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