[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, "luminosity|l=s", "prefix|p=s", "weight|w=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"};
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;
      }
  }
my $nTotalEvents = 0;
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$/)
      {
        my $nEvents = countEvents ($file);
        $nTotalEvents = -1 if $nTotalEvents < 0 || $nEvents < 0;
        $nTotalEvents += $nEvents if !($nTotalEvents < 0 || $nEvents < 0);
      }
  }
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, $opt{"weight"}) if $opt{"weight"};
        push (@weights, ($crossSections{$dir} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && defined $crossSections{$dir};
        push (@weights, 1.0) if !$opt{"weight"} && !(defined $crossSections{$dir});
      }
  }
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");
system ("cutFlowTable $opt{'prefix'}.root cutFlow >& $opt{'prefix'}.tex");
my $goodEvents = countEvents ("$opt{'prefix'}.root");
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 -p PREFIX [OPTION]... DIRECTORIES_AND_FILES\n";
  print "Merges ROOT files containing histograms and produces a cutflow table in the\n";
  print "form of a LaTeX document from the histogram named \"cutFlow\". If there are\n";
  print "Condor logs in the specified directories, checks for nonzero return values. If\n";
  print "the directories where created by \"osusub\", uses the cross section from the\n";
  print "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", "  -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";

  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 $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:	1.5
Committed:	Sun Sep 9 20:05:12 2012 UTC (12 years, 7 months ago) by ahart
Branch:	MAIN
Changes since 1.4:	+59 -15 lines
Log Message:	Automatically weight the histograms using the cross section from the database.
#	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	1.5	GetOptions (\%opt, "luminosity\|l=s", "prefix\|p=s", "weight\|w=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	1.3	foreach my $file (@$files)
28			{
29			next if $file eq ".";
30			next if $file eq "..";
31	ahart	1.5	my $dir = $file;
32			$dir =~ s/^(.)\/[^\/]$/$1/;
33	ahart	1.4	if ($file =~ m/^.\/condor_[^_]\.log$/)
34	ahart	1.3	{
35			my $jobNumber = $file;
36			$jobNumber =~ s/^.\/condor_([^_])\.log$/$1/;
37			open (FILE, "<$file");
38			my @fileContents = <FILE>;
39			close (FILE);
40			my $fileContents = join ("", @fileContents);
41			$fileContents =~ s/\n/ /g;
42			$fileContents =~ s/.\(return value ([^)])\).*/$1/g;
43	ahart	1.5	$nGoodJobs++;
44	ahart	1.3	if ($fileContents != 0)
45			{
46	ahart	1.5	$nBadJobs++;
47	ahart	1.3	print "Skipping job $jobNumber. (return value $fileContents)\n";
48			}
49	ahart	1.5	$exitCodes{$dir}{$jobNumber} = $fileContents;
50	ahart	1.3	$counting = 1;
51			}
52	ahart	1.5	if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
53			{
54			open (CROSS_SECTION, "<$file");
55			my $crossSection = <CROSS_SECTION>;
56			close (CROSS_SECTION);
57			$crossSections{$dir} = $crossSection;
58			}
59	ahart	1.3	}
60	ahart	1.5	my $nTotalEvents = 0;
61	ahart	1.1	foreach my $file (@$files)
62			{
63			next if $file eq ".";
64			next if $file eq "..";
65	ahart	1.5	my $dir = $file;
66			$dir =~ s/^(.)\/[^\/]$/$1/;
67	ahart	1.3	my $badJob = 0;
68			if ($file =~ m/^._[^_]\.root$/)
69	ahart	1.1	{
70	ahart	1.3	my $jobNumber = $file;
71			$jobNumber =~ s/^._([^_])\.root$/$1/;
72	ahart	1.5	$badJob = defined $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
73			}
74			next if $badJob;
75			if ($file =~ m/^.*\.root$/)
76			{
77			my $nEvents = countEvents ($file);
78			$nTotalEvents = -1 if $nTotalEvents < 0 \|\| $nEvents < 0;
79			$nTotalEvents += $nEvents if !($nTotalEvents < 0 \|\| $nEvents < 0);
80			}
81			}
82			foreach my $file (@$files)
83			{
84			next if $file eq ".";
85			next if $file eq "..";
86			my $dir = $file;
87			$dir =~ s/^(.)\/[^\/]$/$1/;
88			my $badJob = 0;
89			if ($file =~ m/^._[^_]\.root$/)
90			{
91			my $jobNumber = $file;
92			$jobNumber =~ s/^._([^_])\.root$/$1/;
93			$badJob = defined $exitCodes{$dir} && defined $exitCodes{$dir}{$jobNumber} && $exitCodes{$dir}{$jobNumber};
94	ahart	1.1	}
95	ahart	1.3	next if $badJob;
96			if ($file =~ m/^.*\.root$/)
97	ahart	1.1	{
98	ahart	1.3	push (@rootFiles, $file);
99			push (@weights, $opt{"weight"}) if $opt{"weight"};
100	ahart	1.5	push (@weights, ($crossSections{$dir} * $integratedLuminosity) / $nTotalEvents) if !$opt{"weight"} && defined $crossSections{$dir};
101			push (@weights, 1.0) if !$opt{"weight"} && !(defined $crossSections{$dir});
102	ahart	1.1	}
103			}
104	ahart	1.3	if (!@rootFiles)
105			{
106			print "Found no ROOT files to merge!\n";
107			exit;
108			}
109	ahart	1.1	my $rootFiles = join (" ", @rootFiles);
110	ahart	1.3	my $weights = join (",", @weights);
111			system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root -w $weights");
112	ahart	1.1	system ("cutFlowTable $opt{'prefix'}.root cutFlow >& $opt{'prefix'}.tex");
113	ahart	1.4	my $goodEvents = countEvents ("$opt{'prefix'}.root");
114	ahart	1.5	print "$nGoodJobs jobs ran successfully over $nTotalEvents ($goodEvents weighted) events.\n" if $counting;
115	ahart	1.4	print "$nBadJobs jobs failed to run.\n" if $counting;
116	ahart	1.1
117			sub
118			processArgs
119			{
120			my $argv = shift;
121
122			my @files;
123			foreach my $arg (@$argv)
124			{
125			$arg =~ s/\/*$//;
126			if (!(-e $arg))
127			{
128			print "$arg does not exist!\n";
129			exit;
130			}
131			next if ($arg =~ m/\/\.$/ \|\| $arg =~ m/\/\.\.$/);
132			if (-d $arg)
133			{
134			opendir (DIR, $arg);
135			my @dirContents = readdir (DIR);
136			closedir (DIR);
137			for (my $i = 0; $i < @dirContents; $i++)
138			{
139			$dirContents[$i] = "$arg/$dirContents[$i]";
140			}
141			my $newFiles = processArgs (\@dirContents);
142			push (@files, @$newFiles);
143			}
144			else
145			{
146			push (@files, $arg);
147			}
148			}
149
150			return \@files;
151			}
152
153			sub
154			printHelp
155			{
156			my $exeName = $0;
157			$exeName =~ s/^.\/([^\/])$/$1/;
158
159	ahart	1.3	print "Usage: $exeName -p PREFIX [OPTION]... DIRECTORIES_AND_FILES\n";
160	ahart	1.1	print "Merges ROOT files containing histograms and produces a cutflow table in the\n";
161			print "form of a LaTeX document from the histogram named \"cutFlow\". If there are\n";
162	ahart	1.5	print "Condor logs in the specified directories, checks for nonzero return values. If\n";
163			print "the directories where created by \"osusub\", uses the cross section from the\n";
164			print "database to weight all histograms.\n";
165	ahart	1.1	print "\n";
166			print "Mandatory arguments to long options are mandatory for short options too.\n";
167			printf "%-29s%s\n", " -h, --help", "print this help message";
168	ahart	1.5	printf "%-29s%s\n", " -l, --luminosity", "integrated luminosity to which the histograms are";
169			printf "%-29s%s\n", " ", "weighted (default: 10000/pb)";
170	ahart	1.1	printf "%-29s%s\n", " -p, --prefix PREFIX", "output files are named PREFIX.root and PREFIX.tex";
171	ahart	1.5	printf "%-29s%s\n", " -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
172			printf "%-29s%s\n", " ", "automatic weighting using the cross section from";
173			printf "%-29s%s\n", " ", "the database";
174	ahart	1.1
175			exit;
176			}
177	ahart	1.3
178			sub
179			getRunList
180			{
181			my $runListFile = shift;
182
183			open (RUN_LIST, "<$runListFile");
184			my @runList0 = <RUN_LIST>;
185			close (RUN_LIST);
186			my @runList;
187			foreach my $file (@runList0)
188			{
189			next if !($file =~ m/^.file:.\.root.*/);
190			$file =~ s/.file:(.)\.root.*/$1.root/;
191			push (@runList, $file);
192			}
193
194			return \@runList;
195			}
196
197			sub
198			countEvents
199			{
200	ahart	1.4	my $file = shift;
201
202			my $output = `getEventsFromCutFlow $file cutFlow`;
203	ahart	1.5	if ($output =~ m/Did not find a histogram named/)
204			{
205			print $output;
206			return -1;
207			}
208	ahart	1.4	$output =~ s/^.: (.)$/$1/;
209			$output =~ s/\n//g;
210
211			return $output;
212	ahart	1.3	}