0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:ExtractFromSDFiles.pl</title>
|
|
4 <meta http-equiv="content-type" content="text/html;charset=utf-8">
|
|
5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css">
|
|
6 </head>
|
|
7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10">
|
|
8 <br/>
|
|
9 <center>
|
|
10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a>
|
|
11 </center>
|
|
12 <br/>
|
|
13 <pre>
|
|
14 1 #!/usr/bin/perl -w
|
|
15 2 <span class="c">#</span>
|
|
16 3 <span class="c"># $RCSfile: ExtractFromSDFiles.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/03/22 19:11:27 $</span>
|
|
18 5 <span class="c"># $Revision: 1.48 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$FindBin::Bin/../lib"</span><span class="sc">;</span>
|
|
44 31 <span class="k">use</span> <span class="w">Getopt::Long</span><span class="sc">;</span>
|
|
45 32 <span class="k">use</span> <span class="w">File::Basename</span><span class="sc">;</span>
|
|
46 33 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
|
|
47 34 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
48 35 <span class="k">use</span> <span class="w">SDFileUtil</span><span class="sc">;</span>
|
|
49 36 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
50 37 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
51 38
|
|
52 39 <span class="k">my</span><span class="s">(</span><span class="i">$ScriptName</span><span class="cm">,</span> <span class="i">%Options</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="cm">,</span> <span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$TotalTime</span><span class="s">)</span><span class="sc">;</span>
|
|
53 40
|
|
54 41 <span class="c"># Autoflush STDOUT</span>
|
|
55 42 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
56 43
|
|
57 44 <span class="c"># Starting message...</span>
|
|
58 45 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
59 46 <span class="k">print</span> <span class="q">"\n$ScriptName:Starting...\n\n"</span><span class="sc">;</span>
|
|
60 47 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
61 48
|
|
62 49 <span class="c"># Get the options and setup script...</span>
|
|
63 50 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
64 51 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
65 52 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">"$FindBin::Bin/$ScriptName"</span><span class="s">)</span><span class="sc">;</span>
|
|
66 53 <span class="s">}</span>
|
|
67 54
|
|
68 55 <span class="k">my</span><span class="s">(</span><span class="i">@SDFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
69 56 <span class="i">@SDFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">"sdf sd"</span><span class="s">)</span><span class="sc">;</span>
|
|
70 57
|
|
71 58 <span class="c"># Process options...</span>
|
|
72 59 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
73 60 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
74 61 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
75 62
|
|
76 63 <span class="c"># Collect information about SD files...</span>
|
|
77 64 <span class="k">print</span> <span class="q">"Checking input SD file(s)...\n"</span><span class="sc">;</span>
|
|
78 65 <span class="k">my</span><span class="s">(</span><span class="i">%SDFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
79 66 <span class="i">RetrieveSDFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
80 67
|
|
81 68 <span class="c"># Generate output files...</span>
|
|
82 69 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
83 70 <span class="k">if</span> <span class="s">(</span><span class="i">@SDFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
84 71 <span class="k">print</span> <span class="q">"\nProcessing SD files...\n"</span><span class="sc">;</span>
|
|
85 72 <span class="s">}</span>
|
|
86 73 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
87 74 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
88 75 <span class="k">print</span> <span class="q">"\nProcessing file $SDFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
89 76 <span class="i">ExtractFromSDFile</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
90 77 <span class="s">}</span>
|
|
91 78 <span class="s">}</span>
|
|
92 79 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
93 80
|
|
94 81 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
95 82 <span class="i">$TotalTime</span> = <span class="w">timediff</span> <span class="s">(</span><span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="s">)</span><span class="sc">;</span>
|
|
96 83 <span class="k">print</span> <span class="q">"Total time: "</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
97 84
|
|
98 85 <span class="c">###############################################################################</span>
|
|
99 86
|
|
100 87 <span class="c"># Extract data from a SD file...</span>
|
|
101 <a name="ExtractFromSDFile-"></a> 88 <span class="k">sub </span><span class="m">ExtractFromSDFile</span> <span class="s">{</span>
|
|
102 89 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
103 90
|
|
104 91 <span class="i">OpenInputAndOutputFiles</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
105 92
|
|
106 93 <span class="j">MODE:</span> <span class="s">{</span>
|
|
107 94 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^AllDataFields$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
108 95 <span class="i">ExtractAllDataFields</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
109 96 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
110 97 <span class="s">}</span>
|
|
111 98 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^CommonDataFields$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
112 99 <span class="i">ExtractCommonDataFields</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
113 100 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
114 101 <span class="s">}</span>
|
|
115 102 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFields$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
116 103 <span class="i">ExtractDataFields</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
117 104 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
118 105 <span class="s">}</span>
|
|
119 106 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^(DataFieldByList|DatafieldUniqueByList)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
120 107 <span class="i">ExtractDataFieldByList</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
121 108 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
122 109 <span class="s">}</span>
|
|
123 110 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFieldNotByList$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
124 111 <span class="i">ExtractDataFieldNotByList</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
125 112 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
126 113 <span class="s">}</span>
|
|
127 114 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFieldsByValue$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
128 115 <span class="i">ExtractDataFieldsByValue</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
129 116 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
130 117 <span class="s">}</span>
|
|
131 118 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFieldsByRegex$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
132 119 <span class="i">ExtractDataFieldsByRegex</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
133 120 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
134 121 <span class="s">}</span>
|
|
135 122 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^RandomCmpds$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
136 123 <span class="i">ExtractRandomCompounds</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
137 124 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
138 125 <span class="s">}</span>
|
|
139 126 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^MolNames$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
140 127 <span class="i">ExtractMolNames</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
141 128 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
142 129 <span class="s">}</span>
|
|
143 130 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^RecordNum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
144 131 <span class="i">ExtractRecordNum</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
145 132 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
146 133 <span class="s">}</span>
|
|
147 134 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^RecordNums$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
148 135 <span class="i">ExtractRecordNums</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
149 136 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
150 137 <span class="s">}</span>
|
|
151 138 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^RecordRange$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
152 139 <span class="i">ExtractRecordRange</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
153 140 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
154 141 <span class="s">}</span>
|
|
155 142 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^2DCmpdRecords$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
156 143 <span class="i">Extract2DCmpdRecords</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
157 144 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
158 145 <span class="s">}</span>
|
|
159 146 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^3DCmpdRecords$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
160 147 <span class="i">Extract3DCmpdRecords</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
161 148 <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span>
|
|
162 149 <span class="s">}</span>
|
|
163 150 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: alldatafields, commondatafields, datafields, datafieldsbyvalue, datafieldbylist, datafielduniquebylist, datafieldnotbylist, molnames, randomcmpds, recordnum, recordnums, recordrange, 2dcmpdrecords, 3dcmpdrecords\n"</span><span class="sc">;</span>
|
|
164 151 <span class="s">}</span>
|
|
165 152
|
|
166 153 <span class="i">CloseInputAndOutputFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
167 154 <span class="s">}</span>
|
|
168 155
|
|
169 156 <span class="c"># Extract all data fields...</span>
|
|
170 <a name="ExtractAllDataFields-"></a> 157 <span class="k">sub </span><span class="m">ExtractAllDataFields</span> <span class="s">{</span>
|
|
171 158 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
172 159 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
173 160
|
|
174 161 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
175 162 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
176 163
|
|
177 164 <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
178 165 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
179 166 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
180 167
|
|
181 168 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
182 169 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
183 170 <span class="i">WriteSDFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
184 171 <span class="s">}</span>
|
|
185 172 <span class="s">}</span>
|
|
186 173
|
|
187 174 <span class="c"># Extract common data fields...</span>
|
|
188 <a name="ExtractCommonDataFields-"></a> 175 <span class="k">sub </span><span class="m">ExtractCommonDataFields</span> <span class="s">{</span>
|
|
189 176 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
190 177 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
191 178
|
|
192 179 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
193 180 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
194 181
|
|
195 182 <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
196 183 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
197 184 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
198 185
|
|
199 186 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
200 187 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
201 188 <span class="i">WriteSDFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
202 189 <span class="s">}</span>
|
|
203 190 <span class="s">}</span>
|
|
204 191
|
|
205 192 <span class="c"># Extract specified data fields...</span>
|
|
206 <a name="ExtractDataFields-"></a> 193 <span class="k">sub </span><span class="m">ExtractDataFields</span> <span class="s">{</span>
|
|
207 194 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
208 195 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
209 196
|
|
210 197 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="sc">;</span>
|
|
211 198 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
212 199
|
|
213 200 <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
214 201 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
215 202 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
216 203
|
|
217 204 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
218 205 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
219 206 <span class="i">WriteSDFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
220 207 <span class="s">}</span>
|
|
221 208 <span class="s">}</span>
|
|
222 209
|
|
223 210 <span class="c"># Extract data fields using a list...</span>
|
|
224 <a name="ExtractDataFieldByList-"></a> 211 <span class="k">sub </span><span class="m">ExtractDataFieldByList</span> <span class="s">{</span>
|
|
225 212 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
226 213 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdNum</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$SpecifiedDataFieldValuesFoundCount</span><span class="cm">,</span> <span class="i">$CurrentValue</span><span class="cm">,</span> <span class="i">$SpecifiedDataFieldLabel</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
227 214
|
|
228 215 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
229 216 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
230 217
|
|
231 218 <span class="k">for</span> <span class="i">$Value</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
232 219 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$Value</span>} = <span class="q">"NotFound"</span><span class="sc">;</span>
|
|
233 220 <span class="s">}</span>
|
|
234 221 <span class="i">$SpecifiedDataFieldValuesFoundCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
235 222 <span class="i">$SpecifiedDataFieldLabel</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabel</span>}<span class="sc">;</span>
|
|
236 223
|
|
237 224 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
238 225 <span class="i">$CmpdNum</span>++<span class="sc">;</span>
|
|
239 226
|
|
240 227 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
241 228 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
242 229
|
|
243 230 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$SpecifiedDataFieldLabel</span>}<span class="s">)</span> <span class="s">{</span>
|
|
244 231 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
245 232 <span class="s">}</span>
|
|
246 233
|
|
247 234 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
248 235
|
|
249 236 <span class="i">$SpecifiedDataFieldLabel</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabel</span>}<span class="sc">;</span>
|
|
250 237 <span class="i">$CurrentValue</span> = <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$SpecifiedDataFieldLabel</span>}<span class="sc">;</span>
|
|
251 238
|
|
252 239 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$CurrentValue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
253 240 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedDataFieldValuesFoundCount</span> < <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesCount</span>}<span class="s">)</span> <span class="s">{</span>
|
|
254 241 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$CurrentValue</span>} <span class="k">eq</span> <span class="q">"NotFound"</span><span class="s">)</span> <span class="s">{</span>
|
|
255 242 <span class="i">$SpecifiedDataFieldValuesFoundCount</span>++<span class="sc">;</span>
|
|
256 243 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$CurrentValue</span>} = <span class="q">"Found"</span><span class="sc">;</span>
|
|
257 244 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFieldUniqueByList$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
258 245 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
259 246 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
260 247 <span class="s">}</span>
|
|
261 248 <span class="s">}</span>
|
|
262 249 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^DataFieldByList$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
263 250 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
264 251 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
265 252 <span class="s">}</span>
|
|
266 253 <span class="s">}</span>
|
|
267 254 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedDataFieldValuesFoundCount</span> >= <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesCount</span>}<span class="s">)</span> <span class="s">{</span>
|
|
268 255 <span class="k">last</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
269 256 <span class="s">}</span>
|
|
270 257 <span class="s">}</span>
|
|
271 258 <span class="s">}</span>
|
|
272 259 <span class="s">}</span>
|
|
273 260
|
|
274 261 <span class="c"># Extract data field whose values are not on the specified list...</span>
|
|
275 <a name="ExtractDataFieldNotByList-"></a> 262 <span class="k">sub </span><span class="m">ExtractDataFieldNotByList</span> <span class="s">{</span>
|
|
276 263 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
277 264 <span class="k">my</span><span class="s">(</span><span class="i">$CurrentValue</span><span class="cm">,</span> <span class="i">$SpecifiedDataFieldLabel</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
278 265
|
|
279 266 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
280 267 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
281 268
|
|
282 269 <span class="i">$SpecifiedDataFieldLabel</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabel</span>}<span class="sc">;</span>
|
|
283 270
|
|
284 271 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
285 272 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
286 273 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
287 274
|
|
288 275 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$SpecifiedDataFieldLabel</span>}<span class="s">)</span> <span class="s">{</span>
|
|
289 276 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
290 277 <span class="s">}</span>
|
|
291 278
|
|
292 279 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
293 280
|
|
294 281 <span class="i">$CurrentValue</span> = <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$SpecifiedDataFieldLabel</span>}<span class="sc">;</span>
|
|
295 282
|
|
296 283 <span class="c"># Make sure the current value is not empty and is not only specified list of values...</span>
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$CurrentValue</span><span class="s">)</span> || <span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$CurrentValue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
298 285 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
299 286 <span class="s">}</span>
|
|
300 287
|
|
301 288 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
302 289 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
303 290 <span class="s">}</span>
|
|
304 291 <span class="s">}</span>
|
|
305 292
|
|
306 293 <span class="c"># Extract data fields by value...</span>
|
|
307 <a name="ExtractDataFieldsByValue-"></a> 294 <span class="k">sub </span><span class="m">ExtractDataFieldsByValue</span> <span class="s">{</span>
|
|
308 295 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
309 296 <span class="k">my</span><span class="s">(</span><span class="i">$Label</span><span class="cm">,</span> <span class="i">$CurrentValue</span><span class="cm">,</span> <span class="i">$SpecifiedCriterion</span><span class="cm">,</span> <span class="i">$SpecifiedValue</span><span class="cm">,</span> <span class="i">$ViolationCount</span><span class="cm">,</span> <span class="i">$Nothing</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
310 297
|
|
311 298 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
312 299 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
313 300
|
|
314 301 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
315 302 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
316 303 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
317 304
|
|
318 305 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
319 306 <span class="i">$ViolationCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
320 307
|
|
321 308 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
322 309 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$Label</span>}<span class="s">)</span> <span class="s">{</span>
|
|
323 310 <span class="i">$CurrentValue</span> = <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
324 311 <span class="i">$SpecifiedCriterion</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldCriteriaMap</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
325 312 <span class="i">$SpecifiedValue</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesMap</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
326 313
|
|
327 314 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">NumericalComparison</span>}<span class="s">)</span> <span class="s">{</span>
|
|
328 315 <span class="j">CRITERION:</span> <span class="s">{</span>
|
|
329 316 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^eq$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> != <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
330 317 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^le$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> > <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
331 318 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^ge$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> < <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
332 319 <span class="i">$Nothing</span> = <span class="n">1</span><span class="sc">;</span>
|
|
333 320 <span class="s">}</span>
|
|
334 321 <span class="s">}</span>
|
|
335 322 <span class="k">else</span> <span class="s">{</span>
|
|
336 323 <span class="j">CRITERION:</span> <span class="s">{</span>
|
|
337 324 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^eq$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> <span class="k">ne</span> <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
338 325 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^le$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> <span class="k">gt</span> <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
339 326 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedCriterion</span> =~ <span class="q">/^ge$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> <span class="k">lt</span> <span class="i">$SpecifiedValue</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
340 327 <span class="i">$Nothing</span> = <span class="n">1</span><span class="sc">;</span>
|
|
341 328 <span class="s">}</span>
|
|
342 329 <span class="s">}</span>
|
|
343 330 <span class="s">}</span>
|
|
344 331 <span class="s">}</span>
|
|
345 332 <span class="k">if</span> <span class="s">(</span><span class="i">$ViolationCount</span> <= <span class="i">$OptionsInfo</span>{<span class="w">Violations</span>}<span class="s">)</span> <span class="s">{</span>
|
|
346 333 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
347 334 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
348 335 <span class="s">}</span>
|
|
349 336 <span class="s">}</span>
|
|
350 337 <span class="s">}</span>
|
|
351 338
|
|
352 339 <span class="c"># Extract data fields by value using regular expression match...</span>
|
|
353 <a name="ExtractDataFieldsByRegex-"></a> 340 <span class="k">sub </span><span class="m">ExtractDataFieldsByRegex</span> <span class="s">{</span>
|
|
354 341 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
355 342 <span class="k">my</span><span class="s">(</span><span class="i">$Label</span><span class="cm">,</span> <span class="i">$CurrentValue</span><span class="cm">,</span> <span class="i">$SpecifiedRegexCriterion</span><span class="cm">,</span> <span class="i">$SpecifiedRegex</span><span class="cm">,</span> <span class="i">$ViolationCount</span><span class="cm">,</span> <span class="i">$Nothing</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
356 343
|
|
357 344 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
358 345 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
359 346
|
|
360 347 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
361 348 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
362 349 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
363 350
|
|
364 351 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
365 352 <span class="i">$ViolationCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
366 353
|
|
367 354 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
368 355 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$Label</span>}<span class="s">)</span> <span class="s">{</span>
|
|
369 356 <span class="i">$CurrentValue</span> = <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
370 357 <span class="i">$SpecifiedRegexCriterion</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexCriteriaMap</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
371 358 <span class="i">$SpecifiedRegex</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexMap</span>}{<span class="i">$Label</span>}<span class="sc">;</span>
|
|
372 359
|
|
373 360 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">RegexIgnoreCase</span>}<span class="s">)</span> <span class="s">{</span>
|
|
374 361 <span class="j">CRITERION:</span> <span class="s">{</span>
|
|
375 362 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedRegexCriterion</span> =~ <span class="q">/^eq$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> !~ <span class="q">/$SpecifiedRegex/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
376 363 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedRegexCriterion</span> =~ <span class="q">/^ne$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> =~ <span class="q">/$SpecifiedRegex/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
377 364 <span class="i">$Nothing</span> = <span class="n">1</span><span class="sc">;</span>
|
|
378 365 <span class="s">}</span>
|
|
379 366 <span class="s">}</span>
|
|
380 367 <span class="k">else</span> <span class="s">{</span>
|
|
381 368 <span class="j">CRITERION:</span> <span class="s">{</span>
|
|
382 369 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedRegexCriterion</span> =~ <span class="q">/^eq$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> !~ <span class="q">/$SpecifiedRegex/</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
383 370 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedRegexCriterion</span> =~ <span class="q">/^ne$/i</span><span class="s">)</span> <span class="s">{</span> <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentValue</span> =~ <span class="q">/$SpecifiedRegex/</span><span class="s">)</span> <span class="s">{</span> <span class="i">$ViolationCount</span>++<span class="sc">;</span> <span class="k">last</span> <span class="j">CRITERION</span><span class="sc">;</span> <span class="s">}</span> <span class="s">}</span>
|
|
384 371 <span class="i">$Nothing</span> = <span class="n">1</span><span class="sc">;</span>
|
|
385 372 <span class="s">}</span>
|
|
386 373 <span class="s">}</span>
|
|
387 374 <span class="s">}</span>
|
|
388 375 <span class="s">}</span>
|
|
389 376 <span class="k">if</span> <span class="s">(</span><span class="i">$ViolationCount</span> <= <span class="i">$OptionsInfo</span>{<span class="w">Violations</span>}<span class="s">)</span> <span class="s">{</span>
|
|
390 377 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
391 378 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
392 379 <span class="s">}</span>
|
|
393 380 <span class="s">}</span>
|
|
394 381 <span class="s">}</span>
|
|
395 382
|
|
396 383 <span class="c"># Extract random compounds...</span>
|
|
397 <a name="ExtractRandomCompounds-"></a> 384 <span class="k">sub </span><span class="m">ExtractRandomCompounds</span> <span class="s">{</span>
|
|
398 385 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
399 386 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdNum</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$RandomCycleCount</span><span class="cm">,</span> <span class="i">$RandomIndex</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">%RandomCmpdIndexMap</span><span class="s">)</span><span class="sc">;</span>
|
|
400 387
|
|
401 388 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
402 389 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
403 390
|
|
404 391 <span class="i">$CmpdCount</span> = <span class="i">$SDFilesInfo</span>{<span class="w">CmpdCount</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
405 392 <span class="k">srand</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Seed</span>}<span class="s">)</span><span class="sc">;</span>
|
|
406 393 <span class="i">$RandomCycleCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
407 394
|
|
408 395 <span class="i">%RandomCmpdIndexMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
409 396 <span class="k">while</span> <span class="s">(</span><span class="i">$RandomCycleCount</span> <= <span class="i">$CmpdCount</span> && <span class="i">$RandomCycleCount</span> <= <span class="i">$OptionsInfo</span>{<span class="w">NumOfCmpds</span>}<span class="s">)</span> <span class="s">{</span>
|
|
410 397 <span class="i">$RandomCycleCount</span>++<span class="sc">;</span>
|
|
411 398 <span class="i">$RandomIndex</span> = <span class="k">int</span> <span class="s">(</span><span class="k">rand</span> <span class="i">$CmpdCount</span><span class="s">)</span> + <span class="n">1</span><span class="sc">;</span>
|
|
412 399 <span class="i">$RandomCmpdIndexMap</span>{<span class="i">$RandomIndex</span>} = <span class="i">$RandomIndex</span><span class="sc">;</span>
|
|
413 400 <span class="s">}</span>
|
|
414 401
|
|
415 402 <span class="i">$CmpdNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
416 403 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
417 404 <span class="i">$CmpdNum</span>++<span class="sc">;</span>
|
|
418 405 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$RandomCmpdIndexMap</span>{<span class="i">$CmpdNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
419 406 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
420 407 <span class="s">}</span>
|
|
421 408
|
|
422 409 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
423 410
|
|
424 411 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
425 412
|
|
426 413 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
427 414 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
428 415 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
429 416 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
430 417 <span class="s">}</span>
|
|
431 418 <span class="s">}</span>
|
|
432 419 <span class="s">}</span>
|
|
433 420
|
|
434 421 <span class="c"># Extract mol names...</span>
|
|
435 <a name="ExtractMolNames-"></a> 422 <span class="k">sub </span><span class="m">ExtractMolNames</span> <span class="s">{</span>
|
|
436 423 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
437 424 <span class="k">my</span><span class="s">(</span><span class="i">$MolName</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
438 425
|
|
439 426 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}}<span class="cm">,</span> <span class="q">"MolName"</span><span class="sc">;</span>
|
|
440 427 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
441 428
|
|
442 429 <span class="i">$NewTextFileRef</span> = <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>}<span class="sc">;</span>
|
|
443 430 <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
444 431 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
445 432 <span class="i">$MolName</span> = <span class="i">QuoteAWord</span><span class="s">(</span><span class="i">ParseCmpdMolNameLine</span><span class="s">(</span><span class="i">$CmpdLines</span>[<span class="n">0</span>]<span class="s">)</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
446 433 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$MolName\n"</span><span class="sc">;</span>
|
|
447 434 <span class="s">}</span>
|
|
448 435 <span class="s">}</span>
|
|
449 436
|
|
450 437 <span class="c"># Extract a specific compound record...</span>
|
|
451 <a name="ExtractRecordNum-"></a> 438 <span class="k">sub </span><span class="m">ExtractRecordNum</span> <span class="s">{</span>
|
|
452 439 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
453 440 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdNum</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
454 441
|
|
455 442 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
456 443 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
457 444
|
|
458 445 <span class="i">$CmpdNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
459 446
|
|
460 447 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
461 448 <span class="i">$CmpdNum</span>++<span class="sc">;</span>
|
|
462 449 <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdNum</span> != <span class="i">$OptionsInfo</span>{<span class="w">RecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
463 450 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
464 451 <span class="s">}</span>
|
|
465 452
|
|
466 453 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
467 454 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
468 455
|
|
469 456 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
470 457 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
471 458 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
472 459 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
473 460 <span class="s">}</span>
|
|
474 461 <span class="k">last</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
475 462 <span class="s">}</span>
|
|
476 463 <span class="s">}</span>
|
|
477 464
|
|
478 465 <span class="c"># Extract a specific compound records...</span>
|
|
479 <a name="ExtractRecordNums-"></a> 466 <span class="k">sub </span><span class="m">ExtractRecordNums</span> <span class="s">{</span>
|
|
480 467 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
481 468 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdNum</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
482 469
|
|
483 470 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
484 471 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
485 472
|
|
486 473 <span class="i">$CmpdNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
487 474 <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
488 475
|
|
489 476 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
490 477 <span class="i">$CmpdNum</span>++<span class="sc">;</span>
|
|
491 478
|
|
492 479 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">RecordNums</span>}{<span class="i">$CmpdNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
493 480 <span class="i">$CmpdCount</span>++<span class="sc">;</span>
|
|
494 481 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
495 482
|
|
496 483 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
497 484
|
|
498 485 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
499 486 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
500 487 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
501 488 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
502 489 <span class="s">}</span>
|
|
503 490 <span class="s">}</span>
|
|
504 491 <span class="k">elsif</span> <span class="s">(</span><span class="i">$CmpdNum</span> > <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsMax</span>} || <span class="i">$CmpdCount</span> >= <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsCount</span>}<span class="s">)</span> <span class="s">{</span>
|
|
505 492 <span class="k">last</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
506 493 <span class="s">}</span>
|
|
507 494 <span class="s">}</span>
|
|
508 495 <span class="s">}</span>
|
|
509 496
|
|
510 497
|
|
511 498 <span class="c"># Extract compounds in a specific record range...</span>
|
|
512 <a name="ExtractRecordRange-"></a> 499 <span class="k">sub </span><span class="m">ExtractRecordRange</span> <span class="s">{</span>
|
|
513 500 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
514 501 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdNum</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
515 502
|
|
516 503 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
517 504 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
518 505
|
|
519 506 <span class="i">$CmpdNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
520 507 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
521 508 <span class="i">$CmpdNum</span>++<span class="sc">;</span>
|
|
522 509
|
|
523 510 <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdNum</span> >= <span class="i">$OptionsInfo</span>{<span class="w">StartRecordNum</span>} && <span class="i">$CmpdNum</span> <= <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
524 511 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
525 512
|
|
526 513 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
527 514
|
|
528 515 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
529 516 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
530 517 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
531 518 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
532 519 <span class="s">}</span>
|
|
533 520 <span class="s">}</span>
|
|
534 521 <span class="k">elsif</span> <span class="s">(</span><span class="i">$CmpdNum</span> > <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
535 522 <span class="k">last</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
536 523 <span class="s">}</span>
|
|
537 524 <span class="s">}</span>
|
|
538 525 <span class="s">}</span>
|
|
539 526
|
|
540 527 <span class="c"># Extract 2D compound records...</span>
|
|
541 <a name="Extract2DCmpdRecords-"></a> 528 <span class="k">sub </span><span class="m">Extract2DCmpdRecords</span> <span class="s">{</span>
|
|
542 529 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
543 530 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
544 531
|
|
545 532 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
546 533 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
547 534
|
|
548 535
|
|
549 536 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
550 537 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
551 538 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsCmpd2D</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
552 539 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
553 540 <span class="s">}</span>
|
|
554 541
|
|
555 542 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
556 543
|
|
557 544 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
558 545 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
559 546 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
560 547 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
561 548 <span class="s">}</span>
|
|
562 549 <span class="s">}</span>
|
|
563 550 <span class="s">}</span>
|
|
564 551
|
|
565 552 <span class="c"># Extract 3D compound records...</span>
|
|
566 <a name="Extract3DCmpdRecords-"></a> 553 <span class="k">sub </span><span class="m">Extract3DCmpdRecords</span> <span class="s">{</span>
|
|
567 554 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
568 555 <span class="k">my</span><span class="s">(</span><span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
569 556
|
|
570 557 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}} = <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$FileIndex</span>]}<span class="sc">;</span>
|
|
571 558 <span class="i">WriteTextFileColLabels</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
572 559
|
|
573 560
|
|
574 561 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>} = <span class="i">ReadCmpdString</span><span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
575 562 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
576 563 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsCmpd3D</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
577 564 <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
578 565 <span class="s">}</span>
|
|
579 566
|
|
580 567 <span class="i">WriteSDFileCmpdString</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
581 568
|
|
582 569 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
583 570 <span class="i">%</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}} = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
584 571 <span class="i">SetupDataValues</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
585 572 <span class="i">WriteTextFileCmpdData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
586 573 <span class="s">}</span>
|
|
587 574 <span class="s">}</span>
|
|
588 575 <span class="s">}</span>
|
|
589 576
|
|
590 577
|
|
591 578 <span class="c"># Open input and output files...</span>
|
|
592 <a name="OpenInputAndOutputFiles-"></a> 579 <span class="k">sub </span><span class="m">OpenInputAndOutputFiles</span> <span class="s">{</span>
|
|
593 580 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
594 581
|
|
595 582 <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
596 583 <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
597 584
|
|
598 585 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>} && <span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
599 586 <span class="k">print</span> <span class="q">"Generating files $SDFilesInfo{NewSDFileName}[$FileIndex] and $SDFilesInfo{NewTextFileName}[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
600 587 <span class="s">}</span>
|
|
601 588 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
602 589 <span class="k">print</span> <span class="q">"Generating file $SDFilesInfo{NewSDFileName}[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
603 590 <span class="s">}</span>
|
|
604 591 <span class="k">else</span> <span class="s">{</span>
|
|
605 592 <span class="k">print</span> <span class="q">"Generating file $SDFilesInfo{NewTextFileName}[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
606 593 <span class="s">}</span>
|
|
607 594
|
|
608 595 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
609 596 <span class="k">open</span> <span class="w">NEWSDFILE</span><span class="cm">,</span> <span class="q">">$SDFilesInfo{NewSDFileName}[$FileIndex]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $SDFilesInfo{NewSDFileName}[$FileIndex]: $! \n"</span><span class="sc">;</span>
|
|
610 597 <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>} = \<span class="i">*NEWSDFILE</span><span class="sc">;</span>
|
|
611 598 <span class="s">}</span>
|
|
612 599 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
613 600 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$SDFilesInfo{NewTextFileName}[$FileIndex]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $SDFilesInfo{NewTextFileName}[$FileIndex]: $! \n"</span><span class="sc">;</span>
|
|
614 601 <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>} = \<span class="i">*NEWTEXTFILE</span><span class="sc">;</span>
|
|
615 602 <span class="s">}</span>
|
|
616 603
|
|
617 604 <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFilesList[$FileIndex]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $SDFilesList[$FileIndex]: $! \n"</span><span class="sc">;</span>
|
|
618 605 <span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>} = \<span class="i">*SDFILE</span><span class="sc">;</span>
|
|
619 606
|
|
620 607 <span class="s">}</span>
|
|
621 608
|
|
622 609 <span class="c"># Close open input and output files...</span>
|
|
623 <a name="CloseInputAndOutputFiles-"></a> 610 <span class="k">sub </span><span class="m">CloseInputAndOutputFiles</span> <span class="s">{</span>
|
|
624 611 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
625 612 <span class="k">close</span> <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>}<span class="sc">;</span>
|
|
626 613 <span class="s">}</span>
|
|
627 614 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
628 615 <span class="k">close</span> <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>}<span class="sc">;</span>
|
|
629 616 <span class="s">}</span>
|
|
630 617
|
|
631 618 <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
632 619 <span class="k">close</span> <span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>}<span class="sc">;</span>
|
|
633 620 <span class="s">}</span>
|
|
634 621
|
|
635 622 <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
636 623 <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
637 624 <span class="i">$SDFilesInfo</span>{<span class="w">InputSDFileRef</span>} = <span class="k">undef</span><span class="sc">;</span>
|
|
638 625 <span class="s">}</span>
|
|
639 626
|
|
640 627 <span class="c"># Write out column labels for text file...</span>
|
|
641 <a name="WriteTextFileColLabels-"></a> 628 <span class="k">sub </span><span class="m">WriteTextFileColLabels</span> <span class="s">{</span>
|
|
642 629 <span class="k">my</span><span class="s">(</span><span class="i">$ColLabelsLine</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span><span class="sc">;</span>
|
|
643 630
|
|
644 631 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
645 632 <span class="k">return</span><span class="sc">;</span>
|
|
646 633 <span class="s">}</span>
|
|
647 634 <span class="i">$NewTextFileRef</span> = <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>}<span class="sc">;</span>
|
|
648 635
|
|
649 636 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputStrDataString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
650 637 <span class="c"># Append structure data string label...</span>
|
|
651 638 <span class="k">my</span><span class="s">(</span><span class="i">@DataLabels</span><span class="s">)</span><span class="sc">;</span>
|
|
652 639
|
|
653 640 <span class="i">@DataLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
654 641 <span class="k">push</span> <span class="i">@DataLabels</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}}<span class="sc">;</span>
|
|
655 642 <span class="k">push</span> <span class="i">@DataLabels</span><span class="cm">,</span> <span class="q">"StructureDataString"</span><span class="sc">;</span>
|
|
656 643
|
|
657 644 <span class="i">$ColLabelsLine</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@DataLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
658 645 <span class="s">}</span>
|
|
659 646 <span class="k">else</span> <span class="s">{</span>
|
|
660 647 <span class="i">$ColLabelsLine</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
661 648 <span class="s">}</span>
|
|
662 649 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$ColLabelsLine\n"</span><span class="sc">;</span>
|
|
663 650 <span class="s">}</span>
|
|
664 651
|
|
665 652 <span class="c"># Setup values for data fields...</span>
|
|
666 <a name="SetupDataValues-"></a> 653 <span class="k">sub </span><span class="m">SetupDataValues</span> <span class="s">{</span>
|
|
667 654 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataValues</span>}} = <span class="k">map</span> <span class="s">{</span> <span class="k">exists</span> <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$_</span>} ? <span class="i">$SDFilesInfo</span>{<span class="w">DataFieldValues</span>}{<span class="i">$_</span>} <span class="co">:</span> <span class="q">""</span> <span class="s">}</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}}<span class="sc">;</span>
|
|
668 655 <span class="s">}</span>
|
|
669 656
|
|
670 657 <span class="c"># Write out structure data and specific data fields to SD file...</span>
|
|
671 <a name="WriteSDFileCmpdData-"></a> 658 <span class="k">sub </span><span class="m">WriteSDFileCmpdData</span> <span class="s">{</span>
|
|
672 659 <span class="k">my</span><span class="s">(</span><span class="i">$MolString</span><span class="cm">,</span> <span class="i">$Count</span><span class="cm">,</span> <span class="i">$NewSDFileRef</span><span class="s">)</span><span class="sc">;</span>
|
|
673 660
|
|
674 661 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
675 662 <span class="k">return</span><span class="sc">;</span>
|
|
676 663 <span class="s">}</span>
|
|
677 664
|
|
678 665 <span class="i">$NewSDFileRef</span> = <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>}<span class="sc">;</span>
|
|
679 666
|
|
680 667 <span class="s">(</span><span class="i">$MolString</span><span class="s">)</span> = <span class="k">split</span> <span class="q">"M END"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
681 668 <span class="i">$MolString</span> .= <span class="q">"M END"</span><span class="sc">;</span>
|
|
682 669 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">"$MolString\n"</span><span class="sc">;</span>
|
|
683 670
|
|
684 671 <span class="k">for</span> <span class="i">$Count</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataLabels</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
685 672 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">"> <$SDFilesInfo{DataLabels}[$Count]>\n$SDFilesInfo{DataValues}[$Count]\n\n"</span><span class="sc">;</span>
|
|
686 673 <span class="s">}</span>
|
|
687 674 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">"\$\$\$\$\n"</span><span class="sc">;</span>
|
|
688 675 <span class="s">}</span>
|
|
689 676
|
|
690 677 <span class="c"># Write out compound string...</span>
|
|
691 <a name="WriteSDFileCmpdString-"></a> 678 <span class="k">sub </span><span class="m">WriteSDFileCmpdString</span> <span class="s">{</span>
|
|
692 679 <span class="k">my</span><span class="s">(</span><span class="i">$NewSDFileRef</span><span class="s">)</span><span class="sc">;</span>
|
|
693 680
|
|
694 681 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
695 682 <span class="k">return</span><span class="sc">;</span>
|
|
696 683 <span class="s">}</span>
|
|
697 684
|
|
698 685 <span class="i">$NewSDFileRef</span> = <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileRef</span>}<span class="sc">;</span>
|
|
699 686 <span class="k">print</span> <span class="i">$NewSDFileRef</span> <span class="q">"$SDFilesInfo{CmpdString}\n"</span><span class="sc">;</span>
|
|
700 687 <span class="s">}</span>
|
|
701 688
|
|
702 689 <span class="c"># Write out data for text file...</span>
|
|
703 <a name="WriteTextFileCmpdData-"></a> 690 <span class="k">sub </span><span class="m">WriteTextFileCmpdData</span> <span class="s">{</span>
|
|
704 691 <span class="k">my</span><span class="s">(</span><span class="i">$DataValuesLine</span><span class="cm">,</span> <span class="i">$NewTextFileRef</span><span class="s">)</span><span class="sc">;</span>
|
|
705 692
|
|
706 693 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
707 694 <span class="k">return</span><span class="sc">;</span>
|
|
708 695 <span class="s">}</span>
|
|
709 696
|
|
710 697 <span class="i">$NewTextFileRef</span> = <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileRef</span>}<span class="sc">;</span>
|
|
711 698 <span class="i">$DataValuesLine</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">DataValues</span>}}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
712 699
|
|
713 700 <span class="c"># Handle multiple lines data values for data fields by joining 'em using semicolons...</span>
|
|
714 701 <span class="k">if</span> <span class="s">(</span><span class="i">$DataValuesLine</span> =~ <span class="q">/\n/</span><span class="s">)</span> <span class="s">{</span>
|
|
715 702 <span class="i">$DataValuesLine</span> =~ <span class="q">s/\n/;/g</span><span class="sc">;</span>
|
|
716 703 <span class="s">}</span>
|
|
717 704
|
|
718 705 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputStrDataString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
719 706 <span class="c"># Append structure data string...</span>
|
|
720 707 <span class="k">my</span><span class="s">(</span><span class="i">$StrDataString</span><span class="cm">,</span> <span class="i">$OutQuoteValue</span><span class="cm">,</span> <span class="i">$OutDelim</span><span class="cm">,</span> <span class="i">$StrDataStringDelimiter</span><span class="s">)</span><span class="sc">;</span>
|
|
721 708
|
|
722 709 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StrDataStringWithFields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
723 710 <span class="i">$StrDataString</span> = <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
724 711 <span class="s">}</span>
|
|
725 712 <span class="k">else</span> <span class="s">{</span>
|
|
726 713 <span class="s">(</span><span class="i">$StrDataString</span><span class="s">)</span> = <span class="k">split</span> <span class="q">"M END"</span><span class="cm">,</span> <span class="i">$SDFilesInfo</span>{<span class="w">CmpdString</span>}<span class="sc">;</span>
|
|
727 714 <span class="i">$StrDataString</span> .= <span class="q">"M END"</span><span class="sc">;</span>
|
|
728 715 <span class="s">}</span>
|
|
729 716 <span class="i">$StrDataStringDelimiter</span> = <span class="i">$OptionsInfo</span>{<span class="w">StrDataStringDelimiter</span>}<span class="sc">;</span>
|
|
730 717 <span class="i">$StrDataString</span> =~ <span class="q">s/\n/$StrDataStringDelimiter/g</span><span class="sc">;</span>
|
|
731 718
|
|
732 719 <span class="i">$OutDelim</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="sc">;</span>
|
|
733 720 <span class="i">$OutQuoteValue</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} ? <span class="q">"\""</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
734 721
|
|
735 722 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$DataValuesLine${OutDelim}${OutQuoteValue}${StrDataString}${OutQuoteValue}\n"</span><span class="sc">;</span>
|
|
736 723 <span class="s">}</span>
|
|
737 724 <span class="k">else</span> <span class="s">{</span>
|
|
738 725 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$DataValuesLine\n"</span><span class="sc">;</span>
|
|
739 726 <span class="s">}</span>
|
|
740 727 <span class="s">}</span>
|
|
741 728
|
|
742 729 <span class="c"># Retrieve information about input SD files...</span>
|
|
743 <a name="RetrieveSDFilesInfo-"></a> 730 <span class="k">sub </span><span class="m">RetrieveSDFilesInfo</span> <span class="s">{</span>
|
|
744 731 <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$NewFileName</span><span class="cm">,</span> <span class="i">$NewSDFileName</span><span class="cm">,</span> <span class="i">$NewTextFileName</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="s">)</span><span class="sc">;</span>
|
|
745 732
|
|
746 733 <span class="i">%SDFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
747 734
|
|
748 735 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
749 736 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CmpdCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
750 737 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileName</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
751 738 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileName</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
752 739
|
|
753 740 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
754 741 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
755 742
|
|
756 743 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
757 744 <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
758 745
|
|
759 746 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
760 747
|
|
761 748 <span class="i">$SDFilesInfo</span>{<span class="w">CmpdCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
762 749 <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileName</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
763 750 <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileName</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
764 751
|
|
765 752 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
766 753 <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
767 754
|
|
768 755 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$SDFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
769 756 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It doesn't exist\n"</span><span class="sc">;</span>
|
|
770 757 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
771 758 <span class="s">}</span>
|
|
772 759
|
|
773 760 <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="q">"sd sdf"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
774 761 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It's not a SD file\n"</span><span class="sc">;</span>
|
|
775 762 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
776 763 <span class="s">}</span>
|
|
777 764
|
|
778 765 <span class="c"># Generate appropriate name for the new output file.</span>
|
|
779 766 <span class="i">$FileDir</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
780 767 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$SDFile</span><span class="s">)</span><span class="sc">;</span>
|
|
781 768 <span class="i">$NewFileName</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
782 769 <span class="i">$NewFileName</span> = <span class="i">$FileName</span> . <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>}<span class="sc">;</span>
|
|
783 770 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} && <span class="s">(</span><span class="i">@SDFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
784 771 <span class="k">my</span> <span class="s">(</span><span class="i">$RootFileDir</span><span class="cm">,</span> <span class="i">$RootFileName</span><span class="cm">,</span> <span class="i">$RootFileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span>
|
|
785 772 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
786 773 <span class="i">$NewFileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
787 774 <span class="s">}</span>
|
|
788 775 <span class="k">else</span> <span class="s">{</span>
|
|
789 776 <span class="i">$NewFileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
|
|
790 777 <span class="s">}</span>
|
|
791 778 <span class="s">}</span>
|
|
792 779 <span class="i">$NewSDFileName</span> = <span class="i">$NewFileName</span> . <span class="q">".$OptionsInfo{SDFileExt}"</span><span class="sc">;</span>
|
|
793 780 <span class="i">$NewTextFileName</span> = <span class="i">$NewFileName</span> . <span class="q">".$OptionsInfo{TextFileExt}"</span><span class="sc">;</span>
|
|
794 781
|
|
795 782 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
796 783 <span class="k">if</span> <span class="s">(</span><span class="k">lc</span><span class="s">(</span><span class="i">$NewSDFileName</span><span class="s">)</span> <span class="k">eq</span> <span class="k">lc</span><span class="s">(</span><span class="i">$SDFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
797 784 <span class="k">warn</span> <span class="q">"Warning: Ignoring input file $SDFile: Same output, $NewSDFileName, and input file names.\n"</span><span class="sc">;</span>
|
|
798 785 <span class="k">print</span> <span class="q">"Specify a different name using \"-r --root\" option or use default name.\n"</span><span class="sc">;</span>
|
|
799 786 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
800 787 <span class="s">}</span>
|
|
801 788 <span class="s">}</span>
|
|
802 789
|
|
803 790 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">Overwrite</span>}<span class="s">)</span> <span class="s">{</span>
|
|
804 791 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
805 792 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewSDFileName</span><span class="s">)</span> <span class="s">{</span>
|
|
806 793 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: New file, $NewSDFileName, already exists\n"</span><span class="sc">;</span>
|
|
807 794 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
808 795 <span class="s">}</span>
|
|
809 796 <span class="s">}</span>
|
|
810 797 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
811 798 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewTextFileName</span><span class="s">)</span> <span class="s">{</span>
|
|
812 799 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: New file, $NewTextFileName, already exists\n"</span><span class="sc">;</span>
|
|
813 800 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
814 801 <span class="s">}</span>
|
|
815 802 <span class="s">}</span>
|
|
816 803 <span class="s">}</span>
|
|
817 804
|
|
818 805 <span class="k">if</span> <span class="s">(</span>!<span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFile"</span><span class="s">)</span> <span class="s">{</span>
|
|
819 806 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Couldn't open it: $! \n"</span><span class="sc">;</span>
|
|
820 807 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
821 808 <span class="s">}</span>
|
|
822 809
|
|
823 810 <span class="k">my</span><span class="s">(</span><span class="i">$CountCmpds</span><span class="cm">,</span> <span class="i">$CollectDataFields</span><span class="s">)</span><span class="sc">;</span>
|
|
824 811 <span class="k">my</span><span class="s">(</span><span class="i">$CmpdString</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">@DataFieldLabels</span><span class="cm">,</span> <span class="i">%DataFieldLabelsMap</span><span class="cm">,</span><span class="i">@CommonDataFieldLabels</span><span class="s">)</span><span class="sc">;</span>
|
|
825 812
|
|
826 813 <span class="i">$CountCmpds</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^randomcmpds$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
827 814
|
|
828 815 <span class="i">$CollectDataFields</span> = <span class="s">(</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^(alldatafields|commondatafields|randomcmpds)$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^(datafieldsbyvalue|datafieldsbyregex)$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^datafieldbylist$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^datafielduniquebylist$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^datafieldnotbylist$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^recordnum$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^recordnums$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span> || <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^recordrange$/i</span> && <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>}<span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
829 816
|
|
830 817 <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
831 818 <span class="k">if</span> <span class="s">(</span><span class="i">$CountCmpds</span> || <span class="i">$CollectDataFields</span><span class="s">)</span> <span class="s">{</span>
|
|
832 819 <span class="i">@DataFieldLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
833 820 <span class="i">@CommonDataFieldLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
834 821 <span class="i">%DataFieldLabelsMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
835 822 <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$CmpdString</span> = <span class="i">ReadCmpdString</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
836 823 <span class="i">$CmpdCount</span>++<span class="sc">;</span>
|
|
837 824 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^recordnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
838 825 <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdCount</span> == <span class="i">$OptionsInfo</span>{<span class="w">RecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
839 826 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span>
|
|
840 827 <span class="i">@DataFieldLabels</span> = <span class="i">GetCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
841 828 <span class="k">last</span> <span class="j">CMPDSTRING</span><span class="sc">;</span>
|
|
842 829 <span class="s">}</span>
|
|
843 830 <span class="s">}</span>
|
|
844 831 <span class="k">if</span> <span class="s">(</span><span class="i">$CollectDataFields</span><span class="s">)</span> <span class="s">{</span>
|
|
845 832 <span class="k">my</span><span class="s">(</span><span class="i">$Label</span><span class="s">)</span><span class="sc">;</span>
|
|
846 833 <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span>
|
|
847 834 <span class="c"># Process compound data header labels and figure out which ones are present for</span>
|
|
848 835 <span class="c"># all the compounds...</span>
|
|
849 836 <span class="k">if</span> <span class="s">(</span><span class="i">@DataFieldLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
850 837 <span class="k">my</span> <span class="s">(</span><span class="i">@CmpdDataFieldLabels</span><span class="s">)</span> = <span class="i">GetCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
851 838 <span class="k">my</span><span class="s">(</span><span class="i">%CmpdDataFieldLabelsMap</span><span class="s">)</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
852 839 <span class="c"># Setup a map for the current labels...</span>
|
|
853 840 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@CmpdDataFieldLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
854 841 <span class="i">$CmpdDataFieldLabelsMap</span>{<span class="i">$Label</span>} = <span class="q">"PresentInSome"</span><span class="sc">;</span>
|
|
855 842 <span class="s">}</span>
|
|
856 843 <span class="c"># Check the presence old labels for this compound; otherwise, mark 'em new...</span>
|
|
857 844 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@DataFieldLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
858 845 <span class="k">if</span> <span class="s">(</span>!<span class="i">$CmpdDataFieldLabelsMap</span>{<span class="i">$Label</span>}<span class="s">)</span> <span class="s">{</span>
|
|
859 846 <span class="i">$DataFieldLabelsMap</span>{<span class="i">$Label</span>} = <span class="q">"PresentInSome"</span><span class="sc">;</span>
|
|
860 847 <span class="s">}</span>
|
|
861 848 <span class="s">}</span>
|
|
862 849 <span class="c"># Check the presence this compound in the old labels; otherwise, add 'em...</span>
|
|
863 850 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@CmpdDataFieldLabels</span> <span class="s">)</span> <span class="s">{</span>
|
|
864 851 <span class="k">if</span> <span class="s">(</span>!<span class="i">$DataFieldLabelsMap</span>{<span class="i">$Label</span>}<span class="s">)</span> <span class="s">{</span>
|
|
865 852 <span class="c"># It's a new label...</span>
|
|
866 853 <span class="k">push</span> <span class="i">@DataFieldLabels</span><span class="cm">,</span> <span class="i">$Label</span><span class="sc">;</span>
|
|
867 854 <span class="i">$DataFieldLabelsMap</span>{<span class="i">$Label</span>} = <span class="q">"PresentInSome"</span><span class="sc">;</span>
|
|
868 855 <span class="s">}</span>
|
|
869 856 <span class="s">}</span>
|
|
870 857 <span class="s">}</span>
|
|
871 858 <span class="k">else</span> <span class="s">{</span>
|
|
872 859 <span class="c"># Get the initial label set and set up a map...</span>
|
|
873 860 <span class="i">@DataFieldLabels</span> = <span class="i">GetCmpdDataHeaderLabels</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span>
|
|
874 861 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@DataFieldLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
875 862 <span class="i">$DataFieldLabelsMap</span>{<span class="i">$Label</span>} = <span class="q">"PresentInAll"</span><span class="sc">;</span>
|
|
876 863 <span class="s">}</span>
|
|
877 864 <span class="s">}</span>
|
|
878 865 <span class="c"># Identify the common data field labels...</span>
|
|
879 866 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^commondatafields$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
880 867 <span class="i">@CommonDataFieldLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
881 868 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@DataFieldLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
882 869 <span class="k">if</span> <span class="s">(</span><span class="i">$DataFieldLabelsMap</span>{<span class="i">$Label</span>} <span class="k">eq</span> <span class="q">"PresentInAll"</span><span class="s">)</span> <span class="s">{</span>
|
|
883 870 <span class="k">push</span> <span class="i">@CommonDataFieldLabels</span><span class="cm">,</span> <span class="i">$Label</span><span class="sc">;</span>
|
|
884 871 <span class="s">}</span>
|
|
885 872 <span class="s">}</span>
|
|
886 873 <span class="s">}</span>
|
|
887 874 <span class="s">}</span>
|
|
888 875 <span class="s">}</span>
|
|
889 876 <span class="s">}</span>
|
|
890 877
|
|
891 878 <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
892 879
|
|
893 880 <span class="i">$SDFilesInfo</span>{<span class="w">NewTextFileName</span>}[<span class="i">$Index</span>] = <span class="i">$NewTextFileName</span><span class="sc">;</span>
|
|
894 881 <span class="i">$SDFilesInfo</span>{<span class="w">NewSDFileName</span>}[<span class="i">$Index</span>] = <span class="i">$NewSDFileName</span><span class="sc">;</span>
|
|
895 882
|
|
896 883 <span class="i">$SDFilesInfo</span>{<span class="w">CmpdCount</span>}[<span class="i">$Index</span>] = <span class="i">$CmpdCount</span><span class="sc">;</span>
|
|
897 884
|
|
898 885 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">AllDataFieldLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@DataFieldLabels</span><span class="sc">;</span>
|
|
899 886 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">CommonDataFieldLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@CommonDataFieldLabels</span><span class="sc">;</span>
|
|
900 887
|
|
901 888 <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span>
|
|
902 889 <span class="s">}</span>
|
|
903 890 <span class="s">}</span>
|
|
904 891
|
|
905 892 <span class="c"># Process options...</span>
|
|
906 <a name="ProcessOptions-"></a> 893 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
907 894 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
908 895
|
|
909 896 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
910 897
|
|
911 898 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="q">"\,"</span><span class="sc">;</span>
|
|
912 899 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} =~ <span class="q">/^semicolon$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
913 900 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="q">"\;"</span><span class="sc">;</span>
|
|
914 901 <span class="s">}</span>
|
|
915 902 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
916 903 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="q">"\t"</span><span class="sc">;</span>
|
|
917 904 <span class="s">}</span>
|
|
918 905
|
|
919 906 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="q">"\,"</span><span class="sc">;</span>
|
|
920 907 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^semicolon$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
921 908 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="q">"\;"</span><span class="sc">;</span>
|
|
922 909 <span class="s">}</span>
|
|
923 910 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
924 911 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="q">"\t"</span><span class="sc">;</span>
|
|
925 912 <span class="s">}</span>
|
|
926 913
|
|
927 914 <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/^yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
928 915
|
|
929 916 <span class="i">$OptionsInfo</span>{<span class="w">RegexIgnoreCase</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">regexignorecase</span>} =~ <span class="q">/^yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
930 917
|
|
931 918 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
932 919 <span class="i">$OptionsInfo</span>{<span class="w">Overwrite</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="i">$Options</span>{<span class="w">overwrite</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
933 920
|
|
934 921 <span class="i">$OptionsInfo</span>{<span class="w">NumOfCmpds</span>} = <span class="i">$Options</span>{<span class="w">numofcmpds</span>}<span class="sc">;</span>
|
|
935 922
|
|
936 923 <span class="i">$OptionsInfo</span>{<span class="w">ValueComparisonMode</span>} = <span class="i">$Options</span>{<span class="w">valuecomparisonmode</span>}<span class="sc">;</span>
|
|
937 924 <span class="i">$OptionsInfo</span>{<span class="w">NumericalComparison</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">valuecomparisonmode</span>} =~ <span class="q">/^Numeric$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
938 925
|
|
939 926 <span class="i">$OptionsInfo</span>{<span class="w">Violations</span>} = <span class="i">$Options</span>{<span class="w">violations</span>}<span class="sc">;</span>
|
|
940 927 <span class="i">$OptionsInfo</span>{<span class="w">Seed</span>} = <span class="i">$Options</span>{<span class="w">seed</span>}<span class="sc">;</span>
|
|
941 928
|
|
942 929
|
|
943 930 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(datafields|datafieldsbyregex|datafieldsbyvalue|datafieldbylist|datafielduniquebylist|datafieldnotbylist)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
944 931 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>} || <span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
945 932 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>} && <span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
946 933 <span class="k">die</span> <span class="q">"Error: For \"-m --mode\" option values of datafields, datafieldsbyvalue, datafieldsbyregex, datafieldbylist, datafielduniquebylist, or datafieldnotbylist specify only one of the \"-d --datafields\" or \"--datafieldsfile\" option.\n"</span><span class="sc">;</span>
|
|
947 934 <span class="s">}</span>
|
|
948 935 <span class="s">}</span>
|
|
949 936 <span class="k">else</span> <span class="s">{</span>
|
|
950 937 <span class="k">die</span> <span class="q">"Error: For \"-m --mode\" option values of datafields, datafieldsbyvalue, datafieldsbyregex, datafieldbylist, datafielduniquebylist, or datafieldnotbylist specify one of the \"-d --datafields\" or \"--datafieldsfile\" option.\n"</span><span class="sc">;</span>
|
|
951 938 <span class="s">}</span>
|
|
952 939 <span class="s">}</span>
|
|
953 940 <span class="i">$OptionsInfo</span>{<span class="w">DataFields</span>} = <span class="i">$Options</span>{<span class="w">datafields</span>} ? <span class="i">$Options</span>{<span class="w">datafields</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
954 941 <span class="i">$OptionsInfo</span>{<span class="w">DataFieldsFile</span>} = <span class="i">$Options</span>{<span class="w">datafieldsfile</span>} ? <span class="i">$Options</span>{<span class="w">datafieldsfile</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
955 942
|
|
956 943 <span class="i">$OptionsInfo</span>{<span class="w">RecordNum</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">StartRecordNum</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
957 944
|
|
958 945 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">RecordNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
959 946 <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsMin</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsMax</span>} = <span class="n">0</span><span class="sc">;</span> <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsCount</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
960 947
|
|
961 948 <span class="i">$OptionsInfo</span>{<span class="w">Record</span>} = <span class="i">$Options</span>{<span class="w">record</span>} ? <span class="i">$Options</span>{<span class="w">record</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
962 949
|
|
963 950 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(recordnum|recordnums|recordrange)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
964 951 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">record</span>}<span class="s">)</span> <span class="s">{</span>
|
|
965 952 <span class="k">my</span><span class="s">(</span><span class="i">$Record</span><span class="cm">,</span> <span class="i">@RecordSplit</span><span class="s">)</span><span class="sc">;</span>
|
|
966 953
|
|
967 954 <span class="i">$Record</span> = <span class="i">$Options</span>{<span class="w">record</span>}<span class="sc">;</span>
|
|
968 955 <span class="i">$Record</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
969 956
|
|
970 957 <span class="i">@RecordSplit</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Record</span><span class="sc">;</span>
|
|
971 958
|
|
972 959 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^recordnum$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
973 960 <span class="k">if</span> <span class="s">(</span><span class="i">@RecordSplit</span> == <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
974 961 <span class="i">$OptionsInfo</span>{<span class="w">RecordNum</span>} = <span class="i">$RecordSplit</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
975 962 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">RecordNum</span>} <= <span class="n">0</span><span class="s">)</span> <span class="s">{</span>
|
|
976 963 <span class="k">die</span> <span class="q">"Error: The value specified, $OptionsInfo{RecordNum}, for option \"--records\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
977 964 <span class="s">}</span>
|
|
978 965 <span class="s">}</span>
|
|
979 966 <span class="k">else</span> <span class="s">{</span>
|
|
980 967 <span class="k">die</span> <span class="q">"Error: Invalid number of values, "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@RecordSplit</span><span class="s">)</span><span class="cm">,</span> <span class="q">", specified using \"--record\" option: only 1 value is allowed.\n"</span><span class="sc">;</span>
|
|
981 968 <span class="s">}</span>
|
|
982 969 <span class="s">}</span>
|
|
983 970 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^recordnums$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
984 971 <span class="k">my</span><span class="s">(</span><span class="i">$RecordNum</span><span class="cm">,</span> <span class="i">$RecordCount</span><span class="cm">,</span> <span class="i">@SortedRecordSplit</span><span class="s">)</span><span class="sc">;</span>
|
|
985 972
|
|
986 973 <span class="i">@SortedRecordSplit</span> = <span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="i">@RecordSplit</span><span class="sc">;</span>
|
|
987 974
|
|
988 975 <span class="i">$RecordCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
989 976 <span class="j">RECORDNUM:</span> <span class="k">for</span> <span class="i">$RecordNum</span> <span class="s">(</span><span class="i">@SortedRecordSplit</span><span class="s">)</span> <span class="s">{</span>
|
|
990 977 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">RecordNums</span>}{<span class="i">$RecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
991 978 <span class="k">next</span> <span class="j">RECORDNUM</span><span class="sc">;</span>
|
|
992 979 <span class="s">}</span>
|
|
993 980 <span class="i">$RecordCount</span>++<span class="sc">;</span>
|
|
994 981 <span class="i">$OptionsInfo</span>{<span class="w">RecordNums</span>}{<span class="i">$RecordNum</span>} = <span class="i">$RecordNum</span><span class="sc">;</span>
|
|
995 982 <span class="s">}</span>
|
|
996 983 <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsCount</span>} = <span class="i">$RecordCount</span><span class="sc">;</span>
|
|
997 984 <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsMin</span>} = <span class="i">$SortedRecordSplit</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
998 985 <span class="i">$OptionsInfo</span>{<span class="w">RecordNumsMax</span>} = <span class="i">$SortedRecordSplit</span>[<span class="i">$#SortedRecordSplit</span>]<span class="sc">;</span>
|
|
999 986 <span class="s">}</span>
|
|
1000 987 <span class="k">else</span> <span class="s">{</span>
|
|
1001 988 <span class="k">if</span> <span class="s">(</span><span class="i">@RecordSplit</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
1002 989 <span class="i">$OptionsInfo</span>{<span class="w">StartRecordNum</span>} = <span class="i">$RecordSplit</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
1003 990 <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>} = <span class="i">$RecordSplit</span>[<span class="n">1</span>]<span class="sc">;</span>
|
|
1004 991 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StartRecordNum</span>} <= <span class="n">0</span> || <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>} <= <span class="n">0</span><span class="s">)</span> <span class="s">{</span>
|
|
1005 992 <span class="k">die</span> <span class="q">"Error: The value pair specified, $Options{record}, for option \"--records\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
1006 993 <span class="s">}</span>
|
|
1007 994 <span class="s">}</span>
|
|
1008 995 <span class="k">else</span> <span class="s">{</span>
|
|
1009 996 <span class="k">die</span> <span class="q">"Error: Invalid number of values, "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@RecordSplit</span><span class="s">)</span><span class="cm">,</span> <span class="q">", specified using \"--record\" option: only 2 values is allowed.\n"</span><span class="sc">;</span>
|
|
1010 997 <span class="s">}</span>
|
|
1011 998 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StartRecordNum</span>} > <span class="i">$OptionsInfo</span>{<span class="w">EndRecordNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1012 999 <span class="k">die</span> <span class="q">"Error: Start record number, $OptionsInfo{StartRecordNum}, must be smaller than end record number, $OptionsInfo{EndRecordNum}.\nSpecify different values using \"--record\" option.\n"</span><span class="sc">;</span>
|
|
1013 1000 <span class="s">}</span>
|
|
1014 1001 <span class="s">}</span>
|
|
1015 1002 <span class="s">}</span>
|
|
1016 1003 <span class="k">else</span> <span class="s">{</span>
|
|
1017 1004 <span class="k">die</span> <span class="q">"Error: For \"-m --mode\" option values recordnum, recordnums or recordrange, specify \"--record\" option value.\n"</span><span class="sc">;</span>
|
|
1018 1005 <span class="s">}</span>
|
|
1019 1006 <span class="s">}</span>
|
|
1020 1007
|
|
1021 1008 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1022 1009
|
|
1023 1010 <span class="k">my</span><span class="s">(</span><span class="i">@Words</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span><span class="sc">;</span>
|
|
1024 1011 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafields$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1025 1012 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1026 1013 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1027 1014 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}} = <span class="k">split</span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="sc">;</span>
|
|
1028 1015 <span class="s">}</span>
|
|
1029 1016 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1030 1017 <span class="k">open</span> <span class="w">DATAFIELDSFILE</span><span class="cm">,</span> <span class="q">"$Options{datafieldsfile}"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $Options{datafieldsfile}: $! \n"</span><span class="sc">;</span>
|
|
1031 1018 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*DATAFIELDSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1032 1019 <span class="i">@Words</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
1033 1020 <span class="k">if</span> <span class="s">(</span><span class="i">@Words</span><span class="s">)</span> <span class="s">{</span>
|
|
1034 1021 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="cm">,</span> <span class="i">@Words</span><span class="sc">;</span>
|
|
1035 1022 <span class="s">}</span>
|
|
1036 1023 <span class="s">}</span>
|
|
1037 1024 <span class="k">close</span> <span class="w">DATAFIELDSFILE</span><span class="sc">;</span>
|
|
1038 1025 <span class="s">}</span>
|
|
1039 1026 <span class="s">}</span>
|
|
1040 1027 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldsbyvalue$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1041 1028 <span class="k">my</span><span class="s">(</span><span class="i">@DataFieldsByValueTriplets</span><span class="s">)</span><span class="sc">;</span>
|
|
1042 1029 <span class="i">@DataFieldsByValueTriplets</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1043 1030 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1044 1031 <span class="i">@DataFieldsByValueTriplets</span> = <span class="k">split</span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="sc">;</span>
|
|
1045 1032 <span class="s">}</span>
|
|
1046 1033 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1047 1034 <span class="k">open</span> <span class="w">DATAFIELDSFILE</span><span class="cm">,</span> <span class="q">"$Options{datafieldsfile}"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $Options{datafieldsfile}: $! \n"</span><span class="sc">;</span>
|
|
1048 1035 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*DATAFIELDSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1049 1036 <span class="i">@Words</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
1050 1037 <span class="k">if</span> <span class="s">(</span><span class="i">@Words</span><span class="s">)</span> <span class="s">{</span>
|
|
1051 1038 <span class="k">push</span> <span class="i">@DataFieldsByValueTriplets</span><span class="cm">,</span> <span class="i">@Words</span><span class="sc">;</span>
|
|
1052 1039 <span class="s">}</span>
|
|
1053 1040 <span class="s">}</span>
|
|
1054 1041 <span class="k">close</span> <span class="w">DATAFIELDSFILE</span><span class="sc">;</span>
|
|
1055 1042 <span class="s">}</span>
|
|
1056 1043 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">@DataFieldsByValueTriplets</span> % <span class="n">3</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1057 1044 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1058 1045 <span class="k">die</span> <span class="q">"Error: Triplets not found in values specified by \"-d --datafields\" option\n"</span><span class="sc">;</span>
|
|
1059 1046 <span class="s">}</span>
|
|
1060 1047 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1061 1048 <span class="k">die</span> <span class="q">"Error: Triplets not found in values specified by \"--datafieldsfile\" option\n"</span><span class="sc">;</span>
|
|
1062 1049 <span class="s">}</span>
|
|
1063 1050 <span class="s">}</span>
|
|
1064 1051 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Label</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$Criterion</span><span class="s">)</span><span class="sc">;</span>
|
|
1065 1052
|
|
1066 1053 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1067 1054 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1068 1055 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldCriteriaMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1069 1056
|
|
1070 1057 <span class="k">for</span> <span class="s">(</span><span class="i">$Index</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$Index</span> < <span class="i">@DataFieldsByValueTriplets</span><span class="sc">;</span> <span class="i">$Index</span> = <span class="i">$Index</span> + <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
1071 1058 <span class="i">$Label</span> = <span class="i">$DataFieldsByValueTriplets</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
1072 1059 <span class="i">$Value</span> = <span class="i">$DataFieldsByValueTriplets</span>[<span class="i">$Index</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
1073 1060 <span class="i">$Criterion</span> = <span class="i">$DataFieldsByValueTriplets</span>[<span class="i">$Index</span> + <span class="n">2</span>]<span class="sc">;</span>
|
|
1074 1061
|
|
1075 1062 <span class="k">if</span> <span class="s">(</span><span class="i">$Criterion</span> =~ <span class="q">/^(eq|le|ge)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1076 1063 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="cm">,</span> <span class="i">$Label</span><span class="sc">;</span>
|
|
1077 1064 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesMap</span>}{<span class="i">$Label</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
1078 1065 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldCriteriaMap</span>}{<span class="i">$Label</span>} = <span class="i">$Criterion</span><span class="sc">;</span>
|
|
1079 1066 <span class="s">}</span>
|
|
1080 1067 <span class="k">else</span> <span class="s">{</span>
|
|
1081 1068 <span class="k">warn</span> <span class="q">"Warning: Ignoring triplet value, $Label $Value $Criterion , specified using \"-d --datafields\" or \"--datafieldsfile\" option: Invalid criterion value: $Criterion\n"</span><span class="sc">;</span>
|
|
1082 1069 <span class="s">}</span>
|
|
1083 1070 <span class="s">}</span>
|
|
1084 1071 <span class="s">}</span>
|
|
1085 1072 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldsbyregex$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1086 1073 <span class="k">my</span><span class="s">(</span><span class="i">@DataFieldsByRegexTriplets</span><span class="s">)</span><span class="sc">;</span>
|
|
1087 1074
|
|
1088 1075 <span class="i">@DataFieldsByRegexTriplets</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1089 1076 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1090 1077 <span class="i">@DataFieldsByRegexTriplets</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span><span class="sc">;</span>
|
|
1091 1078 <span class="s">}</span>
|
|
1092 1079 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1093 1080 <span class="k">open</span> <span class="w">DATAFIELDSFILE</span><span class="cm">,</span> <span class="q">"$Options{datafieldsfile}"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $Options{datafieldsfile}: $! \n"</span><span class="sc">;</span>
|
|
1094 1081 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*DATAFIELDSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1095 1082 <span class="i">@Words</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
1096 1083 <span class="k">if</span> <span class="s">(</span><span class="i">@Words</span><span class="s">)</span> <span class="s">{</span>
|
|
1097 1084 <span class="k">push</span> <span class="i">@DataFieldsByRegexTriplets</span><span class="cm">,</span> <span class="i">@Words</span><span class="sc">;</span>
|
|
1098 1085 <span class="s">}</span>
|
|
1099 1086 <span class="s">}</span>
|
|
1100 1087 <span class="k">close</span> <span class="w">DATAFIELDSFILE</span><span class="sc">;</span>
|
|
1101 1088 <span class="s">}</span>
|
|
1102 1089 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">@DataFieldsByRegexTriplets</span> % <span class="n">3</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1103 1090 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1104 1091 <span class="k">die</span> <span class="q">"Error: Triplet not found in values specified by \"-d --datafields\" option\n"</span><span class="sc">;</span>
|
|
1105 1092 <span class="s">}</span>
|
|
1106 1093 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1107 1094 <span class="k">die</span> <span class="q">"Error: Triplet not found in values specified by \"--datafieldsfile\" option\n"</span><span class="sc">;</span>
|
|
1108 1095 <span class="s">}</span>
|
|
1109 1096 <span class="s">}</span>
|
|
1110 1097
|
|
1111 1098 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Label</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$Criterion</span><span class="s">)</span><span class="sc">;</span>
|
|
1112 1099
|
|
1113 1100 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1114 1101 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1115 1102 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexCriteriaMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1116 1103
|
|
1117 1104 <span class="k">for</span> <span class="s">(</span><span class="i">$Index</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$Index</span> < <span class="i">@DataFieldsByRegexTriplets</span><span class="sc">;</span> <span class="i">$Index</span> = <span class="i">$Index</span> + <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
1118 1105 <span class="i">$Label</span> = <span class="i">$DataFieldsByRegexTriplets</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
1119 1106 <span class="i">$Value</span> = <span class="i">$DataFieldsByRegexTriplets</span>[<span class="i">$Index</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
1120 1107 <span class="i">$Criterion</span> = <span class="i">$DataFieldsByRegexTriplets</span>[<span class="i">$Index</span> + <span class="n">2</span>]<span class="sc">;</span>
|
|
1121 1108
|
|
1122 1109 <span class="k">if</span> <span class="s">(</span><span class="i">$Criterion</span> =~ <span class="q">/^(eq|ne)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1123 1110 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabels</span>}}<span class="cm">,</span> <span class="i">$Label</span><span class="sc">;</span>
|
|
1124 1111 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexMap</span>}{<span class="i">$Label</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
1125 1112 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldRegexCriteriaMap</span>}{<span class="i">$Label</span>} = <span class="i">$Criterion</span><span class="sc">;</span>
|
|
1126 1113 <span class="s">}</span>
|
|
1127 1114 <span class="k">else</span> <span class="s">{</span>
|
|
1128 1115 <span class="k">warn</span> <span class="q">"Warning: Ignoring triplet value, $Label $Value $Criterion , specified using \"-d --datafields\" or \"--datafieldsfile\" option: Invalid criterion value: $Criterion; Supported values: eq or ne\n"</span><span class="sc">;</span>
|
|
1129 1116 <span class="s">}</span>
|
|
1130 1117 <span class="s">}</span>
|
|
1131 1118 <span class="s">}</span>
|
|
1132 1119 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(datafieldbylist|datafielduniquebylist|datafieldnotbylist)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1133 1120 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">@DataFieldAndValuesList</span><span class="s">)</span><span class="sc">;</span>
|
|
1134 1121 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1135 1122 <span class="i">@DataFieldAndValuesList</span> = <span class="k">split</span> <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="i">$Options</span>{<span class="w">datafields</span>}<span class="sc">;</span>
|
|
1136 1123 <span class="s">}</span>
|
|
1137 1124 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1138 1125 <span class="k">open</span> <span class="w">DATAFIELDSFILE</span><span class="cm">,</span> <span class="q">"$Options{datafieldsfile}"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $Options{datafieldsfile}: $! \n"</span><span class="sc">;</span>
|
|
1139 1126 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*DATAFIELDSFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1140 1127 <span class="i">@Words</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>}<span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
1141 1128 <span class="k">if</span> <span class="s">(</span><span class="i">@Words</span><span class="s">)</span> <span class="s">{</span>
|
|
1142 1129 <span class="k">push</span> <span class="i">@DataFieldAndValuesList</span><span class="cm">,</span> <span class="i">@Words</span><span class="sc">;</span>
|
|
1143 1130 <span class="s">}</span>
|
|
1144 1131 <span class="s">}</span>
|
|
1145 1132 <span class="k">close</span> <span class="w">DATAFIELDSFILE</span><span class="sc">;</span>
|
|
1146 1133 <span class="s">}</span>
|
|
1147 1134 <span class="k">if</span> <span class="s">(</span><span class="i">@DataFieldAndValuesList</span> < <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
1148 1135 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafields</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1149 1136 <span class="k">die</span> <span class="q">"Error: Invalid number of values specified by \"-d --datafields\" option\n"</span><span class="sc">;</span>
|
|
1150 1137 <span class="s">}</span>
|
|
1151 1138 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">datafieldsfile</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1152 1139 <span class="k">die</span> <span class="q">"Error: Invalid number values specified by \"--datafieldsfile\" option\n"</span><span class="sc">;</span>
|
|
1153 1140 <span class="s">}</span>
|
|
1154 1141 <span class="s">}</span>
|
|
1155 1142
|
|
1156 1143 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldLabel</span>} = <span class="i">$DataFieldAndValuesList</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
1157 1144 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValuesCount</span>} = <span class="i">@DataFieldAndValuesList</span> - <span class="n">1</span><span class="sc">;</span>
|
|
1158 1145 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1159 1146
|
|
1160 1147 <span class="k">for</span> <span class="s">(</span><span class="i">$Index</span> = <span class="n">1</span><span class="sc">;</span> <span class="i">$Index</span> < <span class="i">@DataFieldAndValuesList</span><span class="sc">;</span> <span class="i">$Index</span>++<span class="s">)</span> <span class="s">{</span>
|
|
1161 1148 <span class="i">$Value</span> = <span class="i">$DataFieldAndValuesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
1162 1149 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedDataFieldValues</span>}{<span class="i">$Value</span>} = <span class="q">"NotFound"</span><span class="sc">;</span>
|
|
1163 1150 <span class="s">}</span>
|
|
1164 1151 <span class="s">}</span>
|
|
1165 1152
|
|
1166 1153 <span class="i">$OptionsInfo</span>{<span class="w">SDFileExt</span>} = <span class="q">"sdf"</span><span class="sc">;</span>
|
|
1167 1154 <span class="i">$OptionsInfo</span>{<span class="w">TextFileExt</span>} = <span class="q">"csv"</span><span class="sc">;</span>
|
|
1168 1155
|
|
1169 1156 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1170 1157 <span class="i">$OptionsInfo</span>{<span class="w">TextFileExt</span>} = <span class="q">"tsv"</span><span class="sc">;</span>
|
|
1171 1158 <span class="s">}</span>
|
|
1172 1159
|
|
1173 1160 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(alldatafields|molnames)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1174 1161 <span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
1175 1162 <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
1176 1163 <span class="s">}</span>
|
|
1177 1164 <span class="k">else</span> <span class="s">{</span>
|
|
1178 1165 <span class="i">$OptionsInfo</span>{<span class="w">OutputSDFile</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(SD|both)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
1179 1166 <span class="i">$OptionsInfo</span>{<span class="w">OutputTextFile</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} =~ <span class="q">/^(text|both)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
1180 1167 <span class="s">}</span>
|
|
1181 1168
|
|
1182 1169 <span class="i">$OptionsInfo</span>{<span class="w">StrDataString</span>} = <span class="i">$Options</span>{<span class="w">strdatastring</span>}<span class="sc">;</span>
|
|
1183 1170 <span class="i">$OptionsInfo</span>{<span class="w">OutputStrDataString</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">strdatastring</span>} =~ <span class="q">/^Yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
1184 1171
|
|
1185 1172 <span class="i">$OptionsInfo</span>{<span class="w">StrDataStringDelimiter</span>} = <span class="i">$Options</span>{<span class="w">strdatastringdelimiter</span>}<span class="sc">;</span>
|
|
1186 1173
|
|
1187 1174 <span class="k">if</span> <span class="s">(</span><span class="i">IsEmpty</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">strdatastringdelimiter</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1188 1175 <span class="k">die</span> <span class="q">"Error: No value specified for \"--StrDataStringDelimiter\" option.\n"</span><span class="sc">;</span>
|
|
1189 1176 <span class="s">}</span>
|
|
1190 1177 <span class="i">$OptionsInfo</span>{<span class="w">StrDataStringMode</span>} = <span class="i">$Options</span>{<span class="w">strdatastringmode</span>}<span class="sc">;</span>
|
|
1191 1178 <span class="i">$OptionsInfo</span>{<span class="w">StrDataStringWithFields</span>} = <span class="i">$Options</span>{<span class="w">strdatastringmode</span>} =~ <span class="q">/^StrAndDataFields$/i</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
1192 1179
|
|
1193 1180 <span class="j">MODE:</span> <span class="s">{</span>
|
|
1194 1181 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^alldatafields$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"AllDataDields"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1195 1182 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^commondatafields$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"CommonDataDields"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1196 1183 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafields$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedDataFields"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1197 1184 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldsbyvalue$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedDataFieldsByValue"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1198 1185 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldsbyregex$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedDataFieldsByRegex"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1199 1186 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldbylist$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedDataField"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1200 1187 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafielduniquebylist$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedUniqueDataField"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1201 1188 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^datafieldnotbylist$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedDataFieldNotByList"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1202 1189 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^molnames$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"MolName"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1203 1190 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^randomcmpds$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"RandomCmpds"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1204 1191 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^recordnum$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"RecordNum$OptionsInfo{RecordNum}"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1205 1192 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^recordnums$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"RecordNums"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1206 1193 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^recordrange$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"RecordNum$OptionsInfo{StartRecordNum}"</span> . <span class="q">"To"</span> . <span class="q">"$OptionsInfo{EndRecordNum}"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1207 1194 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^2dcmpdrecords$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"2DCmpdRecords"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1208 1195 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^3dcmpdrecords$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"3DCmpdRecords"</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span> <span class="s">}</span>
|
|
1209 1196 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: alldatafields, commondatafields, datafields, datafieldsbyvalue, datafieldbylist, datafielduniquebylist, , datafieldnotbylist, molnames, randomcmpds, recordnum, recordnums, recordrange, 2dcmpdrecords, 3dcmpdrecords\n"</span><span class="sc">;</span>
|
|
1210 1197 <span class="s">}</span>
|
|
1211 1198
|
|
1212 1199 <span class="s">}</span>
|
|
1213 1200
|
|
1214 1201 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
1215 <a name="SetupScriptUsage-"></a>1202 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
1216 1203
|
|
1217 1204 <span class="c"># Retrieve all the options...</span>
|
|
1218 1205 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1219 1206 <span class="i">$Options</span>{<span class="w">numofcmpds</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
1220 1207 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">"alldatafields"</span><span class="sc">;</span>
|
|
1221 1208 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
1222 1209 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
1223 1210 <span class="i">$Options</span>{<span class="w">output</span>} = <span class="q">"SD"</span><span class="sc">;</span>
|
|
1224 1211 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">"yes"</span><span class="sc">;</span>
|
|
1225 1212 <span class="i">$Options</span>{<span class="w">regexignorecase</span>} = <span class="q">"yes"</span><span class="sc">;</span>
|
|
1226 1213 <span class="i">$Options</span>{<span class="w">valuecomparisonmode</span>} = <span class="q">"numeric"</span><span class="sc">;</span>
|
|
1227 1214 <span class="i">$Options</span>{<span class="w">violations</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
1228 1215 <span class="i">$Options</span>{<span class="w">seed</span>} = <span class="n">123456789</span><span class="sc">;</span>
|
|
1229 1216
|
|
1230 1217 <span class="i">$Options</span>{<span class="w">strdatastring</span>} = <span class="q">"no"</span><span class="sc">;</span>
|
|
1231 1218 <span class="i">$Options</span>{<span class="w">strdatastringdelimiter</span>} = <span class="q">"|"</span><span class="sc">;</span>
|
|
1232 1219 <span class="i">$Options</span>{<span class="w">strdatastringmode</span>} = <span class="q">"StrOnly"</span><span class="sc">;</span>
|
|
1233 1220
|
|
1234 1221 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"datafields|d=s"</span><span class="cm">,</span> <span class="q">"datafieldsfile=s"</span><span class="cm">,</span> <span class="q">"indelim=s"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"numofcmpds|n=i"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"output=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"quote|q=s"</span><span class="cm">,</span> <span class="q">"regexignorecase=s"</span><span class="cm">,</span> <span class="q">"record=s"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"seed|s=i"</span><span class="cm">,</span> <span class="q">"strdatastring=s"</span><span class="cm">,</span> <span class="q">"strdatastringdelimiter=s"</span><span class="cm">,</span> <span class="q">"strdatastringmode=s"</span><span class="cm">,</span> <span class="q">"valuecomparisonmode=s"</span><span class="cm">,</span> <span class="q">"violations|v=i"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1235 1222 <span class="k">die</span> <span class="q">"\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
1236 1223 <span class="s">}</span>
|
|
1237 1224 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1238 1225 <span class="k">if</span> <span class="s">(</span>! <span class="k">-d</span> <span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1239 1226 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
1240 1227 <span class="s">}</span>
|
|
1241 1228 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't chdir $Options{workingdir}: $! \n"</span><span class="sc">;</span>
|
|
1242 1229 <span class="s">}</span>
|
|
1243 1230 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">numofcmpds</span>} < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
1244 1231 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{numofcmpds}, for option \"-n --numofcmpds\" is not valid. Allowed values: >= 1 \n"</span><span class="sc">;</span>
|
|
1245 1232 <span class="s">}</span>
|
|
1246 1233 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">valuecomparisonmode</span>} !~ <span class="q">/^(Numeric|Alphanumeric)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1247 1234 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{valuecomparisonmode}, for option \"--ValueComparisonMode\" is not valid. Allowed values: Numeric or Alphanumeric\n"</span><span class="sc">;</span>
|
|
1248 1235 <span class="s">}</span>
|
|
1249 1236 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">violations</span>} < <span class="n">0</span><span class="s">)</span> <span class="s">{</span>
|
|
1250 1237 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{violations}, for option \"-v --violations\" is not valid. Allowed values: >= 0 \n"</span><span class="sc">;</span>
|
|
1251 1238 <span class="s">}</span>
|
|
1252 1239 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(alldatafields|commondatafields|datafields|datafieldsbyvalue|datafieldsbyregex|datafieldbylist|datafielduniquebylist|datafieldnotbylist|molnames|randomcmpds|recordnum|recordnums|recordrange|2dcmpdrecords|3dcmpdrecords)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1253 1240 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: alldatafields, commondatafields, datafields, datafieldsbyvalue, datafieldbylist, datafielduniquebylist, datafieldnotbylist, molnames, randomcmpds, recordnum, recordnums, recordrange, 2dcmpdrecords, 3dcmpdrecords\n"</span><span class="sc">;</span>
|
|
1254 1241 <span class="s">}</span>
|
|
1255 1242 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">output</span>} !~ <span class="q">/^(SD|text|both)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1256 1243 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: SD, text, or both\n"</span><span class="sc">;</span>
|
|
1257 1244 <span class="s">}</span>
|
|
1258 1245 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1259 1246 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
1260 1247 <span class="s">}</span>
|
|
1261 1248 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1262 1249 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
1263 1250 <span class="s">}</span>
|
|
1264 1251 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1265 1252 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
1266 1253 <span class="s">}</span>
|
|
1267 1254 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">regexignorecase</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1268 1255 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{regexignorecase}, for option \"--regexignorecase\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
1269 1256 <span class="s">}</span>
|
|
1270 1257 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">strdatastring</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1271 1258 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{strdatastring}, for option \"--StrDataString\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
1272 1259 <span class="s">}</span>
|
|
1273 1260 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">strdatastringmode</span>} !~ <span class="q">/^(StrOnly|StrAndDataFields)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1274 1261 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{strdatastringmode}, for option \"--StrDataStringMode\" is not valid. Allowed values: StrOnly or StrAndDataFields\n"</span><span class="sc">;</span>
|
|
1275 1262 <span class="s">}</span>
|
|
1276 1263 <span class="s">}</span>
|
|
1277 1264
|
|
1278 <a name="EOF-"></a></pre>
|
|
1279 <p> </p>
|
|
1280 <br />
|
|
1281 <center>
|
|
1282 <img src="../../../images/h2o2.png">
|
|
1283 </center>
|
|
1284 </body>
|
|
1285 </html>
|