annotate mayachemtools/bin/JoinTextFiles.pl @ 9:ab29fa5c8c1f draft default tip

Uploaded
author deepakjadmin
date Thu, 15 Dec 2016 14:18:03 -0500
parents 73ae111cf86f
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
1 #!/usr/bin/perl -w
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
2 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
3 # $RCSfile: JoinTextFiles.pl,v $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
4 # $Date: 2015/02/28 20:46:20 $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
5 # $Revision: 1.32 $
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
6 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
7 # Author: Manish Sud <msud@san.rr.com>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
8 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
9 # Copyright (C) 2015 Manish Sud. All rights reserved.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
10 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
11 # This file is part of MayaChemTools.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
12 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
13 # MayaChemTools is free software; you can redistribute it and/or modify it under
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
14 # the terms of the GNU Lesser General Public License as published by the Free
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
15 # Software Foundation; either version 3 of the License, or (at your option) any
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
16 # later version.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
17 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
18 # MayaChemTools is distributed in the hope that it will be useful, but without
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
19 # any warranty; without even the implied warranty of merchantability of fitness
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
20 # for a particular purpose. See the GNU Lesser General Public License for more
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
21 # details.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
22 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
23 # You should have received a copy of the GNU Lesser General Public License
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
26 # Boston, MA, 02111-1307, USA.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
27 #
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
28
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
29 use strict;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
30 use FindBin; use lib "$FindBin::Bin/../lib";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
31 use Getopt::Long;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
32 use File::Basename;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
33 use Text::ParseWords;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
34 use Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
35 use FileUtil;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
36 use TextUtil;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
37
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
38 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
39
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
40 # Autoflush STDOUT
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
41 $| = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
42
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
43 # Starting message...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
44 $ScriptName = basename $0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
45 print "\n$ScriptName:Starting...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
46 $StartTime = new Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
47
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
48 # Get the options and setup script...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
49 SetupScriptUsage();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
50 if ($Options{help} || @ARGV < 1) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
51 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
52 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
53
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
54 my(@TextFilesList);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
55 @TextFilesList = ExpandFileNames(\@ARGV, "csv tsv");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
56
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
57 if (@TextFilesList == 1) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
58 die "Error: Specify more than one Text file.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
59 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
60
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
61 # Process options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
62 print "Processing options...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
63 my(%OptionsInfo);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
64 ProcessOptions();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
65
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
66 # Setup information about input files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
67 print "Checking input text files...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
68 my(%TextFilesInfo);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
69 RetrieveTextFilesInfo();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
70
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
71 # Join files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
72 print "\nGenerating new text file $OptionsInfo{NewTextFile}...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
73 JoinTextFiles();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
74
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
75 print "\n$ScriptName:Done...\n\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
76
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
77 $EndTime = new Benchmark;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
78 $TotalTime = timediff ($EndTime, $StartTime);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
79 print "Total time: ", timestr($TotalTime), "\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
80
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
81 ###############################################################################
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
82
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
83 # Join all valid Text files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
84 sub JoinTextFiles {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
85 my($FileIndex, $TextFile, $NewTextFile, $Line, $FirstColLabelsLine, $OutDelim, $OutQuote, $InDelim, @Words, @ColLabels);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
86
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
87 $NewTextFile = $OptionsInfo{NewTextFile};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
88
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
89 $FirstColLabelsLine = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
90
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
91 $OutDelim = $OptionsInfo{OutDelim}; $OutQuote = $OptionsInfo{OutQuote};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
92
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
93 open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
94 FILELIST: for $FileIndex (0 .. $#TextFilesList) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
95 if (!$TextFilesInfo{FileOkay}[$FileIndex]) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
96 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
97 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
98
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
99 $TextFile = $TextFilesList[$FileIndex];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
100 $InDelim = $TextFilesInfo{InDelim}[$FileIndex];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
101
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
102 print "\nProcessing file $TextFile...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
103
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
104 open TEXTFILE, "$TextFile" or die "Error: Couldn't open $TextFile: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
105
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
106 if ($OptionsInfo{Label}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
107 if ($OptionsInfo{Fast}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
108 $Line = GetTextLine(\*TEXTFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
109 if (!$FirstColLabelsLine) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
110 $FirstColLabelsLine = $Line;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
111 print NEWTEXTFILE "$FirstColLabelsLine\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
112 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
113 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
114 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
115 $Line = GetTextLine(\*TEXTFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
116 if (!$FirstColLabelsLine) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
117 @ColLabels = quotewords($InDelim, 0, $Line);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
118 $FirstColLabelsLine = JoinWords(\@ColLabels, $OutDelim, $OutQuote);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
119 print NEWTEXTFILE "$FirstColLabelsLine\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
120 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
121 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
122 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
123
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
124 while ($Line = GetTextLine(\*TEXTFILE)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
125 if (!$OptionsInfo{Fast}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
126 @Words = quotewords($InDelim, 0, $Line);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
127 $Line = JoinWords(\@Words, $OutDelim, $OutQuote);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
128 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
129 print NEWTEXTFILE "$Line\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
130 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
131 close TEXTFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
132 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
133
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
134 close NEWTEXTFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
135 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
136
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
137 # Retrieve information about Text files...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
138 sub RetrieveTextFilesInfo {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
139 my($Index, $TextFile, $InDelim, $Line, $FirstColLabelsLine, $ColLabelsLine, $FileDir, $FileName, $FileExt, @FirstColLabels, @ColLabels);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
140
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
141 %TextFilesInfo = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
142 @{$TextFilesInfo{FileOkay}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
143 @{$TextFilesInfo{InDelim}} = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
144
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
145 $FirstColLabelsLine = ''; $ColLabelsLine = '';
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
146 @FirstColLabels = (); @ColLabels = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
147
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
148 FILELIST: for $Index (0 .. $#TextFilesList) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
149 $TextFilesInfo{FileOkay}[$Index] = 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
150 $TextFilesInfo{InDelim}[$Index] = "";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
151
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
152 $TextFile = $TextFilesList[$Index];
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
153 if (!(-e $TextFile)) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
154 warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
155 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
156 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
157 if (!CheckFileType($TextFile, "csv tsv")) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
158 warn "Warning: Ignoring file $TextFile: It's not a Text file\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
159 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
160 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
161
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
162 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
163 if ($FileExt =~ /^tsv$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
164 $InDelim = "\t";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
165 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
166 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
167 $InDelim = "\,";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
168 if ($OptionsInfo{InDelim} !~ /^(comma|semicolon)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
169 warn "Warning: Ignoring file $TextFile: The value specified, $OptionsInfo{InDelim}, for option \"--indelim\" is not valid for csv files\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
170 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
171 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
172 if ($OptionsInfo{InDelim} =~ /^semicolon$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
173 $InDelim = "\;";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
174 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
175 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
176
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
177 if (! open TEXTFILE, "$TextFile") {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
178 warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
179 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
180 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
181 $Line = GetTextLine(\*TEXTFILE);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
182 close TEXTFILE;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
183
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
184 if ($OptionsInfo{Label}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
185 if (!$OptionsInfo{Fast}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
186 @ColLabels = quotewords($InDelim, 0, $Line);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
187 if ($FirstColLabelsLine) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
188 if (@ColLabels != @FirstColLabels) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
189 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";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
190 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
191 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
192 $ColLabelsLine = JoinWords(\@ColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
193 if ($ColLabelsLine ne $FirstColLabelsLine) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
194 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";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
195 next FILELIST;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
196 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
197 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
198 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
199 @FirstColLabels = @ColLabels;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
200 $FirstColLabelsLine = JoinWords(\@FirstColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
201 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
202 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
203 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
204
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
205 $TextFilesInfo{FileOkay}[$Index] = 1;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
206 $TextFilesInfo{InDelim}[$Index] = $InDelim;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
207 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
208 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
209
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
210 # Process option values...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
211 sub ProcessOptions {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
212 my($FileDir, $FileName, $FileExt, $NewTextFile, $Index);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
213
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
214 %OptionsInfo = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
215
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
216 $OptionsInfo{Fast} = $Options{fast} ? $Options{fast} : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
217
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
218 $OptionsInfo{InDelim} = $Options{indelim};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
219 $OptionsInfo{Label} = ($Options{label} =~ /^yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
220
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
221 $OptionsInfo{OutFileRoot} = $Options{root} ? $Options{root} : undef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
222 $OptionsInfo{Overwrite} = $Options{overwrite} ? $Options{overwrite} : undef;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
223
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
224 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
225 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
226
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
227 if ($Options{root}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
228 $FileDir = ""; $FileName = ""; $FileExt = "";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
229 ($FileDir, $FileName, $FileExt) = ParseFileName($Options{root});
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
230 if ($FileName && $FileExt) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
231 $NewTextFile = $FileName;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
232 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
233 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
234 $NewTextFile = $Options{root};
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
235 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
236 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
237 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
238 $FileDir = ""; $FileName = ""; $FileExt = "";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
239 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFilesList[0]);
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
240 $NewTextFile = $FileName . "1To" . @TextFilesList . "Joined";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
241 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
242
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
243 if ($Options{outdelim} =~ /^tab$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
244 $NewTextFile .= ".tsv";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
245 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
246 else {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
247 $NewTextFile .= ".csv";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
248 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
249
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
250 if (!$Options{overwrite}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
251 if (-e $NewTextFile) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
252 die "Error: The file $NewTextFile already exists.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
253 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
254 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
255
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
256 if ($Options{root}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
257 for $Index (0 .. $#TextFilesList) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
258 if (lc($NewTextFile) eq lc($TextFilesList[$Index])) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
259 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";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
260 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
261 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
262 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
263
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
264 $OptionsInfo{NewTextFile} = $NewTextFile;
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
265 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
266
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
267 # Setup script usage and retrieve command line arguments specified using various options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
268 sub SetupScriptUsage {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
269
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
270 # Retrieve all the options...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
271 %Options = ();
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
272 $Options{label} = "yes";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
273 $Options{indelim} = "comma";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
274 $Options{outdelim} = "comma";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
275 $Options{quote} = "yes";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
276 if (!GetOptions(\%Options, "fast|f", "help|h", "indelim=s", "label|l=s", "outdelim=s", "overwrite|o", "quote|q=s", "root|r=s", "workingdir|w=s")) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
277 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
278 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
279 if ($Options{workingdir}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
280 if (! -d $Options{workingdir}) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
281 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
282 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
283 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
284 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
285 if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
286 die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
287 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
288 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
289 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
290 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
291 if ($Options{quote} !~ /^(yes|no)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
292 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
293 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
294 if ($Options{label} !~ /^(yes|no)$/i) {
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
295 die "Error: The value specified, $Options{label}, for option \"-l --label\" is not valid. Allowed values: yes or no\n";
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
296 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
297 }
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
298
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
299 __END__
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
300
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
301 =head1 NAME
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
302
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
303 JoinTextFiles.pl - Join multiple CSV or TSV text files into a single text file
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
304
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
305 =head1 SYNOPSIS
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
306
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
307 JoinTextFiles.pl TextFiles...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
308
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
309 JoinTextFiles.pl [B<-f, --fast>] [B<-h, --help>] [B<--indelim> comma | semicolon]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
310 [B<-l, --label> yes | no] [B<-o, --overwrite>] [B<--outdelim> comma | tab | semicolon]
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
311 [B<-q, --quote> yes | no] [B<-r, --root> rootname] [B<-w, --workingdir> dirname] TextFiles...
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
312
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
313 =head1 DESCRIPTION
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
314
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
315 Multiple CSV or TSV I<TextFiles> are joined to generate a single text file. The
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
316 file names are separated by spaces. The valid file extensions are I<.csv> and I<.tsv>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
317 for comma/semicolon and tab delimited text files respectively. All other file names
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
318 are ignored. All the text files in a current directory can be specified by I<*.csv>,
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
319 I<*.tsv>, or the current directory name. The B<--indelim> option determines the
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
320 format of I<TextFiles>. Any file which doesn't correspond to the format indicated
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
321 by B<--indelim> option is ignored.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
322
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
323 =head1 OPTIONS
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
324
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
325 =over 4
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
326
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
327 =item B<-f, --fast>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
328
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
329 In this mode, B<--indelim> and B<-q --quote> options are ignored. The format of
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
330 input and output file(s) are assumed to be similar. And the text lines from I<TextFiles>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
331 are simply transferred to output file without any processing.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
332
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
333 =item B<-h, --help>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
334
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
335 Print this help message.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
336
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
337 =item B<--indelim> I<comma | semicolon>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
338
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
339 Input delimiter for CSV I<TextFile(s)>. Possible values: I<comma or semicolon>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
340 Default value: I<comma>. For TSV files, this option is ignored and I<tab> is used as a
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
341 delimiter.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
342
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
343 =item B<-l, --label> I<yes | no>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
344
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
345 First line contains column labels. Possible values: I<yes or no>. Default value: I<yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
346
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
347 =item B<-o, --overwrite>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
348
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
349 Overwrite existing files.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
350
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
351 =item B<--outdelim> I<comma | tab | semicolon>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
352
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
353 Output text file delimiter. Possible values: I<comma, tab, or semicolon>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
354 Default value: I<comma>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
355
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
356 =item B<-q, --quote> I<yes | no>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
357
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
358 Put quotes around column values in output text file. Possible values: I<yes or
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
359 no>. Default value: I<yes>.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
360
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
361 =item B<-r, --root> I<rootname>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
362
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
363 New text file name is generated using the root: <Root>.<Ext>. Default file
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
364 name: <FirstTextFileName>1To<Count>Joined.<Ext>. The csv, and tsv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
365 <Ext> values are used for comma/semicolon, and tab delimited text files
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
366 respectively.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
367
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
368 =item B<-w, --workingdir> I<dirname>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
369
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
370 Location of working directory. Default: current directory.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
371
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
372 =back
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
373
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
374 =head1 EXAMPLES
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
375
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
376 To join CSV text files, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
377
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
378 % JoinTextFiles.pl -o Sample1.csv Sample2.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
379 % JoinTextFiles.pl -o *.csv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
380
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
381 To join Sample*.tsv TSV text files into a NewSample.tsv file, type:
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
382
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
383 % JoinTextFiles.pl -o -r NewSample Sample*.tsv
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
384
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
385 =head1 AUTHOR
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
386
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
387 Manish Sud <msud@san.rr.com>
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
388
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
389 =head1 SEE ALSO
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
390
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
391 MergeTextFiles.pl, ModifyTextFilesFormat.pl, SplitTextFiles.pl
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
392
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
393 =head1 COPYRIGHT
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
394
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
395 Copyright (C) 2015 Manish Sud. All rights reserved.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
396
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
397 This file is part of MayaChemTools.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
398
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
399 MayaChemTools is free software; you can redistribute it and/or modify it under
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
400 the terms of the GNU Lesser General Public License as published by the Free
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
401 Software Foundation; either version 3 of the License, or (at your option)
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
402 any later version.
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
403
73ae111cf86f Uploaded
deepakjadmin
parents:
diff changeset
404 =cut