1 #!/usr/bin/perl -w 2 # 3 # $RCSfile: ModifySDFilesDataFields.pl,v $ 4 # $Date: 2008/02/02 22:23:45 $ 5 # $Revision: 1.14 $ 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($ModifyMolName, $ModifyDataFields, $KeepAllOldDataFields, $KeepUnMappedOldDataFields, $UseDataFieldForMolName, $AlwaysReplaceMolName, $MolNameDataField, $MolNamePrefix, $CreateDataFieldURL, $URLDataFieldLabel, $URLCGIScriptName, $URLParamName, $URLCmpdIdFieldName, %SpecifiedNewToOldSDFieldMap, %SpecifiedOldToNewSDFieldMap, %SpecifiedCommonFieldMap); 60 ProcessOptions(); 61 62 print "Checking input SD file(s)...\n"; 63 my(@SDFilesOkay, @SDFilesOutFile); 64 RetrieveSDFilesInfo(); 65 66 # Generate output files... 67 my($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 $SDFile...\n" 79 } 80 ModifySDFile($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 # Modify SD file data fields.... 93 sub ModifySDFile { 94 my($Index) = @_; 95 my($SDFile, $NewSDFile); 96 97 $SDFile = $SDFilesList[$Index]; 98 $NewSDFile = $SDFilesOutFile[$Index]; 99 100 print "Generating new SD file $NewSDFile...\n"; 101 open NEWSDFILE, ">$NewSDFile" or die "Error: Couldn't open $NewSDFile: $! \n"; 102 open SDFILE, "$SDFile" or die "Error: Can't open $SDFile: $! \n"; 103 104 my($CmpdCount, $CmpdString, $CmpdData, $MolName, $OldSDField, $NewSDField, $CommonSDField, $Label, $Value, $FieldValues, @CmpdLines, %DataFieldAndValues, @DataFieldLabels); 105 $CmpdCount = 0; 106 107 COMPOUND: while ($CmpdString = ReadCmpdString(\*SDFILE)) { 108 $CmpdCount++; 109 @CmpdLines = split "\n", $CmpdString; 110 if ($UseDataFieldForMolName || $ModifyDataFields) { 111 %DataFieldAndValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines); 112 } 113 if ($ModifyMolName) { 114 if ($AlwaysReplaceMolName || !IsNotEmpty($CmpdLines[0])) { 115 if ($UseDataFieldForMolName && exists($DataFieldAndValues{$MolNameDataField})) { 116 $MolName = $DataFieldAndValues{$MolNameDataField}; 117 if (length($MolName) > 80) { 118 $MolName = substr($MolName, 0, 80); 119 } 120 } 121 else { 122 $MolName = "${MolNamePrefix}${CmpdCount}"; 123 } 124 $CmpdLines[0] = $MolName; 125 $CmpdString = join "\n", @CmpdLines; 126 } 127 } 128 if (!$ModifyDataFields) { 129 # Just write the data and get the next compound... 130 print NEWSDFILE "$CmpdString\n"; 131 next COMPOUND; 132 } 133 # Write out the structure data now and handle the old data fields later... 134 ($CmpdData) = split /\n>/, $CmpdString; 135 print NEWSDFILE "$CmpdData\n"; 136 137 # Modify specified data fields... 138 for $NewSDField (sort keys %SpecifiedNewToOldSDFieldMap) { 139 $FieldValues = ""; 140 for $OldSDField (@{$SpecifiedNewToOldSDFieldMap{$NewSDField}}) { 141 if (exists($DataFieldAndValues{$OldSDField}) && length($DataFieldAndValues{$OldSDField})) { 142 $Value = $DataFieldAndValues{$OldSDField}; 143 $FieldValues .= ($FieldValues) ? "\n$Value" : $Value; 144 } 145 } 146 print NEWSDFILE "> <$NewSDField>\n$FieldValues\n\n"; 147 } 148 # Add specified common fields... 149 for $CommonSDField (sort keys %SpecifiedCommonFieldMap) { 150 $Value = $SpecifiedCommonFieldMap{$CommonSDField}; 151 print NEWSDFILE "> <$CommonSDField>\n$Value\n\n"; 152 } 153 if ($CreateDataFieldURL) { 154 $Value = ""; 155 if (exists($DataFieldAndValues{$URLCmpdIdFieldName}) && length($DataFieldAndValues{$URLCmpdIdFieldName})) { 156 $Value = $DataFieldAndValues{$URLCmpdIdFieldName}; 157 $Value = "${URLCGIScriptName}?${URLParamName}=${Value}"; 158 } 159 print NEWSDFILE "> <$URLDataFieldLabel>\n$Value\n\n"; 160 } 161 162 # Handle old data fields and write 'em in the same order as they appear in the input 163 # files... 164 if ($KeepAllOldDataFields || $KeepUnMappedOldDataFields) { 165 my($KeepLabel); 166 @DataFieldLabels = GetCmpdDataHeaderLabels(\@CmpdLines); 167 LABEL: for $Label (@DataFieldLabels) { 168 $KeepLabel = $KeepAllOldDataFields ? 1 : ( exists($SpecifiedOldToNewSDFieldMap{$Label}) ? 0 : 1 ); 169 if (!$KeepLabel) { 170 next LABEL; 171 } 172 $Value = $DataFieldAndValues{$Label}; 173 print NEWSDFILE "> <$Label>\n$Value\n\n"; 174 } 175 } 176 177 print NEWSDFILE "\$\$\$\$\n"; 178 } 179 close NEWSDFILE; 180 close SDFILE; 181 } 182 183 # Process option values... 184 sub ProcessOptions { 185 186 $ModifyMolName = 1; $ModifyDataFields = 0; 187 if ($Options{mode} =~ /^both$/i) { 188 $ModifyMolName = 1; $ModifyDataFields = 1; 189 } 190 elsif ($Options{mode} =~ /^datafields$/i) { 191 $ModifyMolName = 0; $ModifyDataFields = 1; 192 } 193 194 $KeepAllOldDataFields = ($Options{keepolddatafields} =~ /^all$/i) ? 1 : 0; 195 $KeepUnMappedOldDataFields = ($Options{keepolddatafields} =~ /^unmappedonly$/i) ? 1 : 0; 196 197 $UseDataFieldForMolName = ($Options{molnamemode} =~ /^datafield$/i) ? 1 : 0; 198 $MolNameDataField = ""; $MolNamePrefix = "Cmpd"; 199 if ($Options{molname}) { 200 if ($UseDataFieldForMolName) { 201 $MolNameDataField = $Options{molname}; 202 } 203 else { 204 $MolNamePrefix = $Options{molname}; 205 } 206 } 207 $AlwaysReplaceMolName = ($Options{molnamereplace} =~ /^always$/i) ? 1 : 0; 208 209 if ($Options{datafieldsmap} && $Options{datafieldsmapfile}) { 210 die "Error: Both \"--datafieldsmap\" and \"--datafieldsmapfile\" options specified: only one is allowed at a time\n"; 211 } 212 213 %SpecifiedNewToOldSDFieldMap = (); 214 %SpecifiedOldToNewSDFieldMap = (); 215 216 my($SpecifiedDataFieldMap); 217 $SpecifiedDataFieldMap = ""; 218 if ($Options{datafieldsmap}) { 219 $SpecifiedDataFieldMap = $Options{datafieldsmap}; 220 } 221 elsif ($Options{datafieldsmapfile}) { 222 my($Line, @LineWords); 223 open DATAFIELDSFILE, "$Options{datafieldsmapfile}" or die "Couldn't open $Options{datafieldsmapfile}: $! \n"; 224 while ($Line = GetTextLine(\*DATAFIELDSFILE)) { 225 @LineWords = quotewords(";", 0, $Line); 226 $SpecifiedDataFieldMap .= JoinWords(\@LineWords, ";", 0); 227 } 228 close DATAFIELDSFILE; 229 } 230 231 if ($SpecifiedDataFieldMap) { 232 my($DataFieldMap, $DataField, $NewSDField, @OldSDFields, @DataFieldMapSplit, @DataFieldsSplit, $FirstField); 233 @DataFieldMapSplit = split ";", $SpecifiedDataFieldMap; 234 for $DataFieldMap (@DataFieldMapSplit) { 235 @DataFieldsSplit = split ",", $DataFieldMap; 236 if (@DataFieldsSplit == 1) { 237 die "Error: Invalid number of comma delimited values, ", scalar(@DataFieldsSplit), ", specified, @DataFieldsSplit, using \"--datafieldsmap or --datafieldsmapfile\" option: it must contain more than one value.\n"; 238 } 239 $FirstField = 1; 240 @OldSDFields = (); 241 for $DataField (@DataFieldsSplit) { 242 if (!(defined($DataField) && length($DataField))) { 243 die "Error: One of the comma delimited values, \"", join(",", @DataFieldsSplit), "\", specified using \"--datafieldsmap or --datafieldsmapfile\" option is empty.\n"; 244 } 245 if ($FirstField) { 246 $FirstField = 0; 247 $NewSDField = $DataField; 248 } 249 else { 250 push @OldSDFields, $DataField; 251 } 252 } 253 # Make sure a datafield is only specified once... 254 if (exists $SpecifiedNewToOldSDFieldMap{$NewSDField}) { 255 die "Error: New data field, $NewSDField, specified more than once using \"--datafieldsmap or --datafieldsmapfile\" option.\n"; 256 } 257 $SpecifiedNewToOldSDFieldMap{$NewSDField} = (); 258 push @{$SpecifiedNewToOldSDFieldMap{$NewSDField}}, @OldSDFields; 259 for $DataField (@OldSDFields) { 260 if (exists $SpecifiedOldToNewSDFieldMap{$DataField} ) { 261 die "Error: SD field, $DataField, specified more than once using \"--datafieldsmap or --datafieldsmapfile\" option.\n"; 262 } 263 else { 264 $SpecifiedOldToNewSDFieldMap{$DataField} = $NewSDField; 265 } 266 } 267 268 } 269 } 270 271 %SpecifiedCommonFieldMap = (); 272 if ($Options{datafieldscommon}) { 273 my($DataFieldName, $DataFieldValue, $Index, @CommonDataFieldsSplit); 274 @CommonDataFieldsSplit = split ",", $Options{datafieldscommon}; 275 if (@CommonDataFieldsSplit % 2) { 276 die "Error: Invalid number of comma delimited values, ", scalar(@CommonDataFieldsSplit), ", specified \"", join(",", @CommonDataFieldsSplit), "\" using \"--datafieldscommon\" option: it must contain even number of values.\n"; 277 } 278 for ($Index = 0; $Index < @CommonDataFieldsSplit; $Index += 2) { 279 $DataFieldName = $CommonDataFieldsSplit[$Index]; 280 $DataFieldValue = $CommonDataFieldsSplit[$Index + 1]; 281 if (exists $SpecifiedCommonFieldMap{$DataFieldName}) { 282 die "Error: Common data field, $DataFieldName, specified more than once using \"--datafieldscommon\" option.\n"; 283 } 284 if (exists($SpecifiedNewToOldSDFieldMap{$DataFieldName}) || exists($SpecifiedOldToNewSDFieldMap{$DataFieldName})) { 285 die "Error: Common data field, $DataFieldName, specified using \"--datafieldscommon\" option cannot be specified in \"--datafieldsmap or --datafieldsmapfile\" option.\n"; 286 } 287 $SpecifiedCommonFieldMap{$DataFieldName} = $DataFieldValue; 288 } 289 } 290 291 $CreateDataFieldURL = (exists($Options{datafieldurl}) && length($Options{datafieldurl}) ) ? 1 : 0; 292 $URLDataFieldLabel = ""; $URLCGIScriptName = ""; $URLParamName = ""; $URLCmpdIdFieldName = ""; 293 if ($CreateDataFieldURL) { 294 my(@DataFieldURLSplit, $Value); 295 @DataFieldURLSplit = split ",", $Options{datafieldurl}; 296 if (@DataFieldURLSplit != 4) { 297 die "Error: Invalid number of values, ", scalar(@DataFieldURLSplit), ", specified using \"--datafieldURL\" option: it must contain 4 values.\n"; 298 } 299 for $Value (@DataFieldURLSplit) { 300 if (!IsNotEmpty($Value)) { 301 die "Error: One of the values, $Options{datafieldurl}, specified using \"--datafieldURL\" option is empty.\n"; 302 } 303 } 304 $URLDataFieldLabel = $DataFieldURLSplit[0]; 305 $URLCGIScriptName = $DataFieldURLSplit[1]; 306 $URLParamName = $DataFieldURLSplit[2]; 307 $URLCmpdIdFieldName = $DataFieldURLSplit[3]; 308 } 309 310 } 311 312 # Retrieve information about input SD files... 313 sub RetrieveSDFilesInfo { 314 my($Index, $SDFile, $FileDir, $FileName, $FileExt, $OutFileRoot, $OutFile, $DataFieldName); 315 316 @SDFilesOkay = (); 317 @SDFilesOutFile = (); 318 319 FILELIST: for $Index (0 .. $#SDFilesList) { 320 $SDFile = $SDFilesList[$Index]; 321 $SDFilesOkay[$Index] = 0; 322 $SDFilesOutFile[$Index] = ""; 323 if (!(-e $SDFile)) { 324 warn "Warning: Ignoring file $SDFile: It doesn't exist\n"; 325 next FILELIST; 326 } 327 if (!CheckFileType($SDFile, "sd sdf")) { 328 warn "Warning: Ignoring file $SDFile: It's not a SD file\n"; 329 next FILELIST; 330 } 331 $FileDir = ""; $FileName = ""; $FileExt = ""; 332 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile); 333 if ($Options{root} && (@SDFilesList == 1)) { 334 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root}); 335 if ($RootFileName && $RootFileExt) { 336 $FileName = $RootFileName; 337 } 338 else { 339 $FileName = $Options{root}; 340 } 341 $OutFileRoot = $FileName; 342 } 343 else { 344 $OutFileRoot = $FileName . "ModifiedDataFields"; 345 } 346 347 $OutFile = $OutFileRoot . ".$FileExt"; 348 if (lc($OutFile) eq lc($SDFile)) { 349 warn "Warning: Ignoring file $SDFile:Output file name, $OutFile, is same as input SD file name, $SDFile\n"; 350 next FILELIST; 351 } 352 if (!$Options{overwrite}) { 353 if (-e $OutFile) { 354 warn "Warning: Ignoring file $SDFile: The file $OutFile already exists\n"; 355 next FILELIST; 356 } 357 } 358 359 $SDFilesOkay[$Index] = 1; 360 $SDFilesOutFile[$Index] = "$OutFile"; 361 } 362 } 363 364 # Setup script usage and retrieve command line arguments specified using various options... 365 sub SetupScriptUsage { 366 367 # Retrieve all the options... 368 %Options = (); 369 $Options{detail} = 1; 370 $Options{keepolddatafields} = "none"; 371 $Options{mode} = "molname"; 372 $Options{molnamemode} = "labelprefix"; 373 $Options{molnamereplace} = "empty"; 374 375 if (!GetOptions(\%Options, "detail|d=i", "datafieldscommon=s", "datafieldsmap=s", "datafieldsmapfile=s", "datafieldurl=s", "help|h", "keepolddatafields|k=s", "mode|m=s", "molname=s", "molnamemode=s", "molnamereplace=s", "overwrite|o", "root|r=s", "workingdir|w=s")) { 376 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"; 377 } 378 if ($Options{workingdir}) { 379 if (! -d $Options{workingdir}) { 380 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"; 381 } 382 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n"; 383 } 384 if ($Options{keepolddatafields} !~ /^(all|unmappedonly|none)$/i) { 385 die "Error: The value specified, $Options{keepolddatafields}, for option \"-k --keepolddatafields\" is not valid. Allowed values: all, unmappedonly, or none\n"; 386 } 387 if ($Options{mode} !~ /^(molname|datafields|both)$/i) { 388 die "Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: molname, datafields, or both\n"; 389 } 390 if ($Options{molnamemode} !~ /^(datafield|labelprefix)$/i) { 391 die "Error: The value specified, $Options{molnamemode}, for option \"--molnamemode\" is not valid. Allowed values: datafield or labelprefix\n"; 392 } 393 if ($Options{molnamereplace} !~ /^(always|empty)$/i) { 394 die "Error: The value specified, $Options{molnamereplace}, for option \"--molnamereplace\" is not valid. Allowed values: always or empty\n"; 395 } 396 if (!IsPositiveInteger($Options{detail})) { 397 die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n"; 398 } 399 } 400
