annotate bin/ModifySDFilesDataFields.pl @ 0:4816e4a8ae95 draft default tip

Uploaded
author deepakjadmin
date Wed, 20 Jan 2016 09:23:18 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
1 #!/usr/bin/perl -w
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
2 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
3 # $RCSfile: ModifySDFilesDataFields.pl,v $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
4 # $Date: 2015/02/28 20:46:20 $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
5 # $Revision: 1.27 $
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
6 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
7 # Author: Manish Sud <msud@san.rr.com>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
8 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
9 # Copyright (C) 2015 Manish Sud. All rights reserved.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
10 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
11 # This file is part of MayaChemTools.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
12 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
13 # MayaChemTools is free software; you can redistribute it and/or modify it under
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
14 # the terms of the GNU Lesser General Public License as published by the Free
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
15 # Software Foundation; either version 3 of the License, or (at your option) any
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
16 # later version.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
17 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
18 # MayaChemTools is distributed in the hope that it will be useful, but without
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
19 # any warranty; without even the implied warranty of merchantability of fitness
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
20 # for a particular purpose. See the GNU Lesser General Public License for more
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
21 # details.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
22 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
23 # You should have received a copy of the GNU Lesser General Public License
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
26 # Boston, MA, 02111-1307, USA.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
27 #
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
28
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
29 use strict;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
30 use FindBin; use lib "$FindBin::Bin/../lib";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
31 use Getopt::Long;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
32 use File::Basename;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
33 use Text::ParseWords;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
34 use Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
35 use FileUtil;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
36 use SDFileUtil;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
37 use TextUtil;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
38
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
39 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
40
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
41 # Autoflush STDOUT
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
42 $| = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
43
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
44 # Starting message...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
45 $ScriptName = basename($0);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
46 print "\n$ScriptName: Starting...\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
47 $StartTime = new Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
48
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
49 # Get the options and setup script...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
50 SetupScriptUsage();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
51 if ($Options{help} || @ARGV < 1) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
52 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
53 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
54
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
55 my(@SDFilesList);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
56 @SDFilesList = ExpandFileNames(\@ARGV, "sdf sd");
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
57
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
58 # Process options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
59 print "Processing options...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
60 my(%OptionsInfo);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
61 ProcessOptions();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
62
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
63 print "Checking input SD file(s)...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
64 my(%SDFilesInfo);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
65 RetrieveSDFilesInfo();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
66
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
67 # Generate output files...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
68 my($FileIndex);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
69 if (@SDFilesList > 1) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
70 print "\nProcessing SD files...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
71 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
72 for $FileIndex (0 .. $#SDFilesList) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
73 if ($SDFilesInfo{FileOkay}[$FileIndex]) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
74 print "\nProcessing file $SDFilesList[$FileIndex]...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
75 ModifySDFile($FileIndex);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
76 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
77 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
78 print "\n$ScriptName:Done...\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
79
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
80 $EndTime = new Benchmark;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
81 $TotalTime = timediff ($EndTime, $StartTime);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
82 print "Total time: ", timestr($TotalTime), "\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
83
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
84 ###############################################################################
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
85
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
86 # Modify SD file data fields....
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
87 sub ModifySDFile {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
88 my($Index) = @_;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
89 my($SDFile, $NewSDFile);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
90
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
91 $SDFile = $SDFilesList[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
92 $NewSDFile = $SDFilesInfo{OutFile}[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
93
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
94 print "Generating new SD file $NewSDFile...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
95 open NEWSDFILE, ">$NewSDFile" or die "Error: Couldn't open $NewSDFile: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
96 open SDFILE, "$SDFile" or die "Error: Can't open $SDFile: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
97
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
98 my($CmpdCount, $CmpdString, $CmpdData, $MolName, $OldSDField, $NewSDField, $CommonSDField, $Label, $Value, $FieldValues, $MolNameDataField, $URLCmpdIdFieldName, @CmpdLines, %DataFieldAndValues, @DataFieldLabels);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
99 $CmpdCount = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
100
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
101 COMPOUND: while ($CmpdString = ReadCmpdString(\*SDFILE)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
102 $CmpdCount++;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
103 @CmpdLines = split "\n", $CmpdString;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
104 if ($OptionsInfo{UseDataFieldForMolName} || $OptionsInfo{ModifyDataFields}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
105 %DataFieldAndValues = GetCmpdDataHeaderLabelsAndValues(\@CmpdLines);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
106 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
107 if ($OptionsInfo{ModifyMolName}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
108 if ($OptionsInfo{AlwaysReplaceMolName} || !IsNotEmpty($CmpdLines[0])) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
109 $MolNameDataField = $OptionsInfo{MolNameDataField};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
110 if ($OptionsInfo{UseDataFieldForMolName} && exists($DataFieldAndValues{$MolNameDataField})) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
111 $MolName = $DataFieldAndValues{$MolNameDataField};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
112 if (length($MolName) > 80) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
113 $MolName = substr($MolName, 0, 80);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
114 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
115 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
116 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
117 $MolName = "$OptionsInfo{MolNamePrefix}${CmpdCount}";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
118 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
119 $CmpdLines[0] = $MolName;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
120 $CmpdString = join "\n", @CmpdLines;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
121 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
122 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
123 if (!$OptionsInfo{ModifyDataFields}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
124 # Just write the data and get the next compound...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
125 print NEWSDFILE "$CmpdString\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
126 next COMPOUND;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
127 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
128 # Write out the structure data now and handle the old data fields later...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
129 ($CmpdData) = split /\n>/, $CmpdString;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
130 print NEWSDFILE "$CmpdData\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
131
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
132 # Modify specified data fields...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
133 for $NewSDField (sort keys %{$OptionsInfo{SpecifiedNewToOldSDFieldMap}}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
134 $FieldValues = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
135 for $OldSDField (@{$OptionsInfo{SpecifiedNewToOldSDFieldMap}{$NewSDField}}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
136 if (exists($DataFieldAndValues{$OldSDField}) && length($DataFieldAndValues{$OldSDField})) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
137 $Value = $DataFieldAndValues{$OldSDField};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
138 $FieldValues .= ($FieldValues) ? "\n$Value" : $Value;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
139 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
140 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
141 print NEWSDFILE "> <$NewSDField>\n$FieldValues\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
142 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
143 # Add specified common fields...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
144 for $CommonSDField (sort keys %{$OptionsInfo{SpecifiedCommonFieldMap}}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
145 $Value = $OptionsInfo{SpecifiedCommonFieldMap}{$CommonSDField};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
146 print NEWSDFILE "> <$CommonSDField>\n$Value\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
147 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
148 if ($OptionsInfo{CreateDataFieldURL}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
149 $Value = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
150 $URLCmpdIdFieldName = $OptionsInfo{URLCmpdIdFieldName};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
151 if (exists($DataFieldAndValues{$URLCmpdIdFieldName}) && length($DataFieldAndValues{$URLCmpdIdFieldName})) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
152 $Value = $DataFieldAndValues{$URLCmpdIdFieldName};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
153 $Value = "$OptionsInfo{URLCGIScriptName}?$OptionsInfo{URLParamName}=${Value}";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
154 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
155 print NEWSDFILE "> <$OptionsInfo{URLDataFieldLabel}>\n$Value\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
156 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
157
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
158 # Handle old data fields and write 'em in the same order as they appear in the input
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
159 # files...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
160 if ($OptionsInfo{KeepAllOldDataFields} || $OptionsInfo{KeepUnMappedOldDataFields}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
161 my($KeepLabel);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
162 @DataFieldLabels = GetCmpdDataHeaderLabels(\@CmpdLines);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
163 LABEL: for $Label (@DataFieldLabels) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
164 $KeepLabel = $OptionsInfo{KeepAllOldDataFields} ? 1 : ( exists($OptionsInfo{SpecifiedOldToNewSDFieldMap}{$Label}) ? 0 : 1 );
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
165 if (!$KeepLabel) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
166 next LABEL;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
167 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
168 $Value = $DataFieldAndValues{$Label};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
169 print NEWSDFILE "> <$Label>\n$Value\n\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
170 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
171 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
172
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
173 print NEWSDFILE "\$\$\$\$\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
174 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
175 close NEWSDFILE;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
176 close SDFILE;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
177 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
178
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
179 # Process option values...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
180 sub ProcessOptions {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
181 %OptionsInfo = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
182
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
183 $OptionsInfo{Mode} = $Options{mode};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
184
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
185 $OptionsInfo{ModifyMolName} = 1; $OptionsInfo{ModifyDataFields} = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
186 if ($Options{mode} =~ /^both$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
187 $OptionsInfo{ModifyMolName} = 1; $OptionsInfo{ModifyDataFields} = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
188 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
189 elsif ($Options{mode} =~ /^datafields$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
190 $OptionsInfo{ModifyMolName} = 0; $OptionsInfo{ModifyDataFields} = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
191 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
192
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
193 $OptionsInfo{KeepOldDataFields} = $Options{keepolddatafields};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
194 $OptionsInfo{KeepAllOldDataFields} = ($Options{keepolddatafields} =~ /^all$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
195 $OptionsInfo{KeepUnMappedOldDataFields} = ($Options{keepolddatafields} =~ /^unmappedonly$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
196
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
197 $OptionsInfo{MolNameMode} = $Options{molnamemode};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
198 $OptionsInfo{UseDataFieldForMolName} = ($Options{molnamemode} =~ /^datafield$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
199
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
200 $OptionsInfo{MolName} = $Options{molname};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
201 $OptionsInfo{MolNameDataField} = ""; $OptionsInfo{MolNamePrefix} = "Cmpd";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
202 if ($Options{molname}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
203 if ($OptionsInfo{UseDataFieldForMolName}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
204 $OptionsInfo{MolNameDataField} = $Options{molname};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
205 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
206 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
207 $OptionsInfo{MolNamePrefix} = $Options{molname};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
208 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
209 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
210
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
211 $OptionsInfo{MolNameReplace} = $Options{molnamereplace};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
212 $OptionsInfo{AlwaysReplaceMolName} = ($Options{molnamereplace} =~ /^always$/i) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
213
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
214 if ($Options{datafieldsmap} && $Options{datafieldsmapfile}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
215 die "Error: Both \"--datafieldsmap\" and \"--datafieldsmapfile\" options specified: only one is allowed at a time\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
216 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
217
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
218 $OptionsInfo{DataFieldsMap} = $Options{datafieldsmap} ? $Options{datafieldsmap} : '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
219 $OptionsInfo{DataFieldsMapFile} = $Options{datafieldsmapfile} ? $Options{datafieldsmapfile} : '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
220
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
221 my($SpecifiedDataFieldMap);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
222
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
223 %{$OptionsInfo{SpecifiedNewToOldSDFieldMap}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
224 %{$OptionsInfo{SpecifiedOldToNewSDFieldMap}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
225
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
226 $SpecifiedDataFieldMap = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
227 if ($Options{datafieldsmap}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
228 $SpecifiedDataFieldMap = $Options{datafieldsmap};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
229 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
230 elsif ($Options{datafieldsmapfile}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
231 my($Line, @LineWords);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
232 open DATAFIELDSFILE, "$Options{datafieldsmapfile}" or die "Couldn't open $Options{datafieldsmapfile}: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
233 while ($Line = GetTextLine(\*DATAFIELDSFILE)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
234 @LineWords = quotewords(";", 0, $Line);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
235 $SpecifiedDataFieldMap .= JoinWords(\@LineWords, ";", 0);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
236 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
237 close DATAFIELDSFILE;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
238 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
239
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
240 if ($SpecifiedDataFieldMap) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
241 my($DataFieldMap, $DataField, $NewSDField, @OldSDFields, @DataFieldMapSplit, @DataFieldsSplit, $FirstField);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
242 @DataFieldMapSplit = split ";", $SpecifiedDataFieldMap;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
243 for $DataFieldMap (@DataFieldMapSplit) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
244 @DataFieldsSplit = split ",", $DataFieldMap;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
245 if (@DataFieldsSplit == 1) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
246 die "Error: Invalid number of comma delimited values, ", scalar(@DataFieldsSplit), ", specified, @DataFieldsSplit, using \"--datafieldsmap or --datafieldsmapfile\" option: it must contain more than one value.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
247 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
248 $FirstField = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
249 @OldSDFields = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
250 for $DataField (@DataFieldsSplit) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
251 if (!(defined($DataField) && length($DataField))) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
252 die "Error: One of the comma delimited values, \"", join(",", @DataFieldsSplit), "\", specified using \"--datafieldsmap or --datafieldsmapfile\" option is empty.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
253 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
254 if ($FirstField) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
255 $FirstField = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
256 $NewSDField = $DataField;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
257 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
258 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
259 push @OldSDFields, $DataField;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
260 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
261 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
262 # Make sure a datafield is only specified once...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
263 if (exists $OptionsInfo{SpecifiedNewToOldSDFieldMap}{$NewSDField}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
264 die "Error: New data field, $NewSDField, specified more than once using \"--datafieldsmap or --datafieldsmapfile\" option.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
265 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
266 @{$OptionsInfo{SpecifiedNewToOldSDFieldMap}{$NewSDField}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
267 push @{$OptionsInfo{SpecifiedNewToOldSDFieldMap}{$NewSDField}}, @OldSDFields;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
268 for $DataField (@OldSDFields) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
269 if (exists $OptionsInfo{SpecifiedOldToNewSDFieldMap}{$DataField} ) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
270 die "Error: SD field, $DataField, specified more than once using \"--datafieldsmap or --datafieldsmapfile\" option.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
271 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
272 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
273 $OptionsInfo{SpecifiedOldToNewSDFieldMap}{$DataField} = $NewSDField;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
274 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
275 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
276
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
277 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
278 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
279
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
280 $OptionsInfo{DataFieldsCommon} = $Options{datafieldscommon} ? $Options{datafieldscommon} : '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
281 %{$OptionsInfo{SpecifiedCommonFieldMap}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
282
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
283 if ($Options{datafieldscommon}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
284 my($DataFieldName, $DataFieldValue, $Index, @CommonDataFieldsSplit);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
285 @CommonDataFieldsSplit = split ",", $Options{datafieldscommon};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
286 if (@CommonDataFieldsSplit % 2) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
287 die "Error: Invalid number of comma delimited values, ", scalar(@CommonDataFieldsSplit), ", specified \"", join(",", @CommonDataFieldsSplit), "\" using \"--datafieldscommon\" option: it must contain even number of values.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
288 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
289 for ($Index = 0; $Index < @CommonDataFieldsSplit; $Index += 2) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
290 $DataFieldName = $CommonDataFieldsSplit[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
291 $DataFieldValue = $CommonDataFieldsSplit[$Index + 1];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
292 if (exists $OptionsInfo{SpecifiedCommonFieldMap}{$DataFieldName}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
293 die "Error: Common data field, $DataFieldName, specified more than once using \"--datafieldscommon\" option.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
294 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
295 if (exists($OptionsInfo{SpecifiedNewToOldSDFieldMap}{$DataFieldName}) || exists($OptionsInfo{SpecifiedOldToNewSDFieldMap}{$DataFieldName})) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
296 die "Error: Common data field, $DataFieldName, specified using \"--datafieldscommon\" option cannot be specified in \"--datafieldsmap or --datafieldsmapfile\" option.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
297 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
298 $OptionsInfo{SpecifiedCommonFieldMap}{$DataFieldName} = $DataFieldValue;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
299 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
300 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
301
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
302 $OptionsInfo{DataFieldURL} = $Options{datafieldurl} ? $Options{datafieldurl} : '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
303 $OptionsInfo{CreateDataFieldURL} = (exists($Options{datafieldurl}) && length($Options{datafieldurl}) ) ? 1 : 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
304
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
305 $OptionsInfo{URLDataFieldLabel} = ""; $OptionsInfo{URLCGIScriptName} = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
306 $OptionsInfo{URLParamName} = ""; $OptionsInfo{URLCmpdIdFieldName} = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
307
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
308 if ($OptionsInfo{CreateDataFieldURL}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
309 my(@DataFieldURLSplit, $Value);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
310 @DataFieldURLSplit = split ",", $Options{datafieldurl};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
311 if (@DataFieldURLSplit != 4) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
312 die "Error: Invalid number of values, ", scalar(@DataFieldURLSplit), ", specified using \"--datafieldURL\" option: it must contain 4 values.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
313 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
314 for $Value (@DataFieldURLSplit) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
315 if (!IsNotEmpty($Value)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
316 die "Error: One of the values, $Options{datafieldurl}, specified using \"--datafieldURL\" option is empty.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
317 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
318 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
319 $OptionsInfo{URLDataFieldLabel} = $DataFieldURLSplit[0];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
320 $OptionsInfo{URLCGIScriptName} = $DataFieldURLSplit[1];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
321 $OptionsInfo{URLParamName} = $DataFieldURLSplit[2];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
322 $OptionsInfo{URLCmpdIdFieldName} = $DataFieldURLSplit[3];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
323 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
324
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
325 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
326
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
327 # Retrieve information about input SD files...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
328 sub RetrieveSDFilesInfo {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
329 my($Index, $SDFile, $FileDir, $FileName, $FileExt, $OutFileRoot, $OutFile, $DataFieldName);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
330
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
331 %SDFilesInfo = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
332 @{$SDFilesInfo{FileOkay}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
333 @{$SDFilesInfo{OutFile}} = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
334
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
335 FILELIST: for $Index (0 .. $#SDFilesList) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
336 $SDFile = $SDFilesList[$Index];
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
337
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
338 $SDFilesInfo{FileOkay}[$Index] = 0;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
339 $SDFilesInfo{OutFile}[$Index] = '';
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
340
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
341 if (!(-e $SDFile)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
342 warn "Warning: Ignoring file $SDFile: It doesn't exist\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
343 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
344 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
345 if (!CheckFileType($SDFile, "sd sdf")) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
346 warn "Warning: Ignoring file $SDFile: It's not a SD file\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
347 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
348 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
349 $FileDir = ""; $FileName = ""; $FileExt = "";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
350 ($FileDir, $FileName, $FileExt) = ParseFileName($SDFile);
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
351 if ($Options{root} && (@SDFilesList == 1)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
352 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
353 if ($RootFileName && $RootFileExt) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
354 $FileName = $RootFileName;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
355 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
356 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
357 $FileName = $Options{root};
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
358 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
359 $OutFileRoot = $FileName;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
360 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
361 else {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
362 $OutFileRoot = $FileName . "ModifiedDataFields";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
363 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
364
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
365 $OutFile = $OutFileRoot . ".$FileExt";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
366 if (lc($OutFile) eq lc($SDFile)) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
367 warn "Warning: Ignoring file $SDFile:Output file name, $OutFile, is same as input SD file name, $SDFile\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
368 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
369 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
370 if (!$Options{overwrite}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
371 if (-e $OutFile) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
372 warn "Warning: Ignoring file $SDFile: The file $OutFile already exists\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
373 next FILELIST;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
374 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
375 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
376
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
377 $SDFilesInfo{FileOkay}[$Index] = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
378 $SDFilesInfo{OutFile}[$Index] = $OutFile;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
379 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
380 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
381
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
382 # Setup script usage and retrieve command line arguments specified using various options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
383 sub SetupScriptUsage {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
384
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
385 # Retrieve all the options...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
386 %Options = ();
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
387 $Options{detail} = 1;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
388 $Options{keepolddatafields} = "none";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
389 $Options{mode} = "molname";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
390 $Options{molnamemode} = "labelprefix";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
391 $Options{molnamereplace} = "empty";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
392
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
393 if (!GetOptions(\%Options, "detail|d=i", "datafieldscommon=s", "datafieldsmap=s", "datafieldsmapfile=s", "datafieldurl=s", "help|h", "keepolddatafields|k=s", "mode|m=s", "molname=s", "molnamemode=s", "molnamereplace=s", "overwrite|o", "root|r=s", "workingdir|w=s")) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
394 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
395 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
396 if ($Options{workingdir}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
397 if (! -d $Options{workingdir}) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
398 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
399 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
400 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
401 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
402 if ($Options{keepolddatafields} !~ /^(all|unmappedonly|none)$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
403 die "Error: The value specified, $Options{keepolddatafields}, for option \"-k --keepolddatafields\" is not valid. Allowed values: all, unmappedonly, or none\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
404 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
405 if ($Options{mode} !~ /^(molname|datafields|both)$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
406 die "Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: molname, datafields, or both\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
407 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
408 if ($Options{molnamemode} !~ /^(datafield|labelprefix)$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
409 die "Error: The value specified, $Options{molnamemode}, for option \"--molnamemode\" is not valid. Allowed values: datafield or labelprefix\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
410 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
411 if ($Options{molnamereplace} !~ /^(always|empty)$/i) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
412 die "Error: The value specified, $Options{molnamereplace}, for option \"--molnamereplace\" is not valid. Allowed values: always or empty\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
413 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
414 if (!IsPositiveInteger($Options{detail})) {
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
415 die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n";
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
416 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
417 }
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
418
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
419 __END__
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
420
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
421 =head1 NAME
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
422
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
423 ModifySDFilesDataFields.pl - Modify data fields in SDFile(s)
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
424
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
425 =head1 SYNOPSIS
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
426
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
427 ModifySDFilesDataFields.pl SDFile(s)...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
428
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
429 ModifySDFilesDataFields.pl [B<-d, --detail> infolevel]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
430 [B<--datafieldscommon> newfieldlabel, newfieldvalue, [newfieldlabel, newfieldvalue,...]]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
431 [B<--datafieldsmap> newfieldlabel, oldfieldlabel, [oldfieldlabel,...]; [newfieldlabel, oldfieldlabel, [oldfieldlabel,...]]]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
432 [B<--datafieldsmapfile> filename] [B<--datafieldURL> URLDataFieldLabel, CGIScriptPath, CGIParamName, CmpdIDFieldLabel]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
433 [B<-h, --help>] [B<-k, --keepolddatafields> all | unmappedonly | none] [B<-m, --mode> molname | datafields | both]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
434 [B<--molnamemode> datafield | labelprefix] [B<--molname> datafieldname or prefixstring]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
435 [B<--molnamereplace> always | empty] [B<-o, --overwrite>] [B<-r, --root> rootname]
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
436 [B<-w, --workingdir> dirname] SDFile(s)...
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
437
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
438 =head1 DESCRIPTION
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
439
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
440 Modify molname line and data fields in I<SDFile(s)>. Molname line can be replaced by a
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
441 data field value or assigned a sequential ID prefixed with a specific string. For data
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
442 fields and modification of their values, these types of options are supported: replace
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
443 data field labels by another set of labels; combine values of multiple data fields and
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
444 assign a new label; add specific set of data field labels and values to all compound
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
445 records; and others.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
446
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
447 The file names are separated by space.The valid file extensions are I<.sdf> and I<.sd>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
448 All other file names are ignored. All the SD files in a current directory can be specified
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
449 either by I<*.sdf> or the current directory name.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
450
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
451 =head1 OPTIONS
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
452
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
453 =over 4
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
454
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
455 =item B<-d, --detail> I<infolevel>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
456
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
457 Level of information to print about compound records being ignored. Default: I<1>. Possible
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
458 values: I<1, 2 or 3>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
459
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
460 =item B<--datafieldscommon> I<newfieldlabel, newfieldvalue, [newfieldlabel, newfieldvalue,...]>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
461
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
462 Specify data field labels and values for addition to each compound record. It's a comma delimited
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
463 list of data field label and values pair. Default: I<none>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
464
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
465 Examples:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
466
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
467 DepositionDate,YYYY-MM-DD
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
468 Source,www.domainname.org,ReleaseData,YYYY-MM-DD
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
469
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
470 =item B<--datafieldsmap> I<newfieldlabel, oldfieldlabel, [oldfieldlabel,...]; [newfieldlabel, oldfieldlabel, [oldfieldlabel,...]]>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
471
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
472 Specify how various data field labels and values are combined to generate a new data field
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
473 labels and their values. All the comma delimited data fields, with in a semicolon delimited set,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
474 are mapped to the first new data field label along with the data field values joined via new
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
475 line character. Default: I<none>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
476
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
477 Examples:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
478
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
479 Synonym,Name,SystematicName,Synonym;CmpdID,Extreg
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
480 HBondDonors,SumNHOH
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
481
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
482 =item B<--datafieldsmapfile> I<filename>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
483
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
484 Filename containing mapping of data fields. Format of data fields line in this file corresponds
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
485 to B<--datafieldsmap> option. Example:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
486
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
487 Line 1: Synonym,Name,SystematicName,Synonym;CmpdID,Extreg
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
488 Line 2: HBondDonors,SumNHOH
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
489
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
490
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
491 =item B<--datafieldURL> I<URLDataFieldLabel, CGIScriptPath, CGIParamName, CmpdIDFieldLabel>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
492
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
493 Specify how to generate a URL for retrieving compound data from a web server and add it
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
494 to each compound record. I<URLDataFieldLabel> is used as the data field label for URL value
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
495 which is created by combining I<CGIScriptPath,CGIParamName,CmpdIDFieldLabel> values:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
496 CGIScriptPath?CGIParamName=CmpdIDFieldLabelValue. Default: I<none>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
497
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
498 Example:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
499
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
500 Source,http://www.yourdomain.org/GetCmpd.pl,Reg_ID,Mol_ID
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
501
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
502 =item B<-h, --help>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
503
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
504 Print this help message.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
505
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
506 =item B<-k, --keepolddatafields> I<all | unmappedonly | none>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
507
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
508 Specify how to transfer old data fields from input SDFile(s) to new SDFile(s) during
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
509 I<datafields | both> value of B<-m, --mode> option: keep all old data fields; write out the ones
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
510 not mapped to new fields as specified by B<--datafieldsmap> or <--datafieldsmapfile> options;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
511 or ignore all old data field labels. For I<molname> B<-m --mode>, old datafields are always kept.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
512 Possible values: I<all | unmappedonly | none>. Default: I<none>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
513
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
514 =item B<-m, --mode> I<molname | datafields | both>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
515
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
516 Specify how to modify SDFile(s): I<molname> - change molname line by another datafield or value;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
517 I<datafield> - modify data field labels and values by replacing one label by another, combining
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
518 multiple data field labels and values, adding specific set of data field labels and values to all compound, or
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
519 inserting an URL for compound retrieval to each record; I<both> - change molname line and datafields
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
520 simultaneously. Possible values: I<molname | datafields | both>. Default: I<molname>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
521
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
522 =item B<--molnamemode> I<datafield | labelprefix>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
523
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
524 Specify how to change molname line for B<-m --mode> option values of I<molname | both>: use
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
525 a datafield label value or assign a sequential ID prefixed with I<labelprefix>. Possible values:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
526 I<datafield | labelprefix>. Default: I<labelprefix>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
527
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
528 =item B<--molname> I<datafieldname or prefixstring>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
529
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
530 Molname generation method. For I<datafield> value of B<--molnamemode> option, it corresponds
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
531 to datafield label name whose value is used for molname; otherwise, it's a prefix string used for
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
532 generating compound IDs like labelprefixstring<Number>. Default value, I<Cmpd>, generates
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
533 compound IDs like Cmpd<Number> for molname.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
534
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
535 =item B<--molnamereplace> I<always | empty>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
536
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
537 Specify when to replace molname line for B<-m --mode> option values of I<molname | both>:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
538 always replace the molname line using B<--molname> option or only when it's empty. Possible
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
539 values: I<always | empty>. Default: I<empty>.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
540
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
541 =item B<-o, --overwrite>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
542
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
543 Overwrite existing files.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
544
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
545 =item B<-r, --root> I<rootname>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
546
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
547 New SD file name is generated using the root: <Root>.<Ext>. Default new file
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
548 name: <InitialSDFileName>ModifiedDataFields.<Ext>. This option is ignored for multiple
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
549 input files.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
550
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
551 =item B<-w, --workingdir> I<dirname>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
552
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
553 Location of working directory. Default: current directory.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
554
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
555 =back
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
556
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
557 =head1 EXAMPLES
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
558
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
559 To replace empty molname lines by Cmpd<CmpdNumber> and generate a new SD file
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
560 NewSample1.sdf, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
561
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
562 % ModifySDFilesDataFields.pl -o -r NewSample1 Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
563
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
564 To replace all molname lines by Mol_ID data field generate a new SD file
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
565 NewSample1.sdf, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
566
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
567 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
568 --molnamereplace always -r NewSample1 -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
569
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
570 To replace all molname lines by Mol_ID data field, map Name and CompoundName to
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
571 a new datafield Synonym, and generate a new SD file NewSample1.sdf, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
572
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
573 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
574 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
575 --datafieldsmap "Synonym,Name,CompoundName" -r
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
576 NewSample1 -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
577
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
578 To replace all molname lines by Mol_ID data field, map Name and CompoundName to
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
579 a new datafield Synonym, add common fields ReleaseDate and Source, and
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
580 generate a new SD file NewSample1.sdf without keeping any old SD data fields, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
581
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
582 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
583 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
584 --datafieldsmap "Synonym,Name,CompoundName"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
585 --datafieldscommon "ReleaseDate,yyyy-mm-dd,Source,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
586 www.mayachemtools.org" --keepolddatafields none -r
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
587 NewSample1 -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
588
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
589 B<Preparing SD files PubChem deposition:>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
590
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
591 Consider a SD file with these fields: Mol_ID, Name, Synonyms and Systematic_Name.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
592 And Mol_ID data field uniquely identifies your compound.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
593
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
594 To prepare a new SD file CmpdDataForPubChem.sdf containing only required
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
595 PUBCHEM_EXT_DATASOURCE_REGID field, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
596
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
597 % ModifySDFilesDataFields.pl --m datafields
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
598 --datafieldsmap
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
599 "PUBCHEM_EXT_DATASOURCE_REGID,Mol_ID"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
600 -r CmpdDataForPubChem -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
601
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
602 To prepare a new SD file CmpdDataForPubChem.sdf containing only required
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
603 PUBCHEM_EXT_DATASOURCE_REGID field and replace molname line with Mol_ID, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
604
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
605 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
606 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
607 --datafieldsmap
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
608 "PUBCHEM_EXT_DATASOURCE_REGID,Mol_ID"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
609 -r CmpdDataForPubChem -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
610
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
611 In addition to required PubChem data field, you can also add optional PubChem data
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
612 fields.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
613
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
614 To map your Name, Synonyms and Systematic_Name data fields to optional
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
615 PUBCHEM_SUBSTANCE_SYNONYM data field along with required ID field, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
616
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
617 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
618 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
619 --datafieldsmap
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
620 "PUBCHEM_EXT_DATASOURCE_REGID,Mol_ID;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
621 PUBCHEM_SUBSTANCE_SYNONYM,Name,CompoundName"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
622 -r CmpdDataForPubChem -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
623
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
624 To add your <domain.org> as PUBCHEM_EXT_SUBSTANCE_URL and link substance
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
625 retrieval to your CGI script <http://www.yourdomain.org/GetCmpd.pl,Reg_ID,Mol_ID>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
626 via PUBCHEM_EXT_DATASOURCE_REGID field along with optional and required
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
627 data fields, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
628
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
629 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
630 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
631 --datafieldsmap
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
632 "PUBCHEM_EXT_DATASOURCE_REGID,Mol_ID;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
633 PUBCHEM_SUBSTANCE_SYNONYM,Name,CompoundName"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
634 --datafieldscommon
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
635 "PUBCHEM_EXT_SUBSTANCE_URL,domain.org"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
636 --datafieldURL "PUBCHEM_EXT_DATASOURCE_URL,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
637 http://www.yourdomain.org/GetCmpd.pl,Reg_ID,Mol_ID"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
638 -r CmpdDataForPubChem -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
639
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
640 And to add a publication date and request a release data using
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
641 PUBCHEM_PUBLICATION_DATE and PUBCHEM_DEPOSITOR_RECORD_DATE data fields
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
642 along with all the data fields in earlier examples, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
643 optional fields, type:
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
644
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
645 % ModifySDFilesDataFields.pl --molnamemode datafield
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
646 --molnamereplace always --molname Mol_ID --mode both
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
647 --datafieldsmap
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
648 "PUBCHEM_EXT_DATASOURCE_REGID,Mol_ID;
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
649 PUBCHEM_SUBSTANCE_SYNONYM,Name,CompoundName"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
650 --datafieldURL "PUBCHEM_EXT_DATASOURCE_URL,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
651 http://www.yourdomain.org/GetCmpd.pl,Reg_ID,Mol_ID"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
652 --datafieldscommon
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
653 "PUBCHEM_EXT_SUBSTANCE_URL,domain.org,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
654 PUBCHEM_PUBLICATION_DATE,YYY-MM-DD,
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
655 PUBCHEM_DEPOSITOR_RECORD_DATE,YYYY-MM-DD"
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
656 -r CmpdDataForPubChem -o Sample1.sdf
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
657
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
658 =head1 AUTHOR
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
659
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
660 Manish Sud <msud@san.rr.com>
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
661
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
662 =head1 SEE ALSO
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
663
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
664 InfoSDFiles.pl, JoinSDFiles.pl, MergeTextFilesWithSD.pl, SplitSDFiles.pl, SDFilesToHTML.pl
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
665
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
666 =head1 COPYRIGHT
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
667
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
668 Copyright (C) 2015 Manish Sud. All rights reserved.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
669
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
670 This file is part of MayaChemTools.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
671
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
672 MayaChemTools is free software; you can redistribute it and/or modify it under
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
673 the terms of the GNU Lesser General Public License as published by the Free
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
674 Software Foundation; either version 3 of the License, or (at your option)
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
675 any later version.
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
676
4816e4a8ae95 Uploaded
deepakjadmin
parents:
diff changeset
677 =cut