| 0 | 1 <html> | 
|  | 2 <head> | 
|  | 3 <title>MayaChemTools:Code:SplitSDFiles.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: SplitSDFiles.pl,v $</span> | 
|  | 17    4 <span class="c"># $Date: 2015/02/28 20:46:21 $</span> | 
|  | 18    5 <span class="c"># $Revision: 1.36 $</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">Benchmark</span><span class="sc">;</span> | 
|  | 47   34 <span class="k">use</span> <span class="w">SDFileUtil</span><span class="sc">;</span> | 
|  | 48   35 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span> | 
|  | 49   36 | 
|  | 50   37 <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> | 
|  | 51   38 | 
|  | 52   39 <span class="c"># Autoflush STDOUT</span> | 
|  | 53   40 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span> | 
|  | 54   41 | 
|  | 55   42 <span class="c"># Starting message...</span> | 
|  | 56   43 <span class="i">$ScriptName</span> = <span class="w">basename</span> <span class="i">$0</span><span class="sc">;</span> | 
|  | 57   44 <span class="k">print</span> <span class="q">"\n$ScriptName:Starting...\n\n"</span><span class="sc">;</span> | 
|  | 58   45 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span> | 
|  | 59   46 | 
|  | 60   47 <span class="c"># Get the options and setup script...</span> | 
|  | 61   48 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 62   49 <span class="k">if</span> <span class="s">(</span><span class="i">$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> | 
|  | 63   50   <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> | 
|  | 64   51 <span class="s">}</span> | 
|  | 65   52 | 
|  | 66   53 <span class="k">my</span><span class="s">(</span><span class="i">@SDFilesList</span><span class="s">)</span><span class="sc">;</span> | 
|  | 67   54 <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> | 
|  | 68   55 | 
|  | 69   56 <span class="c"># Process options...</span> | 
|  | 70   57 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span> | 
|  | 71   58 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span> | 
|  | 72   59 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 73   60 | 
|  | 74   61 <span class="c"># Setup information about input files...</span> | 
|  | 75   62 <span class="k">my</span><span class="s">(</span><span class="i">%SDFilesInfo</span><span class="s">)</span><span class="sc">;</span> | 
|  | 76   63 <span class="k">print</span> <span class="q">"Checking input SD file(s)...\n"</span><span class="sc">;</span> | 
|  | 77   64 <span class="i">RetrieveSDFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 78   65 | 
|  | 79   66 <span class="c"># Process input files..</span> | 
|  | 80   67 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 81   68 <span class="k">if</span> <span class="s">(</span><span class="i">@SDFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 82   69   <span class="k">print</span> <span class="q">"\nProcessing SD files...\n"</span><span class="sc">;</span> | 
|  | 83   70 <span class="s">}</span> | 
|  | 84   71 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span> | 
|  | 85   72   <span class="k">if</span> <span class="s">(</span><span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span> | 
|  | 86   73     <span class="k">print</span> <span class="q">"\nProcessing file $SDFilesList[$FileIndex]...\n"</span><span class="sc">;</span> | 
|  | 87   74     <span class="i">SplitSDFile</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 88   75   <span class="s">}</span> | 
|  | 89   76 <span class="s">}</span> | 
|  | 90   77 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span> | 
|  | 91   78 | 
|  | 92   79 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span> | 
|  | 93   80 <span class="i">$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> | 
|  | 94   81 <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">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span> | 
|  | 95   82 | 
|  | 96   83 <span class="c">###############################################################################</span> | 
|  | 97   84 | 
|  | 98   85 <span class="c"># Split a SD file...</span> | 
|  | 99   86 <span class="c">#</span> | 
|  | 100 <a name="SplitSDFile-"></a>  87 <span class="k">sub </span><span class="m">SplitSDFile</span> <span class="s">{</span> | 
|  | 101   88   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 102   89 | 
|  | 103   90   <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^Files$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 104   91     <span class="i">SplitSDFileByNumOfFiles</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 105   92   <span class="s">}</span> | 
|  | 106   93   <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^Cmpds$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 107   94     <span class="i">SplitSDFileByNumOfCmpds</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 108   95   <span class="s">}</span> | 
|  | 109   96 <span class="s">}</span> | 
|  | 110   97 | 
|  | 111   98 <span class="c"># Split SD into specified number of files...</span> | 
|  | 112   99 <span class="c">#</span> | 
|  | 113 <a name="SplitSDFileByNumOfFiles-"></a> 100 <span class="k">sub </span><span class="m">SplitSDFileByNumOfFiles</span> <span class="s">{</span> | 
|  | 114  101   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 115  102   <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$MaxCmpdsPerFile</span><span class="cm">,</span> <span class="i">$MaxNumOfFiles</span><span class="s">)</span><span class="sc">;</span> | 
|  | 116  103 | 
|  | 117  104   <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 118  105 | 
|  | 119  106   <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> | 
|  | 120  107     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Couldn't open it: $! \n"</span><span class="sc">;</span> | 
|  | 121  108     <span class="k">return</span><span class="sc">;</span> | 
|  | 122  109   <span class="s">}</span> | 
|  | 123  110 | 
|  | 124  111   <span class="i">$MaxNumOfFiles</span> = <span class="i">$OptionsInfo</span>{<span class="w">NumOfFiles</span>}<span class="sc">;</span> | 
|  | 125  112 | 
|  | 126  113   <span class="c"># Count number of compounds to figure out maximum number of compound per file...</span> | 
|  | 127  114   <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span> | 
|  | 128  115   <span class="k">while</span> <span class="s">(</span><span class="q"><SDFILE></span><span class="s">)</span> <span class="s">{</span> | 
|  | 129  116     <span class="k">if</span> <span class="s">(</span><span class="q">/^\$\$\$\$/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 130  117       <span class="i">$CmpdCount</span>++<span class="sc">;</span> | 
|  | 131  118     <span class="s">}</span> | 
|  | 132  119   <span class="s">}</span> | 
|  | 133  120   <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span> | 
|  | 134  121 | 
|  | 135  122   <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdCount</span> < <span class="i">$MaxNumOfFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 136  123     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Total number of compounds, $CmpdCount, is smaller than number of new files, $MaxNumOfFiles\n"</span><span class="sc">;</span> | 
|  | 137  124     <span class="k">return</span><span class="sc">;</span> | 
|  | 138  125   <span class="s">}</span> | 
|  | 139  126 | 
|  | 140  127   <span class="i">$MaxCmpdsPerFile</span> = <span class="k">int</span> <span class="i">$CmpdCount</span> / <span class="i">$MaxNumOfFiles</span><span class="sc">;</span> | 
|  | 141  128 | 
|  | 142  129   <span class="i">SplitSDFileByNumOfFilesAndCmpds</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$MaxNumOfFiles</span><span class="cm">,</span> <span class="i">$MaxCmpdsPerFile</span><span class="s">)</span><span class="sc">;</span> | 
|  | 143  130 <span class="s">}</span> | 
|  | 144  131 | 
|  | 145  132 <span class="c"># Split SD into files containing specified number of compounds...</span> | 
|  | 146  133 <span class="c">#</span> | 
|  | 147 <a name="SplitSDFileByNumOfCmpds-"></a> 134 <span class="k">sub </span><span class="m">SplitSDFileByNumOfCmpds</span> <span class="s">{</span> | 
|  | 148  135   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 149  136 | 
|  | 150  137   <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">NumOfCmpds</span>} == <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 151  138     <span class="i">SplitSDFileByOneCmpdPerFile</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 152  139   <span class="s">}</span> | 
|  | 153  140   <span class="k">else</span> <span class="s">{</span> | 
|  | 154  141     <span class="i">SplitSDFileByNumOfCmpdsPerFile</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | 
|  | 155  142   <span class="s">}</span> | 
|  | 156  143 <span class="s">}</span> | 
|  | 157  144 | 
|  | 158  145 <span class="c"># Split SD into files containing one compound per file...</span> | 
|  | 159  146 <span class="c">#</span> | 
|  | 160 <a name="SplitSDFileByOneCmpdPerFile-"></a> 147 <span class="k">sub </span><span class="m">SplitSDFileByOneCmpdPerFile</span> <span class="s">{</span> | 
|  | 161  148   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 162  149   <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$NewSDFile</span><span class="cm">,</span> <span class="i">$NewSDFileRoot</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$OverwriteFiles</span><span class="cm">,</span> <span class="i">$UseDataField</span><span class="cm">,</span> <span class="i">$DataFieldName</span><span class="cm">,</span> <span class="i">$UseMolName</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="cm">,</span> <span class="i">@CmpdLines</span><span class="cm">,</span> <span class="i">%DataFieldValues</span><span class="s">)</span><span class="sc">;</span> | 
|  | 163  150 | 
|  | 164  151   <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 165  152 | 
|  | 166  153   <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> | 
|  | 167  154     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Couldn't open it: $! \n"</span><span class="sc">;</span> | 
|  | 168  155     <span class="k">return</span><span class="sc">;</span> | 
|  | 169  156   <span class="s">}</span> | 
|  | 170  157 | 
|  | 171  158   <span class="k">print</span> <span class="q">"\n"</span><span class="sc">;</span> | 
|  | 172  159 | 
|  | 173  160   <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span> | 
|  | 174  161 | 
|  | 175  162   <span class="i">$FileExt</span> = <span class="i">$SDFilesInfo</span>{<span class="w">FileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 176  163 | 
|  | 177  164   <span class="i">$OutFileRoot</span> = <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 178  165   <span class="i">$OverwriteFiles</span> = <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="sc">;</span> | 
|  | 179  166 | 
|  | 180  167   <span class="i">$UseDataField</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CmpdsMode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | 
|  | 181  168   <span class="i">$DataFieldName</span> = <span class="i">$OptionsInfo</span>{<span class="w">DataField</span>}<span class="sc">;</span> | 
|  | 182  169 | 
|  | 183  170   <span class="i">$UseMolName</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CmpdsMode</span>} =~ <span class="q">/^MolName$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | 
|  | 184  171 | 
|  | 185  172   <span class="j">CMPDSTRING:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$CmpdString</span> = <span class="i">ReadCmpdString</span><span class="s">(</span>\<span class="i">*SDFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 186  173     <span class="i">$CmpdCount</span>++<span class="sc">;</span> | 
|  | 187  174 | 
|  | 188  175     <span class="c"># Setup SD file name...</span> | 
|  | 189  176     <span class="i">$NewSDFileRoot</span> = <span class="q">''</span><span class="sc">;</span> | 
|  | 190  177     <span class="k">if</span> <span class="s">(</span><span class="i">$UseDataField</span><span class="s">)</span> <span class="s">{</span> | 
|  | 191  178       <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span> | 
|  | 192  179       <span class="i">%DataFieldValues</span> = <span class="i">GetCmpdDataHeaderLabelsAndValues</span><span class="s">(</span>\<span class="i">@CmpdLines</span><span class="s">)</span><span class="sc">;</span> | 
|  | 193  180       <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$DataFieldValues</span>{<span class="i">$DataFieldName</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 194  181         <span class="i">$NewSDFileRoot</span> = <span class="i">$DataFieldValues</span>{<span class="i">$DataFieldName</span>}<span class="sc">;</span> | 
|  | 195  182       <span class="s">}</span> | 
|  | 196  183     <span class="s">}</span> | 
|  | 197  184     <span class="k">elsif</span> <span class="s">(</span><span class="i">$UseMolName</span><span class="s">)</span> <span class="s">{</span> | 
|  | 198  185       <span class="i">@CmpdLines</span> = <span class="k">split</span> <span class="q">"\n"</span><span class="cm">,</span> <span class="i">$CmpdString</span><span class="sc">;</span> | 
|  | 199  186       <span class="i">$NewSDFileRoot</span> = <span class="i">$CmpdLines</span>[<span class="n">0</span>]<span class="sc">;</span> | 
|  | 200  187     <span class="s">}</span> | 
|  | 201  188 | 
|  | 202  189     <span class="c"># Check for any invalid file name characters in data field or molname values...</span> | 
|  | 203  190     <span class="k">if</span> <span class="s">(</span><span class="i">$NewSDFileRoot</span> && <span class="i">$NewSDFileRoot</span> =~ <span class="q">/[^a-zA-Z0-9_]/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 204  191       <span class="i">$NewSDFileRoot</span> =~ <span class="q">s/[^a-zA-Z0-9_]//g</span><span class="sc">;</span> | 
|  | 205  192     <span class="s">}</span> | 
|  | 206  193 | 
|  | 207  194     <span class="c"># Fall back plan for SD file name...</span> | 
|  | 208  195     <span class="k">if</span> <span class="s">(</span>!<span class="i">$NewSDFileRoot</span><span class="s">)</span> <span class="s">{</span> | 
|  | 209  196       <span class="i">$NewSDFileRoot</span> = <span class="q">"${OutFileRoot}Cmpd${CmpdCount}"</span><span class="sc">;</span> | 
|  | 210  197     <span class="s">}</span> | 
|  | 211  198 | 
|  | 212  199     <span class="i">$NewSDFile</span> = <span class="q">"${NewSDFileRoot}.${FileExt}"</span><span class="sc">;</span> | 
|  | 213  200 | 
|  | 214  201     <span class="k">if</span> <span class="s">(</span>!<span class="i">$OverwriteFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 215  202       <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewSDFile</span><span class="s">)</span> <span class="s">{</span> | 
|  | 216  203         <span class="k">warn</span> <span class="q">"Warning: Ignoring compound number, $CmpdCount, in $SDFile: New SD file, $NewSDFile, already exists\n"</span><span class="sc">;</span> | 
|  | 217  204         <span class="k">next</span> <span class="j">CMPDSTRING</span><span class="sc">;</span> | 
|  | 218  205       <span class="s">}</span> | 
|  | 219  206     <span class="s">}</span> | 
|  | 220  207 | 
|  | 221  208     <span class="c"># Write out new SD file...</span> | 
|  | 222  209 | 
|  | 223  210     <span class="k">print</span> <span class="q">"Generating $NewSDFile file\n"</span><span class="sc">;</span> | 
|  | 224  211     <span class="k">open</span> <span class="w">NEWSDFILE</span><span class="cm">,</span> <span class="q">">$NewSDFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewSDFile: $! \n"</span><span class="sc">;</span> | 
|  | 225  212     <span class="k">print</span> <span class="i">NEWSDFILE</span> <span class="q">"$CmpdString\n"</span><span class="sc">;</span> | 
|  | 226  213     <span class="k">close</span> <span class="w">NEWSDFILE</span><span class="sc">;</span> | 
|  | 227  214 | 
|  | 228  215   <span class="s">}</span> | 
|  | 229  216   <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span> | 
|  | 230  217 <span class="s">}</span> | 
|  | 231  218 | 
|  | 232  219 <span class="c"># Split SD into files containing specified number of compounds per file...</span> | 
|  | 233  220 <span class="c">#</span> | 
|  | 234 <a name="SplitSDFileByNumOfCmpdsPerFile-"></a> 221 <span class="k">sub </span><span class="m">SplitSDFileByNumOfCmpdsPerFile</span> <span class="s">{</span> | 
|  | 235  222   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 236  223   <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$MaxCmpdsPerFile</span><span class="cm">,</span> <span class="i">$MaxNumOfFiles</span><span class="s">)</span><span class="sc">;</span> | 
|  | 237  224 | 
|  | 238  225   <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 239  226 | 
|  | 240  227   <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> | 
|  | 241  228     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Couldn't open it: $! \n"</span><span class="sc">;</span> | 
|  | 242  229     <span class="k">return</span><span class="sc">;</span> | 
|  | 243  230   <span class="s">}</span> | 
|  | 244  231 | 
|  | 245  232   <span class="i">$MaxCmpdsPerFile</span> = <span class="i">$OptionsInfo</span>{<span class="w">NumOfCmpds</span>}<span class="sc">;</span> | 
|  | 246  233 | 
|  | 247  234   <span class="c"># Count number of compounds to figure out maximum number of files...</span> | 
|  | 248  235   <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span> | 
|  | 249  236   <span class="k">while</span> <span class="s">(</span><span class="q"><SDFILE></span><span class="s">)</span> <span class="s">{</span> | 
|  | 250  237     <span class="k">if</span> <span class="s">(</span><span class="q">/^\$\$\$\$/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 251  238       <span class="i">$CmpdCount</span>++<span class="sc">;</span> | 
|  | 252  239     <span class="s">}</span> | 
|  | 253  240   <span class="s">}</span> | 
|  | 254  241   <span class="k">close</span> <span class="w">SDFILE</span><span class="sc">;</span> | 
|  | 255  242 | 
|  | 256  243   <span class="i">$MaxNumOfFiles</span> = <span class="k">int</span> <span class="i">$CmpdCount</span> / <span class="i">$MaxCmpdsPerFile</span><span class="sc">;</span> | 
|  | 257  244 | 
|  | 258  245   <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$MaxNumOfFiles</span> * <span class="i">$MaxCmpdsPerFile</span><span class="s">)</span> < <span class="i">$CmpdCount</span><span class="s">)</span> <span class="s">{</span> | 
|  | 259  246     <span class="i">$MaxNumOfFiles</span>++<span class="sc">;</span> | 
|  | 260  247   <span class="s">}</span> | 
|  | 261  248 | 
|  | 262  249   <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdCount</span> <= <span class="i">$MaxCmpdsPerFile</span><span class="s">)</span> <span class="s">{</span> | 
|  | 263  250     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Total number of compounds, $CmpdCount, is <= specified number of compunds per file, $MaxCmpdsPerFile\n"</span><span class="sc">;</span> | 
|  | 264  251     <span class="k">return</span><span class="sc">;</span> | 
|  | 265  252   <span class="s">}</span> | 
|  | 266  253 | 
|  | 267  254   <span class="i">SplitSDFileByNumOfFilesAndCmpds</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$MaxNumOfFiles</span><span class="cm">,</span> <span class="i">$MaxCmpdsPerFile</span><span class="s">)</span><span class="sc">;</span> | 
|  | 268  255 <span class="s">}</span> | 
|  | 269  256 | 
|  | 270  257 <span class="c"># Split SD files into specified number of files with specified number of compounds</span> | 
|  | 271  258 <span class="c"># in each file...</span> | 
|  | 272  259 <span class="c">#</span> | 
|  | 273 <a name="SplitSDFileByNumOfFilesAndCmpds-"></a> 260 <span class="k">sub </span><span class="m">SplitSDFileByNumOfFilesAndCmpds</span> <span class="s">{</span> | 
|  | 274  261   <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$NumOfFiles</span><span class="cm">,</span> <span class="i">$NumOfCmpdsPerFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 275  262   <span class="k">my</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="i">$CmpdCount</span><span class="cm">,</span> <span class="i">$NewFileIndex</span><span class="cm">,</span> <span class="i">$NewFileName</span><span class="cm">,</span> <span class="i">$MaxCmpdsCount</span><span class="cm">,</span> <span class="i">@NewSDFilesList</span><span class="s">)</span><span class="sc">;</span> | 
|  | 276  263 | 
|  | 277  264   <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 278  265 | 
|  | 279  266   <span class="k">if</span> <span class="s">(</span>!<span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFile"</span><span class="s">)</span> <span class="s">{</span> | 
|  | 280  267     <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: Couldn't open it: $! \n"</span><span class="sc">;</span> | 
|  | 281  268     <span class="k">return</span><span class="sc">;</span> | 
|  | 282  269   <span class="s">}</span> | 
|  | 283  270 | 
|  | 284  271   <span class="c"># Setup new file names list...</span> | 
|  | 285  272   <span class="i">@NewSDFilesList</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 286  273   <span class="k">for</span> <span class="i">$NewFileIndex</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$NumOfFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 287  274     <span class="i">$NewFileName</span> = <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>] . <span class="q">"Part${NewFileIndex}."</span> . <span class="i">$SDFilesInfo</span>{<span class="w">FileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span> | 
|  | 288  275     <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 289  276       <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewFileName</span><span class="s">)</span> <span class="s">{</span> | 
|  | 290  277         <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: New SD file, $NewFileName, already exists\n"</span><span class="sc">;</span> | 
|  | 291  278         <span class="k">return</span><span class="sc">;</span> | 
|  | 292  279       <span class="s">}</span> | 
|  | 293  280     <span class="s">}</span> | 
|  | 294  281     <span class="k">push</span> <span class="i">@NewSDFilesList</span><span class="cm">,</span> <span class="i">$NewFileName</span><span class="sc">;</span> | 
|  | 295  282   <span class="s">}</span> | 
|  | 296  283 | 
|  | 297  284   <span class="i">$MaxCmpdsCount</span> = <span class="i">$NumOfCmpdsPerFile</span><span class="sc">;</span> | 
|  | 298  285 | 
|  | 299  286   <span class="i">$CmpdCount</span> = <span class="n">0</span><span class="sc">;</span> | 
|  | 300  287   <span class="i">$NewFileIndex</span> = <span class="n">1</span><span class="sc">;</span> | 
|  | 301  288 | 
|  | 302  289   <span class="k">open</span> <span class="w">NEWSDFILE</span><span class="cm">,</span> <span class="q">">$NewSDFilesList[$NewFileIndex - 1]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewSDFilesList[$NewFileIndex -1]: $! \n"</span><span class="sc">;</span> | 
|  | 303  290   <span class="k">print</span> <span class="q">"\nGenerating $NewSDFilesList[$NewFileIndex - 1] file\n"</span><span class="sc">;</span> | 
|  | 304  291 | 
|  | 305  292   <span class="k">open</span> <span class="w">SDFILE</span><span class="cm">,</span> <span class="q">"$SDFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $SDFile: $! \n"</span><span class="sc">;</span> | 
|  | 306  293 | 
|  | 307  294   <span class="k">while</span> <span class="s">(</span><span class="q"><SDFILE></span><span class="s">)</span> <span class="s">{</span> | 
|  | 308  295     <span class="q">s/(\r\n)|(\r)/\n/g</span><span class="sc">;</span> | 
|  | 309  296     <span class="k">print</span> <span class="i">NEWSDFILE</span><span class="sc">;</span> | 
|  | 310  297 | 
|  | 311  298     <span class="k">if</span> <span class="s">(</span> <span class="q">/^\$\$\$\$/</span> <span class="s">)</span> <span class="s">{</span> | 
|  | 312  299       <span class="i">$CmpdCount</span>++<span class="sc">;</span> | 
|  | 313  300       <span class="k">if</span> <span class="s">(</span><span class="i">$NewFileIndex</span> <= <span class="i">$NumOfFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 314  301         <span class="k">if</span> <span class="s">(</span><span class="i">$CmpdCount</span> >= <span class="i">$MaxCmpdsCount</span><span class="s">)</span> <span class="s">{</span> | 
|  | 315  302           <span class="k">if</span> <span class="s">(</span><span class="i">$NewFileIndex</span> < <span class="i">$NumOfFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 316  303             <span class="k">close</span> <span class="w">NEWSDFILE</span><span class="sc">;</span> | 
|  | 317  304           <span class="s">}</span> | 
|  | 318  305           <span class="i">$NewFileIndex</span>++<span class="sc">;</span> | 
|  | 319  306           <span class="i">$MaxCmpdsCount</span> = <span class="i">$NumOfCmpdsPerFile</span> * <span class="i">$NewFileIndex</span><span class="sc">;</span> | 
|  | 320  307 | 
|  | 321  308           <span class="k">if</span> <span class="s">(</span><span class="i">$NewFileIndex</span> <= <span class="i">$NumOfFiles</span><span class="s">)</span> <span class="s">{</span> | 
|  | 322  309             <span class="k">open</span> <span class="w">NEWSDFILE</span><span class="cm">,</span> <span class="q">">$NewSDFilesList[$NewFileIndex - 1]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewSDFilesList[$NewFileIndex - 1]: $! \n"</span><span class="sc">;</span> | 
|  | 323  310             <span class="k">print</span> <span class="q">"Generating $NewSDFilesList[$NewFileIndex - 1] file...\n"</span><span class="sc">;</span> | 
|  | 324  311           <span class="s">}</span> | 
|  | 325  312         <span class="s">}</span> | 
|  | 326  313       <span class="s">}</span> | 
|  | 327  314     <span class="s">}</span> | 
|  | 328  315   <span class="s">}</span> | 
|  | 329  316   <span class="k">close</span> <span class="w">NEWSDFILE</span><span class="sc">;</span> | 
|  | 330  317 <span class="s">}</span> | 
|  | 331  318 | 
|  | 332  319 <span class="c"># Retrieve information about SD files...</span> | 
|  | 333  320 <span class="c">#</span> | 
|  | 334 <a name="RetrieveSDFilesInfo-"></a> 321 <span class="k">sub </span><span class="m">RetrieveSDFilesInfo</span> <span class="s">{</span> | 
|  | 335  322   <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">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="s">)</span><span class="sc">;</span> | 
|  | 336  323 | 
|  | 337  324   <span class="i">%SDFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 338  325   <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 339  326   <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">FileExt</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 340  327   <span class="i">@</span>{<span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 341  328 | 
|  | 342  329   <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SDFilesList</span><span class="s">)</span> <span class="s">{</span> | 
|  | 343  330     <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span> | 
|  | 344  331 | 
|  | 345  332     <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span> | 
|  | 346  333     <span class="i">$SDFilesInfo</span>{<span class="w">FileExt</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span> | 
|  | 347  334     <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span> | 
|  | 348  335 | 
|  | 349  336     <span class="i">$SDFile</span> = <span class="i">$SDFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span> | 
|  | 350  337     <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$SDFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 351  338       <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It doesn't exist\n"</span><span class="sc">;</span> | 
|  | 352  339       <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | 
|  | 353  340     <span class="s">}</span> | 
|  | 354  341     <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$SDFile</span><span class="cm">,</span> <span class="q">"sd sdf"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 355  342       <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SDFile: It's not a SD file\n"</span><span class="sc">;</span> | 
|  | 356  343       <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | 
|  | 357  344     <span class="s">}</span> | 
|  | 358  345 | 
|  | 359  346     <span class="c"># Setup output file root...</span> | 
|  | 360  347     <span class="i">$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> | 
|  | 361  348     <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> | 
|  | 362  349 | 
|  | 363  350     <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} && <span class="s">(</span><span class="i">@SDFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 364  351       <span class="k">my</span> <span class="s">(</span><span class="i">$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">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span> | 
|  | 365  352       <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span> | 
|  | 366  353         <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span> | 
|  | 367  354       <span class="s">}</span> | 
|  | 368  355       <span class="k">else</span> <span class="s">{</span> | 
|  | 369  356         <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span> | 
|  | 370  357       <span class="s">}</span> | 
|  | 371  358       <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span> | 
|  | 372  359     <span class="s">}</span> | 
|  | 373  360     <span class="k">else</span> <span class="s">{</span> | 
|  | 374  361       <span class="i">$OutFileRoot</span> = <span class="q">"$FileName"</span><span class="sc">;</span> | 
|  | 375  362     <span class="s">}</span> | 
|  | 376  363 | 
|  | 377  364     <span class="i">$SDFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span> | 
|  | 378  365     <span class="i">$SDFilesInfo</span>{<span class="w">FileExt</span>}[<span class="i">$Index</span>] = <span class="i">$FileExt</span><span class="sc">;</span> | 
|  | 379  366     <span class="i">$SDFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span> | 
|  | 380  367   <span class="s">}</span> | 
|  | 381  368 <span class="s">}</span> | 
|  | 382  369 | 
|  | 383  370 <span class="c"># Process option values...</span> | 
|  | 384 <a name="ProcessOptions-"></a> 371 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span> | 
|  | 385  372   <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 386  373 | 
|  | 387  374   <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span> | 
|  | 388  375 | 
|  | 389  376   <span class="i">$OptionsInfo</span>{<span class="w">CmpdsMode</span>} = <span class="i">$Options</span>{<span class="w">cmpdsmode</span>}<span class="sc">;</span> | 
|  | 390  377 | 
|  | 391  378   <span class="i">$OptionsInfo</span>{<span class="w">NumOfFiles</span>} = <span class="i">$Options</span>{<span class="w">numfiles</span>}<span class="sc">;</span> | 
|  | 392  379   <span class="i">$OptionsInfo</span>{<span class="w">NumOfCmpds</span>} = <span class="i">$Options</span>{<span class="w">numcmpds</span>}<span class="sc">;</span> | 
|  | 393  380 | 
|  | 394  381   <span class="i">$OptionsInfo</span>{<span class="w">DataField</span>} = <span class="q">''</span><span class="sc">;</span> | 
|  | 395  382   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^Cmpds$/i</span> && <span class="i">$Options</span>{<span class="w">cmpdsmode</span>} =~ <span class="q">/^DataField$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 396  383     <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">datafield</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 397  384       <span class="k">die</span> <span class="q">"Error: You must specify a value for \"-d, --DataField\" option in \"DataField\" value of \"-c, --CmpdsMode\" during \"Cmpds\" \"-m, --mode\" value. \n"</span><span class="sc">;</span> | 
|  | 398  385     <span class="s">}</span> | 
|  | 399  386     <span class="i">$OptionsInfo</span>{<span class="w">DataField</span>} = <span class="i">$Options</span>{<span class="w">datafield</span>}<span class="sc">;</span> | 
|  | 400  387   <span class="s">}</span> | 
|  | 401  388 | 
|  | 402  389   <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | 
|  | 403  390 | 
|  | 404  391   <span class="i">$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="n">0</span><span class="sc">;</span> | 
|  | 405  392 <span class="s">}</span> | 
|  | 406  393 | 
|  | 407  394 | 
|  | 408  395 <span class="c"># Setup script usage  and retrieve command line arguments specified using various options...</span> | 
|  | 409 <a name="SetupScriptUsage-"></a> 396 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span> | 
|  | 410  397 | 
|  | 411  398   <span class="c"># Retrieve all the options...</span> | 
|  | 412  399   <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 413  400 | 
|  | 414  401   <span class="i">$Options</span>{<span class="w">cmpdsmode</span>} = <span class="q">'RootPrefix'</span><span class="sc">;</span> | 
|  | 415  402   <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">'Files'</span><span class="sc">;</span> | 
|  | 416  403 | 
|  | 417  404   <span class="i">$Options</span>{<span class="w">numfiles</span>} = <span class="n">2</span><span class="sc">;</span> | 
|  | 418  405   <span class="i">$Options</span>{<span class="w">numcmpds</span>} = <span class="n">1</span><span class="sc">;</span> | 
|  | 419  406 | 
|  | 420  407 | 
|  | 421  408   <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">"cmpdsmode|c=s"</span><span class="cm">,</span> <span class="q">"datafield|d=s"</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"numfiles|n=i"</span><span class="cm">,</span> <span class="q">"numcmpds=i"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 422  409     <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 $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span> | 
|  | 423  410   <span class="s">}</span> | 
|  | 424  411   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 425  412     <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> | 
|  | 426  413       <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span> | 
|  | 427  414     <span class="s">}</span> | 
|  | 428  415     <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Error: Couldn't chdir $Options{workingdir}: $! \n"</span><span class="sc">;</span> | 
|  | 429  416   <span class="s">}</span> | 
|  | 430  417   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">cmpdsmode</span>} !~ <span class="q">/^(DataField|MolName|RootPrefix)$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 431  418     <span class="k">die</span> <span class="q">"Error: The value specified, $Options{cmpdsmode}, for option \"-c, --CmpdsMode\" is not valid. Allowed values: DataField, MolName, RootPrefix\n"</span><span class="sc">;</span> | 
|  | 432  419   <span class="s">}</span> | 
|  | 433  420   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(Cmpds|Files)$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 434  421     <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m, --mode\" is not valid. Allowed values: Cmpds, Files\n"</span><span class="sc">;</span> | 
|  | 435  422   <span class="s">}</span> | 
|  | 436  423   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">numfiles</span>} < <span class="n">2</span><span class="s">)</span> <span class="s">{</span> | 
|  | 437  424     <span class="k">die</span> <span class="q">"Error: The value specified, $Options{numfiles}, for option \"-n --numfiles\" is not valid. Allowed values: >= 2 \n"</span><span class="sc">;</span> | 
|  | 438  425   <span class="s">}</span> | 
|  | 439  426   <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">numcmpds</span>} < <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 440  427     <span class="k">die</span> <span class="q">"Error: The value specified, $Options{numcmpds}, for option \"-n --numcmpds\" is not valid. Allowed values: >= 1 \n"</span><span class="sc">;</span> | 
|  | 441  428   <span class="s">}</span> | 
|  | 442  429 <span class="s">}</span> | 
|  | 443  430 | 
|  | 444 <a name="EOF-"></a></pre> | 
|  | 445 <p> </p> | 
|  | 446 <br /> | 
|  | 447 <center> | 
|  | 448 <img src="../../../images/h2o2.png"> | 
|  | 449 </center> | 
|  | 450 </body> | 
|  | 451 </html> |