ViewVC Help

View File | Revision Log | Show Annotations | Root Listing

root/cvsroot/UserCode/OSUT3Analysis/AnaTools/scripts/mergeHists

Comparing UserCode/OSUT3Analysis/AnaTools/scripts/mergeHists (file contents):
Revision 1.1 by ahart, Thu Jun 14 17:42:30 2012 UTC vs.
Revision 1.25 by ahart, Tue Jun 4 23:46:12 2013 UTC

#	Line 2 | Line 2
2
3		use strict;
4		use Getopt::Long;
5	+	use POSIX;
6	+
7	+	sub processArgs;
8	+	sub printHelp;
9	+	sub getRunList;
10	+	sub countEvents;
11
12		my %opt;
13		Getopt::Long::Configure ("bundling");
14	<	GetOptions (\%opt, "prefix|p=s", "help|h");
14	>	GetOptions (\%opt, "cutflow|c=s", "luminosity|l=s", "prefix|p=s", "weight|w=s", "xsection|x=s", "help|h");
15
16		printHelp () if $opt{"help"} || !$opt{"prefix"};
17		my $files = processArgs (\@ARGV);
18	<	my @rootFiles;
19	<	my $error = 0;
18	>	my %rootFiles;
19	>	my %weights;
20	>	my @weights;
21	>	my $nGoodJobs = 0;
22	>	my $nBadJobs = 0;
23	>	my $nIncompleteJobs = 0;
24	>	my $counting = 0;
25	>	my %exitCodes;
26	>	my %signals;
27	>	my %partial;
28	>	my %crossSections;
29	>	my %dirs;
30	>	my $integratedLuminosity = 10000;
31	>	$integratedLuminosity = $opt{"luminosity"} if $opt{"luminosity"};
32	>	my $cutFlow = "cutFlow";
33	>	$cutFlow = $opt{"cutflow"} if $opt{"cutflow"};
34		foreach my $file (@$files)
35		{
36		next if $file eq ".";
37		next if $file eq "..";
38	<	if ($file =~ m/\.root$/)
38	>	my $dir = $file;
39	>	$dir =~ s/^(.*)\/[^\/]*$/$1/;
40	>	if ($file =~ m/^.*\/condor_[^_]*\.log$/)
41	>	{
42	>	my $jobNumber = $file;
43	>	$jobNumber =~ s/^.*\/condor_([^_]*)\.log$/$1/;
44	>	open (FILE, "<$file");
45	>	my @fileContents = <FILE>;
46	>	close (FILE);
47	>	my $fileContents = join ("", @fileContents);
48	>	$fileContents =~ s/\n/ /g;
49	>	$counting = 1;
50	>	if ($fileContents =~ m/return value/)
51	>	{
52	>	$fileContents =~ s/.*\(return value ([^)]*)\).*/$1/g;
53	>	$nGoodJobs++;
54	>	print "WARNING: Nonzero exit code for job $jobNumber. (return value $fileContents)\n" if $fileContents != 0;
55	>	$exitCodes{$dir}{$jobNumber} = $fileContents;
56	>	}
57	>	elsif ($fileContents =~ m/signal/)
58	>	{
59	>	$fileContents =~ s/.*\(signal ([^)]*)\).*/$1/g;
60	>	$nBadJobs++;
61	>	print "WARNING: Skipping job $jobNumber. (signal $fileContents)\n";
62	>	$signals{$dir}{$jobNumber} = $fileContents;
63	>	}
64	>	else
65	>	{
66	>	$nIncompleteJobs++;
67	>	$partial{$dir}{$jobNumber} = 1;
68	>	}
69	>	}
70	>	if ($file =~ m/^.*\/crossSectionInPicobarn\.txt$/)
71	>	{
72	>	open (CROSS_SECTION, "<$file");
73	>	my $crossSection = <CROSS_SECTION>;
74	>	close (CROSS_SECTION);
75	>	$crossSections{$dir} = $crossSection;
76	>	}
77	>	}
78	>	foreach my $file (@$files)
79	>	{
80	>	next if $file eq ".";
81	>	next if $file eq "..";
82	>	my $dir = $file;
83	>	$dir =~ s/^(.*)\/[^\/]*$/$1/;
84	>	my $badJob = 0;
85	>	my $jobNumber;
86	>	if ($file =~ m/^.*_[^_]*\.root$/)
87	>	{
88	>	$jobNumber = $file;
89	>	$jobNumber =~ s/^.*_([^_]*)\.root$/$1/;
90	>	$badJob = (defined $signals{$dir} && defined $signals{$dir}{$jobNumber}) || (defined $partial{$dir} && defined $partial{$dir}{$jobNumber});
91	>	}
92	>	next if $badJob;
93	>	if ($file =~ m/^.*\.root$/)
94		{
95	<	push (@rootFiles, $file);
96	<	next;
95	>	foreach my $arg (@ARGV)
96	>	{
97	>	if (substr ($file, 0, length ($arg)) eq $arg)
98	>	{
99	>	$dirs{$arg} = $dir;
100	>	if (countEvents ($file, $cutFlow) < 0)
101	>	{
102	>	$nGoodJobs--;
103	>	$nBadJobs++;
104	>	print "WARNING: Skipping job $jobNumber. (bad ROOT file)\n";
105	>	last;
106	>	}
107	>	push (@{$rootFiles{$arg}}, $file);
108	>	if ($opt{"weight"})
109	>	{
110	>	push (@weights, $opt{"weight"}) if !($opt{"weight"} =~ m/,/);
111	>	push (@weights, 1.0) if $opt{"weight"} =~ m/,/;
112	>	}
113	>	push (@weights, $opt{"xsection"} * $integratedLuminosity) if !$opt{"weight"} && $opt{"xsection"};
114	>	push (@weights, $crossSections{$dir} * $integratedLuminosity) if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dir};
115	>	push (@weights, 1.0) if !$opt{"weight"} && !$opt{"xsection"} && !(defined $crossSections{$dir});
116	>	$weights{$arg} = $weights[-1];
117	>	}
118	>	}
119		}
120	<	open (FILE, "<$file");
121	<	my @fileContents = <FILE>;
122	<	close (FILE);
123	<	my $fileContents = join ("", @fileContents);
124	<	next if !($fileContents =~ m/\(return value [^)]*\)/);
125	<	$fileContents =~ s/\n/ /g;
126	<	$fileContents =~ s/.*\(return value ([^)]*)\).*/$1/g;
127	<	if ($fileContents != 0)
120	>	}
121	>	if (!%rootFiles)
122	>	{
123	>	print "Found no ROOT files to merge!\n";
124	>	exit;
125	>	}
126	>	my %nTotalEvents;
127	>	my $nTotalEvents = 0;
128	>	my @mergedFiles;
129	>	my @mergedWeights;
130	>	foreach my $arg (@ARGV)
131	>	{
132	>	my $rootFiles = join (" ", @{$rootFiles{$arg}});
133	>	my $tmpName = $arg . "_" . "$opt{'prefix'}.root";
134	>	$tmpName =~ s/\//_/g;
135	>	system ("mergeTFileServiceHistograms -i $rootFiles -o $tmpName");
136	>	my $count = countEvents ($tmpName, $cutFlow);
137	>	system ("cutFlowLimits $tmpName");
138	>	$nTotalEvents{$arg} = $count;
139	>	$nTotalEvents += $count;
140	>	$weights{$arg} /= $count if !$opt{"weight"} && $opt{"xsection"};
141	>	$weights{$arg} /= $count if !$opt{"weight"} && !$opt{"xsection"} && defined $crossSections{$dirs{$arg}};
142	>	push (@mergedFiles, $tmpName);
143	>	push (@mergedWeights, $weights{$arg});
144	>	foreach my $rootFile (@{$rootFiles{$arg}})
145		{
146	<	printf "Error: return value of %2d in \"$file\"!\n", $fileContents;
147	<	$error = 1;
146	>	my $weight = $weights{$arg};
147	>	system ("weightTrees $rootFile $weight");
148		}
149		}
150	<	exit if $error;
151	<	my $rootFiles = join (" ", @rootFiles);
152	<	system ("mergeTFileServiceHistograms -i $rootFiles -o $opt{'prefix'}.root");
153	<	system ("cutFlowTable $opt{'prefix'}.root cutFlow >& $opt{'prefix'}.tex");
150	>	my $mergedFiles = join (" ", @mergedFiles);
151	>	my $mergedWeights = join (",", @mergedWeights);
152	>	$mergedWeights = $opt{"weight"} if $opt{"weight"} && $opt{"weight"} =~ m/,/;
153	>	system ("mergeTFileServiceHistograms -i $mergedFiles -o $opt{'prefix'}.root -w $mergedWeights");
154	>	foreach my $mergedFile (@mergedFiles)
155	>	{
156	>	unlink ("$mergedFile");
157	>	}
158	>	print "=============================================\n";
159	>	my $output = sprintf "Cross-section of samples:\n";
160	>	my $printOutput = 0;
161	>	foreach my $arg (keys %weights)
162	>	{
163	>	my $shortArg = $arg;
164	>	$shortArg =~ s/^.*\/([^\/]*)$/$1/;
165	>	$output = sprintf "%s  $shortArg: %.5g pb\n", $output, ($crossSections{$dirs{$arg}});
166	>	$printOutput = 1 if $weights{$arg} != 1;
167	>	}
168	>	print $output if $printOutput;
169	>
170	>	my $output = sprintf "Effective luminosities of samples:\n";
171	>	my $printOutput = 0;
172	>	foreach my $arg (keys %weights)
173	>	{
174	>	my $shortArg = $arg;
175	>	$shortArg =~ s/^.*\/([^\/]*)$/$1/;
176	>	$output = sprintf "%s  $shortArg: %.5g/fb\n", $output, ($integratedLuminosity / (1000.0 * $weights{$arg})) if $weights{$arg} != 1;
177	>	$printOutput = 1 if $weights{$arg} != 1;
178	>	}
179	>	print $output if $printOutput;
180	>
181	>	$output = sprintf "Weights for target luminosity of %g/fb:\n", ($integratedLuminosity / 1000.0);
182	>	$printOutput = 0;
183	>	foreach my $arg (keys %weights)
184	>	{
185	>	my $shortArg = $arg;
186	>	$shortArg =~ s/^.*\/([^\/]*)$/$1/;
187	>	$output = sprintf "%s  $shortArg: %.5g\n", $output, $weights{$arg} if $weights{$arg} != 1;
188	>	$printOutput = 1 if $weights{$arg} != 1;
189	>	}
190	>	print $output if $printOutput;
191	>
192	>	my $goodEvents = countEvents ("$opt{'prefix'}.root", $cutFlow);
193	>	print "$nGoodJobs jobs ran successfully over $nTotalEvents ($goodEvents weighted) events.\n" if $counting;
194	>	print "$nBadJobs jobs failed to run.\n" if $counting;
195	>	print "$nIncompleteJobs jobs have not finished.\n" if $counting;
196	>	print "=============================================\n";
197
198		sub
199		processArgs
#	Line 80 | Line 237 | printHelp
237		my $exeName = $0;
238		$exeName =~ s/^.*\/([^\/]*)$/$1/;
239
240	<	print "Usage: $exeName [OPTIONS] -p PREFIX DIRECTORIES_AND_FILES\n";
241	<	print "Merges ROOT files containing histograms and produces a cutflow table in the\n";
242	<	print "form of a LaTeX document from the histogram named \"cutFlow\". If there are\n";
243	<	print "Condor logs in the specified directories, checks for nonzero return values.\n";
240	>	print "Usage: $exeName [OPTION]... -p PREFIX DIRECTORIES_AND_FILES\n";
241	>	print "Merges ROOT files containing histograms. If there are Condor logs in the\n";
242	>	print "specified directories, checks for nonzero return values. If the directories\n";
243	>	print "where created by \"osusub\", uses the cross section from the database to weight\n";
244	>	print "all histograms.\n";
245		print "\n";
246		print "Mandatory arguments to long options are mandatory for short options too.\n";
247	+	printf "%-29s%s\n", "  -c, --cutflow HISTOGRAM", "name of histogram to use for the cutflow (default:";
248	+	printf "%-29s%s\n", "                  ", "cutFlow)";
249		printf "%-29s%s\n", "  -h, --help", "print this help message";
250	<	printf "%-29s%s\n", "  -p, --prefix PREFIX", "output files are named PREFIX.root and PREFIX.tex";
250	>	printf "%-29s%s\n", "  -l, --luminosity", "integrated luminosity to which the histograms are";
251	>	printf "%-29s%s\n", "                  ", "weighted (default: 10000/pb)";
252	>	printf "%-29s%s\n", "  -p, --prefix PREFIX", "output is  named PREFIX.root";
253	>	printf "%-29s%s\n", "  -w, --weight WEIGHT", "scale the output by WEIGHT, overriding the";
254	>	printf "%-29s%s\n", "                     ", "automatic weighting using the cross section from";
255	>	printf "%-29s%s\n", "                     ", "the database; WEIGHT may be a single number or a";
256	>	printf "%-29s%s\n", "                     ", "comma-separated list, one for each input file";
257	>	printf "%-29s%s\n", "  -x, --xsection XSECTION", "use XSECTION to weight the histograms instead of";
258	>	printf "%-29s%s\n", "                     ", "the value in the database";
259
260		exit;
261		}
262	+
263	+	sub
264	+	getRunList
265	+	{
266	+	my $runListFile = shift;
267	+
268	+	open (RUN_LIST, "<$runListFile");
269	+	my @runList0 = <RUN_LIST>;
270	+	close (RUN_LIST);
271	+	my @runList;
272	+	foreach my $file (@runList0)
273	+	{
274	+	next if !($file =~ m/^.*file:.*\.root.*/);
275	+	$file =~ s/.*file:(.*)\.root.*/$1.root/;
276	+	push (@runList, $file);
277	+	}
278	+
279	+	return \@runList;
280	+	}
281	+
282	+	sub
283	+	countEvents
284	+	{
285	+	my $file = shift;
286	+	my $cutFlow = shift;
287	+
288	+	my $output = `getEventsFromCutFlow $file $cutFlow`;
289	+	if ($output =~ m/Did not find a histogram named/ || $output =~ m/appears to be empty/ || $output =~ m/Failed to open/)
290	+	{
291	+	print $output;
292	+	return -1;
293	+	}
294	+	$output =~ s/^.*: (.*)$/$1/;
295	+	$output =~ s/\n//g;
296	+
297	+	return $output;
298	+	}

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines