1 #!/usr/bin/perl -w 2 # 3 # $RCSfile: SortSDFiles.pl,v $ 4 # $Date: 2008/01/30 21:45:03 $ 5 # $Revision: 1.13 $ 6 # 7 # Author: Manish Sud <msud@san.rr.com> 8 # 9 # Copyright (C) 2004-2008 Manish Sud. All rights reserved. 10 # 11 # This file is part of MayaChemTools. 12 # 13 # MayaChemTools is free software; you can redistribute it and/or modify it under 14 # the terms of the GNU Lesser General Public License as published by the Free 15 # Software Foundation; either version 3 of the License, or (at your option) any 16 # later version. 17 # 18 # MayaChemTools is distributed in the hope that it will be useful, but without 19 # any warranty; without even the implied warranty of merchantability of fitness 20 # for a particular purpose. See the GNU Lesser General Public License for more 21 # details. 22 # 23 # You should have received a copy of the GNU Lesser General Public License 24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or 25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330, 26 # Boston, MA, 02111-1307, USA. 27 # 28 29 use 5.006; 30 use strict; 31 use FindBin; use lib "$FindBin::Bin/../lib"; 32 use Getopt::Long; 33 use File::Basename; 34 use Text::ParseWords; 35 use Benchmark; 36 use FileUtil; 37 use SDFileUtil; 38 use TextUtil; 39 40 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime); 41 42 # Autoflush STDOUT 43 $| = 1; 44 45 # Starting message... 46 $ScriptName = basename($0); 47 print "\n$ScriptName: Starting...\n\n"; 48 $StartTime = new Benchmark; 49 50 # Get the options and setup script... 51 SetupScriptUsage(); 52 if ($Options{help} || @ARGV < 1) { 53 die GetUsageFromPod("$FindBin::Bin/$ScriptName"); 54 } 55 56 my(@SDFilesList); 57 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd"); 58 59 my($SpecifiedDataFieldName, $DetailLevel); 60 ProcessOptions(); 61 62 print "Checking input SD file(s)...\n"; 63 my(@SDFilesOkay, @SDFilesOutFile, @SDFilesKeyDataFieldName); 64 RetrieveSDFilesInfo(); 65 66 # Generate output files... 67 my($Index, $SDFile); 68 if (@SDFilesList > 1) { 69 print "Processing SD file(s)...\n"; 70 } 71 for $Index (0 .. $#SDFilesList) { 72 if ($SDFilesOkay[$Index]) { 73 $SDFile = $SDFilesList[$Index]; 74 if (@SDFilesList > 1) { 75 print "\nProcessing file $SDFile...\n"; 76 } 77 else { 78 print "Processing file $SDFile...\n" 79 } 80 SortSDFile($Index); 81 } 82 } 83 84 print "$ScriptName:Done...\n\n"; 85 86 $EndTime = new Benchmark; 87 $TotalTime = timediff ($EndTime, $StartTime); 88 print "Total time: ", timestr($TotalTime), "\n"; 89 90 ############################################################################### 91 92 # Process option values... 93 sub ProcessOptions { 94 $DetailLevel = $Options{detail}; 95 96 $SpecifiedDataFieldName = ""; 97 if (defined $Options{key}) { 98 $SpecifiedDataFieldName = $Options{key}; 99 } 100 } 101 102 # Retrieve information about input SD files... 103 sub RetrieveSDFilesInfo { 104 my($Index, $SDFile, $FileDir, $FileName, $FileExt, $OutFileRoot, $OutFile, $DataFieldName); 105 106 @SDFilesOkay = (); 107 @SDFilesOutFile = (); 108 @SDFilesKeyDataFieldName = (); 109 110 FILELIST: for $Index (0 .. $#SDFilesList) { 111 $SDFile = $SDFilesList[$Index]; 112 $SDFilesOkay[$Index] = 0; 113 $SDFilesOutFile[$Index] = ""; 114 $SDFilesKeyDataFieldName[$Index] = ""; 115 if (!(-e $SDFile)) { 116 warn "Warning: Ignoring file $SDFile: It doesn't exist\n"; 117 next FILELIST; 118 } 119 if (!CheckFileType($SDFile, "sd sdf")) { 120 warn "Warning: Ignoring file $SDFile: It's not a SD file\n"; 121 next FILELIST; 122 } 123 $FileDir = ""; $FileName = ""; $FileExt = ""; 124 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile); 125 if ($Options{root} && (@SDFilesList == 1)) { 126 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root}); 127 if ($RootFileName && $RootFileExt) { 128 $FileName = $RootFileName; 129 } 130 else { 131 $FileName = $Options{root}; 132 } 133 $OutFileRoot = $FileName; 134 } 135 else { 136 $OutFileRoot = $FileName . "SortedByDataField"; 137 } 138 139 $OutFile = $OutFileRoot . ".$FileExt"; 140 if (lc($OutFile) eq lc($SDFile)) { 141 warn "Warning: Ignoring file $SDFile:Output file name, $OutFile, is same as input SD file name, $SDFile\n"; 142 next FILELIST; 143 } 144 if (!$Options{overwrite}) { 145 if (-e $OutFile) { 146 warn "Warning: Ignoring file $SDFile: The file $OutFile already exists\n"; 147 next FILELIST; 148 } 149 } 150 # Setup data field name... 151 if ($SpecifiedDataFieldName) { 152 $DataFieldName = $SpecifiedDataFieldName; 153 } 154 else { 155 my($CmpdString, @CmpdLines, @DataFieldNames); 156 @DataFieldNames = (); 157 if (!open(SDFILE, "$SDFile")) { 158 warn "Warning: Ignoring file $SDFile: Couldn't open it: $! \n"; 159 next FILELIST; 160 } 161 $CmpdString = ReadCmpdString(\*SDFILE); 162 close SDFILE; 163 164 @CmpdLines = split "\n", $CmpdString; 165 @DataFieldNames = GetCmpdDataHeaderLabels(\@CmpdLines); 166 $DataFieldName = $DataFieldNames[0]; 167 } 168 169 $SDFilesOkay[$Index] = 1; 170 $SDFilesOutFile[$Index] = "$OutFile"; 171 $SDFilesKeyDataFieldName[$Index] = $DataFieldName; 172 } 173 } 174 175 # Sort it out... 176 sub SortSDFile { 177 my($Index) = @_; 178 my($SDFile, $NewSDFile, $KeyDataFieldName); 179 180 $SDFile = $SDFilesList[$Index]; 181 $NewSDFile = $SDFilesOutFile[$Index]; 182 $KeyDataFieldName = $SDFilesKeyDataFieldName[$Index]; 183 184 print "Generating new SD file $NewSDFile...\n"; 185 open NEWSDFILE, ">$NewSDFile" or die "Error: Couldn't open $NewSDFile: $! \n"; 186 open SDFILE, "$SDFile" or die "Error: Can't open $SDFile: $! \n"; 187 188 # Go over all compound records and store 'em using key value as hash... 189 my(%KeyToCompundRecordsMap, @InvalidCompoundRecords, $CmpdCount, $CmpdString, @CmpdLines, %DataFieldValues, $KeyDataFieldValue); 190 %KeyToCompundRecordsMap = (); 191 @InvalidCompoundRecords = (); 192 $CmpdCount = 0; 193 194 COMPOUND: while ($CmpdString = ReadCmpdString(\*SDFILE)) { 195 $CmpdCount++; 196 @CmpdLines = split "\n", $CmpdString; 197 %DataFieldValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines); 198 #Make sure data field value is okay... 199 if (!(IsNotEmpty($DataFieldValues{$KeyDataFieldName}) && ($DataFieldValues{$KeyDataFieldName} !~ /\n/))) { 200 push @InvalidCompoundRecords, $CmpdString; 201 if ($DetailLevel >= 3 ) { 202 print "Ignoring compound record $CmpdCount: Contains empty value for key data field $KeyDataFieldName :\n $CmpdString\n\n"; 203 } 204 elsif ($DetailLevel >= 2) { 205 print "Ignoring compound record $CmpdCount: Contains empty value for key data field $KeyDataFieldName...\n"; 206 } 207 next COMPOUND; 208 } 209 $KeyDataFieldValue = $DataFieldValues{$KeyDataFieldName}; 210 if ($Options{keydata} =~ /^numeric$/i) { 211 if (!IsFloat($KeyDataFieldValue)) { 212 push @InvalidCompoundRecords, $CmpdString; 213 if ($DetailLevel >= 3 ) { 214 print "Ignoring compound record $CmpdCount: Contains non-numerical value for key data field $KeyDataFieldName :\n $CmpdString\n\n"; 215 } 216 elsif ($DetailLevel >= 2) { 217 print "Ignoring compound record $CmpdCount: Contains non-numerical value for key data field $KeyDataFieldName...\n"; 218 } 219 next COMPOUND; 220 } 221 } 222 if (exists($KeyToCompundRecordsMap{$KeyDataFieldValue})) { 223 # Append to existing coompund data... 224 $KeyToCompundRecordsMap{$KeyDataFieldValue} .= "\n" . $CmpdString; 225 } 226 else { 227 $KeyToCompundRecordsMap{$KeyDataFieldValue} = $CmpdString; 228 } 229 } 230 231 if ($Options{sort} =~ /^ascending$/i) { 232 if ($Options{keydata} =~ /^alphanumeric$/i) { 233 for $KeyDataFieldValue (sort { lc($a) cmp lc($b) } keys %KeyToCompundRecordsMap ) { 234 print NEWSDFILE "$KeyToCompundRecordsMap{$KeyDataFieldValue}\n"; 235 } 236 } 237 else { 238 for $KeyDataFieldValue (sort { $a <=> $b } keys %KeyToCompundRecordsMap ) { 239 print NEWSDFILE "$KeyToCompundRecordsMap{$KeyDataFieldValue}\n"; 240 } 241 } 242 } 243 else { 244 if ($Options{keydata} =~ /^alphanumeric$/i) { 245 for $KeyDataFieldValue (sort { lc($b) cmp lc($a) } keys %KeyToCompundRecordsMap ) { 246 print NEWSDFILE "$KeyToCompundRecordsMap{$KeyDataFieldValue}\n"; 247 } 248 } 249 else { 250 for $KeyDataFieldValue (sort { $b <=> $a } keys %KeyToCompundRecordsMap ) { 251 print NEWSDFILE "$KeyToCompundRecordsMap{$KeyDataFieldValue}\n"; 252 } 253 } 254 } 255 # Append the records containing data not appropriate for sorting... 256 if (@InvalidCompoundRecords) { 257 print "Placing ", scalar(@InvalidCompoundRecords)," compound record(s) with invalid data field key data the end...\n"; 258 for $CmpdString (@InvalidCompoundRecords) { 259 print NEWSDFILE "$CmpdString\n"; 260 } 261 } 262 close NEWSDFILE; 263 close SDFILE; 264 } 265 266 # Setup script usage and retrieve command line arguments specified using various options... 267 sub SetupScriptUsage { 268 269 # Retrieve all the options... 270 %Options = (); 271 $Options{detail} = 1; 272 $Options{sort} = "ascending"; 273 $Options{keydata} = "numeric"; 274 if (!GetOptions(\%Options, "detail|d=i", "help|h", "key|k=s", "keydata=s", "overwrite|o", "root|r=s", "sort|s=s", "workingdir|w=s")) { 275 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"; 276 } 277 if ($Options{workingdir}) { 278 if (! -d $Options{workingdir}) { 279 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"; 280 } 281 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n"; 282 } 283 if ($Options{keydata} !~ /^(numeric|alphanumeric)$/i) { 284 die "Error: The value specified, $Options{keydata}, for option \"--keydata\" is not valid. Allowed values: numeric or alphanumeric\n"; 285 } 286 if ($Options{sort} !~ /^(ascending|descending)$/i) { 287 die "Error: The value specified, $Options{sort}, for option \"-s --sort\" is not valid. Allowed values: ascending or descending\n"; 288 } 289 if (!IsPositiveInteger($Options{detail})) { 290 die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n"; 291 } 292 } 293
