|
0
|
1 #!/usr/bin/perl -w
|
|
|
2 #
|
|
|
3 # $RCSfile: TextFilesToSDFiles.pl,v $
|
|
|
4 # $Date: 2015/02/28 20:46:21 $
|
|
|
5 # $Revision: 1.25 $
|
|
|
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 "$FindBin::Bin/../lib";
|
|
|
31 use Getopt::Long;
|
|
|
32 use File::Basename;
|
|
|
33 use Text::ParseWords;
|
|
|
34 use Benchmark;
|
|
|
35 use FileUtil;
|
|
|
36 use TextUtil;
|
|
|
37 use SDFileUtil;
|
|
|
38
|
|
|
39 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
|
|
|
40
|
|
|
41 # Autoflush STDOUT
|
|
|
42 $| = 1;
|
|
|
43
|
|
|
44 # Starting message...
|
|
|
45 $ScriptName = basename $0;
|
|
|
46 print "\n$ScriptName:Starting...\n\n";
|
|
|
47 $StartTime = new Benchmark;
|
|
|
48
|
|
|
49 # Get the options and setup script...
|
|
|
50 SetupScriptUsage();
|
|
|
51 if ($Options{help} || @ARGV < 1) {
|
|
|
52 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
|
|
|
53 }
|
|
|
54
|
|
|
55 my(@TextFilesList);
|
|
|
56 @TextFilesList = ExpandFileNames(\@ARGV, "csv tsv");
|
|
|
57
|
|
|
58 print "Processing options...\n";
|
|
|
59 my(%OptionsInfo);
|
|
|
60 ProcessOptions();
|
|
|
61
|
|
|
62 print "Checking input text file(s)...\n";
|
|
|
63 my(%TextFilesInfo);
|
|
|
64 RetrieveTextFilesInfo();
|
|
|
65
|
|
|
66 # Generate output files...
|
|
|
67 my($FileIndex);
|
|
|
68 if (@TextFilesList > 1) {
|
|
|
69 print "\nProcessing text files...\n";
|
|
|
70 }
|
|
|
71 for $FileIndex (0 .. $#TextFilesList) {
|
|
|
72 if ($TextFilesInfo{FileOkay}[$FileIndex]) {
|
|
|
73 print "\nProcessing file $TextFilesList[$FileIndex]...\n";
|
|
|
74 ConvertTextFile($FileIndex);
|
|
|
75 }
|
|
|
76 }
|
|
|
77 print "\n$ScriptName:Done...\n\n";
|
|
|
78
|
|
|
79 $EndTime = new Benchmark;
|
|
|
80 $TotalTime = timediff ($EndTime, $StartTime);
|
|
|
81 print "Total time: ", timestr($TotalTime), "\n";
|
|
|
82
|
|
|
83 ###############################################################################
|
|
|
84
|
|
|
85 # Convert text file to SD file...
|
|
|
86 sub ConvertTextFile {
|
|
|
87 my($Index) = @_;
|
|
|
88 my($TextFile, $SDFile, $Line, $InDelim, $Label, $Value, $ColIndex, $ColCount, @ColLabels, @LineWords);
|
|
|
89
|
|
|
90 $TextFile = $TextFilesList[$Index];
|
|
|
91 $InDelim = $TextFilesInfo{InDelim}[$Index];
|
|
|
92 $SDFile = $TextFilesInfo{OutSDFile}[$Index];
|
|
|
93 @ColLabels = @{$TextFilesInfo{ColLabels}[$Index]};
|
|
|
94 $ColCount = @ColLabels;
|
|
|
95
|
|
|
96 print "Generating SD file $SDFile...\n";
|
|
|
97 open SDFILE, ">$SDFile" or die "Error: Couldn't open $SDFile: $! \n";
|
|
|
98 open TEXTFILE, "$TextFile" or die "Error: Can't open $TextFile: $! \n";
|
|
|
99 if ($OptionsInfo{ColLabelsPresent}) {
|
|
|
100 # Skip over column labels from old file...
|
|
|
101 $Line = GetTextLine(\*TEXTFILE);
|
|
|
102 }
|
|
|
103 my($Date) = GenerateMiscLineDateStamp();
|
|
|
104 while ($Line = GetTextLine(\*TEXTFILE)) {
|
|
|
105 @LineWords = quotewords($InDelim, 0, $Line);
|
|
|
106
|
|
|
107 # Write out empty CTAB block...
|
|
|
108 print SDFILE GenerateEmptyCtabBlockLines($Date), "\n";
|
|
|
109
|
|
|
110 # Write out data fields and values...
|
|
|
111 for $ColIndex (0 .. $#LineWords) {
|
|
|
112 if ($ColIndex < $ColCount) {
|
|
|
113 $Label = $ColLabels[$ColIndex];
|
|
|
114 $Value = $LineWords[$ColIndex];
|
|
|
115 print SDFILE "> <$Label>\n$Value\n\n";
|
|
|
116 }
|
|
|
117 }
|
|
|
118 print SDFILE "\$\$\$\$\n";
|
|
|
119 }
|
|
|
120 close SDFILE;
|
|
|
121 close TEXTFILE;
|
|
|
122 }
|
|
|
123
|
|
|
124 # Retrieve information about input text files...
|
|
|
125 sub RetrieveTextFilesInfo {
|
|
|
126 my($Index, $TextFile, $FileDir, $FileName, $FileExt, $InDelim, $Line, @LineWords, @ColLabels, $OutFileRoot, $OutFile, $ColNum, $ColLabel);
|
|
|
127
|
|
|
128 %TextFilesInfo = ();
|
|
|
129
|
|
|
130 @{$TextFilesInfo{FileOkay}} = ();
|
|
|
131 @{$TextFilesInfo{ColCount}} = ();
|
|
|
132 @{$TextFilesInfo{ColLabels}} = ();
|
|
|
133 @{$TextFilesInfo{InDelim}} = ();
|
|
|
134 @{$TextFilesInfo{OutSDFile}} = ();
|
|
|
135
|
|
|
136
|
|
|
137 FILELIST: for $Index (0 .. $#TextFilesList) {
|
|
|
138 $TextFile = $TextFilesList[$Index];
|
|
|
139
|
|
|
140 $TextFilesInfo{FileOkay}[$Index] = 0;
|
|
|
141 $TextFilesInfo{ColCount}[$Index] = 0;
|
|
|
142 $TextFilesInfo{InDelim}[$Index] = "";
|
|
|
143 $TextFilesInfo{OutSDFile}[$Index] = "";
|
|
|
144
|
|
|
145 @{$TextFilesInfo{ColLabels}[$Index]} = ();
|
|
|
146
|
|
|
147 if (!(-e $TextFile)) {
|
|
|
148 warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
|
|
|
149 next FILELIST;
|
|
|
150 }
|
|
|
151 if (!CheckFileType($TextFile, "csv tsv")) {
|
|
|
152 warn "Warning: Ignoring file $TextFile: It's not a csv or tsv file\n";
|
|
|
153 next FILELIST;
|
|
|
154 }
|
|
|
155 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
|
|
|
156 if ($FileExt =~ /^tsv$/i) {
|
|
|
157 $InDelim = "\t";
|
|
|
158 }
|
|
|
159 else {
|
|
|
160 $InDelim = "\,";
|
|
|
161 if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
|
|
|
162 warn "Warning: Ignoring file $TextFile: The value specified, $Options{indelim}, for option \"--indelim\" is not valid for csv files\n";
|
|
|
163 next FILELIST;
|
|
|
164 }
|
|
|
165 if ($Options{indelim} =~ /^semicolon$/i) {
|
|
|
166 $InDelim = "\;";
|
|
|
167 }
|
|
|
168 }
|
|
|
169 if (!open TEXTFILE, "$TextFile") {
|
|
|
170 warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
|
|
|
171 next FILELIST;
|
|
|
172 }
|
|
|
173 $Line = GetTextLine(\*TEXTFILE);
|
|
|
174 @LineWords = quotewords($InDelim, 0, $Line);
|
|
|
175 @ColLabels = ();
|
|
|
176 if ($OptionsInfo{ColLabelsPresent}) {
|
|
|
177 push @ColLabels, @LineWords;
|
|
|
178 }
|
|
|
179 else {
|
|
|
180 for $ColNum (1 .. @LineWords) {
|
|
|
181 $ColLabel = "Column${ColNum}Data";
|
|
|
182 push @ColLabels, $ColLabel;
|
|
|
183 }
|
|
|
184 }
|
|
|
185 close TEXTFILE;
|
|
|
186
|
|
|
187 $FileDir = ""; $FileName = ""; $FileExt = "";
|
|
|
188 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
|
|
|
189 if ($Options{root} && (@TextFilesList == 1)) {
|
|
|
190 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
|
|
|
191 if ($RootFileName && $RootFileExt) {
|
|
|
192 $FileName = $RootFileName;
|
|
|
193 }
|
|
|
194 else {
|
|
|
195 $FileName = $Options{root};
|
|
|
196 }
|
|
|
197 $OutFileRoot = $FileName;
|
|
|
198 }
|
|
|
199 else {
|
|
|
200 $OutFileRoot = "${FileName}WithNoStrData";
|
|
|
201 }
|
|
|
202
|
|
|
203 $OutFile = "${OutFileRoot}.sdf";
|
|
|
204 if (!$Options{overwrite}) {
|
|
|
205 if (-e $OutFile) {
|
|
|
206 warn "Warning: Ignoring file $TextFile: The file $OutFile already exists\n";
|
|
|
207 next FILELIST;
|
|
|
208 }
|
|
|
209 }
|
|
|
210 $TextFilesInfo{FileOkay}[$Index] = 1;
|
|
|
211 $TextFilesInfo{InDelim}[$Index] = $InDelim;
|
|
|
212 $TextFilesInfo{OutSDFile}[$Index] = "$OutFile";
|
|
|
213
|
|
|
214 $TextFilesInfo{ColCount}[$Index] = @ColLabels;
|
|
|
215 push @{$TextFilesInfo{ColLabels}[$Index]}, @ColLabels;
|
|
|
216 }
|
|
|
217 }
|
|
|
218
|
|
|
219 # Process option values...
|
|
|
220 sub ProcessOptions {
|
|
|
221 %OptionsInfo = ();
|
|
|
222
|
|
|
223 $OptionsInfo{Label} = $Options{label};
|
|
|
224 $OptionsInfo{ColLabelsPresent} = ($Options{label} =~ /^yes$/i) ? 1 : 0;
|
|
|
225
|
|
|
226 $OptionsInfo{InDelim} = $Options{indelim};
|
|
|
227 $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
|
|
|
228
|
|
|
229 $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
|
|
|
230
|
|
|
231 }
|
|
|
232
|
|
|
233 # Setup script usage and retrieve command line arguments specified using various options...
|
|
|
234 sub SetupScriptUsage {
|
|
|
235
|
|
|
236 # Retrieve all the options...
|
|
|
237 %Options = ();
|
|
|
238 $Options{label} = "yes";
|
|
|
239 $Options{indelim} = "comma";
|
|
|
240 if (!GetOptions(\%Options, "help|h", "indelim=s", "label|l=s", "overwrite|o", "root|r=s", "workingdir|w=s")) {
|
|
|
241 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
|
|
|
242 }
|
|
|
243 if ($Options{workingdir}) {
|
|
|
244 if (! -d $Options{workingdir}) {
|
|
|
245 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
|
|
|
246 }
|
|
|
247 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
|
|
|
248 }
|
|
|
249 if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
|
|
|
250 die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
|
|
|
251 }
|
|
|
252 if ($Options{label} !~ /^(yes|no)$/i) {
|
|
|
253 die "Error: The value specified, $Options{label}, for option \"-l --label\" is not valid. Allowed values: yes or no\n";
|
|
|
254 }
|
|
|
255 }
|
|
|
256
|
|
|
257 __END__
|
|
|
258
|
|
|
259 =head1 NAME
|
|
|
260
|
|
|
261 TextFilesToSDFiles.pl - Generate SD files from CSV or TSV TextFile(s)
|
|
|
262
|
|
|
263 =head1 SYNOPSIS
|
|
|
264
|
|
|
265 TextFilesToSDFiles.pl TextFile(s)...
|
|
|
266
|
|
|
267 TextFilesToSDFiles.pl [B<-h, --help>] [B<--indelim> comma | semicolon] [B<-l, --label> yes | no]
|
|
|
268 [B<-o, --overwrite>] [B<-r, --root> rootname] [B<-w, --workingdir> dirname] TextFile(s)...
|
|
|
269
|
|
|
270 =head1 DESCRIPTION
|
|
|
271
|
|
|
272 Generate SD files from CSV or TSV I<TextFile(s)>. The new SD files contain no structure data
|
|
|
273 as indicated by empty structure data block; Data fields and values in SD files are generated
|
|
|
274 using I<TextFile(s)> column labels and corresponding data values.
|
|
|
275
|
|
|
276 Multiple I<TextFile(s)> names are separated by space. The valid file extensions are
|
|
|
277 I<.csv> and I<.tsv> for comma/semicolon and tab delimited text files respectively.
|
|
|
278 All other file names are ignored. All the text files in a current directory can be specified
|
|
|
279 by I<*.csv>, I<*.tsv>, or the current directory name. The B<--indelim> option
|
|
|
280 determines the format of I<TextFile(s)>. Any file which doesn't correspond to
|
|
|
281 the format indicated by B<--indelim> option is ignored.
|
|
|
282
|
|
|
283 =head1 OPTIONS
|
|
|
284
|
|
|
285 =over 4
|
|
|
286
|
|
|
287 =item B<-h, --help>
|
|
|
288
|
|
|
289 Print this help message.
|
|
|
290
|
|
|
291 =item B<--indelim> I<comma | semicolon>
|
|
|
292
|
|
|
293 Input delimiter for CSV I<TextFile(s)>. Possible values: I<comma or semicolon>.
|
|
|
294 Default value: I<comma>. For TSV files, this option is ignored and I<tab> is used as a
|
|
|
295 delimiter.
|
|
|
296
|
|
|
297 =item B<-l, --label> I<yes | no>
|
|
|
298
|
|
|
299 First line contains column labels. Possible values: I<yes or no>. Default value: I<yes>.
|
|
|
300 Column labels are used to create SD data field labels; otherwise, data field labels look
|
|
|
301 like Column<colnumber>Data.
|
|
|
302
|
|
|
303 =item B<-o, --overwrite>
|
|
|
304
|
|
|
305 Overwrite existing files.
|
|
|
306
|
|
|
307 =item B<-r, --root> I<rootname>
|
|
|
308
|
|
|
309 New SD file names are generated using the root: <Root>.sdf. Default new file names:
|
|
|
310 <TextFileName>WithNoStrData.sdf. This option is ignored for multiple input files.
|
|
|
311
|
|
|
312 =item B<-w, --workingdir> I<dirname>
|
|
|
313
|
|
|
314 Location of working directory. Default: current directory.
|
|
|
315
|
|
|
316 =back
|
|
|
317
|
|
|
318 =head1 EXAMPLES
|
|
|
319
|
|
|
320 To generate NewSample1.sdf file from Sample1.csv file, type:
|
|
|
321
|
|
|
322 % TextFilesToSDFiles.pl -o -r NewSample1 Sample1.csv
|
|
|
323
|
|
|
324 To generate NewSample1.sdf file from Sample1.tsv file which doesn't
|
|
|
325 contain column labels line, type:
|
|
|
326
|
|
|
327 % TextFilesToSDFiles.pl --label no -o -r NewSample1 Sample1.tsv
|
|
|
328
|
|
|
329 =head1 AUTHOR
|
|
|
330
|
|
|
331 Manish Sud <msud@san.rr.com>
|
|
|
332
|
|
|
333 =head1 SEE ALSO
|
|
|
334
|
|
|
335 JoinTextFiles.pl, MergeTextFiles.pl, ModifySDFilesDataFields.pl, ModifyTextFilesFormat.pl
|
|
|
336
|
|
|
337 =head1 COPYRIGHT
|
|
|
338
|
|
|
339 Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
340
|
|
|
341 This file is part of MayaChemTools.
|
|
|
342
|
|
|
343 MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
344 the terms of the GNU Lesser General Public License as published by the Free
|
|
|
345 Software Foundation; either version 3 of the License, or (at your option)
|
|
|
346 any later version.
|
|
|
347
|
|
|
348 =cut
|
