MayaChemTools

   1 #!/usr/bin/perl -w
   2 #
   3 # File: SortTextFiles.pl
   4 # Author: Manish Sud <msud@san.rr.com>
   5 #
   6 # Copyright (C) 2019 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 "$FindBin::Bin/../lib";
  28 use Getopt::Long;
  29 use File::Basename;
  30 use Text::ParseWords;
  31 use Benchmark;
  32 use FileUtil;
  33 use TextUtil;
  34 
  35 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
  36 
  37 # Autoflush STDOUT
  38 $| = 1;
  39 
  40 # Starting message...
  41 $ScriptName = basename($0);
  42 print "\n$ScriptName: Starting...\n\n";
  43 $StartTime = new Benchmark;
  44 
  45 # Get the options and setup script...
  46 SetupScriptUsage();
  47 if ($Options{help} || @ARGV < 1) {
  48   die GetUsageFromPod("$FindBin::Bin/$ScriptName");
  49 }
  50 
  51 my(@TextFilesList);
  52 @TextFilesList = ExpandFileNames(\@ARGV, "csv tsv");
  53 
  54 print "Processing options...\n";
  55 my(%OptionsInfo);
  56 ProcessOptions();
  57 
  58 print "Checking input text file(s)...\n";
  59 my(%TextFilesInfo);
  60 RetrieveTextFilesInfo();
  61 ProcessColumnsInfo();
  62 
  63 # Generate output files...
  64 my($FileIndex);
  65 if (@TextFilesList > 1) {
  66   print "\nProcessing text files...\n";
  67 }
  68 for $FileIndex (0 .. $#TextFilesList) {
  69   if ($TextFilesInfo{FileOkay}[$FileIndex]) {
  70     print "\nProcessing file $TextFilesList[$FileIndex]...\n";
  71     SortTextFile($FileIndex);
  72   }
  73 }
  74 print "\n$ScriptName:Done...\n\n";
  75 
  76 $EndTime = new Benchmark;
  77 $TotalTime = timediff ($EndTime, $StartTime);
  78 print "Total time: ", timestr($TotalTime), "\n";
  79 
  80 ###############################################################################
  81 
  82 # Sort it out...
  83 sub SortTextFile {
  84   my($Index) = @_;
  85   my($TextFile, $NewTextFile, $KeyCol, $Line, $KeyColValue, $InDelim, @ColLabels, @LineWords);
  86 
  87   $TextFile = $TextFilesList[$Index];
  88   $InDelim = $TextFilesInfo{InDelim}[$Index];
  89   $NewTextFile = $TextFilesInfo{OutFile}[$Index];
  90   $KeyCol = $TextFilesInfo{KeyColNum}[$Index];
  91   @ColLabels = @{$TextFilesInfo{ColLabels}[$Index]};
  92 
  93   print "Generating new Text file $NewTextFile...\n";
  94   open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
  95   open TEXTFILE, "$TextFile" or die "Error: Can't open $TextFile: $! \n";
  96 
  97   # Skip over column labels from old file...
  98   $Line = GetTextLine(\*TEXTFILE);
  99 
 100   # Add column lablels in new file...
 101   $Line = JoinWords(\@ColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 102   print NEWTEXTFILE "$Line\n";
 103 
 104   # Go over all rows and store the lines using key value as hash...
 105   my(%KeyToLinesMap, @InvalidDataLines, $LineCount);
 106 
 107   %KeyToLinesMap = ();
 108   @InvalidDataLines = ();
 109   $LineCount = 1;
 110   TEXTLINE: while ($Line = GetTextLine(\*TEXTFILE)) {
 111     @LineWords = quotewords($InDelim, 0, $Line);
 112     $LineCount++;
 113     if ($KeyCol < @LineWords) {
 114       $KeyColValue = $LineWords[$KeyCol];
 115       if (!IsNotEmpty($KeyColValue)) {
 116         $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 117         push @InvalidDataLines, $Line;
 118         if ($OptionsInfo{DetailLevel} >= 3 ) {
 119           print "Ignoring line $LineCount: Contains empty value for key column $ColLabels[$KeyCol]: $Line\n";
 120         }
 121         elsif ($OptionsInfo{DetailLevel} >= 2) {
 122           print "Ignoring line $LineCount: Contains empty value for key column $ColLabels[$KeyCol]...\n";
 123         }
 124         next TEXTLINE;
 125       }
 126       if ($OptionsInfo{KeyData} =~ /^numeric$/i) {
 127         if (!IsFloat($KeyColValue)) {
 128           $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 129           push @InvalidDataLines, $Line;
 130           if ($OptionsInfo{DetailLevel} >= 3 ) {
 131             print "Line number $LineCount: Contains non-numerical value for key column $ColLabels[$KeyCol]: $Line\n";
 132           }
 133           elsif ($OptionsInfo{DetailLevel} >= 2) {
 134             print "Line number $LineCount: Contains non-numerical value for key column $ColLabels[$KeyCol]...\n";
 135           }
 136           next TEXTLINE;
 137         }
 138       }
 139       if (exists($KeyToLinesMap{$KeyColValue})) {
 140         # Append to existing line...
 141         $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 142         $KeyToLinesMap{$KeyColValue} .= "\n" . $Line;
 143       }
 144       else {
 145         $Line = JoinWords(\@LineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 146         $KeyToLinesMap{$KeyColValue} = $Line;
 147       }
 148     }
 149   }
 150   if ($OptionsInfo{Sort} =~ /^ascending$/i) {
 151     if ($OptionsInfo{KeyData} =~ /^alphanumeric$/i) {
 152       for $KeyColValue (sort { lc($a) cmp lc($b) } keys %KeyToLinesMap ) {
 153         print NEWTEXTFILE "$KeyToLinesMap{$KeyColValue}\n";
 154       }
 155     }
 156     else {
 157       for $KeyColValue (sort { $a <=> $b } keys %KeyToLinesMap ) {
 158         print NEWTEXTFILE "$KeyToLinesMap{$KeyColValue}\n";
 159       }
 160     }
 161   }
 162   else {
 163     if ($OptionsInfo{KeyData} =~ /^alphanumeric$/i) {
 164       for $KeyColValue (sort { lc($b) cmp lc($a) } keys %KeyToLinesMap ) {
 165         print NEWTEXTFILE "$KeyToLinesMap{$KeyColValue}\n";
 166       }
 167     }
 168     else {
 169       for $KeyColValue (sort { $b <=> $a } keys %KeyToLinesMap ) {
 170         print NEWTEXTFILE "$KeyToLinesMap{$KeyColValue}\n";
 171       }
 172     }
 173   }
 174   # Write out the lines with invalid data...
 175   if (@InvalidDataLines) {
 176     print "Placing ", scalar(@InvalidDataLines)," line(s) with invalid column key data at the end...\n";
 177     for $Line (@InvalidDataLines) {
 178       print NEWTEXTFILE "$Line\n";
 179     }
 180   }
 181   close NEWTEXTFILE;
 182   close TEXTFILE;
 183 
 184 }
 185 
 186 # Retrieve information about input text files...
 187 sub RetrieveTextFilesInfo {
 188   my($Index, $TextFile, $FileDir, $FileName, $FileExt, $InDelim, $Line, @ColLabels, $OutFileRoot,  $OutFile, $ColNum, $ColLabel);
 189 
 190   %TextFilesInfo = ();
 191 
 192   @{$TextFilesInfo{FileOkay}} = ();
 193   @{$TextFilesInfo{ColCount}} = ();
 194   @{$TextFilesInfo{ColLabels}} = ();
 195   @{$TextFilesInfo{ColLabelToNumMap}} = ();
 196   @{$TextFilesInfo{InDelim}} = ();
 197   @{$TextFilesInfo{OutFile}} = ();
 198 
 199   FILELIST: for $Index (0 .. $#TextFilesList) {
 200     $TextFile = $TextFilesList[$Index];
 201 
 202     $TextFilesInfo{FileOkay}[$Index] = 0;
 203     $TextFilesInfo{ColCount}[$Index] = 0;
 204     $TextFilesInfo{InDelim}[$Index] = "";
 205     $TextFilesInfo{OutFile}[$Index] = "";
 206     @{$TextFilesInfo{ColLabels}[$Index]} = ();
 207     %{$TextFilesInfo{ColLabelToNumMap}[$Index]} = ();
 208 
 209     if (!(-e $TextFile)) {
 210       warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
 211       next FILELIST;
 212     }
 213     if (!CheckFileType($TextFile, "csv tsv")) {
 214       warn "Warning: Ignoring file $TextFile: It's not a csv or tsv file\n";
 215       next FILELIST;
 216     }
 217     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
 218     if ($FileExt =~ /^tsv$/i) {
 219       $InDelim = "\t";
 220     }
 221     else {
 222       $InDelim = "\,";
 223       if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
 224         warn "Warning: Ignoring file $TextFile: The value specified, $Options{indelim}, for option \"--indelim\" is not valid for csv files\n";
 225         next FILELIST;
 226       }
 227       if ($Options{indelim} =~ /^semicolon$/i) {
 228         $InDelim = "\;";
 229       }
 230     }
 231 
 232     if (!open TEXTFILE, "$TextFile") {
 233       warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
 234       next FILELIST;
 235     }
 236 
 237     $Line = GetTextLine(\*TEXTFILE);
 238     @ColLabels = quotewords($InDelim, 0, $Line);
 239     close TEXTFILE;
 240 
 241     $FileDir = ""; $FileName = ""; $FileExt = "";
 242     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
 243     $FileExt = "csv";
 244     if ($Options{outdelim} =~ /^tab$/i) {
 245       $FileExt = "tsv";
 246     }
 247     if ($Options{root} && (@TextFilesList == 1)) {
 248       my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
 249       if ($RootFileName && $RootFileExt) {
 250         $FileName = $RootFileName;
 251       }
 252       else {
 253         $FileName = $Options{root};
 254       }
 255       $OutFileRoot = $FileName;
 256     }
 257     else {
 258       $OutFileRoot = $FileName . "SortedByColumn";
 259     }
 260 
 261     $OutFile = $OutFileRoot . ".$FileExt";
 262     if (lc($OutFile) eq lc($TextFile)) {
 263       warn "Warning: Ignoring file $TextFile:Output file name, $OutFile, is same as input text file name, $TextFile\n";
 264       next FILELIST;
 265     }
 266     if (!$Options{overwrite}) {
 267       if (-e $OutFile) {
 268         warn "Warning: Ignoring file $TextFile: The file $OutFile already exists\n";
 269         next FILELIST;
 270       }
 271     }
 272 
 273     $TextFilesInfo{FileOkay}[$Index] = 1;
 274     $TextFilesInfo{InDelim}[$Index] = $InDelim;
 275     $TextFilesInfo{OutFile}[$Index] = "$OutFile";
 276 
 277     $TextFilesInfo{ColCount}[$Index] = @ColLabels;
 278     push @{$TextFilesInfo{ColLabels}[$Index]}, @ColLabels;
 279     for $ColNum (0 .. $#ColLabels) {
 280       $ColLabel = $ColLabels[$ColNum];
 281       $TextFilesInfo{ColLabelToNumMap}[$Index]{$ColLabel} = $ColNum;
 282     }
 283   }
 284 
 285 }
 286 
 287 # Make sure specified key column are okay...
 288 sub ProcessColumnsInfo {
 289   my($Index, $TextFile, $SpecifiedKeyCol);
 290 
 291   @{$TextFilesInfo{KeyColNum}} = ();
 292 
 293   $SpecifiedKeyCol = $OptionsInfo{SpecifiedKeyCol};
 294 
 295   FILELIST: for $Index (0 .. $#TextFilesList) {
 296     $TextFile = $TextFilesList[$Index];
 297     $TextFilesInfo{KeyColNum}[$Index] = 0;
 298 
 299     if ($TextFilesInfo{FileOkay}[$Index]) {
 300       my($KeyColNum, $KeyColValid);
 301 
 302       $KeyColNum = 0;
 303       $KeyColValid = 1;
 304       if ($SpecifiedKeyCol) {
 305         if ($OptionsInfo{Mode} =~ /^colnum$/i) {
 306           if ($SpecifiedKeyCol <= $TextFilesInfo{ColCount}[$Index]) {
 307             $KeyColNum = $SpecifiedKeyCol - 1;
 308           }
 309           else {
 310             $KeyColValid = 0;
 311           }
 312         }
 313         else {
 314           if (exists($TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedKeyCol})) {
 315             $KeyColNum =  $TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedKeyCol};
 316           }
 317           else {
 318             $KeyColValid = 0;
 319           }
 320         }
 321       }
 322       if ($KeyColValid) {
 323         $TextFilesInfo{KeyColNum}[$Index] = $KeyColNum;
 324       }
 325       else {
 326         warn "Warning: Ignoring file $TextFile: Column key specified, $SpecifiedKeyCol, using \"k --key\" option doesn't exist\n";
 327         $TextFilesInfo{FileOkay}[$Index] = 0;
 328       }
 329     }
 330   }
 331 }
 332 
 333 # Process option values...
 334 sub ProcessOptions {
 335   %OptionsInfo = ();
 336 
 337   $OptionsInfo{Mode} = $Options{mode};
 338 
 339   $OptionsInfo{DetailLevel} = $Options{detail};
 340 
 341   $OptionsInfo{Sort} = $Options{sort};
 342   $OptionsInfo{KeyData} = $Options{keydata};
 343 
 344   $OptionsInfo{InDelim} = $Options{indelim};
 345 
 346   $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
 347   $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
 348 
 349   $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
 350   $OptionsInfo{Root} = defined $Options{root} ? $Options{root} : undef;
 351 
 352   $OptionsInfo{Key} = defined $Options{key} ? $Options{key} : undef;
 353   $OptionsInfo{SpecifiedKeyCol} = "";
 354 
 355   if (defined $Options{key}) {
 356     $OptionsInfo{SpecifiedKeyCol} = $Options{key};
 357     if ($Options{mode} =~ /^colnum$/i) {
 358       if (!IsPositiveInteger($OptionsInfo{SpecifiedKeyCol})) {
 359         die "Error: Invalid value $Options{key} specified using \"-k --key\" option: Allowed values: > 0\n";
 360       }
 361     }
 362   }
 363 }
 364 
 365 # Setup script usage  and retrieve command line arguments specified using various options...
 366 sub SetupScriptUsage {
 367   %Options = ();
 368 
 369   $Options{detail} = 1;
 370   $Options{mode} = "colnum";
 371   $Options{sort} = "ascending";
 372   $Options{keydata} = "numeric";
 373   $Options{indelim} = "comma";
 374   $Options{outdelim} = "comma";
 375   $Options{quote} = "yes";
 376   if (!GetOptions(\%Options, "detail|d=i", "help|h", "indelim=s", "key|k=s", "keydata=s", "mode|m=s", "outdelim=s", "overwrite|o", "quote|q=s", "root|r=s", "sort|s=s", "workingdir|w=s")) {
 377     die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
 378   }
 379   if ($Options{workingdir}) {
 380     if (! -d $Options{workingdir}) {
 381       die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
 382     }
 383     chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
 384   }
 385   if ($Options{mode} !~ /^(colnum|collabel)$/i) {
 386     die "Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: colnum or collabel\n";
 387   }
 388   if ($Options{keydata} !~ /^(numeric|alphanumeric)$/i) {
 389     die "Error: The value specified, $Options{keydata}, for option \"--keydata\" is not valid. Allowed values: numeric or alphanumeric\n";
 390   }
 391   if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
 392     die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
 393   }
 394   if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
 395     die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
 396   }
 397   if ($Options{quote} !~ /^(yes|no)$/i) {
 398     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
 399   }
 400   if ($Options{sort} !~ /^(ascending|descending)$/i) {
 401     die "Error: The value specified, $Options{sort}, for option \"-s --sort\" is not valid. Allowed values: ascending or descending\n";
 402   }
 403   if (!IsPositiveInteger($Options{detail})) {
 404     die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n";
 405   }
 406 }
 407