310			foreach my $file (@runList0)
311			{
312			next if !($file =~ m/^.file:.\.root.*/);
313			$file =~ s/.file:(.)\.root.*/$1.root/;
314			push (@runList, $file);
315			}
316
317			return \@runList;
318			}
319
320			sub
321			countEvents
322			{
323	ahart	1.4	my $file = shift;
324	ahart	1.6	my $cutFlow = shift;
325	ahart	1.4
326	ahart	1.29	my @output = `getEventsFromCutFlow $file $cutFlow`;
327			if ($output[0] =~ m/Did not find a histogram named/ \|\| $output[0] =~ m/appears to be empty/ \|\| $output[0] =~ m/Failed to open/)
328	ahart	1.5	{
329	ahart	1.29	print $output[0];
330	ahart	1.30	return {$cutFlow => -1};
331	ahart	1.5	}
332	ahart	1.29	my %weight;
333			my %count;
334			foreach my $line (@output)
335			{
336			my $cutFlowName = $line;
337			my $weight = $line;
338			my $count = $line;
339			$cutFlowName =~ s/(.): . \/ .\n/$1/;
340			$weight =~ s/.: . \/ (.)\n/$1/;
341			$count =~ s/.: (.) \/ .\n/$1/;
342			$count{$cutFlowName} = $count + 0;
343			$weight{$cutFlowName} = $weight + 0;
344			}
345			foreach my $cutFlowName (keys %count)
346			{
347			$count{$cutFlowName} *= $count{$cutFlow} / $weight{$cutFlowName} if $weight{$cutFlowName};
348			$count{$cutFlowName} = $count{$cutFlow} $weight{$cutFlowName} if !$weight{$cutFlowName};
349			}
350	ahart	1.4
351	ahart	1.29	return \%count;
352	ahart	1.3	}
353	ahart	1.27
354			sub
355			isEDM
356			{
357			my $file = shift;
358
359			my $output = `edmFileUtil $file 2>&1`;
360			return 0 if $output =~ m/appears to be missing/ \|\| $output =~ m/not a ROOT file/;
361	ahart	1.28	$output =~ s/[\f\n\r]//g;
362			$output =~ s/^., ([^,]) events,.*$/$1/;
363			$output += 0;
364			$output-- if $output == 0;
365			return $output;
366	ahart	1.27	}