[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 $nGoodJobs = 0;
my $nBadJobs = 0;
my $counting = 0;
my %exitCodes;
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;
        $fileContents =~ s/.*\(return value ([^)]*)\).*/$1/g;
        $nGoodJobs++;
        if ($fileContents != 0)
          {
            $nBadJobs++;
            print "Skipping job $jobNumber. (return value $fileContents)\n";
          }
        $exitCodes{$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 $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
      }
    next if $badJob;
    if ($file =~ m/^.*\.root$/)
      {
        push (@rootFiles, $file);
        push (@weights, 1.0);
      }
  }
if (!@rootFiles)
  {
    print "Found no ROOT files to merge!\n";
    exit;
  }
my $rootFiles = join (" ", @rootFiles);
my $weights = join (",", @weights);
system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root -w $weights");
my $nTotalEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
unlink ("$opt{'prefix'}.root");
@weights = ();
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 $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
      }
    next if $badJob;
    if ($file =~ m/^.*\.root$/)
      {
        push (@weights, $opt{"weight"}) if $opt{"weight"};
        push (@weights, ($opt{"xsection"} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && $opt{"xsection"};
        push (@weights, ($crossSections{$dir} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
        push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
      }
  }
my $weights = join (",", @weights);
system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root -w $weights");
system ("cutFlowTable $opt{'prefix'}.root $cutFlow >& $opt{'prefix'}.tex");
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 and produces a cutflow table in the\n";
  print "form of a LaTeX document. If there are Condor logs in the specified\n";
  print "directories, checks for nonzero return values. If the directories where created\n";
  print "by \"osusub\", uses the cross section from the database to weight 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 files are named PREFIX.root and PREFIX.tex";
  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:	1.6
Committed:	Mon Sep 10 09:48:56 2012 UTC (12 years, 7 months ago) by ahart
Branch:	MAIN
Changes since 1.5:	+31 -22 lines
Log Message:	Add a couple options for setting the cross section and the cutflow histogram name and use a more efficient method for getting the total number of events.
#	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			my @rootFiles;
19	ahart	1.3	my @weights;
20	ahart	1.5	my $nGoodJobs = 0;
21			my $nBadJobs = 0;
22	ahart	1.3	my $counting = 0;
23	ahart	1.4	my %exitCodes;
24	ahart	1.5	my %crossSections;
25			my $integratedLuminosity = 10000;
26			$integratedLuminosity = $opt{"luminosity"} if $opt{"luminosity"};
27	ahart	1.6	my $cutFlow = "cutFlow";
28			$cutFlow = $opt{"cutflow"} if $opt{"cutflow"};
29	ahart	1.3	foreach my $file (@$files)
30			{
31			next if $file eq ".";
32			next if $file eq "..";
33	ahart	1.5	my $dir = $file;
34			$dir =~ s/^(.)\/[^\/]$/$1/;
35	ahart	1.4	if ($file =~ m/^.\/condor_[^_]\.log$/)
36	ahart	1.3	{
37			my $jobNumber = $file;
38			$jobNumber =~ s/^.\/condor_([^_])\.log$/$1/;
39			open (FILE, "<$file");
40			my @fileContents = <FILE>;
41			close (FILE);
42			my $fileContents = join ("", @fileContents);
43			$fileContents =~ s/\n/ /g;
44			$fileContents =~ s/.\(return value ([^)])\).*/$1/g;
45	ahart	1.5	$nGoodJobs++;
46	ahart	1.3	if ($fileContents != 0)
47			{
48	ahart	1.5	$nBadJobs++;
49	ahart	1.3	print "Skipping job $jobNumber. (return value $fileContents)\n";
50			}
51	ahart	1.5	$exitCodes{$dir}{$jobNumber} = $fileContents;
52	ahart	1.3	$counting = 1;
53			}
54	ahart	1.5	if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
55			{
56			open (CROSS_SECTION, "<$file");
57			my $crossSection = <CROSS_SECTION>;
58			close (CROSS_SECTION);
59			$crossSections{$dir} = $crossSection;
60			}
61	ahart	1.3	}
62	ahart	1.1	foreach my $file (@$files)
63			{
64			next if $file eq ".";
65			next if $file eq "..";
66	ahart	1.5	my $dir = $file;
67			$dir =~ s/^(.)\/[^\/]$/$1/;
68	ahart	1.3	my $badJob = 0;
69			if ($file =~ m/^._[^_]\.root$/)
70	ahart	1.1	{
71	ahart	1.3	my $jobNumber = $file;
72			$jobNumber =~ s/^._([^_])\.root$/$1/;
73	ahart	1.5	$badJob = defined $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
74			}
75			next if $badJob;
76			if ($file =~ m/^.*\.root$/)
77			{
78	ahart	1.6	push (@rootFiles, $file);
79			push (@weights, 1.0);
80	ahart	1.5	}
81			}
82	ahart	1.6	if (!@rootFiles)
83			{
84			print "Found no ROOT files to merge!\n";
85			exit;
86			}
87			my $rootFiles = join (" ", @rootFiles);
88			my $weights = join (",", @weights);
89			system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root -w $weights");
90			my $nTotalEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
91			unlink ("$opt{'prefix'}.root");
92			@weights = ();
93	ahart	1.5	foreach my $file (@$files)
94			{
95			next if $file eq ".";
96			next if $file eq "..";
97			my $dir = $file;
98			$dir =~ s/^(.)\/[^\/]$/$1/;
99			my $badJob = 0;
100			if ($file =~ m/^._[^_]\.root$/)
101			{
102			my $jobNumber = $file;
103			$jobNumber =~ s/^._([^_])\.root$/$1/;
104			$badJob = defined $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
105	ahart	1.1	}
106	ahart	1.3	next if $badJob;
107			if ($file =~ m/^.*\.root$/)
108	ahart	1.1	{
109	ahart	1.3	push (@weights, $opt{"weight"}) if $opt{"weight"};
110	ahart	1.6	push (@weights, ($opt{"xsection"} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && $opt{"xsection"};
111			push (@weights, ($crossSections{$dir} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
112			push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
113	ahart	1.1	}
114			}
115	ahart	1.3	my $weights = join (",", @weights);
116			system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root -w $weights");
117	ahart	1.6	system ("cutFlowTable $opt{'prefix'}.root $cutFlow >& $opt{'prefix'}.tex");
118			my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
119	ahart	1.5	print "$nGoodJobs jobs ran successfully over $nTotalEvents ($goodEvents weighted) events.\n" if $counting;
120	ahart	1.4	print "$nBadJobs jobs failed to run.\n" if $counting;
121	ahart	1.1
122			sub
123			processArgs
124			{
125			my $argv = shift;
126
127			my @files;
128			foreach my $arg (@$argv)
129			{
130			$arg =~ s/\/*$//;
131			if (!(-e $arg))
132			{
133			print "$arg does not exist!\n";
134			exit;
135			}
136			next if ($arg =~ m/\/\.$/ \|\| $arg =~ m/\/\.\.$/);
137			if (-d $arg)
138			{
139			opendir (DIR, $arg);
140			my @dirContents = readdir (DIR);
141			closedir (DIR);
142			for (my $i = 0; $i < @dirContents; $i++)
143			{
144			$dirContents[$i] = "$arg/$dirContents[$i]";
145			}
146			my $newFiles = processArgs (\@dirContents);
147			push (@files, @$newFiles);
148			}
149			else
150			{
151			push (@files, $arg);
152			}
153			}
154
155			return \@files;
156			}
157
158			sub
159			printHelp
160			{
161			my $exeName = $0;
162			$exeName =~ s/^.\/([^\/])$/$1/;
163
164	ahart	1.6	print "Usage: $exeName [OPTION]... -p PREFIX DIRECTORIES_AND_FILES\n";
165	ahart	1.1	print "Merges ROOT files containing histograms and produces a cutflow table in the\n";
166	ahart	1.6	print "form of a LaTeX document. If there are Condor logs in the specified\n";
167			print "directories, checks for nonzero return values. If the directories where created\n";
168			print "by \"osusub\", uses the cross section from the database to weight all histograms.\n";
169	ahart	1.1	print "\n";
170			print "Mandatory arguments to long options are mandatory for short options too.\n";
171	ahart	1.6	printf "%-29s%s\n", " -c, --cutflow HISTOGRAM", "name of histogram to use for the cutflow (default:";
172			printf "%-29s%s\n", " ", "cutFlow)";
173	ahart	1.1	printf "%-29s%s\n", " -h, --help", "print this help message";
174	ahart	1.5	printf "%-29s%s\n", " -l, --luminosity", "integrated luminosity to which the histograms are";
175			printf "%-29s%s\n", " ", "weighted (default: 10000/pb)";
176	ahart	1.1	printf "%-29s%s\n", " -p, --prefix PREFIX", "output files are named PREFIX.root and PREFIX.tex";
177	ahart	1.5	printf "%-29s%s\n", " -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
178			printf "%-29s%s\n", " ", "automatic weighting using the cross section from";
179			printf "%-29s%s\n", " ", "the database";
180	ahart	1.6	printf "%-29s%s\n", " -x, --xsection XSECTION", "use XSECTION to weight the histograms instead of";
181			printf "%-29s%s\n", " ", "the value in the database";
182	ahart	1.1
183			exit;
184			}
185	ahart	1.3
186			sub
187			getRunList
188			{
189			my $runListFile = shift;
190
191			open (RUN_LIST, "<$runListFile");
192			my @runList0 = <RUN_LIST>;
193			close (RUN_LIST);
194			my @runList;
195			foreach my $file (@runList0)
196			{
197			next if !($file =~ m/^.file:.\.root.*/);
198			$file =~ s/.file:(.)\.root.*/$1.root/;
199			push (@runList, $file);
200			}
201
202			return \@runList;
203			}
204
205			sub
206			countEvents
207			{
208	ahart	1.4	my $file = shift;
209	ahart	1.6	my $cutFlow = shift;
210	ahart	1.4
211	ahart	1.6	my $output = `getEventsFromCutFlow $file $cutFlow`;
212	ahart	1.5	if ($output =~ m/Did not find a histogram named/)
213			{
214			print $output;
215			return -1;
216			}
217	ahart	1.4	$output =~ s/^.: (.)$/$1/;
218			$output =~ s/\n//g;
219
220			return $output;
221	ahart	1.3	}