1 #!/usr/bin/perl -w
   2 #
   3 # File: JoinTextFiles.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 "$FindBin::Bin/../lib";
  28 use Getopt::Long;
  29 use File::Basename;
  30 use Text::ParseWords;
  31 use Benchmark;
  32 use FileUtil;
  33 use TextUtil;
  34 
  35 my($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 if (@TextFilesList == 1) {
  55   die "Error: Specify more than one Text file.\n";
  56 }
  57 
  58 # Process options...
  59 print "Processing options...\n";
  60 my(%OptionsInfo);
  61 ProcessOptions();
  62 
  63 # Setup information about input files...
  64 print "Checking input text files...\n";
  65 my(%TextFilesInfo);
  66 RetrieveTextFilesInfo();
  67 
  68 # Join files...
  69 print "\nGenerating new text file $OptionsInfo{NewTextFile}...\n";
  70 JoinTextFiles();
  71 
  72 print "\n$ScriptName:Done...\n\n";
  73 
  74 $EndTime = new Benchmark;
  75 $TotalTime = timediff ($EndTime, $StartTime);
  76 print "Total time: ", timestr($TotalTime), "\n";
  77 
  78 ###############################################################################
  79 
  80 # Join all valid Text files...
  81 sub JoinTextFiles {
  82   my($FileIndex, $TextFile, $NewTextFile, $Line, $FirstColLabelsLine, $OutDelim, $OutQuote, $InDelim, @Words, @ColLabels);
  83 
  84   $NewTextFile = $OptionsInfo{NewTextFile};
  85 
  86   $FirstColLabelsLine = '';
  87 
  88   $OutDelim = $OptionsInfo{OutDelim}; $OutQuote = $OptionsInfo{OutQuote};
  89 
  90   open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
  91   FILELIST: for $FileIndex (0 .. $#TextFilesList) {
  92     if (!$TextFilesInfo{FileOkay}[$FileIndex]) {
  93       next FILELIST;
  94     }
  95 
  96     $TextFile = $TextFilesList[$FileIndex];
  97     $InDelim = $TextFilesInfo{InDelim}[$FileIndex];
  98 
  99     print "\nProcessing file $TextFile...\n";
 100 
 101     open TEXTFILE, "$TextFile" or die "Error: Couldn't open $TextFile: $! \n";
 102 
 103     if ($OptionsInfo{Label}) {
 104       if ($OptionsInfo{Fast}) {
 105         $Line = GetTextLine(\*TEXTFILE);
 106         if (!$FirstColLabelsLine) {
 107           $FirstColLabelsLine = $Line;
 108           print NEWTEXTFILE "$FirstColLabelsLine\n";
 109         }
 110       }
 111       else {
 112         $Line = GetTextLine(\*TEXTFILE);
 113         if (!$FirstColLabelsLine) {
 114           @ColLabels = quotewords($InDelim, 0, $Line);
 115           $FirstColLabelsLine = JoinWords(\@ColLabels, $OutDelim, $OutQuote);
 116           print NEWTEXTFILE "$FirstColLabelsLine\n";
 117         }
 118       }
 119     }
 120 
 121     while ($Line = GetTextLine(\*TEXTFILE)) {
 122       if (!$OptionsInfo{Fast}) {
 123         @Words = quotewords($InDelim, 0, $Line);
 124         $Line = JoinWords(\@Words, $OutDelim, $OutQuote);
 125       }
 126       print NEWTEXTFILE "$Line\n";
 127     }
 128     close TEXTFILE;
 129   }
 130 
 131   close NEWTEXTFILE;
 132 }
 133 
 134 # Retrieve information about Text files...
 135 sub RetrieveTextFilesInfo {
 136   my($Index, $TextFile, $InDelim, $Line, $FirstColLabelsLine, $ColLabelsLine, $FileDir, $FileName, $FileExt, @FirstColLabels, @ColLabels);
 137 
 138   %TextFilesInfo = ();
 139   @{$TextFilesInfo{FileOkay}} = ();
 140   @{$TextFilesInfo{InDelim}} = ();
 141 
 142   $FirstColLabelsLine = ''; $ColLabelsLine = '';
 143   @FirstColLabels = (); @ColLabels = ();
 144 
 145   FILELIST: for $Index (0 .. $#TextFilesList) {
 146     $TextFilesInfo{FileOkay}[$Index] = 0;
 147     $TextFilesInfo{InDelim}[$Index] = "";
 148 
 149     $TextFile = $TextFilesList[$Index];
 150     if (!(-e $TextFile)) {
 151       warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
 152       next FILELIST;
 153     }
 154     if (!CheckFileType($TextFile, "csv tsv")) {
 155       warn "Warning: Ignoring file $TextFile: It's not a Text file\n";
 156       next FILELIST;
 157     }
 158 
 159     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
 160     if ($FileExt =~ /^tsv$/i) {
 161       $InDelim = "\t";
 162     }
 163     else {
 164       $InDelim = "\,";
 165       if ($OptionsInfo{InDelim} !~ /^(comma|semicolon)$/i) {
 166         warn "Warning: Ignoring file $TextFile: The value specified, $OptionsInfo{InDelim}, for option \"--indelim\" is not valid for csv files\n";
 167         next FILELIST;
 168       }
 169       if ($OptionsInfo{InDelim} =~ /^semicolon$/i) {
 170         $InDelim = "\;";
 171       }
 172     }
 173 
 174     if (! open TEXTFILE, "$TextFile") {
 175       warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
 176       next FILELIST;
 177     }
 178     $Line = GetTextLine(\*TEXTFILE);
 179     close TEXTFILE;
 180 
 181     if ($OptionsInfo{Label}) {
 182       if (!$OptionsInfo{Fast}) {
 183         @ColLabels = quotewords($InDelim, 0, $Line);
 184         if ($FirstColLabelsLine) {
 185           if (@ColLabels != @FirstColLabels) {
 186             warn "Warning: Ignoring file $TextFile: The number of columns in this file, ", scalar(@ColLabels), ", is different from the number of columns, ", scalar(@FirstColLabels), ", in the first valid text file. \n";
 187             next FILELIST;
 188           }
 189           $ColLabelsLine = JoinWords(\@ColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 190           if ($ColLabelsLine ne $FirstColLabelsLine) {
 191             warn "Warning: Ignoring file $TextFile: The column names in this file are different from those in first valid text file.\nColumnlabels in first valid text file: $FirstColLabelsLine \nColumnlabels in current text file: $ColLabelsLine\n";
 192             next FILELIST;
 193           }
 194         }
 195         else {
 196           @FirstColLabels = @ColLabels;
 197           $FirstColLabelsLine = JoinWords(\@FirstColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
 198         }
 199       }
 200     }
 201 
 202     $TextFilesInfo{FileOkay}[$Index] = 1;
 203     $TextFilesInfo{InDelim}[$Index] = $InDelim;
 204   }
 205 }
 206 
 207 # Process option values...
 208 sub ProcessOptions {
 209   my($FileDir, $FileName, $FileExt, $NewTextFile, $Index);
 210 
 211   %OptionsInfo = ();
 212 
 213   $OptionsInfo{Fast} = $Options{fast} ? $Options{fast} : 0;
 214 
 215   $OptionsInfo{InDelim} = $Options{indelim};
 216   $OptionsInfo{Label} = ($Options{label} =~ /^yes$/i) ? 1 : 0;
 217 
 218   $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : undef;
 219   $OptionsInfo{Overwrite} = $Options{overwrite} ? $Options{overwrite} : undef;
 220 
 221   $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
 222   $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
 223 
 224   if ($Options{root}) {
 225     $FileDir = ""; $FileName = ""; $FileExt = "";
 226     ($FileDir, $FileName, $FileExt) = ParseFileName($Options{root});
 227     if ($FileName && $FileExt) {
 228       $NewTextFile = $FileName;
 229     }
 230     else {
 231       $NewTextFile =  $Options{root};
 232     }
 233   }
 234   else {
 235     $FileDir = ""; $FileName = ""; $FileExt = "";
 236     ($FileDir, $FileName, $FileExt) = ParseFileName($TextFilesList[0]);
 237     $NewTextFile = $FileName . "1To" . @TextFilesList . "Joined";
 238   }
 239 
 240   if ($Options{outdelim} =~ /^tab$/i) {
 241     $NewTextFile .= ".tsv";
 242   }
 243   else {
 244     $NewTextFile .= ".csv";
 245   }
 246 
 247   if (!$Options{overwrite}) {
 248     if (-e $NewTextFile) {
 249       die "Error: The file $NewTextFile already exists.\n";
 250     }
 251   }
 252 
 253   if ($Options{root}) {
 254     for $Index (0 .. $#TextFilesList) {
 255       if (lc($NewTextFile) eq lc($TextFilesList[$Index])) {
 256         die "Error: Output filename, $NewTextFile, is similar to a input file name.\nSpecify a different name using \"-r --root\" option or use default name.\n";
 257       }
 258     }
 259   }
 260 
 261   $OptionsInfo{NewTextFile} = $NewTextFile;
 262 }
 263 
 264 # Setup script usage  and retrieve command line arguments specified using various options...
 265 sub SetupScriptUsage {
 266 
 267   # Retrieve all the options...
 268   %Options = ();
 269   $Options{label} = "yes";
 270   $Options{indelim} = "comma";
 271   $Options{outdelim} = "comma";
 272   $Options{quote} = "yes";
 273   if (!GetOptions(\%Options, "fast|f", "help|h", "indelim=s", "label|l=s", "outdelim=s", "overwrite|o", "quote|q=s", "root|r=s", "workingdir|w=s")) {
 274     die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
 275   }
 276   if ($Options{workingdir}) {
 277     if (! -d $Options{workingdir}) {
 278       die "Error: The value specified, $Options{workingdir},  for option \"-w --workingdir\" is not a directory name.\n";
 279     }
 280     chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
 281   }
 282   if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
 283     die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
 284   }
 285   if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
 286     die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
 287   }
 288   if ($Options{quote} !~ /^(yes|no)$/i) {
 289     die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
 290   }
 291   if ($Options{label} !~ /^(yes|no)$/i) {
 292     die "Error: The value specified, $Options{label}, for option \"-l --label\" is not valid. Allowed values: yes or no\n";
 293   }
 294 }
 295