1 #!/usr/bin/perl -w
   2 #
   3 # File: ElementalAnalysisTextFiles.pl
   4 # Author: Manish Sud <msud@san.rr.com>
   5 #
   6 # Copyright (C) 2024 Manish Sud. All rights reserved.
   7 #
   8 # This file is part of MayaChemTools.
   9 #
  10 # MayaChemTools is free software; you can redistribute it and/or modify it under
  11 # the terms of the GNU Lesser General Public License as published by the Free
  12 # Software Foundation; either version 3 of the License, or (at your option) any
  13 # later version.
  14 #
  15 # MayaChemTools is distributed in the hope that it will be useful, but without
  16 # any warranty; without even the implied warranty of merchantability of fitness
  17 # for a particular purpose.  See the GNU Lesser General Public License for more
  18 # details.
  19 #
  20 # You should have received a copy of the GNU Lesser General Public License
  21 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
  22 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
  23 # Boston, MA, 02111-1307, USA.
  24 #
  25 
  26 use strict;
  27 use FindBin; use lib "$FindBin::Bin/../lib";
  28 use Getopt::Long;
  29 use File::Basename;
  30 use Text::ParseWords;
  31 use Benchmark;
  32 use FileUtil;
  33 use TextUtil;
  34 use MolecularFormula;
  35 
  36 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
  37 
  38 # Autoflush STDOUT
  39 $| = 1;
  40 
  41 # Starting message...
  42 $ScriptName = basename($0);
  43 print "\n$ScriptName: Starting...\n\n";
  44 $StartTime = new Benchmark;
  45 
  46 # Get the options and setup script...
  47 SetupScriptUsage();
  48 if ($Options{help} || @ARGV < 1) {
  49   die GetUsageFromPod("$FindBin::Bin/$ScriptName");
  50 }
  51 
  52 my(@TextFilesList);
  53 @TextFilesList = ExpandFileNames(\@ARGV, "csv tsv");
  54 
  55 # Process options...
  56 my(%OptionsInfo);
  57 print "Processing options...\n";
  58 ProcessOptions();
  59 
  60 print "Checking input text file(s)...\n";
  61 my(%TextFilesInfo);
  62 RetrieveTextFilesInfo();
  63 RetrieveColumnsAndLabelsInfo();
  64 
  65 # Generate output files...
  66 my($FileIndex);
  67 if (@TextFilesList > 1) {
  68   print "\nProcessing text files...\n";
  69 }
  70 for $FileIndex (0 .. $#TextFilesList) {
  71   if ($TextFilesInfo{FileOkay}[$FileIndex]) {
  72     print "\nProcessing file $TextFilesList[$FileIndex]...\n";
  73     PerformElementalAnalysis($FileIndex);
  74   }
  75 }
  76 print "\n$ScriptName:Done...\n\n";
  77 
  78 $EndTime = new Benchmark;
  79 $TotalTime = timediff ($EndTime, $StartTime);
  80 print "Total time: ", timestr($TotalTime), "\n";
  81 
  82 ###############################################################################
  83 
  84 # Perform elemental analysis...
  85 sub PerformElementalAnalysis {
  86   my($Index) = @_;
  87   my($TextFile, $NewTextFile, $FormulaCol, $Line, $NewLine, $FormulaColValue, $InDelim, $ColNum, $Value, $Status, $ErrorMsg, @ColLabels, @LineWords, @ColNumsBeforeNew, @ColNumsAfterNew);
  88 
  89   $TextFile = $TextFilesList[$Index];
  90   $InDelim = $TextFilesInfo{InDelim}[$Index];
  91   $NewTextFile = $TextFilesInfo{OutFile}[$Index];
  92   $FormulaCol = $TextFilesInfo{FormulaColNum}[$Index];
  93 
  94   @ColNumsBeforeNew = @{$TextFilesInfo{ColNumsBeforeNew}[$Index]};
  95   @ColNumsAfterNew = @{$TextFilesInfo{ColNumsAfterNew}[$Index]};
  96 
  97   print "Generating new Text file $NewTextFile...\n";
  98   open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
  99   open TEXTFILE, "$TextFile" or die "Error: Can't open $TextFile: $! \n";
 100 
 101   # Skip over column labels from old file...
 102   $Line = GetTextLine(\*TEXTFILE);
 103 
 104   # Add column lablels in new file...
 105   @ColLabels = ();
 106   for $ColNum (@ColNumsBeforeNew) {
 107     push @ColLabels, $TextFilesInfo{ColLabels}[$Index][$ColNum];
 108   }
 109   for $Value (@{$OptionsInfo{SpecifiedCalculations}}) {
 110     push @ColLabels, $TextFilesInfo{ValueLabelsMap}[$Index]{$Value};
 111   }
 112   for $ColNum (@ColNumsAfterNew) {
 113     push @ColLabels, $TextFilesInfo{ColLabels}[$Index][$ColNum];
 114   }
 115   $NewLine = '';
 116   $NewLine = JoinWords(\@ColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 117   print NEWTEXTFILE "$NewLine\n";
 118 
 119   # Go over all rows...
 120   my($LineCount, $ElementsRef, $ElementCompositionRef, $CalculationType, $CalculatedValue, @CalculatedValues);
 121 
 122   $LineCount = 1;
 123   TEXTLINE: while ($Line = GetTextLine(\*TEXTFILE)) {
 124     @LineWords = quotewords($InDelim, 0, $Line);
 125     $LineCount++;
 126 
 127     @CalculatedValues = ();
 128     for $Value (@{$OptionsInfo{SpecifiedCalculations}}) {
 129       push @CalculatedValues, '';
 130     }
 131     if ($FormulaCol > @LineWords) {
 132       $ErrorMsg = "Ignoring line $LineCount: Formula column $ColLabels[$FormulaCol] not found";
 133       PrintErrorMsg($Line, $ErrorMsg);
 134       ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
 135       next TEXTLINE;
 136     }
 137 
 138     # Make sure it's a valid molecular formula...
 139     $FormulaColValue = $LineWords[$FormulaCol];
 140     if ($OptionsInfo{CheckFormula}) {
 141       ($Status, $ErrorMsg) = MolecularFormula::IsMolecularFormula($FormulaColValue);
 142       if (!$Status) {
 143         $ErrorMsg = "Ignoring line $LineCount: Formula column $ColLabels[$FormulaCol] value is not valid: $ErrorMsg";
 144         PrintErrorMsg($Line, $ErrorMsg);
 145         ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
 146         next TEXTLINE;
 147       }
 148     }
 149 
 150     # Calculate appropriate values and write 'em out...
 151     @CalculatedValues = ();
 152     for $CalculationType (@{$OptionsInfo{SpecifiedCalculations}}) {
 153       if ($CalculationType =~ /^ElementalAnalysis$/i) {
 154         ($ElementsRef, $ElementCompositionRef) = MolecularFormula::CalculateElementalComposition($FormulaColValue);
 155         $CalculatedValue = (defined($ElementsRef) && defined($ElementCompositionRef)) ? MolecularFormula::FormatCompositionInfomation($ElementsRef, $ElementCompositionRef, $OptionsInfo{Precision}) : '';
 156       }
 157       elsif ($CalculationType =~ /^MolecularWeight$/i) {
 158         $CalculatedValue = MolecularFormula::CalculateMolecularWeight($FormulaColValue);
 159         $CalculatedValue = (defined($CalculatedValue) && length($CalculatedValue)) ? (sprintf("%.$OptionsInfo{Precision}f", $CalculatedValue)) : "";
 160       }
 161       elsif ($CalculationType =~ /^ExactMass$/i) {
 162         $CalculatedValue = MolecularFormula::CalculateExactMass($FormulaColValue);
 163         $CalculatedValue = (defined($CalculatedValue) && length($CalculatedValue)) ? (sprintf("%.$OptionsInfo{Precision}f", $CalculatedValue)) : "";
 164       }
 165       else {
 166         $CalculatedValue = '';
 167       }
 168       push @CalculatedValues, $CalculatedValue;
 169     }
 170 
 171     ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
 172   }
 173   close NEWTEXTFILE;
 174   close TEXTFILE;
 175 
 176 }
 177 
 178 # Write out new line using old and new calculated data...
 179 sub ComposeAndWriteNewLine {
 180   my($NewTextFileRef, $OldLineWordsRef, $ColNumsBeforeNewRef, $ColNumsAfterNewRef, $CalculatedValuesRef) = @_;
 181   my($NewLine, $ColNum, $Value, @NewLineWords);
 182 
 183   @NewLineWords = ();
 184   for $ColNum (@{$ColNumsBeforeNewRef}) {
 185     push @NewLineWords, $OldLineWordsRef->[$ColNum];
 186   }
 187   for $Value (@{$CalculatedValuesRef}) {
 188     push @NewLineWords, $Value;
 189   }
 190   for $ColNum (@{$ColNumsAfterNewRef}) {
 191     push @NewLineWords, $OldLineWordsRef->[$ColNum];
 192   }
 193   $NewLine = JoinWords(\@NewLineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 194   print $NewTextFileRef "$NewLine\n";
 195 }
 196 
 197 # Print out error message...
 198 sub PrintErrorMsg {
 199   my($Line, $ErrorMsg) = @_;
 200 
 201   if ($OptionsInfo{DetailLevel} >= 2 ) {
 202     print "$ErrorMsg: $Line\n";
 203   }
 204   elsif ($OptionsInfo{DetailLevel} >= 1) {
 205     print "$ErrorMsg\n";
 206   }
 207 }
 208 
 209 # Process formula columns and other information...
 210 sub RetrieveColumnsAndLabelsInfo {
 211   RetrieveFormulaColumnsInfo();
 212   RetrieveStartColumnsAndValueLabelsInfo();
 213 }
 214 
 215 # Make sure specified formula column are okay...
 216 sub RetrieveFormulaColumnsInfo {
 217   my($Index, $TextFile);
 218 
 219   @{$TextFilesInfo{FormulaColNum}} = ();
 220 
 221  FILELIST: for $Index (0 .. $#TextFilesList) {
 222     $TextFile = $TextFilesList[$Index];
 223 
 224     $TextFilesInfo{FormulaColNum}[$Index] = 0;
 225 
 226     if ($TextFilesInfo{FileOkay}[$Index]) {
 227       my($FormulaColNum, $FormulaColValid);
 228 
 229       $FormulaColNum = 0;
 230       $FormulaColValid = 0;
 231       if ($OptionsInfo{SpecifiedFormulaCol}) {
 232         if ($OptionsInfo{ColMode} =~ /^colnum$/i) {
 233           if ($OptionsInfo{SpecifiedFormulaCol} <= $TextFilesInfo{ColCount}[$Index]) {
 234             $FormulaColNum = $OptionsInfo{SpecifiedFormulaCol} - 1;
 235             $FormulaColValid = 1;
 236           }
 237         }
 238         else {
 239           if (exists($TextFilesInfo{ColLabelToNumMap}[$Index]{$OptionsInfo{SpecifiedFormulaCol}})) {
 240             $FormulaColNum =  $TextFilesInfo{ColLabelToNumMap}[$Index]{$OptionsInfo{SpecifiedFormulaCol}};
 241             $FormulaColValid = 1;
 242           }
 243         }
 244       }
 245       else {
 246         # Grab the first column with the word Formula in its label...
 247         my($ColLabel);
 248         LABEL: for $ColLabel (@{$TextFilesInfo{ColLabels}[$Index]}) {
 249           if ($ColLabel =~ /Formula/i) {
 250             $FormulaColNum =  $TextFilesInfo{ColLabelToNumMap}[$Index]{$ColLabel};
 251             $FormulaColValid = 1;
 252             last LABEL;
 253           }
 254         }
 255       }
 256       if ($FormulaColValid) {
 257         $TextFilesInfo{FormulaColNum}[$Index] = $FormulaColNum;
 258       }
 259       else {
 260         if ($OptionsInfo{SpecifiedFormulaCol}) {
 261           warn "Warning: Ignoring file $TextFile: Formula column specified, $OptionsInfo{SpecifiedFormulaCol}, using \"f --formulacol\" option doesn't exist\n";
 262         }
 263         else {
 264           warn "Warning: Ignoring file $TextFile: Column label containing the word Formula doesn't exist\n";
 265         }
 266         $TextFilesInfo{FileOkay}[$Index] = 0;
 267       }
 268     }
 269   }
 270 }
 271 
 272 # Setup starting column number for adding calculated values and
 273 # column lables to use for these values...
 274 sub RetrieveStartColumnsAndValueLabelsInfo {
 275   my($Index, $TextFile, $SpecifiedStartColNum, $StartColNum, $Label, $Value, $NewLabel, $Count, $BeforeStartColNum, $AfterStartColNum, $FirstColNum, $LastColNum, $ColNum, $Part1StartColNum, $Part1EndColNum, $Part2StartColNum, $Part2EndColNum, @Part1ColNums, @Part2ColNums);
 276 
 277   # Start column number for inserting new values...
 278   $SpecifiedStartColNum = "last";
 279   if (defined($OptionsInfo{StartCol})) {
 280     if (length($OptionsInfo{StartCol})) {
 281       $SpecifiedStartColNum = $OptionsInfo{StartCol}
 282     }
 283   }
 284 
 285   # Column labels for for new calculated values...
 286   my(%NewValueLabels) = (ElementalAnalysis => 'ElementalAnalysis', MolecularWeight => 'MolecularWeight', ExactMass => 'ExactMass');
 287   if (@{$OptionsInfo{SpecifiedValueLabels}}) {
 288     for ($Index = 0; $Index < @{$OptionsInfo{SpecifiedValueLabels}}; $Index +=2) {
 289       $Value = $OptionsInfo{SpecifiedValueLabels}[$Index];
 290       $Label = $OptionsInfo{SpecifiedValueLabels}[$Index + 1];
 291       if (exists $NewValueLabels{$Value}) {
 292         $NewValueLabels{$Value} = $Label;
 293       }
 294     }
 295   }
 296 
 297   @{$TextFilesInfo{ColNumsBeforeNew}} = ();
 298   @{$TextFilesInfo{ColNumsAfterNew}} = ();
 299   @{$TextFilesInfo{ValueLabelsMap}} = ();
 300 
 301   FILELIST: for $Index (0 .. $#TextFilesList) {
 302     $TextFile = $TextFilesList[$Index];
 303 
 304     @{$TextFilesInfo{ColNumsBeforeNew}[$Index]} = ();
 305     @{$TextFilesInfo{ColNumsAfterNew}[$Index]} = ();
 306     %{$TextFilesInfo{ValueLabelsMap}[$Index]} = ();
 307 
 308     if (!$TextFilesInfo{FileOkay}[$Index]) {
 309       next FILELIST;
 310     }
 311 
 312     if ($SpecifiedStartColNum !~ /^last$/i) {
 313       if ($OptionsInfo{ColMode} =~ /^collabel$/i) {
 314         if (exists($TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedStartColNum})) {
 315           $StartColNum = $TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedStartColNum};
 316         }
 317         else {
 318           die "Error: Invalid value $SpecifiedStartColNum specified using \"-s --startcol\" option: column name doesn't exist in  $TextFile  \n";
 319         }
 320       }
 321       else {
 322         if ($SpecifiedStartColNum > 0 && $SpecifiedStartColNum <= $TextFilesInfo{ColCount}[$Index]) {
 323           $StartColNum -= 1;
 324         }
 325         else {
 326           die "Error: Invalid value $SpecifiedStartColNum specified using \"-s --startcol\" option: column number doesn't exist in  $TextFile  \n";
 327         }
 328       }
 329     }
 330     else {
 331       $StartColNum = $TextFilesInfo{ColCount}[$Index] - 1;
 332     }
 333     # Set up columns lists for before and after the addition of calculated column values
 334     # for each text file...
 335     my($BeforeStartColNum, $AfterStartColNum, $FirstColNum, $LastColNum, $ColNum, $Part1StartColNum, $Part1EndColNum, $Part2StartColNum, $Part2EndColNum, @Part1ColNums, @Part2ColNums);
 336 
 337     $FirstColNum = 0; $LastColNum = $TextFilesInfo{ColCount}[$Index] - 1;
 338 
 339     $BeforeStartColNum = $StartColNum - 1;
 340     $AfterStartColNum = $StartColNum + 1;
 341 
 342     if ($OptionsInfo{StartColMode} =~ /^after$/i) {
 343       $Part1StartColNum = $FirstColNum; $Part1EndColNum = $StartColNum;
 344       $Part2StartColNum = $AfterStartColNum; $Part2EndColNum = $LastColNum;
 345     }
 346     else {
 347       $Part1StartColNum = $FirstColNum; $Part1EndColNum = $BeforeStartColNum;
 348       $Part2StartColNum = $StartColNum; $Part2EndColNum = $LastColNum;
 349     }
 350     @Part1ColNums = (); @Part2ColNums = ();
 351     for $ColNum (0 .. $TextFilesInfo{ColCount}[$Index]) {
 352       if ($ColNum >= $Part1StartColNum && $ColNum <= $Part1EndColNum) {
 353         push @Part1ColNums, $ColNum;
 354       }
 355     }
 356     for $ColNum (0 .. $TextFilesInfo{ColCount}[$Index]) {
 357       if ($ColNum >= $Part2StartColNum && $ColNum <= $Part2EndColNum) {
 358         push @Part2ColNums, $ColNum;
 359       }
 360     }
 361     push @{$TextFilesInfo{ColNumsBeforeNew}[$Index]}, @Part1ColNums;
 362     push @{$TextFilesInfo{ColNumsAfterNew}[$Index]}, @Part2ColNums;
 363 
 364     # Setup column labels for calculated values...
 365     for $Value (keys %NewValueLabels) {
 366       $Label = $NewValueLabels{$Value};
 367 
 368       # Make sure it doesn't already exists...
 369       $Count = 1;
 370       $NewLabel = $Label;
 371       while (exists $TextFilesInfo{ColLabelToNumMap}[$Index]{$NewLabel}) {
 372         $Count++;
 373         $NewLabel = $Label . $Count;
 374       }
 375       $TextFilesInfo{ValueLabelsMap}[$Index]{$Value} = $NewLabel;
 376     }
 377   }
 378 }
 379 
 380 # Retrieve information about input text files...
 381 sub RetrieveTextFilesInfo {
 382   my($Index, $TextFile, $FileDir, $FileName, $FileExt, $InDelim, $Line, @ColLabels, $OutFileRoot,  $OutFile, $ColNum, $ColLabel);
 383 
 384   %TextFilesInfo = ();
 385 
 386   @{$TextFilesInfo{FileOkay}} = ();
 387   @{$TextFilesInfo{ColCount}} = ();
 388   @{$TextFilesInfo{ColLabels}} = ();
 389   @{$TextFilesInfo{ColLabelToNumMap}} = ();
 390   @{$TextFilesInfo{InDelim}} = ();
 391   @{$TextFilesInfo{OutFile}} = ();
 392 
 393   FILELIST: for $Index (0 .. $#TextFilesList) {
 394     $TextFile = $TextFilesList[$Index];
 395 
 396     $TextFilesInfo{FileOkay}[$Index] = 0;
 397     $TextFilesInfo{ColCount}[$Index] = 0;
 398     $TextFilesInfo{InDelim}[$Index] = "";
 399     $TextFilesInfo{OutFile}[$Index] = "";
 400 
 401     @{$TextFilesInfo{ColLabels}[$Index]} = ();
 402     %{$TextFilesInfo{ColLabelToNumMap}[$Index]} = ();
 403 
 404     if (!(-e $TextFile)) {
 405       warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
 406       next FILELIST;
 407     }
 408     if (!CheckFileType($TextFile, "csv tsv")) {
 409       warn "Warning: Ignoring file $TextFile: It's not a csv or tsv file\n";
 410       next FILELIST;
 411     }
 412     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
 413     if ($FileExt =~ /^tsv$/i) {
 414       $InDelim = "\t";
 415     }
 416     else {
 417       $InDelim = "\,";
 418       if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
 419         warn "Warning: Ignoring file $TextFile: The value specified, $Options{indelim}, for option \"--indelim\" is not valid for csv files\n";
 420         next FILELIST;
 421       }
 422       if ($Options{indelim} =~ /^semicolon$/i) {
 423         $InDelim = "\;";
 424       }
 425     }
 426 
 427     if (!open TEXTFILE, "$TextFile") {
 428       warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
 429       next FILELIST;
 430     }
 431 
 432     $Line = GetTextLine(\*TEXTFILE);
 433     @ColLabels = quotewords($InDelim, 0, $Line);
 434     close TEXTFILE;
 435 
 436     $FileDir = ""; $FileName = ""; $FileExt = "";
 437     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
 438     $FileExt = "csv";
 439     if ($Options{outdelim} =~ /^tab$/i) {
 440       $FileExt = "tsv";
 441     }
 442     if ($Options{root} && (@TextFilesList == 1)) {
 443       my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
 444       if ($RootFileName && $RootFileExt) {
 445         $FileName = $RootFileName;
 446       }
 447       else {
 448         $FileName = $Options{root};
 449       }
 450       $OutFileRoot = $FileName;
 451     }
 452     else {
 453       $OutFileRoot = $FileName . "ElementalAnalysis";
 454     }
 455 
 456     $OutFile = $OutFileRoot . ".$FileExt";
 457     if (lc($OutFile) eq lc($TextFile)) {
 458       warn "Warning: Ignoring file $TextFile:Output file name, $OutFile, is same as input text file name, $TextFile\n";
 459       next FILELIST;
 460     }
 461     if (!$Options{overwrite}) {
 462       if (-e $OutFile) {
 463         warn "Warning: Ignoring file $TextFile: The file $OutFile already exists\n";
 464         next FILELIST;
 465       }
 466     }
 467 
 468     $TextFilesInfo{FileOkay}[$Index] = 1;
 469     $TextFilesInfo{InDelim}[$Index] = $InDelim;
 470     $TextFilesInfo{OutFile}[$Index] = "$OutFile";
 471 
 472     $TextFilesInfo{ColCount}[$Index] = @ColLabels;
 473     push @{$TextFilesInfo{ColLabels}[$Index]}, @ColLabels;
 474     for $ColNum (0 .. $#ColLabels) {
 475       $ColLabel = $ColLabels[$ColNum];
 476       $TextFilesInfo{ColLabelToNumMap}[$Index]{$ColLabel} = $ColNum;
 477     }
 478   }
 479 
 480 }
 481 
 482 # Process option values...
 483 sub ProcessOptions {
 484   %OptionsInfo = ();
 485 
 486   $OptionsInfo{Mode} = $Options{mode};
 487 
 488   $OptionsInfo{ColMode} = $Options{colmode};
 489   $OptionsInfo{StartColMode} = $Options{startcolmode};
 490 
 491   $OptionsInfo{Fast} = defined $Options{fast} ? $Options{fast} : undef;
 492 
 493   $OptionsInfo{DetailLevel} = $Options{detail};
 494   $OptionsInfo{CheckFormula} = $Options{fast} ? 0 : 1;
 495   $OptionsInfo{Precision} = $Options{precision};
 496 
 497   $OptionsInfo{InDelim} = $Options{indelim};
 498 
 499   $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
 500   $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
 501 
 502   $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
 503   $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
 504 
 505   $OptionsInfo{StartCol} = defined $Options{startcol} ? $Options{startcol} : undef;
 506 
 507   $OptionsInfo{FormulaCol} = defined $Options{formulacol} ? $Options{formulacol} : undef;
 508   $OptionsInfo{SpecifiedFormulaCol} = "";
 509 
 510   if (defined $Options{formulacol}) {
 511     $OptionsInfo{SpecifiedFormulaCol} = $Options{formulacol};
 512     if ($Options{colmode} =~ /^colnum$/i) {
 513       if (!IsPositiveInteger($OptionsInfo{SpecifiedFormulaCol})) {
 514         die "Error: Invalid value $Options{formulacol} specified using \"-f -formulacol\" option: Allowed values: > 0\n";
 515       }
 516     }
 517   }
 518 
 519   # Setup what to calculate...
 520   @{$OptionsInfo{SpecifiedCalculations}} = ();
 521   if ($Options{mode} =~ /^All$/i) {
 522     @{$OptionsInfo{SpecifiedCalculations}} = qw(ElementalAnalysis MolecularWeight ExactMass);
 523   }
 524   else {
 525     my($Mode, $ModeValue, @SpecifiedModeValues);
 526     $Mode = $Options{mode};
 527     $Mode =~ s/ //g;
 528     @SpecifiedModeValues = split /\,/, $Mode;
 529     for $ModeValue (@SpecifiedModeValues) {
 530       if ($ModeValue !~ /^(ElementalAnalysis|MolecularWeight|ExactMass)$/i) {
 531         if ($ModeValue =~ /^All$/i) {
 532           die "Error: All value for option \"-m --mode\" is not allowed with other valid values.\n";
 533         }
 534         else {
 535           die "Error: The value specified, $ModeValue, for option \"-m --mode\" is not valid. Allowed values: ElementalAnalysis, MolecularWeight, or ExactMass\n";
 536         }
 537       }
 538       push @{$OptionsInfo{SpecifiedCalculations}}, $ModeValue;
 539     }
 540   }
 541 
 542   $OptionsInfo{ValueColLabels} = defined $Options{valuecollabels} ? $Options{valuecollabels} : undef;
 543   @{$OptionsInfo{SpecifiedValueLabels}} = ();
 544 
 545   if ($Options{valuecollabels}) {
 546     my($Value, $Label, @ValueLabels);
 547     @ValueLabels = split /\,/, $Options{valuecollabels};
 548     if (@ValueLabels % 2) {
 549       die "Error: The value specified, $Options{valuecollabels}, for option \"-v --valuecollabels\" is not valid: It must contain even number of comma delimited values\n";
 550     }
 551     my($Index);
 552     for ($Index = 0; $Index < @ValueLabels; $Index +=2) {
 553       $Value = $ValueLabels[$Index];
 554       $Value =~ s/ //g;
 555       $Label = $ValueLabels[$Index + 1];
 556       if ($Value !~ /^(ElementalAnalysis|MolecularWeight|ExactMass)$/i) {
 557         die "Error: The value specified, $Value, using option \"-v --valuecollabels\" is not valid. Allowed values: ElementalAnalysis, MolecularWeight, or ExactMass\n";
 558       }
 559       push @{$OptionsInfo{SpecifiedValueLabels}}, ($Value, $Label);
 560     }
 561   }
 562 }
 563 
 564 # Setup script usage  and retrieve command line arguments specified using various options...
 565 sub SetupScriptUsage {
 566 
 567   # Retrieve all the options...
 568   %Options = ();
 569   $Options{colmode} = "colnum";
 570   $Options{detail} = 1;
 571   $Options{mode} = "All";
 572   $Options{indelim} = "comma";
 573   $Options{outdelim} = "comma";
 574   $Options{precision} = 2;
 575   $Options{quote} = "yes";
 576   $Options{startcolmode} = "after";
 577 
 578   if (!GetOptions(\%Options, "colmode|c=s", "detail|d=i", "fast", "formulacol|f=s", "help|h", "indelim=s", "mode|m=s", "outdelim=s", "overwrite|o", "precision|p=i", "quote|q=s", "root|r=s", "startcol|s=s", "startcolmode=s", "valuecollabels|v=s", "workingdir|w=s")) {
 579     die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
 580   }
 581   if ($Options{workingdir}) {
 582     if (! -d $Options{workingdir}) {
 583       die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
 584     }
 585     chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
 586   }
 587   if ($Options{colmode} !~ /^(colnum|collabel)$/i) {
 588     die "Error: The value specified, $Options{colmode}, for option \"-c --colmode\" is not valid. Allowed values: colnum or collabel\n";
 589   }
 590   if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
 591     die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
 592   }
 593   if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
 594     die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
 595   }
 596   if (!IsPositiveInteger($Options{precision})) {
 597     die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n";
 598   }
 599   if ($Options{quote} !~ /^(yes|no)$/i) {
 600     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
 601   }
 602   if ($Options{startcolmode} !~ /^(before|after)$/i) {
 603     die "Error: The value specified, $Options{quote}, for option \"--startcolmode\" is not valid. Allowed values: before or after\n";
 604   }
 605   if (!IsPositiveInteger($Options{detail})) {
 606     die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n";
 607   }
 608 }
 609